The MOM (mitochondrial outer membrane) contains SA (signal-anchored) proteins that bear at their N-terminus a single hydrophobic segment that serves as both a mitochondrial targeting signal and an anchor at the membrane. These proteins, like the vast majority of mitochondrial proteins, are encoded in the nucleus and have to be imported into the organelle. Currently, the mechanisms by which they are targeted to and inserted into the OM (outer membrane) are unclear. To shed light on these issues, we employed a recombinant version of the SA protein OM45 and a synthetic peptide corresponding to its signal-anchor segment. Both forms are associated with isolated mitochondria independently of cytosolic factors. Interaction with mitochondria was diminished when a mutated form of the signal-anchor was employed. We demonstrate that the signal-anchor peptide acquires an α-helical structure in a lipid environment and adopted a TM (transmembrane) topology within artificial lipid bilayers. Moreover, the peptide's affinity to artificial membranes with OM-like lipid composition was much higher than that of membranes with ER (endoplasmic reticulum)-like lipid composition. Collectively, our results suggest that SA proteins are specifically inserted into the MOM by a process that is not dependent on additional proteins, but is rather facilitated by the distinct lipid composition of this membrane.
Proteins residing in the MOM (mitochondrial outer membrane) facilitate various interactions between the organelle and the rest of the eukaryotic cell. All of these proteins are encoded in the nucleus and synthesized on cytosolic ribosomes. In contrast with most mitochondrial proteins that contain N-terminal cleavable mitochondrial presequences, all MOM proteins carry internal non-cleavable targeting and sorting signals [1–3]. These signals assure their specific targeting to the organelle and insertion into the lipid bilayer.
MOM proteins can be divided into those that contain only one TM (transmembrane) domain and those that are embedded in the membrane with several α-helical segments or as β-barrel structure. The SA (signal-anchored) proteins, which belong to the first group, owe their name to a short portion of their N-terminus that serves as both a mitochondrial targeting signal and an anchor to the MOM . The large remaining portion of the protein is exposed to the cytosol. Verified members of this family in fungi include the receptor components of the TOM (translocase of outer mitochondrial membrane) complex, Tom20 and Tom70, and two additional proteins: OM45 (outer membrane 45) and the OM isoform of Mcr1 (momMcr1). The targeting signal of these proteins consists of the TM segment and positively charged flanking regions [3,4].
A topologically related group are the tail-anchored OM proteins that have an α-helical TM domain in their C-terminus and a large N-terminally positioned domain, exposed to the cytosol . The targeting signal in these proteins and their mechanism of membrane integration was studied in some detail [6–9]. However, whereas the targeting information in tail-anchored proteins resides in the C-terminus, meaning that they have to be inserted into the membrane by a post-translational mechanism, some recognition of the N-terminal signal-anchor segment can occur while the rest of the polypeptide chain is still being synthesized. Hence, the biogenesis process of SA proteins can be completely different from that of the tail-anchored proteins.
Several lines of evidence suggest that import of SA protein does not require any of the known import components. First, it was demonstrated that mitochondrial targeting of Tom20 and Tom70 is not affected by either blocking the import receptors by specific antibodies or removing them upon proteolytic treatment [10,11]. Similarly, membrane insertion of SA proteins appears to be independent of the import pore formed by the TOM complex [11–13]. Despite the aforementioned similarities, it appears as if SA proteins do not share a common mechanism of targeting and insertion. Whereas OM45 and momMcr1 follow a still uncharacterized membrane integration pathway, the Tom receptors seem to use another pathway. It was proposed that another OM protein, Mim1 promotes both insertion of Tom20 and Tom70 into the OM and their final assembly into the TOM core complex [14–16]. Considering the observations described above, we previously suggested that proteins such as OM45 and momMcr1 can be inserted into the OM in a process that does not require the assistance of the known import components [13,17]. However, this model does not exclude the possibility that an as yet unknown OM protein is involved, and our detailed understanding of the mechanisms that govern the membrane integration of these proteins is still rather vague.
