Evidence for the I-Shaped Dimers of a Plant Chloroplast F_OF₁-ATP Synthase in Response to Changes in Ionic Strength

1. Introduction

The ATP synthase is the object of intensive studies for at least half a century. This interest is defined by the fact that the enzyme provides cells with adenosine triphosphate (ATP), which is the key molecule for almost all biochemical reactions in living organisms, being the major source of energy for use and storage at the cellular level. F-type ATP synthases were found in mitochondria of eukaryotes, bacterial cellular membranes, and chloroplasts. ATP synthase couples transmembrane ion transport, established by mechanical rotation of a c-ring in a membrane (F_O) region, with conformational changes in ATP catalytic centers at α/β interfaces (F₁ region).

In mitochondria cristae, ATP synthases (mtF_OF₁) were found to form dimers of four different types (V-shape dimers of types I, II and IV, and U-shape dimer of a type III) [1]. In contrast, in bacteria, ATP synthases (bF_OF₁) are known to be monomers [2]. ATP synthases in thylakoid membranes (cF_OF₁) were shown to be primarily monomeric with approx. 12% fraction of dimers and 3% of higher oligomers [3]. According to what is known, the contacts between two cF_OF₁ monomers are highly-likely non-specific, and such dimerization might inhibit ATP synthases, preventing ATP hydrolysis in the absence of photosynthesis.

A dimerization of cF_OF₁ might be induced by changing of ionic strength in plant chloroplast during the light/dark switches. The internal volume of a chloroplast is separated from a cytoplasm by a lipid membrane, which, together with the pumps and channels, makes it possible to create and maintain a difference between ion concentration inside and outside the chloroplast. This difference can vary significantly from species to species, and between some organisms it can reach more than ten times values [4]. It is known that under light, various ions are redistributed in plants. The phenomenon of light-induced uptake of hydrogen ions by isolated chloroplasts is well studied. It is accompanied by the redistribution of other ions, such as Na, K, Cl, and Mg, implying that the ionic strength of the solution contained in plant chloroplasts can change 1.5 times during the day [5,6].

SAXS is an excellent structural method to study biological macromolecules in solutions; i.e., it allows one to perform the studies under variety of conditions, including different ionic strengths. In our work, we report an evidence of the in vitro dimerization of cF_OF₁ in response to increasing ionic strength. The structural data were obtained by SAXS measurements of solubilized and purified samples of cF_OF₁ from spinach chloroplasts at different NaCl concentrations. SAXS data analysis resulted in a model of an I-shaped dimer of cF_OF₁, which in a mixture with monomeric cF_OF₁, provided the best fit of the experimental data. I-shaped cF_OF₁ dimer is formed by F₁/F₁ contacts, presumably via δ/δ subunit interaction, so that the F_O parts of the monomers are on the opposite sides of the dimer. We used a macromolecular docking to refine our model and improved the quality of the fit of SAXS data for each NaCl concentration thereby generating a high-resolution model of these dimers.

Therefore, we present here a new model of a dimer of F-ATP synthases from spinach chloroplasts, which has a specific F₁/F₁ interaction, presumably via δ-subunits. We speculate that changing of ionic strength in plant chloroplasts during the day (light/dark period of time) might trigger the formation of cF_OF₁ oligomers in chloroplasts and, in particular, I-shaped cF_OF₁ dimers. We hypothesize that chloroplast ATP synthases might also form oligomers in order to inhibit ATP hydrolysis at dark, and I-shaped dimers might possibly connect the neighbor lamellae and stabilize thylakoid stacks in chloroplasts.

The reported here model of the cF_OF₁ I-shaped dimer is unexpected, and raises questions whether such type of dimerization exists in nature. First, such dimer should be inserted in two membranes at a specific distance from each other. Second, it should change the rate of ATP synthesis due to steric hindrance of δ-subunits.

