Early fate decision for mitochondrially encoded proteins by a molecular triage

Folding of newly synthesized proteins poses challenges for a functional proteome. Dedicated protein quality control (PQC) systems either promote the folding of nascent polypeptides at ribosomes or, if this fails, ensure their degradation. Although well studied for cytosolic protein biogenesis, it is not understood how these processes work for mitochondrially encoded proteins, key subunits of the oxidative phosphorylation (OXPHOS) system. Here, we identify dedicated hubs in proximity to mitoribosomal tunnel exits coordinating mitochondrial protein biogenesis and quality control. Conserved prohibitin (PHB)/m-AAA protease supercomplexes and the availability of assembly chaperones determine the fate of newly synthesized proteins by molecular triaging. The localization of these competing activities in the vicinity of the mitoribosomal tunnel exit allows for a prompt decision on whether newly synthesized proteins are fed into OXPHOS assembly or are degraded.


In brief
Dedicated protein quality control systems ensure the folding of newly synthesized proteins at ribosomes.Kohler et al. reveal such a system for mitochondrially encoded proteins in the form of the PHB/ m-AAA protease associated with mitoribosomes.Thereby, interactions with OXPHOS assembly factors or the PHB/m-AAA protease determine the fate of nascent proteins.

INTRODUCTION
Protein biogenesis is a sophisticated and complex process, which can fail at numerous stages.A particular challenge is the folding of the nascent protein, which bears the risk of misfolding, aggregation, and subsequently causing proteotoxic stress and cell death. 1 Hence, all organisms contain protein quality control (PQC) pathways, checking and maintaining the folding of nascent proteins during and after their synthesis. 2The polypeptide tunnel exit (PTE) of the large ribosomal subunit is a key site for protein biogenesis, as it is the place where newly synthesized proteins emerge from the ribosome to start folding.4][5][6] Whereas these processes were extensively studied and well characterized for cytosolic ribosomes, 7 the molecular principles of PQC at mitochondrial ribosomes remain largely elusive.
Mitochondria are essential organelles implicated in many cellular functions and are best known for their role in energy conversion.Thereby, energy from nutrients is converted into ATP in a process called oxidative phosphorylation (OXPHOS), also referred to as cellular respiration. 8This process is conveyed by five large multiprotein complexes, collectively known as the OXPHOS system, which includes the respiratory chain (RC; CI-CIV), ATP synthase, as well as mobile electron carriers within the inner mitochondrial membrane (IMM).The multiprotein complexes of the OXPHOS are unique, as they are encoded by two separate genetic systems, the nuclear and mitochondrial genome. 93][14] The highly hydrophobic, mitochondrially encoded proteins are co-translationally inserted into the IMM 15,16 and assembled into OXPHOS complexes via sophisticated assembly lines. 175][26] Removal of damaged or superfluous proteins from assembly depends on mitochondrial PQC.8][29][30] Significant advances in the understanding of mitochondrial PQC on the molecular level have been made in recent years. 29,31,32However, the precise molecular mechanisms by which the quality of mitochondrially encoded proteins is surveilled and maintained are largely unknown.
Recently, we characterized the proximity interactome of mitoribosomes and reported the positioning of proteins required for mitochondrial gene expression, translation, and OXPHOS assembly at strategically important sites of the mitoribosome. 33articularly, the mitoribosomal PTE was in proximity to protein insertases and OXPHOS assembly factors, in line with previous work. 34,35Remarkably, PQC factors were also detected in this site, including the prohibitin (PHB) complex and the two membrane-bound AAA (ATPases associated with diverse cellular activities) proteases, termed i-AAA and m-AAA proteases. 31,33,36HB complexes serve as membrane scaffolds 37 and form large ring-like structures composed of multiple copies of Phb1/Phb2 heterodimers. 38PHB interacts with the m-AAA protease to form the highly conserved PHB/m-AAA supercomplex, 39,40 composed of multiple Phb1 and Phb2 subunits (PHB1 and PHB2 in humans, respectively) and the hexameric m-AAA protease formed by Yta10 and Yta12 subunits (AFG3L2 and SPG7 in humans).Although the PHB complex is implicated in ageing 41 and various human diseases, including neurodegenerative 42 and cardiac disorders 43 as well as cancer, 44,45 only little is known about its molecular function.Recent structural insights on the bacterial homolog of the PHB-protease complex place the AAA protease in the inside of the ring, suggesting that PHB complexes could separate catalysis from the rest of the IMM. 46,47uch a scenario would be in line with data showing that PHB inhibits degradation by the m-AAA protease. 39ere, we explored the function of the PHB/m-AAA protease complexes in the biogenesis of mitochondrially encoded proteins.We demonstrate that these complexes are part of OXPHOS assembly hubs in the IMM, which gather proteins involved in protein import, mitochondrial translation, OXPHOS biogenesis, and quality control.Absence of PHB leads to a functional impairment of RC complexes, revealing that quality control and proteolytic degradation are integral parts of RC complex assembly.Specifically, the PHB/m-AAA complex is, like a distinct set of OXPHOS assembly factors, localized in the vicinity of the mitoribosomal PTE.This localization allows an early fate decision for newly synthesized mitochondrially encoded proteins, which can either be bound and stabilized by their dedicated assembly factors to be channeled into RC assembly, or by the PHB/m-AAA for degradation.

RESULTS
The PHB/m-AAA protease complex is in proximity to the mitoribosomal PTE We recently used proximity-dependent biotin identification (BioID) (Figure 1A) to characterize the proximity interactions in the mitochondrial gene expression network to reveal how transcription, RNA processing, translation, and early steps of OXPHOS assembly are spatially organized. 334][5][6] To further define proteins in the vicinity of the mitoribosomal PTE, we re-assessed proximity interactions of components of the membrane protein insertion machinery (Oxa1, Mba1, and Mrx15), as well as the mitochondrial translation and assembly factor Cbp3, and combined them with the proximity network of the mitoribosomal PTE 33 (Figures 1B and  1C).Proteins involved in the first steps of OXPHOS biogenesis localize close to the PTE, forming early OXPHOS biogenesis hubs. 33These factors were accompanied by a diverse set of IMM-localized proteins involved in PQC (Figures 1B and 1C).Specifically, we detected all subunits of the PHB/m-AAA complex in proximity to PTE ligands, suggesting their important role in mitochondrial translation and/or OXPHOS assembly.
The proximity of the PHB/m-AAA complex to the PTE prompted us to test whether it could be associated with the large subunit of the mitoribosome.We indeed observed a co-migration of the complex with the large subunit of the mitoribosome in linear sucrose gradients under very mild lysis conditions (Figures 1D,  1E, and S1A-S1D).However, the PHB complex forms a huge ring-like structure approximately 2.2 MDa in size, 48 raising the possibility that these two large protein assemblies migrate independently in the same fraction but do not interact.To examine a possible physical interaction between the PHB/m-AAA and the mitoribosome, we disintegrated the mitoribosome by RNase A treatment, which substantially shifted the PHB/m-AAA complex to lighter fractions, demonstrating a specific assembly with the mitochondrial ribosome (Figure 1F).We next asked whether it is PHB or the m-AAA protease that mediates ribosome interaction.Because the m-AAA protease is required for the assembly of mitoribosomes, 49 we deleted Phb1 and Phb2 and re-evaluated the migration behavior of the m-AAA protease in sucrose gradients.Deletion of one of the PHB subunits destabilizes the other subunit and results in the loss of the PHB complex (Figures S1E and  S1F). 50We observed that the co-migration of the m-AAA protease with the mitoribosome was lost in the absence of the PHB complex (Figure 1G).This shows that PHB is important to localize PHB/m-AAA complexes to the proximity of the PTE.

