Cell’s Insert and Undo Machines Caught Working Together in One ER Supercomplex

Deep inside every eukaryotic cell, the endoplasmic reticulum (ER) operates as the factory floor for membrane proteins, the molecules that make up roughly a quarter of all eukaryotic genes and carry out essential jobs ranging from signal transduction to ion transport. Building these proteins correctly is a formidable challenge: each must be targeted to the right membrane, threaded into the lipid bilayer in the correct orientation, and folded into a functional shape. Mistakes are common because the insertion machinery must remain flexible enough to handle a wide variety of clients. A new study published in Nature Structural & Molecular Biology by Carolin Klose and colleagues has now revealed that two molecular machines with biochemically opposite jobs, one that inserts transmembrane helices into the ER membrane and one that pulls them back out, physically associate into a single, evolutionarily conserved supercomplex, providing the first structural view of how insertion and extraction can be spatially coordinated at the ER membrane.

The first of these machines is the ER membrane protein complex, or EMC, an assembly of eight subunits in yeast and nine in humans that has emerged as a central player in membrane protein biogenesis and cellular homeostasis. Previous cryo-electron microscopy structures revealed a sophisticated three-layered architecture: a lumenal domain built from Emc1, Emc7 and Emc10, a transmembrane core composed of Emc3 through Emc6, and a cytoplasmic cap formed by Emc2, Emc5 and, in humans, either Emc8 or Emc9. The EMC’s best-known function is inserting transmembrane domains, particularly terminal helices and those of moderate hydrophobicity, into the ER membrane. This insertase activity has been largely attributed to Emc3, a member of the evolutionarily ancient YidC/Oxa1 superfamily, which together with Emc4 and Emc6 forms an intramembrane hydrophilic groove that cradles substrate helices before they are released into the bilayer. Beyond insertion, the EMC also acts as a chaperone for membrane proteins and interacts with numerous other biogenesis and quality control factors, positioning it as a hub within the ER membrane proteostasis network.

The second machine is Spf1, known as ATP13A1 in humans, a P5A-type ATPase that performs the opposite task. Because insertases must be somewhat promiscuous to accommodate diverse clients, insertion is inherently error-prone, and cells need surveillance systems to correct the consequences. In mitochondria, for example, the AAA-ATPase Msp1 extracts mistargeted tail-anchored proteins from the outer membrane. In the ER, Spf1 serves an analogous role as a dislocase, pulling out erroneously integrated tail-anchored proteins, signal sequences and transmembrane domains so they can be given another chance to reach their proper destination. Unusually for a P-type ATPase, Spf1 moves its polypeptide cargo in the opposite direction from most family members, extracting substrates from the ER membrane toward the cytosol rather than pumping them into the lumen. This extraction is powered by ATP hydrolysis through the classical Post-Albers cycle, in which the enzyme alternates between two major conformational states, E1 and E2, that expose its substrate-binding cleft to different sides of the membrane.

To discover how these activities might be coordinated, the team began by mapping the EMC’s interaction network in baker’s yeast. They tagged the core subunit Emc5 at its endogenous locus and performed immunoprecipitation coupled to mass spectrometry using a mild isolation strategy with the detergent digitonin, designed to preserve even weak interactions. The screen identified roughly 250 candidate interactors, including previously characterized partners and proteins involved in targeting, lipid metabolism and ER-associated degradation. Among the most strongly enriched was Spf1, with an enrichment level comparable to the EMC subunits themselves. Reciprocal experiments tagging Spf1 confirmed the association from the other direction, and large-scale purification of the endogenous EMC showed that Spf1 co-eluted with the complex at levels matching the EMC subunits after both affinity purification and size-exclusion chromatography. Together, these results demonstrated that EMC and Spf1 form a stable, stoichiometric endogenous complex in the yeast ER membrane.

The structural heart of the study came from cryo-electron microscopy. The researchers produced the complex recombinantly in insect cells and determined its structure to 3.8 Å resolution for the EMC and 4.2 Å for Spf1, verifying that the recombinant assembly faithfully matches a lower-resolution map of the complex purified directly from yeast. They also reconstituted the complex into lipid nanodiscs, where it remained stable, confirming that the interaction survives in a membrane-like environment. In the structure, all EMC subunits adopt an arrangement closely resembling previously reported structures of the isolated complex, while Spf1 sits in its E1 conformation with its intramembrane cleft open to the cytosol. The two machines approach each other at an angle, inducing a pronounced 25-degree bend in the surrounding detergent micelle, a feature also visible in the endogenous reconstruction.