In order to shed light on the biogenesis of this special group of proteins, we used a recombinant OM45 protein and synthesized a peptide corresponding to the signal-anchor segment of the protein. Both forms, but not their mutated variants, were able to associate with isolated mitochondria. Our results indicate that the OM45 peptide can adopt an α-helical conformation in a lipid environment and can tightly bind to protein-free phospholipid bilayers. Importantly, the low ergosterol content of the MOM appears to play an important role in the biogenesis process. Taken together, our findings support a mechanism where SA proteins can be inserted into the MOM in a process that depends on the unique lipid composition of this membrane, but is independent of additional proteins.
Rink amide-MBHA (p-methylbenzhydrylamine) resin and Fmoc (fluoren-9-ylmethoxycarbonyl) amino acids were obtained from Calbiochem-Novabiochem. Other reagents used for peptide synthesis included TFA (trifluoroacetic acid), piperidine, DIEA (N,N-diisopropylethylamine), NMM (N-methylmorpholine), HOBT (N-hydroxybenzotriazole hydrate), HBTU [2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate], DCM (dichlormethan; peptide synthesis grade) and DMF (dimethylformamide; peptide synthesis grade) (all from Biolab). 4-Chloro-NBD-F (7-nitrobenz-2-oxa-1,3-diazole fluoride) was purchased from Sigma and TAMRA (6-carboxytetramethylrhodamine) succinimidyl ester was from Molecular Probes. Egg LPC (L-α-lysophosphatidylcholine) and PI (phosphatidylinositol) were purchased from Sigma, PC (phosphatidylcholine), PE (phosphatidylethanolamine), PS (phosphatidylserine) and CL (cardiolipin) (diphosphatidylglycerol) were from Avanti, and ergosterol was from Larodan.
Construction and purification of recombinant proteins
DNA sequences encoding either full-length OM45, its cytosolic domain (lacking residues 1–38) or a mutant version of the full-length protein (R4E, K26E) were cloned into the vector pET28a (Novagen), which includes a C-terminal His6 tag. The constructs were transformed into Escherichia coli cells (Rosetta2 DE3) and grown at 37°C to D600 of 0.6 before protein expression was induced with 0.5 mM isopropyl β-D-thiogalactoside. Cells were grown for an additional 4 h at 30°C and harvested by centrifugation. Next, the cells were resuspended in lysis buffer (50 mM sodium phosphate, 300 mM sodium chloride and 10 mM imidazole, pH 8.0) and lysis was performed by shaking for 1 h at 4°C in the presence of PMSF, lysozyme and RNase followed by sonification. The homogenate was centrifuged and the supernatant was loaded on to a column containing Ni-NTA (Ni2+-nitrilotriacetate)-agarose beads (Qiagen) and washed with 20 ml of lysis buffer with 20 mM imidazole. Elution of the protein was accomplished with 300 mM imidazole in lysis buffer. Purity and homogeneity of the proteins were confirmed by SDS/PAGE.
Peptide synthesis, fluorescent labelling and purification
Peptides were synthesized by an Fmoc solid-phase method on Rink amide resin. Cleavage of the peptides from the MBHA resin resulted in the amidation of the C-terminus. To label the peptides, the Fmoc protecting group at the N-terminus of the resin-bound peptides was removed by incubation with piperidine. All of the other reactive amine groups of the attached peptides were kept protected under these conditions. The resin-bound peptides were washed twice with DMF, and then treated with either NBD-F or TAMRA in anhydrous DMF containing 2% DIEA, leading to the formation of a resin-bound N-NBD or N-rhodamine peptide respectively. After 1 h (NBD) or 24 h (TAMRA), the resin was washed with DMF and then with methylene chloride. To label the ϵ-amino group of Lys26 with the NBD group, the following procedure was applied. The MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide] side chain protecting group, was removed under mild acidic conditions (2×1 min of 5% TFA in DCM and 30 min of 1% TFA in DCM), enabling the conjugation of the fluorophore after capping of the N-terminus with 1 M acetic anhydride and 0.3 M triethylamine in anhydrous DMF.