The composition of the PHB/m-AAA proximity interactome reveals its involvement in OXPHOS biogenesis
The proximity of the PHB complex to the mitoribosomal PTE, a feature shared with proteins mediating OXPHOS assembly, 33 led us to speculate that the PHB/m-AAA protease complex might be an integral part of early OXPHOS assembly hubs.To test this hypothesis, we determined proteins in the vicinity of ALFA-tagged PHBs (Phb1 ALFA and Phb2 ALFA ) via chemical crosslinking, followed by purification and mass spectrometry (MS; Figures 2A and S2A-S2D).We then combined this dataset with our previous BioID analysis of the PTE to identify hits that are detected by both approaches (Figures 2A and 2B; Tables S1 and S2).We found that 18 of the PTE BioIDome proteins can also be crosslinked to PHBs (Figure 2B), likely presenting a core proximity interactome of the PHB complex.This group included the m-AAA protease subunit Yta12, the i-AAA protease Yme1 and its interactor Mgr1, and, importantly, a set of OXPHOS biogenesis factors.Overall, proteins of very similar functional groups were crosslinked to the PHB complex and (legend continued on next page) found in the mitoribosomal PTE BioIDome, including factors involved in protein import, PQC, OXPHOS biogenesis, and OXPHOS proteins (Figure 2B).To differentiate between direct and proximity interactors of PHB, we next purified the complex under mild lysis conditions and determined proteins by quantitative MS (Figures 2C and 2D; Table S3).This resulted in the identification of many mitoribosomal subunits.By comparing this data to a recent high-resolution mitochochondrial complexome analysis, 48 a common set of nine proteins could be identified (Figure 2D).These nine proteins include the expected subunits of the PHB/m-AAA complex as well as Mrx9, Mrx17, Mdj2, Aim25, and the protease Oma1, which, therefore, likely represent core interactors of the PHB/m-AAA complex.By combining data from our PHB pull-down and crosslinking analysis, together with the BioIDome of the PTE, we therefore obtained a high-confidence PHB-proximity interactome, which was significantly overlapping with direct PHB interactions (Figure 2E, overlapping proteins are marked in bold).At the same time, the proximity interactions also included proteins other than the core PHB/m-AAA interactome, in line with the note that the complex is part of the early OXPHOS assembly hub.To further substantiate this finding, we asked whether Phb1 and Phb2 are transcriptionally coregulated with the proteins implicated in OXPHOS biogenesis.Applying the serial pattern of expression levels locator (SPELL) of the yeast genome database, 51 we compared the expression profiles of PHB1 and PHB2 with mRNAs from 15.475 gene expression microarrays from 576 published studies.Gene Ontology (GO)-term analysis of the 50 most similar co-expressed genes revealed that PHB1 and PHB2 are expressed most similarly as genes involved in RC assembly, cellular protein complex assembly, protein complex subunit organization, mitochondrial translation, and proteasomal ubiquitin-independent protein catabolic processes (Figure 2F).Hence, these analyses corroborate the finding that the PHB complex is an integral part of OXPHOS biogenesis and that it is physically located in the early OXPHOS assembly hubs that organize mitochondrial translation, protein import, PQC components, and OXPHOS biogenesis in common sites in the IMM (Figure 2G).Moreover, these data suggest that the PHB/m-AAA protease complex plays an important role in OXPHOS biogenesis.

The dynamics of the PHB interactome confirms an involvement in OXPHOS biogenesis
We reasoned that if the PHB/m-AAA protease complex was involved in OXPHOS biogenesis, its proximity-interaction patterns should change depending on the metabolic state of the cells.We therefore shifted respiratory growing cells expressing Phb1 ALFA either to fermentable carbon sources or kept them on respiratory medium.After an additional 2 h of growth, mitochondria were isolated, crosslinked, and subjected to purification of Phb1 followed by MS (Figure 3A).The shift to fermentation, which is accompanied by a reduction of OXPHOS capacity, resulted in a loss of crosslinking between the PHB complex and certain assembly factors.Moreover, we observed an accumulation of some nuclear-encoded OXPHOS proteins near the PHB/m-AAA protease complex (Figures 3B, S3A, and S3B; Tables S1 and S4).This included two nuclear-encoded subunits of CIII (Qcr2 and Cor1), as well as Cox4 of CIV, all of which are incorporated early in their respective biogenesis pathways. 22,52Of note, the m-AAA protease and other PQC factors were crosslinked to Phb1 in both metabolic conditions.Hence, a permanent and a dynamic, respiration-dependent interactome of the PHB complex can be differentiated (Figure 3C).Whereas factors for proteostasis, as well as several proteins involved in protein import, mitochondrial translation, and OXPHOS biogenesis are permanently in vicinity to the PHB/m-AAA protease complex, certain factors for protein import and OXPHOS biogenesis are only found under respiratory growth conditions near the PHB/m-AAA protease complex.The presence of these factors decreases during fermentation and an accumulation of nuclear-encoded OXPHOS proteins can be observed (Figure 3C).Taken together, these data show a dynamic proximity-interaction network of the PHB/m-AAA protease complex that mirrors the genetic reprogramming as it occurs during metabolic shifts.

Absence of the PHB complex impairs respiratory function
To test whether the PHB complex plays a prominent role in OXPHOS biogenesis, we next analyzed the physiological consequences of the loss of Phb1 and/or Phb2 on mitochondrial function.Blue-native PAGE (BN-PAGE) analysis of mitochondrial membranes and whole-cell proteomics revealed that steadystate levels of OXPHOS complexes were unaltered in the absence of the PHB complex (Figures 4A and 4B; Table S5).However, when assessing the functionality of the OXPHOS system, we observed that mitochondria of phb1/2DD cells have a severely reduced capacity of NADH oxidation in basal and phosphorylating conditions (Figure 4C), which is not caused by proton leakage over the IMM, as indicated by an unchanged respiratory control ratio (Figure 4D).Rather, the reduction in NADH-driven respiration could be caused by hemylation defects or a loss of (C) Mass spectrometric analysis of PHB complex purifications via FLAG-tagged Phb2 from digitonin-solubilized mitochondria.Volcano plot visualizing the log 2 fold change between Phb2-FLAG vs. wild type (x axis) and the Àlog 10 transformed t test of an unpaired t test.Significance was considered for log 2 fold change greater than 1.5 (>2.9-fold) and a p value smaller than 0.05.(D) Venn diagram to compare high-resolution complexome data for Phb1 and Phb2 (data taken from Schulte et al. 48) with FLAG-tagged Phb2 purification data (this study).
(E) Venn diagram to compare proteomic data from PHB pull-down, PHB crosslinking, and mitoribosomal polypeptide tunnel exit (PTE) BioIDome experiments, to determine a high-confidence PHB proximity interactome.Proteins overlapping between PHB interactome (D) and PHB proximity-interactome (E) are marked in bold.(F) SPELL analyses of co-expression of prohibitins with other transcripts from published datasets.
(G) Schematic model of the early OXPHOS assembly hubs that combine protein import, mitochondrial translation, OXPHOS assembly, and protein quality control in the inner mitochondrial membrane (IMM).IMS, intermembrane space.See also Figure S2 for supplemental information.hemylation of both CIII and CIV (Figure 4E), which is also reflected in significant reductions of CIII and CIV activities (Figure 4F) as well as a respiratory growth defect of the mutant (Figure 4G).Mitochondrial dysfunction can provoke oxidative stress. 53Indeed, phb1/2DD cells display increased ROS levels originating from mitochondria (Figures 4H and 4I).Thus, the loss of PHBs impairs OXPHOS function, which is not associated with a prominent destabilization of RC complexes but rather by a disturbance of assembly leading to defective hemylation and partial loss of OXPHOS capacity.
The PHB/m-AAA protease complex interacts with newly synthesized mitochondrially encoded proteins and is required for efficient OXPHOS complex assembly The observation that the OXPHOS system assembles in a partly non-functional form in the absence of the PHB complex indicates that the PHB/m-AAA protease complex influences OXPHOS assembly accuracy.Likewise, the localization of this complex in the vicinity of the mitoribosomal PTE and OXPHOS biogenesis factors would be in line with an early contact of the complex with newly synthesized mitochondrial translation products.To monitor OXPHOS assembly in wild-type (WT) and phb1/ 2DD cells, we performed 35 S-methionine pulse-chase labeling of mitochondrial translation products in isolated mitochondria.After a pulse for 5 min, labeling was stopped and the assembly of RC complexes was monitored by BN-PAGE (Figure 5A).Newly synthesized proteins can be observed in different assemblies, reflecting assembly intermediates and fully assembled OXPHOS complexes (Figure 5B).Notably, newly labeled proteins were also found in a high molecular weight complex of the same size as PHB/m-AAA, which was lost in phb1/2DD cells, indicating that they were bound to the PHB/m-AAA protease complex (Figure 5B).This conclusion was further supported by crosslinking of newly synthesized proteins to Phb2 and Yta12 (Figures S4A and S4B).Over time, we observed a decrease of signal in the assembly intermediates with a concomitant increase of OXPHOS complexes in both WT and phb1/2DD mitochondria, but the efficiency of OXPHOS assembly was severely reduced in the absence of PHBs (Figure 5C).The overall levels of radiolabeled polypeptides decreased, with comparable rates in WT and phb1/2DD mitochondria, indicating significant degradation of newly synthesized proteins (Figure 5D).Similarly, we observed a prominent signal from newly synthesized proteins associated with PHB/m-AAA protease complexes, which declined over time (Figures 5B and 5D).Two-dimensional BN-PAGE/SDS-PAGE analysis revealed that newly synthesized, radiolabeled proteins found in the complex include completely synthesized, full-length proteins (Figure 5E, left).In addition, a prominent smear could be detected in the PHB/m-AAA protease complex (Figure 5E, left, dark red box), which most likely represents proteins being degraded by the m-AAA protease, as a similar signal was detected together with the m-AAA protease running at a lower molecular weight when PHBs were absent (Figure 5E, right, light red box).
Given the localization of the PHB/m-AAA protease complex in the vicinity of the mitoribosomal PTE and assembly factors, we asked when newly synthesized proteins would start to interact with the complex.We therefore determined the timing of interaction between newly synthesized mitochondrially encoded proteins and the PHB/m-AAA complex in very short pulse-labeling experiments.Consistent with its localization in the vicinity of the mitoribosomal PTE, both full-length proteins and degradation products were found in association with the PHB/m-AAA protease complex as early as 30 s after initiation of labeling and increased over time (Figures 5F-5H).Interaction of the newly synthesized proteins with the PHB/m-AAA protease complex and assembly factors occurred with similar dynamics (Figure 5H).Because degradation products were also found in association with PHB/m-AAA protease already at the earliest time points, these data show that the decision to degrade these proteins occurs very early after their synthesis.
The PHB/m-AAA protease complex is part of a molecular triage that directs newly synthesized proteins to OXPHOS complex assembly or to degradation Our experiments demonstrate that the PHB/m-AAA protease complex is in the vicinity of the mitoribosomal PTE and binds newly synthesized polypeptides early after translation.To further delineate the function of the PHB/m-AAA complexes in the fate decision of newly synthesized polypeptides, we used two independent approaches.First, we delayed degradation using a mutant of the m-AAA protease with reduced proteolytic activity to characterize the interaction of newly synthesized proteins with the PHB/m-AAA complex.Second, we analyzed the fate of newly synthesized polypeptides in the absence of an OXPHOS assembly factor.
In order to study how mitochondrial translation products would engage with the PHB/m-AAA complex, we used a mutant where the metal-binding site in the proteolytic domain of Yta10 is mutated while Yta12 is unchanged.This Yta10 E559Q variant maintains respiratory competence, and the reduced proteolysis was not accompanied by an increased OXPHOS assembly or improved respiration, which was similar in WT cells and in yta10D cells expressing Yta10 E559Q . 54Newly synthesized, radiolabeled proteins persisted much longer in the PHB/m-AAA protease complex in mitochondria containing Yta10 E559Q (Figures 6A and 6B), showing that the partially inactivated m-AAA protease was still sufficiently active to induce degradation, albeit with lower efficiency.Moreover, we identified a similar smear of radiolabeled proteins associated with the complex as in mitochondria from WT cells (Figure 6C), suggesting that translation products engaging with the complex are subjected to degradation.
Next, we asked what would determine whether newly synthesized proteins would become substrates of the PHB/m-AAA protease complex.All mitochondrially encoded proteins are subunits of large complexes, necessitating the aid of dedicated chaperones to bind transiently to the newly synthesized proteins to stabilize their fold to promote assembly.Our previous work has demonstrated that the highly conserved Cbp3/6 complex localizes to the mitoribosomal PTE, where it is optimally positioned to bind to newly synthesized cytochrome b (Cytb) to support its maturation. 34We therefore asked whether the absence of its assembly factor would direct Cytb to the PHB/m-AAA complex.However, Cbp3 is not only required for CIII biogenesis but also for controlling synthesis of Cytb through a translational feedback loop. 26Hence, we uncoupled Cytb expression from Cbp3 by ectopic expression from a mRNA containing the 5 0 untranslated region of COX2. 55This allows synthesis of Cytb in the absence of Cbp3, while assembly of CIII is inhibited at early steps.Strikingly, Cytb synthesized in absence of its assembly factor resulted in an accumulation of newly synthesized proteins bound to PHB/m-AAA protease complexes (Figures 6D-6F), which were effectively degraded by the m-AAA protease upon further incubation, as indicated by significant reduction of the radioactive signal associated with PHB/m-AAA protease complexes after 30 min (Figures 6D and 6E) and the loss of Cytb (Figure 6F, red arrows).This data therefore establishes that it is the availability of a specific assembly chaperone to interact with its client protein that determines the fate of the newly synthesized protein.Because both assembly chaperones and the PHB/m-AAA protease complex associate in vicinity to the PTE, fate decision can occur rapidly upon completion of protein synthesis.

