Insights into conformational changes in cytochrome b during the early steps of its maturation

Membrane proteins carrying redox cofactors are key subunits of respiratory chain complexes, yet the exact path of their folding and maturation remains poorly understood. Here, using cryo‐EM and structure prediction via Alphafold2, we generated models of early assembly intermediates of cytochrome b (Cytb), a central subunit of complex III. The predicted structure of the first assembly intermediate suggests how the binding of Cytb to the assembly factor Cbp3‐Cbp6 imposes an open configuration to facilitate the acquisition of its heme cofactors. Moreover, structure predictions of the second intermediate indicate how hemes get stabilized by binding of the assembly factor Cbp4, with a concomitant weakening of the contact between Cbp3‐Cbp6 and Cytb, preparing for the release of the fully hemylated protein from the assembly factors.

Mitochondria allow eukaryotes to use oxidative phosphorylation for energy conversion, relying on the activities of the electron transport chain and the connected ATP synthase.The respiratory chain is located in the inner mitochondrial membrane and consists, in the yeast Saccharomyces cerevisiae (S. cerevisiae), of three major protein complexes.One of them, the mitochondrial cytochrome bc 1 complex, or complex III, is a homodimer of 10 different protein subunits.This enzyme oxidizes ubiquinol and transfers the electrons to cytochrome c in the intermembrane space (IMS).Reduced cytochrome c diffuses from the bc 1 complex to the terminal respiratory enzyme, cytochrome c oxidase (complex IV), where the electrons are used to reduce O 2 .The electron transfer in the bc 1 complex is coupled to proton translocation over the inner membrane through a process known as the Q-cycle [1], which mainly involves the three protein subunits cytochrome b (Cytb), the Rieske Fe-S cluster protein (Rip1), and cytochrome c 1 (Cyt1) [2].
Assembly of respiratory chain complexes is a sophisticated process, as they are assembled in specific steps with protein subunits expressed from either the nuclear or mitochondrial genome (mtDNA) [3].These subunits are then joined together at specific steps to form assembly intermediates, not unlike an industrial assembly line.To achieve this, numerous assembly factors and chaperones, often specific to one protein, help to obtain both the correct fold and support the acquisition of the redox cofactors [3][4][5][6][7][8].Expression of the subunits is highly regulated to avoid stoichiometry problems, leading to a potentially deleterious accumulation of either nuclear precursors or mitochondrially encoded proteins.Consequently, translational control of mitochondrial protein synthesis ensures that the mitochondrial encoded subunits are not overproduced [9][10][11][12].
In the mitochondrial bc 1 complex, Cytb is the only protein that is expressed from the mitochondrial genome, while the other nine subunits are nuclearencoded and imported into mitochondria [4,13].Work in yeast has shown that assembly of the bc 1 complex is initiated with Cytb, around which the other subunits are added sequentially into the growing complex [4].Cytb is a highly hydrophobic protein consisting of eight transmembrane (TM) helices in a N-in, C-in configuration with three distinct loops facing the mitochondrial matrix [14,15].In addition, Cytb contains two b-type hemes, b L (low potential) and b H (high potential), that are coordinated with conserved histidine residues in a four-helix bundle formed by TM helices 1-4 and are essential for electron transfer via the Q-cycle [16].Heme insertion into Cytb occurs in a strict stepwise fashion [17].The process is aided by the binding of two assembly factors, namely the heterodimer Cbp3-Cbp6 and, upon insertion of the first heme, the protein Cbp4 [17].To coordinate the expression of Cytb with the availability of nuclear-encoded subunits, a feedback loop employing Cbp3-Cbp6 regulates its synthesis [18][19][20].So far, this mechanism of Cytb biogenesis has been best described for yeast mitochondria.However, the high conservation of cytochrome b-type redox proteins and Cbp3-Cbp6 across the domains of life [21][22][23], together with recent mechanistic work in human mitochondria [24], points to a common mechanism of Cytb maturation.Though the different steps of bc 1 complex assembly and the corresponding assembly intermediates have been well described, there are still questions about the exact molecular mechanisms that remain unanswered.Structural determination of complex III [14,16] and more recently the respiratory supercomplexes [15,25,26] revealed important insights into these complexes.Nevertheless, detailed knowledge of assembly intermediates is still lacking.
In the past decade, two methods for structural determination have led to a revolution in structural biology, namely single-particle cryogenic electron microscopy (cryo-EM) [27,28] and algorithm-based structure predicting via, e.g., Alphafold2 [29].Hence, we set out to determine the structure of early Cytb assembly intermediates using these approaches.Attempts to determine the structures of these complexes via cryo-EM proved to be hampered due to sample heterogeneity.We therefore used the AlphaFold2 implementation ColabFold [30] to predict structures of the early assembly intermediates.Analyses of these predicted structures indicate conformational changes in Cytb that are triggered by Cbp3-Cbp6 binding, suggesting complex structural rearrangements during the early stages of maturation.

