Structure of a thermophilic F1-ATPase inhibited by an ε-subunit: deeper insight into the ε-inhibition mechanism.

Structure of a thermophilic F1-ATPase inhibited by an e-subunit: deeper insight into the e-inhibition mechanism Yasuo Shirakihara1, Aya Shiratori1, Hiromi Tanikawa1, Masayoshi Nakasako2,*, Masasuke Yoshida3,4,† and Toshiharu Suzuki3,4,‡ 1 2 3 4

National Institute of Genetics, Mishima, Japan The Institute of Molecular and Cellular Biosciences, The University of Tokyo, Japan The Chemical Resources Laboratory, Tokyo Institute of Technology, Yokohama, Japan ERATO, Japan Science and Technology Corporation (JST), Yokohama, Japan

Keywords F1 from thermophilic bacterium; F1-ATPase; second catch; e-inhibition; e-subunit Correspondence Y. Shirakihara, National Institute of Genetics, Mishima, Shizuoka 411-8540, Japan Fax: +81 559 81 6888 Tel.: +81 559 81 6887 E-mail: [email protected] Present addresses: *Department of Physics, Keio University, Yokohama, Japan † Kyoto Sangyo University, Japan ‡ Department of Applied Chemistry, The University of Tokyo, Japan (Received 28 January 2015, revised 19 May 2015, accepted 26 May 2015) doi:10.1111/febs.13329

F1-ATPase (F1) is the catalytic sector in FoF1-ATP synthase that is responsible for ATP production in living cells. In catalysis, its three catalytic b-subunits undergo nucleotide occupancy-dependent and concerted open–close conformational changes that are accompanied by rotation of the c-subunit. Bacterial and chloroplast F1 are inhibited by their own e-subunit. In the einhibited Escherichia coli F1 structure, the e-subunit stabilizes the overall conformation (half-closed, closed, open) of the b-subunits by inserting its C-terminal helix into the a3b3 cavity. The structure of e-inhibited thermophilic F1 is similar to that of E. coli F1, showing a similar conformation of the e-subunit, but the thermophilic e-subunit stabilizes another unique overall conformation (open, closed, open) of the b-subunits. The e-C-terminal helix 2 and hook are conserved between the two structures in interactions with target residues and in their positions. Rest of the e-C-terminal domains are in quite different conformations and positions, and have different modes of interaction with targets. This region is thought to serve e-inhibition differently. For inhibition, the e-subunit contacts the second catches of some of the b- and a-subunits, the N- and C-terminal helices, and some of the Rossmann fold segments. Those contacts, as a whole, lead to positioning of those b- and a- second catches in e-inhibition-specific positions, and prevent rotation of the c-subunit. Some of the structural features are observed even in IF1 inhibition in mitochondrial F1. Database Structural data are available in the Worldwide Protein Data Bank database under the accession number 4XD7

Introduction FoF1-ATP synthase plays an important role in energy conversion in mitochondria, chloroplasts and bacteria, synthesizing ATP from ADP and inorganic phosphate using energy derived from proton translocation along the transmembrane electrochemical potential gradient (for reviews, see Refs [1–3]). FoF1-ATP synthase con-

sists of a large soluble catalytic sector, F1, and a transmembrane proton-transporting sector Fo; F1 has a subunit composition of a3b3cde (approximate molecular masses in bacterial enzymes: a, 55 000 Da; b, 52 000 Da; c, 32 000 Da; d, 20 000 Da; e, 14 000 Da) [4]. The three a- and b-subunits are arranged alter-

Abbreviations CyDTA, trans-1,2-diaminocyclohexane-N,N,N0 ,N0 -tetraacetic acid, monohydrate; EF1, F1 from E. coli; MF1, F1 from bovine mitochondria; SeMet, selenomethionine; TF1, F1 from thermophilic Bacillus PS3.

FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS

2895

Structure of e-inhibited thermophilic F1-ATPase

nately, forming a barrel into which the coiled-coil portion of the c-subunit penetrates [5]. The rest of the csubunit emerges from the bottom of the barrel. The smallest subunit, e (d in the case of mitochondrial F1) is associated with the c-subunit, forming a stalk. The e-subunit has an inhibitory role in the bacterial and chloroplast enzymes. The three catalytic sites are on the b-subunits, but the a-subunits also contribute. F1 is a motor enzyme and its ATPase activity is accompanied by rotation of the stalk relative to the a3b3 barrel [6]. Rotation of the c-subunit is thought to be a consequence of the combination of the cyclic, concerted conformational changes in the three b-subunits, induced by alternating nucleotide occupancies and asymmetrical association of the b-subunits with the c-subunit [1]. The asymmetry of the three b-subunits in nucleotide binding, conformations and interactions with the c-subunit has been shown by the first crystal structure of bovine F1 [5] and its subsequent structures ([7] and references therein). Two nucleotidebound b-subunits (bTP with Mg–AMP–PNP bound to its ‘loose’ catalytic site, bDP with Mg–ADP bound to its ‘tight’ catalytic site) have a ‘closed’ conformation in which the C-terminal half is swung closer to the N-terminal half, whereas the third b-subunit (bE) has an empty ‘open’ catalytic site and adopts the ‘open’ conformation in which the C-terminal half is positioned off the N-terminal half. This structure, characterized by a close–close–open b-subunit conformation (CCO), has been observed in all F1 structures including bovine (MF1) [7], yeast [8,9] and E. coli (EF1) [10]. There are three exceptions, they are: the three catalytic-sites-filled MF1 structure in which the bE-position b-subunit is in the ‘half-closed’ conformation [11], the EF1 structure in which the bDP-position b-subunit is in the ‘halfclosed’ conformation [12] and the TA2 (thermoalkaliphilic Bacillus sp.TA2.A1) F1 structure in which all the three b-subunits are in the open conformation (OOO) [13]. A range of molecules inhibit the rotational catalytic mechanism by blocking the coordinated structural changes in the b- and a-subunits that interplay with rotation of the c-subunit. In MF1, small molecules like azide, dicyclohexylcarbodiimide, antibiotics (efrapeptins and aurovertins) and polyphenolic inhibitors (resveratrol, quercetin and piceatannol) cause local structural changes specific to each inhibitor, but do not disturb the overall F1 structure [14]. By contrast, protein inhibitors, such as IF1 involved in MF1 inhibition [14] and the e-subunit (e-inhibition) [15], induce changes in the overall F1 structure [12,14]. IF1 changes the structure and properties of the bDP–aDP interface to those of the bTP– aTP interface, thereby eliminating the bDP–aDP interface 2896

Y. Shirakihara et al.

(for the tight catalytic site) seen in most of the F1 structures [14]. The e-subunit of E. coli F1 (EF1) stabilizes a unique overall conformation that also lacks the tight site bDP conformation and instead has the half-closed b-structure in this position [12]. It has been assumed that the physiological role of the inhibition of ATPase activity is to prevent unnecessary ATP hydrolysis [12,14,15]. IF1 inhibition appears to take place when acidification of the mitochondrial matrix occurs in the case of a low oxygen supply. The active dimeric form IF1, formed only under acidic conditions, can bind to MF1 molecules [14]. In bacterial enzymes, the physiology of inhibition is less clear, but based on extensive studies on thermophilic F1 (TF1) e-inhibition [15,16], it is generally assumed that inhibition prevents cellular ATP from being wasted when the proton motive force is very low. TF1 is from a thermophilic Bacillus PS3 and its biochemistry and stalk rotation have been studied extensively [6]. e-Inhibition is associated with conformational and positional changes in the e-subunit’s C-terminal helical domain, which forms the e-subunit with the N-terminal b-sandwich domain [17–19]. These structural changes are thought to depend on ATP concentration, as clearly demonstrated in TF1 [16,20,21]. Under low ATP conditions, where inhibition occurs, the two C-terminal helices are extended so that their end tip is close to the N-terminal end of the c-subunit (an inhibitory ‘up’ conformation). This extended conformation is thought to be similar to that seen in the EF1 structure [12]. Under high ATP conditions, where inhibition is not necessary, the two C-terminal helices are folded to form antiparallel helices and are at the bottom of the F1 assembly without any contact with other subunits (a noninhibitory ‘down’ conformation). The ‘down’ conformation is seen in the crystal structure of MF1 [7], and is very similar to that of the isolated bacterial e-subunit [17–19]. These observations are consistent with earlier cross-linking studies on EF1 [22–24] and TF1 [21]. Transition between the up and down conformations in the isolated bacterial e-subunit is ATP dependent [19]. TF1 forms a stable 1 : 1 complex with Mg-free ADP [25]. The complex can be obtained by applying a nucleotide-removing gel-filtration procedure to a mixture of TF1 and ADP. ADP–TF1 complex is in the inhibited state, and exhibits a longer ATPase lag phase than nucleotide-free TF1. To crystallize the unique complex, an a3b3ce subcomplex, that is almost identical to F1 in terms of its catalytic properties but forms crystals more readily, was used in this study. Here, we present the a3b3ce subcomplex structure (hereafter referred to as the TF1 structure), in the Mg-free ADP resolution. The strucbound form, analysed to 3.9 A FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS



ture presents a well-defined ‘up’ state conformation of the e-subunit that is similar to that observed in the recent EF1 structure [12], which was quite valuable for interpreting the TF1 e-inhibited structure. Despite a general similarity between the two structures, there are significant differences. First, the TF1 structure has an open conformation b-subunit in place of the closed conformation b-subunit (bDP in the MF1 structures). In the EF1 structure, it is the half-closed conformation. Second, the TF1 e-subunit has quite a different conformation and position of the helix 1 of the Cterminal domain from those of the EF1 e-subunit, though it has very similar conformation and position, of the helix 2 and the hook, to those of the EF1 e-subunit. Lastly, but most importantly, thorough comparison of the TF1 structure with the e-inhibited EF1 structure enabled us to identify common structural features relevant to e-inhibition. The comparison was extended to include the MF1–IF1 structure [14], allowing inhibitory features common to the three inhibited structures to be defined, leading to a deeper insight into the inhibition mechanism.