2. Results

2.1. Purification of the monomeric F_OF₁ ATP-synthase from thylakoid membranes

Isolation and purification of cF_OF₁ from spinach chloroplasts was conducted as described [7,8,9], and the final step of purification utilized treatment with a detergent 4-trans-(4-trans-Propylcyclohexyl)-cyclohexyl α-maltoside (t-PCCαM) followed by the anion exchange chromatography (AEX) against the NaCl concentration gradient (50 – 1000 mM) (Figure 1a). The AEX chromatogram showed a peak at ~300mM NaCl (Figure 1b), which corresponds to the whole cF_OF₁ protein complex according to SDS-PAGE (Figure 1c). Blue-Native PAGE was also used to check the completeness and assembly of cF_OF₁ (Figure S1).

The peak fractions of cF_OF₁ were measured by SAXS and showed a good consistensy (χ² = 1.15) with the monomeric model of chloroplast ATP-synthase (PDB ID: 6FKF [10]) with a detergent belt built by MEMPROT program [11] (Figure 1d). Despite the fact that for membrane proteins solubilized in detergent, including large protein complexes, the presence of the local maxima at 0.1 – 0.2 Å^-1 is common for their I(Q) SAXS profile [12,13], our experiments and literature data obtained for an A-type ATP-synthse (SASBDB ID: SASDKK4) [14] show that it is not a rule even for different ATP-synthases. Nevertheless, taking into account a detergent belt allows improving the χ² from 1.35 to 1.15 (Figure S3).

2.2. SAS studies of cF_OF₁ dimerization at different NaCl concentrations

When investigating the samples of purified cF_OF₁ by small-angle neutron scattering (SANS) at 93% D₂O, 300 mM NaCl the increase of the maximum size and a radius of gyration (Rg) for the samples were more than expected from a contrast variation technique (Figure S2). This effect might have pointed towards an increase of an average oligomeric state in the sample, i.e. oligomerization polidispersity [12]. We hypothesized that a change of ionic strength could have led to oligomerization of cF_OF₁. In order to verify this hypothesis, we measured samples of purified cF_OF₁ at a range of NaCl concentrations by small-angle X-ray scattering (SAXS) (Figure 2a).

It is worth to mention that all samples of cF_OF₁ were first purified at the same conditions and after AEX the buffer contained 30 mM HEPES (pH 8.0), 2 mM MgCl₂, 250 – 300 mM NaCl, 0.04 % (w/v), tPCC-α-M (4-trans-(4-trans-Propylcyclohexyl)-cyclohexyl α-maltoside). Then, the samples were dialyzed against different NaCl concentrations (150, 250, 300, 350, 450 mM) and after dialysis they were measured by SAXS.

Pair-distance distribution function P(r) for samples of cF_OF₁ at different NaCl concentrations (Figure 2b) showed the increase of maximum size of the object D_max (Figure 2c), radius of gyration Rg (Figure 2d), and Porod volume Vp (Figure 2e) with changing of ionic strength, reaching the maxima at 150 and 450 mM NaCl and the minimum at 250 – 300 mM NaCl. It implies the increase of oligomerization fraction in the samples with both a decrease and an increase of ionic strength.

The most certainly a dimerization of cF_OF₁ was observed for the least and the most salinity from investigated range (150 and 450 mM NaCl, respectively). The distance between two monomers of cF_OF₁ in the dimer can be estimated as following:

$$D_{m-m}=2\sqrt{\frac{1+\alpha }{2\alpha }\left({Rg}_{mix}^2-{Rg}_{mono}^2\right)},$$

(1)

where Rg_mono and Rg_mix are radii of gyration for monomeric ATP-synthase and average for monomer-dimer mixure, correspondingly, α is a volume fraction of dimers estimated as α = Vp_mix/Vp_mono. Considering Vp errors as 10% and Rg errors (see Table S1) we estimate D_m-m equals to 183 ± 30 Å and 225 ± 24 Å for 150 mM and 450 mM NaCl, correspondingly.

The discrepancy of the values D_m-m might indicate different dimerization mechanisms at low and high salinity. Nevertheless, such values of D_m-m (similar to the size of cF_OF₁) imply that the contacts between two monomers of cF_OF₁ are at the edge of these protein complexes. Thus, only two types of contacts are possible: F_O/F_O (presumably via c-ring) or F₁/F₁ (presumably via δ-subunit).