DISCUSSION
Quality control of newly synthesized proteins is essential to maintain a functional proteome by directing non-functional/superfluous proteins for degradation.In this work, we studied how fate decisions are achieved at mitochondrial ribosomes.We demonstrate that this decision occurs directly upon completion of protein synthesis, in a process where nascent proteins and OXPHOS assembly factors associate with the PHB/m-AAA protease complex in the vicinity of the mitoribosomal PTE (Figure 7).In the case that a functional nascent protein interacts productively with its assembly factor, typically a dedicated chaperone that keeps its client in an assembly competent conformation, the protein will be excluded from proteolysis and assemble (F-H) Pulse 35 S-methionine in-organello radiolabeling of WT mitochondria.First-dimension BN-PAGE membranes, either stained with Coomassie, autoradiography detection, or decorations with indicated antibodies are shown in (F), as well as a quantification (normalized to the 30-s value) of the autoradiography signal from the total or from the PHB/m-AAA complex (H).First-dimension BN-PAGE gels were applied for second-dimension SDS-PAGE and autoradiographs are presented in (G).Data in (H) are represented as mean ± SEM (n = 3 biological replicates).See also Figure S4 for supplemental information.into OXPHOS complexes.However, the protein will be degraded if assembly is delayed, causing a prolonged interaction with the PHB/m-AAA protease.This can occur if the bulk of the assembly factor is occupied with assembly intermediates, if the factor is missing (for example, due to a pathogenic mutation), or if the newly synthesized protein fails to interact with the factor because of aberrant synthesis or a folding problem.This fate decision occurs very early in the biogenesis and is facilitated by the organization of assembly factors and PHB/m-AAA in proximity to the mitoribosomal PTE.
What could be the molecular basis by which proteins are sent for degradation to PHB/m-AAA?If hydrophobic proteins cannot be folded properly and subsequently fail to assemble, hydrophobic stretches will be exposed, representing substrates for the protease.It was, for example, shown that the AAA domain of a mitochondrial AAA protease exerts chaperone-like activity in vitro 56 and that a short protein tail protruding from a membrane protein qualifies this as a substrate for the mitochondrial AAA proteases, which can subsequently extract and degrade this protein. 57One key filter for discriminating which proteins become substrates is likely the PHB complex itself.Consequently, the absence of PHB unleashes the m-AAA protease, leading to increased protein turnover. 39Structural work on the bacterial complex revealed that HflKC, the bacterial homolog of PHBs, forms a spherical, membrane-attached cage that encapsulates the AAA protease FtsH. 46,47It is likely that access to the interior of this cage depends on the folding state of the potential substrate protein.Among others, the availability of OXPHOS biogenesis factors determines the time needed for newly synthesized proteins to be correctly folded, processed, and assembled into complexes. 55,58Hence, the reduced availability of OXPHOS biogenesis factors results in prolonged states of unfolding, which could be a signal for entering the interior of the PHB/m-AAA complex, where the protein is degraded by the m-AAA protease.Moreover, the PHB has previously been shown to be a key regulator of mitochondrial phospholipid metabolism, and its components genetically interact with genes encoding phospholipid metabolism enzymes. 59Combining these functional and structural properties of PHB with our findings, we hypothesize that the PHB complex is a membrane organizer at the PTE with three main functions: (1) regulating access of the m-AAA protease to potential substrate proteins; (2) bringing the m-AAA protease in close proximity to the mitoribosomal PTE, and thus in vicinity to newly synthesized mitochondrially encoded proteins and OXPHOS assembly factors; and (3) providing an optimal lipid environment for membrane protein folding and m-AAA protease activity.The pathway to select newly synthesized proteins for degradation, as suggested in this work, is different from cytosolic ribosome-associated quality control, where ribosomes, stalled for various reasons on the mRNA, are recognized, recycled, and the nascent chains specifically degraded. 7,60To the contrary, the molecular triage deciding the fate of fully synthesized proteins involving PHB/m-AAA is reminiscent of the fate decision for nuclear-encoded, mitochondrial-destined proteins that are synthesized in the cytosol. 61Here, ubiquilins act as chaperons that bind transmembrane domains of these proteins in the cytosol, temporarily providing the opportunity for membrane targeting but, over time, recruiting E3 ligases for client protein ubiquitination and degradation via the proteasome.Hence, this triage system prevents client protein aggregation and is at the crossroad between mitochondrial membrane protein biogenesis and degradation. 61However, while conceptually similar, the triaging system for mitochondrially encoded proteins is spatially organized and relies on the early contact of the protein with either its dedicated chaperone or the protease, which are organized in a common site.Moreover, many nuclear-encoded mitochondrial proteins are subunits of larger enzyme complexes, including the OXPHOS complexes, and need to be imported.Given the proximity of PHB/m-AAA to the protein import machineries, as revealed here, it is tempting to speculate that newly imported proteins could also be subjected to a similar molecular triaging in case they fail to assemble.Consistently, various nonassembled, nuclear-encoded OXPHOS subunits were found to be degraded by the m-AAA protease.
A balanced expression of mitochondrial and nuclear genes is necessary to ensure the availability of OXPHOS subunits for assembly.8][29][30] Consequently, various mechanisms have evolved that allow synchronization of mitochondrial and nuclear gene expression programs, 24 the feedback control of mitochondrial translation, 26,55,58,[62][63][64] and the regulation of mitochondrial import by phosphorylation. 65,66Although these mechanisms have been best documented in baker's yeast, they are appar-ently conserved in mammals, [67][68][69][70][71] but detailed mechanisms and the extent of genetic regulation apparently vary. 72,73The PHB/m-AAA protease complex is highly conserved, and the here-presented molecular triage represents a versatile alternative to control the accumulation of unassembled proteins in the IMM.
Our study links the mitochondrial PQC system to early OXPHOS assembly hubs in the IMM, which are strategically placed close to the PTE of mitoribosomes. 33Apparently, these assembly sites are not equally distributed in the IMM, with early RC assembly occurring preferentially in the inner boundary membrane (the IMM opposing the outer membrane), while ATP synthase assembly is enriched in the cristae, the main sites of OXPHOS. 74Moreover, mitochondrial ribosomes were found to be part of large assemblies 75 with many different interactors, including Phb1 and Yta12 of the PHB/m-AAA complex and their interactors Aim25, Mdj2, Mrx9, and Mrx17.Many steps of gene expression are organized in these assemblies to presumably increase the efficiency of gene expression and assembly.Spatially organizing biogenesis of mitochondrially encoded proteins, their membrane insertion, and assembly into OXPHOS complexes together with PQC and degradation likely provide an efficient solution for maintaining proteostasis of mitochondrially encoded proteins.However, while this study demonstrates a proximal localization of these activities, including protein import by the TIM22 and TIM23 complexes, the exact stoichiometries and distribution of these assemblies, their functional interrelations, and their dynamics, remain to be investigated.
Notably, other proteases specialized in degrading membrane proteins are found in vicinity to PHB/m-AAA, including Yme1 and Oma1.It is therefore likely that additional layers of organization and mechanistic interplay between these proteases shape substrate recognition and membrane protein turnover in mitochondria.We speculate that these hubs might exist in different compositions, tuning PQC and protein degradation at the IMM according to respective cellular needs, metabolic situations, and/or stress conditions.It is well-established that a redundancy in cellular PQC systems exists, which serves as a pitfall strategy Newly synthesized mitochondrially encoded proteins are in proximity to the PHB/m-AAA complex and OXPHOS assembly factors.In case of a functional interaction with a dedicated OXPHOS assembly factor, the newly synthesized protein enters OXPHOS assembly (left).If this interaction with an assembly chaperone cannot occur, the proteins are instead bound by PHB/m-AAA and degraded (right).
7][78][79] It is tempting to speculate that a similar multi-layer system is in charge to maintain the quality of mitochondrially encoded proteins.Indeed, other proteases detected in our proximity-interaction experiments, foremost the i-AAA protease Yme1, might be in a strategic position near the PHB complex in case the degradation of damaged or unassembled mitochondrially encoded proteins by the m-AAA protease fails.In line with this scenario, it was demonstrated that both the m-AAA and i-AAA protease are able to extract substrates from both sides of the IMM. 57Mitochondrial protein turnover is not restricted to newly synthesized or imported proteins but also includes the removal of damaged or surplus proteins to maintain many mitochondrial functions 29,31,32 and the induced removal of protein complexes during metabolic reprogramming. 80How these proteases are engaged in these different activities, how they are coordinated with protein folding machineries, and whether there might be similar ways of organizing these activities to facilitate other aspects of quality control remains an exciting area of future research.

