How does a selectivity filter in Tim17 mediate protein sorting?

The sorting decision sits at the Tim17 vestibule

Most mitochondrial proteins are made on cytosolic ribosomes and must be routed either into the matrix or into the inner membrane. The TIM23 complex handles both destinations, and Tim17 forms the lateral protein-lipid interface through which precursors slide or insert. What was missing was a mechanism explaining how the complex reads a precursor's topogenic signal and commits it to one fate rather than the other.

The paper's answer is structural and electrostatic. Charged residues on the intermembrane-space-facing surface of Tim17, especially the first two negative charges at positions 4 and 8, act as a selectivity filter. The authors propose that these charges interact with and repel the negatively charged arrest signal that follows the transmembrane domain of inner membrane proteins, slowing translocation and giving the precursor time to exit laterally into the bilayer.

A fluorescence reporter turns arrest into a readout

Testing this idea required an assay that could be run across thousands of yeast mutants, which classical in vitro import experiments with radiolabeled precursors could not easily support. The authors built FLASH, a fluorescent arrest sensor for high-throughput analysis of insertion. It fuses a mitochondrial presequence, mCherry, a quencher domain and a TEV protease cleavage site to a short C-terminal sequence.

The quencher is a hydrophobic helix that tetramerizes in aqueous environments and blocks mCherry maturation. When the protein inserts into the inner membrane, the quencher is buried in the lipid bilayer and fluorescence appears. When the protein is instead fully imported into the matrix, the quencher stays in an aqueous environment and fluorescence is quenched. Two versions were compared: an Arrest reporter ending in the Dld1-derived RETKED arrest sequence, and a Transfer reporter carrying a non-arresting RR sequence. The Arrest reporter gave a strong signal, the Transfer reporter did not, and matrix-targeted TEV protease restored fluorescence in both, confirming that the readout reflects membrane insertion rather than import failure.

Dmo2 emerged from a genome-wide depletion screen

The FLASH reporter was integrated into a yeast strain and crossed into a library of 5110 systematic AID-GFP insertions, in which proteins are fused to an auxin-inducible degron and depleted by adding 5-phenyl-auxin. Cells with reduced fluorescence were picked as candidates. Because many hits reflected depletion of proteins needed for synthesis of the reporter itself, the authors used TEV expression to distinguish general import defects from specific sorting defects.

Dmo2, at the time an uncharacterized mitochondrial protein, came out of this screen alongside Tim23. In a DMO2 deletion mutant, the Arrest reporter showed only low fluorescence while the Transfer reporter was unaffected, and matrix-targeted TEV restored wild-type levels. In vitro import experiments with isolated mitochondria supported the same conclusion: the inner membrane protein cytochrome b2 reached a protease-inaccessible location in wild-type mitochondria, but was strongly compromised in the absence of Dmo2. Mass spectrometry of four biological replicates per condition at 37 degrees Celsius showed a significant reduction of inner membrane proteins in the mutant, consistent with a sorting rather than a general import defect.

Dmo2 builds a disulfide bond that clamps the filter

The screen identified a factor, not a mechanism. Co-immunoprecipitation showed that Dmo2 interacts with Tim17, and the interaction with Tim22 was more transient. The relevant modification turned out to be a disulfide bond in Tim17: when the cysteines at positions 10 and 77 were replaced by alanines, the Arrest reporter showed strongly diminished fluorescence, phenocopying the DMO2 deletion. The mutant Tim17 was present at levels comparable to wild type and still formed a complex with Tim23, so the effect was not explained by loss of protein or loss of complex assembly.

The authors interpret Dmo2 as the oxidase that introduces this bond, and the bond as a structural clamp connecting the N-terminus of Tim17 to its intermembrane-space loop. That clamp keeps the negative charges of the filter in position. The bond mattered particularly at higher temperatures, which the authors read as evidence that the covalent link increases the thermostability of Tim17.

Simulations place the arrest signal in the cavity

Atomistic molecular dynamics simulations of the substrate-bound TIM23 complex were used to ask why the disulfide bond matters for sorting. The simulations modeled the Dld1 substrate and compared systems with and without the disulfide bond, applying a membrane potential of about 150 millivolts across the membrane. In the authors' interpretation, the bond positions the negatively charged N-terminus of Tim17 at the intermembrane-space surface, where those charges slow the precursor by interacting with and repelling its charged arrest signal.

Without the bond, or when the negative charges at residues 4 and 8 are mutated, the simulations indicate that precursors are not properly arrested and sit too deep in the open vestibule of Tim17, which would mis-target them into the matrix. These are models built on a published TIM23 structure and simulation, not direct structural observation of each precursor class, and the authors describe them as simplified working models.

The charge rule fits many precursors but is not yet universal

The paper's mechanistic claim rests on a recognizable sequence feature: inner membrane proteins broadly carry clusters of negatively charged residues after their transmembrane domain, and the Dld1 RETKED sequence is one such arrest signal. That pattern is what makes a charge-based filter plausible as a general sorting element rather than a Dld1-specific quirk.

The authors are explicit that universality remains open. Their simulations focused on the Dld1 model substrate, and whether the charge-based arrest mechanism applies across diverse membrane precursors is stated as a question for future experimental and computational work. The filter is also described as a conserved element of the mitochondrial inner membrane translocase, with the related Tim22 subunit carrying a disulfide bond and charged residues in a comparable position, which the authors raise as a suggestion rather than a demonstrated shared mechanism.

Why the filter matters more than the individual parts

The value of this study is the connection it draws between three otherwise separate observations: a conserved cluster of charges on Tim17, an oxidoreductase of previously unclear function, and the long-standing puzzle of how stop-transfer sequences halt translocation. The disulfide bond is not merely a stability feature; in the authors' model it is what allows the charges to sit in the right place to discriminate between precursors.

That reframes what to look for in sorting defects. A precursor that fails to insert is not necessarily failing to be imported; it may be imported too far. The FLASH design makes exactly this distinction measurable, since loss of membrane insertion produces loss of fluorescence while TEV release restores it. For readers evaluating the paper, the decisive evidence is the convergence of the DMO2 deletion, the Tim17 cysteine mutant and the charge mutations at residues 4 and 8 on the same arrest phenotype, with the simulation supplying a positional explanation rather than independent proof. The paper's own priority for follow-up is the substrate spectrum of Dmo2 and whether the charge-based filter holds across the full range of inner membrane precursors.

Sources used in this answer