Insight into the direct interaction of Na+ with NhaA and mechanistic implications

Na+/H+ antiporters comprise a family of membrane proteins evolutionarily conserved in all kingdoms of life that are essential in cellular ion homeostasis. While several human homologues have long been drug targets, NhaA of Escherichia coli has become the paradigm for this class of secondary active transporters as NhaA crystals provided insight in the structure of this molecular machine. However, structural data revealing the composition of the binding site for Na+ (or its surrogate Li+) is missing, representing a bottleneck in our understanding of the correlation between the structure and function of NhaA. Here, by adapting the scintillation proximity assay (SPA) for direct determination of Na+ binding to NhaA, we revealed that (i) NhaA is well adapted as the main antiporter for Na+ homeostasis in Escherichia coli and possibly in other bacteria as the cytoplasmic Na+ concentration is similar to the Na+ binding affinity of NhaA, (ii) experimental conditions affect NhaA-mediated cation binding, (iii) in addition to Na+ and Li+, the halide Tl+ interacts with NhaA, (iv) whereas acidic pH inhibits maximum binding of Na+ to NhaA, partial Na+ binding by NhaA is independent of the pH, an important novel insight into the effect of pH on NhaA cation binding.

eukaryotic Na + /H + antiporters 8 . As a secondary transporter, NhaA harnesses the energy of the H + electrochemical gradient, across the membrane, by allowing downhill movement of H + to drive the uphill antiport of Na + (or Li + ) against their electrochemical gradient 19,20 .
NhaA's transport mechanism is believed to follow the conceptual alternating access model 21 , according to which conformational changes of the transporter expose the active site to either side of the membrane in alternating fashion. Hence, to understand the antiport mechanism, it is critical to gain insight into the individual reaction steps that comprise the transport cycle and their dynamics, the identification of the binding site(s) for Na + and Li + , and how cation binding is regulated and connected to other steps in the antiport cycle. However, to date, crystallographic support for the precise location of the cation binding site is still missing 11,12 . Nevertheless, many indirect observations have predicted that the Na + /Li + binding site includes Asp163 and Asp164 of TM V (Fig. 1). These residues are highly conserved as recently further analyzed by evolutionary analysis of the flood of 6597 putative antiporters sequences 22 . Furthermore, mutagenesis studies revealed that both residues are functionally irreplaceable [23][24][25][26] . Asp164 is exposed to the cytoplasm via the cytoplasmic funnel 11 , and both aspartates are located near the crossing between the extended chains of the TM IV/XI assembly. In ASBT, a homologue that shares the NhaA fold, the Na + binding site was identified in this region 27 . Whereas Li + binding to affinity-purified NhaA in detergent micelles using isothermal calorimetry (ITC) combined with mutagenesis 20,28 revealed, consistent with established uptake studies, strong pH dependence of Li + binding to NhaA, this approach failed to provide data for Na + binding. The failure to obtain ITC data for Na + binding most likely stems from two main reasons: (i) the Na + binding affinity appears to be an order of magnitude lower than that for Li + as deduced from the apparent K m of the ions in transport assays (both in membrane vesicles and purified NhaA reconstituted in proteoliposomes, the apparent K m for Na + transport is 0.2-0.5 mM and that of Li + is 0.02 mM 29 ), and (ii) NhaA is a functional dimer; whereas the ITC measurements were performed in the presence of n-dodecyl-β-D-maltopyranoside (DDM), recent studies revealed that DDM does not support NhaA dimerization that is required for optimal activity and stability 30 . Specifically increasing the concentration of DDM above 0.1% (w/w) delipidates NhaA at the dimer interface and splits the NhaA dimers into monomers while addition of cardiolipin reconstitutes the dimers. Also, using mass spectroscopy showed that cardiolipin binds to the NhaA dimer interface 31 , thereby stabilizing the NhaA dimer 30 .
