Mitochondria serve a multitude of functions, including ATP production, metabolism, and proteostasis that all require import of proteins from the cytosol (Pfanner et al., 2019). To ensure proper function, multiple mechanisms facilitate faithful protein targeting and efficient protein import into the organelle. One of these mechanisms is carried out by a protein on the mitochondrial outer membrane (MOM) named ATAD1 (in humans; Msp1 in yeast). ATAD1 has important roles in various biological contexts (Fresenius and Wohlever, 2019; Wang and Walter, 2020), including clearing out mistargeted proteins from the mitochondria (Chen et al., 2014; Nuebel et al., 2021; Okreglak and Walter, 2014), extracting mitochondrial precursor proteins stuck in the protein translocase channel during protein import overload (Weidberg and Amon, 2018), and mediating apoptosis by removing a BCL-family member protein (Winter et al., 2021). In addition, ATAD1 has also been implicated in the regulation of synaptic plasticity by mediating the endocytosis of neurotransmitter receptors from post-synaptic membranes (Wang and Walter, 2020; Zhang et al., 2011). Among these, the removal of mistargeted tail-anchored (TA) proteins from the MOM is ATAD1’s best understood function to date.

TA proteins are integral membrane proteins that are embedded in the membrane by a hydrophobic stretch at the extreme C-terminus. They are targeted to organellar membranes post-translationally. The endoplasmic reticulum (ER)-targeted TA proteins are delivered to the ER membrane by the TRC proteins (Get proteins in yeast) (Favaloro et al., 2008; Schuldiner et al., 2008; Stefanovic and Hegde, 2007). From there, they either stay on the ER membrane or travel to the peroxisome or other membranes along the secretory pathway. By contrast, TA proteins on the MOM either insert spontaneously (Chio et al., 2017) or are actively inserted by the mitochondrial import machinery (Doan et al., 2020). The partitioning of the ER- and the mitochondria-targeted TA proteins relies on different biophysical properties of their targeting signals (consisting of the transmembrane domain and the short segment C-terminal to it) (Chio et al., 2017). Due to a number of similarities between the ER- and mitochondria-targeted TA proteins, mistargeting happens in both directions. Recently, our lab and the Rutter lab independently discovered that Msp1/ATAD1 could recognize the mislocalized TA proteins and extract them from mitochondria (Chen et al., 2014; Okreglak and Walter, 2014). The extracted proteins are then correctly inserted into the ER membrane, from which they travel to their cognate organelle or become degraded by the proteasome (Dederer et al., 2019; Matsumoto et al., 2019).

ATAD1 is a member of a large family of proteins called the AAA (ATPase Associated with diverse cellular Activities) proteins. It is strictly conserved from yeast to humans. Previously, we determined a series of cryo-EM structures of the cytosolic domain of the Chaetomium thermophilum (C.t.) Msp1 in complex with a peptide substrate (Wang et al., 2020). The structures revealed that Msp1 follows the general principle established for many AAA proteins (Gates et al., 2017; Monroe et al., 2017; de la Peña et al., 2018; Puchades et al., 2020; Puchades et al., 2017; Zehr et al., 2017): six Msp1 subunits form a helical hexamer resembling a right-handed lock-washer that surrounds the substrate in a hydrophobic central pore. Elements at the intersubunit interface couple ATP hydrolysis with stepwise subunit translocation to unfold the substrate peptide in its central pore. The structures also revealed elements in Msp1 that are adapted for its function of removing membrane proteins, one of which is an unusually hydrophobic central pore. Whereas most AAA proteins extend only one short loop (pore-loop 1) containing a conserved aromatic amino acid to directly contact the substrate, Msp1 utilizes a total of three aromatic amino acids within two short loops to contact the substrate. We previously proposed that these aromatic amino acids enhance the bulkiness and hydrophobicity of the pore-loops, giving Msp1 a firm grip on the substrate to prevent it from backsliding, a feature that may be important for pulling hydrophobic membrane proteins out of the lipid bilayer.

To test the importance of the additional aromatic amino acids, we previously mutated the second (Y167 in Saccharomyces cerevisiae (S.c.), corresponding to Y188 in C.t.) and the third (H206 in S.c., corresponding to H227 in C.t.) aromatic amino acids to aliphatic or polar amino acids and measured their effects on yeast growth. Surprisingly, while the Y167A mutation impacted Msp1’s activity, the Y167V or H206A mutation did not have a significant impact on Msp1’s function. We hypothesized that this was perhaps because the yeast growth assay provides an indirect readout (cell death) of Msp1’s function. In other words, deficiencies in the protein’s activity could be masked by compensatory pathways promoting cell survival. Therefore, in this study, we aimed to establish an assay that would allow the direct visualization of protein mislocalization.

In addition to addressing these remaining questions concerning the pore-loops, we also explored unique structural elements in ATAD1 that are not observed in its closely related family members. Within the large AAA protein family, ATAD1 belongs to the ‘meiotic clade’ (AAA_MC). Members of this clade include microtubule severing proteins such as katanin and spastin, as well as the protein that disassembles the ESCRT peptides, Vps4 (Frickey and Lupas, 2004). One of the unique structural elements shared by these family members is a C-terminal extension of the classic AAA protein fold, helix α12. α12 is important in hexamer assembly and protein function for AAA_MC proteins (Sandate et al., 2019; Vajjhala et al., 2008). The secondary structure prediction of ATAD1, however, shows the lack of α12, and instead predicts a longer α11 (the alpha helix proceeding it, Figure 1—figure supplement 1). Of note, there are several disease-relevant mutations clustered around α11, making it particularly important to understand this region in molecular detail: R9H, D221H, and E290K are found in schizophrenia patients (Umanah et al., 2017); H357Rfs*15 (a frame shift mutation resulting in a 10 amino acid extension at the C-terminus) is found in a family of patients with encephalopathy (Piard et al., 2018); and E267stop results in a truncated protein that is missing a large portion of the C-terminal domain and is found in a family of patients suffering from hypertonia, seizure, and death (Ahrens-Nicklas et al., 2017). In the C.t. Msp1 structure, we did not observe density for α11, which could be due to degradation or lack of rigid secondary structure, leaving the structure and function of α11 unresolved. To address these questions, we determined the structure of ATAD1 and tested the individual functional contributions of salient structural features in a microscopy-based mislocalization assay.

