Cryo-EM structure of Neurospora crassa respiratory complex IV

The cryo-EM structure of Neurospora crassa respiratory complex IV was determined to 5.5 Å resolution and is compared with related structures from Saccharomyces cerevisiae and Bos taurus.


Introduction
The mitochondrial respiratory chain consists of NADH dehydrogenase (complex I), succinate dehydrogenase (complex II), cytochrome c reductase (complex III) and cytochrome c oxidase (complex IV). Mitochondrial complex IV is a 200 kDa heme-copper oxidase with up to 14 different polypeptides (Balsa et al., 2012). Complex IV catalyzes the reduction of molecular oxygen to water and conserves free energy by pumping protons across the mitochondrial inner membrane (Wikstrom, 1977). Owing to the high-resolution crystal structure of bovine heart complex IV (Tsukihara et al., 1995(Tsukihara et al., , 1996, the molecular mechanisms of the three central subunits as a redox-coupled proton pump are well understood (Yoshikawa et al., 2011;Shinzawa-Itoh et al., 2007;Liang et al., 2017). There is genetic evidence in yeast (Capaldi, 1990) that most of the peripheral subunits function as assembly factors of complex IV. Null mutants yield enzymes with reduced activities compared with the wild type.
In mammalian mitochondria, complexes I, III and IV are known to form supercomplexes of different compositions (Schä fer et al., 2007;Gu et al., 2016;Althoff et al., 2011;Sousa et al., 2016;Letts et al., 2016) that are sometimes referred to as respirasomes. Other eukaryotes and fungi have characteristic variants. These include the supercomplex of Saccharomyces cerevisiae, a symmetric assembly of complex III and complex IV that has recently been studied by highresolution cryo-EM (Hartley et al., 2019;Rathore et al., 2019). In S. cerevisiae complex I is replaced by three non-protontranslocating alternative peripheral membrane NADH: ubiquinone oxidoreductases exposed on the inner or outer surface of the mitochondrial inner membrane (Joseph-Horne et al., 2001;Iwata et al., 2012;Feng et al., 2012). Unlike S. cerevisiae, its close relative the filamentous fungus Neurospora crassa has a complex I (Videira, 1998;Marques et al., 2005). Biochemical analysis of digitonin-solubilized mitochondria from N. crassa by blue native gel electrophoresis revealed assemblies of complexes I and IV comigrating with ATP synthase dimers, supercomplexes of complexes III and IV, and supercomplexes of complexes I, III and IV with different stoichiometries (Marques et al., 2007). The lowresolution structures of N. crassa complex I and complex III have been studied by electron microscopy of negatively stained particles (Gué nebaut et al., 1997) or electron crystallography of two-dimensional crystals (Leonard et al., 1987). However, the architecture of N. crassa complex IV or its respiratory supercomplexes have not been explored. We now report the 5.5 Å resolution structure of complex IV from N. crassa reconstituted into lipid nanodiscs, as determined by single-particle cryo-EM. Our results underline the conserved molecular mechanism by which complex IV facilitates the reduction of molecular oxygen to water in mammals and fungi.

Materials and methods
2.1. Growth of cells and preparation of mitochondria N. crassa (strain GR-107) was grown and mitochondria were isolated as described previously (Sebald et al., 1979;    Kü nkele et al., 1998; Ahting et al., 1999). Briefly, 1 kg (wet weight) of hyphae were homogenized in 250 mM sucrose, 2 mM EDTA, 20 mM Tris pH 8.5, 1 mM PMSF in a Waring blender at 4 C. 1 kg of quartz sand was added and the cell walls were disrupted by passing the suspension twice through a corundum stone mill. Cellular residues were pelleted and discarded in two centrifugation steps (4000g) for 5 min at 4 C. The mitochondria were sedimented in 250 mM sucrose, 2 mM EDTA, 20 mM Tris pH 8.5, 1 mM PMSF at 17 000g for 80 min. This step was repeated to improve the purity. The isolated mitochondria were suspended in 250 mM sucrose, 20 mM Tris pH 8.5, 1 mM PMSF at a final protein concentration of 50 mg ml À1 , shock-frozen in liquid nitrogen and stored at À20 C.