Results Crystallization and structure determination TF1 (a3b3ce subcomplex) protein, engineered to have a 10-residue histidine tag at the N terminus of the b-subunits, was expressed in FoF1-deficient E. coli cells, and then purified with three steps of column chromatography that included a Ni-NTA column in the first step [26]. Crystals of a 1 : 1 complex of TF1

Fig. 1. Electron-density maps in the TF1 structure analysis. Maps are for e-subunit helix 2 (black sticks), part of helix 1 (black sticks), the hook (black sticks) and the Nterminal domain (green backbone). A final 2Fo – Fc map, calculated with the B-factor sharpening (B = 95) [46] to a resolution of 3.9 A, is shown in cyan. A MAD map is in pink. The latter map is drawn only where the final map does not cover. The two maps have a map correlation coefficient of 0.42. All maps are contoured at 1.0r. Starting residues of the helix 2 (e107) and the loop between the helices (e106), and the residue eMet124 are labelled. A Se position corresponding to eMet124 is also indicated.


and Mg-free-ADP were obtained from partially nucleotide-depleted preparations (nucleotide content < 0.2 mol nucleotidemol1 TF1) under Mg-nucleotide resolution structure of free conditions. An initial 4.2 A TF1 was determined by molecular replacement and rigid body refinement. The structural model contained models of all the subunits and ADP. Although the aand b-subunits show weak side-chain features, these are missing in the c- and e-subunits, and so register errors in those subunits might be suspected. Selenomethionine (SeMet)-substituted TF1 crystals were then grown and analysed. A large increase in the concentration of the precipitant in the cryoprotection solution led to an improvement in the crystal resolu The resultant MAD map gave a strong tion to 3.9 A. support for the backbone part of the initial 4.2 A structure. Also, SeMet positions were consistent with those expected from the earlier model. However, the MAD map was not good enough to show up sidechain features (Fig. 1). Refinement was done using the peak data from the MAD dataset (Table 1). A final map that has better side chain features than those in the molecular replacement map (not shown) is shown in Fig. 1. In the final model, possible register errors in the stalk subunits were found to have been minimized by checking the model Se positions against the MAD selenium positions [c24, c26 and c46 in the c N-terminal helix; c170 in the c Rossmann fold; c24 and c244 in the c C-terminal helix; e22 in the e-barrel; e124 (shown in Fig. 1) in the e-helix 2] and by examining a B-factor sharpened omit map, which showed the side-chain features very clearly but is with less

124M

124M

Se

Se

107

106

107

106

- Refinement map - MAD map

2897


Table 1. X-Ray structure determination. SeMet TF1 Data collection Wavelength 0.97879 Space group I4122 Cell dimensions ( A) 233.5, 233.5, 304.3 Resolution ( A) 69.2–3.9 (4.11–3.9) No. of reflections 254 703/38 420 (measured/unique) Rmerge 0.100 (0.522) Mean [(I)/SD(I)] 14.7 (3.3) Completeness (%) 99.6 (97.9) Mulitiplicity 6.6 (5.7) Anomalous 98.2 (93.0) completeness Anomalous 3.4 (3.0) multiplicity Molecular replacement with GLRF and TF [43] Peak height above the mean in cross rotation search R factor in translation search Rigid body refinement with XPLOR [44] Resolution ( A) No. of reflections R factor Overall B-factor refinement with CNS [47] R factor for 95% data Free-R factor for 5% data Refinement with CNS and PHENIX [49] Resolution ( A) 20–3.9 No. of reflections 71 874 R factor for 0.251 95% data Free-R factor 0.281 for 5% data Average B-factor 123 r.m.s.d. Bond lengths ( A) 0.011 Bond angles (°) 1.5 Ramachandran 86.0 favoured (%) Ramachandran 0.0 disallowed (%)

TF1

1.0 I4122 233.5, 233.5, 303.5 50–4.2 369 016/31 279 0.105 (0.452) 16.3 (2.4) 97.5% (90.3) 12.0 (6.8) – –

6.6r

0.438

12–4.2 30 034 0.415 0.377 (0.418) 0.372 (0.425)

Values in parentheses are for the highest resolution bin.

model bias than the 2Fo – Fc map shown in Fig. 1. Naming of the a- and b-subunits follows that used in the EF1 structure [12], because the TF1 and EF1 structures are very similar. Based on the conserved interactions between the b- and c-subunits, as observed in the EF1 structure, a1, a2, a3, b1, b2 and 2898


b3 correspond to aDP, aE, aTP, bDP, bE and bTP, respectively, in the MF1 structures (Fig. 2B). The final model contains 3156 residues, one ADP molecule and four sulfate molecules: a1 24–502 (483 and 484 missing), a2 26–499 (303–305, 312, 340, 367–368, 378–412, 428, 440–443 and 481–486 missing), a3 26–502 (282 and 351 missing), b1 2–470, b2 2–472, b3 4–470 (19, 24 and 244 missing), c 2–284 (59–68, 105, 131–132, 163 and 193–208 missing) and e 1–133 (30 and 58–60 missing). In each of the subunit chains, 14–25% of the residues have partially or completely missing side-chain atoms. Those numbers are correlated with the average B-factor of the subunits. Features of the TF1 overall structure and of the conformations of the b-, a- and c-subunits The overall TF1 structure is shown in Fig. 2. In side view (A), the a- and b-subunits, which form an F1 head, seem to overlap and the stalk subunits emerge from the head. The structure is generally similar to the previous F1 structures, but has unique differences. These are shown in Fig. 3 where a portion of F1 structures is illustrated for TF1 (A), EF1 (B), MF1–IF1 (C) and the ground state MF1 (D). One of the apparent differences is in the conformation of the e-subunit. As shown in Fig. 2B, the TF1 e-subunit, in red, is located at bottom of the TF1 structure with its C-terminal helices penetrating the a3b3 core from the b1–a2 interface. Figure 3A also shows this clearly. This e-subunit (or d in MF1) conformation, which has extended C-terminal helices, was not observed in any of the previous structures such as those in Fig. 3C,D, with the exception of the EF1 structure (Fig. 3B). The other obvious difference is in the conformation of the b1-subunit, which is open, as identified by a large tilt in the C-terminal domain (Figs 2A and 3A). This is a unique feature, because all F1 structures have the closed b-subunit in this position, as illustrated in Fig. 3C,D, with the exception of the EF1 structure (Fig. 3B). The closed b-subunit conformation in this position has been assumed to show the state having the tight site [5]. In the EF1 structure, the b1-subunit is in the half-closed conformation. The conformations of the TF1 b1- and b2-subunits do not differ from those of other open subunits. For example, they are superimposed onto the open subunit for in the ground state MF1 with an rmsd of ~ 0.7 A between the b1- and b2the a-carbon positions (0.3 A subunits). A sulfate ion is bound on top of the P-loop of those b-subunits, as identified in the TF1 a3b3 subcomplex structure [27]. FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS


A

B


The TF1 b3-subunit has a bound ADP and is in the closed conformation. This ADP has higher temperature factors than those for nearby protein atoms, suggesting weak binding of ADP to this subunit. This is unusual, because bound nucleotides in other F1 structures have much lower temperature factors than those of nearby protein atoms. This is probably because TF1 molecules that retained ADP only weakly after the nucleotide removal operation (essentially a dilution procedure), were crystallized under nucleotide-free conditions. The weak binding of ADP in the b3-subunit suggests that the closed conformation in this position is influenced more by the interaction with the c-subunit. This is consistent with the observation of a closed conformation for the nucleotide-free b-subunit in this position in the nucleotide-free yeast F1 [28] and EF1 [12] structures. The three nucleotide-free TF1 a-subunits are similar to a-subunits in other F1 structures. They are superimposed onto a-subunit structures from other F1 struc tures with rmsd values within the range of 1.1–1.6 A over the entire chain. Similarly, the TF1 c-subunit conformation is generally no different from other c-subunit structures. It is superimposed onto other c-subunit structures with rmsd values within the range but these are for 180–220 overlapping of 1.5–2.0 A, Ca atoms (out of total 285 Ca atoms). Structure of the TF1 e-subunit

Fig. 2. The overall TF1 structure. In the side view (A), the a1subunit (magenta) is at the centre, flanked by the b3 (yellow, left) and b1 (yellow, right) subunits. Other b- and a-subunits are not seen clearly because of overlap. In this orientation, the c-subunit (blue) has its Rossmann fold at left-hand side. The e-subunit (red), with its C-terminal helices penetrating into the a3b3 barrel, is at right-hand side. In the bottom view (B) where all the a- and b-subunits are labelled, the two C-terminal helices of the e-subunit are seen between the b1- and a2subunits.


The e-subunit (e1-133) in the TF1 structure has an N-terminal b-sandwich domain (e1-85), a connecting loop (e86-90) and an extended C-terminal helical domain (e91-133) (Fig. 4), thus being in the ‘up’ conformation. The N-terminal b-sandwich domain is very similar to that in the isolated TF1 e-subunit X-ray structure in the ‘down’ conformation (Ca superimpo e1-85) [19] and to all previous bsition rmsd 0.9 A, sandwich structures (e.g. for the EF1e-subunit [12], e1-80). In Fig. 4, Ca superimposition rmsd = 1.2 A, the e-subunit structure in TF1, the isolated TF1 esubunit structure and the e-subunit structure in EF1 are superimposed using the N-terminal b-sandwich domain. The e-subunit’s structural transition in TF1 from the ‘down’ to the ‘up’ state may involve changes in the connecting loop and the C-terminal domain. The short 3/10 helix (e85–87) in the isolated form is initially disrupted into a loop, and the resultant whole loop (e8690) turns such that it is closer to the viewer in Fig. 4. Helices 1 (e90-104) and 2 (e113-130) in the isolated form then change their orientations so that they resemble those in the TF1 structure. Helix 2 undergoes a 2899


