* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download S C T
Protein (nutrient) wikipedia , lookup
Lipid bilayer wikipedia , lookup
Mechanosensitive channels wikipedia , lookup
Cytokinesis wikipedia , lookup
Protein moonlighting wikipedia , lookup
Membrane potential wikipedia , lookup
Protein phosphorylation wikipedia , lookup
Model lipid bilayer wikipedia , lookup
P-type ATPase wikipedia , lookup
Theories of general anaesthetic action wikipedia , lookup
Nuclear magnetic resonance spectroscopy of proteins wikipedia , lookup
Protein domain wikipedia , lookup
Magnesium transporter wikipedia , lookup
Intrinsically disordered proteins wikipedia , lookup
G protein–coupled receptor wikipedia , lookup
Signal transduction wikipedia , lookup
Protein structure prediction wikipedia , lookup
SNARE (protein) wikipedia , lookup
List of types of proteins wikipedia , lookup
Cell membrane wikipedia , lookup
Endomembrane system wikipedia , lookup
From the Department of Biochemistry & Biophysics Stockholm University, Stockholm, Sweden STUDIES ON THE CONFORMATION OF TRANSMEMBRANE POLYPEPTIDES IN MEMBRANE PROTEINS MARIKA CASSEL Stockholm 2005 2 Marika Cassel Intellecta Docusys AB, Stockholm 2005 ISBN 91-7155-175-1 3 Till Tobias 4 Marika Cassel Abstract The major aim of the studies that this thesis is based on has been to better define the topological determinants of the formation of so-called helical hairpins during membrane protein assembly in the ER membrane. The helical hairpin is a basic folding unit in membrane proteins. It is composed of two closely spaced transmembrane helices with a short connecting loop and it is believed to be inserted into the membrane as one compact unit. It is becoming increasingly clear that the helical hairpin is a very common structural element in membrane proteins and a detailed understanding of its properties is of central importance. We demonstrate that the efficiency of formation of helical hairpins depends both on the overall length of the hydrophobic segment, on the amino acids flanking the transmembrane segment, and on the identity of the central, potentially turn-forming residues. We also show that interhelical hydrogen bonds between pairs of Asn or Asp residues can induce helical hairpin formation. A detailed topology mapping is also reported for the Escherichia coli inner membrane chloride channel YadQ, a protein for which the X-ray structure is known. Our results provide a critical test of the reporter fusion approach and offer new insights into the YadQ folding pathway. In summary, the results present in this thesis have increased our understanding of the determinants of membrane protein topology and structure. Furthermore, the information obtained can be used to improve current models for predictions of membrane protein topology. 5 Main References This thesis is based on the following original papers that are reprinted in the second part of the thesis. In the text, these papers will be referred to by their roman numerals. Paper I: Monné, M. Hermansson, M., von Heijne, G. (1999) A turn propensity scale for transmembrane helices. Journal of Molecular Biology 288:141-145 Paper II: Sääf, A. Hermansson, M., von Heijne, G. (2000) Formation of cytoplasmic turns between two closely spaced transmembrane helices during membrane protein integration into the ER membrane. Journal of Molecular Biology 301:191-197 Paper III: Hermansson, M., Monné, M., von Heijne, G. (2001) Formation of helical hairpins during membrane protein integration into the endoplasmic reticulum membrane. Role of the N and C-terminal flanking regions. Journal of Molecular Biology 314:1171-1179 Paper IV: Hermansson, M., von Heijne, G. (2003) Inter-helical hydrogen bond formation during membrane protein integration into the ER membrane. Journal of Molecular Biology 334:803-809 Paper V: Cassel, M., von Heijne, G. (2005) Confronting fusion protein-based membrane protein topology mapping with reality: the Escherichia coli ClC chloride channel. Manuscript The published articles have been reprinted with permission from Elsevier Ltd (Papers I-IV) 6 Marika Cassel Table of contents Abstract…………………………………………………………………………………………………………………………… 4 Main References………………………………………………………………………………………………………… 5 Abbreviations……………………………………………………………………………………………………………… 8 Introduction………………………………………………………………………………………………………………….. 9 Biological membranes……………………………………………………………………………………………... 9 Membrane proteins………………………………………………………………………………………………….. 10 Peripheral membrane proteins…………………………………………………………………………… 10 Lipid-linked membrane proteins………………………………………………………………………... 10 Facultative membrane proteins…………………………………………………………………………..11 Integral membrane proteins………………………………………………………………………………. 11 Membrane protein structure…………………………………………………………………………………... 12 α-helical membrane proteins……………………………………………………………………………... 13 β-barrel membrane proteins ……………………………………………………………………………… 14 Experimental structure determination of membrane proteins…………………………. 14 Determination of 3D-structure…………………………………………………………………………. 14 Studies of dynamic properties……………………………………………………………………………. 15 Computer simulations……………………………………………………………………………………….. 16 Membrane protein topology…………………………………………………………………………………... 16 Experimental topology determination……………………………………………………………….. 16 Protease-protection assay……………………………………………………………………... 17 Cystein scanning……………………...……………………………………………………………...17 Gene fusion analysis……………………………………………………………………………… 17 Mapping of loop regions………………………………………………………………………. 18 Computerized topology determination……………………………………………………………… 19 Topogenic determinants……………………………………………………………………………………. 20 Topogenic signals………………………………………………………………………………….. 20 Single-spanning membrane proteins…………………………………………………. 20 Charged residues…………………………………………………………………………………… 21 Multi-spanning membrane proteins…………………………………………………… 22 Biosynthesis and targeting to the ER membrane………………………………………………… 23 The protein-conducting translocon………………………………………………………………………. 24 Translocation……………………………………………………………………………………………………... 26 Translocon-mediated membrane insertion………………………………………………………… 26 Folding of extramembraneous domains………………………………………………………………..27 Helix-helix packing in the membrane…………………………………………………………………... 28 Comments on Methodology………………………………………………………………………………….. 29 Model system used to study helical hairpin formation………………………………………29 Model protein – Leader peptidase……………………………………………………………………… 29 Manipulation of the model protein……………………………………………………………………. 30 Glycosylation mapping……………………………………………………………………………………… 30 Alkaline extraction……………………………………………………………………………………………. 31 7 Alkaline phosphatase as a topology reporter protein…………………..……………………. 31 Green fluorescent protein as a topology reporter protein…………...…………………….. 31 Results and Discussion……………………………………………………………………………………………. 32 Paper I……………………………………………………………………………………………………………………….. 32 Paper II………………………………………………………………………………………………………………………. 33 Paper III…………………………………………………………………………………………………………………….. 34 Paper IV…………………………………………………………………………………………………………………….. 35 Paper V……………………………………………………………………………………………………………………… 36 Concluding Remarks………………………………………………………………………………………………… 38 Acknowledgements…………………………………………………………………………………………………… 39 References……………………………………………………………………………………………………………………… 41 8 Marika Cassel Abbreviations AFM ATP CD cryo-EM EM ER FA FTIR GFP HC IF IMP kDa MC MD MP NMR OST PMP PKA rSA SA SP SPase ST SPC SRP TM TRAM TRAP TROSY Atomic force microscopy Adenosine triphosphate Circular dichroism Cryo-electron microscopy Electron microscopy Endoplasmic reticulum Fatty acid Fourier transform infrared Green fluorescent protein Hydrocarbon core Interfacial regions Integral membrane protein kilo Dalton Monte Carlo Molecular dynamics Membrane protein Nuclear magnetic resonance Oligosaccharyl transferase Peripheral membrane protein Protein kinase A Reverse signal anchor Signal anchor Signal peptide Signal peptidase complex Stop-transfer Signal peptidase complex Signal recognition particle Transmembrane Translocating chain associated membrane protein Translocon-associated protein complex Transverse optimized NMR spectroscopy 9 Introduction Biological membranes All living cells are enclosed by a membrane (the plasma membrane) which separates the cell from the surrounding environment. Biological membranes also form boundaries within the eukaryotic cell where they divide the internal space in the cell into distinct compartments, so-called organelles. Each organelle, such as endoplasmic reticulum (ER), golgi network, lysosomes, peroxisomes, nucleus, mitochondria, chloroplasts, etc. contains a specific complement of proteins and other molecules and has a unique role in the cell. Besides acting as selectively permeable barriers, membranes are involved in thousands of essential and diverse functions in the cell, such as transport of nutrients, adenosine triphosphate (ATP) synthesis, ion conduction, signal transduction, cell-cell interactions, respiration, development, vision, hearing, fertilization and many more. Cellular membranes are highly organized structures, built from two basic components, lipids and membrane proteins. The fluid mosaic model describes the organisation and features common to all membranes and predicts free rotational and lateral diffusion of proteins and lipids within the plane of the membrane, resulting in their random distribution (Singer and Nicolson, 1972). However, to account for lateral heterogeneities in membranes, which is a result of proteinprotein, protein–lipid and lipid-lipid interactions, this model has recently been refined (Vereb et al., 2003). Membrane lipids are divided into three classes: glycerophospholipids, sphingolipids and cholesterol, where glycerophospholipids are the main lipid constituent of most biomembranes. Lipids are amphiphilic molecules, i.e. they contain both hydrophobic and hydrophilic regions. In glycerophospholipids and sphingolipids, the hydrophobic part is composed of the fatty acid (FA) hydrocarbon chains, which usually are unbranched and have an even number of carbon atoms, usually 16 or 18. The FA chains are either saturated with hydrogen atoms or have one or more unsaturated double bonds. The polar headgroups of the lipids have different charges and are hydrophilic. In cholesterol the whole molecule except the hydroxyl group on carbon-3 is hydrophobic. Membrane lipids interact noncovalently with each other and are packed side by side forming the “lipid bilayer”, in which the hydrophobic FA chains are oriented towards the interior of the membrane, forming a 30Å wide hydrophobic hydrocarbon core (HC). The hydrophilic lipid headgroups are on the surfaces of the membrane core, forming two polar interfacial regions that are 10-15Å wide. The two layers of lipids in the bilayer are termed the inner and outer leaflet. Embedded in the lipid bilayer are membrane proteins of various shapes and 10 Marika Cassel traits. These proteins are either integrated into the membrane - integral membrane proteins - or peripherally associated with the membrane - peripheral membrane proteins. Many membrane proteins and lipids contain covalently attached oligosaccharides. The weak noncovalent interactions between neighbouring molecules give rise to the membrane fluidity, i.e. the lipids can move easily within the plane of the membrane. The fluidity of the membrane is affected by the length of the FA chains and degree of unsaturation in the FA chains. Maintenance of the bilayer fluidity appears to be essential for normal cell function. The membrane surrounding each type of organelle has organelle-specific lipid and protein composition, which contributes to their unique properties. For example, the ER membrane is composed mainly of the glycerophospholipid phosphatidylcholin and to a lesser extent of phosphatidylethanolamine, phophatidylserin and phosphatidylinositol, while levels of cholesterol and sphingolipids are low (van Meer, 1989). The relative proportions of membrane proteins and lipids vary with the type of membrane. Lipids and membrane proteins are asymmetrically distributed between the inner and outer leaflet and can also be non-randomly distributed within a single leaflet, which is seen in membrane microdomains termed lipid rafts. (Simons and Ikonen, 1997; Simons and Ikonen, 2000). Lipid rafts appear to be enriched in sphingolipids and cholesterol that form a highly ordered structure distinct from the surrounding sea of lipids. In addition, specific types of membrane proteins implicated in signal transduction, membrane trafficking within cells and membrane protein sorting, have been shown to be localized to lipid rafts (Laude and Prior, 2004). Membrane proteins Peripheral membrane proteins Membrane proteins are classified according to how tightly they are associated to the membrane. Peripheral membrane proteins are loosely associated to the membrane by hydrogen bonding and electrostatic interactions and do not interact strongly with the hydrophobic interior of the bilayer. Instead they interact with hydrophilic domains of integral membrane proteins in the membrane and with the polar headgroups of membrane lipids. Some peripheral membrane proteins serve as regulators of membrane spanning enzymes, while others interact with the cytoskeleton and maintain or alter the cellular shape. Structurally, peripheral membrane proteins are similar to globular proteins. Lipid-linked membrane proteins Another class of membrane proteins is the lipid-linked membrane proteins. These proteins have one or more covalently attached lipid that anchors the protein to 11 the lipid bilayer. The anchoring lipids may be of several types: long-chain fatty acids, isoprenoids or sterols. Lipid-protein assemblies containing these three types of lipids are found in the inner leaflet of the plasma membrane, while lipidprotein assemblies containing glycosylated derivatives of phosphatidylinositol are found only in the outer leaflet of the plasma membrane. Facultative membrane proteins In some microorganisms intricate mechanisms have evolved where secreted proteins or insertion peptides, that are soluble in the aqueous phase (so called pore-forming toxins) partition spontaneously into the host cell membrane where they fold and/or assemble and form integral membrane protein pores that promote translocation of toxins into the cytoplasm of the host cell (Krantz et al., 2005). The membrane spanning part of these pores can either have α-helix conformation or form β-barrels (Gouaux, 1997; Song et al., 1996; Stroud, 1995). Integral membrane proteins Approximately 20-30% of all genes in the fully sequenced genome are predicted to code for integral membrane proteins (von Heijne, 1996). They are found in all cellular membranes and carry out a wide range of central functions in the cell. Some act as signal receptors that transduce signals across the membrane, while others catalyze specific transport of nutrients, metabolites and ions across the membrane barriers. In addition, some integral membrane proteins couple the flow of electrons to the synthesis of ATP and specific transmembrane (TM) proteins in the plasma membrane are involved in cell-cell interactions, where they serve as highly active mediators between the cell and its environment or the interior of an organelle and the cytosol. Many membrane proteins play important roles in tumor development and in the immune response. Since integral membrane proteins fulfil many critical functions in the cell, they are of major medical importance and thus provide interesting targets for the pharmaceutical industry. Integral membrane proteins are embedded in the membrane bilayer, in which they fold and function. They are composed of one or more hydrophobic membrane spanning segments, linked by hydrophilic regions of the polypeptide chain, which are exposed either on the extracellular or cytosolic side of the membrane. The TM segments consist mainly of amino acid residues with hydrophobic side-chains that protrude outwards and interact with the FA chains in the lipid bilayer. These hydrophobic protein-lipid interactions tightly anchor the integral membrane proteins to the hydrophobic core of the membrane. The hydrophobic environment of the lipid bilayer strongly limits the range of possible structures for the membrane spanning part of integral membrane proteins. When a polypeptide chain is surrounded by lipids, the polar peptide bonds in the backbone are driven to form hydrogen bonds with one another since there are no water molecules in the