Engineering of IF₁‐susceptive bacterial F₁‐ATPase

Abstract

IF₁, an inhibitor protein of mitochondrial ATP synthase, suppresses ATP hydrolytic activity of F₁. One of the unique features of IF₁ is the selective inhibition in mitochondrial F₁ (MF₁); it inhibits catalysis of MF₁ but does not affect F₁ with bacterial origin despite high sequence homology between MF₁ and bacterial F₁. Here, we aimed to engineer thermophilic Bacillus F₁ (TF₁) to confer the susceptibility to IF₁ for elucidating the molecular mechanism of selective inhibition of IF₁. We first examined the IF₁‐susceptibility of hybrid F₁s, composed of each subunit originating from bovine MF₁ (bMF₁) or TF₁. It was clearly shown that only the hybrid with the β subunit of mitochondrial origin has the IF₁‐susceptibility. Based on structural analysis and sequence alignment of bMF₁ and TF₁, the five non‐conserved residues on the C‐terminus of the β subunit were identified as the candidate responsible for the IF₁‐susceptibility. These residues in TF₁ were substituted with the bMF₁ residues. The resultant mutant TF₁ showed evident IF₁‐susceptibility. Reversely, we examined the bMF₁ mutant with TF₁ residues at the corresponding sites, which showed significant suppression of IF₁‐susceptibility, confirming the critical role of these residues. We also tested additional three substitutions with bMF₁ residues in α and γ subunits that further enhanced the IF₁‐susceptibility, suggesting the additive role of these residues. We discuss the molecular mechanism by which IF₁ specifically recognizes F₁ with mitochondrial origin, based on the present result and the structure of F₁‐IF₁ complex. These findings would help the development of the inhibitors targeting bacterial F₁.

1. INTRODUCTION

F_oF₁‐ATP synthase (F_oF₁) is one of the most ubiquitous enzymes; it is found in the inner membrane of mitochondria, the thylakoid membrane of chloroplast, and the plasma membrane of bacteria. This enzyme catalyzes ATP synthesis reaction from ADP and inorganic phosphate (P_i), using the proton motive force (pmf) across the membranes (Stewart et al., 2014; Walker, ²⁰¹³). F_oF₁ consists of two structurally and functionally distinct rotary motor proteins, F_o and F₁. F_o, the membrane‐embedded domain, mediates proton translocation, driving the rotation of the oligomeric c‐ring against the stator complex (ab ₂). F₁ is the catalytic portion of F_oF₁ and an ATP‐driven motor in which the rotary shaft rotates against the surrounding stator upon catalysis. In cells, F_o and F₁ are connected via the complex of the rotor shafts and the peripheral stalk, which enables inter‐transmission of rotary torque between F_o and F₁. The rotary direction of the rotor complex is determined by the balance of total free energy of ATP hydrolysis and pmf across the membranes. When pmf is sufficiently large, F_o rotates the rotor complex, reversing the rotation of F₁ to induce ATP synthesis. Conversely, when pmf decreases, F₁ hydrolyzes ATP to rotate the rotor complex in reverse, enforcing F_o to pump protons (Boyer, 1997; Junge & Nelson, ²⁰¹⁵).

The minimal complex of bacterial F₁ as a rotary motor is the α₃β₃γ subcomplex in which the rotor γ subunit is accommodated in the central cavity of the stator α₃β₃ ring (Abrahams et al., 1994). The mammalian mitochondrial F₁ has the δ and the ε subunit that form the rotor shaft with the γ subunit. F₁ has three catalytic sites for ATP hydrolysis/synthesis, each of which resides on the αβ interface of the α₃β₃‐ring. Most of the catalytical residues are found in the β subunit. F₁ from bovine mitochondria (bMF₁) in the ground state (Bowler et al., 2007) showed that three β subunits exhibit distinct nucleotide occupancies and conformational states; two β subunits bind to nucleotides, one bound to ATP‐analog, (denoted as β_TP) and the other bound to ADP (β_DP). Both β_TP and β_DP adopt a closed conformation, inwardly swinging the C‐terminal domain. The third β subunit, termed β_E, has no bound nucleotide and takes an open conformation. The three β subunits operate catalysis in a well‐coordinated manner; the β subunits sequentially proceed the catalytic reactions, coupled with conformational transitions between open‐ and closed‐state, that induces the unidirectional rotation of the rotor (Kobayashi et al., 2020; Okazaki & Takada, ²⁰¹¹; Shimabukuro et al., ²⁰⁰³). The recent study found the intermediate conformational states (Sobti et al., 2021), revising the reaction scheme of F₁ (Noji & Ueno, 2022). Due to such a highly cooperative catalytic mechanism, F₁ is known to stop the overall catalysis when one of the β subunits is inhibited by an inhibitory compound or mutation (Amano et al., 1996; Gledhill & Walker, ²⁰⁰⁶).

