Protein production and purification

The gene encoding pea OEP21 (GeneBank: AJ009987.1) was inserted into a modified pET21a vector (AmpR, Merck) by PCR-based restriction-free cloning, resulting in OEP21 harboring a non-cleavable C-terminal 10×His tag. For protein production, Escherichia coli BL21(DE3) were transformed with the described plasmid and grown to an optical density at 600 nm (OD600) of 0.6–0.8 at 37 °C. At this point, protein production was induced by the addition of 1 mM IPTG, and cells were shaken for another 4–5 h at 37 °C, collected by centrifugation (6,000g, 15 min, 4 °C) and stored at −80 °C. For lysis, cells were resuspended and homogenized in buffer A (50 mM Tris pH 8.0, 100 mM NaCl, 1 mM EDTA, 10 mM BME) + 1% Triton X-100, 0.2 mg ml–1 lysozyme (Sigma Aldrich) and one protease inhibitor tablet (Roche, cOmplete). Cells were sonicated (10 min, 30% amplitude, 1-s pulse, 2-s pause) and DNA was digested simultaneously by the addition of 1 U ml–1 DNase I (Roche) and 5 mM MgCl2. After centrifugation (38,769g, 20 min, 4 °C), inclusion bodies were washed with buffer A + 1% Triton X-100 and were centrifuged again. Inclusion bodies were finally dissolved in buffer B (6 M guanidinium chloride, 50 mM Tris pH 8.0, 100 mM NaCl, 10 mM imidazole and 5 mM BME), homogenized, centrifuged and filtered (0.45 μm) before further purification by Ni-NTA affinity chromatography. The supernatant was incubated for 1 h with the Ni-NTA resin (GE Healthcare), washed with 10 column volumes of buffer B and eluted with 10 column volumes of buffer B with 500 mM imidazole. The elution fraction was dialyzed overnight against 20 mM Tris pH 8.0, 50 mM NaCl, 1 mM EDTA and 2.5 mM BME in a 3.5-kDa molecular weight cutoff (MWCO) dialysis tube. Precipitated OEP21-10×His was dissolved in 6 M guanidinium chloride, 50 mM Na-phosphate pH 6.0, 100 mM NaCl, 5 mM EDTA and 10 mM DTT at a concentration of 5 mg ml–1. Protein refolding was performed by rapid dropwise dilution into a tenfold excess volume of refolding buffer consisting of 20 mM Na-phosphate pH 6.0, 50 mM NaCl, 1 mM EDTA, 3 mM DTT, 10% (v/v) glycerol and 0.5 % (m/v) LDAO, with moderate stirring at 4 °C for 2–4 h. To remove residual guanidinium chloride, dialysis (3.5-kDa MWCO) was done overnight in 20 mM Na-phosphate pH 6.0, 50 mM NaCl, 1 mM EDTA, 2.5 mM BME. Final purification was done by SEC (ÄKTA Pure system) using a Superdex 200 10/300 column (GE Healthcare) in 20 mM HEPES-KOH pH 7.5, 50 mM KCl, 0.5 mM EDTA, 5 mM DTT and 0.1% LDAO; 5 mM DTT was added to the sample before it was applied to the column. Protein samples were flash frozen and stored at −80 °C until further use. For buffer exchange, dialysis of the protein samples (6–8 kDa MWCO) in the corresponding buffer was performed.

Isotope-labeled protein (2H,15N; 2H,15N,13C; or selectively ILVAFY-labeled) was produced using published protocols20. In brief, for stereospecific Leu, Val methyl labeling, 300 mg L–1 stereospecific LV precursor ethyl 2-hydroxy 2-13C-methyl 3-oxobutanoate (according to established protocols41,42) and 80 mg L–1 of the Ile precursor α-ketobutyrate43 together with 2.0 g L–1d4-succinate (Eurisotop or Sigma Aldrich) and 0.5 g L–1 3-[13CH3]-2-d-l-Alanine (Eurisotop), as well as 80 mg L–1 of both uniformly 15N labeled l-Phe and l-Tyr (Sigma Aldrich), were added to the bacterial culture 1 h before induction of protein production with 1 mM IPTG. The choice of the Leu/Val precursor resulted in 13CH3 labeling of the pro-S methyl group in both amino acids in an otherwise per-deuterated 12C background.

