In the course of a previous phylogenetic analysis, we were able to spot seven homologues of the prototypic NRAT1 from Oryza sativa (XP_015625418.1, Uniprot: Q6ZG85) in Blast searches of eukaryotic sequence databases (Ramanadane et al., 2022). All identified proteins contain the described hallmarks of NRATs, which distinguishes this clade of the SLC11 family from the bulk of transition metal ion transporters (Figure 1—figure supplement 1). Differences include variations in the putative ion coordinating site, where the conserved DPGN motif on α1 of classical NRAMPs is changed to DPSN and the A/CxxM motif on α6 to AxxT (Lu et al., 2018; Ramanadane et al., 2022). Latter replaces the conserved methionine, whose sulfur atom is involved in transition metal ion coordination, with a threonine, which offers a hard oxygen ligand for metal ion interactions. To identify proteins that are suitable for structural and functional investigations, we have generated constructs of five NRAT homologues, each containing a C-terminal fusion of eGFP and an SBP tag used for affinity purification preceded by a 3C protease cleavage site. We have subsequently investigated their expression upon transient transfection of HEK293 cells followed by an initial biochemical characterization by fluorescence size exclusion chromatography (Hattori et al., 2012). These experiments have singled out the protein from S. italica (termed SiNRAT, XP_004952002.1) as construct with comparably high expression level and promising biochemical behavior. We have subsequently scaled up the expression of SiNRAT in suspension culture and purified it in the detergent n-dodecyl-β-D-maltopyranoside (DDM) for further studies (Figure 1—figure supplement 2A, B).

For a characterization of its transport properties, SiNRAT was reconstituted into proteoliposomes and subjected to previously established fluorescence-based assays that have permitted the functional investigation of other members of the SLC11 family (Ehrnstorfer et al., 2014; Ehrnstorfer et al., 2017; Ramanadane et al., 2022). We have initially studied the transport of Mn²⁺, which is a common substrate of SLC11 transporters that can be monitored using the fluorophore calcein. Although, in a physiological context, SiNRAT presumably operates at mildly acidic extracellular conditions (i.e. at a pH <5.5), all transport experiments were carried out at neutral pH since this resulted in maximum proteoliposome stability and highest sensitivity of the employed fluorophores. The assay revealed a concentration-dependent quenching of calcein as a consequence of the influx of Mn²⁺ into the proteoliposomes (Figure 1A). Transport is strongly dependent on the Mn²⁺ concentration and saturates with an apparent K_m of about 14 µM, reflecting the binding of the transported substrate to a saturable site (Figure 1A and B). We subsequently investigated impairment of Mn²⁺ transport by the alkaline earth metal ions Ca²⁺ and Mg²⁺ and found a concentration-dependent competition by both ions, which is indicative for their specific binding to the protein (Figure 1C and D, Figure 1—figure supplement 2C, D). To investigate whether they would also be transported, we directly monitored their influx into proteoliposomes employing the fluorophores Fura-2 for the detection of Ca²⁺ and Magnesium Green for Mg²⁺. In both cases, we found a weak signal that saturates at µM concentrations, which reflects uptake of both ions (Figure 1E and F, Figure 1—figure supplement 2E-H).

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 1—source data 1

Transport and ITC data of SiNRAT.

https://cdn.elifesciences.org/articles/85641/elife-85641-fig1-data1-v1.zip

Download elife-85641-fig1-data1-v1.zip

After demonstrating that SiNRAT is capable of mediating the transport of divalent metal ions with poor selectivity, we were interested in the transport properties of Al³⁺, which is the proposed physiological substrate. However, assaying this ion under in vitro conditions is complicated by the lack of suitable fluorophores and the fact that the chemical properties of Al³⁺, as strong Lewis acid with an ionic radius of 0.53 Å, leads to the destabilization of liposomes even at low µM concentrations. We thus turned our initial attention to the trivalent cation Ga³⁺, which is one period below Al³⁺ in the same group of the periodic system, whose larger ionic radius (0.62 Å) and consequent weaker Lewis acidity is less disruptive for lipid bilayers. In an assay to investigate the interference with Mn²⁺ transport, we found an inhibition by Ga³⁺, which emphasizes its interaction with the transporter (Figure 1G). Conversely, we did not observe such competition with the prokaryotic transition metal ion transporter EcoDMT, excluding the action of Ga³⁺ as non-specific inhibitor of the family (Figure 1—figure supplement 2I). Thus, despite the low solubility of Ga³⁺ under the assayed conditions (Yu and Liao, 2011), which obscures a definitive assignment of its concentration dependence, our results suggest a specific effect of the ion on SiNRAT-mediated transport. Since equivalent studies with Al³⁺ are prevented by its disruptive properties on membrane integrity, we used isothermal titration calorimetry (ITC) to investigate the binding of the ion to SiNRAT in detergent solution. These experiments showed a specific signal upon Al³⁺ titration as a consequence of its interaction with the transporter (Figure 1H). However, at the neutral pH of the sample, the trivalent ion is at equilibrium with its hydroxyl complexes, which complicates data analysis (Kiss, 2013). The thermograms were thus best fitted to a sum of two binding constants, both in the low micromolar range (Figure 1I). Although precluding a definitive mechanistic interpretation, we use the ITC results of Al³⁺ binding as a qualitative assay of its interaction with SiNRAT and show later that the event is directly associated with its binding to the site that is relevant for ion transport.