2.3. Models of an I-shaped cF_OF₁ dimer

Taking into account the fact that the oligomerization increases at changing the ionic strength and the possible types of contacts derived from the D_m-m value, we built two models of cF_OF₁ which satisfy these conditions. In the first case, we built a model comprising a mixture of a monomeric cF_OF₁ and dimers with F_O/F_O contacts, presumably via c-ring (Figure 3a). In the second case, a model comprises a mixture of a monomeric cF_OF₁ and dimers with F₁/F₁ contacts, presumably via δ-subunit (Figure 3b).

The models of monomeric and dimeric cF_OF₁ contain detergent belts obtained by a program MEMPROT for the monomeric AEX fraction (see Figure 1d). The models of cF_OF₁ dimers were constructed using HDOCK protein–protein docking web-server [15]. Two types of restraints have been used, providing F_O/F_O or F₁/F₁ contacts, as described in Macromolecular docking section in Materials and Methods.

An approximation of SAXS data with these models resulted in χ² = 2.36 in the case of F_O/F_O dimers and χ² = 1.20 for F₁/F₁ dimers with a fraction of dimeric cF_OF₁ 96 ± 3 % and 68 ± 2 % for F_O/F_O and F₁/F₁ dimers, respectively (Figure 3a, b). The approximation by the model containing F₁/F₁ dimers showed better χ² value and relative residues ΔI/σ at small Q, which are within 3 (Figure 3d), in contrast to the approximation by the model containing F_O/F_O dimers (Figure 3c). Therefore, the model comprising a mixture of a monomeric cF_OF₁ and dimers with F₁/F₁ contacts, presumably via δ-subunit, showed the best fit of the SAXS data.

In the case of the cF_OF₁ dimers with F_O/F_O contacts, D_m-m values for top 10 HDOCK predictions (see Macromolecular docking section in Materials and Methods for details) were in the range of 174 – 201 Å. The only model (Figure 3a), which has D_m-m ≥ 201 Å (a lower limit of the range 225 ± 24 Å) (eq. 1), resulted in χ² = 2.36 and a fraction of dimers ~96 % for fitting of SAXS data for purified cF_OF₁ samples at 450 mM NaCl.

In the case of the cF_OF₁ dimers with F₁/F₁ contacts, D_m-m values for top 10 HDOCK predictions (see Macromolecular docking section in Materials and Methods for the details) were in the range of D_m-m 150 – 210 Å. All of these models, including one shown in Figure 3b (Model 4 in Figure S4), in addition to contacts between δ-subunits, demonstrated contacts between other subunits, including α-, β-, b- and b’-subunit. Five out of ten models have D_m-m ≥ 201 Å and demonstrated χ² values in the range of 1.16 – 1.20 (see details in Figure S4) for fitting SAXS data for purified cF_OF₁ samples at 450 mM NaCl. Figure 3b shows one of these five models which corresponds to the best HDOCK confidence score (Figure S4). An additional validation of parameters of macromolecular interfaces estimated using the PDBePISA web-server [16] allowed us to asses this model as reasonable both in terms of HDOCK confidence score values and PDBePISA metric (see Figure S4 for the details).

2.4. A possible physiological role of I-shaped cF_OF₁ dimers

Concerning a physiological relevance, we hypothesize that I-shaped type of dimerization might stabilize the stacks of chloroplasts by possible connection of neighboring lamellae. It is known that a large number of ATP-synthases is located in stroma lamellae and grana membranes [3,17] (Figure 4a). High density of molecules in these membranes leads to formation of contacts between them and lateral dimerization (Figure 4b). At the same time, the distance between neighboring stroma lamellae is about 30 nm, which approximately fits the I-shaped cF_OF₁ dimer (Figure 4c, d, g). It might also indicate a possibility of another type of dimerization via F₁ part of cF_OF₁, which is schematically shown in Figure 4e, f, and might have already been observed in literature (Figure 4g).