lipid environment. Hydrogen bonding 12 Marika Cassel between peptide bonds is maximized if the polypeptide chain forms a regular αhelix as it crosses the lipid bilayer. Thus, the polypeptide chain of most TM proteins is thought to cross the lipid bilayer in an α-helical conformation (Deisenhofer et al., 1985; Henderson et al., 1990; Henderson and Unwin, 1975). An alternative way for the peptide bonds in the lipid bilayer to satisfy the hydrogenbonding requirements is for multiple strands of polypeptide chains to be arranged as a β-sheet, in the form of a closed barrel. Not much is known how βbarrel proteins fold, except that a good part of the structure must be present before the hydrophobic external surface is well defined. Thus, the β-barrel probably forms before or simultaneously with membrane insertion (Tamm et al., 2001). A number of studies suggest that helix bundle proteins fold according to a twostep mechanism. In a first step, individually stable helices are formed and inserted into the membrane. Subsequently in a second step, these helices interact, giving rise to tertiary and quaternary structures and thus to a functional TM structure (Popot and Engelman, 1990). In a membrane the hydrophobic effect contributes only to the formation and insertion of TM helices (step 1). The interactions that drive bundle formation (step 2) are not clearly understood. The formation of salt bridges between oppositely charged residues may be one factor and the burial of polar amino acids in the protein interior may be another (Bowie, 2000; Zhou et al., 2000). Tight packing between TM helices is important for protein stability. Theoretical calculations indicate that van der Waals interactions between apolar TM helices may be enough to drive the formation of helix bundles in a lipid environment (Bowie, 1997; von Heijne, 1994). Membrane protein structure Although many thousands of membrane protein sequences are known, progress in the structural analysis of these proteins has been hampered by difficulties in their purification and crystallization. A comprehensive understanding of fundamental membrane-associated biological processes such as energy conversion, transport and signal transduction cannot be attained unless the highresolution structures of the membrane proteins involved are known. To date, only about 100 high-resolution structures of membrane proteins have been solved [http://blanco.biomol.uci.edu/Membrane_Proteins_xtal.html], compared to about 30,000 globular proteins. Among these around 100 known structures, only two basic structural features have been found: the α-helical conformation and the β-barrel fold (Figure 1) (Cowan and Rosenbusch, 1994). Integral membrane proteins are usually divided into two major classes, each one containing one of these two structural features (von Heijne, 1994). 13 α-helical membrane proteins The membrane proteins that belong to the so-called helix bundle class, have membrane domains composed of two or more TM α-helices usually consisting of a stretch of about 15-30 amino acids with a central hydrophobic domain flanked by more polar residues (Wallin et al., 1997). The loops connecting two TM αhelices can be as short as three residues (Bowie, 1997). The hydrophobic domain of TM α-helices is composed of mainly, but not entirely, of hydrophobic amino acid residues (Ile, Leu, Val, Ala, Phe and Gly ). The TMs may include a limited number of polar or potentially charged groups; Cys, Thr and Ser residues often form H-bonds bonds to the main chain. Aromatic side chains, particularly Tyr and Trp, are often located near the boundary between the hydrophilic and hydrophobic regions of the lipid bilayer, both in helical and βbarrel membrane proteins found within the lipid head group region and can serve as membrane anchors (Schiffer et al., 1992). Pro and Gly are found more commonly in TM α-helices than in soluble protein helices. Pro is sometimes found near the middle of TM α-helices in multispanning membrane proteins (Jones et al., 1994) where it can induce a bend in the helix. Such Pro residues, which tend to be highly conserved, result in a kink and disrupt several helical hydrogen bonds (von Heijne, 1991). Figure 1 . Two membrane proteins containing different basic structural features. A helix bundle protein, bacteriorhodopsin (Belrahli et al, 1999), (left) and a β-barrel protein, maltoporin (Schirmer et al., 1995), (right). The TM α-helices are oriented more or less perpendicular to the plane of the membrane with somewhat variable tilts. Helix bundle membrane proteins are found in almost all membranes except in the bacterial outer membrane. One example of a helix-bundle protein is bacteriorhodopsin, a light-driven proton pump from Halobacterium halobium. It contains seven TM α-helices and was the first integral membrane protein to be shown to consist of a helix bundle. Electron microscopy of two-dimensional crystals of this protein was used to derive the structure (Henderson and Unwin, 1975). Such 2D-crystals or ordered sheets of membrane proteins are often easier to obtain than 3D-crystals, and in some cases 14 Marika Cassel they occur in vivo, in the plane of the membrane (Brändén and Tooze, 1991). Bacteriorhodopsin is one of the best structurally and functionally characterized integral membrane proteins. β-barrel membrane proteins The other class of integral membrane protein is the so-called β-barrel class. In proteins containing this structural feature, the membrane-embedded domain is composed of a large, antiparallel β-sheet barrel. These proteins are found in the outer membrane of Gram-negative bacteria and in the mitochondrial outer membrane (Henderson et al., 1990; Henderson and Unwin, 1975). An example of typical β-barrel proteins in bacteria is the porins. The porins form rather large pores through the membrane, through which specific small substrates can diffuse passively (Buchanan, 1999; Schulz, 2000). In these structures, every second amino acid residue in each strand is hydrophobic and faces out towards the lipid environment, while the surface facing the channel is usually more hydrophilic (Cowan et al., 1992; Weiss et al., 1991). Experimental structure determination of membrane proteins Integral membrane proteins have not been studied as thoroughly as globular proteins, for simple technical reasons. Since membrane proteins have both hydrophobic and hydrophilic regions on their surfaces, they are not soluble in aqueous solution and denature in organic solvents. Because of these amphipatic properties, they are tricky to handle biochemically. In addition, membrane proteins are hard to overexpress and they seldom yield high-quality 2D or 3D crystals. Determination of 3D structures There are three techniques that can be used for determining the 3D structure of protein molecules; X-ray crystallography and nuclear magnetic resonance (NMR), which have yielded by far the largest numbers of protein structures to date, but also electron crystallography (von Heijne, 1996). Each technique has its strengths and weaknesses but they all require large amounts of very pure protein, which often is the bottleneck for membrane proteins because of the difficulties with overexpression, purification and production of 3D-crystals. However, the difficulties in crystallizing membrane proteins are beginning to be overcome and X-ray structures are now available for at least one example of most kinds of transporters, sensors and G-protein coupled receptors. Several different NMR techniques are used for structural studies on membrane proteins. Solution NMR can be used in order to obtain information about membrane proteins in detergent micelles that closely mimic the natural membrane environment. This technique is carried out on macromolecules in solution and crystals are not needed. Solution NMR is generally considered to be 15 restricted to proteins up to 30-50kDa, but recently significantly larger proteins have been studied (Fiaux et al 2000). This technique is preferably used on proteins with β-barrel fold. Recently, larger β-barrel outer membrane proteins such as OmpX (Fernandez et al., 2001; Fernandez et al., 2002), OmpA (Arora et al., 2001; MacKenzie et al., 1997; Tamm et al., 2003) and PagP (Hwang et al., 2002) have been studied using transverse optimized NMR spectroscopy (TROSY) (Pervushin et al., 1997). Solid-state NMR can be applied to membrane proteins in native membranes or in phospholipid bilayers and can reveal the peptide backbone structure and intramolecular distances (Opella, 1997). Conventionally, methods for solubilization, purification and subsequent crystallization of membrane proteins have employed detergents. However, a novel non-detergent concept has been developed recently, using highly viscous, optically transparent, lipidic cubic phases. The materials used are composed of very high amounts of hydrated lipids (Landau and Rosenbusch, 1996) and several integral membrane proteins, from one single mesophase system based on monoolein, have been crystallized (Chiu et al., 2000). This technique provides new possibilities to generate well-ordered 3D crystals of membrane proteins, but it is so far only applicable to coloured membrane proteins, since the formation of uncoloured membrane protein crystals cannot easily be detected in the lipidic cubic phase. Cryo-electron microscopy (cryo-EM) and atomic force microscopy (AFM) can also be used to obtain structural information. These techniques have been used to determine structures of bacteriorhodopsin (Subramaniam and Henderson, 1999), aquaporins (Murata et al., 2000; Scheuring et al., 2002) and Na+, K+-ATPase with resolutions in the order of 3.2-9.2Å. Studies of dynamic properties Although structurally ordered, many membrane proteins are flexible and dynamic molecules and must go through conformational changes in order to perform their functions. Our knowledge of these changes is extremely limited, because of the small number of known membrane protein structures available today. In addition, these high-resolution structures only provide snapshots of single conformations and reveal only one aspect of the behaviour, unless structures of all the intermediates of the conformational rearrangements are available. This is the case of bacteriorhodopsin (Lanyi, 2004) where structural models for all the intermediates in the photochemical cycle are available. This provides a clear picture of the conformational rearrangements resulting from photoisomerization of the retinal chromophore. To obtain information about the dynamic properties of membrane proteins, including conformational changes, protein folding and interactions with other molecules, several methods can be utilized. These methods include for example, circular dichroism (CD) spectroscopy, polarized Fourier transform infrared (FTIR) spectroscopy and fluorescence emission spectroscopy (Hunt et al., 1993). NMR can also illuminate 16 Marika Cassel the dynamic side of membrane protein structure. Conformational dynamics in Ptype ATPases have also been studied under physiological conditions using voltage clamp fluorometry. Computer simulations Computer simulations can complement biophysical experimental approaches to examine membrane protein structures, by extrapolating from static high– resolution structures. In this way, the dynamics of TM peptides can be modelled. Monte Carlo (MC) and molecular dynamics (MD) simulations are used to study membrane proteins structure, dynamics and energetics. Folding kinetics of the first two TM α-helices of bacteriorhodopsin have been studied using computer simulations, which show that the helices form rapidly. However, before the right packing in the membrane is found, a long period of consolidation follows (Orlandini et al., 2000). Membrane protein topology For membrane proteins with unknown three-dimensional structure, it is necessary to use alternative approaches to obtain structural information. By using computerized sequence-based prediction methods or different experimental approaches, it is possible to produce quite reliable topology models for membrane proteins. With such models, it is possible to describe the secondary structure of the membrane embedded domain, the number of peptide segments that traverse the membrane and their respective orientation in the membrane. For most membrane proteins, as a result of the reasons mentioned above, the topology model is the only structural information available. Although it is not as informative as a three-dimensional structure, it still provides important information. This was shown in a recent study where a global topology analysis of the Escherichia coli inner membrane proteome made it possible to categorize proteins according to their different topologies and functions (Daley et al., 2005). Only a small number of membrane proteins are known to adopt two or more distinct topological orientations (Levy, 1996; Ma and Chang, 2004; UbarretxenaBelandia et al., 2003; Ubarretxena-Belandia and Tate, 2004). Examples of homologous proteins with opposite orientations in the membrane have been reported (Sääf et al., 1999). Experimental topology determination Several different biochemical and molecular biology techniques are available for topology determination. In these methods, enzymes or other reagents are involved, which are active on, or have access to, only one side of the membrane (Traxler et al., 1993). Here, some of the most frequently used methods available for topology determination are summarized. 17 Protease-protection assay The protease-protection assay can be used to confirm the topology of a membrane protein and to get information about the localization of membrane protein domains. The proteases trypsin and proteinase K can be added to microsomes or sphaeroplasts, containing the membrane protein under study. These enzymes degrade the exposed domains, while unexposed cytoplasmic and luminal domains remain undegraded. The remaining, protease-protected part of the membrane protein can be detected by SDS-PAGE . Cysteine scanning The cysteine scanning method also provides an excellent tool for determining the localization of intra- and extracellular domains of membrane proteins. In this approach, the amino acid cysteine with its reactive sulfhydryl group is introduced at a specific site of choice, after all naturally occuring cysteine residues have been removed from the target protein (Möller and Rydström, 1999). The sulfhydryl group can be modified by special reagents with different chemical properties. A shift in the appearent molecular weight of the modified protein can be detected by SDS-PAGE analysis, but alternative assays can also be used for detecting the cystein residue exposure to the reagent. The cysteine scanning method has been used in combination with biophysical approaches to study e.g., helix packing in LacY in Escherichia coli (Kaback et al., 1997). Gene fusion analysis The most widely used method for membrane protein topology determination is gene fusion analysis (Ikeda et al., 2003; Möller et al., 2000). In this approach, extramembraneous domains of the membrane protein under study are fused to a reporter protein that is active on only one side of the membrane. Most reporter proteins are enzymes that catalyse a reaction, which can easily be detected by e.g. spectrophotometric measurements. From the reporter activity it can be determined whether the reporter-fused domain of the membrane protein is translocated to the periplasm or if it is retained in the cytoplasm, and a topology model can be produced. Usually, 5´-parts of the gene coding for the membrane protein under study is fused to a reporter gene in a cloning vector (Boyd et al., 1993). The reporter should only be fused to extramembraneous parts of the membrane protein, to avoid ambiguous results. To derive a topology model of a polytopic membrane protein, the reporter has to be fused to each loop region. This is achieved by successively truncating the full-length gene at the 3´end and fusing it to the reporter gene. To produce a reliable topology model it is appropriate to choose two reporter proteins which have opposite activity profiles and thus can give complementary results. As seen in paper V, the topological information obtained from the green 18 Marika Cassel fluorescent protein (GFP) assays, complements the results from the alkaline phosphatase (PhoA) assays. The classic example of a reporter protein is PhoA, which has been widely used for identifying periplasmic loops of membrane proteins