Novel thioester reagents afford efficient and specific S-acylation of unprotected peptides under mild conditions in aqueous solution.

S-acylated peptides have many potential uses for elucidating the biophysical, structural and other properties of the numerous S-acylated proteins of mammalian cells. However, with the currently available reagents, preparation of specifically S-acylated derivatives of peptides is generally laborious or simply unfeasible. We here show that novel, easily preparable aryl and alkyl thioester derivatives of palmitic acid can mediate S-acylation of peptides corresponding to physiologically S-acylated sequences from the proteins p56(lck) and H-ras and the Po glycoprotein of peripheral myelin, with high selectivity for cysteine over other amino acid functional groups (including hydroxyl and both alpha- and epsilon-amino residues), and with much greater efficiency than is obtained using acyl-coenzyme A derivatives. Efficient and selective S-acylation can be accomplished under very mild conditions in aqueous systems containing lipid vesicles or detergent micelles, or in homogenous aqueous/acetonitrile mixtures. Using these novel thioesterifying reagents, we confirm previous suggestions that the N-terminal cysteine residue of Hedgehog proteins can exhibit rapid, uncatalyzed S-to-N acyl transfer following S-acylation to produce the N-palmitoylated amino terminus found in the mature protein. By contrast, we demonstrate that spontaneous S-to-N acyl transfer from the cysteine to the terminal glycine residue in the amino-terminal peptide of G(alphas) is far less rapid and is likely too slow to explain the physiological N-palmitoylation of the amino terminus of this protein.

Mechanisms of Ras protein targeting in mammalian cells.

Many physiological and oncogenic activities of the "classical" Ras proteins (H-Ras, N-Ras and K-Ras4A and -4B) require their correct localization to the plasma membrane. Nascent Ras proteins, however, initially associate with endomembranes (the ER and in some cases the Golgi) to complete the processing of their farnesylated carboxyl-termini before they are delivered to the plasma membrane. Recent work has revealed the outlines of the intracellular pathways by which Ras proteins reach their ultimate plasma membrane destination and has indicated that these pathways differ for different Ras species. Other studies have demonstrated that mature Ras proteins can transfer between the plasma membrane and intracellular membranes, and that Ras proteins may in some cases signal from intracellular compartments. This review will describe recent progress and still-unresolved questions in these areas.

Partitioning of lipidated peptide sequences into liquid-ordered lipid domains in model and biological membranes.

We have used a fluorescence assay and detergent fractionation to examine the partitioning of different fluorescent lipidated peptides, with sequences and lipid substituents matching those found in various classes of lipidated cellular proteins, into liquid-ordered (raft-like) domains in lipid bilayers. Peptides incorporating isoprenyl groups, or multiple unsaturated acyl chains, show negligible affinity for liquid-ordered domains in mixed-phase liquid-ordered/liquid-disordered (l(o)/l(d)) bilayers composed of dipalmitoylphosphatidylcholine, a spin-labeled unsaturated phosphatidylcholine, and cholesterol. By contrast, peptides incorporating multiple S- and/or N-acyl chains, or a cholesterol residue plus an N-terminal palmitoyl chain, show significant partitioning into liquid-ordered domains under the same conditions. Interestingly, the affinity of a lipidated peptide for l(o) domains can be strongly influenced, not only by the structures of the lipid substituents but also by the nature and the positions of their attachment to the peptide chain. These results are well correlated with those obtained from parallel assays based on low-temperature detergent fractionation. Using the latter approach, we further demonstrate that a truly minimal l(o) domain partitioning motif [myristoylGlyCys(palmitoyl)-] can mediate efficient incorporation into the "raft" fraction of COS-7 cell membranes.

Cholesterol does not induce segregation of liquid-ordered domains in bilayers modeling the inner leaflet of the plasma membrane.

A fluorescence-quenching method has been used to assess the potential formation of segregated liquid-ordered domains in lipid bilayers combining cholesterol with mixtures of amino and choline phospholipids like those found in the cytoplasmic leaflet of the mammalian cell plasma membrane. When present in proportions >20-30 mol %, different saturated phospholipids show a strong proclivity to form segregated domains when combined with unsaturated phospholipids and cholesterol, in a manner that is only weakly affected by the nature of the phospholipid headgroups. By contrast, mixtures containing purely unsaturated phospholipids and cholesterol do not exhibit detectable segregation of domains, even in systems whose components differ in headgroup structure, mono- versus polyunsaturation and/or acyl chain heterogeneity. These results indicate that mixtures of phospholipids resembling those found in the inner leaflet of the plasma membrane do not spontaneously form segregated liquid-ordered domains. Instead, our findings suggest that factors extrinsic to the inner-monolayer lipids themselves (e.g., transbilayer penetration of long sphingolipid acyl chains) would be essential to confer a distinctive, more highly ordered organization to the cytoplasmic leaflet of "lipid raft" structures in animal cell membranes.