3. Discussion

Studies of chloroplast ATP synthases already showed evidence of dimerization of cF_OF₁ from plant chloroplasts [3]. This dimerization is believed to be non-specific and is shown to be lateral, which means that two monomers of cF_OF₁ are interacting in one membrane. There is also evidence of cF_OF₁ dimers for single-cell green algae [18,19], however, the authors report only Blue Native PAGE and following SDS PAGE, therefore no structural data has been reported and the exact type of this interaction remains unclear.

Using SAXS studies of purified samples of cF_OF₁ from spinach chloroplasts we showed that the degree of oligomerization increases in response to changing of the NaCl concentration, both a decrease and an increase from the point between 250 and 300 mM NaCl. where the dimeric fraction is minimal. We expected to obtain the model of lateral interaction similar to that reported in literature [3], but this model did not satisfy the distance between the two cF_OF₁ monomers (~200 Å). Thus, we came to a model with edge-to-edge interaction of two cF_OF₁, having an I-shape if visualizing a detergent belt at the membrane F_O part of each monomer. Surprisingly, the model with F_O/F_O’ interaction did not fit well the experimental SAXS data (χ² = 2.3), although satisfied the distance between the two cF_OF₁ monomers, in contrast to the model with F₁/F₁’ interaction.

We showed the first structural evidence of the I-shaped type of dimerization of chloroplast F-ATP synthases from plants. Our model suggests F₁/F₁’ contacts between the cF_OF₁ monomers as it is shown by SAXS data approximation (χ² = 1.24) and it satisfies the value of the distance between the centers of masses of monomers $D_{m-m}$ (eq. 1), which is about 200 Å. This means that the contacts between monomers of cF_OF₁ should be at the edge of the F₁ part to satisfy the conditions of comparatively large distance which is about the size of the whole protein complex cF_OF₁. Taken these facts together we hypothesize that the F₁/F₁’ contacts between cF_OF₁ monomers presumably might be via δ-subunit, which is at the top of the catalytic head (α₃β₃) of the enzyme.

In order to check the possibility of the F₁/F₁ interaction we performed a macromolecular docking using HDOCK protein–protein docking web-server [15]. Considering δ-subunit as a key subunit for cF_OF₁ dimer formation (see Macromolecular docking section in Materials and Methods for the details) we obtained models of a dimer, which, in addition to contacts between δ-subunit, showed contacts between other subunits, including α-, β-, b- and b’-subunit. The resulting models can be assessed as reasonable both in terms of HDOCK confidence score values and parameters of macromolecular interfaces estimated using the PDBePISA web-server [16] (see Figure S4).

To understand which region(s) of a δ-subunit can potentially play a role in the I-shaped dimer formation, we evaluated intrinsic disorder propensity of this protein, since disordered regions are known to often contribute to the protein-protein interactions, being capable of undergoing binding-induced folding [20,21,22,23]. Furthermore, even when a crystal structure of a query protein is solved, one often can find noticeable levels of intrinsic disorder, since very significant fraction of proteins in PDB contains regions with missing electron density, which are potentially intrinsically disordered [24,25]. As we already indicated earlier [1], based on the analysis of the reported crystal structure of cF_OF₁ from the spinach chloroplasts (see PDB ID: 6FKF), each subunit of this protein contains regions of missing electron density, including the chain δ with residues 1-70 and 250-257. This indicates that very significant parts of each cF_OF₁ subunit are expected to be disordered even within the assembled complex [1]. The idea of the presence of significant levels of disorder in the cF_OF₁ δ-subunit is further supported by Figure 5a showing a model 3D-structure generated for the full-length δ-subunit by AlphaFold [26,27], which is the most accurate AI-based platform for the protein structure prediction [28] and Figure 5b representing the disorder profile of this protein generated based on the outputs of six commonly used disorder predictors from the PONDR family, such as PONDR^® VLXT [29], PONDR^® VL3 [30], PONDR^® VLS2 [31], and PONDR^® FIT [32], as well as IUPred2 (Short) and IUPred2 (Long) [33]. Figure 5 clearly shows that the long N-terminal region of the cF_OF₁ δ-subunit is expected to be highly disordered. Utilization of the flDPnn webserver that predicts disorder, disorder-based functions and disordered linkers [34], revealed that residues 1-3, 22-26, and 59-66 of the cF_OF₁ δ-subunit might be involved in the disorder-dependent protein-protein interactions.