Yeast genetics and cloning
All yeast strains used in this study were isogenic to the S. cerevisiae strain W303.To generate a strain with stalled assembly of the bc 1 complex, we used a previously constructed intron-less strain (MRSI) where the mtDNA have been modified to code for a HA-Hisx6 tag on the C-terminal end of Cytb.This COB-HA-HISx6 strain (MOY122) was previously used [17] to study the timing of heme incorporation in Cytb.For the purpose of stalling assembly intermediate 1, QCR7 was deleted using a KanMX4 resistance cassette through homologous recombination.We further deleted QCR8 and COR1 using the URA3 and HPH resistance cassettes, respectively.The COB-HA-HISx6 Dqcr7Dqcr8Dcor1 strain was transformed with the plasmids pYX132-CBP3-3xFLAG and pYX142-CBP6, to overexpress Cbp3-Cbp6 and stabilize Cytb levels.The pYX132-CBP3-3xFLAG was constructed using Gibson cloning with pYX132-CBP3 as a template.The 3xFLAG insert was created using the forward primer 5 0 -gtaggctgt catatacaaacGAGAATCTTTATTTTCAGGGATCCCCTC AACAAAACAAAACCGCCTCTTGCGCAACACGATG AAGCCGTGTTCGAAGGTCATCATCCAGATTATAA AG-3 0 and reverse primer 5 0 -ggtcgacgcgtaagcttttaTTA TTTATCATCATCATCTTTATAATCAATATCATG-AT CTTTATAATCACCATCATGATCTTTATAATCTGGA TGATGACCTTCGAACACGGCTTCATC.The pYX13 2-CBP3 template was linearized using the restriction enzyme HindIII around the site of insertion.

Cell culturing and mitochondrial isolation
Yeast cells were cultured in 36 L of synthetic minimal media containing 1.7 gÁL À1 yeast nitrogen base, 5 gÁL À1 (NH 4 ) 2 SO 4 , 20 lgÁmL À1 of adenine, uracil and arginine, 15 -lgÁmL À1 of histidine and 30 lgÁmL À1 lysine supplemented with 2% glucose.The cultures were grown overnight to an OD 600 of 3 before harvesting at 3000 g for 5 min.Cells were washed with distilled water, resuspended in MP1 buffer (0.1 M Tris-base, 10 mM dithiothreitol (DTT) according to 2 mLÁg À1 wet weight) and incubated for 10 min at 30 °C.Cells were then washed with 1.2 M sorbitol, resuspended in MP2 buffer (6.7 mLÁg À1 wet weight of 20 mM KPi pH 7.4, 0.6 M sorbitol and 3 mgÁg À1 wet weight of zymolyase (Seikagaku Biobusiness, Tokyo, Japan)) and incubated at 30 °C for 1 h to digest the yeast cell wall.Spheroplasts were harvested at 3000 g for 5 min at 4 °C, resuspended in MP3 buffer (13.4 mLÁg À1 wet weight of 0.6 M sorbitol, 10 mM Tris pH 7.4, 1 mM EDTA and 1 mM PMSF) and homogenized with 2 9 10 strokes using a tight-fitting homogenizer (Sartorius Stedim Biotech GMBH, G€ ottingen, Germany).The homogenate was centrifuged two times at 3000 g for 5 min at 4 °C before mitochondria were harvested by centrifugation at 15 000 g for 15 min at 4 °C.The mitochondrial pellet was resuspended in SH buffer (0.6 M sorbitol and 20 mM HEPES pH 7.4) to a final concentration of 10 mgÁmL À1 and frozen in liquid nitrogen before storage at À80 °C.