Use of cyclodextrins to monitor transbilayer movement and differential lipid affinities of cholesterol.

In view of the demonstrated cholesterol-binding capabilities of certain cyclodextrins, we have examined whether these agents can also catalyze efficient transfer of cholesterol between lipid vesicles. We here demonstrate that beta- and gamma-cyclodextrins can dramatically accelerate the rate of cholesterol transfer between lipid vesicles under conditions where a negligible fraction of the sterol is bound to cyclodextrin in steady state. beta- and gamma-cyclodextrin enhance the rate of transfer of cholesterol between vesicles by a larger factor than they accelerate the transfer of phospholipid, whereas, for alpha- and methyl-beta-cyclodextrin, the opposite is true. Analysis of the kinetics of cyclodextrin-mediated cholesterol transfer between large unilamellar vesicles composed mainly of 1-stearoyl-2-oleoyl phosphatidylcholine (SOPC) or SOPC/cholesterol indicates that transbilayer flip-flop of cholesterol is very rapid (halftime < 1-2 min at 37 degrees C). Using beta-cyclodextrin to accelerate cholesterol transfer, we have measured the relative affinities of cholesterol for a variety of different lipid species. Our results show strong variations in cholesterol affinity for phospholipids bearing different degrees of chain unsaturation and lesser, albeit significant, effects of phospholipid headgroup structure on cholesterol-binding affinity. Our findings also confirm previous suggestions that cholesterol interacts with markedly higher affinity with sphingolipids than with common membrane phospholipids.

Different sphingolipids show differential partitioning into sphingolipid/cholesterol-rich domains in lipid bilayers.

Two fluorescence-based approaches have been applied to examine the differential partitioning of fluorescent phospho- and sphingolipid molecules into sphingolipid-enriched domains modeling membrane "lipid rafts." Fluorescence-quenching measurements reveal that N-(diphenylhexatrienyl)propionyl- (DPH3:0-)-labeled gluco- and galactocerebroside partition into sphingolipid-enriched domains in sphingolipid/phosphatidylcholine/cholesterol bilayers with substantially higher affinity than do analogous sphingomyelin, ceramide, or phosphatidylcholine molecules. By contrast, the affinity of sphingomyelin and ceramide for such domains is only marginally greater than that of a phosphatidylcholine with similar hydrocarbon chains. By using direct measurements of molecular partitioning between vesicles of different compositions, we show that the relative affinities of different C(6)-NBD- and C(5)-Bodipy-labeled sphingolipids for sphingolipid-enriched domains are quantitatively, and in most circumstances even qualitatively, quite different from those found for species whose N-acyl chains more closely resemble the long saturated chains of cellular sphingolipids. These findings lend support in principle to previous suggestions that differential partitioning of different sphingolipids into "raft" domains could contribute to the differential trafficking of these species in eukaryotic cells. However, our findings also indicate that short-chain sphingolipid probes previously used to examine this phenomenon are in general ill-suited for such applications.

Fluorescence-based evaluation of the partitioning of lipids and lipidated peptides into liquid-ordered lipid microdomains: a model for molecular partitioning into "lipid rafts".

