Report

Tim50 Maintains the Permeability Barrier of the Mitochondrial Inner Membrane

See allHide authors and affiliations

Science  09 Jun 2006:
Vol. 312, Issue 5779, pp. 1523-1526
DOI: 10.1126/science.1127628

Abstract

Transport of metabolites across the mitochondrial inner membrane is highly selective, thereby maintaining the electrochemical proton gradient that functions as the main driving force for cellular adenosine triphosphate synthesis. Mitochondria import many preproteins via the presequence translocase of the inner membrane. However, the reconstituted Tim23 protein constitutes a pore remaining mainly in its open form, a state that would be deleterious in organello. We found that the intermembrane space domain of Tim50 induced the Tim23 channel to close. Presequences overcame this effect and activated the channel for translocation. Thus, the hydrophilic cis domain of Tim50 maintains the permeability barrier of mitochondria by closing the translocation pore in a presequence-regulated manner.

Most mitochondrial proteins are synthesized as preproteins in the cytosol and must be imported across outer and inner mitochondrial membranes (16). The mitochondrial inner membrane generates and maintains a proton-motive force that is crucial to drive the FoF1-ATP synthase, which is the major machine for cellular adenosine triphosphate (ATP) synthesis (7, 8). How can the permeability barrier of the inner membrane for small ions such as protons be maintained while large hydrophilic channels for the passage of polypeptide chains exist?

The presequence translocase of the inner membrane (TIM23 complex) translocates hundreds of different preproteins into the matrix. The TIM23 complex contains the pore-forming protein Tim23 and three additional membrane proteins, Tim17, Tim21, and Tim50. Tim23 consists of a membrane-embedded domain, containing the large translocation channel, and a domain in the intermembrane space (IMS) that recognizes the N-terminal presequences of preproteins (4, 6, 912). Tim50 and Tim21 each consist of a single transmembrane segment and a large IMS domain, which interact with preproteins and with the translocase of the outer membrane, respectively (1317). Tim17 is largely embedded in the inner membrane and promotes cooperation of the presequence translocase with the associated import motor (PAM complex) (13). PAM is a multisubunit machinery on the matrix side of the inner membrane with the matrix heat shock protein 70 (Ssc1) as the central ATP-consuming subunit. How the presequence channel is regulated to permit translocation of preproteins but prevent leakage of small ions is unknown.

To understand the regulation of the Tim23 channel, we took a combinatorial in vitro and in organello approach. When purified Tim23 was reconstituted into liposomes and subjected to electrophysiological analysis in a planar lipid bilayer system, a cation-preferring channel with the reported characteristics of the presequence channel of the mitochondrial inner membrane was observed; however, the pore remained mainly in an open state (Fig. 1A) (3, 10). The Tim23 channel did not display an intrinsic activity that forced fast channel closure. In organello, an open 450-pS channel, present in about 300 copies per isolated mitochondrion (18, 19), would seriously compromise the permeability barrier and the bioenergetic activity of the inner membrane. We thus asked whether additional components associated with Tim23 might play a role in regulating channel closure by comparing the membrane potential (Δψ) of yeast mutant mitochondria.

Fig. 1.

Impairment of the mitochondrial membrane potential in tim50 mutant mitochondria. (A) Left: Current recording of a planar lipid bilayer containing Tim23 at a constant membrane potential of +150 mV. Right: Count rate histograms of the adjacent current traces. (B and C) Membrane potential (Δψ) of isolated mitochondria from the indicated mutants (21) and their corresponding wild-type (WT) strains was measured by fluorescence quenching using the dye DiSC3(5). Valinomycin was added to dissipate the Δψ. The magnitude of Δψ was assessed by the difference in fluorescence quenching before and after addition of valinomycin. (D) Comparison of fluorescence quenching between tim and pam mutant mitochondria as percentage of the corresponding wild-type value. SEM values were calculated from at least three independent experiments.