We should notice that it is a big challenge to show directly the I-shaped structural organization of cF_OF₁ because of weak non-specific interactions between the cF_OF₁ monomers. For example, a recent study demonstrated a cryo-EM structure of the monomer of ATP synthase from spinach chloroplasts reconstituted into lipid nanodiscs [10]. However, the procedure of grid preparation could expose the sample to harsh conditions, which might break the F₁/F₁’ contacts.

Another way of obtaining high-resolution structural information is protein crystallography. However, crystallization of ATP synthase as well as large membrane protein complexes is the key challenges in structural biology nowadays [1]. A crystallization by using lipid in meso phases (in meso method), e.g., lipid cubic phases (LCP), provides the crystals of membrane proteins in close to native conditions [36]. However, due to limitations of the diameters of LCP water channels the size of a water-exposed part of the protein complex is limited. Potentially perspective can be in bicelles crystallization [37,38], which was successfully applied for several membrane proteins with a large water-soluble part [39,40,41]. However, at the moment of this writing there is no reported successful cases of in bicelles crystallization of an ATP synthase.

ATP synthase crystallization by using vapor diffusion (VD) method, which allows avoiding the protein size limitations, has another problem connected with stabilization of membrane proteins in solution and fast kinetics of crystallization processes. Thus, crystal structures of almost full ATP synthase complexes were obtained only in several cases during the whole period of ATP synthase studies [42,43,44,45,46,47], which indicates a big challenge of obtaining crystals of ATP synthase. It is worth pointing out that even if a crystal of cF_OF₁ is grown it can also rearrange contacts between cF_OF₁ monomers.

In order to directly observe I-shaped dimerization we used SAXS method, which allowed us to obtain structural information in close to native conditions, allowing observing non-specifically connected cF_OF₁ monomers and obtaining the information about their supramolecular arrangement. SAXS does not require crystals or low temperatures, although in case of a mixture of monomers and oligomers SAXS data should be carefully analyzed [12,48].

In literature, there are debates about the possible physiological roles of dimers and higher oligomers of cF_OF₁ [3,49]. Some papers show evidence of the presence of dimers of cF_OF₁ in microorganisms [18,19,50]. Other papers claimed that these dimers might be only aggregates without specific structural arrangement or functional role [49,51]. Commonly, inhibition of ATP hydrolysis in chloroplasts occurs via a redox switch inserted into a γ-subunit of cF_OF₁ [10].

We hypothesize that the I-shaped type of dimerization might stabilize the stacks of thylakoids by possible connection of neighbor lamellae. This might be also connected with an increase of ionic strength in plant chloroplasts at dark [5,6]. Probably, formation of I-shaped dimers might establish an indirect interaction between lamellae (especially close to grana) and help to stabilize thylakoid stacks. Such dimers might be observed in more or less native conditions by cryo-electron tomography (Figure 4). However, reported structural studies of chloroplasts were made during the light phase [3,52], leaving a possibility to find native I-shaped dimers of cF_OF₁ in similar experiments at dark.

If an I-shaped dimer of cF_OF₁ exists in nature, we hypothesize that it might also represent a novel mechanism of a subtle regulation of ATP synthesis in plant chloroplasts and preventing ATP hydrolysis at dark periods. Increasing ionic strength in chloroplasts in the absence of light might trigger the formation of cF_OF₁ I-shaped dimers as well as pairs of cF_OF₁ observed in literature [3].