Protein purification
The following steps were performed either on ice or at 4 °C.Isolated mitochondria were thawed and resuspended in SH-buffer, centrifuged at 10 000 g for 10 min, and resuspended in 100 mL lysis buffer (50 mM KPi pH 7.4, 150 mM KCl, 10 mM imidazole pH 7.4, 1%/0.1% LMNG/CHS, 1 mM PMSF and 19 complete protease inhibitor) for 1 h tumbling.Lysate was clarified through centrifugation at 20 000 g for 10 min and added to 4 mL of pre-equilibrated Ni-NTA slurry (2 mL beads) and incubated for 1 h tumbling.The sample was poured into a 20 mL gravity flow column and beads were washed with 20 cv of wash buffer (50 mM KPi pH 7.4, 150 mM KCl, 20 mM imidazole and 0.01%/0.001%LMNG/CHS) before incubation with elution buffer (wash buffer and 300 mM imidazole) for 2 9 10 min, tumbling.The sample was diluted 1 : 5 with FLAG dilution buffer (50 mM KPi pH 7.4, 150 mM KCl, 1 mM EDTA and 0.01%/0.001%LMNG/CHS) before being added to 800 lL of pre-equilibrated anti-FLAG M2 agarose (Sigma) slurry (400 lL beads) for 2 h tumbling.Beads were collected and washed with 30 cv of dilution buffer before eluting two times with dilution buffer containing 100 lgÁmL À1 3xFLAG peptide, tumbling for 20 min.Eluates were pooled and concentrated to 500 lL using a Vivaspin 500 with 50 kDa molecular weight cut-off (Cytiva, Uppsala, Sweden) in a benchtop cooling centrifuge.The sample was subjected to HPLC-SEC using a Superdex 200 Increase 10/300 GL column (Cytiva) equilibrated in 50 mM KPi pH 7.4, 150 mM KCl and 0.001%/0.0001%LMNG/CHS and with a flow rate of 0.2 mLÁmin À1 .Fractions were collected, concentrated and analyzed by SDS-PAGE before freezing the sample on cryo-EM grids.

Grid preparation, cryo-EM data acquisition and processing
QuantiFoil R1.2/1.3 300 mesh Holey Au grids were glow-discharged at 20 mA for 60 s. 3 lL of the concentrated sample at 15 mgÁmL À1 was applied with 4 s blot time and 5 s wait time before plunge freezing in liquid ethane using a Vitrobot Mark IV (FEI, Thermo Fisher Scientific, Waltham, MA, USA) at 100% humidity and 4 °C.20 237 movies were collected on a Titan-Krios G3 operating at 300 kV equipped with a Gatan K3 detector and Bio-quantum energy filter.Movies were acquired with a nominal magnification of 165 000x and a pixel size of 0.83 A per pixel using EPU (Thermo Fisher Scientific), with 40 frames and a total electron dose of 50 e A À2 with a defocus range of 0.6-2.6 lm.Image processing was performed using CRYOSPARC v4.2.1 (Structura Biotechnology Inc., Toronto, ON, Canada) following a standard processing pipeline consisting of Patch motion correction, Patch CTF estimation and curating of exposures to remove micrographs of low quality or bad CTF.Blob picking was followed by particle extraction and several rounds of 2D classification to acquire classes for template picking.Several additional rounds of 2D classification, 3D ab initio reconstruction and heterogenous refinement resulted in decent 2D classes from a particle stack of 270 510 particles with desired 2D features.However, further processing attempts using ab initio 3D reconstruction followed by heterogenous-, homogenous-and non-uniform refinement demonstrated problems with particle alignment which prevented us from obtaining a high-resolution density map sufficient enough for model building.

Graphical software
All figures were made using AFFINITY DESIGNER (Serif, Nottingham, UK).Cryo-EM densities and predicted structures were visualized using UCSF CHIMERA X (University of California, San Francisco, CA, USA).