Because F_oF₁ is principally obligated to function as ATP synthase, F_oF₁ possesses various mechanisms to suppress or inhibit the hydrolytic activity of F₁ (Feniouk & Yoshida, 2006; Mendoza‐Hoffmann et al., ²⁰²²; Walker, ¹⁹⁹⁴). The most universal suppression mechanism across species is ADP inhibited form that is observed as a transient pause during rotation in single‐molecule rotation assay (Hirono‐Hara et al., 2001; Kobayashi et al., ²⁰²⁰), although the physiological role of ADP inhibition is not well investigated. Mitochondrial F_oF₁ has an endogenous inhibitor protein that blocks ATP hydrolysis by F₁, termed ATPase inhibitory factor 1 (IF₁), which was discovered in 1963 by Pullman and Monroy (Pullman & Monroy, 1963). IF₁ shows highly conserved sequence across species, which allows for an interspecies inhibition in other mitochondrial ATP synthases; bovine IF₁ well suppresses ATPase activity of mitochondrial F₁ from yeast (Klein et al., 1977). The bovine IF₁ is 84 amino acid long. The N‐terminal region is responsible for the inhibition, binding to the αβ interface of F₁ to block conformational transition of the β subunit (Kobayashi et al., 2023), while the C‐terminal region of IF₁ is for a homodimer formation of IF₁ (Domínguez‐Ramírez et al., 2001). When the C‐terminal region (61–84) is deleted, IF₁ is always in the monomeric form retaining its full inhibitory capacity (van Raaij et al., ¹⁹⁹⁶). Due to the simplicity and the ease to handle, the monomeric IF₁ is often used for structural (Bason et al., 2014; Gledhill et al., ²⁰⁰⁷) and biochemical analyses (Bason et al., 2011; Kobayashi et al., ²⁰²¹), including the present study.

The crystal structure of the bMF₁‐IF₁ 1:1 complex provides insights into the atomic details of the interactions between bMF₁ and IF₁ (Gledhill et al., 2007; Figure 1a). The central long helix of residues 21–46 in IF₁ was bound on the C‐terminus domain of the β_DP subunit, seemingly inserted along the crevice of the αβ_DP interface. The central helix has also contact with the β_TP subunit at its N‐terminus end of the helix. The N‐terminus unstructured region (8–13) and the short helix (14–18), which were followed by a distinct kink and the central helix, were deeply inserted in F₁, having contacts with the γ and α_E subunit. Thus, the bMF₁‐IF₁ complex showed the intertwined binding mode, interacting with α_E, α_DP, β_DP, β_TP, and γ subunit. The crystal structure of the bMF₁‐IF₁ 1:3 complex was reported (Bason et al., 2014), in which β_TP and β_E also possessed bound IF₁. IF₁s bound on β_TP and β_E showed partially folded state in comparison with the β_DP‐bound IF₁, suggesting the progressive binding mechanism of IF₁ coupled with folding (Bason et al., 2014); IF₁ loosely associates with the most‐open αβ_E pair, and then it isomerizes to the final‐locked state in the αβ_DP site, coupled with γ rotation driven by ATP hydrolysis (Figure 1b). This progressive inhibition mechanism was in good agreement with biochemical experiments (Corvest et al., 2005; Corvest et al., ²⁰⁰⁷; Kobayashi et al., ²⁰²¹; Milgrom, ¹⁹⁸⁹).

FIGURE 1.

Schematic image of IF₁ inhibition mechanism. (A) IF₁ binding with each subunit in F₁. α_E, α_DP, β_TP, β_DP, and γ interact with the full‐folded IF₁ in the final locked state (PDB: 2v7q). The N‐ and C‐terminus of IF₁ are shown as ‘N’ and ‘C’, respectively. Details are described in the Tables S3 and S4. (B) Two‐step reaction mechanism of IF₁ inhibition: the initial association to β and the following isomerization process to the locked state (PDB: 4tt3). The β, γ, and IF₁ are shown in blue, yellow, and green, respectively.

One of the unique features of IF₁ is the apparent unidirectional inhibition. IF₁ completely blocks the catalysis of F₁ when the chemical equilibrium of ATP hydrolysis/synthesis is toward hydrolysis (De Gómez‐Puyou et al., 1980; Harris & Crofts, ¹⁹⁷⁸; Lippe et al., ¹⁹⁸⁸; Power et al., ¹⁹⁸³). When the chemical equilibrium is on ATP synthesis side with sufficiently high pmf, IF₁ does not interfere with the catalysis. Previously, we showed that this condition‐dependent manner of inhibition is due to the rotation‐direction‐dependent activation; bound IF₁ is ejected from F₁ when torque is applied to F_oF₁ in ATP synthesis direction (Kobayashi et al., 2023). Another remarkable feature of IF₁ is distinctive species‐specific inhibition. IF₁ inhibits catalysis of mitochondrial F₁ but does not interfere with the catalysis of bacterial F₁s (Gledhill & Walker, 2005; Suzuki et al., ²⁰¹⁴; Wu et al., ²⁰¹⁴). The structure of the bMF₁‐IF₁ 1:3 complex showed that, out of 22 residues of bMF₁ in contact with bound IF₁, 17 residues are on β subunit, while two residues on α subunits and three residues on γ subunit (Table S1; Bason et al., ²⁰¹⁴). Consistent with this contention, the F₁–IF₁ interaction has buried surface area of 2200–2400 ${\AA }^2$ in total, with the β_DP subunit contributing to nearly half of this interface, around 1100–1200 ${\AA }^2$. Considering the apparent contribution of the β subunit for IF₁ binding, it would be reasonable to assume that the species‐specificity of IF₁ inhibition primarily arises from the β subunit. On the other hand, the amino acid conservation of the 22 residues in contact with IF₁ between bMF₁ and TF₁ implies the contribution of the α and γ subunits for the species‐specificity of IF₁; the amino acid identity of the contacting residues is the highest for the β subunit (71%), while those are modest for the α subunit (50%) and the lowest for the γ subunit (33%) (Tables S1 and S2). Further, the structural comparison between bMF₁ and TF₁ indicated that the β subunit exhibits the smallest RMSD value (1.27 $\AA$), followed by the α subunit with a second lower value (1.47 $\AA$), whereas the γ subunit displays a markedly higher value (5.02 $\AA$) (Figure S1). The relatively lower sequence and structural identities of the α and the γ subunits among species are considered to imply that the α or the γ subunits are responsible for the species‐specific recognition of IF₁. Thus, the mechanism of the species‐specific inhibition of IF₁ remains elusive.