expressed in Escherichia coli (Manoil et al., 1990). PhoA has to form two disulfide bonds in order to become active and these two bonds can only be formed in the oxidizing environment of the periplasm (Derman and Beckwith, 1991). When a particular substrate is added to a bacterial culture expressing the fusion construct, a chromogenic product is formed if the PhoA domain is in the periplasm, and the PhoA activity can be quantified spectrophotometrically (Manoil, 1991; Manoil and Beckwith, 1986). When PhoA is located in the cytoplasm it is unfolded and therefore inactive. PhoA is highly resistant to proteases when located in the periplasm, but not when located in the cytoplasm (Roberts and Chlebowski, 1984). This feature can be used to further confirm the topology. Another widely used periplasmic reporter protein is β-lactamase (Broome-Smith and Spratt, 1986; Broome-Smith et al., 1990; Prinz and Beckwith, 1994; Tadayyon et al., 1992). This bacterial enzyme hydrolyses and inactivates β-lactam antibiotics such as ampicillin. β-lactamase has a high enzymatic activity when in the periplasm. Growth of bacterial colonies on an agar plate containing ampicillin is thus indicative of a periplasmic location of the reporter protein. β-galactosidase (LacZ) (Froshauer et al., 1988) is another example of a bacterial enzyme that can be used as a reporter. These fusions are only active when the LacZ moiety is present in the cytoplasm (Froshauer et al., 1988). GFP is a thoroughly tested cytoplasmic reporter that can be fused to the Cterminal end of the protein of interest (Drew et al., 2002; Rapp et al., 2004). GFP forms a chromophore when correctly folded but can only fold in a proper manner when on the cytoplasmic side of the inner membrane in Escherichia coli (Feilmeier et al., 2000). Thus, if GFP fluorescence emission can be detected, a cytoplasmic location of the reporter is indicated. When targeted to the periplasm by a Sec-type signal peptide (Drew et al., 2002), GFP does not fold properly and does not fluoresce (Drew et al., 2001; Feilmeier et al., 2000). Mapping of loop regions Mapping of loop regions in a membrane protein can be achieved by introducing consensus sequences for protein kinase A (PKA) (Goder et al., 2000). This cytosolic enzyme only phosphorylates extramembraneous protein domains localized in the cytoplasm, which contain the consensus sequences. By introducing N-glycosylation sites in membrane proteins, useful topological information can be obtained. This will be described in more detail under Comments of methodology. 19 Computerized topology determination An alternative way to gain an idea of a protein’s topology is to use computer programs that predict the topology by analyzing the amino acid sequence of the membrane protein (Claros and von Heijne, 1994; Tusnady and Simon, 1998). All α-helical membrane proteins have some general features like (i) the presence of 20-30 amino acids long TM α-helices with a high overall hydrophobicity and (ii) “the positive-inside rule”(von Heijne, 1986; von Heijne, 1992). These features are taken into consideration when designing the prediction algorithm and common to almost all prediction methods is that they incorporate both hydrophobicity analysis and the positive-inside rule (von Heijne, 1992). The topology of α-helical membrane proteins can be predicted with relatively high accuracy from the amino acid sequence alone. It is more difficult to predict the topology of β-barrel membrane proteins since the membrane spanning β-strands are short and thus hard to detect in the sequence (von Heijne, 1994; von Heijne, 1996). There are several different algorithms available for membrane protein topology predictions. Five of the most widely used topology prediction methods are: TMHMM (Krogh et al., 2001; Sonnhammer et al., 1998), HMMTOP (Tusnady and Simon, 1998; Tusnady and Simon, 2001), MEMSAT (Jones et al., 1994), TOPPRED (Claros and von Heijne, 1994; von Heijne, 1992) and PHD (Rost et al., 1996). All of these methods are designed to identify potential TM α-helices and to predict the overall in-out topology of the extramembraneous parts of the membrane protein. TMHMM and HMMTOP both use a Hidden Markov model formalism to describe the architecture of an integral membrane protein. These approaches generally result in 65-70% correct predictions and are the best-performing methods to date (Ikeda et al., 2001; Möller et al., 2001). MEMSAT uses dynamic programming to optimally thread a polypeptide chain through a set of topology models. PHD is based on a neural network predictor. TOPPRED identifies “certain” and “putative” TM α-helices from a standard hydrophobicity plot and then the most likely topology is suggested based on the positive inside rule (von Heijne, 1986). There are numerous hydrophobicity scales available which can be used for topology prediction (Chen et al., 2002; Cornette et al., 1987). None of the topology prediction methods are perfect. For example, when a TM helix in the analyzed protein contains charged residues, these domains are sometimes identified as soluble loop regions. Another potential inaccuracy is when two closely spaced TM helices, connected by a short loop, sometimes are predicted to be one TM helix. By incorporating experimental topology information into the prediction algorithms, the predictions are significantly improved (Melén et al., 2003). A combination of several different topology prediction algorithms can be used in parallel and when several of those agree on the prediction, highly reliable topology models can be produced. Experimental 20 Marika Cassel efforts can be focused on the less reliably predicted parts of the membrane protein to further substantiate the topology model. Recently it has been shown that topology prediction can be improved considerably, by constraining the topology prediction algorithm with an experimentally determined reference point (Daley et al., 2005). New topology prediction methods are produced at a steady rate. However, the overall performance of these methods appears to improve only slowly (Chen et al., 2002; Chen and Rost, 2002). In addition to topology prediction, there are two other aspects concerning membrane proteins that prediction methods can be applied on: (i) distinguishing membrane proteins from soluble proteins, (ii) modelling of 3D membrane proteins structure. Topogenic determinants Membrane proteins can adopt a variety of TM orientations and several factors are involved in determining the topology. Some of these are quite well characterized, but still many details of how topology is controlled remain to be elucidated. Topogenic signals It is well known that integral membrane proteins contain special sequences, socalled signal sequences or that target them to the ER membrane, where they are cotranslationally integrated (Walter and Johnson, 1994). In addition to their targeting function, these signals also play a role in membrane protein topogenesis and can contribute to the final structure and function of a membrane protein. There are four different, generally accepted kinds of signal sequences. These include signal peptides (SP), signal-anchor (SA) sequences, reverse signal-anchor (rSA) sequences and stop-transfer (ST) sequences (von Heijne, 1996 ; von Heijne and Gavel, 1988). The separate signal sequences all consist of three distinct regions: an NH2terminal hydrophilic domain (N-domain), a central hydrophobic region (Hregion) and a COOH-terminal polar region (C-domain). The ER targeting function of the signal sequences is primarily determined by the H-region and the orientation in the membrane is affected by the flanking regions (Nilsson and von Heijne, 1993). The topology that a membrane protein finally adopts is partially dependent on whether its signal sequence is cleaved off or retained as a TM anchor during membrane protein biosynthesis. Single-spanning membrane proteins Single-spanning membrane proteins containing these signal sequences, can be classified according to their orientation in the membrane (Figure 2) (von Heijne, 1996). Type I membrane proteins are oriented in the membrane with an exoplasmic N-terminus, while the C-terminus is in the cytosol. Type I membrane proteins are targeted to the ER membrane by a cleavable N-terminal signal peptide (a hydrophobic stretch of 7-15 mainly apolar residues) (Blobel, 1983; von 21 Heijne, 1990) and are anchored in the membrane by a subsequent ST-sequence. This ST-sequence is a TM spanning domain composed of around 20 hydrophobic residues that halt a continued translocation of the polypeptide chain. The signal peptide is cleaved off by signal peptidase on the lumenal side of the membrane. Membrane proteins of type II have a cytoplasmic N-terminus and an exoplasmic C-terminus. A signal-anchor sequence composed of 18-25 mainly apolar residues anchors the protein in the membrane and induces the translocation of the Cterminal end across the membrane. Type III membrane proteins are inserted into the lipid bilayer with an exoplasmic N- and a cytoplasmic C-terminus. This type of membrane proteins are anchored in the membrane by a rSA-sequence that induces translocation of the N-terminal end across the membrane. All these three types of membrane proteins are targeted to the ER membrane in an signal recognition particle (SRP) dependent way and are cotranslationally inserted into the membrane by the Sec61 translocon (High and Dobberstein, 1991). Membrane proteins of type IV are anchored to the membrane by a C-terminal hydrophobic sequence and have their N-terminus in the cytoplasm while the C-terminus is exoplasmic. These proteins are posttranslationally inserted into the membrane, using an ATP-requiring mechanism that is SRP- and Sec61-independent (Kutay et al., 1995; Kutay et al., 1993; Whitley et al., 1996). Type I Type II Type III C N Type IV N C exoplasmic N C N cytoplasmic N C Figure 2. Classification of single spanning membrane proteins (adapted from Goder et al, 2001) Charged residues The distribution of charged residues in the sequence flanking a given TM domain is a major factor influencing its orientation in the lipid bilayer (High and Dobberstein, 1992). Positively charged amino acid residues are particularly important. Statistical studies of integral membrane proteins in bacteria have shown that positively charged amino acids are enriched in loop regions on the cytoplasmic side of the membrane (von Heijne, 1986; von Heijne, 1989). This is 22 Marika Cassel known as the “positive-inside rule” i.e., parts of the protein rich in lysine and/or arginine tend to remain non-translocated. Through further statistical studies and experiments, it has been established that the number and location of positively charged residues are crucial for membrane protein topology (Andersson et al., 1992; Boyd and Beckwith, 1989; Gafvelin and von Heijne, 1994; Nilsson and von Heijne, 1990; von Heijne, 1994) and that the “positive inside rule” is a general feature among helix-bundle proteins in a wide range of organisms and organelles, in all three kingdoms of life (Nilsson et al., 2005; Wallin and von Heijne, 1998) It has been shown that negatively charged residues are much less potent than positively charged ones, and only significantly affect the topology if present in high numbers (Nilsson and von Heijne, 1990). The importance of negatively charged residues has been underscored though, as it was observed that they play an active role in the membrane insertion of the small Pf3 coat protein from Pseudomonas aeruginosa (Kiefer et al., 1997). The membrane protein topology can also be influenced by the length and hydrophobicity of a membrane-spanning domain (Wahlberg and Spiess, 1997). It has further been shown that the folding state of the N-terminal domain (Denzer et al., 1995) as well as glycosylation can contribute to the final topology. Multi-spanning membrane proteins Multi-spanning membrane proteins can be classified in a similar way as singlespanning proteins, based on their orientation in the membrane. The first hydrophobic segment in multi-spanning membrane proteins acts as one of the four kinds of signal sequences mentioned earlier (SP, SA sequence, rSA sequence or ST sequence) and targets the nascent polypeptide chain to the ER membrane where it initiates translocation and insertion. In some cases, the orientation of the first TM domain dictates the orientation of the subsequent TM domains, thus defining the overall topology of the protein. However, in most cases the final topology result from cooperation or competition between several topogenic determinants that are distributed throughout the polypeptide sequence (Gafvelin et al., 1997; McGovern and Beckwith, 1991; Sato et al., 1998; Sato and Mueckler, 1999). The impact from two topologically competing or cooperating hydrophobic sequences on the final topology, depends on the length of the connecting loop region (or spacer sequence) between the two TM helices. (Goder et al., 1999; Goder and Spiess, 2001). TM domains connected by a short loop cannot orient themselves independently of each other and are thus inserted into the membrane as one compact unit, a helical hairpin, but with increasing spacer sequence length the cooperativity or competition decreases (Goder and Spiess, 2001). It has been proposed that the TM domains in a polytopic membrane protein are kept together in the translocon and released into the membrane en bloc (Borel and 23 Simon, 1996). However, this is contradictory to cross-linking studies showing that individual TM domains can be cross-linked to lipids as soon as they reach the translocon, which indicates that the TM helices exit into the membrane one-byone (Heinrich et al., 2000; Martoglio et al., 1995; Mothes et al., 1997; Sadlish et al., 2005). Biosynthesis and targeting to the ER membrane In eukaryotic cells most of the membrane proteins are synthesized by ER-bound ribosomes and are then inserted into and folded in the ER membrane, from which they are subsequently transported to the plasma membrane or other organelles within the secretory pathway. There are two different pathways, by which membrane proteins are targeted to and inserted into the ER membrane. These include, (i) the co-translational pathway which is predominant in mammalian cells and (ii) the posttranslational pathway (Johnson and van Waes, 1999; Lecomte et al., 2003; Meacock et al., 2000; Walter and Johnson, 1994). As mentioned earlier, the N-terminal part of newly synthesized membrane proteins destined for the ER contains an ER targeting signal. This signal ensures that the protein is directed to the ER membrane. In the cotranslational pathway, the signal sequence is recognized by and bound to a hydrophobic methionine-rich domain of the cytosolic SRP as soon as it emerges from the ribosome (Keenan et al., 1998; Zopf et al., 1990). This event causes a pause in protein synthesis and translation is halted until the nascent chain– ribosome–SRP complex binds to the SRP-receptor on the rough ER (Keenan et al., 2001). Both the SRP and the SRP receptor are GTPases that regulate the targeting events by binding to and mediate the hydrolysis of GTP (Keenan et al., 2001). The targeting complex then disassembles and the SRP and the SRP receptor are released and become available for further targeting cycles (Pool et al., 2002). Following this, the ribosome docks onto the translocon and the SP is transferred to the recognition site in the Sec61α subunit of the translocon (Mothes et al., 1998; Mothes et al., 1994). Here it is oriented due to topogenic determinants and integrated into the ER membrane (Do et al., 1996; McCormick et al., 2003; Mothes et al., 1998) (Schatz and Dobberstein, 1996). The structure of the ribosometranslocon complex shows that the exit site of the ribosomal tunnel is aligned with the aqueous pore of the translocon (Beckmann et al., 2001). The ribosome and lumenal chaperones provide the driving force for translocation as the translocon channel itself is a passive conduit for polypeptides (Matlack et al., 1998). Several reports indicate that nascent polypeptide chains fold to various extent while in the ribosomal tunnel (Hardesty and Kramer, 2001) and that TM domains may be recognized already inside the ribosome. FRET analysis in combination with cross-linking of a nascent membrane protein support this idea, indicating 24 Marika Cassel that folding of TMs is induced in the ribosome (Woolhead et al., 2004). It has been shown that the nascent chains interact with the