Results
Work in yeast has resulted in the following model for the biogenesis of Cytb [17,19,20] (Fig. 1): Binding of the assembly factor Cbp3-Cbp6 in the vicinity of the mitoribosomal tunnel exit stimulates translation and allows for efficient interaction with the newly synthesized Cytb protein, whereby Cbp3-Cbp6 binds Cytb by interacting with its matrix-facing domains.The Cytb-Cbp3-Cbp6 complex, named intermediate 0 (IME0), is released from the ribosome and heme b L can be incorporated.When heme b L has been inserted, the assembly factor Cbp4 binds with a soluble domain from the IMS side to stabilize the semi-hemylated Cytb, thereby forming intermediate 1 (IME1) [17,19].The configuration of this intermediate favors the insertion of heme b H .When both hemes have been properly incorporated, it was previously proposed that Cytb adopts a conformation that triggers the release of Cbp3-Cbp6 [17], which allows the complex to return to the mitoribosome to activate a new round of COB translation [18][19][20] (Fig. 1).The fully hemylated but unstable [17] Cytb-Cbp4 complex is subsequently joined by the auxiliary subunits Qcr7 and Qcr8, forming the very stable intermediate 2 (IME2), which can continue along the assembly line towards a fully assembled bc 1 complex.
To reveal the structural details of these steps occurring during Cytb assembly, we designed a strategy to efficiently purify IME0/IME1 (Fig. S1A,B).To this end, we utilized a strain with an engineered mitochondrial genome containing an HA-Hisx6 tag on the C-terminal end of Cytb [17].We further modified this strain by deleting the nuclear genes QCR7 and QCR8 through homologous recombination (Fig. S1A).The absence of Qcr7 and Qcr8 stalls the assembly process at IME1 by preventing the formation of IME2 (Fig. S1A).However, stalling assembly at IME1 consequently leads to repression of Cytb synthesis, since the stalled Cbp3-Cbp6 no longer stimulates COB translation at the mitoribosome [19].In addition, stalled and accumulated IME1 is sensitive to proteolysis.To account for this, we overexpressed Cbp3 and Cbp6 to increase and stabilize the levels of Cytb being synthesized (Fig. S1A).This allowed us to purify IME1 in quantities sufficient to attempt structure determination via cryo-EM (Figs S1B-E and S2A,B).However, the low final resolution of this complex, estimated at around 8-9 A (Fig. S2C,D) did not permit us to reveal significant structural insights.
Predicted structures of intermediate 0 and 1 agree well with previous site-directed photo-crosslinking data Since we were not able to determine the structure of assembly IME0 or IME1 using cryo-EM, we employed ColabFold/AlphaFold2 [30] to predict the structures of IME0, IME1 and IME2 (Fig. 2A-C).Likely due to the high conservation of Cytb and the Cbp3-Cbp6 complex, most parts of these structures could be predicted with very high confidence (pLDDT > 90) (Fig. S3A-D).Consequently, all proteins in these predicted structures had the correct topologies with their respective domains found on the correct side of the membrane.Moreover, a comparison between the predicted IME0 structure and our low-resolution cryo-EM densities showed a good agreement in the overall features (Fig. 2D).Specifically, the predicted structure of IME0 was placed in the density from a non-uniform refinement (Fig. 2D), containing an extra density protruding from the micelle.The overall features aligned well, but the part reflecting Cbp3-Cbp6 was considerably smaller than in the predicted structure, pointing to a certain flexibility of the complex or a partial degradation of Cbp3 (Fig. S2D).Likewise, fitting Cytb into the density with visible TM helices confirmed that the overall size matched well, but that the densities of the helices were distorted and more dispersed (Fig. 2E).These observations could point to heterogeneity in the sample, likely deriving from a not-yet strongly folded Cytb which is lacking one or both of its heme cofactors [17].
To further validate the predicted structures, we revisited our previously published data on site-specific photo-crosslinking of Cbp3 residues [33].Because this site-specific crosslinking occurs with very short distances, this assay allows the detection of direct protein-protein contacts.In this work, we found that the conserved chaperone domain of Cbp3 has two separate interaction surfaces for binding to Cbp6 and Cytb, respectively.Hence, Cbp3 crosslinked to Cbp6 at residues K148, P150, R245 and K252, while Cytb could be cross-linked to the residues Q183, Q184, K185, D188, R189, E195 and K215 (Fig. 2F).Intriguingly, when mapping these experimentally identified interacting residues of Cbp3 onto the predicted structure of IME0, they fitted precisely with its predicted interaction surfaces to Cbp6 and Cytb (Fig. 2G), thereby validating the predicted structures.In the Cbp3-Cbp6 interaction surface, the positively charged residues are salt-bridged with negatively charged amino acids in Cbp6.This establishes a strong interaction surface which is in line with a strong complex formation between these two proteins [18].For the Cbp3-Cytb interaction, this surface is mainly composed of two helices arranged as a V-shaped helix-turn-helix motif with charged and polar residues facing the matrix loops of Cytb.Cbp3 interacts in the predicted structure all along the matrix-facing surface of Cytb, even surrounding the C-terminal end to make a strong interaction (Fig. 2A).Mapping of the electrostatic surface potential of IME0 further reveals complementary binding surfaces comprising of charged residues establishing salt-bridges between the two proteins (Fig. S4).In contrast, Cbp6 is not predicted to interact with Cytb to any larger extent (Fig. 2A,B).This could indicate another role for this protein besides the proposed chaperone function, for example binding to the mitoribosome to position the Cbp3 chaperone domain towards the tunnel exit or regulation of translation [18,20].