This study aims to address this question by engineering TF₁ to confer the susceptibility to bovine mitochondrial IF₁. To determine which subunit is responsible for the species specificity, we first tested the hybrid F₁s, which are composed of each α, β, γ subunit from bMF₁ or TF₁ (Watanabe et al., 2023). Then, based on the sequence analysis, we constructed the quintuple and the octuple mutants of TF₁, where the non‐conserved five or eight residues in contact with IF₁ were substituted to the corresponding residues of bMF₁. We also tested the quintuple mutant of bMF₁, where the same five residues were replaced to the TF₁'s counterparts. Our results clearly showed that the five non‐conserved residues on the C‐terminal domain of the β subunit are crucial for species‐specific recognition by IF₁. These findings would provide new insights on the recognition mechanism of IF₁ as well as on the development of inhibitors targeting F₁ with bacterial origin, discriminating from mitochondrial F₁s.

2. RESULTS

2.1. IF₁ inhibition in wild‐type bMF₁ and TF₁

ATPase activity was quantified by NADH absorbance monitored at 340 nm. ATP hydrolysis reaction was initiated upon the addition of F₁ to the assay mixture of an ATP‐regenerating system, leading to an immediate decay in NADH absorbance upon ATP hydrolysis. As previously reported (Hirono‐Hara et al., 2001; Kobayashi et al., ^{2020, 2023}), F₁ showed initial rapid ATP hydrolysis, followed by gradual deceleration to reach the steady‐state catalysis. This time‐dependent inactivation is due to the ADP inhibition, where F₁ molecules transiently lapse into an inactive state during catalysis (Figure S2 and Table S3; Lapashina & Feniouk, ²⁰¹⁸; Turina, ²⁰²²). After confirming the steady‐state catalysis of F₁, that is, 180 s after F₁ injection, IF₁ inhibition measurement was initiated by adding IF₁ into the reaction cuvette. The blue plot in Figure 2a shows the typical time‐course of IF₁ inhibition in bMF₁(WT) in the presence of 1 mM ATP and 1 μM IF₁, where IF₁ inhibition in bMF₁ was almost maximum. After IF₁ injection, the slope of NADH absorbance abruptly went to almost zero, suggesting that IF₁ blocked the catalysis of bMF₁ within the mixing time. This rapid and complete inhibition is a significant feature of IF₁ inhibition in bMF₁. By contrast, TF₁ was not susceptible to IF₁ at all (Figure 2b). One may concern that the residual ATPase activity of TF₁ was decreased down to 80% after IF₁ injection. Notably, the same decay of ATPase activity was also observed in the IF₁‐free condition (Figure S2). Thus, this is due to slow inactivation by ADP inhibition that slowly proceeds even after 180 s.

FIGURE 2.

IF₁ inhibition assay with bMF₁(WT), TF₁(WT) and four hybrid F₁s. (A) Time‐courses of six F₁s standardized by each initial activity. The absorbance value on the y‐axis is taken from the raw data of bMF₁, and the others are standardized. The triplet characters represent the origins (‘b’ from bMF₁ or ‘T' from TF₁) of the three subunits, α, β, and γ, in this order; bbT represents the hybrid F₁ with α and β from bMF₁ and γ from TF₁. Experiments were conducted at 1 mM ATP and 1 μM IF₁. The raw data set of the time‐courses is presented in Figure S4. (B) Residual ATPase Activity after IF₁ injection. Residual ATPase activity was measured at the end of the measurement (300 s after IF₁ injection). Values represent the mean value of three measurements, and error bars represent SD (N = 3 for each measurement). (C) Time‐courses of the bbT hybrid at the indicated [IF₁]s with 1 mM ATP. $\left(\mathrm{D}\right)k_{\text{inhibition}}^{\mathrm{app}}$ plotted against [IF₁]. $k_{\text{inhibition}}^{\mathrm{app}}$ was determined by the fitting of the NADH absorbance change by (Equation 1). The solid lines represent the fitting curve of (Equation 2). Circles and error bars represent the mean value and SD, respectively (N = 3 for each measurement). (E) Residual ATPase activity versus [IF₁] at the end of the measurement. The inset is the expansion of the plots. The residual ATPase activity was <5% in each datapoint, and they mostly overlap in this figure. Circles and error bars represent the mean value and SD, respectively (N = 3 for each measurement).