A

B

C

D

further change such that the helical region itself is shifted in the N-terminal direction so that e107–123 residues now form helix 2. This is accompanied by a shortening of e-loop 2 and formation of a C-terminal e-hook (e124–133). The structure of the e-subunit in TF1 differs markedly from that in the EF1 structure, in terms of the conformation and location of helix 1 and its flanking loops, as shown in Fig. 4. The separation between These struchelices 1 in the two structures is ~ 20 A. tural differences make interactions between the e-subunit and other subunits using that part very different. In TF1, helix 1 and the loops make contact only with cLys150 in the c Rossmann fold (with a contact area 2), whereas their counterparts in EF1 contact of 100 A far more residues in the c Rossmann fold (residues 2), those in the b1-subranging from 83 to 154; 830 A unit (373 and 383; 120 A) and those in the a1-subunit 2). By contrast, helices 2 and the e(414 and 415; 70 A 2900


Fig. 3. Schematic representation of (A) TF1, (B) EF1, (C) MF1–IF1 and (D) MF1 ground state structures. Those structures are aligned using the optimal superimposition method as described in Materials and methods. For clarity, only the b1- and b3-subunits (A and B) and equivalent b-subunits in others (bDP and bTP-subunits, respectively; C and D) are shown. The b-subunits are shown in yellow, c in cyan, bacterial e (TF1 and EF1) and d (MF1–IF1 and MF1) in red, MF1 e (MF1–IF1 and MF1) in yellow and IF1 in orange. The second catch regions of the b-subunits (378–397, M382–401, E368– 387) are highlighted in magenta. The radial helix [5] of the c-subunit is labelled ‘R’.

hooks show fewer differences between the TF1 and EF1 structures. In EF1, the e-subunit’s structural transition from the ‘down’ to the ‘up’ state [12] appears to be similar to that in TF1 described above, and involves broadly similar structural changes such as those around the end of the N-terminal domain and those in the connecting loop in TF1. However, in EF1, these changes occur earlier (Ee82) than those in TF1 (Te85) and the resultant loop runs in a different direction. These features may lead to a large positional and conformation difference between the EF1 e helix 1 and its flanking loops, and their counterparts. Interactions of the TF1 e-subunit with other subunits In the TF1 structure, the two C-terminal helices penetrate the a3b3 cavity, reaching close to its centre (Fig. 3A). These helices contact the b1-, b3-, a1-, a2FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS



133

133

E 138

E 138 124

124 E 126

107

E 103

E 126

E 112

107

E 103

106

E 83 E 87

ATP

E 83 E 87

91 90

113 105

105

86

133

131

E 1 1

ATP 91 90

113

85

1

E 112 106

85

86

133

131

1

E 1 1

Fig. 4. Schematic stereo representation of the e-subunit structures. The TF1 e-subunit, shown in the thickest lines, is presented in colourcoded segments [N-terminal b-sandwich domain (1–85), green; a connecting loop (86–90), orange; helix 1 (91–105), pink; loop 1 (106), orange; helix 2 (107–123), red; hook (124–133) orange], each of which is numbered in black and at an N-terminal end only. EF1 e-subunit is shown in a similar way and in the second thickest line (label colour, red). The X-ray TF1 e-subunit structure (ATP-bound form) in a thin line (label colour, cyan) with colour-coded segments [N-terminal b-sandwich domain (1–84), green; a connecting loop (85–89), orange; helix 1 (90–104), pink; loop 1 (105–112), orange; helix 2 (113–130), red; hook (131–133), orange]. Both are superimposed using the N-terminal barrel domain.

and c-subunits (Figs 5A and 2), but not the b2- and a3-subunits (Fig. 2B). Figure 5A shows the e-source residues and the target segments of other subunits, together with the equivalent segments in the ground state MF1 structure as a reference. The target areas in the b1-, b3-subunits (T382–391 and T379–392, respectively) are parts of the second catch region (T378–397, M382–401, E368–387) of the b-subunit, and are C-terminal three turns of helix 1 (in a light colour; Fig. 5A) in the C-terminal domain and a following loop (in a darker colour). Here, the second catch region is defined as part of the C-terminal domain, which interacts with the c-subunit in the ground state MF1 structure (a sum of the c-contacting regions in the three b-subunits). The second catch regions in other F1s FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS

are defined from the sequence. This region is known to change its position dramatically in the F1 structure when there is a conformational change in the b-subunit due to nucleotide binding [5]. The e-helix 2 and the e-hook appear to stabilize the conformation of the second catch regions of the b-subunits, b1 in the open conformation and b3 in the closed conformation, as described below. The b1-subunit is contacted at three 2) and the b3-subresidues (with a contact area of 170 A 2) (Fig. 5A). unit is contacted at eight residues (270 A The target b-loop contains the so-called DELSEED motif (390–396, M394–400, E380–386). Its second residue, bGlu391, affects F1 catalytic activity [29] and also contributes to the e-inhibition [30]. Both b3Glu391 and b1Glu391 are in contact with the e-subunit. 2901



A

B

C

2902




Fig. 5. Interactions of the C-terminal domain of the e-subunit with other subunits in the TF1 structure (A), their comparison with EF1 counterparts (B) and comparison with IF1 inhibition in the MF1–IF1 structure (C). All are stereo figures. (A) TF1 (thick lines) and ground state MF1 (thin broken lines) structures, when superimposed with the optimum superimposition method. The TF1 e-subunit’s backbone is shown in colour-coded segments and the segments are labelled, as in Fig. 4. TF1 segments from b1 (b381–392), b3 (b378–392), a1 (a393–403), a2 (a396–401) and c (c6-25, c84–88, c239–241 and c254–261) subunits are shown in green, yellow, magenta, magenta and cyan, respectively. The interacting residues, of which the Ca atoms are shown in the figure, are: ΤeGln107–Tb1Glu391; ΤeAsp109–TcArg85; ΤeAsp110– TcArg85; ΤeIle111–Tb1Leu387; ΤeAsp112–Ta2Gln397/Ta2Ser400; ΤePhe113–TcArg85/TcGly86/TcArg240; ΤeLys114–Ta1Phe398/Ta1Asp401/ Ta1Leu402; ΤeAla116–TcMet24; ΤeGlu117–Ta1Gln397/Tb3Glu391/TcGly86/TcLeu87; ΤeLeu118–Ta1Ala394/Ta1Phe395/Ta1Gln397/Ta1Phe3 98/Tb1Asp382; ΤeAla119–TcIle20; ΤeLeu120–TcIle20/TcThr21/TcMet24; ΤeLys121–Ta1Gln397; ΤeArg122–Ta1Phe395/Tb1Asp382; ΤeAla123– TcAla13/TcThr17; ΤeMet124–Tb3Asp382/Tb3Ile383/Tb3Ile386/Tb3Leu387; ΤeLeu127–Tb3Glu379/Tb3Asp382; ΤeSer128–Tb3Asp382/TcArg10/ TcAla13; ΤeVal129–Tb3Asp382/TcArg10/TcLeu254; ΤeAla130–TcArg10; ΤeMet132–TcArg10/TcThr257/TcLeu258; and ΤeLys133–TcAsp6/ TcIle7/TcArg10/TcLeu258/TcSer261. In the ground state MF1 (thin broken lines) structure presentation, the shown segments are equivalent in sequence to those in the TF1 structure, and colour-coded in the same way as for the TF1 structure presentation. The side chains of Tb1Glu391, Ta2Asp401 and TcArg85 are shown, but those in the equivalent residues in the MF1 ground state structures are shown differently, in broken lines. (B) The same view as in Fig. 5(A) of the TF1 (thick lines) and EF1 (thin broken lines, with the same colour coding) structures, when superimposed in the same way as for the TF1 and the ground state MF1 comparison above. The TF1 presentation is the same as that in Fig. 5(A), except in that Ca atoms of residues that do not contribute to the ‘conserved’ interaction are shown in grey. In the EF1 presentation, for clarity, only Ca atoms of residues that contribute to the ‘conserved’ interaction are shown, in red but labelled in cyan. The shown segments in the EF1 structure are equivalent in sequence to those in the TF1 structure. Shown EF1 interacting residues (contributing to the conserved interactions) are: ΕeSer108–Eb1Glu381; ΕeAsp111–EcArg84; ΕeAsp113/Ea2Ser411; ΕeTyr114–EcArg84/ EcGly85; ΕeAla117–EcMet23; ΕeAla119–Ea1Phe406/Eb1Asp372; ΕeGlu120–EcIle19; ΕeLeu121–EcIle19/EcThr20; ΕeAla124–EcSer12/EcThr16; ΕeIle125–Εb3Asp372/Εb3Ile376; ΕeLeu128–Εb3Ile372; ΕeArg129–Εb3Ile372/EcLys9; ΕeVal130–Εb3Ile372/EcLys9/EcLeu256; ΕeLeu133– EcLys9; and ΕeThr134–EcLeu260. The Ca positions of the source and target residues, identified by chemical labelling, cross-linking or FRET, are indicated by a black dot (TF1) or a blue dot (EF1), however, when the position of those dots coincides with red or grey Ca dots (based on structure), they are not shown, but instead are given red labels. Equivalent residues in other structures are also shown. In the case of TeCys134, for which coordinates are unavailable, TeLys133 is marked as the nearest-neighbour marker. (C) The same view as in Fig. 5(A) of TF1 (thick lines), MF1–IF1 (broken lines) and ground state MF1 (backbones only, thin lines) structures, when the MF1–IF1 structure is superimposed on to the TF1 structure in the same way as for the TF1/ground state MF1 comparison above. In the backbone presentation of TF1, MF1–IF1 and the ground state MF, the shown segments [either end of each segment is labelled with a residue number with an attached prefix t (in magenta), i (cyan) or g (green), respectively] are all equivalent in sequence, and colour-coded in the same way as in (A), except for the backbone of IF1, which is in orange–red. The TF1 presentation is the same as in (A), except that Ca atoms are shown only for residues belonging to the inhibition-general core region (red for the conserved interactions, grey otherwise; but both are labelled in cyan). In the MF1–IF1 presentation, Ca atoms are shown only for residues belonging to the inhibition-general core region (same cololuring as in the TF1 presentation; labels in black). Shown interacting residues are given in Table 4.

