Human TOM and TIM22 complexes join forces to import mitochondrial carriers

For the first time, scientists have captured the human machinery that imports metabolite carriers into mitochondria in the act of transferring cargo from the outer membrane to the inner one. A team led by Long Li at Peking University reports in Nature Structural & Molecular Biology that the TOM and TIM22 translocases, long thought to hand carrier proteins across the mitochondrial intermembrane space as independent machines, instead assemble into a single TOM–TIM22 supercomplex that couples translocation across both membranes in one continuous operation. The findings, published on 23 July 2026, are based on cryo-electron microscopy structures and biochemical mapping experiments that trace the substrate pathway at molecular resolution.

Mitochondria are double-membraned organelles that descended from free-living bacteria and still retain their own small genome, yet the overwhelming majority of the roughly thousand different proteins they need to function are encoded in the nucleus and synthesized on cytosolic ribosomes. Getting those proteins across two membranes, each with a distinct lipid composition and protein complement, is one of the defining logistical problems of eukaryotic cell biology. Mitochondria have solved it with a set of dedicated translocases, each specialized for a particular class of cargo, and the accuracy of the system matters enormously: mistargeted or unimported proteins can aggregate, clog membranes, or fail to deliver the metabolite transport that tissues such as heart, brain, and muscle depend on.

Among the most critical cargoes are the metabolite carriers of the SLC25 family, the largest solute transporter family in humans, which shuttle ADP and ATP, phosphate, fatty acids, and other essential metabolites across the otherwise impermeable inner membrane. Without these carriers, the organelle cannot exchange the chemical energy currency it produces with the rest of the cell, and cellular respiration grinds to a halt. These carriers are synthesized without cleavable presequences; instead, their six hydrophobic transmembrane segments must traverse the outer membrane through the TOM complex, cross the aqueous intermembrane space with the help of small Tim chaperones, and insert into the inner membrane through the TIM22 complex. In budding yeast, decades of work suggested that TOM and TIM22 act as separate stations along this pathway, connected only transiently by soluble Tim9–Tim10 chaperone complexes that ferry hydrophobic transmembrane segments through the space between the membranes. Yeast genetics and biochemistry had established the individual components of this route in considerable detail, which made the human system a natural point of comparison — and, as it turns out, a source of surprises.

The new study shows that human mitochondria operate differently. By expressing carrier substrates engineered for stable engagement of the import machinery, the researchers found that human TOM and TIM22 form a stable supercomplex detectable by blue native PAGE and pulldown assays from HEK293 cells. The supercomplex could be assembled in vitro using carrier substrates loaded onto Tim9–Tim10 chaperones, and its formation depended on the presence of an engaged substrate rather than occurring spontaneously in its absence, indicating that the carrier itself nucleates the junction between the two translocases. That substrate-dependence is functionally meaningful: it suggests the cell does not maintain a wasteful permanent bridge between membranes, but instead builds one on demand, precisely when a hydrophobic cargo molecule needs protection.

To visualize the transfer process, the team used cryo-electron microscopy on purified supercomplexes. They obtained maps of the substrate-engaged TOM complex and of the full TOM–TIM22 assembly, with structures deposited in the Protein Data Bank under accession codes 9WV1, 9WV2 and 9WV3 and maps in the EM Data Bank under EMD-66278 through EMD-66281. Cryo-EM has become the method of choice for interrogating membrane-embedded molecular machines of this size and fragility, because it images individual particles frozen in vitreous ice rather than requiring crystallization, and it has repeatedly revealed transient assemblies — like this one — that would never survive the rigors of crystallography. The structures reveal the carrier substrate threaded through the Tom40 channel with its transmembrane segments unpaired — unfolded and stretched out in single-helix form rather than packed as they are in the folded carrier — following a hydrophobic path along the inside of the channel wall. Notably, the substrate does not exit the bottom of the Tom40 pore as might have been expected. Instead, the transmembrane segments leave through an unexpected lateral groove on the side of the channel, emerging directly into the intermembrane space outside the main channel axis.

The researchers used photocrosslinking to map precisely where the substrate sits along its journey. By placing photoreactive probes at defined positions in the carrier transmembrane segments and the surrounding machinery, they showed that the carrier helices contact conserved hydrophobic residues lining the Tom40 channel, then transfer to the membrane-bound small Tim subunits, which in the human system include components that remain tethered to the inner membrane. These membrane-anchored small Tim proteins provide the entry site for the TIM22 complex, positioning the incoming hydrophobic segments for insertion. On the inner membrane side, a groove exposed on the membrane surface of TIM22 serves as the exit route through which carrier transmembrane segments leave the translocase and partition into the lipid bilayer. Together, the pathway forms a continuous, shielded route that keeps the aggregation-prone hydrophobic segments in a proteinaceous environment from the moment they leave the cytosol until they are embedded in the inner membrane.