Limitations of the study
Although we have demonstrated in this work how an early decision can be achieved on whether newly synthesized, mitochondrially encoded proteins directed to OXPHOS assembly or to degradation, we lack insights into the stoichiometry of protein complexes at the PTE and how fate decisions are achieved mechanistically at the level of the PHB/m-AAA complex.For example, it is conceivable that this decision occurs at the stage when proteins are fully synthesized, but it is also possible that this occurs co-translationally.However, our experimental system does not allow differentiation between these possibilities, as degradation at PHB/m-AAA complexes appears to be rapid.In the future, it might be attractive to generate systems where this can be studied, including a structural determination of the yeast and/or human PHB/m-AAA protease complex, which could permit the design of mutants to test various yet mysterious aspects of this mechanism.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for unique reagents should be directed to and will be fulfilled by the lead contact, Martin Ott (martin.ott@gu.se).

Materials availability
All unique materials generated in this study are available upon request to the lead contact.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Yeast strains and culture conditions All strains used in this study were isogenic to W303 MATa {leu2-3.112trp1-1 can1-100 ura3-1 ade2-1 his3-11,195} {phi + } and are listed in the key resources table.Chromosomal modifications were introduced via homologous recombination. 89All primer sequences and plasmids used for this approach are listed in Table S6.
Overnight cultures were grown 16-20 h in respective media and used to inoculate main cultures to a OD 600 of 0.01.Cultures were grown to mid-exponential phase (approx.OD 600 of 1-1.5) and applied for experiments as described below.

METHOD DETAILS
Isolation of mitochondria Mitochondria were isolated as described in Meisinger et al. 90 Therefore, strains grown to mid-logarithmic phase were harvested by centrifugation and washed once in H 2 O.For experiments aiming to analyze OXPHOS assembly, cell cultures were pretreated with 4 mg/ml chloramphenicol for 2 h prior to harvesting cells.After additional centrifugation, pellets were resuspended in 2 ml/g cell wet weight in MP1 buffer (0.1 M Tris, 10 mM dithiothreitol (DTT); pH unadjusted) and incubated for 10 min at 30 C, 170 rpm.The suspension was centrifuged in a Beckman JA-10 rotor for 5 min at 4000 rpm ($2800 rcf), RT and resuspended in 6.7 ml/g cell wet weight MP2 buffer (1.2 M sorbitol, 20 mM KPi and 3 mg/g cell wet weight 20T zymolyase; pH = 7.4).Spheroplasts were created by incubation for 1 h at 30 C, 170 rpm, and subsequently centrifuged in a Beckman JA-10 rotor for 5 min at 4000 rpm ($2800 rcf), 4 C and resuspended in 6.7 ml/g cell wet weight homogenization buffer (10 mM Tris/HCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 1 mM phenylmethylsulfonyl fluoride (PMSF) and 0.6 M sorbitol; pH = 7.4).Homogenization was performed by 10 strokes with a Teflon plunger (Sartorius Stedim Biotech S.A.).After centrifugation at 4 C as described above, the supernatant was collected and the pellet was again resuspended in 6.7 ml/g cell wet weight homogenization buffer.After repeating homogenization and centrifugation, supernatants were merged and centrifuged 3 times with a Beckman JA-10 rotor for 5 min at 4000 rpm ($2800 rcf), 4 C to remove residual cell debris.The final supernatant was centrifuged with a Beckman JA-10 rotor for 12 min at 10,000 rpm ($17,000 rcf), 4 C. Pelleted mitochondria were subsequently resuspended in SH buffer (20 mM HEPES, 0.6 M sorbitol; pH = 7.4) to a concentration of 10 mg/ ml, flash frozen in liquid nitrogen and stored at -80 C until use.

Sucrose gradients
Sucrose density gradients were performed as described in Kehrein et al. 75 with slight adaptations.2 mg of mitochondria were resuspended in 285 ml of lysis buffer (10 mM Tris/HCl, 10 mM KOAc, 0.5 mM Mg(OAc) 2 , 10 mM EDTA, 5 mM 2-mercaptothenaol, 1 mM PMSF, 1% digitonin, 5% glycerol, 1x Complete Protease inhibitor cocktail, 0.1 mM spermidine; pH = 7.4) and lysed for 10 min on ice.285 ml of dilution buffer (same as lysis buffer but without digitonin and Complete Protease inhibitor cocktail) were added and the sample was centrifuged for 10 min at 16,000 rcf, 4 C.The supernatant was transferred into a fresh sample tube and 50 ml thereof were taken as total sample.The rest of the supernatant was carefully layered on top of a 0.3 M -1 M sucrose gradient.These gradients were created with a gradient mixer directly in 4 ml centrifugation tubes suitable for a Beckman SW60 Ti rotor by mixing sucrose gradient solutions (10 mM Tris/HCl, 10 mM KOAc, 0.5 mM Mg(OAc) 2 , 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM PMSF, 0.1% digitonin, 0.1 mM spermidine, 1 M or 0.3 mM sucrose; pH = 7.4).Gradients were centrifuged for 1 h at 60,000 rpm in a Beckman SW60 Ti rotor ($484,000 rcf) at 4 C and subsequently harvested bottom to top in 375 ml fractions with a peristaltic pump.The volume of all fractions was adjusted to 1 ml with ultra-pure H 2 O and 200 ml of 72% TCA were added.Samples were incubated on ice for 30 min, centrifuged for 30 min at 25,000 rcf, 4 C and the resulting pellet was rinsed with 1 ml 100% acetone without resuspending the pellet.After additional centrifugation for 30 min at 25,000 rcf, 4 C, the pellet was resuspended in 70 ml of SDS sample buffer (100 mM Tris/HCl, 4% SDS, 20% glycerol, 0.02% bromophenol blue, 100 mM DTT; pH = 6.8).30 ml of the sample were applied for immunoblotting as described below.
Of note, for additional blue native PAGE analysis, fraction collections after ultracentrifugation were performed at 4 C. 30 ml of each fraction were transferred into a fresh sample tube and 3 ml of NativePAGE Sample additive (5% G-250, Invitrogen) were added.15 ml of each sample were directly applied to blue native PAGE analysis (as described below) and the remaining fractions were TCA precipitated and applied to SDS PAGE as described above.