b1Glu391 is equivalent to Mb1Glu395, which interacts with Ma2Asp409 via McArg75 in the ground state MF1 structure (Fig. 5A) [7]. This McArg75-bridging interaction was observed initially in the dicyclohexylcarbodiimide-inhibited MF1 structure [31]. In the TF1 structure, the equivalent interaction between b1Glu391 and a2Asp401 (via cArg85) is lost, which appears to be due to conformational changes in each of the involved regions, but the greatest contribution is likely to come from a large displacement of b1Glu391, due to insertion of the TF1 e-helix 2 (Fig. 5A). This loss is likely to be one of the e-inhibition-specific structural features. It is noteworthy that, compared with the DELSEED motif, which has been undergone extensive biochemical studies, the current structure suggests that the three C-terminal turns of the precedent helix (helix 1) are involved more in the e-inhibition. The a-target areas (394–402, in a loop), in which four a1- and two a2-target residues are included, are part of the second catch region of the a-subunit (394– 402, M402–410, E405–413). The a second catch region FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS

is defined in the same manner as the b second catch region. They are almost equivalent in sequence to the b-target area. The loop contains part of a well-conserved SDLDxAT motif (400–406, M408–414, E411– 417), which is defined here and is in an equivalent position to the b-DELSEED motif in the sequence. a2Asp401 (Fig. 5A), described above as being involved in the cArg85-bridging interaction, is included in the 2 in the a1motif. The target contact areas are 260 A 2 in the a2-subunit. subunit and 70 A The c-subunit is contacted: (a) at the upper parts of both the N- and C-terminal helices by the e-hook and an upper half of the e-helix 2; and (b) at three segments in the c Rossmann fold and a lower part of the C-terminal helix (containing cArg240) by a lower half of the e-helix 2 and an upper end of e-helix 1. Only one c Rossmann fold segment shows interpretable positional differences in comparison with other F1 structures, in contrast to two other segments (one including Tc142 and the other Tc150), which are omitted in Fig. 5. The segment includes cArg85, described 2903


above, and is in loop 1-b (linking strand 1 and helix b, or so-called the radial helix) [5,31]. cArg240 is bridged to this cArg85 and this part of the C-terminal helix behaves as though it is linked to loop 1-b. Contact 2, by areas between the c- and e-subunits span 840 A far the largest among the target subunits. From these structural features, it can be argued that, for inhibition, the TF1 e-subunit: (a) stabilizes the unusual inhibition-specific b-subunit conformation (b1, open; b3, close) by controlling the second catch region conformations of the a- and b-subunits directly, or indirectly through c-subunit conformational control; and (b) contacts heavily the target subunits (c, b3, b1, a1 and a2), thereby preventing rotation of the c-subunit. The latter point is illustrated in Fig. 6. These descriptions are further examined in comparisons with the MF1 ground state, EF1 and MF1–IF1 structures. Overall structural comparison with other F1 molecules The TF1 structure was superimposed on the structures of EF1, MF1–IF1 and the MF1 ground state to see which part of the TF1 structure differs between those structures (as described in Materials and Methods).


The N-terminal and nucleotide-binding domains of the TF1 a- and b-subunits are similar to their counter for the a-carparts, with average rmsd values of 1.2 A bon positions. The structures are most different in the C-terminal domains of the b1-subunit. This is illustrated in Fig. 6, which shows part of the superimposed C-terminal domains of the b- and a-subunits. The C-terminal domain of the TF1 b1-subunit differs most from that in the ground state MF1 structure with the largest super This reflects the open/ imposition rmsd value of 13.2 A. close conformation difference. Compared with its counterpart in the MF1–IF1 structure, the TF1 b1 C-terminal domain is less different, with an rmsd value of 10.0 A. This is because IF1 significantly distorted the closed conformation of the bDP- (b1) subunit in the MF1–IF1 structure. The TF1 b1 C-terminal domain differs least from that in the EF1 structure, with an rmsd value of This reflects the open/half close conformation 4.7 A. difference. The C-terminal domains of the b2- and b3subunits do not differ as much as those in the b1-sub The unit, with much smaller rmsd values of ~ 1.4 A. C-terminal domains of the three a-subunits differ more than the C-terminal domains of the b2- and b3-subunits, as clearly seen in Fig. 6, giving an average rmsd but with small deviations. value of 3.4 A,

Fig. 6. Comparison between the TF1, EF1, MF1–IF1 and MF1 ground state structures, of the C-terminal domains of the a- and b-subunits and part of the stalk subunits located at similar heights to those of the a and b C-terminal domains. The figure is in stereo. The a1-, b2- and a2-subunits are not shown in full because of space restrictions. The b1-subunits are shown, in dark green and thick lines for TF1, in green and less thick lines for EF1, in light green and thin lines for MF1–IF1 and in yellow–green and thinner lines for MF1 ground state. Other b(yellow tones), a- (magenta tones) subunits, c- [blue tones; the N-terminal (c-N), C-terminal (c-C) and radial (c-R) helices] and e- (in blue tones) subunits are shown in this way, in colours from dark to light tones, and in linewidths from thick to thin, in the order of TF1, EF1, MF1–IF1 and MF1 ground state structures. The second catch regions for the a- and b-subunits are highlighted with broken lines. They are labelled, but only to the TF1 subunits, with ‘C’ in the colour of their chains (except TF1 b1-subunit which is labelled ‘Ct’). In addition, the second catch regions of the other b1-subunits are labelled similarly (EF1 labelled ‘Ce’, MF1–IF1 labelled ‘Ci’ and the ground state MF1 labelled ‘Cg’). The threefold axes of the all the structures, derived from the three a-subunits in each structure, are shown at centre of the figure with a position label of ‘30 , in colurs that are the same as those for the c-subunits, that is, navy blue (TF1), cyan (EF1), dark violet (MF1–IF1) and violet–red (MF1). The view is toward the Fo side.

2904



The TF1 c-subunit structure also differs from the (MF1– (EF1), 5.4 A others. The rmsd values are 3.3 A IF1) and 5.4 A (MF1 ground state) for 285 Ca atoms. These large values are due to positional differences (i.e. different locations of the c-subunit) rather than conformational differences, because, when only the csubunit structures were superimposed the average devi for 180–220 Ca atoms. ation was 1.7 A The same is true for the e-subunit (d-subunit for (EF1), 21.3 A (MF1– MF1). The rmsd values are 8.1 A IF1) and 29.4A (MF1 ground state), whereas the e-subunits barrel structures are superimposed with devia tions of ~ 1.4 A. Taken together, the TF1 structure differs most from the structures in the C-terminal domains of the b1-subunit and three a-subunits, and in the stalk subunits. Statistics for an overall structural comparison are an (for 3117 a-carbon positions) for rmsd value of 2.1 A (2996) for EF1, 2.8 A (3096) for MF1–IF1, and 3.0 A the ground state MF1 structures. When the TF1 structure (Fig. 3A) is compared with the aligned EF1 structure in Fig. 3B, it is seen that areas containing helix 2 (red) and the hook of the e-subunit and the second catches (magenta) of the b1- and b3subunits are very similar between the two. This region is therefore termed an e-inhibition core region, and its details are be described below (Comparison with the EF1 structure). This similarity, characterized by the e-inhibition core region, can be extended to the MF1– IF1 structure (Fig. 3C). The short helix and top half of the long helix of IF1 (lemon) and the second catches (magenta) of the b1- and b3-subunits occupy an area similar to, but smaller than, the e-inhibition core region. This is termed an inhibition-general core region, because it appears to be a common region among the three inhibited structures. It is described in detail below (Comparison with the MF1–IF1 structure). It should be noted that the term ‘inhibition-general’ (i.e. without core) is used to describe a categorical sum of the e-inhibition and the IF1-inhibition. Comparison with the ground state MF1 structure – a comparison with the active state structure When the TF1 and ground state MF1 structures are superimposed using optimum superimposition (see Materials and methods), helix 2 of the TF1 e-subunit is in the position occupied by the second catch of the bDP-subunit (b1; shown in green broken lines) of the MF1 structure (Fig. 5A). This implies that the closed conformation of the tight site b-subunit in the active state is not compatible with the up state conformation of the e-subunit. In other words, for the TF1 e-subunit FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS


to be in the inhibitory ‘up’ conformation, the b1-subunit should be in a conformation other than that of the closed bDP-subunit. In the TF1 structure, that is the open conformation. In Fig. 5A, a large displacement of the b1 second catch (green) during a change from the active state (MF1 ground state; broken lines) to an inhibited state (TF1; continuous line) is clearly shown. A large difference in the N-terminal helix of the c-subunit (Tc6–25) is seen, which is thought to be due to active– inhibited state differences, although the N-terminal helix structure is thought to be changeable among F1 structures [8], in contrast to the C-terminal helix. This point is examined fully in subsequent comparisons with the EF1 and MF1–IF1 structures. There are other smaller conformational differences; the a1 loop (Ta393–403) appears to be pushed by the e helix 2, but the a2 loop (Ta396–401) appears to be pulled in the transition. The c Rossmann fold loop 1-b (Tc83– 89) and the TcArg240-containing segment (Tc239– 241) in the C-terminal helix differ to a similar extent. However, in contrast to the structural parts described above, the b3-subunit shows a much smaller structural difference. The same is true with the C-terminal helix of the c-subunit (Tc254–261). These positional differences in the target segments are summarized in Table 2. Comparison with the EF1 structure and einhibition core region In this comparison with the EF1 structure, the e-inhibition core region mentioned above (Overall structural comparison with other F1 molecules) is examined by highlighting common structural features between the TF1 and EF1 structures. Figure 5B shows the same view, as in Fig. 5A, of the TF1 (thick lines) and the EF1 (thin broken line) structures superimposed using optimum superimposition. Here, EF1 is very similar in structure and position to TF1; it is therefore different from its MF1 ground state counterpart (Fig. 5A), reflecting that they are both in the e-inhibited state. Because helix 2 of the EF1 e-subunit sits in a very similar position to that of TF1, the b1-subunit is again expected to have a conformation other than that of the closed bDP-subunit. In the EF1 structure, that is the half-closed conformation [12]. In the superimposition, the upper half of helix 2 (Te117–123, Ee118–124) and the start of the hook (Te124–126, Ee119–127) overlap well (average of a-carbon separations), although the two were 2.0 A fitted using the overall structures with the optimum superimposition method.