That shielding principle is worth emphasizing, because it recurs throughout mitochondrial biogenesis. Hydrophobic transmembrane helices exposed to the aqueous intermembrane space would be prone to aggregation and misfolding, and the cell deploys chaperones at every vulnerable juncture — cytosolic factors on the way to the organelle, small Tim complexes in the space between membranes, and the translocase channels themselves as conduits. The supercomplex architecture takes this logic to its conclusion by eliminating any interval in which the cargo is not held by some component of the import machinery.

A central question was how the two membrane-embedded translocases physically connect across the intermembrane space. The structures show that Tim29, a subunit of the human TIM22 complex known to be required for carrier transport, participates in mediating supercomplex formation, along with Tim10b and the AGK subunit — the acylglycerol kinase that is mutated in Sengers syndrome and was previously identified as a TIM22 component. The authors tested the functional relevance of these connections by generating knockout and complemented cell lines; disruption of the relevant regions in AGK impaired carrier import, while the presequence-dependent import pathway, tested with the matrix protein SOD2, was unaffected, indicating that the TOM–TIM22 coupling is specific to the carrier route rather than a general collapse of mitochondrial import. This kind of pathway specificity is a hallmark of well-regulated cellular infrastructure: one branch of the import system can be compromised while parallel routes continue to operate, at least initially, a feature that helps explain the tissue-specific patterns of disease seen when individual components are lost.

To trap the supercomplex in a state suitable for imaging, the team also designed a fusion substrate in which Tim22 itself was linked to transmembrane helices 5 and 6 of the carrier GGC1 and a superfolder GFP tag, allowing the assembled complex to be purified through tandem affinity tags and examined by single-particle cryo-EM. This design let the researchers model how the carrier segments occupy the TOM channel while the receiving TIM22 complex stands ready at the inner membrane, and the two sfGFP densities at the TOM entrance provided a consistent handle for alignment during data processing. The strategy reflects a common and often decisive tactic in modern structural biology: when a transient assembly is too short-lived to purify intact, investigators engineer a tether that locks the interaction long enough to image it, then interpret the captured state in light of independent biochemical evidence.

The comparison with yeast is striking. In yeast Tom40, the equivalent exit path is blocked, and the extended C-terminal segment of the yeast channel forces substrates to follow a different route, consistent with the notion that in yeast the two complexes function separately. The human structures therefore suggest convergent but distinct solutions to the same problem: keeping hydrophobic carrier segments soluble as they cross the intermembrane space. The authors propose a sequential insertion model in which carrier transmembrane segments move one at a time from the TOM lateral groove onto the small Tim chaperones and then into the TIM22 exit groove, in a handoff that is tightly coordinated because the two translocases are held in a fixed spatial relationship within the supercomplex. A sequential handoff of this kind also provides a built-in quality-control opportunity: a segment that fails to transfer cleanly can be recognized and the process halted before a partially inserted, potentially harmful carrier accumulates in the inner membrane.

The work also intersects with recent findings on other cargo. A related study reported that the kinase PINK1, central to Parkin-mediated mitophagy and Parkinson's disease biology, is imported through human TOM complexes organized in arrays with the channel VDAC. The new carrier-pathway structures show a distinct substrate conformation and a distinct exit route compared with the PINK1-engaged TOM complex, underscoring that a single outer-membrane channel can accommodate multiple translocation modes with different substrate paths through the same pore. Taken together, these results suggest that Tom40 should be viewed not as a uniform pore but as a versatile platform whose behavior depends on the client it engages and the partner complexes it recruits.

The study has limitations inherent to its approach. The structures were obtained with engineered substrates and, in some cases, fusion proteins designed to stabilize the assembly, so the conformations captured may not represent every transient state that occurs during import in living mitochondria. The supercomplex was detected and enriched in the presence of substrate overexpression or crosslinking reagents such as glutaraldehyde, leaving open how dynamic the TOM–TIM22 association is under native conditions and what fraction of the cellular translocase population participates in supercomplexes at any given time. The resolution of the maps varies locally, and the unstructured or mobile regions of the substrate could not be fully modeled, which the authors acknowledge by marking unmodeled densities in their analyses. These caveats are typical of snapshot structural biology, and resolving them will likely require complementary approaches such as live-cell imaging of tagged import components, quantitative measurements of supercomplex abundance in unmanipulated mitochondria, and trapping of intermediate states along the handoff pathway.

Nevertheless, the implications are broad. Mutations in carrier import machinery components, including AGK, are linked to severe human mitochondrial diseases, and defects in SLC25 carrier biogenesis compromise energy metabolism in every tissue. Sengers syndrome, for example, combines congenital cataracts, skeletal myopathy, and cardiomyopathy, a pattern consistent with the energy-intensive tissues most dependent on carrier-mediated metabolite exchange.