SDS-PAGE and immunoblotting of whole cell extracts and mitochondrial lysates
Whole-cell extracts were obtained by harvesting a culture volume corresponding to OD600 = 4 and resuspending the pellet in 200 ml 0.1 M NaOH.Suspensions were incubated for 5 min shaking at 1400 rpm at 30 C and pelleted by centrifugation at 16,000 rcf for 5 min at RT.The pellet was resuspended in 150 ml of SDS sample buffer (see above) and incubated for 10 min at 65 C, 1400 rpm.15 ml of the samples were applied for SDS PAGE.Mitochondrial lysates were prepared as described in respective sections.
16% polyacrylamide gels were prepared using the Invitrogen SureCast system.Gels were run and transferred to nitrocellulose membranes (Roth) in the Invitrogen XCell SureLock system according to the manufacturer's instructions.Membranes were blocked in 5% milk powder dissolved in TBS buffer (50 mM Tris/HCl, 0.15 M NaCl; pH = 7.4) for at least 1 h and membranes were subsequently probed with indicated antibodies listed in the key resources table.
Blue Native PAGE and second dimension Blue Native PAGE (BN PAGE) analysis was performed as described in Prestele et al. 81 and Dawitz et al. 82 with slight adaptions.Either 500 mg of untreated mitochondria or 200 ml of radiolabeled samples (corresponding to 500 mg of mitochondria as described below) were pelleted at 10,000 rcf for 10 min at 4 C and resuspended in 75 ml lysis buffer (50 mM Bis-Tris, 100 mM KCl, 2 mM Aminohexanoic acid, 1 mM EDTA, 1x Complete protease inhibitor cocktail, 1 mM PMSF, 2% digitonin, 12% glycerol; pH= 7.2) and incubated for 10 min on ice.Samples were centrifuged at 16,000 rcf for 10 min at 4 C and 7.5 ml of NativePAGE Sample additive (5% G-250, Invitrogen) were added to the supernatant.15 ml of the sample was loaded on NativePAGE 3-12% Bis-Tris gels (Invitrogen) and run in an Invitrogen XCell SureLock system using NativePAGE running buffer (Invitrogen) and NativePAGE Cathode Buffer Additive (Invitrogen).Gels were run at 150 V at 4 C until the running front reached approx.1/3 of the gel.Subsequently, the cathode buffer was replaced with NativePAGE running buffer and the gels were run at 200 V until finished.Gels were transferred to PVDF membranes (Roth) and treated as described in the immunoblotting section above.

35
S methionine radiolabeling of mitochondrially-encoded proteins 2 mg of mitochondria were resuspended in 800 ml translation buffer (0.6 M sorbitol, 150 mM KCl, 15 mM KPi, 20 mM HEPES, 12.68 mM MgCl 2 , 4 mM ATP, 0.5 mM GTP, 5 mM a-ketogluterate, 25 mM creatine phosphate, 5 mg/ml creatine kinase, 12.13 mg/ml all amino acids, except 66.67 mM cysteine and no methionine, pH = 7.4).Samples were incubated for 5 min at 30 C and 22 mCi [ 35 S]-Methionine were added to start the labelling reaction and samples were pulse-labelled for 5 min.Puromycin (final concentration of 80 mM) and non-radioactive methionine to a final concentration of 10 mM were added to stop the labelling reaction.200 ml (corresponding to 500 mg mitochondria) were harvested into 1000 ml of cold SH buffer (20 mM HEPES, 0.6 M sorbitol; pH = 7.4) and centrifuged at 10,000 rcf for 10 min at 4 C.The pellet was resuspended in 75 ml BN PAGE lysis buffer and BN PAGE analysis proceeded as described above.

Crosslinking of nascent mitochondrially-encoded proteins
For crosslinking of newly syntheszied mitochondrially-encoded proteins, 2 mg of mitochondria in a volume of 1 ml were pulse-labelled using [ 35 S]-Methionine as described above, but the labelling was performed for 15 min.10 ml of 20 mM 1,5-difluoro-2,4-dinitrobenzene (DFDNB) crosslinker were added to each sample and incubated for 30 min at 30 C. The reaction was stopped by admixing 100 ml of 1M Tris, pH 8 and 10 ml of 0.2 M non-radioactive Methionine.Mitochondria were pelleted for 10 min at 4 C, 25,000 rcf and lysed by resuspension in 100 ml of 1% SDS and boiling for 1 min at 95 C. Next, 2 ml of lysis buffer (20 mM KPi, 300 mM NaCl, 1% DMM; pH = 7.2) were admixed.10 ml of Anti-ALFA beads (for ALFA tag-based purifications) or 5 ml of indicated antibodies and 50 ml of Protein A Sepharose (for immunoprecipitations) were added.Samples were incubated for 2 h at 4 C tumbling and subsequently centrifuged for 2 min at 4 C, 3000 g.Beads were washed 4 times with 500 ml lysis buffer.For ALFA tag-based purifications, elution was performed by boiling the beads in 100 ml of 1x sample buffer (100 mM Tris/HCl, 4% SDS, 20% glycerol, 0.02% bromophenol blue, 100 mM 2-mercaptoethanol; pH = 6.8) for 1 min.For immunoprecipitations, elution was performed with 0.5 ml 0.1 M Glycine, pH 3, followed by TCA precipitation as described above.

Protein Crosslinking
For crosslinking approaches using ALFA-tagged bait proteins, respective isolated mitochondria were resuspended in 1 ml SH buffer (20 mM HEPES/KOH, 0.6 M sorbitol; pH = 7.4).Thereby, 0.6 mg of mitochondria were used for direct analysis of crosslinking products via immunoblotting, and 6 mg for subsequent purification and mass spectrometric analysis.Indicated crosslinkers were equilibrated at RT and dissolved in DMSO to give 20 mM stock solutions.10 ml of the crosslinker were added to each sample (or DMSO as a control, respectively) and incubated for 30 min at 30 C, 600 rpm.The reaction was quenched by the addition of 100 ml of 1 M Tris/HCl, pH = 8 and 10 ml of 14.3 M 2-mercaptoethanol and incubated for further 10 min at 30 C, 600 rpm.Mitochondria were pelleted at 25,000 rcf for 10 min, 4 C.For direct immunoblot analysis, samples were resuspended in 120 ml of SDS sample buffer and 30 ml were applied for immunoblotting as described above.For purification of bait proteins and crosslinked peptides, mitochondria pelleted after crosslinking were resuspended in 1 ml lysis buffer (10 mM Tris/HCl, 300 mM KCl, 1 mM EDTA, 1% DDM, 1 mM PMSF and 1x Complete Protease Inhibitor Cocktail; pH = 7.4) and incubated for 10 min at 4 C tumbling on a spinning wheel. 1 ml of dilution buffer (10 mM Tris/HCl, 300 mM KCl, 1 mM EDTA, 0.1% DMM, 1 mM PMSF, 1x Complete Protease Inhibitor Cocktail; pH = 7.4) was added and samples were centrifuged at 16,000 rcf for 10 min at 4 C.The supernatant was transferred to 40 ml of ALFA Selector ST beads (pre-equilibrated with dilution buffer) and binding was performed for 2 h at 4 C tumbling on a spinning wheel.The samples were centrifuged for 3 min at 3000 rcf, 4 C and the supernatant was removed.Beads were washed 3 times with 1 ml wash buffer 1 (10 mM Tris/HCl, 0.1% SDS; pH = 7.4), 3 times with wash buffer 2 (10 mM Tris/HCl, 6 M urea; pH = 7.4) and once with elution buffer (50 mM ammonium bicarbonate; pH = 8.0).Beads were resuspended in 50 ml of elution buffer and 0.5 mg of sequencing-grade trypsin (Promega) was added and incubated at 37 C, shaking overnight.The beads were pelleted for 3 min at 3000 rcf, RT and the supernatant was harvested.Beads were washed 2 times with 50 mM ammonium bicarbonate and merged with previously collected supernatants.Finally, peptides were lyophilized and analyzed by mass spectrometry.
Liquid Chromatography and Mass Spectrometry (XL-based Immunoprecipitation) LC-MS/MS instrumentation consisted of an Easy-LC 1200 (Thermo Fisher Scientific) coupled via a nano-electrospray ionization source to an Exploris 480 mass spectrometer (Thermo Fisher Scientific, Bremen, Germany).An in-house packed column (inner diameter: 75 mm, length: 40 cm) was used for peptide separation.A binary buffer system (A: 0.1 % formic acid and B: 0.1 % formic acid in 80% acetonitrile) was applied as follows: Linear increase of buffer B from 4% to 27% within 45 min, followed by a linear increase to 45% within 5 min.The buffer B content was further ramped to 65 % within 5 min and then to 95 % within 5 min.95 % buffer B was kept for a further 5 min to wash the column.Prior to each sample, the column was washed using 5 ml buffer A and the sample was loaded using 8 mL buffer A.
The RF Lens amplitude was set to 55%, the capillary temperature was 275 C and the polarity was set to positive.MS1 profile spectra were acquired using a resolution of 60,000 (at 200 m/z) at a mass range of 350-1750 m/z.The maximum injection time was set to 30 ms.Fragmentation MS/MS spectra were acquired using a resolution of 30,000 (at 200 m/z), an isolation width of 1.4Th, an AGC target of 900% and a HCD relative collision energy of 30%.Dynamic exclusion was enabled for 20 sec using a mass tolerance of 10 ppm.FAIMS was enabled and the CV value was set to -50 and -70 measuring each sample twice exclusively for the data of PHB2 crosslinking experiments.
Proteomics Data Analysis (XL-based Immunoprecipitation) MaxQuant 2.0.3.0 88 and the implemented Andromeda search engine 84 were utilized to correlate acquired MS/MS to spectra to insilico trypsin digestion proteins of the Baker's yeast reference proteome from the Uniprot repository (downloaded November 2021, 6050 sequences, uploaded to PRIDE).Protein N-term acetylation and oxidation of methionine residues were defined as variable modifications.Carbamidomethylation was specified as a fixed modification.The mass tolerance for MS/MS spectra was set to 20ppm.The false discovery rate was controlled using the implemented decoy algorithm on the protein and peptide-spectrum match level to 0.01.The MaxLFQ 85 as well as the match-between runs algorithm was enabled using default settings.The LFQ intensities were log 2 transformed and significantly enriched proteins were identified by applying a two-sided t-test followed by controlling the FDR to 0.05 using a permutation-based approach.Gene Ontology annotations were imported from the Uniprot repository.Mitochondrial localization was annotated from a high-confidence mitochondrial proteome set. 91The Instant Clue software 86 was utilized to inspect the proteomics data.