We then investigated whether metal ion transport would be accompanied by a concentration-dependent acidification of vesicles, which is expected in case of coupled proton symport as observed for the prokaryotic SLC11 transporter EcoDMT (Figure 1J; Ehrnstorfer et al., 2017). To this end, we have assayed the import of H⁺ at increasing concentrations of the transported ions Mn²⁺, Ca²⁺, Mg²⁺ and the putative substrate Ga³⁺ but did in no case detect evidence for metal ion concentration-dependent decrease of ACMA fluorescence, despite the negative membrane potential established by an outwardly directed K⁺ gradient, which would facilitate H⁺ transport (Figure 1K and L, Figure 1—figure supplement 2J, K). These findings suggest that, akin to the Mg²⁺ transporter EleNRMT, SiNRAT acts as uncoupled transporter that facilitates the bidirectional transport of metal ions with poor selectivity.

The described transport properties of SiNRAT are unique among characterized members of the SLC11 family. To define the underlying molecular features, we set out to determine its three-dimensional (3D) structure. Since the protein did not crystallize, we generated nanobody-based binders that, in complex with SiNRAT, increase the size of the protein sufficiently to permit its structure determination by cryo-electron microscopy (cryo-EM). These binders were obtained by immunization of alpacas with the purified protein and subsequently selected by phage display from a library generated from a blood sample (Pardon et al., 2014). The procedure allowed the identification of the nanobody Nb1^SiNRAT (short Nb1) as promising interaction partner that does not dissociate during size exclusion chromatography, and which inhibits Mn²⁺ transport in a concentration-dependent manner when added to the outside of liposomes (Figure 2—figure supplement 1). The incomplete inhibition of transport by Nb1 presumably reflects the mixed distribution of reconstituted SiNRAT residing in either inside-out or outside-out orientation, where the epitopes recognized by the nanobody are only accessible in one orientation. For the SiNRAT-Nb1 complex, we have prepared samples for cryo-EM. Since data recorded in the detergent DDM did not permit a 3D reconstruction at high resolution, we have reconstituted the complex into amphipols (Kleinschmidt and Popot, 2014) and collected a large dataset, which ultimately yielded a cryo-EM density map at 3.66 Å (Figure 2—figure supplements 1–3, Table 1, Table 2). Due to the challenges associated with the amphipol reconstitution process, the sample used for imaging did not contain transported ions. In case of trivalent ions this would have required a lowering of the pH to increase their solubility, which was incompatible with nanobody binding. The resulting map was of high quality and allowed the unambiguous interpretation with an atomic model revealing details of the 3D organization of SiNRAT (Figure 2A, Figure 2—figure supplement 3).

Figure 2 with 4 supplements see all

Download asset Open asset

The SiNRAT structure in complex with Nb1 is depicted in Figure 2. It shows a protein sharing the conserved architecture of the SLC11 family with a core consisting of a pair of structurally related repeats of five transmembrane spanning segments that are oriented in the membrane with opposite directions in an arrangement that was initially observed in the amino acid transporter LeuT (Yamashita et al., 2005). In this architecture, the unwound centers of the first helix of each repeat together constitute a substrate binding site (Figure 2—figure supplement 4A). The protein contains a total of 12 membrane-inserted helices akin to other eukaryotic family members and the prokaryotic EcoDMT (PDBID: 5M87) (Ehrnstorfer et al., 2017), whose Cα-positions superimpose with an RMSD of 2.66 Å (Figures 2B and 3A, Figure 2—figure supplement 4B). The nanobody binds to an extended interface located at the extracellular side. These interactions involve contacts of all three CDRs of the nanobody, which bridge the core domain of SiNRAT consisting of α-helices 1–10 with α-helices 11–12 located at its periphery, thereby contacting the loops α1–2, α5–6, α7–8 and residues on α11 and burying 1682 Ų of the combined molecular surface (Figure 2—figure supplement 4C, D). Unlike in EcoDMT, in this complex α11 and α12 are detached from the core domain on the extracellular part in a conformation that is presumably stabilized by Nb1 binding (Figure 2B, Figure 2—figure supplement 4B, C).