Notwithstanding the importance of the NhaA Li + binding results, Na + , rather than Li + , is the most prevalent cation in nature, and excreting Na + from the cytoplasm is the main physiological role of NhaA. NhaA maintains a cytoplasmic Na + concentration of around 0.5 mM 32 , a value which is consistent with the apparent K m for Na + transport 29 . However, an apparent Na + binding affinity of 1-10 mM was previously estimated using solid supported membranes (SSM)-based electrophysiological measurements and competition between Na + and H + at the binding site has been suggested to underpin the pH regulation of NhaA 33,34 , excluding the previously suggested option of a pH sensor 35 . Hence, there is no consensus regarding the precise location, organization and the determinants of the Na + (and Li + ) binding site of NhaA, and whether the pH controls Na + binding and/or transport at physiological cytoplasmic Na + concentrations by controlling a rate-limiting step in the Na + transport cycle. The putative Na + binding site of NhaA. The crystal structure of NhaA (PDB entry 4ATV) is viewed parallel to the membrane as cylinder representation. The TMs are numbered by Roman numerals and packed into two domains (interface, left side and core, right side). Their separation is indicated by the blue dotted line. The NhaA fold, with topologically inverted TM IV and TM XI colored cyan is crossing each other where they are interrupted by an unwound portion (extended chains) that split them into cytoplasmic (c) and periplasmic (p) segments. The other TMs are colored grey. The cytoplamsic and periplasmic funnels are indicated by red lines, and the membrane by broken black lines. Functional important residues (Asp133, Asp163, Asp164, and Lys300) at the proposed binding site are shown in stick representation. www.nature.com/scientificreports/ To address these questions, here we applied the scintillation proximity-based radioligand-binding assay (SPA) to follow direct 22 Na + binding 36 to affinity-purified NhaA. This assay was previously successfully used for the determination of the Na + binding kinetics of many Na + -dependent transport proteins [37][38][39][40][41] . Our results provide insight into the molecular determinants of the long-sought direct interaction of Na + with NhaA. These data have wide ramifications for studies focusing on Na + resistance in plants or the development of drugs that target Na + /H + antiporters in humans.

Results and discussion
Assay optimization. We chose to determine the binding properties of 22 Na + to affinity-purified NhaA in DDM micelles using the SPA technique, which allows for the rapid and sensitive measurement of radioisotope binding without the need for a separation step 36 . For this purpose, we first optimized the amount of affinitypurified NhaA in the presence of 0.1% (w/v) DDM for the binding assay (Fig. 2a). Since it was shown that high concentrations of DDM delipidate NhaA, leading to the separation of the NhaA dimer into the monomeric form and reduced Na + /H + antiport activity, while cardiolipin restores dimerization in vivo and in vitro 30 , we performed this assay in the presence of 0.015% DDM and 5% cardiolipin (Fig. 2a). Whereas half-maximum saturation of 125 µg of the His-tag polyvinyl toluene (PVT) SPA beads in the assay format was reached between 150 and 200 ng of NhaA, a value that was independent of the experimental condition, maximum binding in 0.015% DDM/5% cardiolipin was about 10% higher than that observed when the assay was performed in the presence of 0.1% DDM. Testing the effect of the protein amount used in the binding assay for NhaA-D163N, a mutant of NhaA that features impaired Na + /H + antiport activity 42 revealed that only negligible binding of 50 µM 22 Na + above the non-proximity signal was observed, even when 1 µg of the mutant protein was used in the assay (Fig. 2a). The results of these pilot experiments documented that the SPA is a suitable platform to determine the direct binding of Na + to NhaA, an approach that was hitherto not possible with other methods.