Blue Native PAGE analysis of OEP21 in chloroplasts

Chloroplasts were isolated as described in Stengel et al.44 and mixed with outer envelope vesicles in a ratio of 5:1. The mixture was centrifuged for 5 min at 6,000g at 4 °C, and the resulting pellet was resuspended in 750 mM aminocaproic acid, 50 mM Bis-Tris and 0.5 mM EDTA pH 7.0, supplemented with 1% DDM (dodecyl-β-d-maltoside, Roth). After a 10-min incubation on ice, the suspension was centrifuged for 10 min at 15,000g at 4 °C, and the supernatant was mixed with 10× loading dye (750 mM amino caproic acid, 5% Coomassie G-250). This was loaded onto a 7.5–15% acrylamide gel with 20 µg chlorophyll per lane. For subsequent western blotting, lanes were incubated in Towbin buffer (25 mM Tris pH 8.3, 192 mM glycine, 0.1% (w/v) SDS, 20% (v/v) methanol) with additional 0.9% (w/v) SDS and 50 mM DTT for 30 min before blotting.

Proteolytic digest of outer envelope membranes and immunoprecipitation of OEP21

Isolated OEMs21, equivalent to 20 µg total protein per sample, were centrifuged at 256,000g for 10 min at 4 °C. The pellet was resuspended in 20 mM tricine pH 8.0 and 0.5 mM CaCl2 and treated with 200 ng trypsin from bovine pancreas (T1426, Sigma Aldrich) for 90 s at room temperature (RT). The digestion was stopped by addition of 1× cOmplete protease inhibitor (Roche), and proteins were solubilized by adding 1 volume of 2× SDS Laemmli loading buffer. Samples were run on 10% or 15% SDS gels. Immunoblotting was performed o PVDF membranes in Towbin buffer (25 mM Tris pH 8.3, 192 mM glycine, 1% SDS, 20% methanol) in a wet transfer apparatus for 1 h at RT at 300 mA or overnight at 4 °C at 60 mA. The membranes were blocked for at least 30 min with 1% skim milk in 50 mM Tris pH 7.6, 150 mM NaCl and 0.05% Tween 20 (TBS-T), and then were incubated with specific antibodies in 1:1,000 dilution in TBS-T for 2 h at room temperature. Subsequently, membranes were washed 3 times in TBS-T, then incubated with HRP-coupled secondary antibody (goat-anti-rabbit, Sigma Aldrich) in blocking buffer for 1 h at RT, and then were washed 3 times in TBS-T. Proteins were detected by chemiluminescence after incubating the membranes in 100 mM Tris pH 8.5, 25 mM luminol, 4 mM coumaric acid and 0.2% H2O2 for 1 min. Immunoprecipitation of trypsin-treated outer envelope membranes was performed with 1 mg total protein, digested as described above. Membranes were solubilized in 1% SDS, diluted 1:10 with 50 mM Tris pH 7.6, 150 mM NaCl, 1× cOmplete and incubated with OEP21 antiserum coupled to ProteinA Sepharose beads (GE Healthcare) for 1 h at RT. Beads were washed twice with 50 mM Tris pH 7.6, 150 mM NaCl, 0.1% SDS, 1× cOmplete and bound proteins were eluted by boiling in Laemmli loading buffer. Proteins were separated on a 12.5% Tricine gel, blotted onto PVDF as described above and subsequently stained with Coomassie Brilliant Blue (G). The upper band corresponding to the digestion fragment identified in the immunoblot was cut out and subjected to Edman sequencing (TopLab).

Protein crosslinking

Protein crosslinking in detergent micelles was performed in 20 mM HEPES-KOH, pH 7.5, and 50 mM KCl, with a final OEP21 concentration of 20 µM in the presence of 0.1% of LDAO. A 50× molar excess of the amino-selective crosslinker BS3 (ThermoFisher Scientific, cat. no. 21580) was applied in the absence or presence of the respective metabolite at a 5 mM concentration for 30 min at room temperature, following 5 min of preincubation with the metabolites. The reaction was quenched by the addition of 50 mM Tris pH 7.5 for 15 min. For crosslinking in liposomes, liposomes were prepared as described in the next section in the absence or presence of 5 mM GAP or ATP using 10 μM OEP21. A 50× molar excess of BS3 was applied to liposomes for 30 min at RT. The reaction was quenched by addition of 50 mM Tris pH 7.5 for 15 min at RT. Finally, the samples were analyzed by SDS–PAGE. For crosslinking of cysteines by Cu2+, 20 µM of protein samples in 20 mM HEPES-KOH, pH 7.5, 50 mM KCl and 0.1% LDAO were incubated in the absence or presence of 1 mM CuSO4 for 15 min. Finally, the samples were analyzed by SDS–PAGE under non-reducing conditions. To detect the effect of cysteine oxidation on overall protein oligomerization, we applied Cu2+-crosslinking and then preincubated the samples in the absence or presence of 5 mM GAP or ATP for 5 min at RT. Finally, the BS3 crosslinking experiment was performed as described above, followed by analysis by SDS–PAGE.