Figure 3 with 1 supplement see all

Download asset Open asset

The observed structure resembles an inward-occluded state of the classical SLC11 transporter DraNRAMP (PDBID: 8E60, DraNRAMP^occ), where most of the structure retains a conformation observed in the inward-facing structure of the same protein (PDBID: 6D9W, DraNRAMP^inw) except for a movement of the intracellular part of α-helix 1 (α1a), which has rearranged to close the access path to the ion binding site (Figure 3, Figure 3—figure supplement 1A, B; Bozzi et al., 2019b; Ray et al., 2023). This resemblance is reflected in an RMSD of Cα atoms of 1.57 Å upon a superposition of SiNRAT with DraNRAMP^occ and 1.74 Å with DraNRAMP^inw, whereas the RMSD of 2.71 Å compared to the outward-facing conformation of the same protein (PDBID: 6BU5, DraNRAMP^out) is considerably larger (Figure 3C, Figure 3—figure supplement 1C). As in DraNRAMP^occ, the shorter α1a of SiNRAT is located closer to α6b compared to DraNRAMP^inw, resulting in a narrowing of an aqueous cavity leading toward the supposed ion binding site, thereby separating this site from the intracellular environment (Figure 3C, Figure 3—figure supplement 1D). We thus assume that the SiNRAT structure represents an intermediate between both endpoints of a transporter, functioning by an alternate access mechanism, that is closer to an inward-facing state. In its conformation, the binding site is sealed to both sides of the membrane by a thin gate composed of a single layer of residues (i.e. Asp 66, Asp 140, Thr 241, Tyr 243, and Thr 346) toward the inside and a thick gate (encompassing several layers of residues and spanning about 15 Å) toward the outside (Figure 3—figure supplement 1D).

Although for the reasons described above, we have not explicitly added transported ions to our sample, we find residual cryo-EM density bridging the neighboring residues Asp 66 and Ser 68 which, as part of the supposed ion binding site, are both located on the unwound loop connecting α1a and α1b (Figure 4A). This density presumably either originates from water, the monovalent cation Na⁺ that is contained in the sample at a concentration of 200 mM, or traces of divalent metal ions found in our solution. Due to its proximity to binding sites identified in structures of other family members, we use this position as reference to analyze plausible protein substrate interactions, which nevertheless should be considered as tentative. In this putative site, Asp 66 is particularly striking, since it is universally conserved and was shown to constitute a key component for ion coordination in the entire SLC11 family. In contrast, Ser 68 is unique to NRATs and substituted by a Gly in all other members (Figure 1—figure supplement 1). The comparison of the SiNRAT site with the ion interaction in DraNRAMP^occ shows common features but also differences that may underlie the altered substrate preference of SiNRAT. In both structures, the water-filled paths to the metal ion binding site are sealed from both sides of the membrane and bound ions would thus be segregated from their aqueous environment, although the occlusion toward the intracellular solution is more pronounced in DraNRAMP^occ (Figures 3C, 4B and C, Figure 3—figure supplement 1B, D). In its binding site, the interacting transition metal ion in DraNRAMP^occ is predominantly enclosed by protein residues except for one contact with a small pocket harboring trapped water molecules (Ray et al., 2023). The side chains of Asp 56 and Asn 59 on α1, in conjunction with the backbone carbonyls of Ala 53 and Ala 227, tightly surround the bound Mn²⁺. Additionally, the thioether group of Met 230 contributes a soft ligand for ion interactions (Figure 4C). Finally, a water-mediated interaction with the side chain of Gln 378 on α10 completes the octahedral coordination of the ion (Figure 4C). In contrast, the equivalent region in SiNRAT contains an elongated, approximately cylindrical pocket of predominant polar character that is about 10 Å long and 4–5 Å wide and that encloses a three times larger volume than in DraNRAMP (Figure 4B). This pocket appears of sufficient size to mediate metal ion interactions with protein residues and surrounding water molecules. Despite the inability to pinpoint the exact location of bound ions, we find residues of the consensus binding site to line about half of this pocket (Figure 4B). These include the previously mentioned Asp 66 and Ser 68, both bordering residual density in our cryo-EM data and the conserved Asn 69, all located on α1. On α6, the interacting residues include the conserved Ala 238, whose backbone carbonyl points toward the pocket and Thr 241, which replaces the conserved methionine in NRAMP transporters and whose side chain could contribute a hard ligand for either direct or water-mediated ion interactions (Figure 4B). Distinct from DraNRAMP, the glutamine on α10 involved in ion binding is replaced by a serine (Ser 403), whose shorter side chain does not contact the aqueous pocket and thus would not be available for metal ion interactions. Instead, we find the acidic side chain of Asp 140 located on α3 to line the cavity, close to the predicted ion binding site, at a position that contains a threonine (Thr 130) in DraNRAMP. This acidic side chain, which is conserved in NRATs, would increase the negative charge density of the region by offering favorable coulombic interactions that stabilize the additional positive charge of Al³⁺ (Figure 4B, Figure 1—figure supplement 1). Thus, besides the replacement of the conserved methionine of the transition metal ion binding site, we find an expanded interaction region with increased polar character and negative charge density, which together may account for the altered substrate preference and stabilize trivalent cations.