2905



In the superimpostion, the b1-subunits differ between the two structures because of differences in the conformation type, as described above and to a less extent, so do the a1-subunits. These b1- and a1-subunits do not contribute very much to the e-inhibition core region because of these conformational differences. Most of the rest of the structure in Fig. 5B fits well in the superimposition; these are all of the c-egments and the b3- and a2-subunits. Combined with results from the superipmposition of TF1 and the ground state MF1 structures, it appears that the N-terminal helix, the c Rossmann fold segment, loop 1-b, the TcArg240-containing segment (lower part of the C-terminal helix) and the a2-subunit segment are parts that change during e-inhibition. The EF1 segment equivalent to the TcArg240-containing segment is not contacted by the EF1 e-subunit, suggesting that the position of that segment in the EF1 structure is determined passively. The c C-terminal helix and the b3segment are unchanged during e-inhibition. Those positional differences of the target segments are summarized in Table 2. Many of the interactions of the e-helix 2 and e-hook with target subunits in the TF1 structure have EF1 counterparts (equivalent-in-sequence source target pair), and are therefore termed ‘conserved’ interactions. The conserved interactions defined here use the equivalent-in-sequence residue pairs, not necessarily equivalent atoms. As clearly shown in Fig. 5B, the conserved source residues cover an entire region consisting of the e-helix 2 and hook. Nine residues in ehelix 2 of TF1 contribute to the conserved interactions, whereas six residues in the hook do. The e-inhibition

core region may be defined as a region containing ehelix 2 (Te107–123, Ee108–124), the hook (Te124–133, Ee125–134) and the targets. It should be noted that the e-inhibition core region (Te107–133, Ee108–134) is much more extensive than the stretch of well-fitted residues (Te117–126, Ee118–127) in the superimposition. The conformations and positions of rest of the e-subunit structures, omitted in Fig. 5B (Te99–106), differ between the two structures, leading to no conserved interactions (Table 3), as also expected from comparison of the e-subunit structures (Fig. 4). A list of the source and target residues involved in the conserved interactions is given in Table 3, but this is only for residues that have been described in biochemical studies or are in the characterized motifs. The list is short, giving the expectation that a full residue analysis on these findings will be done. The major targets of the conserved interactions are the c-subunit (total 15 interactions; 10 for the N-terminal helix, 3 for loop 1-b, and 2 for the C-terminal helix), and the b3 helix 1 (6 interactions). The N-terminal helix of the c-subunit and c loop 1-b are the parts that change in e-inhibition. The b3 helix 1 is unchanged in e-inhibition. Other minor targets are the b1-subunit (3 interactions), and the a1- and a2-subunits (one interaction each). The b1- and a1- target regions are less fitted between the TF1 and EF1 structures than those for the major targets just described, although the a2-subunit target region is fitted (Table 2). A conserved interaction, ΤeGln107–Tb1Glu391 and ΕeSer108–Eb1Glu381, is unusual because the largely separated source (ΤeGln107 and ΕeSer108) and target (Tb1Glu391 and Eb1Glu381) residues make equivalent

Table 2. Relative positions of the target segments - inhibition-specific positions. Segments (Unchanging group) b3 c C-terminal helix (changing group) c N-terminal helix c Rossmann fold loop 1-b Τc Arg240 segment a2 b1 a1

ΤF1

EF1

MF1/IF1

MF1

Source residue’s location in TF1and comments

M M

M M

M M

M M

Upper helix 2 and hook Hook end

T T

T T

T/M T

M M

Upper helix 2 and hook Lower helix 2

T T T T

T T E E

T T/M M M

M M M M

Middle helix 2 Lower helix 2 Lower helix 2 Entire helix 2

Positions of the target segments in Fig. 5 are classified. M stands for a case in which the segment position is similar to or the same as that of the segment in the ground state MF1 structure. This is the base in this comparison. T stands for a case in which its segment position is not the M type, but is similar to or the same as that of the segment in the TF1 structure. E stands for a case in which the segment position is not M or T type, but is similar to or the same as that of the segment in the EF1 structure. T/M stands for a case inw hcih the segment position is not M, E or T type, but is in between T and M.

2906




Table 3. Some of the conserved interactions in the e-inhibition core area in TF1 and EF1. TF1 source

TF1 target

EF1 source

EF1 target

Comments on target and others

ΤeGln107

Tb1Glu391

ΕeSer108

Eb1Glu381

ΤeAsp110 ΤeAsp112 ΤePhe113

TcArg85 Ta2Ser400 TcArg85

ΕeAsp111 ΕeAsp113 EeTyr114

EcArg84 Ea2Ser411 EcArg84c

Backbone conformations different. In b DFELSEED motif. ΕbGlu381 cross-linked to EeSer108. TbGlu391 affects catalysis and inhibition. Equivalent to MbGlu395 which is bridged to Ma2Asp409 by McArg75. c Rossmann fold loop 1-b a2 SDLDxAT motif c Rossmann fold loop 1-b

contacts (Fig. 5B) The rest of the conserved interactions have their residues in proximity. Separation of the target residues is due to differences in the ‘open’ and ‘half-closed’ conformations of the b1-subunits. Separation of the source residues is due to the large conformational difference between the structures of the two e-subunits (Fig. 4). As mentioned previously, the target TbGlu391 residue affects catalysis [30] and also inhibition [32], and is equivalent to MbGlu395, which is bridged to Ma2Asp409 by McArg75 [31]. Taking all these lines of evidence together, this conserved interaction is likely to play a role in e-inhibition. It is interesting to note that the ΕeSer108– Eb1Glu381 interaction is evidenced by a cross-linking study [12,22], which is described below. A number of source e-residues have been identified, with their targets, by chemical labelling, cross-linking or fluorescence resonance energy transfer. They are EeSer108–ΕbGlu381 [22], EeSer108–EaSer411 [23] and EeSer118–EcLeu99 [24] in EF1, and TeCys134–TcSer3 in TF1 [21]. Most of the residues are mapped in Fig. 5B, together with their equivalent residues in counterpart structures. TeGln107–Tb1Glu391 is rela tively close with an a-carbon separation of 9.4 A, implying that the TF1 local conformation, described above, is consistent with the EF1 biochemical data for EeSer108–Εb1Glu381 [12]). For TeCys134– (7.3 A TcSer3, because TeCys134 was not observed, a value for TeLys133–TcSer3 may give a hint to the of 8.3 A for EeThr134–EcSer2). Again, the separation (15.5 A TF1 structure appears to be consistent with the TF1 biochemical data. Other separation values are 12.9 A for TeSer107–Ta2Ser400 (11.9 A for EeSer108– for TeVal117–TeAla100 Ea1Ser411) and 21.0 A (15.4 A for EeSer118–EeAla99). Although the former separations might be consistent with the EF1 biochemical data to a similar extent between the TF1 and EF1 structures, the latter are inconsistent with the EF1 biochemical data, suggesting that the data reflect a different state in the F1 catalytic cycle.


Comparison with the MF1–IF1 structure and inhibition-general core region A simple look at the e-inhibited TF1 and IF1-inhibited MF1–IF1 structures gives the impression that they are unrelated. However, superimposition of the two structures shows that parts of the inhibitory proteins (TF1 e-subunit and IF1) and their targets are closely located in the two. This leads to an examination of the two structures for the inhibition-general core region, with available additional information from comparison of the TF1 and EF1 structures. As shown in Fig. 5C, in the MF1–IF1 structure, the b1 position is occupied by a closed b-subunit. However, the inserted IF1 long helix appears to pull the second catch region of the b1-subunit and to push that of the neighbouring a1subunit (only a1:401–404 and a1:410–411 are shown). This makes the catalytic interface of the bDP-subunit (b1) look like that of bTP-subunit (b3), leading to loss of the structure and properties of the catalytic site interface of the tight site b-subunit [14]. In the TF1 structure with the open b1-subunit, such a closed catalytic site interface cannot be formed in that position, thus the e-inhibition shares with the IF1 inhibition a loss of the properties of the tight site catalytic interface. Only the b3-segment, the c C-terminal helix segment, the c Rossmann fold loop 1-b and the Tc240 containing segment fit well in the superimposition of the TF1 and MF1–IF1 structures (Fig. 5C and Table 2). The b3-segment and the c C-terminal helix segment are thought to be unchanged in the inhibition-general. By contrast, the c Rossmann fold loop 1-b and Tc240-containing segment (see Table 2 for a note on the segment) are the parts that change in the inhibition-general (Table 2). The c N-terminal helix segment and the a2-segment of the MF1–IF1 structure do not fit to the TF1 counterpart (and therefore not to the EF1 counterpart either) and are positioned in between the ground state MF1 and TF1 counterparts (Table 2).

2907



Figure 5C clearly shows that in the superimposition, part of IF1, I22–32 in the long helix and I14–18 in the short helix, are close to part of the TF1 e-subunit (Te118–128, an upper half of e-helix 2 and a lower part of the hook). This region, easily recognized by noting Te118 and Te128, is the inhibition-general core region. It is almost identical to the well-fitted region (Te117–126) in the TF1 and EF1 superimposition, further supporting the idea of the inhibition-general core region. To characterize the inhibition-general core region, TF1 target residues were examined for an equivalent MF1 residue that is: (a) also a target of the IF1; and (b) located close to the TF1 target residue. This led to a correlation of seven source residues in the TF1 e-subunit and in IF1 (Table 4). Most of the listed TF1 target residues contribute to the ‘conserved’ interactions in the TF1/EF1 comparison; therefore, the correlation is generally endorsed by the EF1 structure. The correspondence between the source residues in the e-subunit and IF1 (Table 4) may be somewhat obscure because of the unrelated sequences and structures. More important may be the presence and properties of the common targets. The inhibition-general core targets are in the b3 helix 1 (three contact occurrences in Table 4), the c N-terminal helix (two contact occurrences) and the a1 loop between helices 1 and 2 (two contact occurrences). The b3 helix 1, which is unchanged in the e-inhibition and has the heaviest econtacts among the a- and b-second catches, shows similar properties in the inhibition-general. In the c Nterminal helix, the portions involved in the inhibitiongeneral core (ΤcAla13, TcThr17 and their MF1–IF1 counterparts) are in similar positions in the TF1 and MF1–IF1 structures, thus leading to inhibition-general involvement, although, as described, the c N-terminal helices as a whole are not in similar positions (Table 2). The a1 loops occupy different positions in

the two structures (Table 2), but the interactions are conserved. This feature is reminiscent of conserved interaction, despite the much changed residue positions, in the TF1 and EF1 structures (ΤeGln107– Tb1Glu391; ΕeSer108–Eb1Glu381).