Ion recognition by NhaA. Under physiological conditions NhaA specifically exports 1 Na + or 1 Li + from the cell in exchange for the cellular import of 2 H +19,20 , and, thus, is electrogenic as in each reaction cycle one net positive charge enters the cell. As mentioned above, the apparent antiport K m for Na + is ~ 0.5 mM, whereas that of Li + is ten-fold lower (~ 0.02 mM) in isolated everted membrane vesicles and NhaA proteoliposomes 2 .  Table 1 for the results in log and non-log scale. www.nature.com/scientificreports/ Building on the feasibility of the SPA to measure equilibrium binding of 22 Na + with NhaA, we next performed measurements to determine the interaction of different ions with NhaA. To test whether non-radioactive ions compete for binding of radiolabeled Na + with NhaA, we performed a competion assay in which NaCl, LiCl, KCl, and TlCl were added to the assay, and the reduction of 22 Na + binding was monitored. Previous transport measurements suggest that Na + and Li + bind to the same site in NhaA 19 . In contrast, there is no evidence that supports the interaction of K + with NhaA. However, although the ion radius of the monovalent halide Tl + (1.5 Å) is similar to that of K + (1.44 Å) but significantly larger than that of Na + (1.12 Å) 43,44 , Tl + was identified in the Na + binding sites of the Na + /H + antiporter of Pyrococcus abyssi (PaNhaP) 45 and in the Na + -dependent aspartate transporter Glt Ph 46 . Thus, it seemed feasible to speculate that Tl + can also interact with NhaA of E. coli. Binding of 50 µM [ 22 Na]Cl was measured in the presence of non-labeled NaCl, LiCl, KCl, and TlCl at 10 mM, a concentration that is about 20-and 500-fold larger than the reported apparent affinities for the antiport reaction of Na + and Li + , respectively. Based on the results of the pilot experiments (Fig. 2a), and to maintain a membranemimicking environment that was used for the cell (or proteoliosome)-based uptake studies 19,20 , we performed the assay in the presence of 0.015% DDM and 5% cardiolipin. Figure 2b shows that whereas binding of 50 µM [ 22 Na] Cl in the presence or absence of 10 mM KCl exhibited virtually identical levels, 10 mM NaCl or LiCl reduced binding by 95.1% and 99.8%, respectively. Binding of 50 µM [ 22 Na]Cl was reduced to 42.1% when the assay was performed in the presence of 10 mM TlCl, revealing that Tl + can indeed compete with Na + for binding to NhaA. To further investigate the binding kinetics of Tl + we determined the apparent affinity of Tl + binding to NhaA (expressed as the IC Tl + 50 , the concentration of Tl + that yields 50% reduction of 22 Na + binding) by measuring the effect of increasing concentrations of Tl + on 50 µM [ 22 Na]Cl binding. Fitting the isotherm of 22 Na + displacement by increasing concentrations of Tl + (Fig. 2c) yielded an IC Tl + 50 of ~ 11 mM (see Table 1 for the results of the nonlinear regression fitting in log and non-log scale). Furthermore, the calculated Hill coefficient of ~ 0.61, i.e., below unity, indicates the lack of cooperativity for Tl + binding, suggesting the presence of only one Tl + binding site. Given the complexity associated with the determination of the accurate Hill coefficient, it must be noted that this coefficient reflects cooperativity rather than the absolute number of binding sites in proteins [37][38][39][40][41] .