To obtain a homogenous sample for structural analysis, we expressed human ATAD1 lacking the first 40 amino acids (consisting of the transmembrane helix) with an N-terminal His-tag and bearing a commonly used ‘Walker B’ mutation that inactivates ATP hydrolysis but retains ATP binding (Δ40-ATAD1^E193Q). Δ40-ATAD1^E193Q formed oligomers as assessed by size exclusion chromatography (Figure 1—figure supplement 2). We took the fraction corresponding to hexamers and incubated it with ATP before preparing samples for cryo-EM imaging.

3D classification of the particles resulted in two distinct hexameric structures. In the first structure, the six ATAD1 subunits (M1–M6) are arranged in a right-handed spiral with an open seam between the top and the bottom subunits (45,003 particles analyzed, Figure 1—figure supplement 3B, Table 1), an arrangement resembling the ‘open state’ structure of Msp1 (Wang et al., 2020), as well as other related AAA proteins (Cooney et al., 2019; Han et al., 2020; Su et al., 2017; Sun et al., 2017; Twomey et al., 2019; Zehr et al., 2017; Figure 1A). Refinement of this structure resulted in a map with an average resolution of 3.7 Å. In the second structure, the subunits are arranged in a similar fashion as the first, with the exception of M6, which is in transition to convert into M1 as part of the AAA protein functional ATPase cycle and now bridges M1 and M5, closing the hexameric ring (96,577 particles analyzed). This structure closely resembles the ‘closed state’ structure observed in Msp1 (Figure 1B). Refinement of this structure yielded a map with an average resolution of 3.2 Å with most of the side chain densities clearly visible, which allowed us to build an atomic model. Noticeably, the density of the M6 subunit in the ‘closed state’ structure was less well defined than the rest of the complex (local resolution ranging from 4 to 6 Å, Figure 1—figure supplement 3C), indicating a mixture of different states. A similar dynamic nature of this subunit was also observed in the Msp1 structure (Wang et al., 2020) but the density was too poorly resolved to be interpreted. By contrast, for ATAD1, we could identify secondary structure elements clearly, which enabled us to build a polyalanine model for this subunit. In doing so, we were able to analyze the interactions between M6 and neighboring subunits in the closed state, which was not possible for Msp1.

Figure 1 with 4 supplements see all

Download asset Open asset

A unique α-helix at the C-terminus mediates intersubunit interactions

While examining the interactions of the mobile subunit (M6) in the closed conformation, we noticed that it is held in place by contacts on both sides: on the side of M5, substantial contacts exist between the large and the small subdomains of the two subunits (Figure 1—figure supplement 4D). By contrast, on the side of M1, M6 is held in place by two contact points. One structural element stood out as it was not observed previously in the C.t. Msp1 structure (Wang et al., 2020): an α-helix at the extreme C-terminus of M6 points toward the core β sheet of the large subdomain of M1. We next asked whether this α-helix also mediates intersubunit contacts within the other subunits in the spiral. Indeed, slightly extended EM density for this α-helix exists in other subunits as well (Figures 1C and 2A). We modeled an α-helix (α11, amino acids 333–346) followed by a short turn (amino acids 347–351) into this density. Density beyond that (amino acids 352–361) was not clearly visible indicating its lack of rigid structure. A similar α-helix existed in the crystal structure of the monomeric S.c. Msp1 (Wohlever et al., 2017). However, the last few amino acids in that α-helix in the S.c. Msp1 were replaced by amino acids left from protease cleavage, making it difficult to interpret the original structure of this region.

Figure 2 with 2 supplements see all

Download asset Open asset

In ATAD1’s helical assembly, six α11s form a staircase beneath pore-loop 3, which does not contact the substrate directly but forms an interconnected network with pore-loops 1 and 2 that directly engage the substrate (Figure 2—figure supplement 1). Zooming in on α11 reveals that it lies at the interface between two adjacent subunits (Figure 2A). It forms extensive contacts with both the rest of the subunit and the counterclockwise (when viewed from the cytosol-facing side, as shown in Figure 4) adjacent subunit, bridging between the two (Figure 2). In α11_M4, for example, L336, A339, and I340 pack against the core of M4 (Figure 2A); M343 rests on the interface created by M4 and M5, and A349 points to a hydrophobic groove formed by pore-loop 3 of the M5. Finally, both the side chain and the backbone carbonyl of S346 form hydrogen bonds with R254 of M5 (Figure 2D). A few charged amino acids (R254, K342, K344, K345, K347, E341, and D348) point into the cytosol (Figure 2C and E).

Interestingly, other AAA_MC proteins such as katanin, spastin, and Vps4 are structured differently in this region (Han et al., 2020; Han et al., 2017; Sandate et al., 2019; Shin et al., 2019). For those proteins, α11 is shorter (consisting of three helical turns instead of four as seen in ATAD1). It is followed by a loop and another short helix, α12, that reaches across the intersubunit interface to contact the clockwise adjacent subunit (Figure 2—figure supplement 2). α12 is functionally important as its deletion impacts oligomer assembly and protein function both in vitro and in vivo (Vajjhala et al., 2008). Since ATAD1 lacks α12, and instead has a longer α11, we hypothesized that α11 could also mediate hexamer assembly. If so, we expected that a mutant version might fail to remove mislocalized membrane proteins in cells lacking wild-type (WT) ATAD1, but be unable to poison WT ATAD1’s activity, as it would not be able to incorporate into WT hexamers. In other words, we expected it to display a recessive phenotype, rather than a dominant one. To test this notion, we next sought to establish an assay that distinguishes between dominant and recessive ATAD1 mutations and allows us to measure the activity of ATAD1 variants in a direct and quantifiable way.