ribosomal interior and thus induce conformational changes in the highly dynamic ribosome (Frank and Agrawal, 2000; Gao et al., 2003). These conformational changes may be transferred to the translocon, which in turn influence the topogenesis of the membrane protein (Tenson and Ehrenberg, 2002). This notion is supported by a fluorescence quenching study, which shows that a TM inside the ribosome induces conformational changes in the translocon (Liao et al., 1997). The protein-conducting translocon The translocon is a membrane-embedded protein-conducting channel, found in the ER membrane of eukaryotes and in the plasma membrane of prokaryotes. Translocons are conserved and built from the same basic components in both the inner membrane of Escherichia coli and in the ER membrane (Rapoport et al., 1996). The translocon works in concert with the ribosome and is responsible both for the integration of membrane proteins and for the translocation of secretory proteins (Figure 3) (Laird and High, 1997; Johnson and van Waes, 1999;Rapoport et al., 2004 ;Matlack et al., 1998; von Heijne, 1997). Figure 3. Schematic representation of a ribosome-translocon complex. Soluble proteins pass through the translocon channel whereas TM segments of membrane proteins exit laterally into the membrane. The structural core of the Sec61 translocon in the eukaryotic ER membrane, which is homologous to the prokaryotic SecY translocon, is built from three different integral membrane proteins. These include Sec61α, Sec61β and Sec61γ, where Sec61α is the largest subunit and is thought to form the central aqueous pore through which the nascent polypeptide chain is translocated (Crowley et al., 1994; Gorlich and Rapoport, 1993; Johnson and van Waes, 1999; Rapoport et al., 1996; Van den Berg et al., 2004). Sec61α is also one of the first translocon components to interact with the polypeptide chain as it emerges from the ribosome (High et al., 25 1993; McCormick et al., 2003; Mothes et al., 1994; Sadlish et al., 2005; Thrift et al., 1991). The Sec 61 complex functions in association with a variety of other membrane proteins with specialized functions, such as oligosaccharyl transferase (OST), the signal peptidase complex (SPase), the translocon-associated protein complex (TRAP) and the translocating chain associated membrane protein (TRAM). OST has its active site on the luminal side of the ER membrane where it transfers oligosaccharides to acceptor sites for N-linked glycosylation (Asn-X-Thr) on growing nascent polypeptide chains. The SPase removes signal peptides from secretory proteins (Dalbey et al., 1997). TRAP is a tetrameric membrane protein complex of unknown function (Fons et al., 2003) and TRAM is thought to function as a chaperone during membrane protein integration as well as during early stages of secretory protein synthesis. The whole assembly has an outer diameter of about 100Å and the translocation pore has a diameter of about 20Å, as shown by cryo-EM image reconstructions (Beckmann et al., 1997; Menetret et al., 2000; Meyer et al., 1999). However, earlier studies suggested it to be wider (Hamman et al., 1997). Electrophysiology and fluorescence lifetime measurements have shown that the translocation pore has a hydrophilic interior (Crowley et al., 1994; Simon and Blobel, 1991). Whether the translocon channel forms at the interface between multiple copies of the heterotrimeric complex or is intrinsic to a monomeric Sec61 complex, as suggested by the recently solved X-ray structure of the archaeal SecY translocon from Methanococcus jannaschii (Van den Berg et al., 2004), is a controversial issue. It has been proposed that the channel is formed at the interface between three to four copies of the Sec61/SecY complex, based on EM of the yeast and mammalian translocons (Beckmann et al., 1997; Beckmann et al., 2001; Hamman et al., 1997; Hanein et al., 1996; Manting et al., 2000; Menetret et al., 2000). Model building from cryo-EM images suggests that the translocon assembly consists of four copies of heterotrimeric Sec61 and two copies of TRAP. This study also provides evidence for passage of the elongating chain through a single SecY heterotrimer (Menetret et al., 2005). This is also seen in the X-ray structure of SecY (Van den Berg et al., 2004), which suggests that a translocation pore is formed by a single monomer of the SecY complex, rather than at the interface of multiple SecY molecules. Recent crosslinking studies also support this model (Cannon et al., 2005). The structure shows that the translocon is organized in two halves where the first half, containing TM helices 1-5, is antiparallell to the second one, containing TM helices 6-10. The lateral gate of the translocon where hydrophobic sequences can exit into the lipid bilayer, is found between helices 2/3 and 7/8. In the absence of a translocating substrate, the central pore of this complex is surprisingly small with a diameter of 3Å and is plugged by a short helix (Van den Berg et al., 2004). 26 Marika Cassel When actively translocating, the plug would move away and the structure would alter so as to allow the vectorial transfer of polypeptide chains. In the structure, a ring of hydrophobic Ile residues is seen, which is thought to bind to hydrophobic segments in the polypeptide. This ring is also thought to provide a flexible seal around the translocating polypeptide preventing ion leakage through the membrane (Dobberstein and Sinning, 2004; Van den Berg et al., 2004). A recent study supports earlier indications that the ER membrane is nevertheless rather leaky to small molecules and suggests that this may be of functional significance for protein folding in the lumen of the ER (Le Gall et al., 2004). Translocation While the translocon provides a lateral exit to the lipid bilayer and rapidly orients and integrates the TM domains emerging from the ribosome one after the other, it simultaneously orients hydrophilic luminal and cytosolic domains of the membrane protein. At the same time it maintains the permeability barrier of the ER membrane, preventing ions and small molecules from freely crossing the membrane (Gilbert et al., 2004). To date little is known regarding how the translocon syncronizes this process. It is nevertheless clear that as the elongating polypeptide chain passes through the channel, the translocon interprets the code within the amino acid sequence of the nascent chain and decides whether protein segments should be translocated through the membrane or be selected for insertion. The basic features of this “sorting code” have recently been determined showing that membrane insertion depends strongly on the position of polar residues within TM segments. This study also implies that direct protein-lipid interactions are critical during translocon-mediated membrane insertion (Hessa et al., 2005). Translocon-mediated membrane insertion In several studies the environment of the polypeptide as it is integrated into the ER membrane has been analysed in vitro showing that every TM in a polypeptide chain is associated with the Sec61 complex during their transition into the lipid bilayer (Heinrich and Rapoport, 2003; Laird and High, 1997; Meacock et al., 2002; Mothes et al., 1997). The TRAM protein is also contacted by some TM domains in polytopic membrane proteins (Meacock et al., 2002). The integration process for polytopic membrane proteins can be described by two theoretical models: (i) the “en masse” model and (ii) the “sequential” model (High and Laird, 1997; Lecomte et al., 2003). In the “en masse” model, all the TM domains of the nascent polypeptide chain are proposed to accumulate within the ER translocon, before they are released together into the lipid bilayer. It has been shown in one study that a polytopic protein containing 5 TM helices behaves as though all the TM helices were in an aqueous pore (Borel and Simon, 1996). However, this model is contradictory to the X-ray structure of SecY which indicates that the pore may be too small to accommodate substantial numbers of TM domains inside it (Van den Berg et al., 2004). 27 Other studies support the “sequential” model (High and Laird, 1997; Lecomte et al., 2003), where each of the TM domains exit the Sec61 translocon individually during the biosynthesis of the membrane protein. In a recent study, where the integration process for the aquaporin-4 water channel (6TMs) was examined by photo-cross-linking, it was shown that the individual TM helices in this polytopic membrane protein are delivered from the Sec61 translocon into the lipid bilayer in succession, where TM2 follows TM1 and so on (Sadlish et al., 2005). A recent study of a membrane protein containing seven-TM helices has shown that the first and second TM domains leave the translocon before the synthesis of the protein is completed (Meacock et al., 2002). Newly synthezised TM domains may not exit the translocon in the order of their synthesis. Yet another possibility concerning integration of polytopic membrane proteins is that a single polypeptide chain may have different TM domains accommodated by distinct copies of the Sec61 complex during integration into the ER membrane (Dobberstein and Sinning, 2004). In some cases the integration of downstream TM domains in polytopic membrane proteins can be SRP-dependent, when for example a long cytosolic domain preceeds a TM helix (Kuroiwa et al., 1996). However, this is not the case for all polytopic membrane proteins (Wessels and Spiess, 1988). It can be concluded that too few examples of the synthesis and membrane integration processes of polytopic membrane proteins have been studied so far to allow general principles to be formulated. Folding of extramembraneous domains It is still not known precisely where the soluble loops flanking TM domains are located during synthesis, nor how their presence affects the gating of the pore. It has been suggested that cytosolic loops could protrude out of gaps at the cytosolic junction between the ribosome and the Sec61 complex (Beckmann et al., 2001). A similar scenario may occur on the luminal side of the ER membrane. The intracellular and extracellular hydrophilic regions of polytopic membrane proteins often have important biological functions and it is important that they are correctly folded and assembled. This is achieved by interactions with chaperones in the ER lumen and in the cytosol that assist the folding. Examples of lumenal chaperones are BiP and calnexin, which have been shown to be transiently associated with polytopic membrane proteins after their synthesis (Kopito, 1999; Swanton et al., 2003). Cytosolic chaperones Hsc70 and Hsp90 are involved in the folding of CFTR (Chapple and Cheetham, 2003; Kopito, 1999; Loo et al., 1998; Meacham et al., 1999; Swanton et al., 2003). It has been suggested that specific chaperones may act to facilitate the integration of TM domains as they exit the ER translocon (Meacock et al., 2002) and assist the folding and assembly of the membrane spanning part of membrane proteins. However, to date only calnexin has been implicated in this process. It associates with a misfolded 28 Marika Cassel domain of a polytopic membrane protein preventing its degradation (Swanton et al., 2003). Helix-helix packing in the membrane After being integrated into the membrane the TM helices of a polytopic membrane protein associate with each other. Several factors are thought to be responsible for the association of helices in membrane proteins. These include surface complementarity, the presence of polar residues in the TM region (Choma et al., 2000; Johnson et al., 2004; Zhou et al., 2001) and certain specific motifs such as the well-known GXXXG pattern (MacKenzie et al., 1997; Senes et al., 2001). It is likely that several of these factors act in concert to determine the final folded structure, as well as the association of monomers to form an oligomer. In addition to helix-helix interactions, interactions between the polypeptide chain with itself, water, the bilayer and in some cases cofactors also play a role in the organization and assembly of TM helices. 29 Comments on Methodology Model system used for the study of helical hairpin formation Model protein –Leader peptidase For the study of helical hairpin formation in the ER membrane, we have used the well-characterized protein leader peptidase (Lep) as a model membrane protein. Lep is a 323 amino acid protease found in the inner membrane of Escherichia coli, where it serves to remove signal peptides from secretory proteins. It contains two TM helices (H1 and H2), a small polar domain (P1) and a large C-terminal domain (P2) (Figure 4). The X-ray crystal structure of a catalytically active soluble fragment of Lep has been determined at 1.9Å resolution (Paetzel et al., 1998). Y When expressed in vitro in the presence of dog pancreas microsomes, Lep has been shown to insert into the microsomal membrane with the P1 loop facing the cytoplasm and the N-terminus as well as the P2 domain in the extra-cytoplasmic space, i.e. it adopts the same topology as in the Escherichia coli inner membrane (Wolfe et al., 1983). The H1 segment is needed for the correct membrane insertion and H2 is an internal uncleaved signal sequence, which is needed for translocation of the C-terminal domain. The P2 domain contains the catalytic site. To study helical hairpin formation, a unique acceptor site for N-linked glycosylation (Asn-Ser-Thr) has been introduced in the P2 domain, 20 residues downstream of H2 (Nilsson and von Heijne, 1993). P2 X N H1 N lumen H1 Y cy tosol P2 Figure 4. Model protein used to study helical hairpin formation. The H2 domain of Lep has been replaced by an artificial poly-Leu segment. Depending on the lumenal (no helical hairpin formed) or cytoplasmic (helical hairpin formed) localization of the P2 domain, the glycosylation acceptor site will either be modified (Y) or not ( ). 30 Marika Cassel Manipulation of the model protein In order to simplify our analysis of helical hairpin formation, the H2 domain of Lep was replaced by a specifically designed poly-Leu (39L+V) TM segments, that is long enough to form either a single TM helix or a helical hairpin (Monné et al., 1999b). Site-directed mutagenesis was then used to introduce mutations in the model protein. All the mutants were expressed in vitro in presence of dog pancreas microsomes and the proteins were analyzed by SDS-PAGE. The gels were scanned and quantified on a phosphoimager (Fuji BAS 1000). Following this, the glycosylation efficiency of all mutants were calculated. For the studies in paper II and III, we introduced an extra TM segment (H3) into the P2 domain of Lep. This H3 segment is composed of 40 Leu´s. Two glycosylation acceptor sites were introduced in position 96 and 258. If H3 spans the membrane once, only one acceptor site will be modified, whereas if H3 forms a helical hairpin with a cytoplasmic turn, both sites will be exposed to the lumenal side of the membrane and hence glycosylated. Singly and doubly glycosylated proteins are easily distinguished by SDS-PAGE. In paper III, the H2 TM domain of Lep was replaced by three different artificial segments, one at a time, each compriing a stretch of hydrophobic amino acids, containing 39LV W22 (39 Leu, 1 Val and Trp in position 22), 39LV and 30LV P18, respectively (31 residues is close to the minimum length required for helical hairpin formation). Glycosylation mapping Segments of eukaryotic membrane proteins that are translocated across the ER membrane can become modified by N-linked glycosylation on Asn-X-Thr/Ser acceptor sites (where X is any amino acid except proline.) Almost as soon as the polypeptide chain enters the ER lumen, it is glycosylated on the target Asn. The oligosaccharide is transferred onto the nascent chain in a reaction catalysed by the enzyme OST. There is one copy of this enzyme associated with each translocon in the ER membrane and it has its catalytic site located in the lumen of the ER (Palade, 1975). The active site of OST is placed at a fixed distance of about 30-40 Å from the lipid bilayer and is oriented roughly parallel to the membrane surface (von Heijne, 1996). By introducing glycosylation acceptor sites in the protein of interest, this defined position of the OST active site makes it possible to study how proteins integrate into and translocate across the ER membrane. By in vitro transcription and translation of Lep, in the presence of dog pancreas microsomes (Liljeström and Garoff, 1991), lumenally