2.2. IF₁ inhibition in hybrid F₁s

We prepared four kinds of hybrid F₁s composed of subunits from bMF₁ or TF₁ along our previous article, bbT, bTb, bTT, and TTb (Watanabe et al., 2023). The triplet characters represent the origins (‘b’ from bMF₁ or ‘T' from TF₁) of the three subunits, α, β, and γ, in this order; bbT represents the hybrid F₁ with α and β from bMF₁ and γ from TF₁. Analysis of ATPase activity in hybrid F₁s is summarized in Figure S3 and Table S3. The IF₁ susceptibility of these hybrid F₁s was investigated under the same condition for bMF₁(WT) and TF₁(WT). Figure 2a shows the time‐courses standardized with the initial activities (=slope before IF₁ injection). The raw data set of the time‐courses is presented in Figure S4. The residual ATPase activities at 300 s after IF₁ injection are summarized in Figure 2b. The hybrid F₁s did not show the IF₁ susceptibility. The exception is bbT with β from bMF₁ (the green plot in Figure 2a). All of the other hybrids: bTb, bTT, and TTb carry β from TF₁ while α and/or γ originate from bMF₁. Thus, it is evident that the β subunit is the determinant subunit for the IF₁ susceptibility.

We have investigated IF₁ inhibition in bbT hybrid at various concentrations of IF₁ in the presence of 1 mM ATP (Figure 2c–e). The time‐courses of IF₁ inhibition in bbT hybrid show clear inhibition at 0.1–1 μM IF₁ (Figure 2c). We also analyzed the residual ATPase activity at 300 s after IF₁ injection, and found that IF₁ completely suppressed ATPase activity of bbT hybrid (Figure 2e). These features are well similar to the inhibition in bMF₁(WT). To estimate the rate constant of IF₁ inhibition, $k_{\text{inhibition}}^{\mathrm{app}}$, we have fitted the time‐courses with the exponential decay function (Kobayashi et al., 2021; see Section 5),

$$y\left(t\right)-y_0=V_{\infty }t+\frac{V_0-V_{\infty }}{k_{\text{inhibition}}^{\mathrm{app}}}\left\{1-\exp \left(-k_{\text{inhibition}}^{\mathrm{app}}t\right)\right\}$$ (1)

The resultant $k_{\text{inhibition}}^{\mathrm{app}}$ plotted against [IF₁] shows the typical saturation curve, indicating that IF₁ inhibition in bbT hybrid follows the two‐step mechanism as found for bMF₁(WT) (Figure 2d). According to the equation, [IF₁]‐dependent $k_{\text{inhibition}}^{\mathrm{app}}$ is expressed as follows:

$$k_{\text{inhibition}}^{\mathrm{app}}=\frac{k_{\text{lock}}\left[\mathrm{I}{\mathrm{F}}_1\right]}{K_M^{\mathrm{I}{\mathrm{F}}_1}+\left[\mathrm{I}{\mathrm{F}}_1\right]}$$ (2)

where $k_{\text{lock}}$ is the rate constant for isomerization to the fully inhibited state, and $K_M^{\mathrm{I}{\mathrm{F}}_1}$ represents an affinity with IF₁, respectively. These values characterize kinetic features of IF₁ inhibition; higher $k_{\text{lock}}$ explains rapid inhibition and lower $K_M^{\mathrm{I}{\mathrm{F}}_1}$ explains tight inhibition. The comparison of these kinetic parameters between bbT hybrid and bMF₁ provides important insights on the molecular mechanism of IF₁ inhibition. At low [IF₁], where the initial IF₁ association is the rate‐limiting step, bMF₁ and bbT hybrid showed the similar inhibition rate, that is seen in the productive binding rate of IF₁ estimated from $k_{\text{lock}}/K_M^{I{F}_1}$ (Table S4; Kobayashi et al., ²⁰²¹). It suggests that the interspecies substitution of γ subunit does not affect the association step of IF₁. By contrast, at high [IF₁], where the locking process to the final state is dominant in the overall reaction, bMF₁ reached the final inhibited state 1.6 times faster than bbT, as seen in the estimation of $k_{\mathit{\text{lock}}}$. This result indicates that γ subunit can enhance the rate of the final locking process to some extent.

2.3. IF₁ ‐susceptive TF₁ mutants

The experiments with hybrid F₁s highlighted the dominant role of β subunit in IF₁ susceptibility. According to the bMF₁‐(IF₁)₃ structure, where three IF₁s were bound to each αβ pair, 22 residues in α_E, α_DP, β_DP, β_TP, and γ subunit were identified as the IF₁‐contacting residues (Table S1): 17 residues are on β subunit, two residues on α subunit and three residues on γ subunit (Figure 3a). The sequence comparison between bMF₁ and TF₁ shows that five out of 17 residues in β subunit are non‐conserved, one out of two residues in α subunit, and two out of three residues in γ subunit, respectively (Figure 3a).

FIGURE 3.

Design concept of TF₁(β⁵) and TF₁(α¹β⁵γ²). (A) Sequence comparison between bMF₁ and TF₁. The colored characters indicate the amino acid residues involved in IF₁ binding, as found in the bMF₁‐(IF₁)₃ structure (Table S1). The red ones represent the different residues between bMF₁ and TF₁, whereas the blue ones represent the identical residues. (B) Side (left) and top (right) views of the positions of mutations in the TF₁(β⁵) and TF₁(α¹β⁵γ²). The structural data of PDB: 2v7q for the bMF₁‐IF₁ complex was used for these illustrations. α_DP, β_DP, γ, and IF₁ are shown in pink, cyan, yellow, and green, respectively. The following mutations are included in the TF₁(β⁵) or TF₁(α¹β⁵γ²): H401S, K467D, M469L, G470A, V473H in the β subunit, Q345E in the α subunit, K16N and T17I in the γ subunit (TF₁ numbering).