A fluorescence-quenching assay is described that can directly monitor the relative extents of partitioning of different but structurally homologous fluorescent molecules into liquid-ordered (l(o)) domains in lipid vesicles exhibiting liquid-ordered/liquid-disordered (l(o)/l(d)) phase coexistence. Applying this assay to a series of bimane-labeled diacyl phospholipid probes in cholesterol-containing ternary lipid mixtures exhibiting l(o)/l(d) phase separation, we demonstrate that partitioning into l(o)-phase domains is negligible for diunsaturated species and greatest for long-chain disaturated species. These conclusions agree well with those derived from previous studies of the association of lipids and lipid-anchored molecules with l(o)-phase domains, using methods based on the isolation of a detergent-insoluble fraction from model or biological membranes at low temperatures. However, we also find that monounsaturated and shorter-chain saturated species partition into l(o) phases with significant, albeit modest affinities, and that the level of partitioning of these latter species into l(o)-phase domains is significantly underestimated (relative to that of their long-chain saturated counterparts) by the criterion of low-temperature detergent insolubility. Finally, applying the fluorescence-quenching method to a family of lipid-modified peptides, we demonstrate that the S-palmitoyl/S-isoprenyl dual-lipidation motif found in proteins such as H- and N-ras and yeast Ste18p does not promote significant association with l(o) domains in l(o)/l(d)-phase-separated bilayers.

Mutational and biochemical analysis of plasma membrane targeting mediated by the farnesylated, polybasic carboxy terminus of K-ras4B.

Mutational analysis and in vitro assays of membrane association have been combined to investigate the mechanism of plasma membrane targeting mediated by the farnesylated, polybasic carboxy-terminal sequence of K-ras4B in mammalian cells. Fluorescence-microscopic localization of chimeric proteins linking the enhanced green fluorescent protein (EGFP) to the K-ras4B carboxy-terminal sequence, or to variant forms of this sequence, reveals that the normal structure of this targeting motif can be greatly altered without compromising plasma membrane-targeting activity so long as an overall strongly polybasic/amphiphilic character is retained. An EGFP/K-ras4B(171-188) chimeric protein was readily abstracted from isolated cell membranes by negatively charged lipid vesicles, and this abstraction was markedly enhanced by the anionic lipid-binding agent neomycin. Our results strongly favor a mechanism in which at the plasma membrane the carboxy-terminal sequence of K-ras4B associates not with a classical specific proteinaceous receptor but rather with nonspecific but highly anionic 'sites' formed at least in part by the membrane lipid bilayer. Our findings also suggest that the recently demonstrated prenylation-dependent trafficking of immature forms of K-ras4B through the endoplasmic reticulum [Choy et al. (1999) Cell 98, 69-80], while required for maturation of the protein, beyond this stage may not be essential to allow the ultimate delivery of the mature protein to the plasma membrane.

Differential effects of acyl-CoA binding protein on enzymatic and non-enzymatic thioacylation of protein and peptide substrates.

Both enzymatic and autocatalytic mechanisms have been proposed to account for protein thioacylation (commonly known as palmitoylation). Acyl-CoA binding proteins (ACBP) strongly suppress non-enzymatic thioacylation of cysteinyl-containing peptides by long-chain acyl-CoAs. At physiological concentrations of ACBP, acyl-CoAs, and membrane lipids, the rate of spontaneous acylation is expected to be too slow to contribute significantly to thioacylation of signaling proteins in mammalian cells (Leventis et al., Biochemistry 36 (1997) 5546-5553). Here we characterized the effects of ACBP on enzymatic thioacylation. A protein S-acyltransferase activity previously characterized using G-protein alpha-subunits as a substrate (Dunphy et al., J. Biol. Chem., 271 (1996) 7154-7159), was capable of thioacylating short lipid-modified cysteinyl-containing peptides. The minimum requirements for substrate recognition were a free cysteine thiol adjacent to a hydrophobic lipid anchor, either myristate or farnesyl isoprenoid. PAT activity displayed specificity for the acyl donor, efficiently utilizing long-chain acyl-CoAs, but not free fatty acid or S-palmitoyl-N-acetylcysteamine. ACBP only modestly inhibited enzymatic thioacylation of a myristoylated peptide or G-protein alpha-subunits under conditions where non-enzymatic thioacylation was reduced to background. Thus, protein S-acyltransferase remains active in the presence of physiological concentrations of ACBP and acyl-CoA in vitro and is likely to represent the predominant mechanism of thioacylation in vivo.

Direct membrane insertion of voltage-dependent anion-selective channel protein catalyzed by mitochondrial Tom20.