All of the peptides were purified by RP-HPLC (reversed-phase-HPLC) on a C4 RP Bio-Rad semi-preparative column. The column was eluted with a 40 min linear gradient of acetonitrile in water, containing 0.05% TFA, at a flow rate of 1.8 ml/min. The purified peptides were further subjected to electrospray MS to confirm their molecular mass.
Binding of recombinant proteins and synthetic peptides to isolated mitochondria
Isolated mitochondria (20 μg) were incubated in SEM buffer (250 mM sucrose, 1 mM EDTA and 10 mM Mops-KOH, pH 7.2) in the presence or absence of 100 mM KCl with 4 μg of recombinant purified OM45 proteins for 20 min at room temperature (21°C) and subsequently layered over a 60% (w/v) sucrose cushion. After centrifugation (40000 g, 40 min, 4°C) the supernatant was removed and proteins were precipitated. The layer between the cushion and the supernatant (5% of total volume) was diluted and centrifuged (20000 g, 15 min, 4°C). All fractions were subjected to SDS/PAGE and Western blot analysis. For the mitochondria–peptide interaction, 1 μM fluorescently labelled peptide was used and 100 μl fractions were collected. All fractions were analysed for their fluorescence and were analysed by SDS/PAGE and immunodecoration.
Preparation of LUVs (large unilamellar vesicles)
Thin films of a mixture of PC, PE, PI, CL and PS (molar ratio of 46:35:13:4:2 respectively) for OM-like LUVs or of PC, PE, PI, PS (54:25:11:10 respectively) containing 0.18 mol of ergosterol/mol of phospholipids for ER (endoplasmic reticulum)-like LUVs were generated from a solution of the lipids in a 2:1 (v/v) mixture of chloroform/methanol that was dried during rotation. The films were freeze-dried overnight, sealed with Ar(g), and stored at −20°C. To form vesicles, the films were suspended in PBS and vortex-mixed for 1.5 min. The lipid suspension underwent five cycles of freezing–thawing and then vesicles were extruded through polycarbonate membranes with 1.0 and then 0.2 μm pore diameters. The shape and size of the formed LUVs were verified by negative-staining electron microscopy.
Changes in the fluorescence of NBD-labelled peptides were measured upon binding to vesicles. LUVs (100 μM final concentration) were added to 400 μl of PBS containing 1 μM NBD-labelled peptide. Fluorescence measurements were performed at room temperature, with λex set at 467 nm (10 nm slit) and λem scan at 500–600 nm (10 nm slits) or set constant at 530 nm. The extent of insertion was measured by adding increasing amounts of LUVs to a fixed amount of peptide (0.2–1 μM). After the signal reached a plateau, proteinase K (125 μg/ml final concentration) was added.
ITC (isothermal titration calorimetry)
ITC was performed using an ITC200 microcalorimeter (MicroCal; GE Healthcare). Small aliquots (3 μl) of the SA-peptide solution (20 μM) were injected into a solution of LUVs (2 mM) in a cell volume of ~200 μl. Alternatively, small aliquots of an LUV solution (4 mM) were injected into a solution of peptide (10 μM). Control experiments included titration of peptide into buffer solution, injection of vesicle solution into buffer and the addition of buffer into LUV solution. The heat released upon dilution, determined in these control experiments was subtracted from the heat of the peptide–liposome binding reactions. Binding isotherms were integrated and analysed using Origin®7 software according to a ‘one binding site’ model.
Further methods are described in the Supplementary Online Data (available at http://www.BiochemJ.org/bj/442/bj4420381add.htm).