To assess the role of these non‐conserved residues in IF₁ susceptibility, we prepared two kinds of TF₁ mutants, by substituting these non‐conserved amino acid residues of TF₁ with the corresponding bMF₁ residues. The resultant mutants were named TF₁(β⁵) and TF₁(α¹β⁵γ²) (Figure 3b); TF₁(β⁵) is a quintuple mutant that possesses H401S, K467D, M469L, G470A, and V473H on the β subunit, and TF₁(α¹β⁵γ²) is an octuple mutant that possesses Q345E on the α subunit, and K16N and T17I on the γ subunit, in addition to the five mutations of TF₁(β⁵). ATPase activity measurement of these mutants showed the maximum ATPase activity at 3 mM for TF₁(β⁵) and 1 mM for TF₁(α¹β⁵γ²), although the ATPase activity of TF₁(α¹β⁵γ²) was a few times smaller than TF₁(β⁵) and TF₁(WT) (Figure S5). Under the above conditions, the mean time for ADP inhibition was 20 s for TF₁(β⁵) and 90 s for TF₁(α¹β⁵γ²), respectively (Table S3). To minimize the possible effect of ADP inhibition on IF₁ inhibition measurements, IF₁ was injected at 240 s after F₁ injection.

Biochemical assay of IF₁ inhibition with TF₁(β⁵) and TF₁(α¹β⁵γ²) was conducted under the saturating ATP conditions (Figure 4). The time‐courses of IF₁ inhibition with the mutants (Figure 4a) showed gradual deceleration of ATPase activity after IF₁ injection, while TF₁ showed continuous ATPase activity (left panel in Figure 4a). These results confirmed that the engineered TF₁s acquired IF₁ sensitivity as expected. The kinetic analysis of the time‐courses revealed that the mutant TF₁s also obey the two‐step inhibition mechanism, giving $k_{\text{inhibition}}^{\mathrm{app}}$, (Figure 4b and Table S4). The resultant $k_{\text{lock}}$ was determined as 0.015 s⁻¹ of TF₁(β⁵) and 0.022 s⁻¹ of TF₁(α¹β⁵γ²), which were evidently lower than that of bMF₁, 0.032 s⁻¹. $K_M^{\mathrm{I}{\mathrm{F}}_1}$ was estimated as 3.6 μM of TF₁(β⁵) and 2.2 μM of TF₁(α¹β⁵γ²), which were higher than that of bMF₁, 0.3 μM. Thus, although TF₁(β⁵) and TF₁(α¹β⁵γ²) acquired IF₁‐susceptibility, the affinity with IF₁ and the rate of the locking process are both lower than those of bMF₁. The comparison between TF₁(β⁵) and TF₁(α¹β⁵γ²) shows that TF₁(α¹β⁵γ²) is more susceptive to IF₁ than TF₁(β⁵), suggesting the additional role of the mutations at α and γ subunit. It should be noted that both of the mutants did not show complete inhibition that was seen in the IF₁ inhibition in bMF₁(WT) (Figure 4c). The remarkable example is shown in the measurement of TF₁(α¹β⁵γ²) at 10 μM IF₁, where 20% of the residual ATPase was still observed. These observations indicate that the complex of the mutants and IF₁ is not sufficiently stable, so that the mutants undergo reversible association and dissociation with IF₁.

FIGURE 4.

IF₁ inhibition assay with TF₁(β⁵) and TF₁(α¹β⁵γ²). (A) Time‐courses of TF₁(WT) (left), TF₁(β⁵) (center), and TF₁(α¹β⁵γ²) (right) at the indicated concentrations of IF₁. IF₁ inhibition measurement was performed at 1 mM for TF₁(WT) and TF₁(α¹β⁵γ²) and 3 mM for TF₁(β⁵), respectively. $\left(\mathrm{B}\right)k_{\text{inhibition}}^{\mathrm{app}}$ plotted against [IF₁] in TF₁(β⁵) (yellow), TF₁(α¹β⁵γ²) (red), and bMF₁(WT) (blue). $k_{\text{inhibition}}^{\mathrm{app}}$ was determined by the fitting of the NADH absorbance change by (Equation 1). The solid lines represent the fitting curve of (Equation 2). Circles and error bars represent the mean value and SD, respectively (N = 3 for each measurement). (C) Residual ATPase activity of TF₁(β⁵) (yellow), TF₁(α¹β⁵γ²) (red), and TF₁(WT) (gray) versus [IF₁] at the end of the measurement (300 s after IF₁ injection). Circles and error bars represent the mean value and SD, respectively (N = 3 for each measurement).

2.4. IF₁ ‐less‐susceptive bMF₁ (β⁵)

For further confirmation about the important role of the five non‐conserved residues in the IF₁ inhibition, we prepared bMF₁(β⁵), in which these residues were substituted with the corresponding TF₁ residues (Figure 5). The resultant mutant, termed bMF₁(β⁵), includes S405H, D471K, L473M, A474G, and H477V in the β subunit (in bMF₁ numbering). Figure 5a shows the time‐courses of IF₁ inhibition in bMF₁(β⁵) at 1 mM ATP. The mutant required evidently higher concentration of IF₁ than bMF₁(WT); the mutant required 10 μM IF₁ to reach the maximum inhibition. Figure 5b shows [IF₁]‐dependent $k_{\text{inhibition}}^{\mathrm{app}}$ plot in bMF₁(β⁵), which follows a typical hyperbolic curve. The resultant $k_{\text{lock}}$ was determined as 0.021 s⁻¹, which is also lower than that in bMF₁(WT). Furthermore, the affinity parameter, $K_M^{\mathrm{I}{\mathrm{F}}_1}$ was determined as 2.3 μM, which was seven times larger than bMF₁(WT). Thus, these analyses again confirmed the crucial role of these five non‐conserved residues in IF₁‐susceptibility. Interestingly, the residual ATPase activity of bMF₁(β⁵) can reach almost zero (Figure 5c), which is in contrast to the TF₁(β⁵) and the TF₁(α¹β⁵γ²) mutants. This observation suggests that non‐mutated residues in the β‐subunit, or residues in the α and/or γ subunit, support complete inhibition.