Protein purification
Complex IV was accidentally copurified from mitochondrial membranes of N. crassa (Kü nkele et al., 1998;Ahting et al., 1999) by immobilized metal-affinity chromatography (IMAC) without introducing an affinity tag, most likely on account of the internal histidine-rich region in the Cox4 subunit ( Supplementary Fig. S3). Isolated N. crassa mitochondria (2 g) were solubilized in 20% glycerol, 20 mM Tris pH 8.5, 300 mM NaCl, 20 mM imidazole, 1%(w/v) DDM and four tablets of cOmplete (EDTA-free) protease inhibitor (Roche) for 1 h at 4 C. The solution was centrifuged at 150 000g for 40 min at 4 C and the resulting supernatant was incubated with 10 ml Nickel-complexed Chelating Sepharose (GE Healthcare) for 30 min. The resin was washed with 10% glycerol, 50 mM HEPES pH 7.2, 0.1%(w/v) DDM, 1 mM PMSF. The protein was eluted with 300 mM imidazole in the same buffer. The protein concentration of the yellow protein fraction was 1.9 mg ml À1 . A 5 ml volume was used for reconstitution into nanodiscs with the membrane-scaffold protein MSP2N2.

Membrane-scaffold protein purification
The gene for MSP2N2 was cloned into a pET-28a vector with an N-terminal His tag and a TEV cleavage site. The protein was expressed in Escherichia coli BL21(DE3) cells for 3 h in TB medium at 37 C after induction with IPTG. The cells were disrupted in 50 mM Tris pH 8, 150 mM NaCl, 1 mM PMSF, 1% Triton X-100. After high-speed centrifugation, the supernatant was incubated with Nickel-complexed Chelating Sepharose (GE Healthcare) and washed in three steps with 40 mM Tris pH 8, 300 mM NaCl, 1% Triton X-100, then with 40 mM Tris pH 8, 300 mM NaCl, 50 mM sodium cholate, 20 mM imidazole and finally with 40 mM Tris pH 8, 300 mM NaCl, 40 mM imidazole. MSP2N2 was purified by IMAC using the N-terminal His tag. The protein was eluted with 40 mM Tris pH 8, 300 mM NaCl, 400 mM imidazole. After dialysis against 20 mM Tris pH 8, 100 mM NaCl overnight, the His tag was cleaved off using TEV protease and separated from the cleaved MSP2N2 by a second IMAC step (His-Select, Sigma) using the unbound flowthrough.

Reconstitution of complex IV into lipid nanodiscs
For reconstitution, 950 ml of lipids [4 mM POPC, 3.5 mM POPE, 2.5 mM cholesterol in 0.5%(w/v) DDM] and 1.97 ml MSP2N2 (4 mg ml À1 ) were added to the eluted mitochondrial protein, resulting in a final molar protein:MSP2N2:lipids ratio of 1:3:150. 7.5 ml of the mixture was incubated for 1 h at room temperature before the detergent was removed by incubation with 500 mg BioBeads SM-2 (Bio-Rad) for 2.5 h. The solution was dialyzed against 20 mM HEPES pH 7.2, 20 mM KCl, 5 mM -mercaptoethanol overnight. The sample was centrifuged for 5 min to remove aggregated protein and concentrated to a final volume of 300 ml using a 100 kDa cutoff Amicon centrifugation tube (Merck). 50 ml of sample was applied onto a Superose 6 Increase 3.2/300 gel-filtration column (GE Healthcare) and yellow complex IV fractions were collected for cryo-EM sample preparation.

Cryo-EM data acquisition
The purified protein-nanodisc complex was applied onto C-Flat 1/1 grids (400 mesh) and plunge-frozen in a Vitrobot (ThermoFisher) with 7 s blotting. Data were automatically acquired at 300 kV with Leginon (Suloway et al., 2005) in a Polara electron microscope (FEI) equipped with an energy filter (slit width of 70 eV) and a K2 direct electron detector (Gatan) in counting mode. A total of 3467 micrographs were recorded, of which 2835 were selected manually for further processing. A typical micrograph is shown in Fig. 1(a). Dataacquisition and processing parameters are summarized in Table 1.