Recombinant OM45 and a signal-anchor peptide specifically bind isolated mitochondria
To study binding of SA proteins to the MOM in the absence of cytosolic chaperones, we used the MOM SA protein OM45 as a model protein. To follow binding, we analysed the sedimentation behaviour on a sucrose cushion of isolated organelle, recombinant His6-tagged OM45 purified from E. coli cells, and a mixture containing both components. Three fractions were collected after centrifugation: (i) an upper fraction that contains soluble unbound proteins, (ii) a middle fraction, where mitochondria was enriched, collected from the interface between the sucrose cushion and the buffer and (iii) a lower fraction that contained the pellet. Free OM45 was mainly present at the top fraction of the gradient, where as a subpopulation of free OM45 molecules was also detected in the lower pellet fraction (Figure 1A). We suggest that this latter population represents aggregated molecules. Incubation of the recombinant protein with isolated mitochondria in the presence of 100 mM salt prior to centrifugation resulted in co-migration of these components in the middle fraction. In contrast, a protein corresponding to the cytosolic domain of OM45 did not bind to the isolated organelle under these conditions (Figure 1A). Notably, the binding of full-length OM45 was not affected by trypsin-mediated removal of the exposed domains of protease-sensitive mitochondrial surface receptors (Figure 1A).
Binding of the OM45 protein to mitochondria
To verify the specificity of this binding assay, we used a recombinant variant of OM45 where Arg4 and Lys26 within the signal-anchor segment were changed to Glu4 and Glu26, thus changing the net charge of the regions flanking the TM helix from +5 to +1. We previously observed that these mutations dramatically reduced the ability of the corresponding signal-anchor to mediate insertion of a model protein into the MOM . In contrast with the native protein that was detected primarily in the bound fraction, a large portion of the mutated protein remained in the upper unbound fraction (Figure 1B). Hence, in accordance with our previous observation, the mutated protein demonstrates a reduced capacity to interact with isolated mitochondria. We next asked whether the mitochondria-associated OM45 is only attached on the surface of the organelle or also integrated into the membrane. To that end, the mitochondrial fraction was extracted after the binding reaction with alkaline solution and centrifuged. To allow comparison with the variants that lack the SA domain, the binding experiment was performed in a buffer lacking salt where small amounts of the cytosolic domain of OM45 can associate with the organelle (Figure 1C). Notably, whereas the cytosolic domain of OM45 is found solely in the soluble fraction, approximately a third of the full-length protein molecules are embedded in the membrane (Figure 1C). We conclude that binding and integration of recombinant OM45 into the OM can occur in the absence of cytosolic chaperones, and the assembly into the membrane depends on the SA domain.
The aforementioned experiments demonstrate that the signal-anchor segment is necessary for mitochondrial binding. In addition, we have previously demonstrated that the first 32 amino acid residues of OM45 are sufficient to target in vivo a passenger protein to mitochondria . To further investigate the mechanism by which the SA segment integrates into the organelle OM, we used a synthetic peptide corresponding to residues 2–32 of OM45 (OM45-SA). The peptide was selectively labelled at either its N-terminus or at Lys28 with the fluorophore NBD. This fluorophore can be utilized for binding studies since its fluorescence spectrum and intensity reflects the environment in which the NBD group is located. In addition a variant where the N-terminus is labelled with tetramethyl rhodamine was prepared. To serve as control peptides, we also synthesized and labelled a peptide harbouring the charge mutations (R4E and K26E) in the SA segment (Figure 2A). Purified peptides were shown to be homogeneous (>95%) by analytical HPLC and MS (results not shown).
