Cardiolipin enhances the enzymatic activity of cytochrome bd and cytochrome bo3 solubilized in dodecyl-maltoside

Cardiolipin (CL) is a lipid that is found in the membranes of bacteria and the inner membranes of mitochondria. CL can increase the activity of integral membrane proteins, in particular components of respiratory pathways. We here report that CL activated detergent-solubilized cytochrome bd, a terminal oxidase from Escherichia coli. CL enhanced the oxygen consumption activity ~ twofold and decreased the apparent KM value for ubiquinol-1 as substrate from 95 µM to 35 µM. Activation by CL was also observed for cytochrome bd from two Gram-positive species, Geobacillus thermodenitrificans and Corynebacterium glutamicum, and for cytochrome bo3 from E. coli. Taken together, CL can enhance the activity of detergent-solubilized cytochrome bd and cytochrome bo3.


Results
CL enhances the activity and decreases the K M value of purified E. coli cytochrome bd. Cytochrome bd was purified from E. coli strain MB43 using streptactin affinity chromatography and β-D-dodecyl-maltoside (DDM) as detergent, as performed earlier 32 . The purity and spectroscopic properties of the isolated protein were comparable to previous results 31,32 (data not shown), Blue-Native PAGE showed that the sample was devoid of large-scale aggregation (Suppl. Figure 1). In line with earlier data 21,31,32 , the purified enzyme showed a specific oxygen consumption activity of ~ 110 μmol O 2 *mg −1 *min −1 in buffer containing 0.025% DDM, using ubiquinol-1 as substrate. We examined the effect of CL and observed ~ twofold activation of the oxygen consumption activity (Fig. 1B). The activity of cytochrome bd in both the absence and the presence of CL was strongly suppressed by aurachin D (Fig. 1C), an inhibitor of E. coli cytochrome bd 33 .
We then investigated whether activation of cytochrome bd by CL is only observed at saturating substrate concentrations or if the K M value changes as well. In the absence of CL, cytochrome bd showed a K M value of 95 ± 16 µM for ubiquinol-1 as substrate, in line with previously published results 25,26,34 . In the presence of 10 μM The effect of CL (10 μM) on oxygen consumption activity by cytochrome bd (final conc. 2 nM) purified from E. coli was determined using a Clark-type electrode. The reaction was initiated by addition of ubiquinone-1 + DTT (arrow), the negative control contained ubiquinone-1 and DTT, but no cytochrome bd. Inset: Dependency of activation on the pre-incubation time at the indicated CL concentrations. The enzyme was incubated with CL prior to starting the reaction for either 3 min (black bars) or 60 min (white bars). (C) Impact of the inhibitor aurachin D (400 nM) on oxygen consumption by cytochrome bd in the presence or absence of 10 (Fig. 1D). These results show that CL can influence enzymatic parameters of DDM-solubilized E. coli cytochrome bd.
CL activates purified cytochrome bd from Gram-positive bacteria. Next, we evaluated if activation by CL can also be found for cytochrome bd purified from other bacteria. Genetic classification analyses indicated that two basic types of cytochrome bd can be distinguished, based on the length of a hydrophilic loop (Q-loop) close to the substrate binding site 14,[35][36][37] . Whereas E. coli cytochrome bd displays a long Q-loop, cytochrome bd from Gram-positive bacteria harbors a short version 14,[35][36][37] . Previously, purification of cytochrome bd from the two Gram-positive strains Geobacillus thermodenitrificans (formerly called Bacillus stearothermophilus) 29,30 and Corynebacterium glutamicum 28 was described. As observed above for the E. coli enzyme, purity and spectroscopic properties of these isolated proteins were comparable to previous results [28][29][30] (data not shown) and the samples were devoid of large-scale aggregation (Suppl. Figure 1). We examined the oxygen consumption activity of purified cytochrome bd from both strains with the same protocol as for E. coli cytochrome bd, except for using menaquinol-1 instead of ubiquinol-1 as substrate, as these Gram-positive bacteria use menaquinone as main constituent of the quinone pool. Cytochrome bd from G. thermodenitrificans showed lower oxygen consumption activity (~ 18 μmol O 2 *mg −1 *min −1 in the initial phase) as compared to the E. coli enzyme, consistent with previous data 38 . After the initial phase of the reaction, timedependent inactivation was observed ( Fig. 2A). CL significantly increased the activity of cytochrome bd from this strain ( Fig. 2A). As observed above for the E. coli enzyme, the activity of G. thermodenitrificans cytochrome bd was sensitive to inhibition by aurachin D in the presence and absence of CL (Fig. 2B).