Discussion Formation of the inhibited complex Formation of the TF1 inhibited complex should start with ATP depletion (a suggested threshold ATP concentration is 700 lM [32]). ATP depletion influences TF1 in two ways. First by changing the TF1 e-conformation to the ‘up’ conformation from the ‘down’ [21]. Second, by changing the nucleotide occupancy and conformation of the b- and a-subunits during the consumption of bound ATP. A starting state is the one with two nucleotide-bound b-subunits and with the e-subunit in the ‘down’ state, as in ground state MF1. When the ATP concentration becomes low, nucleotides bound to the a-subunits detach first [25]. The ATP molecule on the b1-subunit (the tight site b1-subunit) is then converted to ADP, and ADP is subsequently released. The b1-subunit with no bound nucleotide is likely to change its conformation to the open form. The resultant structure may have an a3b3 portion similar to that of the current TF1 structure, with an open b1–a2 interface and a less open b1–a1 interface (Fig. 6). While such structural changes occur in the a3b3 portion, the e-subunit may tend to change its structure because of low ATP concentrations. Initially, helices 1 (90–104) and 2 (113–130) may get apart, as a bridging ATP detaches. Before ATP is released from the e-subunit, the down conformation must be stable with the two domains and the linker loop together forming a compact rigid body via ATP bridging [19]. On ATP release, loss of ATP-mediated interactions is likely to lead to: (a) detachment, from

Table 4. TF1 and MF1–IF1 residues in the inhibition- general core area. TF1 target

TF1 source

MF1 target

IF1 source

TF1/EF1

Target site and comments

Tb3Asp382 Tb3Ile386 Tb3Leu387 Ta1Ala394

TeSer128 TeMet124 TeMet124 TeLeu118

Mb3Asp386 Mb3Ile390 Mb3Leu391 Ma1Ala402

I22Phe I22Phe I22Phe, I25Arg I28Ala

Conserved Conserved No No

Ta1Phe395 ΤcAla13 TcThr17

TeLeu118 TeAla123 TeAla123

Ma1Phe403 ΜcSer12 McIle16

I28Ala, I29Glu, I32Arg I14Ala, I15Val, I18Ala IPhe22

Conserved Conserved Conserved

b3 helix 1 b3 helix 1 b3 helix 1 a1 loop between helices 1 and 2. Ta1Ala394 equivalent in sequence to TbAsp382. a loop between helices 1 and 2 c N-terminal helix c N-terminal helix

The TF1 e-subunit residues can be correlated to the IF1 counterparts. The correspondences are: TeSer128 to I22Phe; TeMet124 to I22Phe and I25Arg; TeAla123 to I14Ala, I15Val, I18Ala and I22Phe; TeLeu118 to I28Ala, I29Gly and I32Arg.

2908



the N-terminal domain, of the interdomain loop and the whole C-terminal domain; (b) rearrangement of the interdomain loop; and (c) separation of helices 1 and 2 and their shortening from the ATP-bound state (helix 1 90–107 from 90–104, helix 2 110–127 from 113–130). These structural changes may be responsible for the increased mobility of the linker loop and the whole C-terminal domain in the ATP-free state [19]. In their new positions, the connecting loop and helix 1 are stabilized by interaction with the e N-terminal domain, resulting in further conformational changes that give new positions for the connecting loop (86– 90) and helix 1 (91–105). These structural changes of the e-subunit from the ‘down’ state, combined with the changes in the a3b3 structure described above, allow e-helix 2 to push into the core of the a3b3 portion via the more open b1a2 interface. Subsequently, after some conformational adjustment by interactions with other target subunits, the e-inhibition core region settles in its final position, as in the current TF1 structure, with the implication that it works as a brake impeding rotation of the c-subunit. Mechanism of e-inhibition The structural comparison described so far provides clues to how bacterial F1 is inhibited by the e-subunit. The e-subunits in the structures of both TF1 and EF1 occupy the position of the b1 second catch region in the superimposed ground state MF1 structure (Fig. 5A,B), thus they prevent the b1-subunit from adopting the tight-site-specific closed conformation (bDP’s). This is thought to be the essential feature of the e-inhibition. Loss of the tight-site-specific conformation, accompanied by loss of the tight-site-specific catalytic site interface in the b1-subunit is seen in the MF1–IF1 [14], EF1 [12] and TF1 structures. In e-inhibited F1 structures, the b1-subunits can be in either the half-closed conformation, as in the EF1 structure, or in the open conformation, as in the TF1 structure. The e-inhibition core region (e-helix 2 and hook) contacts the target subunits in a very conserved manner between the EF1 and TF1 structures (conserved interactions), appearing to result in stabilization of an e-inhibition-specific conformation of the target subunits. In the b- and a-subunits, each second catch in the b3, a2, b1 and a1-subunits is in a specific position; either unchanged (b3, a2) or changed (b1, a1) depending on the subunit (Table 2), thus controlling each subunit’s overall conformation. In the c-subunit, each of the target parts also has its specific position. The N-terminal helix and Rossmann FEBS Journal 282 (2015) 2895–2913 ª 2015 FEBS


fold 1-b loop are each in e-inhibition-specific positions that are different from those of the counterpart in the ground state MF1 structure (Table 2). The c N-terminal helix was contacted by the second catch of the MbDP-subunit in the ground state MF1 structure, but in the e-inhibition, it is contacted by the e-subunit (Fig. 5A). This e-contact is only feasible with a new position of the c N-terminal helix, which is e-inhibition specific. The Rossmann fold 1-b loop, before e-inhibition, should have bridged b1Glu391 and a2Asp401 (in the DELSEED and SDLDxAT regions, respectively), thus keeping the second catches in positions favoured by the active state, as in the ground state MF1 structure [7,31]. But in e-inhibition, because there is no more bridging, its loss would help the second catches of the b1- and a2-subunits adopt their e-inhibition-specific conformations. This change in the c-structure appears to help control the b1-DELSEED and a2-SDLDxAT conformations. By contrast to those changing target parts, the c Cterminal helix is unchanged. This is like the b3 second catch region described above. In summary, the e-inhibition core region appears to make the conformations of the a- and b- second catch regions specific to e-inhibition, either via direct contacts or indirectly through contacts with the c-subunit. It also appears to prevent rotation of the c-subunit through many interactions with the target subunits. This view is supported further by the structural comparison with MF1–IF1 described above. The e-inhibition seems compatible with the open or half-closed conformations of the b1-subunit. Despite the ‘open’ and ‘half-closed’ difference in the b1-subunit conformation, the relative positions of some parts of the b1-, e- and a2-segments are rather similar between the TF1 and EF1 structures, as shown in Fig. 5B. This may lead to observation of a conserved interaction (ΤeGln107–Tb1Glu391/ΕeSer108– Eb1Glu381), accompanied by close placement of the e-source residues induced by e-backbone structural changes (Table 3). The similarity in the conformations of the e C-terminal domains between the TF1 and EF1 structures is limited to the e-inhibition core region. Below it, the backbones run quite differently, and therefore are accompanied by very different interactions with the targets. In the EF1 structure, helix 1 and its flanking loops form strong contacts with the c-subunit, with a 2. By contrast, the equivalent contact area of 830 A region in the TF1 structure has only weak contacts 2). Because this is the single major significant (100 A difference in the e structures between TF1 and EF1, it 2909



might be related to the major difference in the e-inhibition properties between TF1 and EF1, for example, by preventing c-subunit rotation to different degrees. In TF1, inhibition is apparent only at low concentrations of Mg-ATP (below 50 lM) [16,32]. By contrast, in EF1, inhibition is apparent at any concentration of Mg-ATP [33–35]. Finally, there is the question of why the synthesis reaction takes place in the TF1 up state [21]. We speculate that the TF1 structure gives a hint and this shown in Fig. 6, which includes a top view of the TF1 lower part structure. When a small degree (say 15°) of clockwise rotation (ATPase reaction) of the stalk about the indicated threefold axis is imagined, it can be clearly seen that the e-subunit will collide the b3-subunit, and to a less extent, the a1-subunit. Then, when a similarly small degree of counter clockwise rotation (ATP synthesis reaction) of the stalk about the threefold axis is imagined, the e-subunit will have fewer blockers, if any. This is because the b1- and a2subunits are clearly off the e-subunit’s rotation course (note that TF1 e-subunit is shown in dark red). Thus, counter-clockwise rotation appears to be favoured. However, our speculation assumes that there are no major structural changes in the C-terminal domains of the b- and a-subunits during this small angle rotation of the stalk and therefore it has some limitations.

firmed by measuring the bound nucleotide [37] for the TF1 molecules contained in crystals. Crystals of TF1 were flash-cooled with nitrogen gas at resolu100 K after Paratone treatment [38]. Data to 4.2 A tion were collected from a single crystal on BL41XU at SPring8, Sayou, Japan. Crystals were tetragonal, space group I4122, with cell dimensions of a = b = 233.3 A, c = 305.3 A. An asymmetric unit contained one TF1 molecule. The diffraction data were integrated with HKL2000 [39] and processed further with the CCP4 [40] program suite (Table 1). Crystals of SeMet TF1 were flash-cooled after harvesting and kept for 2 h in a cryo-buffer containing 32.5% poly (ethylene glycol) 6000, 0.20 M sodium chloride, 0.05 M NaPipes buffer (pH 7.0), 2 mM dithiothreitol, 5 mM CyDTA and 10% (v/v) ethylene glycol treatment. An increase in the poly(ethylene glycol) 6000 concentration in the cryo-buffer to 32.5% was optimal to improve the crystal resolution among various poly(ethylene glycol) concentrations tried resolution were collected (18–42%). MAD data to 3.9 A from a single crystal on BL6A at the Photon Factory (Tsukuba, Japan) at wavelengths of 0.97879 (peak), 0.97926 (edge), 0.994 (remote 1) and 0.96403 (remote 2). It was essential that the weak beam at the beam line was used to minimize radiation damage of the very radiation-sensitive SeMet TF1 crystals. The diffraction data sets were integrated with MOSFLM [41], scaled together over wavelengths with SCALA [42], and processed further with the CCP4 program suite (Table 1).