FIGURE 5.

IF₁ inhibition assay with bMF₁(β⁵). (A) Time‐courses of bMF₁(β⁵) at the indicated concentrations of IF₁ with 1 mM ATP. $\left(\mathrm{B}\right)k_{\text{inhibition}}^{\mathrm{app}}$ plotted against [IF₁] in bMF₁(β⁵) (purple) and bMF₁(WT) (blue). $k_{\text{inhibition}}^{\mathrm{app}}$ was determined by the fitting of the NADH absorbance change by (Equation 1). The solid lines represent the fitting curve of (Equation 2). Circles and error bars represent the mean value and SD, respectively (N = 3 for each measurement). (C) Residual ATPase activity of bMF₁(β⁵) (purple) and bMF₁(WT) (blue) versus [IF₁] at the end of the measurement (300 s after IF₁ injection). Circles and error bars represent the mean value and SD, respectively (N = 3 for each measurement).

3. DISCUSSION

In this study, we have successfully conferred IF₁‐susceptibility to TF₁(WT), by replacing the five non‐conserved amino acids at the C‐terminus of the β subunit to the corresponding residues of bMF₁ (S405, D471, L473, A474, and H477 in bMF₁ numbering). In the previous study with yeast mitochondrial F₁ (yMF₁) and IF₁ (Wu et al., 2014), E471 and A474 that correspond to D471 and A474 of bMF₁ were substituted to Lys and Glu, respectively. The yMF₁ with double mutations decreased the rate constant of IF₁ association by 7‐fold. This result was consistent with the present study where we observed the 7‐fold decrement in $K_M^{\mathrm{I}{\mathrm{F}}_1}$ of bMF₁(β⁵) in comparison with that of bMF₁(WT) (Figure 5), highlighting the important role of D471 and A474 of bMF₁ for the IF₁ recognition.

Here, we discuss the possible molecular mechanism on the IF₁ recognition based on the present results, with the sequence comparison of the β subunits derived from mitochondrial and non‐mitochondrial enzymes (Figure S6). One straightforward idea is the contribution of the electrostatic interactions. Assuming that electrostatic interactions work as a long‐range attraction force, it is reasonable to consider that D471 of bMF₁ β has an important role in the recognition process. The sequence alignment shows that Asp or Glu at this position are highly conserved in mitochondrial F₁s, whereas these are not conserved among bacterial F₁s (Figure S6). The negative charge of Asp/Glu found in mitochondrial F₁s is likely to facilitate the initial association via electrostatic interactions with the positive charges of IF₁. K46 and K47 of IF₁ are the possible counterpart residues to form electrostatic interaction with Asp/Glu of the β subunit in MF₁s (Figure S7).

Another possible mechanism for the IF₁ recognition is that these mutations increase hydrophobic interaction with IF₁, considering that D471 and A474 of bMF₁ β make direct contacts with IF₁ residues such as L42 and L45 in IF₁ (Figure 6a and Table S1). Interestingly, the sequence alignment of β subunit showed that the A474 of bMF₁ is strictly conserved in other mitochondrial F₁s, whereas the corresponding residue in bacterial ones are non‐conserved (Figure S6). It should be noted that some bacterial enzymes lack a few residues after the residue 474 in bMF₁ numbering. In addition, the corresponding residue in TF₁ is Gly, which is likely to prevent helix formation after this residue. These would suggest the important role of the helix formation at the C‐terminus of the β subunit; the hydrophobic interaction of the C‐terminal helix with IF₁ has the crucial role for the specific‐recognition of mitochondrial F₁ by IF₁. For further investigation of this possibility, we performed the structural prediction of a single β subunit in the IF₁‐free condition using AlphaFold2 (Jumper et al., 2021; Mirdita et al., ²⁰²²). As expected, the mutant β subunit in TF₁(β⁵) took the α‐helix at the C‐terminus, although the wild‐type β subunit failed (Figure 6b). These would implicate that a helical structure at the C‐terminus of β subunit may facilitate the formation of the F₁‐IF₁ complex. One may consider why bMF₁(β⁵) still kept IF₁‐susceptibility despite the introduction of helix‐breaking glycine. The structural prediction suggests that bMF₁(β⁵) can retain the C‐terminal helix even with the Gly mutation. These might be a clue for the molecular mechanism of the species‐specific recognition by IF₁. The precise role of these residues in IF₁ inhibition would be tested in future experiments.

FIGURE 6.

The C‐terminus of the β subunit. (A) Interaction of IF₁ with the C‐terminus of the β_DP subunit in the final inhibited form (PDB: 2v7q). The side chains of D471 and A474 in β_DP and L42 and L45 in IF₁ are represented by stick form. (B) Structural prediction of a single β subunit by AlphaFold2. The last 16 residues are illustrated in these figures.