Direct visualization and quantification of ATAD1’s activity in cells

One of ATAD1’s established substrates is Gos28, a Golgi-localized TA SNARE protein. In cells lacking functional ATAD1, Gos28 localizes to mitochondria (Chen et al., 2014). We thus sought to use the localization of Gos28 as a readout for ATAD1’s activity (Figure 3A). To analyze the activity of ATAD1 mutants in the absence of the WT enzyme, we deleted both alleles of ATAD1 in HeLa cells using CRISPR/Cas9 (Figure 3—figure supplement 1). We next stably expressed Gos28 with an N-terminal EGFP tag in the ATAD1^-/- cells and used confocal microscopy to visualize the localization of EGFP-Gos28. In ATAD1^-/- cells, we observed two populations of Gos28 molecules: one visualized as an extended perinuclear region characteristic of the Golgi apparatus and the other in a spread-out network (Figure 3B, top row). Overlap of the EGFP signal with the signal from the mitochondria stain, MitoTracker Deep Red, showed that the latter population corresponds to the mitochondrial network, indicating that in ATAD1^-/- cells, a portion of EGFP-Gos28 molecules are mislocalized to mitochondria, in line with a previous study (Chen et al., 2014). We then expressed WT ATAD1 labeled with a C-terminal HaloTag in the ATAD1^-/- cells (Figure 3B, middle row) and observed a prominent shift of EGFP-Gos28 signal toward a Golgi-like distribution, indicating that ATAD1 cleared mislocalized EGFP-Gos28 from the mitochondria. Finally, we expressed the ATAD1 bearing a mutation that inactivates its ATPase activity, ATAD1^E193Q. The result mimicked the localization pattern seen in the ATAD1^-/- cells (Figure 3B, top row), indicating that the ability of clearing mislocalized EGFP-Gos28 was dependent on ATAD1’s enzymatic activity. We similarly observed ATAD1-dependent removal of mistargeted Pex26, a peroxisomal TA protein (Matsumoto et al., 2003), from mitochondria, suggesting that monitoring the clearance of a mislocalized protein is a reliable method to examine ATAD1-dependent TA protein extraction (Figure 3—figure supplement 2).

Figure 3 with 3 supplements see all

Download asset Open asset

Figure 3—source data 1

Data analysis pipeline for live-cell imaging results.

https://cdn.elifesciences.org/articles/73941/elife-73941-fig3-data1-v2.zip

Download elife-73941-fig3-data1-v2.zip

In addition to expressing the reporter in the ATAD1^-/- cell line, we stably expressed Gos28 in a WT HeLa cell line. We expected ATAD1 variants that impact its ability to form proper oligomers to induce Gos28 mislocalization in the ATAD1^-/- cell line, but not in the WT cell line. By contrast, inactive ATAD1 variants that retain their ability to assemble into hexamers should act as dominant inhibitors of the enzyme and induce substrate mislocalization.

To test the reliability of the system, we first expressed a known dominant-negative mutant, ATAD1^E193Q, within the WT reporter cells (Figure 3—figure supplement 3). Expression of ATAD1^E193Q induced mislocalization of EGFP-Gos28 to mitochondria, indicating that ATAD1^E193Q was incorporated into hexamers, poisoning WT activity. As expected, expression of the WT ATAD1 or an empty vector did not induce EGFP-Gos28 mislocalization in this background.

Before proceeding to test the function of α11, we wanted to see if we could quantify the degree of EGFP-Gos28 mislocalization in an unbiased fashion and use it as a readout for ATAD1’s activity. To this end, we developed a data analysis pipeline using CellProfiler (McQuin et al., 2018; Figure 3C, Figure 3—source data 1). In brief, we first identified cells expressing ATAD1-HaloTag, using the cell-permeable JF549 dye (Grimm et al., 2017) to label and visualize Halo-tagged ATAD1. We then measured the colocalization of EGFP-Gos28 and mitochondria (stained with MitoTracker) in ATAD1-expressing cells using the Pearson correlation coefficient (PCC) as the metric, measured between the two channels for each cell (Figure 3D). As shown in Figure 3D, expression of an empty vector or ATAD1^E193Q in the ATAD1^-/- reporter cell line led to a PCC of 0.31±0.022 and 0.31±0.017, respectively. By contrast, transiently expressing WT ATAD1 showed a significantly lower PCC of 0.13±0.015 (p<0.0001), confirming the visually evident changes in EFGP-Gos28 localization and validating this approach for the evaluation of functional mutants of ATAD1.

With the EGFP-Gos28 reporter assay established in both an ATAD1^-/- background and a WT background, we next used it to test α11’s function. To this end, we made two truncated versions of ATAD1: one in which we deleted 30 amino acids at the C-terminus (removing α11 and everything C-terminal to it, ATAD1Δα11, Figure 1C), and another in which we kept α11 and removed everything C-terminal to it (ATAD1_1-351, Figure 1C). We first expressed these variants in the ATAD1^-/- cells. As shown in Figure 4A, EFGP-Gos28 showed significant mislocalization to mitochondria in cells expressing ATAD1Δα11 but not ATAD1_1-351, suggesting that the α11 helix is required for its function. By contrast, when expressed in the WT cells, ATAD1Δα11 did not induce significant mislocalization of EGFP-Gos28 (Figure 4B), indicating its recessive phenotype, that is, its lack of incorporation into and inactivating the WT hexamers present in these cells (the average values for the biological replicates are shown in Figure 4—figure supplement 1A and B). While expression of ATAD1Δα11 did not induce substrate mislocalization on average, it was evident by eye that a small population of WT cells expressing ATAD1Δα11 showed EGFP-Gos28 mislocalization (Figure 4—figure supplement 2A). We assume that this effect resulted from high ATAD1Δα11 expression in these outlier cells that compensated for ATAD1Δα11’s reduced oligomerization ability and allowed assembly into WT hexamers. Indeed, we observed a positive correlation between ATAD1Δα11 expression level and the degree of Gos28 mislocalization: cells that expressed more ATAD1Δα11 showed a higher degree of Gos28 mislocalization (Figure 4—figure supplement 2B), suggesting that ATAD1Δα11 incorporated into WT hexamers and inactivated them, or that the ATAD1Δα11 subunits prevented the WT enzyme from forming stable hexamers. These data combined showed that α11 plays an important role in both subunit oligomerization and hexamer function.

Figure 4 with 5 supplements see all

Download asset Open asset

To further establish α11’s impact on ATAD1 oligomerization, we examined the elution profile of recombinantly expressed ATAD1 with and without the C-terminal helix. To this end, we purified the cytosolic domain of ATAD1 lacking the α11 helix (Δ40-ATAD1Δα11). On size exclusion, Δ40-ATAD1Δα11 eluted exclusively as a monomer. At similar concentrations, Δ40-ATAD1 formed oligomers (Figure 4—figure supplement 3), indicating that α11 mediates oligomerization in vitro. To evaluate the functional impact of a defect in hexamer formation, we next asked if the ability to bind a peptide substrate was affected for Δ40-ATAD1Δα11. To this end, we sought to identify a peptide that binds ATAD1. First, we designed a peptide array by stepping through ATAD1’s known substrates and measured the binding of individual peptides to ATAD1. We picked a few peptides that showed the strongest binding and tested their affinity to ATAD1 in solution using fluorescence polarization. Out of this group, a peptide (P13) that is derived from the sequence of the C-terminus of Pex26 (an established ATAD1 substrate) emerged as the strongest binder, which we then used to test the effect of truncating α11 on ATAD1’s substrate-binding affinity. As expected and consistent with its inability to form hexamers, Δ40-ATAD1Δα11 also showed substantially reduced peptide-binding affinity (Figure 4C).