The presequence translocase exists in two forms. One form (TIM23SORT) contains Tim23, Tim17, Tim50, and Tim21 and can direct preproteins into the inner membrane. The second form transports preproteins into the mitochondrial matrix and is associated with the motor PAM but lacks Tim21 (fig. S1A) (13, 20). We assessed the Δψ values of mitochondria isolated from different tim and pam mutants with the use of the potential-sensitive fluorescent dye 3,3′-dipropylthiadicarbocyanine iodide [DiSC3(5)] (21) (Fig. 1, B to D). tim17 and tim21 mutant mitochondria displayed Δψ values comparable to that of wild-type mitochondria (13), whereas tim50 mutant mitochondria (14) showed a severe reduction of Δψ (Fig. 1, B and D). In the translocon of the endoplasmic reticulum, the luminal Hsp70 (BiP) and a DnaJ protein are involved in regulation of the Sec61 channel (22, 23). Because PAM contains the Hsp70 Ssc1 and an associated DnaJ protein (Pam18), we asked whether the Δψ across the inner membrane was affected in mitochondria containing mutant versions of PAM subunits. However, neither ssc1, nor pam18, nor tim44 mutant mitochondria (21, 2428) exhibited a significant reduction in Δψ relative to wild-type mitochondria (Fig. 1, C and D), consistent with the observation that only two components, Tim50 and Tim17, are associated with Tim23 in both forms of the presequence translocase (13). Because the Tim23 channel is active in both forms, a factor that regulates channel closure should be present in either form. We conclude that functional Tim50 is required to maintain the Δψ in intact mitochondria.

The IMS domain of Tim50 interacts with the IMS domain of Tim23 (1416). [Tim50 does not directly interact with the TOM complex (13), nor is it present in any other translocase of the inner membrane (fig. S1B).] We asked whether Tim50IMS affected the gating of the Tim23 channel. We expressed and purified Tim50IMS from Escherichia coli cells (Fig. 2A) and added it to the reconstituted Tim23 channel (21). Nanomolar concentrations of Tim50IMS led to rapid and efficient closure of the channel (Fig. 2B). The reversal potential was not changed by addition of Tim50IMS (Fig. 2C). Similarly, the conductance states of the Tim23 channel, as shown for the main conductance of 445 ± 11 pS and the most frequent subconductance state of 150 ± 7 pS, were not altered by Tim50IMS (450 ± 8 pS, 155 ± 6 pS) (Fig. 2D), indicating that the basic channel properties were not affected.

Fig. 2.

The intermembrane space domain of Tim50 (Tim50IMS) induces closure of the Tim23 channel. (A) Tim50IMS was expressed and purified from E. coli cells. Tim50IMS-containing fractions were separated by SDS-polyacrylamide gel electrophoresis (PAGE) to assess purity of the preparation. (B) Tim23 was incorporated into planar lipid bilayers and analyzed in the presence or absence of Tim50IMS (trans chamber) at a potential of +100 mV. (C) Current-voltage relationship of Tim23 alone or after addition of the indicated amounts of Tim50IMS added at the trans side of the membrane. (D) Current-voltage relationship of Tim23 in the absence or presence of 12.5 nM Tim50IMS. The main conductance (squares) and the most frequent subconductance state (circles) are shown. SEM values were calculated from at least 10 independent measurements. (E) Tim50IMS was added to the reconstituted Tim23 channel from the cis side of the membrane. (F) As in (C), with the exception that 1.5 mM bovine serum albumin or 1.5 mM lysozyme were added instead of Tim50IMS to the trans chamber. (G) Voltage-dependent open probability of the Tim23 channel alone or in the presence of the indicated amounts of Tim50IMS. Quantification was performed by comparing the mean current determined over a time range of 1 min at a constant holding potential.

The reconstitution of Tim23 into small unilamellar liposomes leads to an asymmetric insertion such that the IMS domain is exposed to the trans compartment of the planar bilayer (10). To test the specificity of Tim50IMS-induced channel closure, we added Tim50IMS to the cis compartment. The Tim23 channel was not affected (Fig. 2E), demonstrating that the hydrophilic Tim50 domain affects gating of Tim23 only from the IMS side. Moreover, control proteins did not affect the reconstituted Tim23 channel, even at millimolar concentrations (Fig. 2F). The Δψ value of intact mitochondria has been estimated to be about 150 mV (29). Determination of the Tim23 channel open probability (Popen) confirmed that the reconstituted Tim23 was mainly in the open state at this physiological voltage, whereas the addition of Tim50IMS drastically reduced Popen at the relevant voltage range (Fig. 2G).