Figure 4 with 1 supplement see all

Download asset Open asset

Besides the altered ion selectivity, the second feature that distinguishes SiNRAT from classical NRAMPs is the apparent lack of H⁺ coupling, which is also shared by prokaryotic NRMTs of the same family (Ramanadane et al., 2022). Although the detailed mechanism of H⁺ coupling in the SLC11/NRAMP family is still a matter of debate, there are remarkable structural differences between classical H⁺-coupled metal ion transporters, SiNRAT, and the Mg²⁺ transporter EleNRMT (Figure 4—figure supplement 1). In NRAMPs, H⁺ transport was associated with acidic and basic residues located on the α-helices 1, 3, 6, and 9, most of which are replaced by hydrophobic side chains in NRMTs (Bozzi et al., 2019a; Bozzi et al., 2020; Ehrnstorfer et al., 2017; Ramanadane et al., 2022; Figure 1—figure supplement 1, Figure 4—figure supplement 1B, C). In contrast, many of these residues are conserved in NRATs, which are evolutionally closer to NRAMPs (Figure 1—figure supplement 1, Figure 4—figure supplement 1A). A pronounced difference between NRATs and NRAMPs concerns the substitution of a conserved histidine on α6b (i.e. His 232 in DraNRAMP) (Ehrnstorfer et al., 2017; Lam-Yuk-Tseung et al., 2003) with a tyrosine (Tyr 243 in SiNRAT), whereas the equivalent position in NRMTs is occupied by a tryptophane (i.e. W229 in EleNRMT) (Figure 1—figure supplement 1, Figure 4—figure supplement 1). However, the exact relation of the described structural features to the observed transport mechanism is still unclear and requires further investigation.

After defining the structure of the putative metal ion binding site of SiNRAT, we decided to study the effect of point mutants on the transport properties of the protein. To this end, we have put our focus on residues of the consensus binding site located on α-helices 1 and 6, whose equivalent positions were found to contribute to ion interaction in transition metal ion transporters of the SLC11 family (Figure 5A). All investigated mutants were purified and either reconstituted into liposomes for uptake studies of Mn²⁺ or used for ITC experiments in their detergent-solubilized form. Upon mutation of the residues Asp 66 or Asn 69 to alanine, we found a severely compromised transport of Mn²⁺, consistent with equivalent results obtained for other pro- and eukaryotic transition metal ion transporters (Figure 5B and C; Bozzi et al., 2019b; Ehrnstorfer et al., 2014; Ehrnstorfer et al., 2017; Ramanadane et al., 2022). We next turned our attention toward Thr 241, which has substituted the conserved methionine found in transition metal ion transporters to provide a potential hard ligand for direct or water-mediated metal ion interactions. In this case, we have mutated the residue to alanine to truncate the side chain and valine to preserve its volume but remove its polar character. In both cases, we found a similar strong negative impact of the respective mutations on Mn²⁺ transport (Figure 5D and E). For this residue, we were also interested in whether a mutation would affect interactions with trivalent cations and thus investigated binding of Al³⁺ to the mutant T241V by ITC. Unlike WT, we did not detect any specific heat exchange, which emphasizes the importance of the residue for Al³⁺ interactions (Figure 5F). The absence of a detectable signal in T241V also underlines the correspondence of the ITC data of WT to Al³⁺ binding to the consensus site (Figure 1H). Finally, we have studied a mutation of Ser 68 to alanine, which concerns a residue that is unique to NRATs and which was found in the SiNRAT structure to contribute to potential ion interactions (Figure 4A, Figure 1—figure supplement 1). Unlike previous mutations, we did in this case not observe an obvious effect on Mn²⁺ transport, which proceeds with similar kinetics as WT (Figure 5G and H) and whose transport is competed by the addition of Mg²⁺ and Ca²⁺ in a similar manner (Figure 5I and J). Unlike WT, we did not detect a competition by Ga³⁺ in equivalent experiments (Figure 5K). We thus set out to investigate the binding of Al³⁺ by ITC but did not find any indication for its interaction with the mutant (Figure 5L). Taken together, our experiments on the SiNRAT mutant S68A showed distinct phenotypes for the interaction with di- and trivalent cations, which points toward structural differences in their coordination with more stringent requirements for Ga³⁺ and Al³⁺ binding.

Figure 5

Download asset Open asset

Figure 5—source data 1

Transport and ITC data of SiNRAT mutants.