Insertion of newly synthesized proteins into or across the mitochondrial outer membrane is initiated by import receptors at the surface of the organelle. Typically, this interaction directs the precursor protein into a preprotein translocation pore, comprised of Tom40. Here, we show that a prominent beta-barrel channel protein spanning the outer membrane, human voltage- dependent anion-selective channel (VDAC), bypasses the requirement for the Tom40 translocation pore during biogenesis. Insertion of VDAC into the outer membrane is unaffected by plugging the translocation pore with a partially translocated matrix preprotein, and mitochondria containing a temperature-sensitive mutant of Tom40 insert VDAC at the nonpermissive temperature. Synthetic liposomes harboring the cytosolic domain of the human import receptor Tom20 efficiently insert newly synthesized VDAC, resulting in transbilayer transport of ATP. Therefore, Tom20 transforms newly synthesized cytosolic VDAC into a transmembrane channel that is fully integrated into the lipid bilayer.

Experimental and Monte Carlo simulation studies of the thermodynamics of polyethyleneglycol chains grafted to lipid bilayers.

Experimental measurements of the affinity of binding of fluorescent acylated polyethyleneglycol (PEG) conjugates to bilayers containing varying levels of phosphatidylethanolamine-PEGs (PE-PEGs) have been combined with Monte Carlo simulations to investigate the properties of the polymer chains at a PEG-grafted lipid interface. The affinity of binding of such conjugates to large unilamellar phosphatidylcholine/phosphatidylethanolamine (9:1) vesicles decreases 27-fold as the size of the coupled PEG chain increases from 1 to 114 monomer units. Incorporation of increasing amounts of PE-PEG2000 or PE-PEG5000 into the vesicles progressively reduces the affinity of binding of acylpeptide-PEG2000 or -PEG5000 conjugates. Monte Carlo simulations of surfaces with grafted PEG chains revealed no significant dependence of several characteristic properties of the polymer chains, including the average internal energy per polymer and the radii of gyration, on the grafting density in the range examined experimentally. The average conformation of a surface-grafted PEG2000 or PEG5000 chain was calculated to be fairly extended even at low grafting densities, and the projected cross-sectional areas of the grafted PEG chains are considerably smaller than those predicted on the basis of the estimated Flory radius. The experimental variation of the binding affinity of acylated conjugates for bilayers containing varying mole fractions of PE-PEG2000 or -PEG5000 is well explained by expressions treating the surface-grafted PEG polymers either as a van der Waals gas or as a system of rigid discs described by scaled particle theory. From the combined results of our experimental and simulation studies we conclude that the grafted PEG chains exist in a "mushroom" regime throughout the range of polymer densities examined experimentally and that the diminished affinity of binding of acylated-PEG conjugates to bilayers containing PE-PEGs results from occlusion of the surface area accessible for conjugate binding by the mobile PE-PEG polymer chains.

Lipid-binding characteristics of the polybasic carboxy-terminal sequence of K-ras4B.

We have examined the association with lipid vesicles of fluorescent lipidated peptides based on the farnesylated, polybasic carboxy-terminal region of mature K-ras4B, which functions physiologically as an autonomous plasma membrane-targeting motif. While the peptides bind to neutral lipid (phosphatidylcholine/phosphatidylethanolamine) vesicles with relatively low affinity, the vesicle-binding affinity increases exponentially as increasing amounts of anionic lipids are incorporated into the vesicle bilayers. Competitive vesicle-binding experiments reveal that the K-ras4B carboxy-terminal sequence accordingly discriminates strongly between lipid surfaces of differing surface charge, such that two lipid bilayers differing in anionic lipid content by 10 mol % will show a 45-fold preferential accumulation of the lipidated peptide in the more negatively charged surface. At the same time, the carboxyl-terminal region of K-ras4B exhibits no preferential binding to particular anionic lipids, including the polyanionic species phosphatidylinositol-4'-phosphate and phosphatidylinositol-4',5'-bisphosphate, beyond that predicted on the basis of surface-charge effects. The K-ras4B carboxyl-terminal sequence dissociates rapidly (with half-times of seconds or less) from lipid bilayers containing up to 40 mol % anionic lipid. These results suggest that the targeting of the mature K-ras4B carboxy-terminus to the plasma membrane, if it is based on interactions with plasma membrane lipids, is not mediated by a kinetic-trapping mechanism or by specific binding to particular anionic lipids but may rest on the sensitive surface potential-sensing function of this region of the protein.

Comprehensive geriatric assessment. Helping your elderly patients maintain functional well-being.

In this article, three geriatricians describe an approach to comprehensive geriatric assessment that takes into account the multiple social and medical problems that affect the functional well-being of frail elderly patients. With a 45- to 90-minute time investment, physicians can obtain an inventory of the factors that threaten an elderly patient's independence and gain a fuller understanding of the patient's complex needs.