Preparation of liposomes and metabolite transport assays

Soybean polar lipid extract (Avanti Polar Lipids, 541602C) was first dried under a stream of nitrogen gas and solubilized at a concentration of 10 mg ml–1 in 20 mM HEPES-KOH pH 7.5, 250 mM KCl and 1% LDAO using an ultrasonic bath. After addition of 2 mM ATP on ice, OEP21 was added into the lipid–ATP mix. For every reconstitution experiment, a control liposome sample was prepared in parallel using the same lipid mix but an identical volume of SEC buffer (20 mM HEPES-KOH pH 7.5, 50 mM KCl, 0.5 mM EDTA, 5 mM DTT and 0.1% LDAO) was added instead of the purified protein. After shaking the sample for 30 min at 10 °C, liposomes were formed by gradual removal of the detergent from the mixture using 3 rounds (2 × 100 mg ml–1, 1 × 200 mg ml–1) of Bio-Beads SM2 resin (Bio-Rad) for 1.5 h at 4 °C. Following 3 freeze–thaw cycles, liposomes were passed through a 0.2-µm filter 15 times in a mini extruder (Avanti Polar Lipids).

For transport assays, the molecule to be tested (2 mM GAP, 2 mM ATP (already contained for each reconstitution except for the BS3 crosslinking experiment in the presence of only GAP), 200 µM MANT-ATP, 2 mM NADH, 140 µM bromophenol blue, 100 µM magnesium green (Thermo Fisher Scientific, M3733), 2 mM NADPH, 1.5 mM FAD, 30 µM Atto532 or 2.5 mM vitamin B12) was added into liposomes just before the 3 freeze–thaw cycles and extrusion through a 0.2-µm membrane. Subsequently, the liposome samples were applied to PD10 desalting columns equilibrated with 20 mM HEPES-NaOH pH 7.5 and 250 mM NaCl, and the liposome fractions were collected. The content of each molecule inside liposomes was quantified in the presence of 0.2% Triton X-100 to disintegrate the liposomes. The 846-Da monomeric ‘Atto532′ molecule was generated by incubating Atto532-maleimide (ATTO-TEC, AD532-41) with a twofold excess of β-mercaptoethanol, and the dimeric 1688 Da ‘2× Atto532′ molecule was generated by incubating DTT and Atto532-maleimide in a 1:2 molar ratio for at least 2 h at RT.

The GAP content was measured through NADH fluorescence (excitation, 340 nm; emission, 450 nm), which was generated by an enzyme-coupled NADH assay in the presence of 1 mM NAD+, 10 mM potassium phosphate pH 7.5, 0.02 mg ml–1 GAPDH. Excitation 340 nm and emission 450 nm were also used for detection of NADH and NADPH. ATP and MANT-ATP content of the liposomes were measured using an Invitrogen ATP determination kit (Thermo Fisher Scientific). Magnesium green, FAD and Atto532 contents were measured by fluorescence detection at an excitation of 506 nm and an emission of 546 nm, an excitation of 450 nm and an emission of 520 nm and an excitation of 532 nm and an emission of 572 nm, respectively. Bromophenol blue and vitamin B12 contents were measured by absorbance at 590 nm and 363 nm, respectively.

For the detection of GAP translocation in the presence of ATP or MgCl2, 5 mM GAP and 3 mM ATP (final 5 mM) and/or 10 mM MgCl2 were added just before the 3 freeze–thaw cycles, and the extrusion step. The PD10 SEC column runs were conducted in the presence or absence of 5 mM ATP and/or 10 mM MgCl2. For ΔΨ measurements, 1.5 µl of liposomes in 20 mM HEPES-KOH pH 7.5, 250 mM KCl were diluted in 1.5 ml of 20 mM HEPES-NaOH pH 7.5, 250 mM NaCl containing 0.5 µM of potentiometric fluorescent dye DiSC3(5). Time course fluorescent measurements were done using spectrofluorometer FP-8300 (Jasco) at 622 nm excitation and 670 nm emission wavelengths. Then, 2 nM valinomycin and 250 mM KCl were added at the indicated time points for generation and dissipation of ΔΨ, respectively. Data were plotted with OriginPro (OriginLab).