oriented domains will be glycosylated whereas cytoplasmically oriented domains will not be modified. An advantage of this method is that the microsomal in vitro system closely mimics the conditions of in vivo membrane protein assembly into the ER membrane. The presence of the N-linked oligosaccharide can easily be detected, as glycosylated substrate 31 molecules have a lower migration rate than non-glycosylated molecules when separated by SDS-PAGE. The glycosylation mapping approach is a simple experimental system for the analysis of turn formation in TM helices embedded in the ER membrane (Nilsson et al., 1998). This technique has allowed us to perform detailed studies of the factors that determine how helical hairpins are formed. Alkaline extraction To show that our poly-Leu based helical hairpins are properly assembled in the ER membrane and to rule out that low levels of glycosylation are caused by a failure in membrane integration rather than helical hairpin formation, we have used sodium carbonate extraction on some of our constructs as previously described (Hunt et al., 1997). This was done by expressing a truncated form of several constructs forming helical hairpins, leaving H2 as the only potential membrane-spanning segment. As expected, the proteins were not glycosylated and were associated with the membrane pellet after alkaline extraction of the microsomes. Alkaline phosphatase as a topology reporter protein In paper V we have used a combination of two topology reporter fusion methods, where the protein under study is fused to a reporter protein, PhoA or GFP (described below). PhoA is a commonly used reporter protein for identifying extracellular loops of membrane proteins expressed in Escherichia coli. PhoA is only enzymatically active on the periplasmic side of the inner membrane. In order to be active, the protein has to form two disulfide bonds (Akiyama and Ito, 1993; Derman and Beckwith, 1991). These two bonds can only be formed in the oxidizing environment of the periplasm. When a substrate analogue is added to a bacterial culture expressing the fusion construct, a chromogenic product is formed only when the PhoA domain is in the periplasm. Its activity can be quantified spectrophotometrically (Manoil, 1991; Manoil and Beckwith, 1986). Green flourescent protein as a topology reporter protein GFP is another example of a thoroughly tested topology reporter protein that can be fused to the C-terminal part of the protein of interest. GFP forms a chromophore when correctly folded but the protein can only fold in a proper manner when on the cytoplasmic side of the inner membrane in Escherichia coli. A correct folding is needed in order for GFP to fluoresce. When targeted to the periplasm by a Sec-type signal peptid (Feilmeier et al., 2000), GFP does not fold properly and remains non-fluorescent. 32 Marika Cassel Results and Discussion Formation of helical hairpins during membrane protein integration into the ER membrane The major aim of this thesis has been to better define the topological determinants underlying the formation of helical hairpins in membranes. Helical hairpins are basic folding units in membrane proteins during membrane protein assembly in the ER membrane. The helical hairpin is a structural unit in multi-spanning integral membrane proteins. It is composed of two closely spaced hydrophobic TM α-helices connected by a short turn (Engelman and Steitz, 1981). The helical hairpin is formed during translocon assisted membrane protein integration and is thought to be inserted into the membrane as one compact unit. It is becoming increasingly clear that the helical hairpin is a very common structural element in membrane proteins and a detailed understanding of its properties is of central importance. Formation of helical hairpins with a lumenal turn, induced by turn promoting residues (paper I) The formation of tight turns in globular proteins has been studied for decades both experimentally and by statistical analysis of known structures. As a result of this research, reliable turn propensity scales have been established. However, before we started this project nothing was really known about the residue characteristics responsible for the formation of tight turns between transmembrane α-helices in integral membrane proteins. Previous studies demonstrated that, when a single Pro residue is introduced near the middle of a 40 residues long poly(Leu) TM helix, a turn is induced and the polyLeu stretch is efficiently converted from a long, single TM helix into a helical hairpin (Nilsson et al., 1998). Because of these findings, in paper I we extended the study by measuring the ability of all the 20 naturally occurring amino acids to induce the formation of a tight turn and consequently a helical hairpin, when placed in the middle of a 40 amino acids long hydrophobic TM helix. We used a glycosylation mapping approach that can distinguish between one long TM helix and a helical hairpin (see previous section). This study revealed that the amino acid residues have either a high or a low turn propensity and a turn propensity scale for TM helices was derived. The amino acid residues with high turn propensity are the charged (Arg, Lys, His, Glu and Asp) and the highly polar residues (Asn and Gln) together with the classical helix breakers, Pro and Gly. This means that these residues efficiently promote formation of helical hairpins 33 with a tight lumenal turn, when placed in the middle of a hydrophobic TM helix. The hydrophobic amino acid residues (Leu, Phe, Ala, Met, Val and Ile), together with the weakly polar amino acids (Ser, Thr, Tyr and Cys), have no or low turn propensity and do not promote helical hairpin formation. In summary, most of the results from paper I can be explained by hydrophobicity, i.e. all hydrophobic residues prefer the membrane environment over the membrane-water interface region and have low turn propensities. Conversely, the charged and highly polar residues induce turn formation in order to avoid the membrane interior. From our results we can conclude that turn propensities in TM helices are markedly different from those of globular proteins, and in most cases correlate closely with the hydrophobicity of the residue (Monné et al., 1999a). Following this, research in our laboratory was designed to derive a more finegrained turn-propensity scale. This was based on (i) measurements of the turninducing potential of single residues placed in the middle of a 31 residue poly(Leu) segment and (ii) measurements of the turn-inducing potential of pairs of residues placed in the middle of a 40 residue poly(Leu) segment (Monné et al., 1999b). This study showed that if the hydrophobic segment is long enough and contains a turn-inducing residue in the middle, it forms a helical hairpin. The shortest TM segment that can form a helical hairpin is around 30 residues long (Monné et al., 1999b). Formation of helical hairpins with a cytoplasmic turn, induced by turn promoting residues (paper II) In previous studies, the turn of the helical hairpin was always placed on the lumenal side of the ER membrane. To expand these studies, in paper II we performed similar measurements of turn propensities, but in this study the turn of the helical hairpin has a cytoplasmic location. To study formation of helical hairpins with a cytoplasmic turn, we introduced one or a pair of each of the 20 naturally occurring amino acids in the middle of the polyLeu stretch (H3). Subsequently, their ability to induce a cytoplasmic turn was assessed. The result showed that a single Pro residue induces a helical hairpin in only 36% of the molecules and 3 Pro are needed to reach full degree of helical hairpin formation. This is in contrast to our previous studies of lumenal turns, where one Pro was sufficient to induce full helical hairpin formation. We demonstrated that none of the singly inserted residues were able to promote efficient formation of a helical hairpin with a cytoplasmic turn to any significant extent. However, charged and polar amino acids all gave some degree of helical hairpin formation. Pairs of polar or charged residues had a higher turn inducing 34 Marika Cassel effect but the molecules do not reach full helical hairpin formation in most cases. All the hydrophobic residues plus Thr, Ser, Cys, Gly, Trp and Tyr had low or no turn-inducing effect even when placed in tandem. Only in constructs with 2 Lys or 2 Arg in the middle, a full degree of helical hairpins was formed. To summarize, from the results in paper I and paper II we can conclude that hydrophobic amino acids do not induce turns, charged residues are strong turn promoters and most polar amino acids are in-between. Among the non-charged residues, Pro and Asn have the highest turn propensities. This general picture holds true both for helical hairpins with turns on the cytoplasmic and lumenal side of the ER membrane. However, there are some differences. In the case of turns on the cytoplasmic side of the ER membrane, a higher number of turn-promoting residues are required to induce a helical hairpin. Arg and Lys are the strongest turn-promoters on the cytoplasmic side, but on the lumenal side, Pro, Asn and Arg are the strongest ones. The high cytoplasmic turn propensities for Arg and Lys may be related to the positive inside rule. The differences in the number of turn–promoting residues required to induce a helical hairpin with a cytoplasmic respectively a luminal turn may be an effect of the translocon mediated insertion event. During insertion into the ER membrane an internal signal-anchor sequence like H2 presumably enters the translocon in a loop or hairpin conformation which may facilitate the formation of a permanent helical hairpin in the presence of a turn-promoting residue. The H3 stop-transfer sequence on the other hand presumably enters the translocon in a stretched conformation and the effect of turn-promoting residues may be weaker. This may explain why a smaller number of turn-promoting residues are required when the turn is on the lumenal side. Helical hairpin formation promoted by flanking charged residues (paper III) In paper III, we shifted our attention to the possible role of the residues that flank the TM helices. As charged residues are known to act as powerful topological determinants, in both prokaryotic and eukaryotic membrane proteins, we considered the possibility that charged residues in the flanking region of a long hydrophobic TM helix may specifically affect helical hairpin formation. The results from paper III showed that both Lys and Asp residues, introduced in the C-terminally flanking region, promote the formation of helical hairpins with a lumenally oriented turn. A helical hairpin can thus be induced even in a uniformly hydrophobic segment, provided that the number of C-terminally flanking charged residues is sufficiently high. At the same time, N-terminally 35 flanking residues have no effect on helical hairpin formation in our model protein. It was further shown that C-terminally flanking Lys and Asp residues inhibit the formation of helical hairpins with a cytoplasmic turn. The effect of Lys was somewhat stronger than that of Asp, which is in accordance with the positive inside rule. The effect of increasing the separation between clusters of charged residues (four Lys or four Asp) and the hydrophobic stretch for helical hairpins with lumenal and cytoplasmic turn was also studied. We found that a cluster of four consecutive Asp or Lys residues can affect the efficiency of helical hairpin formation even when placed 30 residues or more downstream of the hydrophobic stretch. This may suggest that when the helical hairpin forms in the translocon, the whole length of the nascent chain present in the ribosomal tunnel (about 40 residues) at the time can have an impact on the process of helical hairpin formation. Helical hairpin formation induced by inter-helical hydrogen bonds (paper IV) As mentioned previously, the polypeptide chain is threaded through the translocon during membrane protein biosynthesis. In this process, hydrophobic regions of the nascent chain are recognised by the translocon and transferred laterally into the membrane, either one by one, or associated in pairs (Do et al., 1996; Heinrich et al., 2000; Kanki et al., 2003; Oliver et al., 1995) or as higher–order multimers (Borel and Simon, 1996). Several factors are thought to be responsible for the association of TM helices in membrane proteins, including surface complementarity, the GXXXG-motif (MacKenzie et al., 1997; Senes et al., 2001) and the presence of polar residues in the TM region. (Choma et al., 2000; Johnson et al., 2004; Zhou et al., 2001). It is likely that several of these factors act in concert to determine the final folded structure of membrane proteins. However, before the membrane protein folding pathway is fully understood many details remain to be elucidated of how and to what extent different interactions direct the folding and association of TM segments. In several studies, it has been shown that inter-helical hydrogen bond formation between pairs of the polar residues Asn or Asp in TM α-helices can drive efficient oligomerization in detergent micelles, model membranes, and in the Escherichia coli inner membrane (Choma et al., 2000; Gratkowski et al., 2001; Howard et al., 2002). These observations inspired us to examine whether pairs of Asn or Asp residues in a model TM helix also could promote the formation of helical hairpins during membrane protein assembly into the ER membrane. 36 Marika Cassel For the study in paper IV, one or a pair of Asn or Asp residues were introduced in different positions in a poly(Leu) stretch in the H2 position of Lep. Helical hairpin formation was assayed using the glycosylation mapping approach. Our results showed that none of the single Asn-mutations induced efficient helical hairpin formation, while single Asp-mutations have some effect. We further showed that interhelical hydrogen bonds between Asn-Asn and Asp-Asp pairs can drive helical hairpin formation. Asn-Asn and Asp-Asp pairs increase the efficiency of helical hairpin formation by 1.5-2-fold over that seen for the corresponding single Asn or Asp residues. From our results, it is also evident that the effects on helical hairpin formation by Asn-Asn and Asp-Asp pairs are highly position specific. From the results of paper IV, we can conclude that the helical hairpin conformation is stabilized by hydrogen bonding. The results further imply that the formation of helical hairpins must take place in a non-aqueous environment in the Sec61 translocon, since stable inter-residue hydrogen bonding would not be expected in an aqueous environment. Instead residues would prefer to form hydrogen bonds with the surrounding water molecules. A non-aqueous environment in the Sec61 translocon has been implied in a previous study (Nilsson et al., 2003). The results further implies that the two halves of the helical hairpin formed in the TM segment cannot rotate freely relative to one another, as only residues in positions on one face of the two helices are implicated. Confronting fusion protein-based membrane protein topology mapping with reality: the Escherichia coli ClC chloride channel (paper V) In the most recent study of this thesis, a detailed fusion-based topology mapping of the Escherichia coli chloride channel YadQ was performed. This was done to evaluate GFP and PhoA topology mapping assays and topology prediction methods. The X-ray structure of YadQ was reported in 2002 (Dutzler et al., 2002). YadQ is a homodimer, with an ion pore in each of the monomers (Figure 5). Each monomer contains 14 TM helices of varying lengths of which some are highly inclined relative to the membrane normal. Considering the complicated structural arrangement of the α-helices, this protein was considered to be a challenge for topology mapping by reporter fusions. From the YadQ structure optimal fusion joints were chosen and a set of fusions were made to each loop in the structure. The reporter fusion activities were mapped onto the 3D structure of YadQ and on the corresponding topology map. We found that the reporter fusion data accurately reflects the topology of 10 of the 14 TM helices, but do not score two short, weakly hydrophobic helical 37 hairpins on the periphery of the protein. The fully constrained topology prediction method TMHMM only predicts 4 of the 7 helical hairpins correctly. We suggest that the reporter fusion results reflect not only the final topology of the protein but also pin-point parts parts of the protein that may insert into the membrane at a late stage in the folding process. The reporter fusion approach thus offer new insight into the folding pathway of YadQ. Figure 5. 