Signal-anchor peptide can bind to mitochondria
To address the question whether the signal-anchor segment is able by itself to bind to mitochondria, we mixed the rhodamine-labelled peptides with isolated mitochondria. Next, the organelle was re-isolated by sedimentation on a sucrose cushion and fractions were collected. Importantly, rhodamine fluorescence of the native peptide and mitochondrial proteins were enriched in the same fractions. In contrast, the fluorescence of the mutated peptide was not detected in the mitochondria-containing fractions (Figure 2B). Thus this assay demonstrates specific binding of the native SA peptide to the organelle. To monitor whether the peptide is indeed integrated into the membrane, we incubated the isolated organelle with NBD-labelled peptide and then followed the emission spectrum of the NBD moiety. The addition of mitochondria to an aqueous solution of the peptide resulted in a dramatic increase in the fluorescence intensity of the NBD moiety and in the so-called ‘blue-shift’, a shift of the maximal emission to shorter wavelengths (Figure 2C). Of note these changes that demonstrate a shift of the NBD group to a more hydrophobic environment apparently upon integration into the OM were observed to a highly reduced extent with the mutant peptide (Figure 2C). To verify the insertion into the membrane, samples containing mitochondria-bound native peptide were treated with proteinase K. It is expected that a cleavage of an exposed N-terminus would result in the release of an NBD-containing fragment into the aqueous environment and subsequently a major reduction in the NBD fluorescence. Nevertheless, we observed that approximately 70% of the fluorescence signal remained (Figure 2D). These results suggest that the majority of the peptide molecules were inserted into the membrane.
Secondary structure of the OM45-SA peptide and its orientation within membranes
To better understand the structural requirements for membrane integration, we analysed by the CD method the secondary structure of the peptide in buffer or in the presence of LPC which provides a lipid-like environment, whereas in the former conditions the peptide acquired a mainly β-sheet conformation or was unstructured, the addition of the LPC resulted in conformations with mostly α-helical structure (Supplementary Figure S1 and Supplementary Table S1 available at http://www.BiochemJ.org/bj/442/bj4420381add.htm).
To further study the structure of the peptide in a lipid environment, we utilized ATR-FTIR (attenuated total reflectance-Fourier transform IR) spectroscopy. We first tested whether the addition of the peptide to the membrane affects the acyl-chain order of the phospholipid molecules. To that end, polarized ATR-FTIR was used to determine the orientation of the lipid membrane. The phospholipid composition (PC 46%, PE 35%, PI 13%, CL 4% and PS 2%) was according to the published composition of the yeast MOM . The symmetric [υsym(CH2)≈2853 cm−1] and the antisymmetric [υantisym(CH2)≈2922 cm−1] vibrations of lipid C−H bonds are perpendicular to the molecular axis of a fully extended hydrocarbon chain. The position of these peaks indicated that the membranes are predominantly in a native-like liquid-crystalline phase . Measurement of the dichroism of IR light absorbance can reveal the order and orientation of the membrane sample relative to the prism surface . The R value based on the stronger υantisym(CH2) was 1.18 and the corresponding orientation order parameter, f, was calculated (based on υantisym) to be 0.62. The effect of OM45-SA peptide on the multibilayer acyl-chains order can be estimated by comparing the CH2-stretching dichroic ratio of peptide-free multibilayers with that obtained with membranes containing incorporated peptide. The results indicate that integration of the SA peptide into the membranes changed somewhat the lipid order (Figures 3A and 3B). The R values and the corresponding order parameter f, at peptide/lipid molar ratios of 1:50 was 1.12 and 0.59 respectively. Hence the peptide decreased the lipid order, reflecting a reorganization of the lipid molecules probably due to the presence of oblique-inserted peptide molecules.
FTIR measurements of OM45-SA peptide in membrane environment
To determine directly the orientation of the peptides within lipid bilayers, polarized ATR-FTIR was used. Spectra of the amide I region of the peptide bound to the multibilayers and their deconvoluted components are shown in Figures 3(C) and 3(D). According to this analysis 52% of the peptide molecules are in an α-helical structure. This value is in a good agreement with the 58% α-helical structure that was estimated according to the CD measurements (Supplementary Table S1). The R value was 1.63 and the order parameter, f (−0.18) was slightly negative, typical of helices oriented oblique to the membrane surface . According to these parameters, an ‘average’ angle of 62° between the main axis of the peptide and the bilayer normal was obtained. We propose that this value actually results from a situation where approximately half of the molecules are in a surface conformation and the other half acquired the TM configuration. Such an average angle is also similar to that of peptides known to acquire a TM conformation . Taken together, these structural studies suggest that the peptide can adopt an α-helical TM conformation in the presence of artificial lipid membranes.