Materials and methods Molecular replacement Crystallization and X-ray data collection TF1 (a3b3ce sub-complex), with b-subunits having a 10-residue histidine tag at the N terminus, was expressed in FoF1deficient E. coli cells and was purified with Ni-NTA, DEAE Toyopearl, Phenyl–Toyopearl columns and with an additional heat treatment, using the E. coli over-expression system [26]. Its SeMet-substituted TF1 was expressed and purified in the same way as TF1, except that E. coli methionine auxotrophic B834(DE3) cells were used in the expression and M9 medium supplemented with SeMet and 19 other amino acids [36]. Crystals of TF1 and SeMet-substituted TF1 were grown at 25 °C using the sitting drop technique. A 10 lL drop contained 9–12.5% poly(ethylene glycol) 6000, 0.20 M sodium chloride, 0.05 M Tris-sulfate buffer (pH 8.0), 2 mM dithiothreitol, 5 mM trans-1,2-diaminocyclohexane-N,N,N’,N’-tetraacetic acid, monohydrate (CyDTA), 10% (v/v) ethylene glycol and 10 mgmL1 protein, and a 1 mL reservoir contained 14–16% poly(ethylene glycol) 6000, 0.2 M sodium chloride, 0.05 M Tris-sulfate buffer (pH 8.0), 5 mM CyDTA and 10% (v/v) ethylene glycol. Bipyramidal crystals (0.4 9 0.4 9 0.2 mm) grew in 2 weeks. The 1 : 1 ADP binding stoichiometry was con-

2910

In a molecular replacement calculation using GLRF package [43], a cross-rotation function and translation search using the TF1 a3b3 structure [27] (with three b-subunits in the open conformation) gave a single significant solution (Table 1). In the positional rigid body refinement with XPLOR-3.851 [44], an overall B-factor (54.8) was added initially to the model coordinates to compensate for the difference in resolution between the model and the data. In stages of refinement where either one of b-subunits was changed into the closed conformation or new structural elements such as c- and e-subunit structures were incorporated, the initial model was treated as a rigid entity, but was then broken into smaller units such as subunits, domains and large secondary elements. This was accompanied by fine stepwise increases in the resolution range from to 15–4.2 A to achieve a large radius of conver15–8 A gence [45]. An initial map, calculated after the starting TF1 a3b3 structure [27] model had been located correctly (R = 0.542), showed that one of the b-subunits is in the closed conformation, and the model was changed accordingly and refined (R = 0.433). The c-subunit model was then built using the F1-dicyclohexylcarbodiimide structure



[31] and refined (R = 0.428). The b-sandwich domain of the e-subunit was built using the E. coli c/e-complex structure [46]. There remained only a clear rod-like density between the end tip of the e b-sandwich and the a3b3 barrel. The two C-terminal helices derived from the TF1-dicyclohexylcarbodiimide structure were placed in the density map, and the refinement was continued (R = 0.415). ADP bound to the b3-subunit was built into the density map. Finally, overall B-factor refinement, for domains or large secondary structure elements using CNS 1.21 [47], reduced an R-factor to 0.377 (Table 1).

MAD analysis of the SeMet TF1 crystal MAD data were analysed in SHARP [48]. Seventy-nine expected Se positions from the molecular replacement model were incorporated into the program and refined. Seventy-seven Se positions were refined. The resultant MAD map gave good density for the backbone, but not for the side chains (Fig. 1, in blue contours).

Refinement and final structural model Refinement was done using CNS 1.21 [47] and PHENIX-REFINE v. 760 [49], as summarized in Table 1. A weighting factor between X-ray and geometry terms was carefully set, otherwise the main chain geometry deteriorated easily despite improvement in both the R-work and R-free factors and the geometry terms such as a bond length and a bond angle. A Ramachandran restraint in PHENIX-REFINE was useful to keep the main chain geometry satisfactory. A combination of model and MAD phases was tried in the refinement at one stage, but it was found to give slightly worse R-free, R-work and Ramachandran favoured values by 0.001, 0.004 and 0.2%, respectively, than those in the model only refinement, and therefore was not adopted. 2Fo – Fc and omit maps, both with or without B-factor sharpening (–95) [47] were usually used to check the model. Occasionally, MAD and MAD phase combined maps were consulted. Evaluation of the model was done using POLYGON [50], with results where most of the refinement quality criteria were better than the average. The secondary structure elements were identified with an aid of PROCHECK [51]. Buried surface areas were calculated using PISA [52]. Pictures were drawn with BOBSCRIPT [53] (Fig. 1) and MOLSCRIPT [54] (others).

Structural comparison The structural comparison was carried out using O [55]. To compare a pair of the F1 structures, alignment was done initially using the P-loop motif of the b2-subunit and then refined with all closely located a-carbon positions (optimum superimposition). This was done using lsq_exp and



lsq_imp functions, respectively, in O. The lsq_imp function in O gives an rmsd value for closely positioned a-carbons only. For an rmsd value for all a-carbon pairs, a local program was used. In comparisons of, for example, the TF1 and EF1 structures, the local program gave an rmsd value for 3117 a-carbon positions, whereas lsq_imp gave of 2.1 A for 2737 close a-carbon pairs. Structure an rmsd of 1.6 A comparsion using the c-core [12], which may be useful for determining the F1 rotational position, was not used here. When the TF1 and EF1 structures were compared using the c-core, the final rmsd value for all the a-carbons was much larger than that in the optimum superimposi4.5 A, tion. In this superimposition, although a-carbon positional differences for the c-subunit are small, those for the a- and b-subunits are very large (by about three times). This resulted in a 5° deviation of the threefold axes of F1 between the two structures, which is physically unlikely. The results of traditional fitting using the barrel domain [28] are very similar to those of the optimum superimposition, in statistics and with no deviation of the threefold axes.

Acknowledgements We thank Drs Andrew G.W. Leslie, Liang Tong, Eiro Muneyuki and Atsushi Nakagawa for helpful suggestions. We thank the beamline staff at SP-ring 8 (BL44XU, BL41XU) and Photon Factory (BL6A) for help in data collection. A financial support was provided by MEXT-supported Program for the Strategic Research Foundation at Private Universities, 2011– 2016.

Author contributions YS, TS and MY designed the project, supervised research and wrote the article. AS, HT and TS prepared and crystallized protein. AS prepared frozen crystals. YS, AS and MN collected diffraction data. YS analysed the data. HT prepared paper figures.

References 1 Boyer PD (1997) The ATP synthase - a splendid molecular machine. Annu Rev Biochem 66, 717–749. 2 Weber J & Senior AE (1997) Catalytic mechanism of F1-ATPase. Biochim Biophys Acta 1319, 19–58. 3 Yoshida M, Muneyuki E & Hisabori T (2001) ATP synthase-a marvelous rotary engine of the cell. Nat Rev Mol Cell Biol 2, 669–677. 4 Yoshida M, Sone N, Hirata H & Kagawa Y (1975) A highly stable adenosine triphosphatase from a thermophilic bacterium. J Biol Chem 250, 7910–7916.

2911



5 Abrahams JP, Leslie AGW, Lutter R & Walker JE resolution of F1–ATPase from (1994) Structure at 2.8 A bovine heart mitochondria. Nature 370, 621–628. 6 Noji H, Yasuda R, Yoshida M, Kinoshita KJ, Noji H, Yasuda R, Yoshida M & Kinoshita KJ (1997) Nature 386, 299–302. 7 Bowler MW, Montgomery MG, Leslie AGW & Walker JE (2007) Ground state structure of F1–ATPase from resolution. J Biol bovine heart mitochondria at 1.9 A Chem 282, 14238–14242. 8 Kabaleeswaran V, Puri N, Walker JE, Leslie AGW & Mueller DM (2006) Novel features of the rotary catalytic mechanism revealed in the structure of yeast F1 ATPase. EMBO J 25, 5433–5442. 9 Dautant A, Velours J & Giraud MF (2010) Crystal structure of the MgADP-inhibited state of the yeast F1c10–ATP synthase. J Biol Chem 285, 29502–29510. 10 Hausrath AC, Capaldi RA & Matthews BW (2001) The conformation of the e- and c–subunits within the Escherichia coli F1 ATPase. J Biol Chem 276, 47227–47232. 11 Menz RI, Walker JE & Leslie AGW (2001) Crystal structure of bovine mitochondrial F1ATPase with nucleotide bound to all three catalytic sites: implications for the mechanism of rotary catalysis. Cell 106, 331–341. 12 Cingolanil G & Duncan TM (2011) Structure of the ATP synthase catalytic complex (F1) from Escherichia coli in an autoinhibited conformation. Nat Struct Mol Biol 18, 701–708. 13 Stocker A, Keis S, Vonck J, Cook GM & Dimroth R (2007) The structural basis for unidirectional rotation of thermoalkaliphilic F1-ATPase. Structure 15, 904–914. 14 Gledhill JR, Montgomery MG, Leslie AGW & Walker JE (2007) How the regulatory protein, IF1, inhibits F1ATPase from bovine mitochondria. Proc Natl Acad Sci USA 104, 15671–15676. 15 Feniouk BA, Suzuki T & Yoshida M (2006) The role of subunit e in the catalysis and regulation of FoF1-ATP synthase. Biochim Biophys Acta 1757, 326–338. 16 Saita E, Iino R, Suzuki T, Feniouk BA, Kinosita KJ & Yoshida M (2010) Activaion and stiffness of the inhibited states of F1-ATPase probed by single-molecule manipulation. J Biol Chem 285, 11441–11447. 17 Wilkens S, Dahlquist FW, Mcintosh LP, Donaldson LW & Capaldi RA (1995) Structural features of the Escherichia coli e subunit of the ATPsynthase determined by NMR spectroscopy. Nat Struct Biol 2, 961–967. 18 Uhlin U, Cox GB & Guss JM (1997) Crystal structure of the e subunit of the proton-translocating ATP synthase from Escherichia coli. Structure 5, 1219–1230. 19 Yagi H, Kajiwara N, Tanaka H, Tsukihara T, KatoYamada Y, Yoshida M & Akutsu H (2007) Structures of the thermophilic F1-ATPase e subunit suggesting