To compare the apparent affinity of Na + and Li + , the two only cations that were previously shown to be exchanged with H + by NhaA 19,20 , we performed assays in which increasing concentrations of NaCl or LiCl were used to displace 50 µM [ 22 Na]Cl bound to NhaA. Like for Tl + , we determined the IC Li + 50 as measure for the apparent affinity of Li + , whereas we determined the EC Na + 50 , representing the effective concentration of NaCl to isotopically replace 50% of bound 22 Na + . Taking into account recent findings that highlight the importance of the lipidation state of NhaA 30 , and the results from the pilot experiments (Fig. 2a), implying that the binding activity of NhaA is correlated to the assay conditions, we determined these measurements in the presence of two different concentrations of DDM (0.1% or 0.015%) and in the presence or absence of 5% cardiolipin. Binding of 50 µM [ 22 Na]Cl to 200 ng NhaA increased in the following order: 0.1% DDM (Fig. 3a) < 0.015% DDM (Fig. 3b) < 0.1% DDM and 5% cardiolipin (Fig. 3c) < 0.015% DDM and 5% cardiolipin (Fig. 3d); likewise, the EC Na + 50 and the IC Li + 50 decreased in the same order, meaning that the highest affinity for Na + and Li + was observed in 0.015% DDM plus 5% cardiolipin (Table 1). Also, the Hill coefficients of the fits for 50 µM [ 22 Na]Cl replacement by Na + or Li + were close to unity, reflecting the absence of binding cooperativity as would be expected for a single cation binding site. Importantly, the EC Na + 50 for Na + was found to be ~ 0.4 mM when the assay was performed with a low DDM concentration (0.015%) and in the presence of cardiolipin, similar to the apparent K m for Na + transport determined in proteoliposomes and membrane vesicles. In contrast, when the assay was performed in the presence of a higher concentration (0.1%) of DDM and in the absence of cardiolipin, the EC Na + 50 increased about five-fold (~ 2 mM). Likewise, a similar trend was observed when LiCl was used to displace bound 50 µM [ 22 Na]Cl to determine the IC Li + 50 , yielding values of ~ 0.5 mM or ~ 0.17 mM in 0.1% DDM or 0.015% DDM/5% cardiolipin, respectively. Hill coefficients of ~ 1 suggest the presence of one Li + binding site (Table 1). These data point out that the higher binding activity observed in the pilot experiments when the assay was performed in 0.015% DDM/5% cardiolipin (Fig. 2a) cannot be attributed to the specific effect of these two compounds on Na + binding www.nature.com/scientificreports/ per se in the assay, but they rather reflect that lipids play a critical role in the functionality of NhaA, consistent with a recent study 30 . Remarkably, the apparent affinity found here for Na + (~ 0.4 mM) equals the physiological intracellular Na + concentration 32 , supporting the contention that NhaA is the main antiporter in Na + homeostasis of E. coli and possibly similar bacterial cells. On the other hand, for unknown reason, the apparent affinity of Li + (0.17 mM) found here is ten-fold lower than expected from the apparent K m for Li + transport in isolated membranes or NhaA proteoliposomes (0.02 mM).
Effect of pH on Na + binding by NhaA. Since it was shown that the antiport acitivity of NhaA is drastically dependent on the pH 18 , this dependence may reflect the pH-mediated regulation of either the passage of Na + through NhaA during the transport process across the membrane or the binding of Na + to NhaA or both. All options seem feasible based on described mechanisms of ion-coupled transport 13 that invloves the exchange of two H + with one Na + (or Li + ) 19,20 . Since it has been suggested that the Na + (or Li + ) binding site also binds H +34 , a high H + concentration (low pH) may block access of Na + (or Li + ) to such a site that is shared for binding with H + , thereby reflecting an apparent pH regulation through competetion for the same site by the different ions. Consistent with this notion, previous ITC measurements 20,28 showed that Li + binding occurs only at pH values ≥ 8.0. However, antiport data showing significant H + /Na + (or Li + ) antiport activity by NhaA wild-type at pH ≥ 6.5, thus suggesting that differences in the experimental conditions between the different assays may (partially) obscure the pH dependence of NhaA. In particular, the ITC experiments were performed at higher detergent concentration and in the absence of cardiolipin, conditions that were shown to interfere with the functionally important dimerization of NhaA 30 . Whereas such a competetive mechanism between H + and Na + (or Li + ) binding for the same site may exist, it cannot be the only scenario of pH regulation as our SPA analysis shows that about 50% of Na + binds in a pH independent fashion (Fig. 4). Our SPA study reports equilibrium binding of 22 Na + to detergent-solubilized NhaA under non-compartmentalized conditions; it displays the interaction of Na + with NhaA in the absence of any vectorial transmembrane ion gradient (i.e., ΔpH, � ∼ µ H + ) that may control the state of the protein in Na + -binding or non-binding conformation but rather reflects the protonation state of the cation active