Interaction of assembly factors with Cytb
Furthermore, in the predictions of IME0 and IME1, the C-terminal end of Cytb was inserted into a cleft of Cbp3 formed by its N-terminal helices and loops (Fig. 3A).This interaction is likely part of the chaperone function of Cbp3, where binding and stabilizing the unstable Cytb until full hemylation is necessary.The C-terminal end of Cytb has previously been shown to be important for the translational feedback regulation and full assembly of the bc 1 complex [34].Cytb is locked in its final conformation through the binding of Qcr7 and the matrix-localized soluble domain of Qcr8 [15,16].Strikingly, Cbp3 encircles the C-terminus of Cytb in a similar fashion as Qcr7, keeping it firmly bound in place.The TM helix of Qcr8 is binding to the last three TMs of Cytb, thereby further stabilizing their conformation.Qcr8, together with Qcr7, then binds to the loops previously coordinated by Cbp3, forming the highly stable IME2, an interaction that persists in the fully assembled bc 1 complex [19].
Cbp4 is predicted to bind to the opposite side of Cytb compared to Qcr8 (Fig. 3B).Here, Cbp4 binds to the IMS-located loops to stabilize the acquired heme b L and the conformation of Cytb.Furthermore, the single TM of Cbp4 interacts with the hemecoordinating four-helix bundle.This suggests that this interaction could further support a tight folding of this domain, which should support heme b H insertion and coordination, and the subsequent retention of the heme cofactors.Moreover, the four-helix bundle is one of the main dimerization surfaces in the fully assembled bc 1 dimer [14,15,35].The TM of Cbp4 is predicted to block this interaction surface by diagonally crossing the bundle (Fig. 3B), thereby preventing premature interaction of two Cytb proteins.Consequently, dissociation of Cbp4 from IME2 would then liberate this interaction surface, allowing dimerization with another Cytb-Qcr7-Qcr8-module.

Matrix-facing loop 4 of Cytb could play an important role in the release of Cbp3-Cbp6 during assembly
The predicted structure of IME1, containing Cytb--Cbp3-Cbp6-Cbp4, was very similar to IME0 with regards to the structure and interaction between Cytb and Cbp3-Cbp6 (Fig. 2A,B).Overall, the conformation of Cytb in these intermediates is very similar, particularly between IME0 and IME1, and between IME2 and the finally assembled Cytb (Fig. 4A).For the IME0 conformation, there is a slight lateral change in the small transverse domain 1 at the N-terminal end.This was accompanied by a larger conformational change of the matrix-facing loop 4 of Cytb that connects TM4 and TM5 (Fig. 4A,B).In IME0 and IME1, this loop is predicted to fit and bind in an interfacial cleft formed between Cbp3 and Cbp6 (Fig. 4B).Here, residues D217 and R218 of Cytb and residues D216 and R245 of Cbp3 are predicted to form two salt-bridges (Fig. S5A).Interestingly, these residues are well conserved, particularly R245 in Cbp3, which always carries a positive charge (Fig. S5B) and is located in the bottom of the cleft formed between Cbp3 and Cbp6, binding a negatively charged residue in loop 4 of Cytb.However, in IME2 and the fully assembled Cytb, loop 4 takes up a different, more central, conformation that would clash with its predicted position when bound to Cbp3 (Fig. 4B).As expected, the prediction score of this loop in the early intermediates is substantially lower compared to that of Cytb found in IME2, where it adopts a more mature conformation (Fig. S5C).The low prediction score of loop 4 in IME0/1 could be an indication of a more flexible configuration that is fortified during the assembly process.
In the structure of IME2, both the transverse domain 1 and loop 4 have adopted conformations similar to the assembled Cytb (Fig. 4A).This conformational switch likely reflects that this form is further down the assembly line and, consequently, after the release of Cbp3-Cbp6 has occurred.The nature of the signal leading to the release of Cbp3-Cbp6 from Cytb is yet unknown, but it likely involves full hemylation accompanied by a conformational change [17].Likewise, there are no obvious or direct interactions of Cbp4 with either domain 1 or loop 4, suggesting that Cbp4 might not directly influence Cbp3-Cbp6 release (Fig. 4C).Interestingly, a closer look at loop 4 revealed a possible explanation behind this conformational change (Fig. 4D).In the fully assembled and hemylated Cytb, S207, located in this loop, is one of the residues coordinating and binding the porphyrin ring of heme b H .In the semi-hemylated Cytb of IME1, however, heme b H has not yet been incorporated.This means that S207 cannot participate in the coordination of the porphyrin ring, and, in turn, the loop is free to interact with Cbp3-Cbp6 instead.It is possible that this alternate binding and conformational change of loop 4 is a major part of the mechanism behind the Cbp3-Cbp6 release from Cytb and, as a consequence, the transition from IME1 to IME2 to trigger further steps in the assembly line.

