Introduction

In the last steps of aerobic respiration, exergonic electron transfer drives proton translocation or pumping across the mitochondrial inner membrane or bacterial cytoplasmic membrane to maintain a transmembrane proton electrochemical gradient. Free energy stored in this gradient is used to drive transmembrane transport processes and the synthesis of ATP. In mammalian mitochondria, electron-transport chain complexes (C) I and II reduce the membrane-bound two-electron carrier ubiquinone (UQ), also known as coenzyme Q, to ubiquinol (UQH2), which donates electrons to dimeric CIII2, also known as cytochrome bc1. Complex III is an electron donor to cytochrome c, which reduces CIV (cytochrome c oxidase) that transfers electrons to molecular oxygen, which is reduced to water at a heme-copper catalytic site (for recent reviews, see e.g., refs. 1,2,3).

In CIII2, UQH2 binds in a quinone-binding site called Qo (also called QP, Fig. 1A). Crystal structures of canonical CIII2s have the Qo site typically empty and its position was originally suggested based on observation of binding of Qo-site inhibitors such as myxothiazol or stigmatellin4,5. More recent cryo-EM data revealed endogenous UQ in the Qo sites in CIII2, e.g., from the pathogenic yeast C. albicans6 and in one monomer of CIII2 in a mammalian CI-CIII2 supercomplex7. In canonical CIII2 the UQ benzoquinone head group is located deeply inside the Qo site cavity to allow oxidation of UQH2 by the Rieske head domain. Density for an endogenous menaquinone (MQ) was found at the same position in the Qo site as in canonical CIII2 (Qo1a) in the actinobacterial C. glutamicum CIII28,9. However, MQ in the M. smegmatis CIII2 occupies a more distal position in the Qo site, referred to as Qo1b, at the entrance to the Qo site cavity, with the naphthoquinone head group found ~15 Å from the FeS cluster electron acceptor10,11.

Fig. 1: Overview of the studied complexes.
Fig. 1: Overview of the studied complexes.The alternative text for this image may have been generated using AI.
Full size image

Overall structures of the CIII2CIV2 supercomplex (A) and cytochrome bd-I (B). Electron and proton-transfer pathways are indicated with black and colored dashed arrows, respectively. The negatively and positively charged sides of the membrane are indicated by N and P, respectively. The M. smegmatis CIII2CIV2 supercomplex and cytochrome bd-I structural models are PDB ID 6ADQ11 and 7D5I77, respectively.

The quinol is oxidized in a branched electron-transfer reaction referred to as the Q-cycle (for review, see refs. 5,12). One electron is transferred from QH2 to a Rieske iron-sulfur site (FeS), which transfers electrons consecutively to cytochrome c1 and a water-soluble cytochrome c that docks near cytochrome c1. The second electron is transferred sequentially to hemes bL, bH and then to a Q bound in the Qi (also called QN) site, which forms a semiquinone. Protons are released from QH2 in the Qo site to the membrane positive (p) side. Upon binding of a second UQH2 in the Qo site, the same sequence of reactions is repeated, leading to reduction of the semiquinone in the Qi site, uptake of two protons from the membrane negative (n) side and formation of QH2, which equilibrates with the Q/QH2 pool in the membrane. The overall Q-cycle process results in a transmembrane charge separation that is equivalent to the translocation of one proton from the n to the p side of the membrane for each electron that is transferred from QH2 to cytochrome c. In CIV, electrons are initially transferred from cytochrome c to a copper center, CuA, and then consecutively to heme a and the heme a3-CuB catalytic site, which, in its reduced state, binds O2. Reduction of O2 to H2O is linked to proton uptake from the n side of the membrane, which also in CIV results in a charge separation across the membrane that is equivalent to the transfer of one positive charge across the membrane for each electron transferred from cytochrome c to O2. In addition, each electron transfer to the catalytic site is linked to proton pumping across the membrane, which doubles the charge-separation stoichiometry (for review, see refs. 13,14,15,16). In the CIII2CIV2-mediated menaquinol (MQH2) oxidation pathway in M. smegmatis, MQH2 binds near the FeS domain in each monomer of CIII2 (Fig. 1A), and electrons are presumably transferred utilizing a Q-cycle mechanism, similar to that in the mitochondrial CIII217. Mycobacteria do not harbor a water-soluble cytochrome c and electrons from the FeS site are instead transferred to CuA via a cytochrome cc domain of each monomer of CIII218.

Mycobacteria, including the human pathogen Mycobacterium tuberculosis, belong to the phylum Actinobacteria. These Gram-positive bacteria utilize aerobic respiration to extract energy from various carbon sources during growth19,20. In contrast to eukaryotes and many other bacteria, the electron-transfer chain in mycobacteria is exceedingly branched20. In the last steps of the mycobacterial respiratory chain, electrons from MQH2 are transferred to O2 through two pathways that involve different types of oxidoreductases: cytochrome bd-type quinol oxidases and a heme-copper aa3-type CIV, which is part of an obligate CIII2CIV2 supercomplex (Fig. 1A). Cytochrome bd oxidases are found only in prokaryotes where they catalyze oxidation of QH2 and reduction of O2 to H2O, but they are evolutionarily unrelated to heme-copper oxidases (for a recent review, see ref. 21). The bd oxidases do not pump protons and energy conservation arises only from a transmembrane charge separation upon proton release from QH2 to the p side and proton uptake upon O2 reduction from the n side, linked to intramolecular electron transfer. The enzymes harbor three heme groups, two heme bs (termed b558 and b595) and a heme d, which is the site of reduction of O2 to H2O. The relative position of heme b595 and heme d varies between organisms and their involvement in electron transfer and catalysis is not fully understood1,21. Structures of the E. coli cytochrome bds22,23,24 and actinobacterial cytochrome bds (Fig. 1B)25,26,27 are overall similar and suggest that electrons from QH2 are initially transferred consecutively to heme b558, heme b595 and heme d. M. smegmatis, like E. coli, harbors two cytochrome bd gene clusters, the cydAB (bd-I) and appBC (bd-II)28,29, but evidence for a biochemical role exists only for bd-I. Analysis of the cytochrome bd structures suggests variability among the QH2 binding sites in E. coli and also among different actinobacteria21,22,24,25,26,27,30.