Several disease-related mutations are close to the α11 region at the C-terminus (see Introduction). We tested these ATAD1 variants using our imaging assay to assess if their ability to remove mislocalized TA proteins would be impacted. Out of the variants tested, the E267stop variant impacted ATAD1’s function significantly, whereas the other variants (H357Rfs*15, D221H, R9H, and E290K) displayed WT-like phenotypes (Figure 4—figure supplement 4). These results indicate that the mechanism that underlies neurological disorders may be separate from the extraction of Gos28.

As previously mentioned, α11 packs against pore-loop 3 from the adjacent subunit and constitutes the additional layer of the interconnected network involving three pore-loops (Figure 2—figure supplement 1). The fact that it lies at the oligomerization interface and also contacts pore-loop 3 prompted us to ask whether α11 could couple oligomerization with substrate binding. Given that oligomerization usually enhances the ATPase activity of AAA proteins, we sought to measure the ATPase activity of Δ40-ATAD1 and Δ40-ATAD1Δα11 and asked whether the activities can be stimulated by the addition of substrate. As shown in Figure 4—figure supplement 5, Δ40-ATAD1Δα11 displays a much lower ATPase activity compared to Δ40-ATAD1, indicating the lack of functional hexamers. With the addition of a peptide substrate (P13), the activity of Δ40-ATAD1 is stimulated by around 41%, whereas no significant stimulation was observed for Δ40-ATAD1Δα11 (Figure 4—figure supplement 5). This results suggest that Δ40-ATAD1Δα11’s ability to oligomerize is significantly impacted and the addition of a substrate is not able to bring the monomeric subunits into a functional oligomeric form. Having examined the oligomer interface in detail, we next examined the substrate interactions within the central pore.

Multiple aromatic amino acids in the central pore are essential for ATAD1’s function

We next revisited our hypothesis that ATAD1 function, like Msp1 function, necessitates engagement of multiple aromatic amino acids in its central pore. To this end, we tested the effect of pore-loop mutations in the ATAD1^-/- EGFP-Gos28 reporter cell line (Figure 5A) using our cell-based assay. Expression of ATAD1 bearing either the W166A (the first aromatic amino acid in pore-loop 1) or Y167A (the second aromatic amino acid in pore-loop 1) mutation led to significant mislocalization of EGFP-Gos28, suggesting that these mutations inactivate ATAD1. By contrast, expression of the aromatic mutant Y167F cleared Gos28 from mitochondria, indicating functional ATAD1, which is consistent with what we observed for Msp1. Interestingly, changing W166 or Y167 to an aliphatic amino impacted ATAD1’s activity: ATAD1^W166V, ATAD1^Y167V, and ATAD1 ^Y167L were inactive. The activity of ATAD1^W166L was also impacted, albeit to a lesser degree (Figure 5A, Figure 5—figure supplements 1 and 2), supporting the notion that the aromaticity of this position is important. This observation, namely that an aliphatic amino acid did not functionally replace the aromatic amino acid in this position, contrasts with our data from corresponding mutations in Msp1 in the yeast growth assay (Wang et al., 2020), suggesting either a better sensitivity of the microscopy-based assay or an inherit difference between the yeast and the mammalian system. We also tested the effect of mutations on H206 (pore-loop 2) in our assay. While our structural data clearly showed an interaction between H206 and the substrate backbone, mutations in this position did not impact the function of ATAD1 in this context.

Figure 5 with 2 supplements see all

Download asset Open asset

Having established the functional importance of the pore-loop 1 side chains in substrate extraction, we further examined their specific impact on ATAD1’s ability to bind a substrate in vitro. Consistent with the cell-based assay, ATAD1^W166A, ATAD1^W166V, ATAD1^W166L, ATAD1^Y166A, and ATAD1^Y167V showed dramatically reduced (estimated >100-fold) binding to P13 compared to WT ATAD1. Both ATAD1^H206A and ATAD1^Y167F showed WT-like activities in the cell-based assay, but bound P13 with much weaker affinity than the WT enzyme in this assay. We reasoned that this difference arose either because the in vitro assay is more sensitive to small changes in activity that cell-based assays fail to capture, or because a different substrate was used in the cell-based assay (Gos28) than in the in vitro binding assay (P13, derived from Pex26). Taken together, the in vitro peptide-binding results, implications from the structure, and the cell-based assay all converge on the conclusion that the aromaticity of the second amino acid in pore-loop 1 is important for ATAD1’s function.

In this work, we present the cryo-EM structures of the soluble domain of the mitochondrial AAA protein ATAD1 in complex with a peptide substrate. Our structures show that the overall architecture of ATAD1 is very similar to that of its yeast homolog Msp1. ATAD1’s architecture fully agrees with the conserved mechanism of hand-over-hand spiral propagation established for many AAA proteins in substrate translocation. We also discovered a distinct structural element, helix α11, that was present but remained structurally unresolved in Msp1, which differentiates ATAD1/Msp1 from other AAA_MC proteins. Multiple lines of evidence including cell-based mislocalization, substrate binding, and ATPase activity all suggest that α11 is required for the formation of functional oligomers. Although we do not know why it is structured differently than the other AAA_MC proteins (which adopt an α11–α12 organization), one possibility is that α11 is useful in adaptor interaction. In particular, yeast Msp1 docks to the proteosome to channel its substrate directly for degradation (Basch et al., 2020). Another study showed that during mitochondrial import stress, Msp1 is recruited to the TOM complex by the adaptor Cis1 to remove precursor proteins stuck in the TOM translocase (Weidberg and Amon, 2018). We reason that structural elements facing the cytosolic side of Msp1/ATAD1 (such as α11) may be evolutionarily adapted to interact with these factors. In this view, the α11–α12 organizational differences in different AAA_MC protein subfamilies may have evolved to suit their specialized functions, such as binding to microtubules and ESCRT proteins for katanin and Vps4, respectively. This view is supported by the fact that ATAD1/Msp1 occupy the basal position in the evolutionary tree of this clade, and the other members (katanin, spastin, and Vps4) share a common evolutionary ancestor (Frickey and Lupas, 2004).