Binding of the OM45-SA peptide to pure lipid vesicles
We next addressed the question whether a proteinaceous element has to be postulated in the insertion of SA proteins. To that end, the interaction of the OM45 peptides with protein-free lipid vesicles was analysed. LUVs were prepared from phospholipids in a composition similar to that of the MOM in yeast . Addition of these LUVs to the NBD-labelled native or mutant peptide resulted in an enhancement in the fluorescence of the NBD group only in the case of the former peptide (Figures 4A and 4D). Hence addition of vesicles to the native peptide resulted in membrane integration of the NBD moiety. Interestingly, although the NBD group at the C-terminus of the native peptide displayed an enhancement of the fluorescence intensity in methanol, it failed to do so upon addition of LUVs (Figure 4B). The absence of such an increase probably indicates that the NBD group and the hydrophobic core of the membrane are distant from each other. While at the N-terminus only three residues separate the NBD group from the last amino acid of the putative TM segment, in the C-terminus this distance corresponds to six amino acids. Of note, we cannot exclude the unfavourable possibility that the NBD group at the C-terminus interferes with peptide binding to the membrane.
NBD-labelled OM45-SA peptide can integrate into artificial membrane vesicles
We then tested whether the peptide is indeed integrated into the lipid bilayer. The addition of proteinase K to the samples with the bound peptide resulted in an approximately 50% reduction in the fluorescence intensity (Figure 4C). Hence it appears that approximately half of the peptide molecules are in the TM configuration. We excluded the possibility of peptide aggregation since no self-quenching of the rhodamine-labelled peptide was observed under these conditions ( and results not shown).
Is the unique lipid composition of the MOM important for the membrane integration of the OM45 peptide? The N-terminally NBD-labelled peptide was titrated with LUVs and the increase in the fluorescence intensity was monitored. Plotting this enhancement as a function of the vesicle concentration allowed us to determine the affinity of the peptide to the vesicles. This affinity was calculated to be 223±12 μM when LUVs lacking ergosterol and having the phospholipid composition of the MOM were used (Figure 4E). The binding of the peptide to CL-free vesicles was comparable with CL-containing LUVs, and the binding of the peptide to mitochondria lacking CL (isolated from the cdr1▵ strain) was very similar to its binding to WT (wild-type) organelle (results not shown). Thus CL does not appear to play an important role in binding and integration of the OM45 peptide. We previously observed that elevated ergosterol levels interfere with the insertion of tail-anchored protein into lipid vesicles . To check whether a similar effect can be observed with the signal-anchor peptide, we included 2, 10 or 20% ergosterol in the mixture of the lipid vesicles. Adding these ergosterol-containing vesicles to the NBD-labelled peptide resulted in a strongly reduced enhancement in the fluorescence signal as compared with addition of ergosterol-free LUVs (Figures 4E and 4F).
To further investigate, in a quantitative manner, the role of the lipid composition in the membrane integration process, we studied its thermodynamics using ITC. To this end, a solution containing lipid vesicles with the composition of either the MOM or ER membrane was titrated into an ITC cell containing the native OM45 peptide, and the heat amount released upon the peptide–lipid binding reaction was recorded (Figures 5A and 5B). Progressive vesicle injections produced decreasing exothermic heats as the availability of the peptide in the cell was gradually exhausted by association with the liposomes. We observed an elevated heat release upon binding of the SA-peptide to the vesicles with the MOM-like lipid composition. Analysis of the data resulted in enthalpy (▵H) of −31 kcal/mol (1 kcal=4.184 kJ) for these membranes, whereas the binding to the ER-like vesicles was too weak to allow any fitting (Figures 5C and 5D). These results indicate that the peptide interaction with MOM-like membranes is thermodynamically favoured as compared with its interaction with ER-like membrane.