Tim50IMS-induced closure of the Tim23 channel occurred in distinct steps, reflecting the described conductance states of Tim23 (Fig. 2, B to D) (10). This finding suggests that Tim50IMS does not lead to a direct physical block of the channel, but rather induces distinct gating steps of Tim23 toward the closed state. Tim23 has been reported to form a dimer in mitochondria (11). The dimer apparently reflects the inactive state in the absence of preprotein, because Tim23 is monomeric when preproteins accumulate in the presequence translocase (11). We established an assay to monitor oligomerization of Tim23 in tim50 mutant mitochondria (21). Tim23 with a protein A tag was coexpressed in yeast cells containing an untagged wild-type copy of Tim23. Isolated mitochondria were lysed in digitonin and TIM23-PAM was purified by IgG-affinity chromatography. TIM and PAM components analyzed, including untagged Tim23, were copurified with tagged Tim23 when wild-type mitochondria were used (Fig. 3A) (13, 14). In conditional tim50 mutant mitochondria [with partial inactivation of Tim50 (13)], the copurification of various TIM and PAM components with tagged Tim23 was comparable to that in wild-type mitochondria, with one exception: The yield for the untagged form of Tim23 was strongly reduced in the copurification (Fig. 3A) (the mitochondrial levels of Tim23 were unchanged in tim50 mutants). Thus, a functional Tim50 is selectively required for oligomerization of Tim23 but not for the association of other TIM or PAM components with Tim23.

Fig. 3.

Presequences of mitochondrial precursor proteins overcome Tim50IMS-induced closure of the Tim23 channel. (A) Tim23ProtA was expressed from a plasmid in wild-type and tim50 mutant cells carrying a wild-type copy of TIM23 on the chromosome. Isolated mitochondria were solubilized in digitonin buffer and the TIM23 complex isolated via immunoglobulin G affinity chromatography (21). Bound proteins were eluted with sample buffer, separated by SDS-PAGE, and analyzed by immunoblotting. Signals were quantified with Scion Image 1.62a, standardized to the amount of isolated Tim23ProtA. The amount of protein purified from wild-type mitochondria was set to 100% (control). SEM was calculated from at least three independent experiments. (B) Current recording of Tim23 alone (upper panel), Tim23 plus Tim50IMS (second panel), and Tim23 plus Tim50IMS upon addition of P2 carrier-peptide (middle panel) or CoxIV-peptide (bottom panels) under asymmetrical buffer conditions; no holding potential applied (21).

These results indicate that Tim50 promotes the oligomerization and closure of the Tim23 channel in the absence of preproteins, allowing a tight regulation of the Tim23 pore in the physiologically relevant range of the membrane potential. However, for protein translocation the channel needs to be activated and reopened. The observation that presequences triggered a dissociation of the Tim23 dimer (11) led us to ask whether presequences were able to activate the closed Tim23 channel. We used the planar bilayer system with reconstituted Tim23 and added Tim50IMS to close the channel (Fig. 3B). Upon addition of a synthetic peptide corresponding to the presequence of cytochrome oxidase subunit IV, we observed opening of the channel with fast flickering gating (Fig. 3B). Thus, the channel was not simply shifted to a permanently open form but to a highly active form with rapid gating transitions. A signal peptide specific for mitochondrial carrier proteins (30) did not activate the Tim23 channel (Fig. 3B).

Our work reveals a molecular mechanism of how the proton-motive force across the mitochondrial inner membrane can be maintained despite the presence of large channels that transport entire precursor polypeptides across the membrane. The Tim23 channel is tightly regulated so as to maintain the permeability barrier of the inner mitochondrial membrane in its inactive state while being opened for translocation of polypeptide chains. Tim50 and presequences act in an antagonistic manner in this process. Whereas the hydrophilic IMS domain of Tim50 promotes oligomerization and voltage-dependent closure of the channel, presequences selectively override the Tim50-induced closure and activate the channel. This mechanism ensures selective on-demand opening of the Tim23 channel when a preprotein needs to be translocated through the presequence translocase and allows closure of the channel after transport is completed.

Supporting Online Material

www.sciencemag.org/cgi/content/full/312/5779/1523/DC1

Materials and Methods

Fig. S1

References

References and Notes

View Abstract

Navigate This Article