MQ has a lower midpoint potential (Em7 ≅ −80 mV31,32) than the mammalian electron carrier, UQ (Em7 ≅ +100 mV31,33). Some Gram-negative bacteria are known to exchange between UQ and MQ depending on oxygen concentration. This switch between different isoprenoid quinones does not take place in Gram-positive bacteria31,34. On the other hand, it has been shown that mycobacteria maintain cellular bioenergetics in oxygen-deficient habitats such as biofilms by switching to the synthesis of polyketide quinones35. Because under low-oxygen concentrations M. smegmatis switches from using the aa3-type heme-copper CIV to cytochrome bd36, the polyketide quinone is presumably used as an electron donor to the latter35.

A number of drug candidates that target oxidative phosphorylation in M. tuberculosis have been identified. These drugs inhibit selectively either electron transfer to O2 or the ATP synthase37, thereby impairing ATP synthesis. Bedaquiline inhibits mycobacterial ATP synthase38 by inducing structural changes that result in tight binding at the interface of two of the enzyme’s subunits39. The electron-transport chain branch through CIII2CIV2 is inhibited by telacebec (Q203)40 or Lansoprazole (LPZ) metabolized to its sulfide form (LPZS)41 by competing with MQ/MQH2 binding to the Qo site in CIII210,42. However, killing of M. tuberculosis may require blocking both the CIII2CIV2 and cytochrome bd electron transfer branches to O243,44. In the current study, we investigated oxidation of a menaquinone analog, dimethyl-naphthoquinone (DMNQH2), by O2, catalyzed by either the isolated M. smegmatis cytochrome bd-I or CIII2CIV2 supercomplex in the presence of the UQH2 analog decylubiquinone (DCQ). Kinetic data show that both CIII2CIV2 and cytochrome bd-I are inhibited by DCQ, and cryo-EM data reveal that DCQ binds to the Qo site in CIII2. 44Furthermore, we show that DCQ inhibits the growth of M. smegmatis cells and impairs the intracellular growth of virulent M. tuberculosis in human primary macrophages, suggesting that DCQ could be explored as a potential drug targeting pathogenic mycobacteria.

Results and discussion

Naphthoquinol oxidation activity of the isolated CIII2CIV2 supercomplex

We measured the quinol oxidation: O2 reduction activity of the M. smegmatis CIII2CIV2 supercomplex by monitoring the oxygen-reduction rate upon addition of CIII2CIV2 to a solution containing reduced 2,3-dimethyl-[1,4] naphthoquinone (DMNQH2), which is used as an analog of MQ. The maximum supercomplex activity was measured at 100 μM DMNQH2 and was found to be 80 ± 10 s−1 (one measurement on each of three different supercomplex preparations).

DMNQH2 oxidation activity of the isolated cytochrome bd-I

The oxidoreductase activity of the M. smegmatis cytochrome bd-I was observed by measuring the oxygen-reduction rate upon addition of cytochrome bd-I to a solution containing DMNQH2. As described above, we monitored the O2-reduction rate before the addition of cytochrome bd-I, and this slope was subtracted from that obtained after the addition of cytochrome bd-I. The activity observed at 300 μM DMNQH2 was 18 ± 1 s−1 (n = 3).

Activities of isolated CIII2CIV2 and cytochrome bd-I in the presence of decylubiquinone

We used decylubiquinone (DCQ), which is an ethanol-soluble, commonly used (e.g., ref. 45) analog of UQ46. The CIII2CIV2 supercomplex activity with reduced DCQ was ≤2 s−1, i.e., ≤3% of the maximum activity measured with DMNQH2, both at a concentration of 100 μM. The activity of cytochrome bd-I with DCQ was too low to be measured. Figure 2A shows the DMNQH2 oxidation: O2 reduction activity of the supercomplex as a function of the DMNQH2 concentration. As seen in the figure, the activity increased with increasing DMNQH2 concentration up to ~100 μM DMNQH2 and then decreased slightly above 100 μM DMNQH2. In the presence of DCQH2 at 20 μM or 50 μM, in general, a lower turnover rate was observed. In addition, with 50 μM DCQH2, a decrease in activity at DMNQH2 concentrations >100 μM was observed. The reason for this decrease is unknown, but we note that under native conditions the Q-cycle of CIII2 involves oxidation of MQH2 at the Qo site as well as reduction of MQ the Qi site. In other words, during turnover, two different sites compete for Q/QH2 and most likely display different binding constants for the different redox states and the two different quinones, giving potentially rise to complicated kinetics.

Fig. 2: Supercomplex and cytochrome bd-I activities.
Fig. 2: Supercomplex and cytochrome bd-I activities.The alternative text for this image may have been generated using AI.
Full size image

CIII2CIV2 supercomplex (A) and cytochrome bd-I (B) activity as a function of the DMNQH2 concentration, in the absence and presence of DCQH2. Supercomplex (C) and cytochrome bd-I (D) activity as a function of the DCQH2 concentration, with 100 μM MQH2. All measured values are shown in the figure. Where applicable, the solid line is drawn through the average of the measured values.

Figure 2B shows the DMNQH2 oxidation: O2 reduction activity of cytochrome bd-I as a function of DMNQH2 concentration. As seen in the figure, the activity increased with increasing DMNQH2 to reach ~18 s−1 at 300 μM DMNQH2. This activity would presumably increase further above 300 μM DMNQH2, but these measurements were precluded by the too low solubility of DMNQH2. In the presence of 50 μM DCQH2, the DMNQH2-oxidation activity was slower and at 300 μM it was ~45% of that in the absence of DCQH2 (Fig. 2B).