Purification of native PHB complexes
Mitochondria isolated from wild-type yeast strains and Phb2 FLAG expressing cells were subjected to FLAG-resin purification in triplicates.For each sample, 11 mg of isolated mitochondria stored in SH buffer were pelleted, resuspended in 1 mL activation buffer (20 mM HEPES/KOH pH 7.4, 0.6 M sorbitol, 150 mM KCl, 15 mM KPi pH 7.4, 15 mM MgCl2, 66.7 mM cysteine, 10 mg/mL of alanine, arginine, aspartic acid, asparagine, glutamic acid, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophane, tyrosine and valine, 4 mM ATP, 0.5 mM GTP, 5 mM a-Ketoglutarate, 25 mM creatine phosphate, 0.6 U/mL creatine kinase, 1X EDTA-free protease inhibitor) and incubated at 30 C for 30 min at 400 rpm.Activated mitochondria were centrifuged at 4 C for 10 min at 10,000 rcf followed by resuspension of the pellet in 2 mL lysis buffer (40 mM HEPES, 150 mM KCl, 5 mM ATP, 5 mM Mg acetate, 1 mM PMSF, 5% glycerol, 1% Digitonin; pH = 7.4).Mitochondria were lysed at 4 C for 30 min with constant tumbling followed by clarification of the lysate at 4 C for 10 min, 25000 rcf.The supernatant was diluted with 3 mL dilution buffer (40 mM HEPES pH 7.4, 150 mM KCl, 5 mM Mg acetate, 1 mM PMSF, 0.1% Digitonin; pH = 7.4) and added to 150 ml Anti-FLAG M2 affinity gel (Sigma), pre-equilibrated with dilution buffer.After over night binding at 4 C and constant tumbling, the FLAG resin was subjected to three rounds of washes in 3 ml dilution buffer and one round of FLAG elution by adding 0.3 mL elution buffer (dilution buffer supplemented with FLAG elution peptide at 1X final concentration).The eluate was separated from FLAG beads by centrifugation at 4 C for 10 min at 100 rcf and mixed with 4x NuPAGE LDS sample buffer (ThermoFisher Scientific) for mass-spectrometry analysis.
Protein Digestion (Native PHB complex) FLAG-based enriched eluates were digested using the SP3 digestion technique as described.In brief, washed SP3 beads (Sera-Mag(TM) Magnetic Carboxylate Modified Particles (Hydrophobic, GE44152105050250), Sera-Mag(TM) Magnetic Carboxylate Modified Particles (Hydrophilic, GE24152105050250) from Sigma Aldrich) were mixed equally, and 3 mL of bead slurry were added to each eluate.Acetonitrile was added to a final concentration of 50% and washed twice using 70 % ethanol (V=200 mL) on an in-house made magnet.After an additional acetonitrile wash (V=200 mL), 5 mL digestion solution (10 mM HEPES pH = 8.5 containing 0.1mg Trypsin (Sigma) and 0.1mg LysC (Wako)) was added to each sample and incubated overnight at 37 C. Peptides were desalted on a magnet using 2 x 200 mL acetonitrile.Peptides were eluted in 10 mL 5% DMSO in LC-MS water (Sigma Aldrich) in an ultrasonic bath for 10 min and further cleaned using the StageTip technique (SDB-RP material, Affinisep, France).Dried peptides were then reconstituted in 2.5% formic acid and 2% acetonitrile directly prior to LC-MS/MS analysis.
Liquid Chromatography and Mass Spectrometry (Native PHB complex) Peptides were separated using an in-house column (40cm, 75mm inner diameter, filled with C18 PoroShell 2.7 mm beads).The column temperature was controlled at 50 C using a SON-VARII oven.A binary buffer system (A: 0.1 % formic acid and B: 0.1 % formic acid in 80% acetonitrile) was applied for a total gradient time of 90 min as follows: Linear increase of buffer B from 4% to 27% within 70 min, followed by a linear increase to 45% within 5 min.The buffer B content was further ramped to 65 % within 5 min and then to 95 % within 5 min.95 % buffer B was kept for a further 5 min to wash the column.
The capillary transfer temperature was set to 280 C and the RF lens was set to 55%.MS1 profile spectra were acquired using a resolution of 120,000 (at 200 m/z) at a mass range of 320-1250 m/z and an AGC target of 1 3 106.For MS/MS independent spectra acquisition, 48 equally spaced windows were acquired at an isolation m/z range of 15 Th, and the isolation windows overlapped by 1 Th.The fixed first mass was 200 m/z.The isolation center range covered a mass range of 357-1060 m/z.Fragmentation spectra were acquired at a resolution of 15,000 at 200 m/z using a maximal injection time of 22 ms and stepped normalized collision energies (NCE) of 26, 28, and 30.