4. Materials and Methods

4.1. cF_OF₁ isolation and purification

The protein complex cF_OF₁ was isolated and purified from spinach chloroplasts as described [7,8] with minor modifications. Briefly, thylakoid membranes were solubilized in a buffer containing 60 mM n-Octyl-β-D-glucopyranoside and 25 mM Sodium Cholate v/v = 1/1, ultracentrifuged at 200 000 g for 1 h and a pellet was discarded. Then the solution underwent ammonium sulfate (AS) precipitation in the range of 1.2 – 1.8 M of AS. The precipitate at 1.8 M AS was either resuspended in a buffer containing ~2 mM tPCC-α-M, ultracentrifuged at 45 000 g and a pellet was discarded, a solution after 0.45 μm filtering was uploaded onto a column with POROS™ 20 HQ Strong Anion Exchange Resin and anion exchange chromatography was performed in a gradient of NaCl (50 – 1000 mM) in a buffer: 30 mM HEPES (pH 8.0), 2 mM MgCl₂, 0.04% (w/v) tPCC-α-M; or alternatively precipitate at 1.8 M AS was resuspended in a buffer containing 16 mM N-dodecyl β-D-maltoside (DDM), ultracentrifuged in a sucrose gradient (15 – 50 % w/v) at 200 000 g for 16 h, and the fraction between 29 and 36 % w/v of sucrose was desalted and underwent red-120 dye-ligand chromatography. Both methods provided ATP synthase applicable for further structural studies.

4.2. Blue Native Polyacrylamide Gel Electrophoresis

BN-PAGE was performed as described [53,54]. Briefly, the samples of cF_OF₁ were analysed in linear gradient polyacrylamide gel (from 4 to 14% (w/v) BN-PAGE, the separation gel was overlayed with an 3% (w/v) sample gel) (according 10.1080/13813455.2016.1218518, 10.1016/0003-2697(91)90094-A). An electrophoresis apparatus (SE 400, Cytiva) was used with cathode buffer contained 0.002% coomassie brilliant blue G-250 (Bio-Rad), phoresis duration was 16h. All native electrophoresis runs were performed at 100V and at 4°C. Hight molecular weight marker (HMW, #17044501, Cytiva) were used with the 10 ul load on a line.

4.3. Small-angle scattering measurements

SANS experiments were done on the YuMO spectrometer (IBR-2, Dubna, Russia) [55] with two-detector system [56,57]. For SANS measurements, two samples of cF_OF₁ were prepared. The first sample was obtained by anion-exchange chromatography (АЕХ). The second sample was obtained by dialysis of AEX-purified protein in 93% D₂O-buffer with 300 mM NaCl, 30 mM HEPES (pH ~8.0 [58]), 2 mM MgCl₂, 0.04 % (w/v) tPCC-α-M. Each sample (V = 400 uL) was poured into a Helma quartz cuvette (path length 1 mm) and placed to the temperature-controlled sample chamber [57] for further SANS measurements. Total exposure time for each sample was 80 min.

SAXS measurements were performed on the instrument Rigaku MicroMax-007 HF (MIPT, Dolgoprudny, Russia), which was used and described previously [38,59]. SAXS data were obtained for six samples of ATP synthase. First sample with protein concentration of ~5 mg/ml was obtained by purification using anion-exchange chromatography (see Figure 1). Other five samples were obtained by overnight dialysis in buffers containing 30 mM HEPES, 2 mM MgCl₂, 0.04% (w/v) tPCC-α-M, and different NaCl concentrations: 150, 250, 300, 350, and 450 mM. Each sample (V = 30 uL) was poured into a glass capillary, which was sealed by gas-burner or wax and placed into a vacuum chamber at a distance of 2.0 m from a multiwire gas-filled detector Rigaku ASM DTR Triton 200. All measurements were done at room temperature. A thermal stability of cF_OF₁ at room temperature was checked by Dynamic Scanning Fluorimetry (see the section Dynamic Scanning Fluorimetry). See Table S1 for other details of SAXS measurements and data treatment.