https://cdn.elifesciences.org/articles/85641/elife-85641-fig5-data1-v1.zip

Download elife-85641-fig5-data1-v1.zip

EcoDMT and its mutants were expressed and purified as described previously (Ehrnstorfer et al., 2017). Briefly, the vector EcoDMT-pBXC3GH was transformed into E. coli MC1061 cells and a preculture was grown in TB-Gly medium overnight using ampicillin (100 μg/ml) as a selection marker for all expression cultures. The preculture was used at a 1:100 volume ratio for inoculation of TB-Gly medium. Cells were grown at 37°C to an OD₆₀₀ of 0.7–0.9, after which the temperature was reduced to 25°C. Protein expression was induced by addition of arabinose to a final concentration of 0.004% (wt/vol) for 12–14 hr at 18°C. Cells were harvested by centrifugation for 20 min at 5000 × g, resuspended in buffer RES (200 mM NaCl, 50 mM KPi pH 7.5, 2 mM MgCl₂, 40 μg/ml DNAseI, 10 μg/ml lysozyme, 1 μM leupeptin, 1 μM pepstatin, 0.3 μM aprotinin, 1 mM benzamidine, 1 mM PMSF) and lysed using a high-pressure cell lyser (Maximator HPL6). After low-speed centrifugation (10,000 × g for 20 min), the supernatant was subjected to a second centrifugation step (200,000 × g for 45 min). The pelleted membrane was resuspended in buffer EXT (200 mM NaCl, 50 mM KPi pH 7.5, 10% glycerol) at a concentration of 0.5 g of vesicles per ml of buffer. The purification of EcoDMT and its mutants was carried out at 4°C. Isolated membrane fractions were diluted 1:2 in buffer EXT supplemented with protease inhibitors (Roche cOmplete EDTA-free) and 1% (wt/vol) n-decyl-β-D-maltopyranoside (DM, Anatrace). The extraction was carried out under gentle agitation for 2 hr at 4°C. The lysate was cleared by centrifugation at 200,000 × g for 30 min. The supernatant supplemented with 15 mM imidazole at pH 7.5 was subsequently loaded onto NiNTA resin and incubated for at least 1 hr under gentle agitation. The resin was washed with 20 column volumes (CV) of buffer W (200 mM NaCl, 20 mM HEPES pH 7, 10% glycerol, 50 mM imidazole pH 7.5, 0.25% DM) and the protein was eluted with buffer ELU (200 mM NaCl, 20 mM HEPES pH 7, 10% glycerol, 200 mM imidazole pH 7.5, 0.25% DM). Tag cleavage proceeded by addition of HRV 3C protease at a 3:1 molar ratio and dialysis against buffer DIA (200 mM NaCl, 20 mM HEPES pH 7, 10% glycerol, 0.25% DM) for at least 2 hr at 4°C. After application to NiNTA resin to remove the tag, the sample was concentrated by centrifugation using a 50 kDa molecular weight cut-off concentrator (Amicon) and further purified by size exclusion chromatography using a Superdex S200 column (GE Healthcare) pre-equilibrated with buffer SEC (200 mM NaCl, 20 mM HEPES pH 7, 0.25% DM). The peak fractions were pooled, concentrated, and used directly for further experiments.

HEK293S GnTI^– and HEK293T cells were obtained from ATCC. HEK293S GnTI^– cells were grown in HyCell HyClone Trans Fx-H media supplemented with 1% FBS, 4 mM L-glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin, and 1.5 g/l Kolliphor-188 at 37°C and 5% CO₂. Adherent HEK293T cells were grown in DMEM-Glutamax, supplemented with 10% FBS, 100 U/ml penicillin and 100 µg/ml streptomycin. All cell lines were regularly tested for prevention of mycoplasma contamination.

The DNA coding for NRATs from different plants (i.e. the proteins from O. sativa, Uni prot: Q6ZG85; D. oligosanthes, NCBI: OEL35611.1; S. bicolor, NCBI: XP_002451480.2; S. italica, NCBI: XP_004952002.1; Z. mays, NCBI: PWZ19830.1) was synthesized by Genscript and cloned into a pcDNA3.1 vector (Invitrogen) using the FX cloning technique (Geertsma and Dutzler, 2011) yielding a final construct where each NRAT1 contains an HRV 3C protease cleavage site, eGFP, a Myc tag, and a streptavidin binding protein (SBP) attached to either its N- or C-terminus. Adherent HEK 293T cells (ATCC CRL-1573) were used for expression screening of plant NRAT1 homologues. Cells were grown on plates in DMEM-Glutamax High Glucose supplemented with 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin. Cells were grown and split to 60–70% confluency 1 day prior to transfection. Cells were transiently transfected using polyethyleneimine (PEI) 25 K. Briefly, for each 10 ml plate, 10 µg of DNA in 500 µl of non-supplemented DMEM was mixed with 40 µg of PEI 25 K in 500 µl of non-supplemented DMEM. After 15 min incubation, the transfection mixture was added to the plate together with 3 mM valproic acid. Cells were grown at 37°C and 5% CO₂ for 48–60 hr and harvested by centrifugation at 1000 × g for 15 min. Pellets were resuspended in buffer LYS (200 mM NaCl, 20 mM HEPES pH 7, 10% glycerol, 2 mM MgCl₂, 40 µg/ml DNAse I, protease inhibitors [Roche Complete, EDTA-free], and 1% DDM). The lysate was incubated at 4°C for 1.5 hr under gentle agitation and clarified by centrifugation at 200,000 × g for 15 min. The clarified whole-cell extract was subjected to fluorescent size exclusion chromatography (Agilent HPLC) using a Superdex S200 column (GE Healthcare) and detected with a fluorescence detector set to λ_ex = 489 nm/λ_em = 510 nm. SiNRAT from S. italica (Uniprot KB K3YRE7) with a C-terminal tag was selected as the biochemically best behaving homologue based on the monodispersity of the eluted peak (Hattori et al., 2012).