Binding of the SA-peptide to MOM-like and ER-like vesicles
Next, we turned to investigate the kinetics of membrane binding of the SA peptide by utilizing a stopped-flow setup. The addition of the vesicles to the NBD-labelled peptide resulted in a biphasic increase in the fluorescence intensity. We observed a first phase where the fluorescence increased very fast (within <1 s) to approximately 90% of the final intensity and a second slower phase that lasted several minutes (Figure 6A). Indeed, the first association rate constant was over 100-fold larger than the second one (Figure 6B). Similar biphasic behaviour was observed also for the membrane-active peptide mastoparan X . Both the initial kinetics and the final fluorescence level were dependent on the amounts of the added vesicles. According to this dependency and considering only kobs1 we could calculate the Kd to be 244 μM, which is in an excellent agreement with the value obtained in the previous experiments (223 μM). In agreement with the aforementioned results, addition of vesicles containing 10% ergosterol resulted in slower kinetics and a reduction in the fluorescence increase (results not shown).
Kinetics of peptide binding to LUVs monitored using a stopped-flow setup
A TM configuration of OM45-SA is thermodynamically more stable than a surface configuration
The results of the ATR-FTIR spectroscopy and the protection of the NBD probe from externally added protease suggest a TM configuration of the peptide. We asked whether this topology is indeed thermodynamically favoured. To that goal we used MC (Monte Carlo) simulations to calculate the free energy of the peptide in two configurations: parallel to the membrane surface and in TM orientation. For comparison, similar simulations were performed for the charge variant (R4E, K26E) peptide. The results show that the surface orientation of the native peptide is rather stable (Table 1), explaining the efficient and fast initial binding to lipid membranes (Figure 6). Interestingly, the values of the free energy decomposition obtained by the MC simulations suggest that the surface conformation of the peptide is stabilized by both solvation and electrostatic attractions (Table 1). However, the TM configuration is by far more stable, and therefore a reorganization of the peptide to the TM conformation would be favoured. This transformation is driven mostly by solvation since the electrostatic component is approximately the same in both configurations (Table 1). As the peptide has to cross the lipid core of the membrane, this re-orientation is expected to be slower than the initial binding; however, once occurred it will essentially not be reversible. In sharp contrast, the mutant peptide has less favourable electrostatic interactions with the lipid molecules and low stability in both the surface and TM orientations (Table 1). Taken together, these results in combination with the stopped-flow data can support a mechanism where the signal-anchor segment first binds at the surface of the organelle and then gets inserted into the membrane and acquires a TM conformation.
In the present study, we investigated the mechanism by which SA proteins are delivered from the cytosol into their final location at the MOM and subsequently inserted into this membrane. We choose OM45 as a model protein since, in contrast with Tom20 and Tom70, it is not a subunit of the TOM complex. Subunits of this complex might represent a special case and indeed, the correct topology and assembly of the latter two proteins depends on pre-existing TOM complex [11,12]. Our results suggest that cytosolic factors are not required for the membrane integration of OM45. Interestingly, similar independency of cytosolic chaperones was suggested for some ER and mitochondrial tail-anchored proteins [7,26]. Of note, although in vivo such cytosolic chaperones might bind newly synthesized SA proteins, the results of the present study suggest that they are not strictly required for specific targeting and integration into the OM.
Our working hypothesis suggests that the SA domain interacts with the membrane according to a two-state model that was suggested for many membrane-active peptides [27–30]. According to this model, the peptide first binds rather fast to the lipid–water interface and forms α-helical structure. Induction of an α-helical structure upon interaction of membrane-active peptides with lipids is a common observation for such peptides [28,30,31]. Previous work [31,32], as well as our MC simulations and present findings with the charge variant, suggest that electrostatic interactions among the positively flanking regions of the SA segment and the anionic headgroups of the phospholipids probably contribute to this interaction. After the initial absorption, the SA segment reorganizes itself within the membrane and adopts a TM configuration. The driving force for this step is the free energy gain thanks to the hydrophobic effect, i.e. the non-polar interactions between the TM region and tails of phospholipid molecules surrounding it. Favourable electrostatic interactions between the flanking regions and the headgroups on both sides of the membrane further stabilize the TM configuration. The importance of the electrostatic interactions is reflected by a reduced calculated thermodynamic stability of a charge variant of OM45 and a reduced insertion of the corresponding protein or peptide into natural and artificial membranes. Furthermore, this working model is also in line with a previous study reporting that physiological membrane concentrations of anionic lipids greatly stabilize the TM configuration of a model hydrophobic segment with positively charged flanking regions .
