Atomistic determinants of co-enzyme Q reduction at the Qi-site of the cytochrome bc1 complex

The cytochrome (cyt) bc1 complex is an integral component of the respiratory electron transfer chain sustaining the energy needs of organisms ranging from humans to bacteria. Due to its ubiquitous role in the energy metabolism, both the oxidation and reduction of the enzyme’s substrate co-enzyme Q has been studied vigorously. Here, this vast amount of data is reassessed after probing the substrate reduction steps at the Qi-site of the cyt bc1 complex of Rhodobacter capsulatus using atomistic molecular dynamics simulations. The simulations suggest that the Lys251 side chain could rotate into the Qi-site to facilitate binding of half-protonated semiquinone – a reaction intermediate that is potentially formed during substrate reduction. At this bent pose, the Lys251 forms a salt bridge with the Asp252, thus making direct proton transfer possible. In the neutral state, the lysine side chain stays close to the conserved binding location of cardiolipin (CL). This back-and-forth motion between the CL and Asp252 indicates that Lys251 functions as a proton shuttle controlled by pH-dependent negative feedback. The CL/K/D switching, which represents a refinement to the previously described CL/K pathway, fine-tunes the proton transfer process. Lastly, the simulation data was used to formulate a mechanism for reducing the substrate at the Qi-site.


Table of Contents
Text S1.
Semiquinone binds firmly to canonical residues at the Qi-site Text S2. Quinone binding does not promote Lys251 rotation inward Table S1.
Substrate binding distances at the Qi-site in X-ray crystal structures of cytochrome bc1 complexes Table S2.
Distances at the Qi-site with semiquinone from the conf1 simulation Table S3.
Distances at the Qi-site with semiquinone from the conf2 simulation Table S4.
Distances at the Qi-site with quinone from the conf3 simulation Table S5.
The empirical pKa values for the substrate-bound Qi-site in the cyt bc1 complex X-ray crystal structures Table S6.
Distances at the Qi-site with quinone from the conf4 simulation Table S7.
The empirical pKa values for the apo Qi-site in the cyt bc1 complex X-ray crystal structures Figure S1.
The substrate binding at the Qi-site in the cyt bc1 complex X-ray crystal structures Text S1. Semiquinone binds firmly to canonical residues at the Qi-site The reaction radical semiquinone (SQ) binding at the Qi-site was robust in the simulation ( Fig. 2A-B). This stability originated from the half-protonated state of the quinone ring and the charged state of Lys251 and Asp252 side chains (conf1 in Table 1).
The C4-hydroxyl was able to H-bond with the Asp252 COO-, which in turn made it possible for the quinone ring to orient its C1-carbonyl to H-bond with the epsilon protonated His217 ( Fig. 2B; Fig. S3A; Table  S2). The connection between the C1-hydroxyl and Asp252 could also be water-mediated occasionally ( Fig. 2A). Although Lys251 is not participating directly in the substrate binding in any of the X-ray crystal structures, the Lys251 NH3+ remained turned towards the Qi-site with the bound SQ, where it Hbonded continuously with the Asp252 COO- (Fig. S3; Table S2). On the B side of the dimer, Lys251 formed even a direct H-bond with the C1-hydroxyl ( Fig. 2B; Fig. S3; Table S2).
The other major difference between the SQ simulation ( Fig. 2A-B) and the previously reported X-ray crystal structures with the bound substrate (Table S1; Fig. S1) was the positioning of Asn221 side chain at the Qi-site ( Fig. 2A-B; Fig. S3; Table S2). The Asn221 is interacting favorably with the substrate's C6-methoxy group in the X-ray crystal structures ( Fig. S1; also for Q in Fig. 2C). However, Asn221 also H-bonds with the C4-carbonyl either directly (Q in Fig. 2B) or it forms a water-mediated connection with SQ in the simulations (Fig. 2B). The Asn221 side chain is not necessarily competing with His217 for the substrate's C4-carbonyl ( Fig. S3; Table S2), instead the H-bond donating residue (asparagine/serine/glutamine; Fig. S2) could be needed to coordinate His217 positioning.
Notably, the neutral SQ binding involves less waters on the B side than on the A side. Although the SQ could H-bond directly with the canonical residues, water can provide necessary flexibility for quinone ring positioning. Nevertheless, the relative stability of SQ binding at the Qi-site is not unexpected. The reaction radical is known to reside at the Qi-site from various experimental studies 1,2 . For comparison, it is still under contentious debate whether SQ is even momentarily formed during the oxidation reaction at the Qo-site 1,3 . The sequential model (Fig. 5) dictates that the half-protonated substrate would stay in this stalemate state waiting for the second proton that can be acquired only after another electron has been dispatched into the Qi-site ( Fig. 1A-B). It would be very unfavorable, if the substrate could break its ties before acquiring both electrons/protons as the reaction radical SQ is linked to detrimental superoxide generation.