The cell-based imaging assay allowed us to directly quantify ATAD1-dependent substrate mislocalization and thereby revisit a hypothesis raised in the previous study (Wang et al., 2020) on the functional importance of the pore-loop amino acids. The notion that aromaticity of pore-loop/substrate-interacting side chains is important is supported by single molecule force measurements of other AAA protein translocases. A study on the bacterial AAA proteins ClpXP (Rodriguez-Aliaga et al., 2016), for example, showed that the flat, bulkier pore-loop amino acids give the protein a stronger grip, that is higher coupling efficiency between ATP hydrolysis and substrate unfolding, yet are compromised in the rate of substrate translocation (measured as protein unfolding with ClpXP). Given ATAD1’s function is to extract hydrophobic membrane proteins from the lipid membrane, having a tighter grip on the substrate may be important, even if it comes at the cost of a slower translocation speed. Similar to ATAD1, Cdc48, a protein that removes misfolded membrane proteins from the ER membrane, also possesses two aromatic amino acids in its pore-loop 1 (Cooney et al., 2019; Twomey et al., 2019), suggesting that the additional aromatic amino acids in the central pore evolved to aid in the removal of membrane proteins.

To conclude, while ATAD1/Msp1 utilize a conceptually similar mechanism for substrate translocation as many other AAA proteins, the high conservation of several unique features between ATAD1 and Msp1 suggests that evolution fine-tuned these enzymes early in eukaryotic cell evolution for their special role in membrane protein extraction and protein quality control. The phylogenetic comparison of structure and function, and the complementary experimental opportunities afforded by structural and cell-level analysis, allowed us to extract hints regarding important functional principles and assess their generality over a wide span of evolutionary time. The neurological phenotypes associated with a number of ATAD1 mutations still remain poorly understood, yet they serve to underscore the importance of proteostasis in human physiology and raise hope that understanding these principles may help to develop new treatments for combating devastating pathological dysfunctions arising from proteostasis imbalances.

All data generated or analyzed during this study are included in the manuscript and supporting files. The imaging analysis pipeline is described in Figure 3-source data 1. The raw images used for data analysis are deposited into dryad (https://doi.org/10.5061/dryad.q2bvq83n3). The final models are deposited into the protein data bank (PDB) under the accession codes 7UPR (closed conformation) and 7UPT (open conformation). The final reconstructed maps are deposited into the EMDB under the accession codes EMD-26674 (closed conformation) and EMD-26675 (open conformation).

1. 1. Wang L
   2. Toutkoushian H
   3. Belyy V
   4. Kokontis CY
   5. Walter P

   (2022) Dryad Digital Repository

   Conserved structural elements specialize ATAD1 as a membrane protein extraction machine.

   https://doi.org/10.5061/dryad.q2bvq83n3
2. 1. Wang L
   2. Toutkoushian H
   3. Belyy V
   4. Kokontis CY
   5. Walter P

   (2022) RCSB Protein Data Bank

   ID 7UPR. Human mitochondrial AAA protein ATAD1 (with a catalytic dead mutation) in complex with a peptide substrate (closed conformation).

   https://www.rcsb.org/structure/7UPR
3. 1. Wang L
   2. Toutkoushian H
   3. Belyy V
   4. Kokontis CY
   5. Walter P

   (2022) RCSB Protein Data Bank

   ID 7UPT. Human mitochondrial AAA protein ATAD1 (with a catalytic dead mutation) in complex with a peptide substrate (open conformation).

   https://www.rcsb.org/structure/7UPT
4. 1. Wang L
   2. Toutkoushian H
   3. Belyy V
   4. Kokontis CY
   5. Walter P

   (2022) Electron Microscopy Data Bank

   ID EMD-26674. Human mitochondrial AAA protein ATAD1 (with a catalytic dead mutation) in complex with a peptide substrate (closed conformation).

   https://www.ebi.ac.uk/emdb/EMD-26674
5. 1. Wang L
   2. Toutkoushian H
   3. Belyy V
   4. Kokontis CY
   5. Walter P

   (2022) Electron Microscopy Data Bank

   ID EMD-26675. Human mitochondrial AAA protein ATAD1 (with a catalytic dead mutation) in complex with a peptide substrate (open conformation).

   https://www.ebi.ac.uk/emdb/EMD-26675

1. 1. Adams PD
   2. Afonine PV
   3. Bunkóczi G
   4. Chen VB
   5. Davis IW
   6. Echols N
   7. Headd JJ
   8. Hung LW
   9. Kapral GJ
   10. Grosse-Kunstleve RW
   11. McCoy AJ
   12. Moriarty NW
   13. Oeffner R
   14. Read RJ
   15. Richardson DC
   16. Richardson JS
   17. Terwilliger TC
   18. Zwart PH

   (2010) PHENIX: a comprehensive Python-based system for macromolecular structure solution

   Acta Crystallographica. Section D, Biological Crystallography 66:213–221.

   https://doi.org/10.1107/S0907444909052925

   • PubMed
   • Google Scholar
2. 1. Ahrens-Nicklas RC
   2. Umanah GKE
   3. Sondheimer N
   4. Deardorff MA
   5. Wilkens AB
   6. Conlin LK
   7. Santani AB
   8. Nesbitt A
   9. Juulsola J
   10. Ma E
   11. Dawson TM
   12. Dawson VL
   13. Marsh ED

   (2017) Precision therapy for a new disorder of AMPA receptor recycling due to mutations in ATAD1

   Neurology. Genetics 3:e130.

   https://doi.org/10.1212/NXG.0000000000000130

   • PubMed
   • Google Scholar
3. 1. Basch M
   2. Wagner M
   3. Rolland S
   4. Carbonell A
   5. Zeng R
   6. Khosravi S
   7. Schmidt A
   8. Aftab W
   9. Imhof A
   10. Wagener J
   11. Conradt B
   12. Wagener N

   (2020) Msp1 cooperates with the proteasome for extraction of arrested mitochondrial import intermediates

   Molecular Biology of the Cell 31:753–767.

   https://doi.org/10.1091/mbc.E19-06-0329

   • PubMed
   • Google Scholar
4. 1. Chen Y-C
   2. Umanah GKE
   3. Dephoure N
   4. Andrabi SA
   5. Gygi SP
   6. Dawson TM
   7. Dawson VL
   8. Rutter J

   (2014) Msp1/ATAD1 maintains mitochondrial function by facilitating the degradation of mislocalized tail-anchored proteins

   The EMBO Journal 33:1548–1564.

   https://doi.org/10.15252/embj.201487943

   • PubMed
   • Google Scholar
5. 1. Chio US
   2. Cho H
   3. Shan SO

   (2017) Mechanisms of Tail-Anchored Membrane Protein Targeting and Insertion

   Annual Review of Cell and Developmental Biology 33:417–438.