Data Analysis (Native PHB complex)
The acquired data were analyzed using DIA-NN 1.8 The spectral library was created in-silico using the reviewed-only Uniport reference protein (Saccharomyces cerevisiae (strain ATCC 204508 / S288c), 6060 entries, downloaded December 2022) with the 'Deep learning-based spectra and RTs prediction' turned on.Protease was set to Trypsin and a maximum of 1 miss cleavage was allowed.N-term M excision was set as a variable modification and carbamidomethylation at cysteine residues was set as a fixed modification.The option 'Quantitative matrices' was enabled.The FDR was set to 1 % and the mass accuracy (MS2 and MS1) as well as the scan window was set to 0 (automatic inference via DIA-NN).Match between runs (MBR) was enabled.The Neuronal network classifier worked in 'double pass mode' and protein interference was set to 'Isoform IDs'.The quantification strategy was set to 'robust LC (high accuracy)' and cross-run normalization was defined as 'RT-dependent'.To identify significantly enriched protein groups, a two-sided t-test was utilized.To correct the multiple testing problem, the FDR was controlled using a permutation-based approach in the Perseus software. 92ole proteome analysis Five individual colonies of wild type and prohibitin deletion strains were used to inoculate precultures with YP media containing 0.1% glucose and 2% galactose.Main cultures containing the 10 ml of fresh media with the same composition were inoculated to an OD 600 = 0.1, and grown for approx.9 hours to an OD 600 = 1.Samples were harvested by centrifugation, washed once in 1ml of ultra-pure water and pellets stored at -80 C until further use.
40 ml of 4% SDS in 100 mM HEPES (pH = 8.5) were pre-heated to 70 C and added to the cell pellet for further 10 min incubation at 70 C on a ThermoMixer (shaking: 550 rpm).The protein concentration was determined using the 660 nm Protein Assay (Thermo Fisher Scientific, #22660).20 mg of protein was subjected to tryptic digestion.Proteins were reduced (10 mM TCEP) and alkylated (20 mM CAA) in the dark for 45 min at 45 C. Samples were subjected to an SP3-based digestion. 93Washed SP3 beads (Sera-MagÔ Magnetic Carboxylate Modified Particles (Hydrophobic, GE44152105050250), SeraÔ(TM) Magnetic Carboxylate Modified Particles (Hydrophilic, GE24152105050250) from Sigma Aldrich) were mixed equally, and 3 ml of bead slurry were added to each sample.Acetonitrile was added to a final concentration of 50% and washed twice using 70 % ethanol (V=200 ml) on an in-house made magnet.After an additional acetonitrile wash (V=200ml), 5 ml digestion solution (10 mM HEPES pH = 8.5 containing 0.5mg Trypsin (Sigma) and 0.5mg LysC (Wako)) was added to each sample and incubated overnight at 37 C. Peptides were desalted on a magnet using 2 x 200 ml acetonitrile.Peptides were eluted in 10 ml 5% DMSO in LC-MS water (Sigma Aldrich) in an ultrasonic bath for 10 min.Formic acid and acetonitrile were added to a final concentration of 2.5% and 2%, respectively.Samples were stored at -20 C before subjection to LC-MS/MS analysis.
Liquid Chromatography and Mass Spectrometry (Whole proteome) LC-MS/MS instrumentation consisted of an Easy-LC 1200 (Thermo Fisher Scientific) coupled via a nano-electrospray ionization source to an Exploris 480 mass spectrometer (Thermo Fisher Scientific, Bremen, Germany).An in-house packed column (inner diameter: 75 mm, length: 40 cm) was used for peptide separation.A binary buffer system (A: 0.1 % formic acid and B: 0.1 % formic acid in 80% acetonitrile) was applied as follows: Linear increase of buffer B from 4% to 27% within 70 min, followed by a linear increase to 45% within 5 min.The buffer B content was further ramped to 65 % within 5 min and then to 95 % within 5 min.95 % buffer B was kept for a further 5 min to wash the column.
Prior to each sample, the column was washed using 5 ml buffer A and the sample was loaded using 8 ml buffer A. The RF Lens amplitude was set to 55%, the capillary temperature was 275-280 C and the polarity was set to positive.MS1 profile spectra were acquired using a resolution of 120,000 (at 200 m/z) at a mass range of 320-1150 m/z and an AGC target of 1 3 106.For MS/MS independent spectra acquisition, 48 equally spaced windows were acquired at an isolation m/z range of 15 Th, and the isolation windows overlapped by 1 Th.The fixed first mass was 200 m/z.The isolation center range covered a mass range of 357-1060 m/z.Fragmentation spectra were acquired at a resolution of 15,000 at 200 m/z using a maximal injection time of 22 ms and stepped normalized collision energies (NCE) of 26, 28, and 30.The default charge state was set to 3. The AGC target was set to 3e 6 (900% -Exploris 480).MS2 spectra were acquired in centroid mode.
Proteomics Data Analysis (Whole proteome) DIA-NN (Data-Independent Acquisition by Neural Networks) v 1.8 87 was used to analyze data-independent raw files.The spectral library was created using the reviewed-only Uniport reference protein (Saccharomyces cerevisiae (strain ATCC 204508 / S288c), 6060 entries, downloaded December 2022) with the 'Deep learning-based spectra and RTs prediction' turned on.Protease was set to Trypsin and a maximum of 1 miss cleavage was allowed.N-term M excision was set as a variable modification and carbamidomethylation at cysteine residues was set as a fixed modification.The peptide length was set to 7 -30 amino acids and the precursor m/z range was defined from 340 -1200 m/z.The option 'Quantitative matrices' was enabled.The FDR was set to 1 % and the mass accuracy (MS2 and MS1) as well as the scan window was set to 0 (automatic inference via DIA-NN).Match between runs (MBR) was enabled.The Neuronal network classifier worked in 'double pass mode' and protein interference was set to 'Isoform IDs'.The quantification strategy was set to 'robust LC (high accuracy)' and cross-run normalization was defined as 'RT-dependent'.
The 'pg' (protein group) output (MaxLFQ intensities were further processed using Instant Clue including and pairwise comparison using a t-test followed by a permutation-based FDR correction (5%).Uniprot-based Gene Ontology annotations were added based on the protein group identifiers.Mitochondrial localization was annotated from a high-confidence mitochondrial proteome set. 91

SPELL analysis
The Serial Pattern of Expression Levels Locator (SPELL) tool of the yeast genome database 51 was searched for the genes PHB1 and PHB2, presenting the Prohibitin complex and 50 was used as the number of result genes to show.No additional filtering tools with dataset tags was used to perform an unbiased analysis.The automatically conducted GO-term enrichment analysis was used and reported p-values as well as the number of genes per group are presented as bubble blot.

Polarographic measurement of substrate oxidation
Measurement of NADH oxidation with isolated mitochondria was performed with a protocol adapted from Berndtsson et al. 94 A Clark-type oxygen electrode (Dual Digital Model 20, Rank Brothers LTD; 4 ml perspex dissolved oxygen electrodes with a PicoLog ADC-20, Pico Technology) was prepared according to the supplied manual, using respiration buffer (20 mM HEPES, 0.6 M sorbitol, 10 mM H 3 PO 4 , 2 mM MgCl 2 , 1 mM EGTA, 0.1% BSA; pH = 7.4) or 30 g/l Na 2 SO 3 dissolved in H 2 O to set the system to 100% or 0% oxygen concentration, respectively.1.5 ml of respiration buffer was filled into each oxygen electrode chamber and after reaching a constant baseline, 60 mg of coupled isolated mitochondria where added.Subsequently 20 ml of 0.1 M NADH where supplemented and basal respiration (state 2) was recorded.Next, 20 ml of 20 mM ADP was admixed and the respiration under phosphorylating conditions (state 3) was measured.Finally, 6 ml of 80 mM KCN were added to the chamber to stop mitochondrial respiration and assess the specificity of the assay.Data was recorded with PicoLog 6 data logging software using 1 sec measurement intervals.

Spectrophotometry and respiratory chain activity
To create mitoplasts, 2 mg of isolated mitochondria were pelleted by centrifugation (25,000 rcf, 10 min, 4 C) and resuspended in 1 ml of swelling buffer (20 mM HEPES/KOH; pH = 7.4).After 30 min of incubation on ice, mitoplasts were collected by centrifugation at 25,000 g, 20 min at 4 C. Pelleted mitoplasts were resuspended in swelling buffer to reach a final concentration of 10 mg/ml.
For the spectroscopic measurement of CIII activity, 5 ml of mitoplasts were added to 915 ml of swelling buffer and incubated at 30 C for 2 min.Subsequently 10 ml of KCN, 50 ml cytochrome c and 10 ml BSA were admixed.Absorption at 550 nm was measured for 1 min, 2.5 ml of decylubiquinol was added and absorption at 550 nm measured for additional 5 min.To analyze specificity of the assay, the measurement was repeated for each sample by adding 10 ml of 0.4 mM antimycin A to the swelling buffer/mitoplast mixture prior to 2 min incubation at 30 C.
To spectroscopically analyze CIV activity, 915 ml swelling buffer were incubated for 2 min at 30 C. Subsequently, 10 ml of Antimycin A, 50 ml of reduced cytochrome c and 10 ml BSA were added and the solution was incubated for additional 4 min at 30 C. The activity within the solution was recorded for 1min by analyzing the absorption at 550 nm, before adding 5 ml of mitoplasts (10 mg/ml) and measuring the activity for another 5 min.Specificity of the activity measurement was analyzed by repeating the measurement for each sample with the addition of 10 ml KCN to the 915 ml swelling buffer before incubation at 30 C.

Analysis of cellular growth
Cells were cultivated as described above and washed once in YP media without carbon source.Cells were pelleted and resuspended in YP media without carbon source to a final concentration of OD 600 =1 /ml.Serial 1:10 dilutions were prepared and 3 ml of each dilution were spotted on YP agar plates containing either 2% of fermentable or 2% of non-fermentable carbon source (glucose vs. glycerol).Plates were incubated at 30 C for 2 days and photographed.

Confocal microscopy
Oxidative stress was monitored via the reactive oxygen species (ROS) driven conversion of non-fluorescent dihydroethidium (DHE) to fluorescent ethidium (Eth).Approximatively, 1x10 6 cells were harvested at indicated timepoints and incubated with phosphate-buffered saline (PBS, 25 mM potassium phosphate; 0.9% NaCl; pH = 7.2) containing 2.5 mg/l DHE for 10 min in the dark.Samples were then pelleted via centrifugation and washed once in PBS.Cells were pipetted on 3% agar slides and visualized as Z-stacks using an LSM 800 Airyscan confocal microscope (Zeiss) equipped with an 63x/1.40 oil objective.The open-source software Fiji was used to process and analyze the obtained micrographs. 833-dimensional Gaussian filtering (xs | ys | zs = 1) was applied to decrease image noise, followed by background subtraction (rolling ball radius = 50 pixels) Three-dimensional data was projected using the maximumintensity projection method.Pictures within an experiment were captured and processed in the same way.
Quantification of Eth intensity within mitochondria was conducted in Fiji.Therefore, images where segmented by using the Threshold function.Individual mitochondria were separated by applying the watershed function.Particle analysis was performed using standard settings and the ''exclude on edges'' function.The resulting mask was used to measure integrated densities by redirecting to the original micrograph and fold change values using wild type cells for normalization were calculated.

