NMR structural analysis of the yeast cytochrome c oxidase subunit Cox13 and its interaction with ATP

Background Mitochondrial respiration is organized in a series of enzyme complexes in turn forming dynamic supercomplexes. In Saccharomyces cerevisiae (baker’s yeast), Cox13 (CoxVIa in mammals) is a conserved peripheral subunit of Complex IV (cytochrome c oxidase, CytcO), localized at the interface of dimeric bovine CytcO, which has been implicated in the regulation of the complex. Results Here, we report the solution NMR structure of Cox13, which forms a dimer in detergent micelles. Each Cox13 monomer has three short helices (SH), corresponding to disordered regions in X-ray or cryo-EM structures of homologous proteins. Dimer formation is mainly induced by hydrophobic interactions between the transmembrane (TM) helix of each monomer. Furthermore, an analysis of chemical shift changes upon addition of ATP revealed that ATP binds at a conserved region of the C terminus with considerable conformational flexibility. Conclusions Together with functional analysis of purified CytcO, we suggest that this ATP interaction is inhibitory of catalytic activity. Our results shed light on the structural flexibility of an important subunit of yeast CytcO and provide structure-based insight into how ATP could regulate mitochondrial respiration. Supplementary Information The online version contains supplementary material available at 10.1186/s12915-021-01036-x.

-Representative sequential backbone assignment of Cox13. Strips for residues F106-G111 from the overlaid 3D-HNCACB (C α in orange and C β in pink) and 3D-HNCOCACB (resonance in green) spectra of [ 15 N, 13 C]-labelled Cox13 at a concentration of 0.4 mM in 20 mM NaP i pH 6.5, containing 50 mM L-Arg, 50 mM L-Glu, 1 mM DTT and 30 mM DPC, recorded at 40 °C and 900 MHz. The lines indicate sequential connectivities. Example of an overlaid 1 H-1 H slice taken from the 3D 15 N-edited NOESY spectrum (green) and the 3D H(CCO)NH spectrum (pink). The assigned side-chain resonances of residue N100 and NOE resonances of residue I101 are indicated. The observed long-distance NOE correlation between residue I101 and L115 is shown in the 3D 13 C-edited NOESY spectrum (blue). The diagonal peak is marked with a red cross, and correlations are indicated by red dashed lines. Figure S4 -Representative slices from the 3D 15 N-edited/filtered NOESY spectra. Examples of overlaid 1 H-1 H slices for residue F72 (A) W51 (B) from the 3D 15 N-edited NOESY spectrum (cyan) and the 3D F1-13 C/ 15 N-filtered, F3-15 N-edited-NOESY spectrum (red). The corresponding inter-monomer NOE correlations are mapped onto the Cox13 structure by pink dashed lines. The monomer subunits of Cox13 structure are shown in cyan and grey, respectively. The diagonal peak in (B) is marked with a red cross.

Figure S5 -Paramagnetic spin-label titrations.
Resonance intensity ratios as a function of sequence calculated from 1 H-15 N TROSY-HSQC spectra recorded on samples of 0.5 mM 15 N-labelled Cox13 in DPC micelles in the presence or absence of 5 mM gadodiamide (upper) or 5 mM 16-DSA (lower). For more details, see Additional file 4. Residues of regions assigned as random coil, soluble helix, and transmembrane helix are coloured purple, orange, and red, respectively. Error bars indicate relative uncertainties calculated by error propagation and curves are weighted moving averages, which take sequential distance and uncertainties into account. The weighted moving averages' colour scales go from blue to green to indicate regional partitioning into aqueous or hydrophobic environments, respectively. To illustrate the overall partitioning suggested by the Gd and DSA data, which is similar with only minor local differences, the Cox13 dimer structure has been colour-coded according to the values of the moving averages: the left subunit according to the upper (Gd) panel, and the right subunit according to the lower (DSA) panel. illustrating the range of possible fits. For the wildtype with ADP, the data was too scattered to give a reliable fit.

Figure S7 -Turnover in wildtype and Cox13Δ CytcO as a function of added yeast cytochrome c.
The same data as in Figure 4 (main text) but plotted comparing the data with and without ATP in each variant directly and without normalising to the maximum activity at 100 µM cyt. c. The fits are the same as those shown in Figure 4.

Figure S8 -Interaction of Cox13 with ATP.
Overlaid 15 N-HSQCs from an ATP titration to Cox13. Spectra are colour-coded with ATP/Cox13 ratios according to: 0 (grey), 5 (cyan), 10 (green), 25 (yellow), 50 (orange), and 100 (red). Arrows indicate directions of the chemical shift change for selected residues.  respectively. For more details, see Additional file 6. (B) CSP as a function of ADP concentration for selected residues. The curves are fitted as in Figure 5 (main text) with a global K d = 17±2 mM. The fitted maximum CSPs for residues R81, H83, Y96 and I125 are 0.53±0.03, 0.28±0.02, 0.19±0.02, and 0.27±0.02 ppm, respectively. (C) Parts of 31 P spectra showing signals from the two phosphate groups of ADP in solutions with (red) and without (grey) Cox13. The red spectrum was recorded on the same sample as used in A. The grey spectrum was obtained on a sample where the ADP stock solution was added in the corresponding amount to NMR-buffer only.

NMR distance and dihedral constraints (a)
Distance constraints