Discussion
In this work, we used recent breakthroughs in structure prediction to generate models of assembly intermediates to shed light on conformational changes as they occur during the early maturation of Cytb.These structure predictions resulted in models that substantially corroborate with published experimental data, including the topology of domains relative to the membrane and the pattern of protein-protein interactions as revealed through previous site-specific photo-crosslinking data [33].Moreover, the high conservation of Cytb, Cbp3 and Cbp6 facilitates high-quality structure prediction.This, together with the validation by previous experimental data, supports the notion that these predicted structures are biologically meaningful and can be interpreted to generate new insights or to formulate new hypotheses.
Cbp3-Cbp6 binds the matrix-facing loops to stabilize the newly synthesized Cytb, which is in line with its localization and its role as a specific chaperone supporting hemylation.Hemylation of Cytb is essential for bc 1 complex function.The two hemes are not covalently attached but coordinated by conserved histidine residues in a four-helix bundle [16].For this purpose, the four TMs must come into the correct spatial arrangement, which is then also important to retain the acquired hemes.It is still not clear whether heme b L , which is the first to be bound to Cytb, is incorporated co-translationally, during initial folding or first in IME0 where Cbp3-Cbp6 has already bound.Moreover, it is possible that other factors facilitate heme b L insertion by aiding Cytb to adopt a favorable conformation.Such a factor is likely Cbp3-Cbp6, but it is also possible that other proteins like the ferrochelatase Hem15 [36] or the yet poorly understood early bc 1 complex assembly factor Fmp25/Bca1 [37] are involved.In particular, Bca1 is interesting as it can be found in the proximity to Cbp3, the mitoribosome and the insertion machinery [38].Interestingly, the predicted structure of IME1 places the IMS domain of Cbp4 in an ideal position for stabilizing the semi-hemylated Cytb by locking its conformation around the heme b L site.The single TM helix of Cbp4 binds along the heme-coordinating helices of Cytb, placing it in a position to affect conformation and incorporation of heme b H .Moreover, the binding of Cbp4 is likely inhibiting the dimerization of Cytb, allowing a control mechanism that ensures that only fully hemylated Cytb is further assembled.Such a scenario is in agreement with the previously elaborated sequence, where Cbp4 is recruited to the semihemylated IME1 to stabilize hemylation and guide Cytb to IME2 [17].
The predicted structure of IME1 also agrees with the role of Cbp3-Cbp6 binding to the matrix-facing domains to stabilize the semi-hemylated Cytb.Moreover, Cbp3-Cbp6 binding to Cytb would be favorable to aid in the insertion of heme b H .By binding to loop 4 and, in the process, displacing the transverse domain 1 of Cytb, this would introduce an open conformation for heme b H incorporation.It has previously been shown that insertion of heme b H triggers the release of Cbp3-Cbp6 from Cytb, upon which the heterodimer can return to the mitoribosome and activate a new round of COB translation [17,19,20].It is therefore intriguing to observe the predicted shift of loop 4 and domain 1 of Cytb in the presence of Cbp3-Cbp6.S207 on loop 4 of Cytb participates in the coordination of heme b H by binding to a substituent group of the porphyrin ring [15].In the absence of the heme, the loop is instead predicted to bind in a cleft formed by Cbp3 and Cbp6.It is tempting to speculate that the incorporation of heme b H and the subsequent conformational shift of loop 4 weakens the Cytb interaction with Cbp3-Cbp6, thereby facilitating its release.
It should be noted that these predicted structures, while apparently biologically interpretable, are not fully verified despite their agreement with experimental data on membrane topology, localization in correct compartments or protein-protein contacts.To further study these insights experimentally, we aimed to resolve the cryo-EM structures of these assembly intermediates.Unfortunately, even though we managed to successfully purify sufficient quantities of an assembly intermediate containing Cytb, Cbp3-Cbp6 and Cbp4, determining the structure by using cryo-EM proved difficult.These efforts resulted in a density that could only partly be fitted with the predicted structures.Since it was difficult to align the particles during both 2D and 3D classifications, it is possible that these problems reflect a heterogeneous sample with Cytb in several different conformations.A disordered Cytb inside the micelle could cause problems with particle alignment and be the reason behind broken and dispersed TM helices seen in our cryo-EM densities.During early assembly, in the absence of the two b-type hemes, Cytb is in an unstable and likely disordered state.These flexible states, which can be functionally important for hemylation, could also be difficult to predict using Alphafold2, which uses structures of fully folded and assembled proteins determined from classical structural biology approaches as input.It is therefore possible that the predicted structures represent the most stable conformations, neglecting other, more flexible states of Cytb.To identify and describe these, it would be exciting for future research to, for example, employ molecular simulations on Cytb early assembly intermediates.