Image processing
Micrographs were processed with RELION-3 (Zivanov et al., 2018). Particles were picked automatically and extracted with a box size of 192 Â 192 pixels. Subsequent classification in 2D (100 classes) and selection resulted in 2 418 748 particles with no obvious ice contamination or carbon features. Particles were 3D-classified into six classes, with a first model generated ab initio in RELION-3 using 109 979 particles from ten selected 2D class averages as a reference [see Fig. 1(b)]. 1 656 842 selected particles were classified again into six classes, resulting in one single, homogeneous class with 426 592 particles that were refined by gold-standard autorefinement in RELION-3 to 7 Å resolution. Bayesian polishing was applied to the refined particles (Zivanov et al., 2019). The polished particles were classified into three classes, resulting in 134 734 particles (31%) with lower noise levels, clearer protein features and a homogeneous distribution of orientations. A final 3D refinement step resulted in a map with a gold-standard resolution of 5.5 Å [FSC cutoff at 0.143; see Fig. 1(c)]. The final map was sharpened with a B factor of À370 Å 2 . The map has been deposited in the EMDB under code EMD-4720.

Model fitting and structure comparison
Atomic models of complex IV from S. cerevisiae (PDB entry 6hu9, chains a-l) and Bos taurus (PDB entry 1occ, chains A-M) were fitted into the density by rigid-body docking in UCSF Chimera (Pettersen et al., 2004;Figs. 3 and 4). The two models were converted to a simulated map at 5.5 Å resolution before fitting. The correlation values were 0.78 for the S. cerevisiae model and 0.72 for the B. taurus model. The N. crassa map density was colored by subunit (S. cerevisiae) and visualized with ChimeraX (Goddard et al., 2018; Fig. 2). For comparison, homologous protein sequences (from UniProtKB) of N. crassa complex IV subunits were aligned with the S. cerevisiae subunits using Clustal Omega 1.2.4 (Sievers & Higgins, 2014; see Supplementary Figs. S1-S5).

Overall structure
The complex that was accidentally copurified from N. crassa mitochondria was identified as complex IV by cryo-EM and image processing. The 5.5 Å resolution map showed clear density for ten different complex IV subunits and the membrane-scaffold protein (Fig. 2). The transmembrane core subunit Cox1 has 12 transmembrane helices (see also Supplementary Fig. S1) surrounding two strong, globular densities for the two bound hemes that were already visible in 2D class averages [ Fig. 1(b)]. The second core subunit Cox2 consists of two transmembrane helices ( Supplementary Fig.  S2) and a large domain in the intermembrane space (IMS). With seven transmembrane helices ( Supplementary Fig. S2) Cox3 is the second-largest subunit, with one loop protruding into the matrix. Cox4, Cox6 and Cox6b do not have transmembrane helices (Supplementary Fig. S3) and are either attached on the IMS or the matrix side of the complex. Cox4 and Cox6 are the major soluble domains on the matrix side. transmembrane spans (Supplementary Fig. S4). Cox5a has domains on both the matrix and the IMS sides.