Another interesting finding in the present study is that the relatively small number of mutations has a remarkable impact on the IF₁‐susceptibility. To discuss the impact of the introduced five mutations from structural point of view, we estimated the buried surface area of hydrophobic residues of IF₁ in the bMF₁‐IF₁ complex using PyMOL and PDBePISA (Krissinel & Henrick, 2007; Table S5 and Section 5). The difference between bMF₁(WT) and bMF₁(β⁵) was 16 ${\AA }^2$, which was so small compared to the total buried surface area of the β_DP subunit with IF₁, 1200 ${\AA }^2$. Furthermore, the resultant energy difference by hydrophobic interaction was estimated to be <0.6 kcal/mol (1 k _BT), implying that the stabilization/destabilization of the complex by these mutations should be, at most, a 3‐fold change in the equilibrium constant. Thus, it is difficult to account for the observed 7‐fold difference in $K_M^{\mathrm{I}{\mathrm{F}}_1}$ between bMF₁(WT) and bMF₁(β⁵). The more evident impact of the five mutations on the association was found in the difference in the productive binding rate of IF₁ that is estimated from $k_{\text{lock}}/K_M^{\mathrm{I}{\mathrm{F}}_1}$; bMF₁(WT) has 11 times higher rate than bMF₁(β⁵). Taking this into account, it is likely that the principal role of the five residues is to stabilize the initial association complex with IF₁ rather than to stabilize the final locked state. The positions of these residues in the three‐dimensional structure are along this contention; these residues are located at the entrance of the crevice of the αβ interface, from which IF₁ would be inserted.

Probably, relevant to the above argument, the mutant TF₁(β⁵) did not achieve complete inhibition. In the assay with TF₁(β⁵), we observed some residual ATPase activity remaining under the saturating IF₁ conditions, even after sufficient time has elapsed after IF₁ injection (Figure 4c). This means that the five residues are not sufficient for the stable formation of the final locked state. The additional mutations into α and γ does not evidently enhance the stability of the locked state. Considering that the bbT hybrid achieved complete inhibition even with γ subunit from TF₁ (Figure 2d), the incomplete inhibition in TF₁(β⁵) and TF₁(α¹β⁵γ²) is attributable to more detailed difference in the structure of the β subunit between bMF₁ and TF₁; more bMF₁‐like structure of the β subunit is required for the stable formation of the locked state. The crystal structure of bMF₁‐(IF₁)₃ revealed that Glu30 of IF₁ and the Arg408 in β subunit make a salt bridge that augments the binding of IF₁ to the β subunit (Gledhill et al., 2007). The importance of this salt bridge for irreversible inhibition was demonstrated by several biochemical assays (Bason et al., 2011; Kobayashi et al., ²⁰²¹). Thus, more optimized interaction between IF₁ and the β subunit is a key to achieve complete IF₁ inhibition.

4. CONCLUSION

Overall, we investigated the molecular origins of species‐specific IF₁ inhibition by engineering bacterial F₁ to confer the susceptibility to IF₁. We found that the non‐conserved residues on the C‐terminus helix of the β subunit are responsible for the IF₁‐susceptibility. We discuss the molecular mechanism by which IF₁ specifically recognizes F₁ with mitochondrial origin, based on the present result and the crystal structure of F₁‐IF₁ complex. These findings would provide new insights into the development of inhibitors targeting F₁ with bacterial origin. The findings of the present study also prove important implications for the de novo design of the inhibitory peptides against F₁s with bacterial origin.

5. MATERIALS AND METHODS

5.1. Construct and purifications of proteins

The proteins of bMF₁(WT), TF₁(WT), four hybrid F₁s, and IF₁ were expressed and purified as described in the reference article (Watanabe et al., 2023). For TF₁(β⁵), TF₁(α¹β⁵γ²), and bMF₁(β⁵), the artificially synthesized construct with mutations was introduced into the wild‐type TF₁ and bMF₁ plasmids, respectively. Purifications were performed according to the method of TF₁ (for TF₁(β⁵) and TF₁(α¹β⁵γ²)) and bMF₁ (for bMF₁(β⁵)). The sequence of the mutant proteins was confirmed using the Fasmac sequencing service (Fasmac, Japan). The purified mutant F₁s were confirmed by sodium dodecyl sulfate poly‐acrylamide gel electrophoresis (SDS‐PAGE) analysis (Figure S8).

5.2. Biochemical assay and analysis

The ATPase activity of F₁ was quantified from the oxidation rate of NADH, monitored at 340 nm (Kobayashi et al., 2021). Experiments were performed at 25°C, using a UV/VIS spectrophotometer equipped with a temperature controller. The buffer contained 50 mM HEPES‐KOH (pH 7.5; for bMF₁(WT), hybrid F₁s, bMF₁(β⁵)) or 50 mM MOPS‐KOH (pH 7.5; for TF₁(WT), TF₁(β⁵), TF₁(α¹β⁵γ²)), 50 mM KCl, an excess amount of MgCl₂ over [ATP], 0.2 mg/mL pyruvate kinase, 2.5 mM phosphoenolpyruvate, 0.2 mM NADH and 50 μg/ml lactate dehydrogenase. NADH was added during the measurement if necessary.