Fig. 1 .Fig. 2 .
Fig. 1.Formation of assembly intermediate 1 of the bc 1 complex.Schematic showing the early steps of Cytb biogenesis.As the nascent Cytb polypeptide emerges from the tunnel exit, Cbp3-Cbp6 binds to the matrix-facing loops to stabilize Cytb and form assembly intermediate 0. After insertion of heme b L , Cbp4 joins to form assembly intermediate 1, which facilitates the incorporation of heme b H . Cbp3-Cbp6 is then released and returns to the mitoribosome to activate a new round of COB translation.Qcr7 and Qcr8 then join Cytb and Cbp4 to form the more stable assembly intermediate 2 on the path to a fully assembled bc 1 complex.

Cytb from dimeric bc 1 90°Fig. 3 .
on the dimeric Cytb as found in the respiratory supercomplex (PDB: 6GIQ) Interaction of assembly factors with Cytb.(A) Structure of Cytb in complex with Qcr7 and Qcr8 (representing the state of Cytb in IME2) from the assembled supercomplex (PDB ID: 6GIQ) compared to the predicted IME0.In IME0, the C-terminal helix of Cytb is encircled by Cbp3 in the same manner as Qcr7 binds this helix in the fully assembled complex.This interaction indicates the importance of the Cytb C-terminal helix during the assembly of the bc 1 complex.(B) The TM of Cbp4 stabilizes the heme-containing four-helix bundle and likely inhibits premature dimerization of Cytb.Left, depiction of how the TM of Cbp4 runs diagonally over the four-helix bundle.Right, the Cytb dimer from the respiratory supercomplex (PDB ID: 6GIQ) was superimposed on Cytb in the predicted IME2.The TM of Cbp4 occupies a position blocking the dimerization interface between two Cytb proteins.

Fig. 4 .
Fig. 4. Conformational changes during early Cytb maturation.(A) Structures of Cytb from the predicted assembly intermediates and from the assembled supercomplex (PDB ID: 6GIQ).The most prominent changes in Cytb conformation are in the matrix-localized N-terminal domain 1 and loop 4. (B) Cytb from the assembled supercomplex (in red) (PDB ID: 6GIQ) superimposed on the predicted structure of Cytb in IME0 (yellow).Overall, the structures are very similar, with the only noticeable differences in the position of domain 1 and loop 4 of Cytb, which is bound to Cbp3-Cbp6 in IME0.(C) The structure of Cytb (red) superimposed on the predicted structure of Cytb in IME1 (orange).The conformation of Cytb in IME1 shows high similarity to its conformation in IME0, with loop 4 still being bound to Cbp3-Cbp6.(D) Comparison of the different positions of domain 1 and loop 4 of Cytb in IME0/1 and the fully assembled Cytb.S207 in loop 4 normally participates in the coordination of heme b H by binding to the porphyrin ring.In IME0 lacking heme b H , this loop and S207 are instead bound in the interface between Cbp3 and Cbp6.This conformational change is possibly part of the mechanism by which Cbp3-Cbp6 is released from IME1 during assembly.