Figure 2C shows the supercomplex activity measured with 50 μM DMNQH2 as a function of added DCQH2. As seen in the figure, the activity dropped, with an IC50 of ~70 μM, to reach ~3% of the maximum activity at 500 μM DCQH2. Also, with cytochrome bd-I, the activity dropped upon addition of an increasing amount of DCQH2 (Fig. 2D), but at a saturating concentration of ~20 μM DCQH2, it remained at ~10 s−1, i.e., ~42% of the maximum activity.

In the measurements described above, the inhibitory DCQ was added in the reduced form, i.e., DCQH2. Because the midpoint potential of DCQ/DCQH2 is higher than that of DMNQ/DMNQH2, in the presence of DCQ, any added DMNQH2 would otherwise be oxidized directly by DCQ.

Modulation of DCQH2 inhibition by engineering the Qo site

To further analyze the inhibitory effect of DCQH2, we used a structural variant of CIII2 in which residues near the Qo site were modified42: the M337A, M305A, F153A triple mutation of the QcrB subunit42. The structural variant (termed AAA) was expressed and purified using the same approach as that used with the wild-type supercomplex. The DMNQH2:O2 oxidoreduction activity of the structural variant was ~50 s−1 (at 100 μM DMNQH2), i.e., approximately 60% of the wild-type activity. Titration of the CIII2CIV2 supercomplex activity as a function of added DCQH2 showed that for the triple mutant, the IC50 increased from ~70 μM (wild-type) to ~130 μM (Fig. 3), suggesting that DCQH2 binds at the same site as DMNQH2.

Fig. 3: Activity of the wild-type (WT, grey squares) and M337A, M305A, F153A triple (referred to as AAA, black circles) mutant supercomplex as a function of the concentration of added DCQH2.
Fig. 3: Activity of the wild-type (WT, grey squares) and M337A, M305A, F153A triple (referred to as AAA, black circles) mutant supercomplex as a function of the concentration of added DCQH2.The alternative text for this image may have been generated using AI.
Full size image

The DMNQH2 concentration was 100 μM. The solid line is a fit of the data with the equation: \(\theta =\left({1+\left(\displaystyle \frac{{{{\rm{K}}}}_{{{\rm{A}}}}}{\left[{{\mathrm{DCQH}}}_{2}\right]}\right)}\right)^{-1}\) where θ is the percent fraction maximum activity, KA is the DCQH2 concentration at which half the maximum rate is obtained, [DCQH2] is the concentration added DCQH2. The maximum activity in the absence of added DCQH2 is 80 s−1 and 50 s−1 for the WT and AAA variant, respectively. The supercomplex activity with 100 μM DCQH2 was ≤3% of the maximum activity with 100 μM DMNQH2, therefore, the two solid lines are assumed to approach zero at the highest DCQH2 concentrations.

Reduced growth of M. smegmatis in the presence of DCQ

To test the effect of DCQ or DMNQ on the growth of M. smegmatis, cells were plated on LB medium + kanamycin, supplemented with different concentrations of DCQ or DMNQ. In Fig. 4, each section of each plate represents different bacterial culture dilutions, marked with numbers ranging from 1 (undiluted) to 100,000. As observed in panels Fig. 4C and D, addition of DCQ resulted in impaired bacterial growth. Addition of 25 µM DMNQ did not exhibit any effect, while at 50 µM DMNQ, growth was somewhat impaired at bacterial culture dilutions by a factor of >1000 (Fig. 4E, F).

Fig. 4: Growth of M. smegmatis cells in the presence of DCQ or DMNQ.
Fig. 4: Growth of M. smegmatis cells in the presence of DCQ or DMNQ.The alternative text for this image may have been generated using AI.
Full size image

A Control plate with 75 μl 98% ethanol, added to the liquid agar before it set. B Control plate without any additions. C 25 μM or D 50 μM decylubiquinone (DCQ) added. E 25 μM or F 50 μM naphthoquinone (DMNQ) added. Each plate was divided into six parts where each part represents a different cell dilution with LB by factors, from “1” (undiluted) to “100,000”.

Reduced growth of virulent M. tuberculosis cells in human macrophages upon treatment with DCQ

Next, the potency of DCQ to reduce growth of virulent M. tuberculosis expressing green fluorescent protein (GFP) was explored using a well-established macrophage infection model and high-content imaging47. As mycobacteria are facultative intracellular pathogens, this represents a physiological model to study inhibition of intracellular growth. Different concentrations of DCQ and DMNQ were tested for the ability to inhibit intracellular growth of H37Rv in human monocyte-derived macrophages after 4 days culture (Fig. 5A). Accordingly, we observed a dose-dependent reduction of H37Rv growth in the presence of DCQ, while there was a corresponding increase of mycobacterial growth in the presence of DMNQ (Fig. 5A). The antimycobacterial effect of DCQ at 50 µM and 250 µM was significantly higher compared to 25 µM (p < 0.016 and p < 0.0054, respectively). Instead, there was a significantly increased growth of H37Rv upon treatment with 25 µM, 50 µM and 250 µM DMNQ (p < 0.0094–0.0001) as compared to the corresponding concentrations of DCQ (Fig. 5A). Lower concentrations of DCQ (1 and 10 μM) did not significantly affect intracellular M. tuberculosis growth compared to the MOI infection control. Overall, DCQ-mediated growth inhibition of H37Rv at doses of 50 µM and 250 µM (53–64%) was also significant compared to the MOI1 infection control (p < 0.003–0.0001), while there was no significant difference between the 25 µM DCQ treatment and the MOI1 infection control (Fig. 5A). In contrast, H37Rv growth was significantly enhanced by different concentrations of DMNQ (p < 0.041–0.0001).