A protein-free membrane-insertion mechanism was proposed to describe the integration of some tail-anchored proteins into the membranes of the ER and mitochondria [9,26,33]. It is also in line with previous studies that failed to identify a mitochondrial protein which is required for the membrane integration of OM45 and momMcr1 [12,13]. In contrast, Mim1 was reported to be required for the membrane insertion of Tom20 and Tom70, and/or for their assembly with the TOM complex [14–16]. However, the request for Mim1 and Tom40 in the biogenesis route of Tom receptors does not have to exclude a unified mechanism of initial insertion of SA proteins into the MOM. It might well be that all SA proteins follow initially a similar pathway of insertion that does not require additional proteins. According to this scenario, Mim1 is required for subsequent steps in the assembly of Tom20 and Tom70 with the rest of the TOM complex [15,34].
The apparently minor importance of proteins in assuring the specificity of the targeting and recognition processes raises the question of what other elements can mediate such distinct mitochondrial location. One such element can be the unique lipid composition of the MOM. Indeed, the SA peptide associated much better with lipid vesicles with the OM composition as compared with vesicles with ER-like composition. Thus we looked more closely at unique features of the lipid content of the MOM. For example, the MOM is the only one among all membranes facing the cytosol that contain the mitochondrial-specific phospholipid CL [19,35]. However, whereas this lipid seems to play a role in the association of apoptosis-regulating proteins with the OM [36,37], our results do not support a crucial role of CL in assuring specific membrane integration of OM45.
Low levels of ergosterol are another unique feature of the yeast MOM [38,39]. Indeed, our previous results with the tail-anchored protein Fis1  and current findings with OM45 suggest that low ergosterol content is crucial for efficient membrane integration of OM single-span proteins. In addition to its well-known enhancement of the rigidity of the membrane, ergosterol might also influence the compressibility of the headgroups region. It has a relatively small headgroup allowing a more compact packing of the headgroup region of the bilayer [32,40]. Hence the elevated rigidity and reduced compressibility of sterol-containing membrane can result in compromised productive binding of signal-anchor peptides to this membrane.
In summary, we propose that newly synthesized mitochondrial SA protein might associate non-specifically and reversibly with various cellular membranes, but would bind productively, irreversibly and in a TM conformation only to the MOM. Although additional proteins can enhance the membrane integration process or preserve the SA protein in an insertion-competent conformation, we suggest that a distinct lipid composition of the MOM plays a primary role in the specific integration of SA proteins into this membrane.
attenuated total reflectance-Fourier transform IR
isothermal titration calorimetry
large unilamellar vesicle
mitochondrial outer membrane
translocase of outer mitochondrial membrane
Elisa Merklinger, Arnold Driessen, Nir Ben-Tal, Yechiel Shai and Doron Rapaport designed the research. Elisa Merklinger, Yana Gofman, Alexej Kedrov and Yechiel Shai performed the research. All authors analysed the data. Elisa Merklinger and Doron Rapaport wrote the paper.
We thank K. Rehn, E. Kracker and B. Zarmi for technical support, G. Schreiber and M. Harel for help with the stopped-flow setup, and K.S. Dimmer and M. Schuldiner for helpful discussions.
This work was supported by the Deutsche Forschungsgemeinschaft [grant number RA 1048/4–1 to D.R.] and a short-term fellowship from the Minerva Stiftung (to E.M.).