S-Acylation and plasma membrane targeting of the farnesylated carboxyl-terminal peptide of N-ras in mammalian fibroblasts.

We have used a series of fluorescent lipid-modified peptides, based on the farnesylated C-terminal sequence of mature N-ras [-GCMGLPC(farnesyl)-OCH3], to investigate the membrane-anchoring properties of this region of the protein and its reversible modification by S-acylation in cultured mammalian fibroblasts. The farnesylated peptide associates with lipid bilayers (large unilamellar phospholipid vesicles) with high affinity but in a rapidly reversible manner. Additional S-palmitoylation of the peptide suppresses its ability to desorb from, and hence to diffuse between, lipid bilayers on physiologically significant time scales. NBD-labeled derivatives of the farnesylated N-ras C-terminal heptapeptide, when incubated with CV-1 cells in culture, are taken up by the cells and reversibly S-acylated in a manner similar to that observed previously for the parent protein. The S-acylation process is highly specific for modification of a cysteine rather than a serine residue but tolerates replacement of the peptide-linked farnesyl moiety by other hydrophobic groups. Fluorescence microscopy reveals that in CV-1 cells the S-acylated form of the peptide is localized preferentially to the plasma membrane, as has been observed for N-ras itself. This plasma membrane localization is unaffected by either reduced temperature (15 degrees C) or exposure to brefeldin A, treatments which inhibit various trafficking steps within the secretory pathway. These results suggest that in mammalian cells the plasma membrane localization of mature N-ras is maintained by a 'kinetic trapping' mechanism based on S-acylation of the protein at the level of the plasma membrane itself.

Acyl-CoA binding proteins inhibit the nonenzymic S-acylation of cysteinyl-containing peptide sequences by long-chain acyl-CoAs.

Acyl-CoA binding proteins (ACBPs) from rat and bovine liver were found to inhibit the nonenzymic S-acylation of two representative types of peptides by long-chain acyl-CoAs. As demonstrated previously [Quesnel, S. & Silvius, J. R. (1994) Biochemistry 33 13340-13348; Bharadwaj, M., & Bizzozero, O. A. (1995) J. Neurochem. 65, 1805-1815], peptides with the sequences myristoyl-GCG, myristoyl-GCV, and IRYCWLRR-NH2, all representing physiological S-acylation sites in mammalian proteins, become S-acylated at appreciable rates in the presence of long-chain acyl-CoAs and large unilamellar lipid vesicles. Addition of ACBP at physiological molar ratios with respect to long-chain acyl-CoAs strongly inhibits the spontaneous S-acylation reaction, in a manner that can be quantitatively described by assuming that the ACBP sequesters the acyl-CoA with nanomolar affinity in a complex unable to serve as an S-acyl donor. From these results, we calculate that at physiological (intracellular) concentrations of ACBP, long-chain acyl-CoAs, and membrane lipids the expected half-times for spontaneous S-acylation of such protein sequences by long-chain acyl-CoAs will lie in the range of several tens of hours. The nonenzymic reaction of protein cysteine residues with long-chain acyl-CoAs is thus unlikely to contribute significantly to the physiological modification of signaling and other proteins that show relatively rapid rates of S-acylation in mammalian cells. However, it cannot be excluded that a nonenzymic reaction with long-chain acyl-CoAs could contribute to the physiological S-acylation of certain membrane proteins if the latter exhibit very slow kinetics of S-acylation in vivo.

Cholesterol at different bilayer concentrations can promote or antagonize lateral segregation of phospholipids of differing acyl chain length.

Fourier-transform infrared-spectroscopic and fluorescence measurements have been combined to examine the effect of cholesterol on the intermixing of short-chain dilauroyl phosphatidylcholine (DLPC) and its bromo-substituted derivative (12BrPC) with longer-chain (C16- or C18-) phosphatidylcholines (PCs) in hydrated lipid bilayers. Infrared spectroscopy of mixtures combining protonated DLPC or 12BrPC with chain-perdeuterated dipalmitoyl PC reveals that cholesterol at lower concentrations in the bilayer modifies the resolved thermal melting profiles for both phospholipid components and, at high bilayer concentrations, produces a convergence of the thermal transitions for the two PC species. Fluorescence-quenching measurements using a short-chain fluorescent PC (1-dodecanoyl-2-[8-[N-indolyl]octanoyl] PC) in ternary mixtures combining 12BrPC, dipalmitoyl or distearoyl PC, and cholesterol confirm that very high cholesterol levels (50 mol %) abolish the lateral segregation of the PC components at 25 degrees C, a temperature where the phospholipids extensively phase-separate in the absence of sterol. By contrast, under these same conditions cholesterol at lower concentrations in the bilayer is found to enhance the tendency of the PC components to exhibit lateral segregation. We show that these seemingly contradictory effects of cholesterol can be readily explained in the light of a ternary phase diagram that is fully consistent with out current understanding of the nature of cholesterol-phospholipid interactions in binary mixtures.