Fig. 5: Efficacy of DCQ and DMNQ to reduce growth of M. tuberculosis in primary human macrophages.
Fig. 5: Efficacy of DCQ and DMNQ to reduce growth of M. tuberculosis in primary human macrophages.The alternative text for this image may have been generated using AI.
Full size image

A High-content analysis with IncuCyte at day 4 was used to quantify intracellular H37Rv growth in infected macrophages treated with 25, 50, or 250 µM doses of DCQ or DMNQ compared to the MOI1 infection control. The dotted lines indicate the MOI1 infection control set to 100% growth, as well as 50% Mtb growth in macrophages. Statistical differences comparing different doses of compounds (median ± range) were determined using a 2-way ANOVA and Tukey’s multiple comparisons test. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05. One representative experiment out of three similar, including n = 6 replicates from two donors, is shown. Data are displayed in a dot-and-bar plot showing median and range. B Microscopy images showing macrophages (light gray color) infected with H37Rv-GFP (green color) and treated with different concentrations of DCQ (upper panel) or DMNQ (lower panel). Samples from whole well images, magnification ×20. Observe the changes in morphological features as the concentrations of UDCQ and DMNQ increase, ranging from preserved cell integrity (characterized by relatively large, irregular cells with ovoid nuclei) to reduced or lost cell integrity (evidenced by small, rounded, shrunken, and condensed cells). Arrows indicate H37Rv-GFP infected cells that were numerous upon DMNQ treatment. C Flow cytometry was used to assess intracellular H37Ra bacteria in primary human macrophages after 48 h treatment with DCQ and DMNQ using concentrations that maintain cellular integrity. Results were compared to MOI5 infection control as well as the uninfected control. The geometric mean of the GFP-positive (+) subset gated on live cells is shown in the histogram. Mtb M. tuberculosis.

Microscopic examination of M. tuberculosis-infected macrophages revealed altered cellular morphology at higher concentrations of DCQ and especially DMNQ, including rounded and shrunken cells (Fig. 5B). These changes could indicate incipient cytotoxicity induced by higher doses of the compounds. Notably, macrophage morphology was preserved at DCQ doses of 25 and 50 µM (Fig. 5B, upper panel), while DMNQ treatment reduced cellular integrity even at the lower dose of 50 µM (Fig. 5B, lower panel). Despite the compromised morphology of M. tuberculosis-infected cells at the higher 250 µM dose (Fig. 5B, upper panel), DCQ clearly maintained its antimycobacterial activity (Fig. 5A), indicating direct effects on the bacteria and the ability to inhibit both intra- and extracellular M. tuberculosis growth.

Consistent with the high-content imaging results (Fig. 5A, B), flow cytometry analysis quantifying avirulent H37Ra-GFP inside human macrophages confirmed that DCQ could inhibit mycobacterial growth at 48 h (Fig. 5C). This effect was observed comparing DCQ (geometric mean (GM): 2814) to both DMNQ (GM: 4992) and the untreated infection control (GM: 3534), using compound doses that maintained cell integrity (Fig. 5C). Altogether, these data suggest that DCQ possesses potent antimycobacterial activity, whereas mycobacteria may exploit DMNQ to enhance their survival in host cells.

Cryo-EM structure of the CIII2CIV2 supercomplex with bound DCQ

In order to verify the binding site of DCQ in CIII2, we determined a cryo-EM structure of the M. smegmatis CIII2CIV2 supercomplex solubilized in GDN, which was incubated with DCQH2 in the presence of O2 prior to cryo-EM grid preparation. The structure of the supercomplex was solved to an overall resolution of 2.5 Å without symmetry (C1) (Fig. 6A). Its architecture and subunit composition agree with already published M. smegmatis supercomplex structures11,48,49. The complex is composed of centrally positioned CIII2 dimer where each CIII monomer consist of three main functional subunits: QcrB, which harbors two cytochrome bs and the quinol/quinone binding sites, QcrA Rieske subunit, which harbors the FeS iron-sulfur cluster and the membrane-bound cytochrome cc subunit (QcrC) with two heme c containing domains. One of these domains is attached to the Rieske subunit, the other is in direct interface with CIV. The cytochrome cc subunit creates an electron-transfer bridge to the two copies of CIV on each side of the CIII2, forming a direct path for electron transfer from CIII to CIV. CIV is composed of main functional subunits CtaC, which harbors a dinuclear CuA and CtaD with heme a and the heme a3-CuB binuclear active site, where O2 is reduced to H2O. Additional CIV subunits are CtaE and CtaF and actinobacteria-specific CtaI and CtaJ. Accessory subunits LpqE and AscA stabilize the connection between CIII2 and CIV. In addition, a superoxide dismutase is part of the supercomplex.

Fig. 6: Structure of the supercomplex with bound DCQ.
Fig. 6: Structure of the supercomplex with bound DCQ.The alternative text for this image may have been generated using AI.
Full size image

A MQ-binding active sites Qo and Qi, respectively, and additional binding sites in CIII2 (pink) of the supercomplex. The two subsites, Qo1a and Qo1b, of the of Qo site are marked with a blue dashed line. MQ and DCQ in its binding site (Qo1b position) are shown in dark and light green, respectively. B Qo cavity with bound DCQ (light green) and interacting residues. Residues from the Rieske subunit are marked by an asterisk. C Comparison of the binding poses of MQ and DCQ in the Qo1b binding site. D, E Comparison of the binding of DCQ and known Qo site inhibitors, lansoprazole sulfide (D, red42) and Q203 (E, pink10). The panels are an overlay of the current structure and those published earlier with these two inhibitors. Cryo-EM electron density maps of the DCQ (F) or MQ (G) in the Qo1b site of the supercomplex (σ = 2 level).