3D structure of the YadQ homodimer. The two subunits are shaded in gold and white. The periplasmic side is up. 38 Marika Cassel Concluding Remarks We have shown that the formation of helical hairpins during membrane protein integration into the ER membrane can be dependent both on the overall length of the hydrophobic segment and on the amino acids flanking the TM segment. Furthermore, the identity of the central, potentially turn-forming residues is of importance. Thus, charged, polar, and classic helix-breaking residues are good turn-formers and effectively induce helical hairpin formation when placed near the middle of a sufficiently long hydrophobic stretch. In contrast, centrally placed apolar residues do not induce helical hairpin formation, but rather cause the hydrophobic stretch to insert as a single, long TM helix that spans the membrane once. Hydrophobic amino acids do not induce turns, charged residues are strong turn promoters and most polar amino acids are in-between. Among the non-charged residues, Pro and Asn have the highest turn propensities. This general picture holds both for helical hairpins with turns on the cytoplasmic and lumenal side of the ER membrane. However, a higher number of turn-promoting residues are required to induce a helical hairpin with a cytoplasmic turn. Arg and Lys are the strongest turn-promoters on the cytoplasmic side of the ER membrane, and on the lumenal side, Pro, Asn and Arg are the strongest ones. C-terminally flanking Lys and Asp residues promote the formation of helical hairpins with a lumenal turn, while N-terminally flanking residues have no effect on helical hairpin formation. In contrast, C-terminally flanking Lys and Asp residues inhibit the formation of helical hairpins with a cytoplasmic turn. The effect on helical hairpin formation of C-terminally flanking Lys and Asp residues decrease with the distance from the hydrophobic stretch. We have also shown that interhelical hydrogen bonds between Asn-Asn and Asp-Asp pairs can drive helical hairpin formation in a position-specific manner. Some of the topological determinants underlying the formation of helical hairpins during membrane protein integration into the ER membrane have been clarified in this thesis. These observations, together with future results, will increase our understanding of what determines membrane protein topology and structure. In addition, the information obtained may be relevant for improving current methods of membrane protein topology prediction. 39 Acknowledgements Många är ni som förgyllt min tillvaro under doktorandtiden och fått mig att nå ända fram. Ett speciellt VARMT TACK till: Gunnar, min handledare, för all din hjälp och uppmuntan som jag fått under doktoreringen. Tack, för att du lotsat mig i rätt riktning och för att du gett mig frihet i arbetet, och för att du alltid blivit så glad och entusiastisk, även för de allra minsta resultat som jag lyckats frambringa. Tack också till både dig och Anna för trevliga middagar och annandags-fika! Stefan, för ditt stöd och uppmuntran och för att du är så juste och bryr dig om och gör ett så himla bra jobb som institutionschef. Anki, Ann, Kicki, Bogos, Eddie och Peter för att ni är så vänliga och hjälpsamma och alltid ställer upp. Mikaela, för att du accepterat att ständigt bli ihopblandad med mig. För ditt ständiga stöd och din positiva syn. För alla oändligt många gånger du fått mig att skratta. För att du är en god vän och pärla i största allmänhet. Karin, för att du är min bästa lunchkompis och en god vän, som alltid är snäll och vänlig och ger stöd. Tack för alla mysiga o trevliga fikor under mammaledigheten och tack för allt kul på Korsika. Tack också för hjälp med sista peket. Malin, för att du är så snäll och bryr dig om. För att du lyssnar och stöttar. För allt kul vi hade på Korsika. Tack till både dig och Johan för mysiga supermiddagar! Sån´t får det bli mer av. Annika, min vän och fd labkompis. Vilket samarbete vi hade! Tack för att du alltid stöttat mig och bryr dig om. Tack för allt kul vi haft. Önskar att du kunde flytta tillbaka hem snart igen! Magnus, för att du introducerade mig i labbandet, för att du fick mig att ”förstå mitt eget bästa” och ta klivet att doktorera. Tack för allt kul vi haft på den goda tiden då du fortfarande var kvar i gruppen. Tara, för ditt stöd och din omtanke och all hjälp och framförallt för allt kul vi haft. Mina labkompisar: Marie, Joy, Dan, Carolina, Filippa, Susanna, Yoko, för all er hjälp och för all den kunskap ni delar med er av, men framförallt tack för att ni är så vänliga och gör dagarna på labbet extra roliga. Louise, David, Sam och David och Jan-Willem för att ni är så trevliga labgrannar och bjuder på så mycket goda grejor. 40 Marika Cassel Alla ni som tidigare har varit i min grupp, för all er hjälp och kunskap ni delar med er av: Marie, Chen-Ni, Nadja, Thomas, Susana, Susanna, Paula IngMarie, för att du alltid är hjälpsam. Din arbetsamhet och ditt vetenskapliga engagemang är inspirerande. Åke, min bihandledare, för att du varit nyfiken på och intresserat dig för min forskning. Marlene, för att du är en så god vän som jag tycker så mycket om. Anna, Linus och Alice, för att ni är så härliga och omtänksamma och vet hur livet ska levas. Camilla, Lars och Isak, Sofia, Fredrik och Eskil, Ingeborg, Magnus och Melker, för att ni förgyller tillvaron med vänskap och goda middagar. Framför allt vill jag tacka min underbara familj: Mina svärföräldrar: Margareta och Lars, Claes och Marie för att ni är så varma, generösa och fantastiska människor. Calle, för att du är så juste, rolig, snäll och hjälpsam. Kalle o Elin, för att ni är så snälla och gulliga och bryr er om och för att det är så kul att vara med er. Samt resten av bonusfamiljen: Caroline, Paul, Daisy, Louisa, Thomas, Clara, Jakob, Christine, Emmy och Emil, vilken familj! Mamma och Pappa, för er outtömliga omtanke, värme och godhet. Tack för alla oändligt många gånger som ni ställt upp för mig och Tobias. Tack för att ni alltid tror på mig och ger mig stöd i mot- och medgång. Stellan, världens bästa bror, för allt ditt stöd. För att du är så genuint snäll och bryr dig om. Tack för att du får mig att tänka på härliga saker som segling och havet. Filippa, söta lilla solstråle, som är så härligt glad, god och empatisk. Du skänker så mycket glädje och lycka och får oss att skratta. Du är en pärla. Tobias, min Älskling. Det svåraste med den här avhandlingen har faktiskt varit att komma på hur jag i ord skall kunna tacka Dig... för all oändlig kärlek, lycka och glädje som Du ger mig, för att Du gör livet så härligt att leva, för all hjälp och allt Ditt stöd, för att Du får mig att skratta. Du har gett mig allt som jag någonsin drömt om. Du är helt fantastisk, en pärla av stora mått. Du är mitt allt. 41 References Akiyama, Y. and Ito, K. (1993) Folding and assembly of bacterial alkaline phosphatase in vitro and in vivo. J Biol Chem, 268, 8146-8150. Bowie, J.U. (2000) Understanding membrane protein structure by design. Nat Struct Biol, 7, 91-94. Andersson, H., Bakker, E. and von Heijne, G. (1992) Different positively charged amino acids have similar effects on the topology of a polytopic transmembrane protein in Escherichia coli. J Biol Chem, 267, 1491-1495. Boyd, D. and Beckwith, J. (1989) Positively charged amino acid residues can act as topogenic determinants in membrane proteins. Proc Natl Acad Sci USA, 86, 9446-9450. Arora, A., Abildgaard, F., Bushweller, J.H. and Tamm, L.K. (2001) Structure of outer membrane protein A transmembrane domain by NMR spectroscopy. Nat Struct Biol, 8, 334338. Beckmann, R., Bubeck, D., Grassucci, R., Penczek, P., Verschoor, A., Blobel, G. and Frank, J. (1997) Alignment of conduits for the nascent polypeptide chain in the ribsomeSec61 complex. Science, 278, 2123-2126. Beckmann, R., Spahn, C.M., Eswar, N., Helmers, J., Penczek, P.A., Sali, A., Frank, J. and Blobel, G. (2001) Architecture of the protein-conducting channel associated with the translating 80S ribosome. Cell, 107, 361-372. Belrhali, H., Nollert, P., Royant, A., Menzel, C., Rosenbusch, J. P., Landau, E. M. and Pebay-Peyroula, E. (1999) Protein, lipid and water organization in bacteriorhodopsin crystals: a molecular view of the purple membrane at 1.9 A resolution. Structure Fold Des, 7, 909-917 Blobel, G. (1983) Control of intracellular protein traffic. Methods Enzymol, 96, 663-682. Borel, A.C. and Simon, S.M. (1996) Biogenesis of polytopic membrane proteins: membrane segments assemble with translocation channels prior to membrane integration. Cell, 85, 379389. Bowie, J.U. (1997) Helix packing in membrane proteins. J Mol Biol, 272, 780-789. Boyd, D., Traxler, B. and Beckwith, J. (1993) Analysis of the topology of a membrane protein by using a minimum number of alkaline phosphatase fusions. J Bacteriol, 175, 553-556. Broome-Smith, J.K. and Spratt, B.G. (1986) A vector for the construction of translational fusions to TEM ß-lactamase and the analysis of protein export signals and membrane protein topology. Gene, 49, 341-349. Broome-Smith, J.K., Tadayyon, M. and Zhang, Y. (1990) Beta-Lactamase As a Probe of Membrane Protein Assembly and Protein Export. Mol Microbiol, 4, 1637-1644. Brändén, C. and Tooze, J. (1991) Introduction to Protein Structure. Garland Publishing, New York. Buchanan, S.K. (1999) β-barrel proteins from bacterial outer membranes: structure, function and refolding. Curr Opin Struct Biol, 9, 455-461. Cannon, K.S., Or, E., Clemons, W.M., Jr., Shibata, Y. and Rapoport, T.A. (2005) Disulfide bridge formation between SecY and a translocating polypeptide localizes the translocation pore to the center of SecY. J Cell Biol, 169, 219-225. Chapple, J.P. and Cheetham, M.E. (2003) The chaperone environment at the cytoplasmic face of the endoplasmic reticulum can modulate rhodopsin processing and inclusion formation. J Biol Chem, 278, 19087-19094. 42 Marika Cassel Chen, C.P., Kernytsky, A. and Rost, B. (2002) Transmembrane helix predictions revisited. Protein Sci, 11, 2774-2791. Chen, C.P. and Rost, B. (2002) Long membrane helices and short loops predicted less accurately. Protein Sci, 11, 2766-2773. Chiu, M.L., Nollert, P., Loewen, M.C., Belrhali, H., Pebay-Peyroula, E., Rosenbusch, J.P. and Landau, E.M. (2000) Crystallization in cubo: general applicability to membrane proteins. Acta Crystal D Biol Crystal, 56 ( Pt 6), 781-784. enzymology of the type I signal peptidases. Protein Sci, 6, 1129-1138. Daley, D.O., Rapp, M., Granseth, E., Melén, K., Drew, D. and von Heijne, G. (2005) Global topology analysis of the Escherichia coli inner membrane proteome. Science, 308, 1321-1323. Deisenhofer, J., Epp, O., Miki, K., Huber, R. and Michel, H. (1985) Structure of the protein subunits in the photosynthetic reaction centre of Rhodopseudomonas viridis at 3Å resolution. Nature, 318, 618-624. Choma, C., Gratkowski, H., Lear, J.D. and DeGrado, W.F. (2000) Asparagine-mediated self-association of a model transmembrane helix. Nat Struct Biol, 7, 161-166. Denzer, A.J., Nabholz, C.E. and Spiess, M. (1995) Transmembrane orientation of signalanchor proteins is affected by the folding state but not the size of the N- terminal domain. EMBO J, 14, 6311-6317. Claros, M.G. and von Heijne, G. (1994) TopPred II: An improved software for membrane protein structure prediction. CABIOS, 10, 685-686. Derman, A.I. and Beckwith, J. (1991) Escherichia coli alkaline phosphatase fails to acquire disulfide bonds when retained in the cytoplasm. J Bacteriol, 173, 7719-7722. Cornette, J.L., Cease, K.B., Margalit, H., Spouge, J.L., Berzofsky, J.A. and DeLisi, C. (1987) Hydrophobicity scales and computational techniques for detecting amphipathic structures in proteins. J Mol Biol, 195, 659-685. Do, H., Falcone, D., Lin, J., Andrews, D.W. and Johnson, A.E. (1996) The cotransational integration of membrane proteins into the phospholipid bilayer is a multistep process. Cell, 85, 369-378. Cowan, S.W. and Rosenbusch, J.P. (1994) Folding pattern diversity of integral membrane proteins. Science, 264, 914-916. Cowan, S.W., Schirmer, T., Rummel, G., Steiert, M., Ghosh, R., Pauptit, R.A., Joansonius, J.N. and Rosenbusch, J.P. (1992) Crystal structures explain functional properties of two E. coli porins. Nature, 358, 727-733. Crowley, K.S., Liao, S., Worrell, V.E., Reinhart, G.D. and Johnson, A.E. (1994) Secretory proteins move through the endoplasmic reticulum membrane via an aqueous, gated pore. Cell, 78, 461-471. Dalbey, R.E., Lively, M.O., Bron, S. and Van Dijl, J.M. (1997) The chemistry and Dobberstein, B. and Sinning, I. (2004) Structural biology. Surprising news from the PCC. Science, 303, 320-322. Drew, D., Sjöstrand, D., Nilsson, J., Urbig, T., Chin, C.N., de Gier, J.W. and von Heijne, G. (2002) Rapid topology mapping of Escherichia coli inner-membrane proteins by prediction and PhoA/GFP fusion analysis. Proc Natl Acad Sci USA, 99, 2690-2695. Drew, D., von Heijne, G., Nordlund, P. and de Gier, J.W.L. (2001) Green fluorescent protein as an indicator to monitor membrane protein overexpression in Escherichia coli. FEBS Letters, 507, 220-224. Dutzler, R., Campbell, E.B., Cadene, M., Chait, B.T. and MacKinnon, R. (2002) X-ray structure of a ClC chloride channel at 3.0 Å 43 reveals the molecular basis of anion selectivity. Nature, 415, 287-294. Engelman, D.M. and Steitz, T.A. (1981) The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell, 23, 411-422. Feilmeier, B.J., Iseminger, G., Schroeder, D., Webber, H. and Phillips, G.J. (2000) Green fluorescent protein functions as a reporter for protein localization in Escherichia coli. J Bacteriol, 182, 4068-4076. Fernandez, C., Hilty, C., Bonjour, S., Adeishvili, K., Pervushin, K. and Wuthrich, K. (2001) Solution NMR studies of the integral membrane proteins OmpX and OmpA from Escherichia coli. FEBS Lett, 504, 173-178. Fernandez, C., Hilty, C., Wider, G. and Wuthrich, K. (2002) Lipid-protein interactions in DHPC micelles containing the integral membrane protein OmpX investigated by NMR spectroscopy. Proc Natl Acad Sci U S A, 99, 13533-13537. Fons, R.D., Bogert, B.A. and Hegde, R.S. (2003) Substrate-specific function of the translocon-associated protein complex during translocation across the ER membrane. J Cell Biol, 160, 529-539. Frank, J. and Agrawal, R.K. (2000) A ratchetlike inter-subunit reorganization of the ribosome during translocation. Nature, 406, 318-322. Froshauer, S., Green, G.N., Boyd, D., McGovern, K. and Beckwith, J. (1988) Genetic analysis of the membrane insertion and topology of MalF, a cytoplasmic membrane protein of Escherichia coli. J Mol Biol, 200, 501511. Gafvelin, G., Sakaguchi, M., Andersson, H. and von Heijne, G. (1997) Topological rules for membrane protein assembly in eukaryotic cells. J Biol Chem, 272, 6119-6127. Gafvelin, G. and von Heijne, G. (1994) Topological "frustration" in multi-spanning E. coli inner membrane proteins. Cell, 77, 401-412. Gao, H., Sengupta, J., Valle, M., Korostelev, A., Eswar, N., Stagg, S.M., Van Roey, P., Agrawal, R.K., Harvey, S.C., Sali, A., Chapman, M.S. and Frank, J. (2003) Study of the structural dynamics of the E coli 70S ribosome using real-space refinement. Cell, 113, 789-801. Gilbert, R.J., Fucini, P., Connell, S., Fuller, S.D., Nierhaus, K.H., Robinson, C.V., Dobson, C.M. and Stuart, D.I. (2004) Threedimensional structures of translating ribosomes by Cryo-EM. Mol Cell, 14, 57-66. Goder, V., Bieri, C. and Spiess, M. (1999) Glycosylation can influence topogenesis of membrane proteins and reveals dynamic reorientation of nascent polypeptides within the translocon. J Cell Biol, 147, 257-266. Goder, V., Crottet, P. and Spiess, M. (2000) In vivo kinetics of protein targeting to the endoplasmic reticulum determined by sitespecific phosphorylation. EMBO J, 19, 67046712. Goder, V. and Spiess, M. (2001) Topogenesis of membrane proteins: determinants and dynamics. FEBS Lett, 504, 87-93. Gorlich, D. and Rapoport, T.A. (1993) Protein translocation into proteoliposomes reconstituted from purified components of the endoplasmic reticulum membrane. Cell, 75, 615-630. Gouaux, E. (1997) Channel-forming toxins: tales of transformation. Curr Opin Struct Biol, 7, 566-573. Gratkowski, H., Lear, J.D. and DeGrado, W.F. (2001) Polar side chains drive the association of model transmembrane peptides. Proc Natl Acad Sci U S A, 98, 880-885. Hamman, B., Chen, J.