Lipid-modified, cysteinyl-containing peptides of diverse structures are efficiently S-acylated at the plasma membrane of mammalian cells.

A variety of cysteine-containing, lipid-modified peptides are found to be S-acylated by cultured mammalian cells. The acylation reaction is highly specific for cysteinyl over serinyl residues and for lipid-modified peptides over hydrophilic peptides. The S-acylation process appears by various criteria to be enzymatic and resembles the S-acylation of plasma membrane-associated proteins in various characteristics, including inhibition by tunicamycin. The substrate range of the S-acylation reaction encompasses, but is not limited to, lipopeptides incorporating the motifs myristoylGC- and -CXC(farnesyl)-OCH3, which are reversibly S-acylated in various intracellular proteins. Mass-spectrometric analysis indicates that palmitoyl residues constitute the predominant but not the only type of S-acyl group coupled to a lipopeptide carrying the myristoylGC- motif, with smaller amounts of S-stearoyl and S-oleoyl substituents also detectable. Fluorescence microscopy using NBD-labeled cysteinyl lipopeptides reveals that the products of lipopeptide S-acylation, which cannot diffuse between membranes, are in almost all cases localized preferentially to the plasma membrane. This preferential localization is found even at reduced temperatures where vesicular transport from the Golgi complex to the plasma membrane is suppressed, strongly suggesting that the plasma membrane itself is the preferred site of S-acylation of these species. Uniquely among the lipopeptides studied, species incorporating an unphysiological N-myristoylcysteinyl- motif also show substantial formation of S-acylated products in a second, intracellular compartment identified as the Golgi complex by its labeling with a fluorescent ceramide. Our results suggest that distinct S-acyltransferases exist in the Golgi complex and plasma membrane compartments and that S-acylation of motifs such as myristoylGC- occurs specifically at the plasma membrane, affording efficient targeting of cellular proteins bearing such motifs to this membrane compartment.

An amphiphilic lipid-binding domain influences the topology of a signal-anchor sequence in the mitochondrial outer membrane.

Mas70p is targeted and inserted into the mitochondrial outer membrane in the N(in)-C(cyto) orientation, via an NH2-terminal signal-anchor sequence. The signal-anchor is comprised of two domains: an NH2-terminal hydrophilic region which is positively charged (amino acids 1-10), followed by the predicted transmembrane segment (amino acids 11-29). Substitution of the NH2-terminal hydrophilic domain with a matrix-targeting signal caused the signal-anchor to adopt the reverse orientation in the membrane (N(cyto)-C(in)). This substitution resulted in an increase in the net positive charge of the hydrophilic region, from +4 to +8. In contrast to the endoplasmic reticulum and the bacterial inner membrane, where the net positive charge is an important determinant in conferring protein topology in the lipid bilayer, we show here that the reversal of the Mas70p signal-anchor was not due to differences in the number and positions of basic amino acids in the hydrophilic domain. However, a reduction in the hydrophobic moment of predicted amphiphilic helices containing an arginine, obtained by converting the apolar amino acids flanking the arginine to polar residues, caused the otherwise N(cyto)-C(in) signal-anchor to re-adopt the original N(in)-C(cyto) orientation of Mas70p. The reduced hydrophobic moment at the NH2-terminus significantly reduced the ability of this domain to bind to synthetic liposomes whose lipid composition reflected that of the outer membrane. These results identify amphiphilicity as an important determinant in causing retention of the NH2-terminus of a mitochondrial signal-anchor on the cytosolic side of the outer membrane. In addition to potential interactions between this domain and cytosolic-exposed components of the import machinery, this retention may result as well from interaction of the NH2-terminus with the surrounding membrane surface.