As mentioned above, there are two substrate binding sites in each CIII monomer, an electron donor site Qo where the quinone can adopt a proximal Qo1a position and a more distal Qo1b position towards FeS center (Fig. 6B), and an electron acceptor site Qi. Because CIII2 is a dimer, there are two copies of each quinone-binding site. The cryo-EM data identified densities corresponding to substrate quinones bound at each Qo site. However, the density in one monomer of the CIII2 dimer could be interpreted as the externally added DCQ (Fig. 6C) while in the other site it was more similar to the native MQ (Fig. 6D), which is typically copurified with the supercomplex. Hence, the densities are most likely a combination of both molecules. For this reason, a 50% occupancy was assigned to each of MQ and DCQ. The DCQ density is seen in the more distal Qo1b position, i.e., the same as that of the native MQ (see superposition of the two quinones, DCQ and MQ, in Fig. 6E), in accord with the earlier observations10,11. MQ and DCQ display the same orientation in the binding site, with the aromatic head pointing towards the Qo1a site and the aliphatic tail placed outside the cavity in the membrane space. After binding at the Qo1b position, the native MQH2 presumably moves to the more proximal Qo1a position (c.f. Fig. 6B) to transfer an electron and proton to the FeS Rieske domain. The inhibitory action of DCQ is explained by competition of DCQ and MQ for the same site, where DCQ would be unable to move to QO1a or adopt a correct orientation to transfer an electron to FeS. In Fig. 6B–D, some residues that define the MQo1b site are shown; Phe153, Phe156, Met305, Met337 and Leu 175, form hydrophobic interactions with the substrate quinones, and a possible hydrogen bond is formed with Thr179. Because the sample was prepared under aerobic conditions, the oxidized state is more likely and, therefore, both MQ and DCQ are modeled in their oxidized states.

Structures of CIII2CIV2 with bound lansoprazole sulfide (LPZS)42 or telacebec (Q203)10 indicate that these inhibitors block MQH2 binding to both Qo1a and Qo1b sites37. Figure 6FG shows an overlay of DCQ (shown in yellow) along with LPZS (panel F, red) and telacebec (panel G, pink).

In earlier studies, the electron density at the Qi site was found to be too disordered for distinction of the orientation of the MQ headgroup, however, the tail part of the molecule was clearly resolved. In the current structure, the situation is very similar, while it is difficult to resolve the headgroup, there is a density for the aliphatic tail, which is consistent with the native MQ (menaquinone-9, with nine isoprene units50). In other words, it is unlikely that DCQ is bound in the Qi site.

As observed earlier, a number of additional ligands were resolved in the supercomplex structure, including extra MQ-binding sites, with unknown function (see Fig. 6A)10,11, together with multiple cardiolipin molecules located in cavities between CIII2 and CIV.

The combined cryo-EM and kinetic data indicate competitive binding of DCQH2 and DMNQH2 to the same Q-binding site. The dual action of DCQ, impairing both CIII2CIV2 and the cytochrome bd branches of the electron-transport chain, makes this compound a promising inhibitor of mycobacterial respiration, which, in combination with e.g., Bedaquiline, could potentially synergize to abolish oxidative phosphorylation44.

Summary and conclusions

Oxidation of both UQH2 (Em7 ≅ +100 mV31,33) and MQH2 (Em7 ≅ −80 mV31,32) catalyzed by the M. smegmatis CIII2CIV2 supercomplex or cytochrome bd-I is a highly exergonic reaction because the reduction of the final electron acceptor, O2, to H2O has an Em7 = + 820 mV51. However, for the supercomplex, the kinetics of the QH2 oxidation: O2 reduction is determined by the steady state fraction of reduced electron acceptors, FeS and heme bL in CIII2. Assuming that the Em7 values of these electron acceptors are the same in M. smegmatis as in C. glutamicum17, oxidation of MQH2 in M. smegmatis is slightly exergonic, but would become endergonic with UQH2, which would result in a smaller fraction of reduced initial acceptors and thereby a lower activity. Hence, since in the current experiments DCQH2 and DMNQH2 compete for binding to the same (Qo) site, the difference in their Em7 values would explain the difference in turnover numbers. Yet another possibility is that, in contrast to DMNQH2, DCQH2 is stabilized at the more distant MQo1b position and does not move to MQo1a, which is required to allow electron transfer to the FeS center. Qualitatively, the same explanation would explain the decrease in activity upon addition of DCQH2 to cytochrome bd-I, but we were not able to establish whether or not DCQH2 and DMNQH2 bind to the same site in this enzyme. Inhibition by ubiquinol of the M. tuberculosis cytochrome bd was also recently indirectly observed52.

As previously mentioned, targeting respiration with antimycobacterial drugs is complicated because of the branched respiratory chains of actinobacteria, which also adapt to environmental conditions36,43,53. In other words, more potent killing of these bacteria would require inhibition of both branches43,54,55. Identifying inhibitors of mycobacterial respiration is also challenging due to the complexity and redundancy of the M. tuberculosis respiratory chain, which allows the bacterium to adapt to various stress conditions56. The pathogen’s metabolic heterogeneity and ability to persist in different physiological states further complicate drug targeting. Additionally, ensuring that inhibitors are effective within the hostile intracellular environment of macrophages, while remaining selective enough to avoid host mitochondrial toxicity, is a major hurdle. Resistance development and limitations in current screening assays also pose significant obstacles to discovering effective and safe respiratory inhibitors.

Earlier studies demonstrated partial inhibition of M. smegmatis growth when the CIII2CIV2 respiratory branch was blocked57, supporting the feasibility of potential drug candidates such as telacebec (Q203) or lansoprazole sulfide (LPZS)10,37,40,41,42. In the current study, we observed an almost complete growth inhibition of M. smegmatis in the presence of DCQ, although inhibition of the CIII2CIV2 respiratory branch by DCQ occurred at higher IC50s compared to those of e.g., Q203 or LPZS37. Remarkably, DCQ also inhibited the growth of virulent M. tuberculosis bacilli in human macrophages, which suggests that the compound may be used as an adjuvant during treatment of tuberculosis (TB) disease. The mechanism of this inhibition is presently unknown, but it is possibly related to the effect of DCQ on DMNQH2 oxidation: O2 reduction activities of both the CIII2CIV2 supercomplex and cytochrome bd branches. A previous study reported that coenzyme Q10 stimulated the growth of M. tuberculosis H37Rv58, which contrasts sharply with our findings that DCQ inhibits M. tuberculosis growth. However, the experimental conditions differed significantly. The earlier work assessed planktonic M. tuberculosis growth in liquid medium supplemented with both pyruvate and succinate, and observed growth stimulation only when both substrates were present58. In contrast, our study investigated intracellular M. tuberculosis growth within human monocyte-derived macrophages at a low MOI, a model that more closely reflects physiological conditions. Notably, our culture medium (RPMI) did not contain succinate. Although we did not directly assess the effects of DCQ on human macrophages in this study, it is also plausible that appropriate doses of DCQ may support immune cell function. Native coenzyme Q10 has been reported to enhance phagocytic and antimicrobial activity, improve mitochondrial energy production, and exhibit antioxidant and anti-inflammatory properties, and support of lysosomal function59,60. Similarly, its analog DCQ has been shown to reduce reactive oxygen species (ROS) production in human cells and mitigate oxidative stress in rat serum samples61. DCQ has also demonstrated a potent ability to inhibit angiogenesis62, which may interfere with the chronic inflammatory milieu63 that is central in TB pathogenesis. As such, these mechanisms could contribute to improved control of intracellular M. tuberculosis growth.