A plasmid carrying the sequence coding for the NRAT from the plant S. italica (SiNRAT) (Uniprot KB K3YRE7) cloned into a pcDNA3.1 vector (Invitrogen) containing an HRV 3C protease cleavage site, eGFP, a Myc and an SBP tag attached to its C-terminus was used for all further experiments. All mutants were generated using the QuickChange site-directed mutagenesis method (Agilent) (Table 2). SiNRAT was expressed by transient transfection using HEK293S GnTI^- cells diluted to 0.6–0.8·10⁶ cells/ml 1 day prior to transfection. DNA for transfection was purified from E. coli MC1061 cell line using the Nucleobond Xtra Maxi Kit (Macherey-Nagel); 1.3 µg DNA was used per 10⁶ cells. For that purpose, the DNA was diluted to 16 µg/ml in a volume of non-supplemented DMEM corresponding to 1/14 of the transfected cell culture volume. In parallel, PEI-max 40 kDA was diluted to 0.04 mg/ml in the same volume of non-supplemented DMEM. The two solutions were mixed and after 15–20 min incubation, the final transfection mixture was added to the suspension cells. Finally, the cell culture was supplemented with 3 mM valproic acid. After 48–60 hr incubation, cells were harvested by centrifugation at 1500 × g for 20 min and washed with PBS. All following steps of the purification were carried out at 4°C. The cell pellets were resuspended in buffer RES (200 mM NaCl, 50 mM KPi pH 7.5, 10% glycerol, 2 mM MgCl₂, 40 µg/ml DNAse I, and protease inhibitor [Roche Complete, EDTA-free]). The protein was extracted by adding DDM (Anatrace) to a final concentration of 2%. After an incubation for 2 hr under gentle agitation, the lysate was clarified by centrifugation at 200,000 × g for 30 min. The supernatant was loaded on Streptactin Superflow affinity resin (IBA Lifesciences) and incubated for 2 hr under gentle agitation. The resin was washed with 25 CV of buffer W (200 mM NaCl, 20 mM HEPES pH 7, 10% glycerol, 0.04% DDM) and eluted with 20 ml of buffer ELU (buffer W supplemented with 5 mM D-desthiobiotin). The eGFP-Myc-SBP tag was removed by adding HRV 3C protease at a 1:3 molar ratio with SiNRAT and incubated on ice for 1 hr. The protein was concentrated by centrifugation to 500 µl using a 50 kDa molecular weight cut-off concentrator (Amicon) and further purified by size exclusion chromatography on a Superdex S200 column (GE Healthcare) pre-equilibrated with buffer SEC (200 mM NaCl, 20 mM HEPES pH 7, 0.04% DDM).

A MicroCal ITC200 system (GE Healthcare) was used for ITC experiments. All titrations were performed at 6°C in buffer SEC (200 mM NaCl, HEPES pH 7, 0.04% DDM). The syringe was filled with 1 mM AlCl₃ and the titration was initiated by the sequential injection of 2 μl aliquots into the cell filled with SiNRAT or mutants at a concentration of 30 μM. Data were analyzed with the Origin ITC analysis package and WT protein data were fitted using the integrated two sides model. Each experiment was repeated with independently purified protein samples with similar results. Experiments with buffer not containing any protein were performed as controls.

SiNRAT and its mutants were reconstituted into detergent destabilized liposomes (Geertsma et al., 2008). The lipid mixture consisting of POPE, POPG, and EggPC (Avanti Polar Lipids) at a weight ratio of 3:1:1 was washed with diethylether and dried under a nitrogen stream and by exsiccation overnight. The dried lipids were resuspended into 100 mM KCl and 20 mM HEPES pH 7. After three freeze-thaw cycles, the lipids were extruded through a 400 nm polycarbonate filter (Avestin, LiposoFast-BAsic) and aliquoted into samples at a concentration of 45 mg/ml. The extruded lipids were destabilized by addition of Triton X-100 and the protein was reconstituted at a protein to lipid ratio of 1:50. After several incubation steps with biobeads SM-2 (Bio-Rad), the proteoliposomes were harvested the next day, resuspended in 100 mM KCl and 20 mM HEPES pH 7, and stored at –80°C.