Comparison with complex IV structures from S. cerevisiae and B. taurus
The structure of N. crassa complex IV is very similar to that from S. cerevisiae, which was purified as part of a supercomplex (Hartley et al., 2019;Rathore et al., 2019). The atomic model of the yeast complex (PDB entry 6hu9) fitted the cryo-EM density without adjustment. The N. crassa complex was present in lipid nanodiscs in one single conformation. Complexes of different subunit compositions were not identified in the 3D classes, indicating that the lipid environment did not result in significant structural differences compared with complex IV in the S. cerevisiae supercomplex. The membranescaffold protein (MSP) was easily visible as a double belt around the transmembrane helices (Figs. 2 and 3), which rules out the presence of additional subunits in the N. crassa complex. As in other cryo-EM structures of membraneprotein complexes (Hahn et al., 2018), a single scaffold protein chain wraps around the assembly twice. In our case, it is roughly twice as long as the circumference of complex IV.
The N. crassa complex IV has only ten subunits, two fewer than that from S. cerevisiae. As shown in Fig. 3, our map had no densities for S. cerevisiae subunit Cox13 or the recently identified subunit Cox26 (Levchenko et al., 2016). The N. crassa genome contains a homologous sequence for Cox13 but not for Cox26 ( Supplementary Fig. S5), which is therefore absent, as is also the case in the bovine complex (Luo et al., 2017;Fig. 4). Cox6a in the bovine structure is homologous to Cox13 and is found in an analogous location, but its structure is slightly different. In S. cerevisiae it is present in one supercomplex structure (Hartley et al., 2019) but not in the other (Rathore et al., 2019), suggesting a possible regulatory role. Cox13 may be involved in complex IV dimer formation and might not always be expressed (Hartley et al., 2019). This would explain its absence in our map of the isolated monomer.
The crystal structure of B. taurus complex IV (PDB entry 1occ; Tsukihara et al., 1996) fits our map less closely. Three subunits of the bovine crystal structure, Cox6a, Cox7b and Cox8b, do not fit into the N. crassa map (Fig. 4). Cox7b and Cox8b are absent in the S. cerevisiae model, indicating that they are not needed in fungi. The N. crassa structure therefore shows the ten core subunits of complex IV that are present in most, if not all, other eukaryotes and appear to be a minimal set. It is safe to assume that the function of the central subunits Cox1, Cox2 and Cox3 in catalyzing electron transfer and proton pumping is identical in S. cerevisiae, N. crassa and mammals. Presumably, the seven other subunits are necessary for complex IV assembly or supercomplex formation. It will be interesting to investigate whether some of these subunits mediate interaction with complex I in the N. crassa supercomplex, as in mammalian respirasomes.
A respiratory supercomplex for N. crassa has been described biochemically (Marques et al., 2007), whereas cryo-EM maps have been determined for the S. cerevisiae and B. taurus supercomplexes (Hartley et al., 2019;Rathore et al., 2019;Sousa et al., 2016;Gu et al., 2016). A major difference in the structures of complex IV is visible in the matrix domain of Cox5a. It has been suggested that this domain mediates interaction with complex III in the S. cerevisiae supercomplex (Hartley et al., 2019). The overall interaction between complex III and complex IV in the S. cerevisiae supercomplex is remarkably different from that in the bovine supercomplex, possibly because of a difference in the structure of Cox5a. Since Cox5a of yeast and N. crassa have the same structure (Fig. 5) and the N. crassa complex IV is unassociated and incorporated into lipid nanodiscs, the observed difference is unlikely to be owing to an interaction with complex III. However, such an interaction is likely to occur in the N. crassa supercomplex.
Another striking difference in the N. crassa structure compared with that from S. cerevisiae is an additional rodshaped density between the matrix domains of Cox4 and Cox5a (gray density in Fig. 2). Since the sequence of N. crassa Cox4 is longer by 29 residues than that from S. cerevisiae ( Supplementary Fig. S3), we suggest that this density belongs to the C-terminal tail of N. crassa Cox4, although at 5.5 Å resolution we cannot rule out that it belongs to another     subunit. The density may be unique to N. crassa, as it is also absent in bovine complex IV (Fig. 4).

Conclusion
With recent technological and computational developments (Kü hlbrandt, 2014;McMullan et al., 2016;Elmlund et al., 2017;Zivanov et al., 2018), cryo-EM is developing into a method that can identify protein complexes of suitable size in heterogenous preparations quickly and unambiguously: in our case, N. crassa complex IV that was inadvertently copurified by nickel-affinity chromatography was identified by its similarity to the known structures of the S. cerevisiae and B. taurus complexes. It is particularly similar to that from S. cerevisiae, even though it lacks two of its subunits. The ten core subunits of the N. crassa complex are present in both the S. cerevisiae and B. taurus complexes. The major difference compared with the B. taurus complex is the structure of the matrix domain of the Cox5a subunit, which appears to be typical for a fungal complex IV. The subunit may mediate contact with complex III in a supercomplex. Unlike S. cerevisiae, N. crassa has complex I and may therefore form a supercomplex that combines features of the S. cerevisiae and mammalian supercomplexes. The major difference of the N. crassa complex IV is the density between the Cox4 and Cox5a subunits, which presumably belongs to the histidine-rich C-terminus of Cox4. The functional role of this feature is unknown.
Although the structures of N. crassa and S. cerevisiae complex IV are largely identical, it would be interesting to investigate the structure of a fungal supercomplex that includes complex I, which is absent in S. cerevisiae (Sousa & Vonck, 2019) but is present in N. crassa. The N. crassa supercomplex may be more similar to the mammalian respirasome than to the S. cerevisiae supercomplex.