While coenzyme Q10 is generally regarded as safe and beneficial at physiological or therapeutic concentrations60, the use of coenzyme Q10 or DCQ at elevated doses in cell culture may raise concerns, including oxidative stress, mitochondrial dysfunction, altered cell signaling, and potential disruption of lysosomal integrity61,64,65. Accordingly, our results indicate that both DCQ and DNMQ can compromise host cell integrity at higher concentrations. However, we did not observe such adverse effects within the lower dose range of DCQ, up to 100 µM. Notably, these lower concentrations were equally effective in reducing M. tuberculosis growth compared to the higher dose of 250 µM. In macrophage-based cell culture experiments, DCQ is typically used at concentrations ranging from 1 µM to 100 µM. Further investigations are warranted to explore the potential benefits as well as adverse effects of DCQ on the viability and function of M. tuberculosis-infected macrophages across a range of concentrations.

As ubiquinone plays a central role in the mitochondrial electron-transfer chain, therapeutic regimens that target bioenergetics have shown promising potential to enhance bactericidal activity against M. tuberculosis, with the added prospect of shortening treatment duration66. UQ analogs or modulators could contribute to such approaches, particularly if bacterial respiration is selectively impaired without harming host cells. Some anti-TB drugs, such as bedaquiline, directly target components of the bacterial electron-transfer chain, while others, like rifampicin, isoniazid and streptomycin, interfere with cell wall synthesis and protein translation. Combining agents that disrupt different aspects of M. tuberculosis physiology, including electron-transfer chain components, can result in additive or synergistic killing effects. Coenzyme Q10 and its analogs may enhance the efficacy of existing anti-TB drugs not only by targeting M. tuberculosis bioenergetics but also by supporting host immune responses, reducing pathological inflammation and reducing drug-induced host toxicity59,60,67,68,69. Coenzyme Q10 has also been shown to improve mitochondrial and lysosomal function as well as phagocytic activity in macrophages59,70,71, potentially boosting antimicrobial activity60,72. Its antioxidant properties may contribute to protecting host cells from the cytotoxic effects associated with certain TB drugs60, thereby contributing to more effective and better-tolerated combination therapies. It remains unclear whether DCQ retains these beneficial effects in Mtb-infected immune cells. Further studies will deepen our understanding of the mechanistic action of DCQ to block mycobacterial respiration and uncover potential synergistic or additive effects with conventional antibiotics for TB treatment.

Materials and methods

Mycobacterial strains and cultures

M. smegmatis mc2155 cells transformed with qcrCAB/pST-KT (C-terminally FLAG-tagged on QcrB), were grown on 7H10 73 agar plates supplemented with 10% albumin-dextrose-saline (ADS) and antibiotics (25 μg ml−1 kanamycin) at 37 °C. Single colonies were picked from the agar plates, inoculated into 25 ml culture, and shaken at 180 rpm, 30 °C. The pre-cultures were diluted after 48 h into 1 L 7H9 medium in a 2-L flask, and shaken at 180 rpm at 30 °C for ~72 h. The 7H973 medium was supplemented with ADS, 25 μg ml−1 kanamycin, 50 μg ml−1 hygromycin, 0.2% glycerol, and 0.05% Tween 80. Cells were harvested (6000 x g, 30 min; JLA 8.1 rotor (Beckman CoulterTM)) when the OD600 reached 2.

M. smegmatis mc2155 cells transformed with pST-KcydABTwinStrep (Strep-tagged on CydB) were grown essentially the same way, except that the large cultures were in 2 L medium in 2.8 L flasks.

M. tuberculosis laboratory strains H37Ra (avirulent) or H37Rv (virulent), transfected with a plasmid carrying a gene encoding GFP (pFPV2), were from ATCC (Rockville, MD). Sub-cultures of mycobacteria in Middlebrook 7H9 media supplemented with 10% Oleic Albumin Dextrose Catalase (OADC), 0.05% Tween-80, and 20 µg/ml kanamycin (Karolinska University Hospital, Stockholm, Sweden) were incubated at 37 °C for 2 weeks, after which an aliquot was transferred to fresh growth medium for an additional week of expansion. Mycobacterial cultures were washed in 0.05% PBS-Tween-80 followed by pulse-chase sonication to disrupt bacterial clumps. The optical density of the bacterial suspension was measured at 600 nm and adjusted to 0.2–0.3, then used to infect human monocyte-derived macrophages as described below.

Growth of M. smegmatis on plates supplemented with DCQ/DMNQ

The effect of Decylubiquinone (DCQ, CAS no. 55486-00-5) and 2,3-Dimethyl-1,4-naphthoquinone (DMNQ, CAS no. 2197-57-1) on the growth of M. smegmatis cells on solid LB agar was tested by inoculating a 5 ml liquid LB medium with a single colony of M. smegmatis mc2155 strain and growing it for two days at 37 °C with constant shaking at 180 RPM. After 2 days, a series of dilutions was prepared using a liquid LB medium ranging from no dilution to a dilution by a factor of 105. Stocks of 50 mM DCQ/DMNQ were prepared by dissolving 16 mg of DCQ powder (Sigma) in 1 ml ethanol (99.45% ethanol, Sigma) or 16 mg of DMNQ powder (Sigma) in 2 ml ethanol. Subsequently, plates containing either DCQ or DMNQ were prepared by adding a fixed volume of DCQ/DMNQ stocks (7.5 μl for 25 μM, 15 μl for 50 μM) to 15 ml of LB + agar (per plate) before it solidified. Finally, 5 μl of each dilution was plated on the corresponding part of the plate and grown for another two days at 37 °C.