2912

20

21

22

23

24

25

26

27

28

29

30

31

ATP-regulated arm motion of its C-terminal domain in F1. Proc Natl Acad Sci USA 104, 11233–11238. Kato-Yamada Y, Yoshida M & Hisabori T (2000) Movement of the helical domain of the e subunit is required for the activation of thermophilic F1-ATPase. J Biol Chem 275, 35746–35750. Suzuki T, Murakami T, Iino R, Suzuki J, Ono S, Shirakihara Y & Yoshida M (2003) FoF1-ATPase/ Synthase is geared to the synthesis mode by conformational rearrangement of e subunit in response to proton motive force and ATP/ADP balance. J Biol Chem 278, 46840–46846. Aggeler R, Haughton MA & Capaldi RA (1995) Disulfide bond formation between the COOH-terminal domain of the b-subunits and the c-subunits and esubunits of the Escherichia coli F1-ATPase _ Structural implications and functional consequences. J Biol Chem 270, 9185–9191. Aggeler R & Capaldi RA (1996) Nucleotide-dependent movement of the e subunit between a and b subunits in the Escherichia coli FoF1-type ATPase. J Biol Chem 271, 13888–13891. Tsunoda SP, Rodgers AJW, Aggeler R, Wilce MCJ, Yoshida M & Capaldi RA (2001) Large conformational changes of the e subunit in the bacterial FoF1 ATP synthase provide a ratchet action to regulate this rotary motor enzyme. Proc Natl Acad Sci USA 98, 6560–6564. Yoshida M & Allison WS (1986) Characterization of the catalytic and noncatalytic ADP binding sites of the F1-ATPase from the thermophilic bacteirum, PS3. J Biol Chem 261, 5714–5721. Suzuki T, Suzuki J, Mitome N, Ueno H & Yoshida M (2000) Second stalk of ATP synthase: cross-linking OF c subunit in F1 to truncated Fo b subunit prevents ATP hydrolysis. J Biol Chem 275, 37902–37906. Shirakihara Y, Leslie AGW, Abrahams JP, Walker JE, Ueda T, Sekimoto Y, Kambara M, Saika K, Kagawa Y & Yoshida M (1997) The crystal structure of the nucleotide free a3b3 sub-complex of F1-ATPase from the thermophilic Bacillus PS3 is a symmetric trimer. Structure 5, 825–836. Kabaleeswaran V, Shen H, Symersky JN, Walker JE, Leslie AGW & Mueller DM (2009) Asymmetric structure of the yeast F1 in the Absence of bound nuleotides. J Biol Chem 284, 10546–10551. Hara KY, Noji H, Bald D, Yasuda R, Kinoshita KJ & Yoshida M (2000) The role of the DELSEED motif of the b subunit in rotation of F1-ATPase. J Biol Chem 275, 14260–142633. Hara KY, Kato-Yamada Y, Kikuchi Y, Hisabori T & Yoshida M (2001) The role of the bDELSEED motif of F1-ATPase: propagation of the inhibitory effect of the e subunit. J Biol Chem 276, 23969–23973. Gibbons C, Montgomery MG, Leslie AGW & Walker JE (2000) The structure of the central stalk in bovine



32

33

34

35

36

37

38

39

40

41

42 43

resolution. Nat Struct Biol 7, F1-ATPase at 2.4-A 1055–1061. Kato Y, Matsui T, Tanaka N, Muneyuki E, Hisabori T & Yoshida M (1997) Thermophilic F1-ATPase is activated without dissociation of an endogenous inhibitor, e subunit. J Biol Chem 272, 24906–24912. Laget PP & Smith JB (1979) Inhibitory properties of endogenous subunit e in the Escherichia coli F1ATPase. Arch Biochem Biophys 197, 83–89. Nakanishi-Matsui M, Sekiya M, Yano S & Futai M (2014) Inhibition of F1-ATPase rotational catalysis by the carboxyl-terminal domain of the e subunit. J Biol Chem 289, 30822–30831. Sekiya M, Hosokawa H, Nakanishi-Matsui M, AlShawi MK, Nakamoto RK & Futai M (2010) Single molecule behavior of inhibited and active states of Escherichia coli ATP synthase F1 rotation. J Biol Chem 285, 42058–42067. Doublie S & Carter CW Jr (1992) Preparation of selenomethionyl protein crystals. In Crystallization of Nucleic Acids and Proteins. A Pracitical Approach (Ducruix A & Giegg R, eds), pp. 311–317. Oxford University Press, Oxford, UK. Hisabori T, Muneyuki E, Odaka M, Yokoyama K, Mochizuki K & Yoshida M (1992) Single site hydrolysis of 20 ,30 -O-(2,4,6–trinitrophenyl) -ATP by the F1-ATPase from the thermophilic Bacterium PS3 is accelerated by the chase-addition of excess ATP. J Biol Chem 267, 4551–4556. Riboldi Tunnicliffe A & Higenfeld R (1999) Cryocrystallography with oil – an old idea revived. J Appl Crystallogr 32, 1003–1005. Otwinowski Z & Minor W (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol 276, 307–326. Collaborative Computational Project Number 4 (1994) The CCP4 suite: programs for Protein Crystallography. Acta Crystallogr D 50, 760–763. Leslie AGW & Powell HR (2007) Processing diffraction data with Mosflm. Evol Methods Macromol Crystallogr. Springer 245, 41–51. Evans PR (2006) Evolving methods for macromolecular crystallography. Acta Crystallogr D 62, 72–82. Tong L & Rossman MG (1997) Rotation function calculations with GLRF program. Methods Enzymol 276, 594–611.



44 Brunger AT (1992) XPLOR Version 3.1 Manual. Yale University Press, New Haven, CT. 45 Leslie AGW & Wonacott AJ (1984) Structural evidence for ligand-induced sequential conformational changes in glyceraldehyde 3-phosphate dehydrogenase. J Mol Biol 176, 743–772. 46 Rodgers AJ & Wilce MC (2000) Structure of the c-e complex of ATP synthase. Nat Struct Biol 7, 1051–1054. 47 Brunger AT, Adams PD, Clore GM, Gros P, GrosseKunstleve RW, Jiang JS, Kuszewski N, Nilges N, Pannu NS, Read RJ et al. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr D 54, 905–921. 48 Bricogne G, Vonrhein C, Flensburg C, Schiltz M & Paciorek W (2003) Generation, representation and flow of phase information in structure determination: recent developments in and around SHARP 2.0. Acta Crystallogr D 59, 2023–2030. 49 Adams PD, Grosse-Kunstleve RW, Hung LW, Ioerger TR, McCoy AJ, Moriarty NW, Read RJ, Sacchettini JC, Sauter NK & Terwilliger TC (2002) PHENIX; building new software for automated crystallographic structure determination. Acta Crystallogr D 58, 1948–1954. 50 Urzhumtseva L, Afonine PV, Adams PD & Urzhumtsev A (2009) Crystallographic model quality at a glance. Acta Crystallogr D 65, 297–300. 51 Laskowski RA, MacArthur MW, Moss D & Thornton JM (1993) PROCHECK – a program to check the stereochemical quality of protein structures. J Appl Crystallogr 26, 283–291. 52 Krissinel E & Henrick K (2007) Inference of macromolecular assemblies from crystalline state. J Mol Biol 372, 774–797. 53 Esnouf RM (1997) An extensively modified version of Molscript that includes greatly enhanced coloring capabilities. J Mol Graph Model 5, 132–134. 54 Kraulis PJ (1991) MOLSCRIPT: a program to produce both detailed and schematic structures. J Appl Crystallogr 24, 946–950. 55 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991) Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr A 47, 110–119.

2913

Nbs1 complex.

Deeper insight into HBsAg--anti-albumin antibody correlations.

A Preliminary Insight into an Islamic Mechanism for Neuroethics.

Looking ahead from age 6 to 13: a deeper insight into the development of planning ability.

An insight into the mechanism of the aerobic oxidation of aldehydes catalyzed by N-heterocyclic carbenes.

Insight into the Mechanism of Intramolecular Inhibition of the Catalytic Activity of Sirtuin 2 (SIRT2).

The structure of the pleiotropic transcription regulator CodY provides insight into its GTP-sensing mechanism.

Insight into the structure and mechanism of nickel-containing superoxide dismutase derived from peptide-based mimics.

Insight into the mechanism of aminomutase reaction: a case study of phenylalanine aminomutase by computational approach.

The Structure of a Sugar Transporter of the Glucose EIIC Superfamily Provides Insight into the Elevator Mechanism of Membrane Transport.

Computational insight into p21-activated kinase 4 inhibition: a combined ligand- and structure-based approach.

CO2 conversion to methanol on Cu(I) oxide nanolayers and clusters: an electronic structure insight into the reaction mechanism.

Insight into the molecular mechanism of a herbal injection by integrating network pharmacology and in vitro.

A new crystal structure of the bifunctional antibiotic simocyclinone D8 bound to DNA gyrase gives fresh insight into the mechanism of inhibition.

Insight into the structural mechanism for PKBα allosteric inhibition by molecular dynamics simulations and free energy calculations.

New insight into the mechanism of mitochondrial cytochrome c function.

Insight into the antifungal mechanism of Neosartorya fischeri antifungal protein.

Insight into the mechanism of polyphenols on the activity of HMGR by molecular docking.

Structural insight into the mechanism of stabilization of the 7SK small nuclear RNA by LARP7.

quaternary chalcopyrite colloidal nanocrystals.

The crystal structure of D-threonine aldolase from Alcaligenes xylosoxidans provides insight into a metal ion assisted PLP-dependent mechanism.

A new insight into an old mystery.

An insight into internal resorption.

S205A.