Consistent with previous results 28 , the oxygen consumption activity of cytochrome bd from C. glutamicum (~ 50 μmol O 2 *mg 1 *min −1 ) was lower than that of the E. coli enzyme, but higher than that of G. thermodenitrificans cytochrome bd. Importantly, the activity was significantly enhanced by CL (Fig. 2C). We confirmed that the observed oxygen consumption activity in the presence and absence of CL was sensitive to inhibition by aurachin D (Fig. 2D). These results reveal that activation by CL is not restricted to cytochrome bd from E. coli, but can also be found for this enzyme isolated from two Gram-positive bacteria.
CL activates enzymatic activity of cytochrome bo 3 from E. coli. We then extended our efforts to the second terminal oxidase found in E. coli, cytochrome bo 3 . Cytochrome bo 3 is a heme-copper-type quinol oxidase and evolutionary is not related to cytochrome bd 14,39 . Cytochrome bo 3 was purified from E. coli strain GO105/pJRhisA using DDM as detergent without significant aggregation (Suppl. Figure 1), displaying similar spectroscopic properties as described earlier 40,41 (data not shown). Like cytochrome bd, cytochrome bo 3 can accept ubiquinol-1 as electron donor and reduces molecular oxygen (Fig. 3A). In the absence of CL, cytochrome bo 3 displayed a specific oxygen consumption activity of 47 μmol O 2 *mg −1 *min −1 , comparable to previously reported values 39,40 . Addition of CL caused a pronounced increase in activity (Fig. 3B). Oxygen consumption by cytochrome bo 3 in both the absence and in the presence of CL was highly susceptible to the inhibitor potassium cyanide (KCN) (Fig. 3C). The K M value decreased from 56 ± 13 μM in the absence of CL to 38 ± 4 μM in the presence of CL (Fig. 3D). Previously, a K M of 59 μM has been reported for cytochrome bo 3 in the presence of CL in the detergent-free state 8 . Taken together, our results show that CL can activate both terminal oxidases in E. coli.

Discussion
It has been established that CL can enhance the activity of various bacterial membrane proteins, including complexes of both aerobic and of anaerobic respiration 1,2,5,9,11,12 . Previously, activation of purified cytochrome bo 3 by CL and activation of purified cytochrome bd by asolectin was reported 9,26 . However, these experiments were carried out in detergent-free state. In the absence of detergent, membrane protein aggregation likely causes a significant decrease in activity, which subsequently is relieved by addition of lipid. In this study, we found that CL enhanced the activity of both terminal oxidases of the E. coli respiratory chain in the detergent-solubilized state. In line with our results, recently high enzymatic activity (889 e − *s −1 ≅ 135 μmol O 2 *mg −1 *min −1 ) has been reported for E. coli cytochrome bd solubilized in MSP1D1/POPC-containing nano-discs 42 , likely reflecting the importance of the lipid environment for the performance of this enzyme. CL can be located at the outer surface of a detergent-solubilized membrane protein, enabling proper vertical positioning of the protein, or it may bind to clefts or cavities on the protein surface 1,3 . CL may play a structural role, e.g. by binding at the interface between individual subunits, as previously reported for formate dehydrogenase N 12 . Alternatively, CL may enhance the interaction with the quinol substrate and/or facilitate the electron transfer reaction, as reported for nitrate reductase, where CL binds to a niche near the quinol-binding site 11 . In the respiratory chain of Saccharomyces cerevisiae CL stabilizes the super-complex formed by the cytochrome bc 1 complex and cytochrome c oxidase, binding at the interface of the two components 5,43 . In case of mitochondrial ATP synthase, CL transiently binds to conserved lysine residues in subunit c, possibly lubricating the motion of this membrane-embedded rotary machine 44 . As found for DDM-purified cytochrome c oxidase from bovine heart mitochondria, CL can be functionally required for optimal electron transports and proton translocation 43,44 . Three-dimensional structures are available for cytochrome bd from Geobacillus thermodenitrificans 30 and from E. coli 42,45 , however, the presently achieved resolution might not allow for identification of all bound lipid molecules. The decreased K M value of E. coli cytochrome bd for ubiquinol-1 measured here indicates that CL influences the substrate binding process.