M. smegmatis membrane preparation

For preparation of the CIII2CIV2 supercomplex, cells were homogenized in cell lysis buffer (50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.5 mM EDTA) in the presence of phenyl-methanesulfonyl fluoride and DNase I (Roche), and crushed with a cell disrupter (Constant Systems Ltd.) with four cycles at 35 kPsi. Debris material was removed by centrifugation at 18,600 × g for 30 min. This step was followed by membrane separation using ultracentrifugation at 147,000 × g for 90 min.

For preparation of cytochrome bd-I, approximately 50 g of mc2155 pST-KcydABTwinStrep cells (wet weight) were homogenized in 300 ml of 50 mM HEPES at pH 7.4, 150 mM NaCl, in the presence of one tablet of cOmpleteTM Mini Protease Inhibitor Cocktail (Roche) and a small amount of DNase I. Cell disruption was performed through three to four passes at 35 kpsi using a cell disruptor (Constant Systems Ltd.), followed by centrifugation to pellet cell debris at 15 500 × g for 15 min. The resulting cell membranes were then separated via ultracentrifugation at 215,000 × g for 90 min.

Isolation of M. smegmatis supercomplexes and cytochrome bd

To isolate the CIII2IV2 supercomplex, membranes (1 g) were mixed with 10 ml of a solubilization buffer 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5 mM EDTA, 2% (w/v) GDN (Anatrace) and incubated at 4 °C for 24 h under stirring. Insolubilized debris material was removed by ultracentrifugation at 147 000 × g for 30 min. The supernatant was then applied to a previously equilibrated anti-FLAG M2 column (1 ml, Sigma Aldrich). The column was washed with four column volumes of washing buffer containing 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5 mM EDTA, 0.01% (w/v) GDN. Five column volumes of elution buffer composed of the washing buffer supplemented with 5 mg ml−1 FLAG peptide (Sigma Aldrich) were used to elute the protein, which was further concentrated with a 100-kDa molecular weight cut-off concentrator (Merck Millipore). The protein samples were aliquoted, flash frozen in liquid N2 and stored at −80 °C until use.

To isolate cytochrome bd-I, harvested membranes (typically 5–10 g) were resuspended in 50 mM HEPES at pH 7.4, 150 mM NaCl to achieve a final protein concentration of 10 mg/ml, and were solubilized by the addition of 1% (w/v) N-Dodecyl-β-D-maltoside (DDM) with gentle stirring for 60 min. Any insolubilized material was removed by centrifugation at 215,000 × g for 60 min, after which the supernatant was subjected to affinity chromatography. The solubilized protein was then applied to a gravity column (2 ml Strep-tactin Superflow Plus, VWR) that had been pre-equilibrated with a buffer composed of 50 mM HEPES at pH 7.4, 150 mM NaCl, and 0.05% DDM. After sample loading, the column was washed with 10 ml of the buffer used for pre-equilibration of the column and the elution was performed using 10 ml of the same buffer supplemented with 10 mM desthiobiotin.

Activity measurements

The activity of the purified M. smegmatis supercomplex (O2 reduction: MQH2 oxidation) was measured by following the O2 reduction rate upon addition of reduced 2,3-dimethyl-[1,4] naphthoquinone (DMNQH2) (Rare Chemicals GmbH) using a Clark-type oxygen electrode (Hansatech instruments). The reaction was started by the addition of 5 μL of 20 mM DMNQH2 into an electrode chamber containing 1 ml protein solution in a buffer composed of 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5 mM EDTA, 0.01% (w/v) GDN at 25 °C. The activity was obtained from the initial slope of the graph, where the O2 concentration was linearly dependent on time. DMNQH2 is spontaneously oxidized by O2, but the auto-oxidation rate decreases upon addition of 0.5 μM bovine superoxide dismutase (SOD), which was used routinely in our measurements according to the assay described in detail in ref. 10. We first monitored the O2-reduction rate by DMNQH2 before addition of the supercomplex and this slope was subtracted from that obtained after addition of the supercomplex. The difference reflects the quinol oxidation: O2 reduction activity of the supercomplex10.

The oxidoreductase activity of cytochrome bd-I was measured in a solution composed of 50 mM HEPES pH 7.4, 150 mM NaCl, 0.05% DDM at 25 °C with the use of a Clark-type electrode (Hansatech instruments). Reduced DMNQH2 was used as the substrate. A volume of 2.5–15 μL of a 20 mM stock solution of reduced DMNQH2 was added in order to monitor the background oxidation rate. The reaction was then initiated by addition of cytochrome bd-I (~100 nM). Reduction of DMNQ and DCQ were performed as described in ref. 8. Briefly, quinone was dissolved in ethanol to yield a 20 mM solution. Several crystals of sodium borohydride (NaBH4) were added to reduce the quinone to quinol. The solution was kept on ice until transparent and HCl was added until formation of small bubbles in the solution ended. The sample was centrifuged for 10 min at 10,000 x g, and the supernatant containing reduced quinol was aliquoted, flash frozen, and stored in −80 °C until use.

M. tuberculosis macrophage infection model

Buffy-coat blood was obtained from healthy adult donors at the Karolinska Hospital Blood Bank, Stockholm, Sweden, with informed consent and ethical approval (2010/603-31/4). Lymphoprep™ (Alere technologies, Norway) is a density gradient medium that was used for the isolation of mononuclear cells from peripheral blood. Monocytes were separated from bulk mononuclear cells using 2–3 h plastic adherence in T75 flasks (Thermo Fisher Scientific, Waltham, MA) or 6-well plates at 37 °C in serum-free media (RPMI 1640, VWR, Radnor, PA). Non-adherent cells were washed off, and Macrophage colony-stimulating factor (M-CSF, 50 ng/ml) was supplemented to the cell culture media (RPMI including 10% fetal calf serum (FCS), 2 mM L-glutamine, 1 mM sodium pyruvate, 5 mM HEPES) to allow differentiation of monocytes to macrophages for 7 days. Human monocyte-derived macrophages were detached from the cell culture flasks or plates using 2 mM EDTA buffer including 2.5% FCS for 30 min at 37 °C and used in experiments as described below.