Proteolytic cleavage of recombinant OEP21 in liposomes

Liposomes containing 10 µM OEP21 were prepared as described above in the absence of ATP. Trypsin was added to samples in the presence or in absence of 0.2% TX100, 5 mM GAP and/or 5 mM ATP. Sample aliquots were quenched by addition of 1 mM PMSF at the indicated time points and were analyzed by SDS–PAGE.

Circular dichroism spectroscopy

CD spectra and thermal transitions were recorded on a Jasco J-715 spectropolarimeter (Jasco Deutschland). Far-UV CD spectra were recorded at 20 °C from 190 to 260 nm at a scanning speed of 50 nm min–1 and 5 accumulations in a 1 mm path-length cuvette. Melting temperatures were obtained by monitoring the CD signal at 215 nm during continuous heating from 20 to 100 °C, with a heating rate of 1 °C min–1. Curve fitting to a custom Boltzmann equation45 and plotting was done with the software ProFit 7 (QuantumSoft). All protein samples had a concentration of 10 µM in 10 mM Na-phosphate pH 7.0, 0.5 mM DTT and 0.1% LDAO.

Isothermal titration calorimetry

ITC experiments were performed with a MicroCal PEAQ-ITC instrument (Malvern Panalytical) at 25 °C in 10 mM HEPES pH 7.0, 20 mM NaCl, 0.5 mM EDTA, 1 mM DTT, 50 mM LDAO. For titrations with the metabolites, the OEP21 concentration in the cell was 75 µM, whereas the concentration in the syringe was 1.5 mM for ATP, CTP, UTP, 2.5 mM for ADP and 5 mM for AMP. To determine the affinity of ATP, GAP or phosphate to magnesium, a buffer containing 10 mM HEPES pH 7.0 and 20 mM NaCl was used. Curve fitting and data analysis were done with the Malvern PEAQ-ITC software.

Fluorescence polarization

FP assays were performed on a spectrofluorometer FP-8300 (Jasco) equipped with a water bath cooling system MCM-100 (Jasco). Samples were put in a 1-cm path-length quartz cuvette, and the FP of the fluorescently labeled ATP, 2′-(or-3′)-O-(N-methylanthraniloyl) adenosine 5′-triphosphate (MANT-ATP, Thermo Scientific) was measured at 25 °C in 10 mM HEPES pH 7.0, 20 mM NaCl, 0.5 mM EDTA, 1 mM DTT and 0.1% LDAO. A buffer containing no EDTA was used when MgCl2 was present. Excitation was done at 356 nm and the emission was detected at 448 nm with a 5-nm bandwidth. To determine the affinity of MANT-ATP for OPE21 variants, increasing concentrations of the protein were titrated stepwise to 200 nM MANT-ATP. Depending on the binding affinity, the final protein concentration was between 500 nM and 800 nM. KD values were derived by fitting the data to a one-site binding model. Using the same workflow and adjusted final OEP21 concentrations, the affinities of OEP21-WT for MANT-ATP in presence of different concentrations of NaCl (0, 0.5, 1, 2 M) and MgCl2 (0, 1, 10, 50 mM) were determined. To determine IC50 values for the competitive binding of MANT-ATP and Na-phosphate, GAP or GAP + MgCl2 (2:1 concentration ratio), FP was measured of a preformed complex of 1 µM MANT-ATP and 1 µM OEP21-WT or OEP21ΔL5 in the presence of increasing concentrations of the competitor. In addition, the binding affinity for fluorescently labeled MANT-5′-guanylyl imidodiphosphate (GMPPNP, Thermo Scientific), was determined as described for MANT-ATP. All FP experiments were performed as triplicates.