   https://doi.org/10.1146/annurev-cellbio-100616-060839

   • PubMed
   • Google Scholar
6. 1. Cooney I
   2. Han H
   3. Stewart MG
   4. Carson RH
   5. Hansen DT
   6. Iwasa JH
   7. Price JC
   8. Hill CP
   9. Shen PS

   (2019) Structure of the Cdc48 segregase in the act of unfolding an authentic substrate

   Science 365:502–505.

   https://doi.org/10.1126/science.aax0486

   • PubMed
   • Google Scholar
7. 1. de la Peña AH
   2. Goodall EA
   3. Gates SN
   4. Lander GC
   5. Martin A

   (2018) Substrate-engaged 26S proteasome structures reveal mechanisms for ATP-hydrolysis-driven translocation

   Science 362:eaav0725.

   https://doi.org/10.1126/science.aav0725

   • PubMed
   • Google Scholar
8. 1. Dederer V
   2. Khmelinskii A
   3. Huhn AG
   4. Okreglak V
   5. Knop M
   6. Lemberg MK

   (2019) Cooperation of mitochondrial and ER factors in quality control of tail-anchored proteins

   eLife 8:e45506.

   https://doi.org/10.7554/eLife.45506

   • PubMed
   • Google Scholar
9. 1. Doan KN
   2. Grevel A
   3. Mårtensson CU
   4. Ellenrieder L
   5. Thornton N
   6. Wenz LS
   7. Opaliński Ł
   8. Guiard B
   9. Pfanner N
   10. Becker T

   (2020) The Mitochondrial Import Complex MIM Functions as Main Translocase for α-Helical Outer Membrane Proteins

   Cell Reports 31:107567.

   https://doi.org/10.1016/j.celrep.2020.107567

   • PubMed
   • Google Scholar
10. 1. Edelstein A
    2. Amodaj N
    3. Hoover K
    4. Vale R
    5. Stuurman N

    (2010) Computer control of microscopes using µManager

    Current Protocols in Molecular Biology Chapter 14:Unit14.

    https://doi.org/10.1002/0471142727.mb1420s92

    • PubMed
    • Google Scholar
11. 1. Emsley P
    2. Lohkamp B
    3. Scott WG
    4. Cowtan K

    (2010) Features and development of Coot

    Acta Crystallographica. Section D, Biological Crystallography 66:486–501.

    https://doi.org/10.1107/S0907444910007493

    • PubMed
    • Google Scholar
12. 1. Favaloro V
    2. Spasic M
    3. Schwappach B
    4. Dobberstein B

    (2008) Distinct targeting pathways for the membrane insertion of tail-anchored (TA) proteins

    Journal of Cell Science 121:1832–1840.

    https://doi.org/10.1242/jcs.020321

    • PubMed
    • Google Scholar
13. 1. Fresenius HL
    2. Wohlever ML

    (2019) Sorting out how Msp1 maintains mitochondrial membrane proteostasis

    Mitochondrion 49:128–134.

    https://doi.org/10.1016/j.mito.2019.07.011

    • PubMed
    • Google Scholar
14. 1. Frickey T
    2. Lupas AN

    (2004) Phylogenetic analysis of AAA proteins

    Journal of Structural Biology 146:2–10.

    https://doi.org/10.1016/j.jsb.2003.11.020

    • PubMed
    • Google Scholar
15. 1. Gates SN
    2. Yokom AL
    3. Lin J
    4. Jackrel ME
    5. Rizo AN
    6. Kendsersky NM
    7. Buell CE
    8. Sweeny EA
    9. Mack KL
    10. Chuang E
    11. Torrente MP
    12. Su M
    13. Shorter J
    14. Southworth DR

    (2017) Ratchet-like polypeptide translocation mechanism of the AAA+ disaggregase Hsp104

    Science 357:273–279.

    https://doi.org/10.1126/science.aan1052

    • PubMed
    • Google Scholar
16. 1. Grimm JB
    2. Brown TA
    3. English BP
    4. Lionnet T
    5. Lavis LD

    (2017) Super-Resolution Microscopy, Methods and Protocols

    Methods Mol Biology 1663:179–188.

    https://doi.org/10.1007/978-1-4939-7265-4

    • PubMed
    • Google Scholar
17. 1. Han H
    2. Monroe N
    3. Sundquist WI
    4. Shen PS
    5. Hill CP

    (2017) The AAA ATPase Vps4 binds ESCRT-III substrates through a repeating array of dipeptide-binding pockets

    eLife 6:e31324.

    https://doi.org/10.7554/eLife.31324

    • PubMed
    • Google Scholar
18. 1. Han H
    2. Schubert HL
    3. McCullough J
    4. Monroe N
    5. Purdy MD
    6. Yeager M
    7. Sundquist WI
    8. Hill CP

    (2020) Structure of spastin bound to a glutamate-rich peptide implies a hand-over-hand mechanism of substrate translocation

    The Journal of Biological Chemistry 295:435–443.

    https://doi.org/10.1074/jbc.AC119.009890

    • PubMed
    • Google Scholar
19. 1. Matsumoto N
    2. Tamura S
    3. Fujiki Y

    (2003) The pathogenic peroxin Pex26p recruits the Pex1p-Pex6p AAA ATPase complexes to peroxisomes

    Nature Cell Biology 5:454–460.

    https://doi.org/10.1038/ncb982

    • PubMed
    • Google Scholar
20. 1. Matsumoto S
    2. Nakatsukasa K
    3. Kakuta C
    4. Tamura Y
    5. Esaki M
    6. Endo T

    (2019) Msp1 Clears Mistargeted Proteins by Facilitating Their Transfer from Mitochondria to the ER

    Molecular Cell 76:191–205.

    https://doi.org/10.1016/j.molcel.2019.07.006

    • PubMed
    • Google Scholar
21. 1. McQuin C
    2. Goodman A
    3. Chernyshev V
    4. Kamentsky L
    5. Cimini BA
    6. Karhohs KW
    7. Doan M
    8. Ding L
    9. Rafelski SM
    10. Thirstrup D
    11. Wiegraebe W
    12. Singh S
    13. Becker T
    14. Caicedo JC
    15. Carpenter AE

    (2018) CellProfiler 3.0: Next-generation image processing for biology

    PLOS Biology 16:e2005970.

    https://doi.org/10.1371/journal.pbio.2005970

    • Google Scholar
22. 1. Monroe N
    2. Han H
    3. Shen PS
    4. Sundquist WI
    5. Hill CP

    (2017) Structural basis of protein translocation by the Vps4-Vta1 AAA ATPase

    eLife 6:e24487.