The ATPase reaction was initiated by adding purified F₁ into the buffer including ATP. Time‐course of ATPase activity is exemplified in Figures S2 and S3. IF₁ inhibition measurement was started after the activity of F₁ reached a steady state, that is, 180 s for bMF₁(WT), TF₁(WT), four hybrids, and bMF₁(β⁵), 240 s for TF₁(β⁵) and TF₁(α¹β⁵γ²). As described previously (Kobayashi et al., 2021), the rate constant of IF₁ inhibition ($k_{\text{inhibition}}^{\mathrm{app}}$) was quantified by fitting the time‐course using a following single‐exponential decay function:

$$y\left(t\right)-y_0=V_{\infty }t+\frac{V_0-V_{\infty }}{k_{\text{inhibition}}^{\mathrm{app}}}\left\{1-\exp \left(-k_{\text{inhibition}}^{\mathrm{app}}t\right)\right\}$$ (3)

The residual ATPase activity was determined from the activity 300 s after IF₁ injection relative to that just before IF₁ injection.The progressive inhibition equation of IF₁ with the initial association process followed by the isomerization process is as follows:

$$F_1+\mathrm{I}{\mathrm{F}}_1\begin{array}{c}\stackrel{k_{\mathrm{on}}}{\to }\\ \underset{k_{\mathrm{off}}}{\leftarrow }\end{array}{F}_1\bullet I{F}_1\stackrel{k_{\text{lock}}}{\to }{F}_1\bullet \mathrm{I}{\mathrm{F}}_1^{\text{lock}}$$ (4)

where $F_1\bullet I{F}_1^{\text{lock}}$ represents the final inhibited form, and $F_1\bullet \mathrm{I}{\mathrm{F}}_1$ is the intermediate state before isomerization. The kinetic parameters, $k_{\mathrm{on}}$ and $k_{\mathrm{off}}$ represent the rate constants of association and dissociation with F₁, respectively. $k_{\text{lock}}$ is the rate constant of the isomerization to the fully folded state of IF₁ in the 2nd process. According to this equation, $k_{\text{inhibition}}^{\mathrm{app}}$ and $K_M^{\mathrm{I}{\mathrm{F}}_1}$ are defined as follows:

$$k_{\text{inhibition}}^{\mathrm{app}}=\frac{k_{\text{lock}}\left[I{F}_1\right]}{K_M^{I{F}_1}+\left[I{F}_1\right]}$$ (5)

$$K_M^{\mathrm{I}{\mathrm{F}}_1}\equiv \frac{k_{\mathrm{off}}+k_{\text{lock}}}{k_{\mathrm{on}}}$$ (6)

The resultant $k_{\text{inhibition}}^{\mathrm{app}}$ plot against [IF₁] follows a hyperbolic curve.

5.3. Structural comparison between bMF₁ and TF₁

The catalysis‐waiting structures of bMF₁ (PDB: 2jdi; Bowler et al., ²⁰⁰⁷) and TF₁ (PDB:7l1r; Sobti et al., ²⁰²¹) were used for analysis (Figure S1). The RMSD (root mean square deviation) for each subunit was estimated from the Cα atoms of the backbone using the Python packages, MDAnalysis (Liu et al., 2010; Michaud‐Agrawal et al., ²⁰¹¹; Theobald, ²⁰⁰⁵). The amino acid residues analyzed here are as follows. For the α subunit, residues 28–187, 196–401, 413–483, 494, and 497–509 in bMF₁, and residues 28–393, 405–475, 486–499 in TF₁ were analyzed. For the β subunit, residues 10–126, 129–206, 213–310, 312–387, 396–474 in bMF₁, and residues 3–21, 23–35, 43–115, 117–128, 131–306, 308–383, 392–470 were analyzed. For the γ subunit, residues 1–47, 67–69, 72–86, 105–116, 130–148, 159–173, 206, 211–271 in bMF₁, and residues 6–49, 61–63, 83–100, 119–130, 140–145, 148–157, 161–163, 173–187, 226–287 in TF₁ were analyzed.

5.4. Estimation of buried surface area and binding energy estimation

Buried surface areas of hydrophobic amino acids were calculated with PDBePISA (Krissinel & Henrick, 2007; Table S5). For this purpose, modeling of bMF₁(β⁵) was performed using the PyMOL mutagenesis tool, with the least clashed rotamer candidate selected for each mutation.

5.5. Structural prediction by AlphaFold2

Structures of a single β subunit of bMF₁(WT), bMF₁(β⁵), TF₁(WT), and TF₁(β⁵) (Figure 6b) were predicted using AlphaFold2 in the Google Collaboratory with default parameters (ColabFold v1.5.2‐patch: AlphaFold2 using MMseqs2; Jumper et al., 2021; Mirdita et al., ²⁰²²).

AUTHOR CONTRIBUTIONS

Ryohei Kobayashi: Investigation; visualization; conceptualization; methodology; writing – original draft; writing – review and editing. Yuichiro Hatasaki C: Investigation; methodology; writing – review and editing. Ryo Watanabe R: Investigation; writing – review and editing. Mayu Hara: Investigation. Hiroshi Ueno: Investigation; conceptualization; writing – review and editing. Hiroyuki Noji: Writing – original draft; supervision; conceptualization; writing – review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Supporting information

DATA S1. Supporting Information.

Hatasaki YC, Kobayashi R, Watanabe RR, Hara M, Ueno H, Noji H. Engineering of IF₁‐susceptive bacterial F₁‐ATPase. Protein Science. 2024;33(4):e4942. 10.1002/pro.4942