QUANTIFICATION AND STATISTICAL ANALYSIS
Quantitative statistically analyzed data are presented as dot plots with mean (square) ± standard error of the mean (SEM) and median (centre line), as well as single data points.Presented number of n describes the number of individual clones (=biological replicates) used for the analysis.Outliers were identified with the 2.2-fold interquartile range labelling rule and normality of the data was tested via Shapiro-Wilk's test.Homogeneity of variances was evaluated by a Levene's test.The means of two independent groups were compared by a two-tailed independent sample t-text.Of note, all assumptions to perform this test were met in all data sets.A detailed description of statistical tests performed and respective results are listed in Table S7.

Figure 1 .Figure 2 .
Figure1.The PHB/m-AAA protease complex is in proximity to the mitoribosomal PTE (A) Scheme of experimental approach for BioID assays.Bait proteins were chromosomally equipped with the mutant biotin ligase BirA*.Strains were grown in respiratory conditions and in the presence of biotin, prior to isolation of mitochondria.Biotinylated proteins were purified using streptavidin beads and identified via mass spectrometry.(B) Volcano plot for the BioIDome of Mrp20 (uL23) and Oxa1.In both cases, the soluble matrix protein Kgd4 was used as a background control.Deposited data from Singh et al.33 were re-evaluated for this study.(C) Combined BioIDome of the mitoribosomal polypeptide tunnel exit (PTE) and indicated PTE ligands.Identified prey proteins were grouped according to their physiological function.Analysis was performed with deposited data from Singh et al.33 (D) Scheme of the experimental approach used for sucrose density gradient centrifugation.Isolated mitochondria were lysed using digitonin and the lysate was subjected to density gradient centrifugation.(E) SDS-PAGE (upper), Ponceau S staining of the membrane after transfer (middle), and BN-PAGE (lower) of sucrose gradients performed with mitochondrial extracts of wild-type (WT) cells.(F and G) SDS-PAGE of sucrose gradients performed with WT mitochondrial extract (F) and from PHB1 and PHB2 double knockout cells (phb1/2DD, G).Extracts in (F) were treated with RNase A prior to density gradient centrifugation to selectively destroy the mitoribosome.See also FigureS1for supplemental information.

Figure 3 .
Figure 3.The dynamics of the prohibitin interactome confirm an involvement in OXPHOS biogenesis (A) Scheme of the experiment to follow the dynamics of the proximity interactome of Phb1 with chemical crosslinking during metabolic reprogramming.(B) Combined data of crosslinking/MS of Phb1 under respiration or fermentation.Mitochondria from 5 biological replicates were exposed to crosslinkers and processed as depicted in (A).The identified proteins were grouped and color-coded according to their molecular function.(C) Schematic model of the metabolism-dependent proximity interactomes of the prohibitin complex.See also Figure S3 for supplemental information.

Figure 4 .
Figure 4. Absence of the PHB complex impairs respiratory function (A) BN-PAGE of mitochondria isolated from wild-type (WT) and the prohibitin deletion (phb1/2DD) strain, lysed with digitonin (Dig) or n-Dodecyl-beta-D-Maltoside (DDM).Proteins were detected with indicated antibodies.(B) Quantitative mass spectrometry of wild-type (WT) and prohibitin deletion (phb1/2DD) cells.Five replicates of cells were grown to logarithmic phase in galactose and their proteome was determined.Boxplot analysis of log 2 fold changes of cellular or mitochondrial proteins and subunits of CII-CV comparing WT and phb1/2DD cells.Dashed lines indicate 1.5-fold up-and down-regulation.(C) Basal or phosphorylating NADH-driven oxygen consumption of mitochondria from strains described in (A).Data are represented as mean ± SEM (n = 3 biological replicates).(D) Respiratory control ratio from samples analyzed in (C), determined as the ratio of ADP-stimulated oxygen consumption over basal respiration.Data are represented as mean ± SEM (n = 3 biological replicates).(E) Heme absorbance, as determined in detergent lysates of strains described above by ultraviolet-visible (UV-vis) spectroscopy.Data represent the mean of 3 biological replicates.(F) Quantification of respiratory chain complex activity analyzed for cytochrome c reductase (CIII) and cytochrome c oxidase (CIV).Data are represented as mean ± SEM (n = 3 biological replicates).(G) Growth analysis of WT and phb1/2DD strain, each either overexpressing Phb1 and Phb2 (Phb1/2 OE ) or harboring empty vector controls (Ctrl.).Strains were spotted in 10-fold serial dilution onto media allowing fermentation or respiration, respectively.(H and I) Fluorescence microscopic analysis of the reactive oxygen species (ROS)-driven conversion of non-fluorescent dihydroethidium (DHE) to fluorescent ethidium (Eth) in WT and phb1/2 DD strains.Representative fluorescence microscopic pictures (H) and quantification of Eth fluorescence intensity (I) are shown.Data in (I) are represented as mean ± SEM (n = 4 biological replicates).

Figure 5 .
Figure 5.The PHB/m-AAA protease complex interacts with newly synthesized mitochondrially encoded proteins and is required for efficient OXPHOS complex assembly (A) Schematic representation of the experimental strategy used to radioactively label newly synthesized, mitochondrially encoded proteins and visualize their assembly into respiratory chain complexes.(B-E) Pulse-chase35 S-methionine in-organello radiolabeling of wild-type (WT) and prohibitin deletion (phb1/2DD) strains, as illustrated in (A).Immunoblots of firstdimension BN-PAGE membranes, either stained with Coomassie, autoradiography detection, or decorations with indicated antibodies are shown in (B), as well as quantifications of the autoradiography signals in (C) and (D).First-dimension BN-PAGE gels were applied for second-dimension SDS-PAGE and autoradiographs and immunoblots are presented in (E).Data in (C) and (D) are represented as mean ± SEM (n = 3 biological replicates).(F-H) Pulse35 S-methionine in-organello radiolabeling of WT mitochondria.First-dimension BN-PAGE membranes, either stained with Coomassie, autoradiography detection, or decorations with indicated antibodies are shown in (F), as well as a quantification (normalized to the 30-s value) of the autoradiography signal from the total or from the PHB/m-AAA complex (H).First-dimension BN-PAGE gels were applied for second-dimension SDS-PAGE and autoradiographs are presented in (G).Data in (H) are represented as mean ± SEM (n = 3 biological replicates).See also FigureS4for supplemental information.

Figure 6 .
Figure 6.The PHB/m-AAA protease complex is part of a molecular triage that directs newly synthesized proteins to OXPHOS complex assembly or to degradation (A-C) Pulse-chase 35 S-methionine in-organello radiolabeling of wild-type (WT) strains, as well as of the proteolysis-compromised Yta10 E559Q mutant.Firstdimension BN-PAGE membranes, either stained with Coomassie, autoradiography detection, or decorations with indicated antibodies are shown in (A), together with the quantifications of total and PHB/m-AAA-associated autoradiography signal (B), as well as second-dimension autoradiographs (C).Data in (B) are represented as mean ± SEM (n = 3 biological replicates).(D-F) Pulse-chase 35 S-methionine in-organello radiolabeling of strains with Cytb expression uncoupled from Cbp3 (cox2::COB, cob::ARG8 m ).This strain contains a modified mitochondrial genome, resulting in the ectopic expression of Cytb from a mRNA containing the 5 0 untranslated region of COX2.Translation products of a strain with additional deletion of CBP3 (cbp3D) were compared with isogenic control cells (Ctrl.)Coomassie-stained membranes of first-dimension BN-PAGE, and autoradiographs are shown (D), together with respective quantifications of total and PHB/m-AAA-associated autoradiography signal (E), as well as seconddimension autoradiographs (F).Data in (E) are represented as mean ± SEM (n = 3 biological replicates).

Figure 7 .
Figure 7. Model of early fate decision of mitochondrial encoded OXPHOS

TABLE d
B Yeast strains and culture conditions d METHOD DETAILS B Isolation of mitochondria B Sucrose gradients B SDS-PAGE and immunoblotting of whole cell extracts and mitochondrial lysates B Blue Native PAGE and second dimension B Proteomics Data Analysis (Whole proteome) B SPELL analysis B Polarographic measurement of substrate oxidation B Spectrophotometry and respiratory chain activity B Analysis of cellular growth B Confocal microscopy d QUANTIFICATION AND STATISTICAL ANALYSIS (Continued on next page) Data and code availability d Original acrylamide gel and immunoblot images, autoradiograms and microscopic images are deposited in Mendeley Data.In addition, all quantifications made from these images are provided as raw data in the form of EXCEL files.Raw data from spectrophotometric quantifications of heme levels and measurements of oxygen levels are provided in the form of EXCEL files in Mendeley Data.Images from agar plates (growth assays) are also deposited in Mendeley Data.The DOI is listed in the key resources table.Mass spectrometry data is deposited in the Proteomics Identification Database (PRIDE; https://www.ebi.ac.uk/ pride/).Respective accession numbers are listed in the key resources table.d This paper does not report original code.d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.