-C., Johnson, E. and Johnson, A. (1997) The aqueous pore through 44 Marika Cassel the translocon has a diameter of 40-60 Å during cotrsnaltional protein translocation at the ER membrane. Cell, 89, 535-544. Hanein, D., Matlack, K.E.S., Jungnickel, B., Plath, K., Kalies, K.-U., Miller, K.R., Rapoport, T. and Akey, C.W. (1996) Oligomeric rings of the Sec61p complex induced by ligands required for protein translocation. Cell, 87, 721-732. Hardesty, B. and Kramer, G. (2001) Folding of a nascent peptide on the ribosome. Prog Nucleic Acid Res Mol Biol, 66, 41-66. Heinrich, S., Mothes, W., Brunner, J. and Rapoport, T. (2000) The Sec61p complex mediates the integration of a membrane protein by allowing lipid partitioning of the transmembrane domain. Cell, 102, 233-244. Heinrich, S.U. and Rapoport, T.A. (2003) Cooperation of transmembrane segments during the integration of a double-spanning protein into the ER membrane. EMBO J, 22, 3654-3663. Henderson, R., Baldwin, J.M., Ceska, T.A., Zemlin, F., Beckmann, E. and Downing, K.H. (1990) A Model for the structure of bacteriorhodopsin based on high resolution electron cryo-microscopy. J Mol Biol, 213, 899929. Henderson, R. and Unwin, P.N.T. (1975) Three-dimensional model of purple membrane obtained by electron microscopy. Nature, 257, 28-32. Hessa, T., Kim, H., Bihlmaier, K., Lundin, C., Boekel, J., Andersson, H., Nilsson, I., White, S.H. and von Heijne, G. (2005) Recognition of transmembrane helices by the endoplasmic reticulum translocon. Nature, 433, 377-381. High, S. and Dobberstein, B. (1991) The Signal Sequence Interacts with the Methionine-Rich Domain of the 54-kD Protein of Signal Recognition Particle. J Cell Biol, 113, 229-233. High, S. and Dobberstein, B. (1992) Mechanisms that determine the transmembrane disposition of proteins. Curr Opin Cell Biol, 4, 581-586. High, S. and Laird, V. (1997) Membrane protein biosynthesis - all sewn up? Trends Cell Biol., 7, 206-210. High, S., Martoglio, B., Görlich, D., Andersen, S.S.L., Ashford, A.J., Giner, A., Hartmann, E., Prehn, S., Rapoport, T.A., Dobberstein, B. and Brunner, J. (1993) Sitespecific photocrosslinking reveals that Sec61p and TRAM contact different regions of a membrane-inserted signal sequence. J Biol Chem, 268, 26745-26751. Howard, K.P., Lear, J.D. and DeGrado, W.F. (2002) Sequence determinants of the energetics of folding of a transmembrane four-helixbundle protein. Proc Natl Acad Sci U S A, 99, 8568-8572. Hunt, J., Rath, P., Rotschild, K. and Engelman, D. (1993) Spontaneous, pHdependent membrane insertion of a transbilayer α-helix. Biochemistry, 36, 1517715192. Hunt, J.F., Rath, P., Rothschild, K.J. and Engelman, D.M. (1997) Spontaneous, pHdependent membrane insertion of a transbilayer α-helix. Biochemistry, 36, 1517715192. Hwang, P.M., Choy, W.Y., Lo, E.I., Chen, L., Forman-Kay, J.D., Raetz, C.R., Prive, G.G., Bishop, R.E. and Kay, L.E. (2002) Solution structure and dynamics of the outer membrane enzyme PagP by NMR. Proc Natl Acad Sci U S A, 99, 13560-13565. Ikeda, M., Arai, M., Lao, D. and Shimizu, T. (2001) Transmembrane topology prediction methods: A re-assessment and improvement by a consensus method using a data-set of experimentally characterized transmembrane topologies. In Silico Biol, 2, 1-15. 45 Ikeda, M., Arai, M., Okuno, T. and Shimizu, T. (2003) TMPDB: a database of experimentally-characterized transmembrane topologies. Nucl Acids Res, 31, 406-409. Johnson, A.E. and van Waes, M.A. (1999) The translocon: A dynamic gateway at the ER membrane. Annu Rev Cell Dev Biol, 15, 799-842. Johnson, R.M., Heslop, C.L. and Deber, C.M. (2004) Hydrophobic helical hairpins: design and packing interactions in membrane environments. Biochemistry, 43, 14361-14369. Jones, D.T., Taylor, W.R. and Thornton, J.M. (1994) A model recognition approach to the prediction of all-helical membrane protein structure and topology. Biochemistry, 33, 30383049. Kaback, H.R., Voss, J. and Wu, J.H. (1997) Helix packing in polytopic membrane proteins: The lactose permease of Escherichia coli. Curr Opin Struct Biol, 7, 537-542. Kanki, T., Young, M.T., Sakaguchi, M., Hamasaki, N. and Tanner, M.J.A. (2003) The N-terminal region of the transmembrane domain of human erythrocyte band 3 Residues critical for membrane insertion and transport activity. J Biol Chem, 278, 5564-5573. Keenan, R.J., Freymann, D.M., Stroud, R.M. and Walter, P. (2001) The signal recognition particle. Annual Review of Biochemistry, 70, 755775. Keenan, R.J., Freymann, D.M., Walter, P. and Stroud, R.M. (1998) Crystal structure of the signal sequence binding subunit of the signal recognition particle. Cell, 94, 181-191. Krantz, B.A., Melnyk, R.A., Zhang, S., Juris, S.J., Lacy, D.B., Wu, Z., Finkelstein, A. and Collier, R.J. (2005) A phenylalanine clamp catalyzes protein translocation through the anthrax toxin pore. Science, 309, 777-781. Krogh, A., Larsson, B., von Heijne, G. and Sonnhammer, E. (2001) Predicting transmembrane protein topology with a hidden Markov model. Application to complete genomes. J Mol Biol, 305, 567-580. Kuroiwa, T., Sakaguchi, M., Omura, T. and Mihara, K. (1996) Reinitiation of protein translocation across the endoplasmic reticulum membrane for the topogenesis of multispanning membrane proteins. J Biol Chem, 271, 6423-6428. Kutay, U., Ahnert-Hilger, G., Hartmann, E., Wiedenmann, B. and Rapoport, T. (1995) Transport route for synaptobrevin via a novel pathway for insertion into the endoplasmic reticulum. EMBO J, 14, 217-223. Kutay, U., Hartmann, E. and Rapoport, T.A. (1993) A class of membrane proteins with a Cterminal anchor. Trends Cell Biol, 3, 72-75. Laird, V. and High, S. (1997) Discrete crosslinking products identified during membrane protein biosynthesis. J Biol Chem, 272, 19831989. Landau, E.M. and Rosenbusch, J.P. (1996) Lipidic cubic phases: a novel concept for the crystallization of membrane proteins. Proc Natl Acad Sci U S A, 93, 14532-14535. Lanyi, J.K. (2004) X-ray diffraction of bacteriorhodopsin photocycle intermediates. Mol Membr Biol, 21, 143-150. Kiefer, D., Hu, X.T., Dalbey, R. and Kuhn, A. (1997) Negatively charged amino acid residues play an active role in orienting the Secindependent Pf3 coat protein in the Escherichia coli inner membrane. EMBO J., 16, 2197-2204. Laude, A.J. and Prior, I.A. (2004) Plasma membrane microdomains: organization, function and trafficking. Mol Membr Biol, 21, 193-205. Kopito, R.R. (1999) Biosynthesis and degradation of CFTR. Physiol Rev, 79, S167-173. Le Gall, S., Neuhof, A. and Rapoport, T. (2004) The endoplasmic reticulum membrane 46 Marika Cassel is permeable to small molecules. Mol Biol Cell, 15, 447-455. Lecomte, F.J., Ismail, N. and High, S. (2003) Making membrane proteins at the mammalian endoplasmic reticulum. Biochem Soc Trans, 31, 1248-1252. Levy, D. (1996) Membrane proteins which exhibit multiple topological orientations. Essays Biochem, 31, 49-60. Liao, S., Lin, J., Do, H. and Johnson, A. (1997) Both lumenal and cytosolic gating of the aqueous ER translocon pore are regulated from inside the ribosome during membrane protein integration. Cell, 90, 31-41. Liljeström, P. and Garoff, H. (1991) Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J Virol, 65, 147-154. Manoil, C., Mekalanos, J.J. and Beckwith, J. (1990) Alkaline Phosphatase Fusions - Sensors of Subcellular Location. J Bact, 172, 515-518. Manting, E.H., van der Does, C., Remiy, H., Engel, A. and Driessen, A.J.M. (2000) SecYEG assembles into a tetramer to form the active protein translocation channel. EMBO J, 19, 852861. Martoglio, B., Hofmann, M.W., Brunner, J. and Dobberstein, B. (1995) The proteinconducting channel in the membrane of the endoplasmic reticulum is open laterally toward the lipid bilayer. Cell, 81, 207-214. Matlack, K., Mothes, W. and Rapoport, T. (1998) Protein translocation: Tunnel vision. Cell, 92, 381-390. McCormick, P.J., Miao, Y., Shao, Y., Lin, J. and Johnson, A.E. (2003) Cotranslational protein integration into the ER membrane is mediated by the binding of nascent chains to translocon proteins. Mol Cell, 12, 329-341. Loo, M.A., Jensen, T.J., Cui, L., Hou, Y., Chang, X.B. and Riordan, J.R. (1998) Perturbation of Hsp90 interaction with nascent CFTR prevents its maturation and accelerates its degradation by the proteasome. EMBO J, 17, 6879-6887. McGovern, K. and Beckwith, J. (1991) Membrane insertion of the Escherichia coli MalF protein in cells with impaired secretion machinery. J Biol Chem, 266, 20870-20876. Ma, C. and Chang, G. (2004) Structure of the multidrug resistance efflux transporter EmrE from Escherichia coli. Proc Natl Acad Sci U S A, 101, 2852-2857. Meacham, G.C., Lu, Z., King, S., Sorscher, E., Tousson, A. and Cyr, D.M. (1999) The Hdj2/Hsc70 chaperone pair facilitates early steps in CFTR biogenesis. EMBO J, 18, 1492-1505. MacKenzie, K.R., Prestegard, J.H. and Engelman, D.M. (1997) A transmembrane helix dimer: structure and implications. Science, 276, 131-133. Meacock, S.L., Greenfield, J.J. and High, S. (2000) Protein targeting and translocation at the endoplasmic reticulum membrane-through the eye of a needle? Essays Biochem, 36, 1-13. Manoil, C. (1991) Analysis of membrane protein topology using alkaline phosphatase and ß-galactosidase gene fusions. Methods in Cell Biol, 34, 61-75. Manoil, C. and Beckwith, J. (1986) A genetic approach to analyzing membrane protein topology. Science, 233, 1403-1408. Meacock, S.L., Lecomte, F.J., Crawshaw, S.G. and High, S. (2002) Different transmembrane domains associate with distinct endoplasmic reticulum components during membrane integration of a polytopic protein. Mol Biol Cell, 13, 4114-4129. Melén, K., Krogh, A. and von Heijne, G. (2003) Reliability measures for membrane 47 protein topology prediction algorithms. J Mol Biol, 327, 735-744. Menetret, J.F., Hegde, R.S., Heinrich, S.U., Chandramouli, P., Ludtke, S.J., Rapoport, T.A. and Akey, C.W. (2005) Architecture of the ribosome-channel complex derived from native membranes. J Mol Biol, 348, 445-457. Menetret, J.F., Neuhof, A., Morgan, D.G., Plath, K., Radermacher, M., Rapoport, T.A. and Akey, C.W. (2000) The structure of ribosome-channel complexes engaged in protein translocation. Mol Cell, 6, 1219-1232. Meyer, T.H., Menetret, J.F., Breitling, R., Miller, K.R., Akey, C.W. and Rapoport, T.A. (1999) The bacterial SecY/E translocation complex forms channel-like structures similar to those of the eukaryotic Sec61p complex. J Mol Biol, 285, 1789-1800. Monné, M., Hermansson, M. and von Heijne, G. (1999a) A turn propensity scale for transmembrane helices. J Mol Biol, 288, 141145. Murata, K., Mitsuoka, K., Hirai, T., Walz, T., Agre, P., Heymann, J.B., Engel, A. and Fujiyoshi, Y. (2000) Structural determinants of water permeation through aquaporin-1. Nature, 407, 599-605. Möller, J. and Rydström, J. (1999) The membrane topology of proton-pumping Escherichia coli transhydrogenase determined by cysteine labeling. J Biol Chem, 274, 1907219080. Möller, S., Croning, M. and Apweiler, R. (2001) Evaluations of methods for the predictive evaluation of membrane spanning regions. Bioinformatics, 17, 646-653. Möller, S., Kriventseva, E. and Apweiler, R. (2000) A collection of well-characterised integral membrane proteins. Bioinformatics, 16, 1159-1160. Nilsson, I., Johnson, A.E. and von Heijne, G. (2003) How hydrophobic is alanine? J Biol Chem, 278, 29389-29393. Monné, M., Nilsson, I., Elofsson, A. and von Heijne, G. (1999b) Turns in transmembrane helices: Determination of the minimal length of a ''helical hairpin'' and derivation of a finegrained turn propensity scale. J Mol Biol, 293, 807-814. Nilsson, I., Sääf, A., Whitley, P., Gafvelin, G., Waller, C. and von Heijne, G. (1998) Prolineinduced disruption of a transmembrane αhelix in its natural environment. J Mol Biol, 284, 1165-1175. Mothes, W., Heinrich, S., Graf, R., Nilsson, I., von Heijne, G., Brunner, J. and Rapoport, T. (1997) Molecular mechanisms of membrane protein integration into the endoplasmic reticulum. Cell, 89, 523-533. Nilsson, I. and von Heijne, G. (1993) Determination of the distance between the oligosaccharyltransferase active site and the endoplasmic reticulum membrane. J Biol Chem, 268, 5798-5801. Mothes, W., Jungnickel, B., Brunner, J. and Rapoport, T.A. (1998) Signal sequence recognition in cotranslational translocation by protein components of the endoplasmic reticulum membrane. J Cell Biol, 142, 355-364. Nilsson, I.M. and von Heijne, G. (1990) Finetuning the Topology of a Polytopic Membrane Protein. Role of Positively and Negatively Charged Residues. Cell, 62, 1135-1141. Mothes, W., Prehn, S. and Rapoport, T.A. (1994) Systematic probing of the environment of a translocating secretory protein during translocation through the ER membrane. EMBO J, 13, 3937-3982. Nilsson, J., Persson, B. and von Heijne, G. (2005) Comparative analysis of amino acid distributions in integral membrane proteins from 107 genomes. Proteins, 60, 606-616. 48 Marika Cassel Oliver, J., Jungnickel, B., Görlich, D., Rapoport, T. and High, S. (1995) The Sec61 complex is essential for the insertion of proteins into the membrane of the endoplasmic reticulum. FEBS Lett, 362, 126130. Opella, S.J. (1997) NMR and membrane proteins. Nature Struct Biology, 4, 845-848. Orlandini, E., Seno, F., Banavar, J.R., Laio, A. and Maritan, A. (2000) Deciphering the folding kinetics of transmembrane helical proteins. Proc Natl Acad Sci U S A, 97, 1422914234. Paetzel, M., Dalbey, R.E. and Strynadka, N.C.J. (1998) Crystal structure of a bacterial signal peptidase in complex with a β-lactam inhibitor. Nature, 396, 186-190. Palade, G. (1975) Intracellular aspects of the process of protein synthesis. Science, 189, 347358. Pervushin, K., Riek, R., Wider, G. and Wuthrich, K. (1997) Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc Natl Acad Sci USA, 94, 12366-12371. Pool, M.R., Stumm, J., Fulga, T.A., Sinning, I. and Dobberstein, B. (2002) Distinct modes of signal recognition particle interaction with the ribosome. Science, 297, 1345-1348. Popot, J.L. and Engelman, D.M. (1990) Membrane protein folding and oligomerization - The 2-stage model. Biochemistry, 29, 4031-4037. Prinz, W.A. and Beckwith, J. (1994) Gene fusion analysis of membrane protein topology: A direct comparison of alkaline phosphatase and ß-lactamase fusions. J Bacteriol, 176, 64106413. Rapoport, T.A., Goder, V., Heinrich, S.U. and Matlack, K.E. (2004) Membrane-protein integration and the role of the translocation channel. Trends Cell Biol, 14, 568-575. Rapoport, T.A., Jungnickel, B. and Kutay, U. (1996) Protein transport across the eukaryotic endoplasmic reticulum and bacterial inner membranes. Annu Rev Biochem, 65, 271-303. Rapp, M., Drew, D.E., Daley, D.O., Nilsson, J., Carvalho, T., Melén, K., de Gier, J.W. and von Heijne, G. (2004) Experimentally based topology models for E. coli inner membrane proteins. Prot Sci, 13, 937-945. Roberts, C.H. and Chlebowski, J.F. (1984) "PhoA+T X-100+trypsin". J Biol Chem, 259, 729733. Rost, B., Fariselli, P. and Casadio, R. (1996) Topology prediction for helical transmembrane proteins at 86% accuracy. Prot Sci, 5, 1704-1718. Sadlish, H., Pitonzo, D., Johnson, A.E. and Skach, W.R. (2005) Sequential triage of transmembrane segments by Sec61α during biogenesis of a native multispanning membrane protein. Nat Struct Mol Biol, 12,870878. Sato, M., Hresko, R. and Mueckler, M. (1998) Testing the charge difference hypothesis for the assembly of a eucaryotic multispanning membrane protein. J Biol Chem, 273, 2520325208. Sato, M. and Mueckler, M. (1999) A conserved amino acid motif (R-X-G-R-R) in the GLUT1 glucose transporter is an important determinant of membrane topology. J Biol Chem, 274, 24721-24725. Schatz, G. and Dobberstein, B. (1996) Common principles of protein translocation across membranes. Science, 271, 1519-1526. Scheuring, S., Muller, D.J., Stahlberg, H., Engel, H.A. and Engel, A. (2002) Sampling the conformational space of membrane protein surfaces with the AFM. Eur Biophys J, 31, 172178. 