    https://doi.org/10.7554/eLife.24487

    • PubMed
    • Google Scholar
23. 1. Nuebel E
    2. Morgan JT
    3. Fogarty S
    4. Winter JM
    5. Lettlova S
    6. Berg JA
    7. Chen Y-C
    8. Kidwell CU
    9. Maschek JA
    10. Clowers KJ
    11. Argyriou C
    12. Chen L
    13. Wittig I
    14. Cox JE
    15. Roh-Johnson M
    16. Braverman N
    17. Bonkowsky J
    18. Gygi SP
    19. Rutter J

    (2021) The biochemical basis of mitochondrial dysfunction in Zellweger Spectrum Disorder

    EMBO Reports 22:e51991.

    https://doi.org/10.15252/embr.202051991

    • PubMed
    • Google Scholar
24. 1. Okreglak V
    2. Walter P

    (2014) The conserved AAA-ATPase Msp1 confers organelle specificity to tail-anchored proteins

    PNAS 111:8019–8024.

    https://doi.org/10.1073/pnas.1405755111

    • PubMed
    • Google Scholar
25. 1. Pettersen EF
    2. Goddard TD
    3. Huang CC
    4. Couch GS
    5. Greenblatt DM
    6. Meng EC
    7. Ferrin TE

    (2004) UCSF Chimera--A visualization system for exploratory research and analysis

    Journal of Computational Chemistry 25:1605–1612.

    https://doi.org/10.1002/jcc.20084

    • PubMed
    • Google Scholar
26. 1. Pfanner N
    2. Warscheid B
    3. Wiedemann N

    (2019) Mitochondrial proteins: from biogenesis to functional networks

    Nature Reviews. Molecular Cell Biology 20:267–284.

    https://doi.org/10.1038/s41580-018-0092-0

    • PubMed
    • Google Scholar
27. 1. Piard J
    2. Umanah GKE
    3. Harms FL
    4. Abalde-Atristain L
    5. Amram D
    6. Chang M
    7. Chen R
    8. Alawi M
    9. Salpietro V
    10. Rees MI
    11. Chung S-K
    12. Houlden H
    13. Verloes A
    14. Dawson TM
    15. Dawson VL
    16. Van Maldergem L
    17. Kutsche K

    (2018) A homozygous ATAD1 mutation impairs postsynaptic AMPA receptor trafficking and causes A lethal encephalopathy

    Brain 141:651–661.

    https://doi.org/10.1093/brain/awx377

    • PubMed
    • Google Scholar
28. 1. Puchades C
    2. Rampello AJ
    3. Shin M
    4. Giuliano CJ
    5. Wiseman RL
    6. Glynn SE
    7. Lander GC

    (2017) Structure of the mitochondrial inner membrane AAA+ protease YME1 gives insight into substrate processing

    Science 358:eaao0464.

    https://doi.org/10.1126/science.aao0464

    • PubMed
    • Google Scholar
29. 1. Puchades C
    2. Sandate CR
    3. Lander GC

    (2020) The molecular principles governing the activity and functional diversity of AAA+ proteins

    Nature Reviews. Molecular Cell Biology 21:43–58.

    https://doi.org/10.1038/s41580-019-0183-6

    • PubMed
    • Google Scholar
30. 1. Punjani A
    2. Rubinstein JL
    3. Fleet DJ
    4. Brubaker MA

    (2017) cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination

    Nature Methods 14:290–296.

    https://doi.org/10.1038/nmeth.4169

    • PubMed
    • Google Scholar
31. 1. Rodriguez-Aliaga P
    2. Ramirez L
    3. Kim F
    4. Bustamante C
    5. Martin A

    (2016) Substrate-translocating loops regulate mechanochemical coupling and power production in AAA+ protease ClpXP

    Nature Structural & Molecular Biology 23:974–981.

    https://doi.org/10.1038/nsmb.3298

    • PubMed
    • Google Scholar
32. 1. Sandate CR
    2. Szyk A
    3. Zehr EA
    4. Lander GC
    5. Roll-Mecak A

    (2019) An allosteric network in spastin couples multiple activities required for microtubule severing

    Nature Structural & Molecular Biology 26:671–678.

    https://doi.org/10.1038/s41594-019-0257-3

    • PubMed
    • Google Scholar
33. 1. Scheres SHW

    (2012) RELION: implementation of a Bayesian approach to cryo-EM structure determination

    Journal of Structural Biology 180:519–530.

    https://doi.org/10.1016/j.jsb.2012.09.006

    • PubMed
    • Google Scholar
34. 1. Schuldiner M
    2. Metz J
    3. Schmid V
    4. Denic V
    5. Rakwalska M
    6. Schmitt HD
    7. Schwappach B
    8. Weissman JS

    (2008) The GET complex mediates insertion of tail-anchored proteins into the ER membrane

    Cell 134:634–645.

    https://doi.org/10.1016/j.cell.2008.06.025

    • PubMed
    • Google Scholar
35. 1. Shin SC
    2. Im SK
    3. Jang EH
    4. Jin KS
    5. Hur EM
    6. Kim EE

    (2019) Structural and Molecular Basis for Katanin-Mediated Severing of Glutamylated Microtubules

    Cell Reports 26:1357–1367.

    https://doi.org/10.1016/j.celrep.2019.01.020

    • PubMed
    • Google Scholar
36. 1. Stefanovic S
    2. Hegde RS

    (2007) Identification of a targeting factor for posttranslational membrane protein insertion into the ER

    Cell 128:1147–1159.

    https://doi.org/10.1016/j.cell.2007.01.036

    • PubMed
    • Google Scholar
37. 1. Su M
    2. Guo EZ
    3. Ding X
    4. Li Y
    5. Tarrasch JT
    6. Brooks CL
    7. Xu Z
    8. Skiniotis G

    (2017) Mechanism of Vps4 hexamer function revealed by cryo-EM

    Science Advances 3:e1700325.

    https://doi.org/10.1126/sciadv.1700325

    • PubMed
    • Google Scholar
38. 1. Sun S
    2. Li L
    3. Yang F
    4. Wang X
    5. Fan F
    6. Yang M
    7. Chen C
    8. Li X
    9. Wang HW
    10. Sui SF

    (2017) Cryo-EM structures of the ATP-bound Vps4^E233Q hexamer and its complex with Vta1 at near-atomic resolution

    Nature Communications 8:16064.