The most striking architectural feature is the arrangement of the two substrate-binding sites. The EMC’s insertase cavity and Spf1’s dislocase cleft are positioned directly opposite each other within the membrane, creating a large composite intramembrane space that resembles a funnel spanning the entire transmembrane region, roughly 25 Å wide at the lumenal side and 50 Å wide at the cytoplasmic side. Critically, the two cavities remain structurally distinct rather than merging into a single catalytic chamber, and they display complementary electrostatic properties: the Spf1 cleft is lined with negative charges, while the EMC’s hydrophilic vestibule carries numerous positive residues. This shared logic of charge-based discrimination, in which both machines use charged residues as selectivity filters for substrate helices, suggests a mechanism for synergistic proofreading on both sides of the ER membrane, allowing freshly inserted or misassembled transmembrane domains to be rapidly sampled and extracted if necessary.

The main interface between the two machines, termed the lumenal dock, forms on the ER lumenal side and involves the lumenal portions of Spf1’s support domain together with a loop in Emc7 and a segment of the Emc10 linker. Both Emc7 and Emc10 are accessory subunits whose functions had been poorly understood, and the structure now assigns them a clear biochemical role as the recruitment platform for Spf1. Mutational dissection in yeast confirmed the interface: deleting Emc7 or Emc10 weakened the EMC-Spf1 interaction to roughly 40 and 48 percent of wild type, respectively, while point mutations in the Emc7 contact loop and charge-reversal substitutions in the Emc10 linker significantly reduced Spf1 binding without disrupting EMC assembly. A complementary mutation in Spf1, E1087A, reduced the interaction to about 30 percent. Importantly, these interface mutants left Spf1’s core dislocase activity intact, as shown by normal rescue of client protein stability assays and cellular phenotypes, indicating that the two machines can assemble into a higher-order complex without compromising their individual functions.

The team then asked whether the assembly is conserved beyond yeast. Tagging the human EMC subunit hEMC5 in HEK293T cells and performing interactomics revealed ATP13A1, the human Spf1 ortholog, as a major interaction partner. An AlphaFold 3 model of the human complex, validated experimentally by cysteine crosslinking and mutagenesis, placed the two machines in a similar orientation with facing cavities, although the precise interfaces differ between species. In human cells, the lumenal dock appears less important, while an extended intramembrane interface involving the transmembrane domains of hEMC7 and hEMC10 dominates; deleting the hEMC7 transmembrane region abolished ATP13A1 binding entirely. These findings establish that the insertase-dislocase supercomplex is conserved from yeast to humans, with species-specific adaptations in how the interaction is stabilized.

Perhaps the most mechanistically revealing results came from trapping the complex in different nucleotide states. In the presence of the ATP analog ATPγS, mimicking the E1-ATP state, the structure revealed a previously poorly resolved feature called the Spf1 arm, an insertion within the P-domain exclusive to P5A-ATPases, extending across the cytoplasmic space to contact Emc3 and the Emc4 gate. In this conformation, the arm acts like a lid, occluding cytoplasmic access to both substrate-binding cavities. In the presence of aluminum fluoride, likely mimicking the E1-P transition state, the arm density disappeared and Spf1 shifted about 10 Å toward the EMC, reducing the cavity entrance by 10 Å. Mutations at the arm tip destabilized the complex and affected client stability, particularly under ER stress conditions, suggesting that ATP-driven conformational cycling of Spf1 remodels the supercomplex and gates access to the shared intramembrane space, potentially preventing freshly extracted helices from being immediately reinserted.

Taken together, the study provides a structural framework for integrated membrane protein quality control at the ER. It shows that functionally distinct proteostasis factors can assemble into higher-order membrane hubs that spatially coordinate opposing activities, that the EMC engages partners through multiple distinct interfaces allowing simultaneous interactions, and that ATP hydrolysis, a hallmark of high-fidelity molecular machines, may be functionally coupled to regulating protein insertion. The work also raises new questions: what drives progression through the Spf1 ATPase cycle, how the different EMC interfaces are regulated in vivo, and how the supercomplex behaves in a native lipid bilayer. As researchers continue to map the dynamic equilibrium between the growing network of insertases and dislocases, this insertase-dislocase supercomplex stands as a vivid example of how cells turn error-prone biogenesis into reliable molecular assembly.

Subject of Research: Structural basis of the ER membrane protein complex and Spf1 insertase-dislocase supercomplex in membrane protein quality control

Article Title: Structural basis of an endoplasmic reticulum EMC:Spf1 insertase–dislocase complex

Article References: Structural basis of an endoplasmic reticulum EMC:Spf1 insertase–dislocase complex. (n.d.). https://doi.org/10.1038/s41594-026-01883-8

Image Credits: AI Generated

DOI: 10.1038/s41594-026-01883-8

Keywords: endoplasmic reticulum, EMC, Spf1, ATP13A1, membrane protein insertion, dislocase, cryo-EM, proteostasis, P-type ATPase, transmembrane domains, quality control, supercomplex