site at the experimentally selected pH. The results summarized in Fig. 4 show that: (i) direct Na + binding to NhaA is drastically dependent on the pH, consistent with the results of the antiport measurements 18 . Maximum Na + binding was observed when the H + concentration was low (pH ≥ 8.0), indicating a tight regulation of Na + binding to NhaA by H + ; (ii) between pH 6.0 and 7.0 the binding is about 50% of that observed at pH values ≥ 8.0, with a progressive increase in binding between pH 7.0 and pH 8.0. However, given the fact that 22 [Na + ] binding of about 40% was observed at high H + concentration (pH ≤ 6.0) when compared to that measured at pH ≥ 8.0, this result points out that fractional Na + binding to NhaA can occur in pH-independent manner. This finding contradicts a mechanism in which Na + binding is exclusively controlled by the deprotonation of residues in Na + coordination, i.e., by ion competetion 12 ; (iii) testing 22 Na + binding by NhaA-D163N at different pH values (Fig. 4) resulted in 22 Na + binding signals that were virtually indistinguishable from those obtained in the presence of imidazole, the agent that competes with the His-tagged NhaA for binding to the PVT His-tag  Table 1 www.nature.com/scientificreports/ SPA beads (Fig. 2a). This result revealed this mutant's inability to bind Na + at any pH tested and implies that the lack of Na + /H + antiport activity 42 can be attributed to this NhaA mutant's inability to bind Na + , thus identifying Asp163 as direct Na + coordinating partner. Since fractional Na + binding by NhaA wild-type was observed at low pH, fractional Na + binding would be expected for NhaA-D163N if the mutation would impact H + binding, thus ruling out an effect of the mutation on H + binding that, in turn, would impair Na + binding or transport. Taken together, these results raise the question as to which residues are involved in pH dependent and pH independent Na + binding. Despite the fact that to date a high-resolution structure of Na + (or Li + )-bound NhaA is missing, biochemical and biophysical measurements, as well as molecular dynamics (MD) simulations indicate that Asp163 and Asp164 are involved in Na + binding 12,28 . Since Asp164 is strictly conserved in all Na + /H + antiporters (including those that mediate the electroneutral stoichiometric exchange of 1 Na + and 1 H + ), this residue is generally accepted, in addition to being a Na + binding site, as the conserved H + binding site in this class of proteins 22,28,34,47 . Previous studies suggest that the second H + (in the electrogenic Na + /H + antiporters like NhaA) may bind to Asp133 or Asp163 48 . However, Asp133 is despensable for function 25,28 and has been shown to contribute its peptide-bond carbonyl to bind Na +49 . Furthermore, state-of-the-art continuous constant pH molecular dynamics simulations of NhaA K300 mutants suggested that in these mutants the electrogenic transport is maintained by utilizing Asp133 as an alternative proton-binding residue 50 . Therefore, and based on the recent evolutionary anlysis and metadynamics simulations 51 , it seemed feasible to speculate that the irreplaceable Asp163 (like Asp164) possesses a central dual function as binding partner for H + and Na + of the electrogenic (2 H + /1 Na + ) exchange mechanism 52 . The crystal structure of the dimeric NhaA and MD simulations point out the formation of a salt bridge between Asp163 and Lys300 in NhaA that is interrupted when Asp163 participates, together with Asp164, in the binding of Na +12 . In this vein, Lys300 was shown to be important in pH regulation and antiport activity of NhaA 26 mainly because the salt bridge is important for NhaA stability 53 . In fact, in the NhaA homolog Na + /H + antiporter NhaP from Pyrococcus abyssi (PaNhaP), which was crystallized in ion-bound state (here, the Na + surrogate Tl + was used), the ion binding site was characterized as distorted trigonal bpyramide 45 . Here, ion coordination is achieved through the direct interaction of the carboxyl groups of Glu73 and Asp159, whereas the ion interacts through a bound water molecule with Asp130 and with the hydroxyl side chain of Ser155 and the main-chain carbonyl of Thr129. A similar trigonal bpyramidal Na + coordination was identified for the ion in the Na2 site in LeuT, whereas Na + in the Na1 site of LeuT features an octahedral coordiantion 54 , pointing to the involvement of ≥ six coordination partners for Na + . As our data shows that Tl + binds to NhaA (Fig. 2), it seems interesting to propose that Na + (or Li + ) coordination in NhaA, like that in PaNhaP 45 or LeuT 54 , involves additional interactions with amino acid side chains and main chains that cannot be protonated/deprotonated. Consequently, such a coordination network for Na + (or Li + ) that is insensitive to H + may serve as binding scaffold for Na + (Li + ) in pH independent manner. In such a scenario it appears possible that deprotonation of protonable side chains of residues that are involved in Na + binding provide the activation energy that is required for the execution of the full H + /Na + exchange cycle.