4.4. Small-angle scattering data treatment

In the case of SAXS data, first, SAXSGui software was used to convert 2D images of scattering intensity vs transmitted momentum in reciprocal space Q = 4𝜋 sin(𝜃)/𝜆 (where 𝜆 = 1.5405 Å (K_α of Cu), 2𝜃 – a scattering angle) to 1D-profiles I(Q) by using a radially integration. For Q-calibration a powder of silver behenate (AgBh) was used (Q = 0.1076 Å^-1, d = 58.38 Å) [60]. In the case of SANS data, raw data were converted to one-dimensional data set I(Q) with a program SAS (version 5.1.5) [61]. Then, in both cases a regularized model fit of I(Q) 1D-profiles was used for calculation of a pair-distance distribution function P(R) by a program GNOM from ATSAS program package [62,63]. Visualizing of high-resolution models of cF_OF₁ dimers was done by PyMOL software [64]. MEMPROT software [11] was used to fit an experimental SAXS data for AEX-purified cF_OF₁ using a high-resolution model of cF_OF₁ from spinach chloroplasts (PDB ID: 6FKF) with a pseudo-atomic model of the detergent belt surrounding its transmembrane part. For proper orienting of the cF_OF₁ model before running MEMPROT (place the center of the transmembrane part at the origin (zero) and set the normal vector to the membrane plane along z-axis) we used a PPM web server [65]. A program OLIGOMER [66] from ATSAS was used to fit experimental SAXS data from a two-component mixture of monomers and dimers of cF_OF₁ and to obtain the volume fractions of each component in the mixture. A program CRYSOL [67] (command line mode) from ATSAS was used for evaluating the solution scattering from macromolecules in order to fit experimental SAXS data and/or prepare a set of form-factors for subsequent run of OLIGOMER program. See Table S1 for other details of SAXS measurements and data treatment.

4.5. Macromolecular docking

For validation of possible F₁/F₁ interaction of cF_OF₁ monomers we performed a macromolecular docking using HDOCK protein–protein docking web-server [68]. Considering δ-subunit as a key subunit for dimer formation observed by SAXS, we used the following residue distance restraints: 1-257:d 1-257:d 50. Here, letter “d” corresponds to δ-subunit chain, “1-257” are numbers that cover all residues presented in δ-subunit, and a number “50” corresponds to condition that residues 1-257 of a chain “d” on the receptor and on the ligand will be within 50 Å. Screening of the distance value in the range of 10 – 150 with a step = 10 shows the same docking results. Top 10 obtained HDOCK models, in addition to contacts between δ-subunit, demonstrated contacts between other subunits, including α, β, b and b’, and shown the values of the HDOCK confidence score in the range 0.39 – 0.52 (Figure S4). Accordingly to the description of HDOCK it is considered that two molecules would be possible to bind, when the confidence score is between 0.5 and 0.7. However, taking into account the comment of the authors: “the confidence score here should be used carefully due to its empirical nature”, the lower value of 0.5 is empirical and models with scores slightly below 0.5 can also be considered as possible and reasonble. Additionally, we checked the parameters of macromolecular dimerization interfaces using the PDBePISA web-server [16]. Results are shown in Figure S4.

In order to check the possibility of F_O/F_O interaction we also performed a macromolecular docking using HDOCK web-server. Considering c-subunit as a key subunit for dimer formation observed by SAXS, we used the residue distance restraints correspoding to 10 Å distance between residue 3 in c-subunits on the receptor and on the ligand (3:g 3:q 10). The only model that satisfies the condition D_m-m ≥ 201 Å (eq. 1), has a high value of the HDOCK confidence score = 0.78 (Docking Score = -212.65), which corresponds to a high probability of the formation of such a contact. However, dimeriation via F_O/F_O interaction was not confirmed using SAXS data.

4.6. Dynamic Scanning Fluorimetry

We used Prometheus Panta from NanoTemper Technologies to check the thermostability of the protein complex cF_OF₁. Experiments were carried out in standard capillaries Prometheus NT.48 (Cat.# PR-C002). A temperature scan was conducted with 4 ℃ / min slope for temperature range 15 – 75 ℃, an excitation LED power was 70 %. Thermal stability parameters (T_onset, T_m) were calculated by Panta.Analysis software (NanoTemper Technologies). Folding state transition was monitored by the ratio of fluorescence intensity at 330 nm and 350 nm as a function of temperature, where T_onset (onset temperature of thermal unfolding) was 51.6 °C and T_m (inflection temperature of thermal unfolding) was 66.9 °C (Figure S5). Thus, at 20°C a stability of the protein was not disrupted.