How strong is a mitochondrial targeting signal?

Eukaryotic cells are highly compartmentalized, yet most proteins are synthesized on cytosolic ribosomes and must be targeted to specific organelles. This requires accurate sorting and translocation across cellular membranes. Classical approaches to study protein import rely on isolated organelles or reconstituted membrane systems and monitor the time-dependent uptake of radiolabeled or purified precursor proteins. While these assays have provided mechanistic insight, they offer limited kinetic resolution and are difficult to scale for high-throughput analyses. Fluorescent reporters and split-GFP approaches have been developed as alternatives, but they either introduce bulky modifications or depend on chromophore maturation, which occurs more slowly than the translocation process itself. Therefore, new tools are needed that minimally perturb substrates while enabling quantitative measurements of import kinetics.

The development of a split-luciferase assay has enabled highly sensitive and kinetically resolved measurements of protein import (NanoBiT) (1, 2, 3). In this system, the luciferase NanoLuc is divided into a large fragment (LgBiT) targeted to the compartment of interest and a small 11–amino acid peptide (HiBiT) fused to the precursor protein. Upon successful translocation, HiBiT complements LgBiT to reconstitute an active luciferase, generating luminescence in the presence of furimazine. Because complementation occurs faster than translocation, luminescence directly reports import kinetics.

The NanoBiT-based assay now enables experimental exploration of a fundamental question in mitochondrial biology: do mitochondrial presequences differ in strength? Presequences are N-terminal targeting signals, and although they typically form positively charged amphipathic helices that are proteolytically removed upon import (4), they lack a conserved sequence motif and vary widely in composition, suggesting different import efficiencies. The idea that presequences differ in “strength” has been proposed in several biological models, particularly mitochondrial stress signaling. The transcription factor ATFS-1 was proposed to carry a weak presequence that fails to support efficient mitochondrial import when membrane potential is compromised, allowing nuclear accumulation of the protein and activation of the mitochondrial unfolded protein response (5). In contrast, mitochondrial chaperones and proteases are thought to contain stronger presequences that still support import under stress conditions (6). However, these classifications have largely relied on prediction algorithms that estimate targeting probability rather than import efficiency. Quantitative experimental approaches are therefore required to determine whether presequences indeed differ in strength.

This was addressed by Yan et al. (7), who applied the MitoLuc assay to systematically compare mitochondrial presequences. Individual presequences were fused to a common cargo protein (DHFR) carrying a C-terminal HiBiT tag and incubated with isolated yeast mitochondria containing matrix-localized LgBiT (Fig. 1). Successful matrix translocation triggered luciferase complementation and luminescence. Compared with radiolabeled assays, MitoLuc provided superior kinetic resolution and dynamic range. Import curves were fitted to a generalized logistic function to extract parameters including maximal import signal, maximal velocity, lag time, and duration of active import, defining several metrics of presequence strength.

Applying this approach to seven different Saccharomyces cerevisiae presequences—each predicted by MitoFates to contain a “strong” targeting signal (score >0.99) (8)—revealed unexpectedly diverse import behaviors. Some presequences produced rapid import and saturation of kinetic parameters at higher precursor concentrations, whereas others showed slower import with linear responses across the tested range. Thus, presequence strength reflects differences in import amount, rate, and duration rather than a single parameter. Correlating presequence features with the import metrics revealed that the combination of net charge and hydrophobic moment—an amphiphilicity score—best predicted import efficiency. Importantly, extending the weak Mrpl36 presequence to increase amphiphilicity markedly improved import kinetics, validating this predictive principle.

Yan et al. (7) next revisited the proposed role of presequence strength in mitochondrial stress signaling by comparing the presequences of the Caenorhabditis elegans proteins ATFS-1 and HSP-60, previously suggested to be “weak” and “strong”, respectively. Surprisingly, both presequences showed similar basal import efficiencies in the MitoLuc assay. Differences emerged when membrane potential was perturbed with the ionophore valinomycin: import driven by the ATFS-1 presequence was markedly more sensitive to depolarization than that mediated by HSP-60.

To directly connect import kinetics with membrane potential, the authors developed PotLuc, combining luminescence-based import measurements with simultaneous monitoring of membrane potential using the potentiometric dye DiSC₃. This enabled concurrent quantification of protein import and mitochondrial depolarization. PotLuc experiments confirmed that the ATFS-1 presequence is highly sensitive to decreases in membrane potential, whereas other presequences maintain import even under strongly compromised membrane potential. These results suggest that differential stress sensitivity, rather than large differences in basal import efficiency, may underlie the regulatory role proposed for ATFS-1 in mitochondrial stress responses.

Finally, the authors asked whether differences in presequence strength influence mitochondrial function in vivo. They used a cox4 deletion yeast strain lacking a cytochrome c oxidase subunit and therefore being unable to grow on nonfermentable carbon sources. Respiratory growth can be restored by expressing Cox4 fused to a targeting signal. When different presequences were used to target Cox4, clear differences in growth recovery emerged: strong yeast presequences restored growth similar to the endogenous Cox4 signal, whereas weaker presequences supported delayed recovery. Both C. elegans signals complemented the defect slowly, with HSP-60 supporting recovery earlier than ATFS-1, consistent with their different sensitivities to membrane potential changes and the reduced membrane potential of cox4 mutant mitochondria. Interestingly, a substantial fraction of ATFS-1^preseq-Cox4 accumulated as an unprocessed precursor in mitochondria, potentially preventing proper assembly into respiratory complexes. This observation suggests that parameters beyond targeting efficiency—such as precursor processing—may influence functional rescue. Taken together, the elegant growth assays place the quantitative import differences measured in organello into a physiological context.

Collectively, the work by Yan et al. establishes a quantitative approach to analyze mitochondrial presequences and demonstrates the power of the MitoLuc and PotLuc assays to dissect multiple dimensions of protein import, providing an important conceptual advance and a foundation for future studies. Moving forward, analyses of authentic precursor proteins and matched presequence–mitochondrial systems will be important, and in vivo approaches will be essential to capture the full protein biogenesis pathway, including cytosolic chaperone-assisted precursor delivery and regulatory modifications of the TOM complex under different metabolic or stress conditions (9). Extending these analyses to living cells will allow these regulatory layers to be studied in their physiological context. Encouragingly, the MitoLuc assay has already enabled quantitative measurements of mitochondrial protein import in semi-intact mammalian cells (10), illustrating the potential of this approach to dissect mitochondrial protein biogenesis quantitatively in the cellular context.