In our study we investigated cytochrome bd and cytochrome bo 3 in the detergent-solubilized state and our results therefore do not clarify if CL has a similar effect on these enzymes in the native membrane. CL as highcurvature lipid is predominantly localized at the poles in rod-shaped bacteria and may thereby influence the cellular localization of membrane protein complexes 46    www.nature.com/scientificreports/ again at 37 °C, 200 rpm until reaching OD 600 ~ 2.0. Cells were sedimented by centrifugation at 6000 g for 20 min (JA-10 rotor). The pellets were washed by phosphate buffer saline, pH 7.4, and spun down at 6000 g for 20 min. Each 15 g of wet cells were re-suspended with 75 ml of MOPS solution (50 mM 3-N-morpholino-propanesulfonic acid, 100 mM NaCl and protease inhibitor (cOmplete, Roche). The cells were disrupted by passing three times though a Stansted cell homogenizer at 1.8 kb. Unbroken cells were centrifuged at 9500 g (Ja-3050-ti rotor) for 20 min. Subsequently, the supernatant was pelleted by ultracentrifugation 250,000 g (70-ti rotor) for 75 min at 4 °C. The pellet was re-suspended in MOPS solution and the protein concentration was measured using the BCA Protein Assay kit (Pierce) as described by the manufacturer. The concentration was adjusted to 10 mg/ ml and incubated in MOPS solution containing 1% DDM (final conc.) at 4 °C for an hour with gentle shaking. Un-solubilized material was sedimented by ultracentrifugation at 250,000 g at 4 °C for 15 min (70-ti rotor). The collected supernatant was applied on streptactin column at 4 °C (cold room) and the flowthrough was collected.
The column was washed with washing buffer (50 mM sodium phosphate, 300 mM NaCl, protease inhibitor (cOmplete), containing 0.01% DDM, pH 8.0) to remove unspecific protein binding and the flow-through was collected again. The elution buffer (50 mM sodium phosphate, 300 mM NaCl, protease inhibitor (C0mplete EDTA free), 0.01% DDM, and 2.5 mM desthiobiotin pH 8.0) was added to the column at 4 °C to elute the protein.   3 containing fractions were pooled and concentrated to 6.57 mg mL −1 using an Amicon Ultra centrifugal filter devices with100,000 Da molecular weight cutoff.

Purification of cytochrome bd from Geobacillus thermodenitrificans and from
Oxygen consumption activity assay. Oxygen consumption by purified cytochrome bd and cytochrome bo 3 was measured using a Clark-type electrode as previously described in Lu et al. 51 , with modifications as in Goojani et al. 31 . Briefly, the electrode was fully aerated (212 μM O 2 at 37 °C) and calibrated with sodium hydrosulfite. The purified enzymes (final conc: 2 nM for cytochrome bd from E. coli, 10 nM for cytochrome bd from G. thermodenitricifans, 2.8 nM for cytochrome bd from C. glutamicum, 5 nM for cytochrome bo 3 ) were preincubated for three minutes with CL (and with inhibitors, if applicable) in a pre-warmed (37 °C) buffer containing 50 mM 3-N-morpholino-propanesulfonic acid (MOPS), 100 mM NaCl and 0.025% DDM, pH 7.5. Ubiquinone-1 (Sigma) and menaquinone-1 (Santa Cruz Biotechnology) were dissolved in absolute ethanol (20 mM stock) and the reducing agent dithiothreitol (1 M stock) in 50 mM HEPES (4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid), pH 7.75. Quinone stock and DTT stock were mixed in 1:1 volume ratio and incubated for 3 min (ubiquinone-1/DTT) or 6 min (menaquinone-1/DTT) at 37 °C. The oxygen consumption reaction was initiated by adding the quinone/DTT mixture (final concentration 200 μM quinone and 10 mM DTT) to the assay mixture, respiration was measured for 3 min. The enzymatic activity was calculated from the slope in the period 30 s-60 s after starting the reaction (linear approximation).