    https://doi.org/10.1038/ncomms16064

    • PubMed
    • Google Scholar
39. 1. Twomey EC
    2. Ji Z
    3. Wales TE
    4. Bodnar NO
    5. Ficarro SB
    6. Marto JA
    7. Engen JR
    8. Rapoport TA

    (2019) Substrate processing by the Cdc48 ATPase complex is initiated by ubiquitin unfolding

    Science 365:eaax1033.

    https://doi.org/10.1126/science.aax1033

    • PubMed
    • Google Scholar
40. 1. Umanah GKE
    2. Pignatelli M
    3. Yin X
    4. Chen R
    5. Crawford J
    6. Neifert S
    7. Scarffe L
    8. Behensky AA
    9. Guiberson N
    10. Chang M
    11. Ma E
    12. Kim JW
    13. Castro CC
    14. Mao X
    15. Chen L
    16. Andrabi SA
    17. Pletnikov MV
    18. Pulver AE
    19. Avramopoulos D
    20. Bonci A
    21. Valle D
    22. Dawson TM
    23. Dawson VL

    (2017) Thorase variants are associated with defects in glutamatergic neurotransmission that can be rescued by Perampanel

    Science Translational Medicine 9:eaah4985.

    https://doi.org/10.1126/scitranslmed.aah4985

    • PubMed
    • Google Scholar
41. 1. Vajjhala PR
    2. Nguyen CH
    3. Landsberg MJ
    4. Kistler C
    5. Gan A-L
    6. King GF
    7. Hankamer B
    8. Munn AL

    (2008) The Vps4 C-terminal helix is a critical determinant for assembly and ATPase activity and has elements conserved in other members of the meiotic clade of AAA ATPases

    The FEBS Journal 275:1427–1449.

    https://doi.org/10.1111/j.1742-4658.2008.06300.x

    • PubMed
    • Google Scholar
42. 1. Wang L
    2. Myasnikov A
    3. Pan X
    4. Walter P

    (2020) Structure of the AAA protein Msp1 reveals mechanism of mislocalized membrane protein extraction

    eLife 9:e54031.

    https://doi.org/10.7554/eLife.54031

    • PubMed
    • Google Scholar
43. 1. Wang L
    2. Walter P

    (2020) Msp1/ATAD1 in Protein Quality Control and Regulation of Synaptic Activities

    Annual Review of Cell and Developmental Biology 36:141–164.

    https://doi.org/10.1146/annurev-cellbio-031220-015840

    • PubMed
    • Google Scholar
44. 1. Waterhouse A
    2. Bertoni M
    3. Bienert S
    4. Studer G
    5. Tauriello G
    6. Gumienny R
    7. Heer FT
    8. de Beer TAP
    9. Rempfer C
    10. Bordoli L
    11. Lepore R
    12. Schwede T

    (2018) SWISS-MODEL: homology modelling of protein structures and complexes

    Nucleic Acids Research 46:W296–W303.

    https://doi.org/10.1093/nar/gky427

    • PubMed
    • Google Scholar
45. 1. Weidberg H
    2. Amon A

    (2018) MitoCPR-A surveillance pathway that protects mitochondria in response to protein import stress

    Science (New York, N.Y.) 360:eaan4146.

    https://doi.org/10.1126/science.aan4146

    • PubMed
    • Google Scholar
46. Preprint

    1. Winter JM
    2. Fresenius HL
    3. Keys HR
    4. Cunningham CN
    5. Ryan J
    6. Sirohi D
    7. Berg JA
    8. Agarwal P
    9. Letai N
    10. Sabatini A
    11. Wohlever DM

    (2021) Co-Deletion of ATAD1 with PTEN Primes Cells for BIM-Mediated Apoptosis

    bioRxiv.

    https://doi.org/10.1101/2021.07.01.450781

    • Google Scholar
47. 1. Wohlever ML
    2. Mateja A
    3. McGilvray PT
    4. Day KJ
    5. Keenan RJ

    (2017) Msp1 Is a Membrane Protein Dislocase for Tail-Anchored Proteins

    Molecular Cell 67:194–202.

    https://doi.org/10.1016/j.molcel.2017.06.019

    • PubMed
    • Google Scholar
48. 1. Zehr E
    2. Szyk A
    3. Piszczek G
    4. Szczesna E
    5. Zuo X
    6. Roll-Mecak A

    (2017) Katanin spiral and ring structures shed light on power stroke for microtubule severing

    Nature Structural & Molecular Biology 24:717–725.

    https://doi.org/10.1038/nsmb.3448

    • PubMed
    • Google Scholar
49. 1. Zhang J
    2. Wang Y
    3. Chi Z
    4. Keuss MJ
    5. Pai Y-ME
    6. Kang HC
    7. Shin J-H
    8. Bugayenko A
    9. Wang H
    10. Xiong Y
    11. Pletnikov MV
    12. Mattson MP
    13. Dawson TM
    14. Dawson VL

    (2011) The AAA+ ATPase Thorase regulates AMPA receptor-dependent synaptic plasticity and behavior

    Cell 145:284–299.

    https://doi.org/10.1016/j.cell.2011.03.016

    • PubMed
    • Google Scholar
50. 1. Zhang K

    (2016) Gctf: Real-time CTF determination and correction

    Journal of Structural Biology 193:1–12.

    https://doi.org/10.1016/j.jsb.2015.11.003

    • PubMed
    • Google Scholar

Author details

1. Lan Wang

   1. Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco, United States
   2. Howard Hughes Medical Institute, University of California at San Francisco, San Francisco, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review and editing

   Contributed equally with

   Hannah Toutkoushian

   For correspondence

   lan@walterlab.ucsf.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-8931-7201

2. Hannah Toutkoushian

   1. Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco, United States
   2. Howard Hughes Medical Institute, University of California at San Francisco, San Francisco, United States

   Present address

   Altos Labs, San Francisco, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review and editing

   Contributed equally with

   Lan Wang

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7461-2005

3. Vladislav Belyy

   1. Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco, United States
   2. Howard Hughes Medical Institute, University of California at San Francisco, San Francisco, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – review and editing

   Competing interests

   No competing interests declared

4. Claire Y Kokontis

   Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco, United States

   Contribution

   Data curation, Formal analysis, Methodology, Visualization, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9397-651X

5. Peter Walter

   1. Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco, United States
   2. Howard Hughes Medical Institute, University of California at San Francisco, San Francisco, United States

   Present address

   Altos Labs, San Francisco, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Supervision, Validation, Visualization, Writing – original draft, Writing – review and editing

   For correspondence

   pwalter@altoslabs.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6849-708X

• 2,509

  views

• 468

  downloads

• 8

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.