Text S2. Quinone binding does not promote Lys251 rotation inward
The quinone ring positioning or C1-and C4-group coordination for nonprotonated substrate Q at the Qisite ( Fig. 2C-D) is less coordinated than that of neutral SQ.
When Lys251 and Asp252 side chains were set charged (conf1 in Table 1), a salt bridge was formed during the equilibration phase of the simulations (0-10 ns) with Q at the Qi-site. On the B side, the Lys251-Asp252 salt bridge broke during the restraint-free production simulation as water entered between the residues (Fig. S4A). On the A side, the salt bridge lasted throughout the simulation; however, this particular binding mode is not considered here, because it lacked any kind of C4-coordination ( Fig.  S4A; Table S4). On the B side of the dimer, the C1-carbonyl of Q H-bonded with His217. Meanwhile, the amine group of Asn221 side chain H-bonded with His217 and the C6-methoxy of Q. The C1-carbonyl formed a water bridge with the Asp252 COO- (Fig. S4A). Thus, the logical conclusion is that without a Hbond donor at the C1-position, Lys251 would not be participating in the substrate binding.
The Q binding is less coordinated with the canonical residues ( Fig. S2) than that of half-protonated SQ, even if the Lys251 and Asp252 side chains were set neutral (conf4 in Table 1). The underlying thought behind this set-up was that the proton would be transferred between Lys251 and Asp252 prior to Q binding. On the B side, the Lys251 NH2 assumed quickly the outward rotamer pose and the C1-or C4coordination was lost equally fast ( Fig. S4B; Table S5). On the A side, the C1-carbonyl could H-bond with the Asp252 COOH , while the C4-carbonyl H-bonded with the Asn221 side chain ( Fig. S4B; Table S5). Although Asn221 might be crucial for the initial stages of the Q binding, the overwhelming amount of X-ray crystallographic data suggest that the C4-carbonyl H-bonds to the His217 side chain instead ( Fig.  S1; Table S1).
The simulations contain variability regarding Q binding (Fig. S4). Only with bound SQ both Asp252 and His217 side chains could H-bond directly with the C1-and C4-groups simultaneously ( Fig. 2B; Fig. S3; Table S2). The difference between the SQ and Q binding modes must arise mainly from the Lys251 protonation state and rotamer pose changes. The outward pose of the Lys251 side chain is preferred, when Q binding is at least partially coordinated at the Qi-site (Fig. 2C-D; Table S4). Accordingly, the simulations imply that the inward pose of Lys251 would be needed to stabilize the H-bonding between neutral SQ's C4-carbonyl and His217 ( Fig. 2A-B). Table S1. Substrate binding distances at the Qi-site in X-ray crystal structures of cytochrome bc1 complexes.  (1) Cardiolipin (CL) head group (phosphate groups) in the general vicinity of the Lys251 side chain and the Qi-site. (2) The residue numbering/naming from R. capsulatus (Fig. S2). Table S2. Distances at the Qi-site with semiquinone from the conf1 simulation. (1) The polar atoms of the H-bonding groups were used in the distance measurements. (2) Percentage of the simulation that a H-bond was formed. The upper limit for a H-bond was 3.4 Å. The H-bonding angle was not considered. Table S3. Distances at the Qi-site with semiquinone from the conf2 simulation.  Table S4. Distances at the Qi-site with quinone from the conf3 simulation.   (1) Cardiolipin (CL) head group (phosphate groups) in the general vicinity of the Lys251 side chain and the Qi-site. (2) The residue numbering/naming from R. capsulatus (Fig. S2). The alternative titration state pKa values (PROKA3.1) are shown in parentheses.