High-content imaging with IncuCyte

For high-content analysis of macrophages with IncuCyte, cells from two healthy donors (n = 2) were prepared in T75 flasks and re-seeded at a density of 4000 cells/well in optilux Black/Clear Flat bottom, TC-treated 384-well plates (Falcon Corning, Corning, NY). Cell cultures were infected with H37Rv-GFP at a multiplicity of infection (MOI) of 1 and treated with DCQ or DMNQ in triplicates. The untreated MOI1 infection control was cultured in medium with diluent only. Mycobacterial growth (green fluorescent signal) and cell morphology were assessed using high-content imaging with the IncuCyte® S3 Live-Cell Analysis System (Sartorius, Göttingen, Germany) after 4 days of incubation of the plates at 37 °C. Image acquisition was based on a single image of the whole well that was captured using a 4× objective (original magnification) at a single time point on day 4 post-infection. The fluorescent signal from the mycobacteria was quantified as total integrated intensity (GCU x µm2/image) using the IncuCyte S3 software. Data (median and range) from n = 2–4 donors was plotted and analyzed in GraphPad Prism 9 (version 9.5.1) as the percent (%) intracellular M. tuberculosis growth in macrophages relative to the untreated MOI infection control (set to 100%). IncuCyte results were analyzed using a 2-way ANOVA and Tukey’s multiple comparisons test to calculate the indicated p-values.

Flow cytometry

For flow cytometry analysis of macrophages, cells from two healthy donors (n = 2) were cultured in 6-well plates at a density of 1 million cells per well (Falcon Corning). Cell cultures were infected with H37Ra-GFP at an MOI of 5 for 4 h, followed by washing and treatment with either DCQ or DMNQ. Untreated M. tuberculosis-infected macrophages cultured in medium with diluent only served as positive controls, while uninfected macrophages were used as negative controls. After 48 h of incubation, the cells were washed with FACS buffer (PBS, 2% FCS, 0.05 mM EDTA) and stained for 15 min with LIVE/DEAD Near IR Viability Dye (APC-Cy7) (Thermo Fisher Scientific). An Fc-blocking cocktail (Miltenyi, Germany) was used to minimize nonspecific binding. Red blood cells were lysed using a fixation/permeabilization working solution (BD Cytofix/Cytoperm™, Franklin Lakes, NJ), and the cells were resuspended in FACS buffer. All stained samples were acquired on an LSR Fortessa (BD Biosciences) at low to medium speed. Macrophages infected with H37Ra-GFP bacteria were detected in the FITC channel during flow cytometry analysis (see Supplementary Fig. S1). Gating strategies included singlet isolation (to exclude doublets) and viability gating (to exclude dead cells). The fluorescence intensity of the GFP signal was visualized in a histogram and quantified as the geometric mean (GM).

Cryo-EM sample preparation and data collection

GDN solubilized purified supercomplex protein sample (3 μl of 0.5 mg/ml) was mixed with 30 μL 200 mM stock solution of DCQ, incubated for 5 min at room temperature and concentrated to a final concentration of 3.5 mg/ml. This DCQ-containing sample was used for cryo-EM experiments. Copper EM grids coated with holy carbon film (300 mesh R2/2 grid, Quantifoil, Micro Tools GmbH, Germany) were used for the sample preparation. The grids were glow-discharged in air at 20 mA for 120 s (PELCO easiGlow). Vitrobot Mark VI (Thermo Fisher Scientific) was used for the sample blotting (2 μL of the protein, 3 s at 4 °C, 100% humidity) and plunge-freezing in liquid ethane. Titan Krios G3i electron microscope (Thermo Fisher Scientific) Ceta-D camera and a Gatan K3 BioQuantum detector was used for data collection. The dataset was acquired in electron-counting mode at a nominal magnification of 10,5000 (0.825 Å/pixel) and contained 15,000 exposures (40 exposure fractions each). EPU software package (v2.12.1; Thermo Fisher Scientific) was used for automated data collection with camera exposure rate was 16.5 e−/pixel/s and total exposure of the sample 40 e−/Å2 (Supporting information Supplementary Fig. S2 and Table 1).

Table 1 Data collection and processing of cryo-EM data
Full size table

Cryo-EM data processing, structure building, and refinement

CryoSPARC v474 was used for the processing of the dataset, with motion correction and CTF estimation performed using the Patch motion correction and Patch CTF estimation, as implemented in cryoSPARC. Initial particles used for the generation of initial 2D classes, later used by the Template picker job, were picked from 1000 images using the Blob picker job in cryoSPARC. The initial set of particles (4,072,098) was reduced by 2D classification to 1,521,965 particles, of which 55,9531 were selected for another round of 2D classification. Resulting group of 187,202 particles was used for ab initio reconstruction set for 3 classes. The best class of ab initio reconstruction contained 99,481 particles, which were re-extracted from micrographs with box size 540 px without Fourier crop. The reextracted particles were separated into another two classes using ab initio reconstruction. The best class contain final set of 79,220 particles. These were further refined using homogenous refinement, followed by local CTF refinement without symmetry enforced and a second round of homogenous refinement. The resolved final map had an overall resolution of 2.54 Å. The initial structure model used for map interpretation was from a previously solved structure of the supercomplex with the PDB code 8OVD42. The model was refined against the map using Coot75 and real-space refinement in Phenix76.

Statistical analysis and reproducibility

Statistical differences comparing different doses of compounds (median ± range) were determined using a 2-way ANOVA and Tukey’s multiple comparisons test. All values are presented as mean ± SD, derived from at least three independent experiments.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.