Nuclear magnetic resonance structure determination

NMR experiments were done at 308 K on Bruker AvanceIII spectrometers operating at 800, 900 or 950 MHz proton frequency with cryogenic probes and were controlled with Topspin 4.0 (Bruker Biospin). For backbone resonance assignments, a set of TROSY-type 3D-experiments was recorded46 as well as a 3D-15N-edited-[1H,1H]-NOESY-TROSY (200 ms mixing time) with a 400 µM U-2H,13C,15N-labeled OEP21 sample in 20 mM Na-phosphate, pH 6.0, 50 mM NaCl, 0.5 mM EDTA, 5 mM DTT, 5 mM ATP, 300 mM per-deuterated d31-LDAO (FB Reagents). All 3D-NOESY (types HNH, HCH, CCH; 200 ms mixing time) experiments for structure determination were run in a non-uniformly sampled (NUS) manner with 15–20% sampling density with a uniformly 2H,15N-labeled OEP21 sample containing selective methyl group labels (Ile-δ1, Leu-δ2, Val-γ2, Ala-β) as well as incorporated 1H,15N-labeled aromatic amino acids (Phe and Tyr). The NUS sampling schedule was obtained by the Poisson-gap method47. For rapid spectra reconstruction, we employed iterative soft thresholding (IST)48. All NUS-3D spectra were processed with NMRpipe49. All other spectra were processed with Topspin3.5 (Bruker Biospin). Resonance assignment and NMR data analysis were done with NMRFAM-Sparky50. Chemical-shift-based backbone dihedral angle restraints were calculated with the program TALOS+ (ref. 51). Structure calculation and refinement were performed with Xplor-NIH52 using standard protocols. Structural statistics (Table 1) are reported for the ten lowest-total-energy structures. Ramachandran analysis of backbone angles was done with PROCHECK-NMR53 using the best-energy structure (most favored regions: 81.3%; additionally allowed regions: 12.9%; generously allowed regions: 4.5%; disallowed regions: 1.3%). Calculation of the electrostatic potential of the OEP21 pore interior was done in PyMol (Schrödinger) and visualized with the program HOLLOW54. The pore diameter was analyzed with the program CHEXVIS55. A structural model of the dimeric, disulfide-bridged (through Cys109) form of OEP21 was calculated with Xplor-NIH using standard scripts using non-crystallographic symmetry restraints. The resulting structural model was subjected to a MD simulation of 100 ns duration at 303 K in a DMPC/DMPG lipid bilayer with the program NAMD56 on an in-house CPU cluster.

Nuclear magnetic resonance titrations

Metabolite and detergent titrations were monitored by a series of 2D-[15N,1H]-TROSY experiments at 303 K with LDAO-solubilized 400 µM 2H,15N-labeled OEP21 in 20 mM HEPES pH 7.0, 50 mM NaCl, 0.5 mM EDTA, 5 mM DTT. Typically, 48 transients were recorded per increment, with 128 complex points in the indirect 15N dimension. Curve fitting of NMR binding isotherms was done with a full binding model accounting for the relatively high protein concentration (>KD) required for NMR experiments57. Chemical shift perturbations in the 1H and15N spectral dimensions were scaled using the distribution of nucleus-specific chemical shift changes in proteins58.

Paramagnetic relaxation enhancements with the spin-labeled fatty acid 16-doxyl-stearic acid (16-DSA) were monitored with 2D-[15N,1H]-TROSY experiments at 308 K using a recycle delay of 2.5 s. Two experiments were recorded with 2H,15N-labeled OEP21 in 20 mM Na-phosphate pH 6.0, 50 mM NaCl, 0.5 mM EDTA, 5 mM DTT, 150 mM LDAO with and without 2 mM 16-DSA. For visualization, peak intensity ratios (±2 mM 16-DSA) were plotted against the residue number.

Amide hydrogen exchange nuclear magnetic resonance experiments

Amide hydrogen exchange was quantified with CLEANEX NMR experiments59 using different mixing times (2–100 ms) at 303 K. Fitting of the built-up curves was done with a mono-exponential equation. In addition, hydrogen exchange was qualitatively monitored with 3D 15N-edited-[1H,1H]-NOESY-TROSY experiments with a mixing time of 40 ms. The occurrence of sequential amide-amide cross peaks as well as the hydrogen exchange with water was analyzed. These experiments were recorded with 400 µM 2H,15N-labeled OEP21 in the apo form at a high LDAO concentration (300 mM), as well as in presence of 2 mM ATP or 5 mM GAP in 20 mM Na-phosphate pH 6.0, 50 mM NaCl, 0.5 mM EDTA and 5 mM DTT.

Molecular dynamics simulations

OEP21 was embedded in a bilayer consisting of 202 DMPC molecules and solvated in an aqueous 0.15 M KCl solution, using the membrane builder of the CHARMM-GUI web-server60,61,62. The ligands GAP2− and ATP4− were placed outside the pore. All simulations (duration of 2–4 µs) were performed using the GPU accelerated CUDA version of PMEMD63, part of the AMBER18 package64 (see also Supplementary Methods). The unbiased simulations were carried out at a pressure of 1 bar. Simulations with an exterior electric field of 5.625 mV A–1 were performed without pressure coupling (NVT ensemble). The target temperature for all simulations was 303 K.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.