The proteoliposomes were incubated with buffer IN (100 mM KCl, 20 mM HEPES pH 7, containing either 250 μM Calcein (Invitrogen), 100 μM Fura-2 (Thermo Fisher Scientific), or 400 μM MagGreen (Thermo Fisher Scientific) depending on the assayed ion). After three cycles of freeze-thawing, the liposomes were extruded through a 400 nm filter and the proteoliposomes were centrifuged and washed three times with buffer IN not containing any fluorophores. The proteoliposomes were subsequently diluted to 0.025 mg/ml in buffer OUT (100 mM NaCl, 20 mM HEPES pH 7) and aliquoted in 100 μl batches into a 96-well plate (Thermo Fisher Scientific) and the fluorescence was measured every 4 s using a Tecan Infinite M1000 fluorimeter. An inside negative membrane potential was generated by addition of valinomycin to a final concentration of 100 nM. After 20 cycles of incubation, the substrate was added at different concentrations. The transport reaction was terminated by addition of calcimycin or ionomycin at a final concentration of 100 nM. The initial rate of transport was calculated by linear regression within the initial 100 s of transport after addition of the substrate. The calculated values were fitted to a Michaelis-Menten equation. Fura-2 was used to assay the transport of Ca²⁺. For detection, the ratio between calcium bound Fura-2 (λ_ex = 340 nm; λ_em = 510 nm) and unbound Fura-2 (λ_ex = 380 nm; λ_em = 510 nm) was monitored during kinetic measurements. Calcein was used to assay the transport of Mn²⁺ (λ_ex = 492 nm; λ_em = 518 nm) and Magnesium Green was to assay the transport of Mg²⁺ (λ_ex = 506 nm; λ_em = 531 nm).

For assaying coupled proton transport, the proteoliposomes were mixed with buffer 1 (100 mM KCl, 5 mM HEPES pH 7, 50 μM ACMA). After clarification by sonication, the proteoliposomes were diluted to a concentration of 0.15 mg/ml in buffer 2 (100 mM NaCl, 5 mM HEPES pH 7) and aliquoted into batches of 100 μl into a flat black 96-well plate (Thermo Fisher Scientific). The fluorescence was measured every 4 s using a Tecan Infinite M1000 fluorimeter (λ_ex = 412 nm; λ_em = 482 nm). The transport reaction was initiated after addition of the substrate and valinomycin at a final concentration of 100 nM. The reaction was terminated by addition of CCCP at 100 nM. For experiments involving Ga³⁺, GaI₃ was used. For compensation, NaI was added to reach a final I-concentration of 45 µM, which did not exert any effect on the measurement.

SiNRAT-specific nanobodies were generated using a method described previously (Pardon et al., 2014). Briefly, an alpaca was immunized four times with 200 µg of detergent-solubilized SiNRAT (in intervals of four weeks). Two days after the final injection, the peripheral blood lymphocytes were isolated, and the RNA total fraction was extracted and converted into cDNA by reverse transcription. The nanobody library was amplified and cloned into the phage display pDX vector (Invitrogen). After two rounds of phage display, ELISA assays were performed on the periplasmic extract of 198 individual clones. A single nanobody was identified. For all steps of phage display and ELISA, SiNRAT was chemically biotinylated using an EZ-Link NHS-PEG4 Biotinylation Kit (Thermo Fisher Scientific) and bound to neutravidin-coated plates (Fairhead and Howarth, 2015) . The biotinylation was confirmed by total mass spectrometry analysis.

Nb1 was cloned into the pBXNPHM3 vector (Invitrogen) using the FX cloning technique. The expression construct contained the nanobody with a fusion of a pelB sequence, a 10-His tag, a maltose binding protein (MBP) and an HRV 3C protease site to its N-terminus. The vector was transformed into E. coli MC1061 cells and a preculture was grown in TB-Gly medium overnight using ampicillin (100 μg/ml) as a selection marker in all expression cultures. The preculture was used for inoculation of TB-Gly medium at a 1:100 volume ratio. Cells were grown to an OD₆₀₀ of 0.7–0.9 and the expression was induced by the addition of arabinose to a final concentration of 0.02% (wt/vol) for 4 hr at 37°C. Cells were harvested by centrifugation for 20 min at 5000 × g, resuspended in buffer A (200 mM KCl, 50 mM KPi pH 7.5, 10% glycerol, 15 mM imidazole pH 7.5, 2 mM MgCl₂, 1 μM leupeptin, 1 μM pepstatin, 0.3 μM aprotinin, 1 mM benzamidine, 1 mM PMSF) and lysed using a high-pressure cell lyser (Maximator HPL6). The purification was carried out at 4°C. The lysate was cleared by centrifugation at 200,000 × g for 30 min. The supernatant was loaded onto NiNTA resin and incubated for at least 1 hr under gentle agitation. The resin was washed with 20 CV of buffer B (200 mM KCl, 20 mM Tris-HCl pH 8, 10% glycerol, 50 mM imidazole, pH 7.5). The protein was eluted with buffer C (200 mM KCl, 20 mM Tris-HCl pH 8, 10% glycerol, 300 mM imidazole). The tag was cleaved by addition of HRV 3C protease at a 3:1 molar ratio and dialyzed against buffer D (200 mM KCl, 20 mM Tris-HCl pH 8, 10% glycerol) for at least 2 hr at 4°C. After application of NiNTA resin to remove the tag, the sample was concentrated by centrifugation using a 10 kDa molecular weight cut-off concentrator (Amicon) and further purified by size exclusion chromatography using a Superdex S200 column (GE Healthcare) pre-equilibrated with buffer E (200 mM KCl, 20 mM Tris-HCl pH 8). The peak fractions were pooled, concentrated, and flash-frozen and stored at –20°C for subsequent experiments.