H-bond partners (1) AVG (Å) MAX (Å) MIN (Å) MED (Å) STDEV (Å) H-BOND (%) (2)
(3) The difference between Lys251 and Asp252 pKa values.  (1) The polar atoms of the H-bonding groups were used in the distance measurements. (2) Percentage of the simulation that a H-bond was formed. The upper limit for a H-bond was 3.4 Å. The H-bonding angle was not considered.  (1) Cardiolipin (CL) head group (phosphate groups) in the general vicinity of the Lys251 side chain and the Qi-site. (2) The residue numbering/naming from R. capsulatus (Fig. S2).   Table S1). The limit for H-bonding is set to ~3.6 Å (magenta dotted line). Also the oxygen atom in the C5-methoxy group is shown to H-bond, although strictly speaking, it forms somewhat weaker electrostatically favorable interactions with asparagine or serine side chains instead.   Table 1) included the epsilon protonated His217 (or Hse217), positively charged Lys251 (or Lys251 NH3+ ), and negatively charged Asp252 (or Asp252 COO-). On the A side, the neutral SQ (blue lines) is forming more water mediated interactions than on the B side (red lines), which is reflected as longer distances for the interacting partners on the B side. Both A and B side arrangements assured good coordination for the C1-and C4-groups with the Asp252 and His217 side chains, respectively. Despite this, the H-bonding between His217 and the C4-carbonyl of SQ is clearly disturbed in the latter part of the simulation on the B side lacking water (red line). B) The second set-up (conf2 in Table 1) included the double protonated His217 (Hsp217), the positively charged Lys251 (or Lys251 NH3+ ), and the negatively charged Asp252 (or Asp252 COO-). Note that on the A side (blue line) the C1-carbonyl is not H-bonding (or forming a water bridge) with the Asp252 COO-(binding mode not discussed elsewhere), although both the C4-and C1-carbonyls are connected to His217 and Asp252 on the B side (red line), respectively. The H-bonding distance of 3.4 Å is indicated with a black line. For clarity, the results are shown as 10-point moving averages.

Figure S4. Distances between potential H-bonding partners at the Qi-site with nonprotonated quinone.
A) The first setup included the epsilon protonated His217 (or Hse217), positively charged Lys251 (or Lys251 NH3+ ), and negatively charged Asp252 (or Asp252 COO-). On the A side, the C4-carbonyl of Q is not H-bonding (or forming water bridges) neither with His217 or the amine group of Asn221 side chain (or Asn221 NH2 ; blue lines; binding mode not shown elsewhere). In contrast, on the B side, His217 is H-bonding with the C4-carbonyl (red lines) as the Asn221 NH2 H-bonds with the histidine side chain and the C5-methoxy group (Fig. 2C). Accordingly, stable C1-and C4-group coordination resulted in breaking of the Lys251 NH3+ -Asp252 COOsalt bridge on the B side of the dimer with bound Q. B) The second set-up included the epsilon protonated His217 (or Hse217), the neutral Lys251 (or Lys251 NH2 ), and the neutral Asp252 (or Asp252 COOH ). Only on the A side, the C1-carbonyl of was able to H-bond directly with the Asp252 COO-, and while in this position, the C4-carbonyl Hbonded with the Asn221 side chain instead of His217 (Fig. 2D). On the B side both the C1-and C4-groups were not connected to the Asn221, Asp252 or His217 side chains. The H-bonding distance of 3.4 Å is indicated with a black line. For clarity, the results are shown as 10-point moving averages. Figure S5. The proposed effects of mutations on the CL/K/D switching and proton shuttling. A) Mutating His217 to alanine (H217A mutant) is not expected to prevent the switching and proton shuttling, but alanine cannot H-bond or donate protons to the substrate (not shown). B) Substituting Asp252 with alanine (D252A mutant) is expected to make the Qi-site more hydrophobic and prevent the switching of Lys251 between the cardiolipin (CL) and Asp252. Also binding should be weakened as the substrate is unable lock its C1-group at the site with the Ala252 side chain (not shown). C) With D252N mutant, the Lys251 NH3+ should be able to form water bridge or even a direct H-bond with the asparagine side chain (not shown). Although Lys251 could participate in the neutral SQ binding with D252N mutant, asparagine is unble to accept protons. D) Mutating Lys251 to methionine (K251M mutant) should deactive the K/D switching. The proton transfers would have to happen entirely via interconnected water molecules directly from the cardiolipin (CL).