In summary, our SPA-based direct Na + binding to NhaA data revealed that (i) the Na + binding affinity of NhaA is ~ 0.4 mM, implying that NhaA is the main antiporter for Na + homeostasis at 0.4 mM, the cytoplasmic concentration of E. coli and possibly similar cells, (ii) the membrane lipid cardiolipin plays a critical role in the cation binding functionality of NhaA, (iii) the halide Tl + interacts with NhaA, (iv) whereas acidic pH inhibits maximum binding of Na + to NhaA, partial Na + binding by NhaA is independent of the pH, an important novel insight into the effect of pH on NhaA cation binding.
By gaining insight into the molecular determinants of the interaction of Na + with NhaA, these findings provide valuable information for the generation of more accurate mechanistic models of an essential reaction that is maintained throughout organisms from bacteria to man. Thus, these data have wide ramifications for studies focusing on Na + resistance in plants or the development of drugs that target Na + /H + antiporters in humans, e.g., in the treatment of hypertension and congestive heart failure.  Detection and quantification of NhaA. Total quantities of membrane-and affinity-purified proteins were determined using the Bradford assay 60 . In certain cases, quantitation of the affinity-purified NhaA was carried out by resolving the sample and a sample of known NhaA concentration on SDS-PAGE, staining the gels with Coomassie Blue and quantifying the band densities using Image Gauge (Fuji) software.

SPA-based binding assay.
Binding of radiolabeled Na + to His-tagged NhaA was measured with the SPA method 36 . The assay was modified to test the effect of detergent (DDM) and cardiolipin (CL) on NhaA's binding activity. Basic assay conditions, unless otherwise indicated, involved the immobilization of 200 ng of affinitypurified NhaA to Cu 2+ -coated polyvinyl toluene (PVT) SPA beads (at a concentration of 1.25 mg ml −1 ) in 100 µL assay buffer composed of 200 mM tris(hydroxymethyl)aminomethane (Tris)/2-(N-morpholino)ethanesulfonic acid (Mes), pH 8.0, 20% glycerol, 0.1 mM tris(2-carboxyethyl)phosphine (TCEP; Sigma), 0.1% (w/v) DDM (Anatrace). When indicated, the concentration of DDM was reduced to 0.015% (w/v) and/or cardiolipin was added at a concentration of 5% (w/v). Optimized binding conditions were evaluated by varying the amount of NhaA from 25 to 1000 ng per 100-µL assay (Fig. 2a) MicroBeta photomultiplier tube (PMT) plate counter. The non-proximity background signal was determined for each sample in the presence of 800 mM imidazole, which prevents the interaction of the His-tagged protein with the Cu 2+ -coated SPA beads, and the non-proximity cpm was subtracted from the cpm determined in the absence of imidazole to obtain the specific binding activity. Data were normalized with regard to the activity of wild-type NhaA in the absence of non-labelled ligand and was set as 100%. All experiments were performed at least in triplicate with technical replicas of 3. Data were expressed as mean ± s.e.m. Data fits of binding isotherms were performed using unweighted, unconstrained nonlinear regression algorithms in Prism 8 (GraphPad), and errors represent the s.e.m. of the fit. The goodness of the fit (R 2 ) and the range are shown in Table 1.