For structure determination, the SiNRAT-Nb1 complex was reconstituted into amphipols. To ensure the stability of the proteins during this process, purified SiNRAT and Nb1 were mixed at a 1:2.5 molar ratio for 30 min prior to amphipol reconstitution. The SiNRAT-Nb1 complex and amphipol PMAL-C8 were mixed at a mass ratio of 1:5 and gently stirred overnight at 4°C. DDM was removed by the addition of a 100-fold excess (wt/wt) of biobeads SM-2 (Bio-Rad) compared to protein for 4 hr at 4°C under gentle agitation. The biobeads were removed by filtration and the protein-amphipol mixture was concentrated before purification by size exclusion chromatography using a Superdex S200 column (GE Healthcare) pre-equilibrated with buffer SEC2 (200 mM NaCl, 20 mM HEPES pH 7). The protein peak fractions were pooled and concentrated to 2.7 mg/ml for cryo-EM grid preparation.

For sample preparation for cryo-EM, 2.5 μl of the complex were applied to glow-discharged holey carbon grids (Quantifoil R1.2/1.3 Au 200 mesh). Samples were blotted for 2–4 s at 4°C and 100% humidity. The grids were frozen in liquid propane-ethane mix using a Vitrobot Mark IV (Thermo Fisher Scientific). All datasets were collected on a 300 kV Titan Krios (Thermo Fisher Scientific) with a 100 μm objective aperture and using a post-column BioQuantum energy filter with a 20 eV slit and a K3 direct electron detector (Gatan) in super-resolution mode. All datasets were recorded automatically using EPU2.9 (Thermo Fisher) with a defocus ranging from –1 to –2.4 μm, a magnification of 130,000× corresponding to a pixel size of 0.651 Å per pixel (0.3255 Å in super-resolution mode) and an exposure of 1.01 s (36 frames). Two datasets were collected for SiNRAT-Nb1 with respective total doses of 69.739 and 69.63 e^–/Ų.

Datasets were processed using Cryosparc v3.2.0 (Punjani et al., 2017) following the same processing pipeline (Figure 2—figure supplement 2). All movies were subjected to motion correction using patch motion correction with a Fourier crop factor of 2 (pixel size of 0.651 Å/pix). After patch CTF estimation, high-quality micrographs were identified based on relative ice thickness, CTF resolution, and total full frame motion and micrographs not meeting the specified criteria were rejected. For the SiNRAT-Nb1 complex, 2D classes generated from the final map of rXkr9 in complex with Sb1^rXkr9 (Straub et al., 2021) were used for template-driven particle picking on the whole dataset.

The 2D classes generated from particles selected from the entire dataset were extracted using a box size of 360 pix and down-sampled to 180 pixels (pixel size of 1.32 Å/pix). These particles were used for generation of four ab initio classes. Promising ab initio models were selected based on visual inspection and subjected to heterogenous refinement using one of the selected models as ‘template’ and an obviously bad model as decoy model. After several rounds of heterogenous refinement, the selected particles and models were subjected to non-uniform refinement (input model filtering to 8 Å) followed by local CTF refinement and another round of non-uniform refinement. Finally, the maps were sharpened using the sharpening tool from the Cryosparc package. The quality of the map was evaluated validated using 3DFSC (Tan et al., 2017) for FSC validation and local resolution estimation.

Model building was performed in Coot (Emsley and Cowtan, 2004). Initially, the structure of ScaDMT (PDB 5M94) was rigidly fitted into the density and served as template for map interpretation. The quality of the map allowed for the unambiguous assignment of residues 54–497. The structure of Nb16 (PDB 5M94) was used to build Nb1. The model was iteratively improved by real space refinement in PHENIX (Afonine et al., 2018) maintaining secondary structure constrains throughout. Figures were generated using ChimeraX (Pettersen et al., 2021) and Dino (http://www.dino3d.org). Surfaces were generated with MSMS (Sanner et al., 1996).

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

This research was supported by the Swiss National Science Foundation (SNF) through the National Center of Competence in Research TransCure. We thank Dr. Marta Sawicka for input in cryo-EM and help during initial sample characterization. Nanobodies were generated at the Nanobody Service Facility of UZH with the help of Dr. Sasa Stefanic. The assistance of Yvonne Neldner during library generation is acknowledged. The cryo-electron microscope and K3-camera were acquired with support of the Baugarten and Schwyzer-Winiker foundations and a Requip grant of the Swiss National Science Foundation. We thank Simona Sorrentino and the Center for Microscopy and Image Analysis (ZMB) of the University of Zurich for their support and access to the electron microscope. All members of the Dutzler lab are acknowledged for help in various stages of the project.

1. Received: December 21, 2022
2. Preprint posted: December 23, 2022 (view preprint)
3. Accepted: March 31, 2023
4. Version of Record published: April 19, 2023 (version 1)

© 2023, Ramanadane et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.