49 Schiffer, M., Chang, C.H. and Stevens, F.J. (1992) The Functions of Tryptophan Residues in Membrane Proteins. Prot Eng, 5, 213-214. Schirmer, T., Keller, T. A., Wang, Y. F. and Rosenbusch, J. P. (1995) Structural basis for sugar translocation through maltoporin channels at 3.1 Ångstrom resolution Science, 267, 512-514 Schulz, G.E. (2000) β-Barrel membrane proteins. Curr Opin Struct Biol, 10, 443-447. Senes, A., Ubarretxena-Belandia, I. and Engelman, D.M. (2001) The Calpha ---H...O hydrogen bond: a determinant of stability and specificity in transmembrane helix interactions. Proc Natl Acad Sci U S A, 98, 90569061. Simon, S.M. and Blobel, G. (1991) A ProteinConducting Channel in the Endoplasmic Reticulum. Cell, 65, 371-380. Simons, K. and Ikonen, E. (1997) Functional rafts in cell membranes. Nature, 387, 569-572. Simons, K. and Ikonen, E. (2000) How cells handle cholesterol. Science, 290, 1721-1726. Singer, S.J. and Nicolson, G.L. (1972) The fluid mosaic model of the structure of cell membranes. Science, 175, 720-731. Song, L., Hobaugh, M.R., Shustak, C., Ceheley, S., Bayley, H. and Gouaux, J.E. (1996) Structure of Staphylococcal αhemolysin, a heptameric transmembrane pore. Science, 274, 1859-1866. Sonnhammer, E., von Heijne, G. and Krogh, A. (1998) A hidden Markov model for predicting transmembrane helices in protein sequences. Intell Syst Mol Biol, 6, 175-182. Stroud, R. (1995) Ion channel forming colicins. Curr Opin Struct Biol, 5, 514-520. Subramaniam, S. and Henderson, R. (1999) Electron crystallography of bacteriorhodopsin with millisecond time resolution. J Struct Biol, 128, 19-25. Swanton, E., High, S. and Woodman, P. (2003) Role of calnexin in the glycan-independent quality control of proteolipid protein. EMBO J, 22, 2948-2958. Sääf, A., Johansson, M., Wallin, E. and von Heijne, G. (1999) Divergent evolution of membrane protein topology: The Escherichia coli RnfA and RnfE homologues. Proc Natl Acad Sci USA, 96, 8540-8544. Tadayyon, M., Zhang, Y., Gnaneshan, S., Hunt, L., Mehraein-Ghomi, F. and BroomeSmith, J.K. (1992) beta-Lactamase fusion analysis of membrane protein assembly. Biochem Soc Trans, 20, 598-601. Tamm, L.K., Abildgaard, F., Arora, A., Blad, H. and Bushweller, J.H. (2003) Structure, dynamics and function of the outer membrane protein A (OmpA) and influenza hemagglutinin fusion domain in detergent micelles by solution NMR. FEBS Lett, 555, 139143. Tamm, L.K., Arora, A. and Kleinschmidt, J.H. (2001) Structure and assembly of beta-barrel membrane proteins. J Biol Chem, 276, 3239932402. Tenson, T. and Ehrenberg, M. (2002) Regulatory nascent peptides in the ribosomal tunnel. Cell, 108, 591-594. Thrift, R.N., Andrews, D.W., Walter, P. and Johnson, A.E. (1991) A Nascent Membrane Protein Is Located Adjacent to ER Membrane Proteins Throughout Its Integration and Translation. J Cell Biol, 112, 809-821. Traxler, B., Boyd, D. and Beckwith, J. (1993) The topological analysis of integral cytoplasmic membrane proteins. J Membr Biol, 132, 1-11. Tusnady, G.E. and Simon, I. (1998) Principles governing amino acid composition of integral 50 Marika Cassel membrane proteins: Application to topology prediction. J Mol Biol, 283, 489-506. Tusnady, G.E. and Simon, I. (2001) The HMMTOP transmembrane topology prediction server. Bioinformatics, 17, 849-850. Weiss, M.S., Abele, U., Weckesser, J., Welte, W., Schiltz, E. and Schulz, G.E. (1991) Molecular architecture and electrostatic properties of a bacterial porin. Science, 254, 1627-1630. Ubarretxena-Belandia, I., Baldwin, J.M., Schuldiner, S. and Tate, C.G. (2003) Threedimensional structure of the bacterial multidrug transporter EmrE shows it is an asymmetric homodimer. EMBO J, 22, 61756181. Vereb, G., Szollosi, J., Matko, J., Nagy, P., Farkas, T., Vigh, L., Matyus, L., Waldmann, T.A. and Damjanovich, S. (2003) Dynamic, yet structured: The cell membrane three decades after the Singer-Nicolson model. Proc Natl Acad Sci U S A, 100, 8053-8058. Ubarretxena-Belandia, I. and Tate, C.G. (2004) New insights into the structure and oligomeric state of the bacterial multidrug transporter EmrE: an unusual asymmetric homo-dimer. FEBS Lett, 564, 234-238. Wessels, H.P. and Spiess, M. (1988) Insertion of a multispanning membrane protein occurs sequentially and requires only one signal sequence. Cell, 55, 61-70. Wahlberg, J.M. and Spiess, M. (1997) Multiple determinants direct the orientation of signalanchor proteins: the topogenic role of the hydrophobic signal domain. J Cell Biol, 137, 555-562. Wallin, E., Tsukihara, T., Yoshikawa, S., von Heijne, G. and Elofsson, A. (1997) Architecture of helix bundle membrane proteins: An analysis of cytochrome c oxidase from bovine mitochondria. Protein Sci, 6, 808815. Wallin, E. and von Heijne, G. (1998) Genomewide analysis of integral membrane proteins from eubacterial, archaean, and eukaryotic organisms. Protein Sci, 7, 1029-1038. Walter, P. and Johnson, A.E. (1994) Signal sequence recognition and protein targeting to the endoplasmic reticulum membrane. Annu Rev Cell Biol, 10, 87-119. Van den Berg, B., Clemons, W.M., Jr., Collinson, I., Modis, Y., Hartmann, E., Harrison, S.C. and Rapoport, T.A. (2004) Xray structure of a protein-conducting channel. Nature, 427, 36-44. van Meer, G. (1989) Lipid traffic in animal cells. Annu Rev Cell Biol, 5, 247-275. Whitley, P., Grahn, E., Kutay, U., Rapoport, T. and von Heijne, G. (1996) A 12 residues long poly-leucine tail is sufficient to anchor syntaptobrevin to the ER membrane. J Biol Chem, 271, 7583-7586. Wolfe, P.B., Wickner, W. and Goodman, J.M. (1983) Sequence of the leader peptidase gene of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J Biol Chem, 258, 12073-12080. von Heijne, G. (1986) The distribution of positively charged residues in bacterial inner membrane proteins correlates with the transmembrane topology. EMBO J, 5, 3021-3027. von Heijne, G. (1989) Control of topology and mode of assembly of a polytopic membrane protein by positively charged residues. Nature, 341, 456-458. von Heijne, G. (1990) The signal peptide. J Membr Biol, 115, 195-201. von Heijne, G. (1991) Proline kinks in transmembrane α-helices. J Mol Biol, 218, 499503. von Heijne, G. (1992) Membrane protein structure prediction - Hydrophobicity analysis 51 and the positive-inside rule. J Mol Biol, 225, 487-494. secretory proteins differ in FRET-detected folding. Cell, 116, 725-736. von Heijne, G. (1994) Membrane proteins: From sequence to structure. Annu Rev Biophys Biomol Struct, 23, 167-192. Zhou, F.X., Cocco, M.J., Russ, W.P., Brunger, A.T. and Engelman, D.M. (2000) Interhelical hydrogen bonding drives strong interactions in membrane proteins. Nature Struct Biol, 7, 154-160. von Heijne, G. (1996) Principles of membrane protein assembly and structure. Progr Biophys Mol Biol, 66, 113-139. von Heijne, G. (1997) Getting greasy: How transmembrane polypeptide segments integrate into the lipid bilayer. Mol Microbiol, 24, 249-253. von Heijne, G. and Gavel, Y. (1988) Topogenic signals in integral membrane proteins. Eur J Biochem, 174, 671-678. Woolhead, C.A., McCormick, P.J. and Johnson, A.E. (2004) Nascent membrane and Zhou, F.X., Merianos, H.J., Brunger, A.T. and Engelman, D.M. (2001) Polar residues drive association of polyleucine transmembrane helices. Proc Natl Acad Sci USA, 98, 2250-2255. Zopf, D., Bernstein, H.D., Johnson, A.E. and Walter, P. (1990) The Methionine-Rich Domain of the 54 kd Protein Subunit of the Signal Recognition Particle Contains an RNA Binding Site and Can Be Crosslinked to a Signal Sequence. EMBO J, 9, 4511-4517. I Article No. jmbi.1999.2657 available online at http://www.idealibrary.com on J. Mol. Biol. (1999) 288, 141–145 A Turn Propensity Scale for Transmembrane Helices Magnus MonneÂ, Marika Hermansson and Gunnar von Heijne* Department of Biochemistry Stockholm University S-106 91 Stockholm, Sweden Using a model protein with a 40 residue hydrophobic transmembrane segment, we have measured the ability of all the 20 naturally occurring amino acids to form a tight turn when placed in the middle of the hydrophobic segment. Turn propensities in a transmembrane helix are found to be markedly different from those of globular proteins, and in most cases correlate closely with the hydrophobicity of the residue. The turn propensity scale may be used to improve current methods for membrane protein topology prediction. # 1999 Academic Press *Corresponding author Keywords: membrane protein; turn propensity; secondary structure; transmembrane helix Introduction The formation of tight turns in globular proteins has been studied for decades, both experimentally and by statistical analysis of known structures, and reliable turn propensity scales have been established (Creighton, 1993; von Heijne, 1987). Remarkably, however, essentially nothing is known about the residue characteristics responsible for the formation of tight turns between transmembrane a-helices in integral membrane proteins. This is due in part to the paucity of high-resolution structural information for this class of proteins, but it is nevertheless surprising that no direct means of measuring turn propensities in membrane proteins has been established. We recently developed a simple experimental system for the analysis of turn formation in transmembrane helices embedded in the membrane of the endoplasmic reticulum (Nilsson & von Heijne, 1998), and could show that a single proline residue placed near the middle of a 40 residue poly(Leu) transmembrane helix ef®ciently converts the poly(Leu) segment from a single, long transmembrane helix to a tightly spaced pair of transmembrane helices (a ``helical hairpin''). We have now used this system to measure the turn propensities for all the 20 naturally occurring amino acid residues, and present the ®rst experimental propensity scale for the formation of tight turns between transmembrane a-helices. This scale is shown to differ in important respects from turn propensities in globular proteins, and in general Abbreviations used: BSA, bovine serum albumin. E-mail address of the corresponding author: [email protected] 0022-2836/99/160141±05 $30.00/0 correlates better with residue hydrophobicity than with ``classical'' secondary structure propensities. Results A system for measuring turn propensities in transmembrane helices under in vivolike conditions For these studies, we have used the well-characterized Escherichia coli protein Lep, which contains two transmembrane helices (H1 and H2) and a large C-terminal domain (P2). When expressed in vitro in the presence of dog pancreas microsomes, Lep has been shown to insert into the microsomal membrane with both the N and C termini on the luminal side (Nilsson & von Heijne, 1993), i.e. in the same orientation as it normally inserts into the inner membrane of E. coli (Wolfe et al., 1983). Translocation of the P2 domain to the lumenal side is conveniently assayed by the glycosylation of a unique acceptor site for N-linked glycosylation (Asn-Ser-Thr) placed 20 residues downstream of H2 (Figure 1(a)). An advantage of this approach is that the microsomal in vitro system closely mimics the conditions of in vivo membrane protein assembly into the endoplasmic reticulum membrane. For the studies reported here, H2 was replaced by a 40 residue poly(Leu) segment (including one Val) ¯anked by four lysine residues on the N-terminal end and by a Gln-Gln-Gln-Pro stretch on the C-terminal end. Given that typical transmembrane helices are 20-30 residues long (Bowie, 1997), this poly(Leu) stretch should, in principle, be able to form either one long or two closely spaced transmembrane helices. Indeed, the poly(Leu) segment has previously been shown to insert into the micro# 1999 Academic Press 142 Membrane Protein Turn Propensities somal membrane as a single transmembrane helix with the P1 loop in the cytoplasm and the P2 domain in the microsomal lumen, and it was observed that the introduction of a Pro residue near the middle of the poly(Leu) stretch results in the formation of a helical hairpin in the membrane and localization of the P2 domain to the cytoplasmic side (Nilsson & von Heijne, 1998). Using the same poly(Leu) construct, we have now measured turn propensities for all the 20 naturally occurring amino acids by substituting each residue (X) for Leu22 near the middle of the poly(Leu) segment, and expressing the L22 ! X constructs in vitro in the presence of dog pancreas microsomes (Figure 1(b)). Based on the quanti®cation of the glycosylation ef®ciencies given in Figure 1(c), a scale of turn propensities can be derived from this set of data (Table 1). To rule out that lack of glycosylation is the result of inef®cient insertion of the poly(Leu) stretch into the microsomal membrane rather than formation of a helical hairpin structure, a segment encompassing H1 and part of the P1 domain (residues 5-46) was deleted from two poorly glycosylated constructs (L22 ! R and L22 ! E), and membrane insertion of the L22 ! R(5-46) and L22 ! E(5-46) constructs was monitored by alkaline extraction of the microsomes. This treatment is known to remove peripherally bound membrane proteins, but leaves properly inserted transmembrane proteins in the membrane pellet (Fujiki et al., 1982). In addition, the P2 domain (residues 79-323) was expressed alone to make sure that any membrane association observed for the 5-46 constructs was due only to the poly(Leu) segment. As seen in Figure 2, the two 5-46 constructs remained with the membrane pellet, whereas the P2 domain, Table 1. Turn propensities for amino acid residues in a transmembrane helix Residue Figure 1. (a) Model protein used in this study. The H2 transmembrane segment in Lep was replaced with a stretch of residues of the general design LIK4L21XL7VL10Q3P, where X is one of the 20 naturally occurring amino acid residues. A glycosylation acceptor site was placed 20 residues downstream of H2 (counting from the ®rst Gln residue after the hydrophobic stretch). Depending on the luminal or cytoplasmic localization of the P2 domain, the glycosylation acceptor site will either be modi®ed (Y) or not ( ). Note that the tilted conformation of the model single-spanning transmembrane H2 helix has not been experimentally proven. (b) The indicated L22 ! X mutants were translated in vitro in the presence of rough microsomes and analyzed by SDS-PAGE. Black and white dots indicate the glycosylated and non-glycosylated forms of the proteins, respectively. (c) Quanti®cation of the gels shown in (b). The percentage glycosylation was calculated as 100 I/(I Iÿ), where I (Iÿ) is the intensity of the glycosylated (non-glycosylated) band. From duplicate experiments on all the 20 constructs, the typical error in the determination of glycosylation ef®ciency was 45 % (bars), except for W where the error was 10 %. A C D E F G H I K L M N P Q R S T V W Y Turn propensity 0.5 0.6 1.6 1.6 0.4 1.3 1.6 0.6 1.6 0.4 0.5 1.7 1.7 1.6 1.7 0.7 0.4 0.5 0.7 0.6 The turn propensity is de®ned as (1 ÿ fX)/m(1 ÿ fX), where fX is the fraction of glycosylated molecules in the L22 ! X mutant and m(1 ÿ fX) is the mean value of 1 ÿ fX over all 20 residues. The typical error in the propensity values is 0.05. Membrane Protein Turn Propensities 143 Figure 2. Alkaline extraction of the L22 ! R(5-46) (lanes 2, 5, 8), L22 ! E(5-46) (lanes 3, 6, 9), and P2 (lanes 4, 7, 10) constructs. Constructs were translated in vitro in the absence (lanes 2-4) or presence (lanes 5-10) of rough microsomes. In lanes 5-10, microsomes were subjected to a sodium carbonate wash before loading onto the gel. p, pellet; s, supernatant. Black and white dots indicate glycosylated and non-glycosylated forms of the proteins, respectively. The 5-46 constructs lack H1 and about two-thirds of the P1 domain, and the P2 construct lacks residues 2-78, i.e. the entire H1-P1-H2 domain. Lane 1 contains molecular mass markers as indicated. when expressed alone, was found exclusively in the supernatant. We conclude that the degree of glycosylation seen for the different L22 ! X mutants accurately re¯ects the fraction of molecules that insert with a single transmembrane segment versus a helical hairpin, and that it can thus be used as a basis for the turn propensity scale presented in Table 1. Discussion We have used a simple in vivo-like system where the membrane topology adopted by a 40 residue long model transmembrane segment can be used to directly infer a turn propensity scale relevant for transmembrane a-helices. As seen in Figure 1(c), hydrophobic residues (L, F, A, Y, V, and I) do not induce a turn in the poly(Leu) helix, whereas charged or highly polar residues do. In addition, the two classical helix breakers Pro and Gly both induce a turn (Pro somewhat more ef®ciently than Gly). Interestingly, despite their polar nature, Ser and Thr do not have high turn propensities. It is known from helices in globular proteins that Ser, Thr, and Cys side-chains can form hydrogen bonds to the polypeptide backbone (Gray & Matthews, 1984), which might increase their apparent hydrophobicity when present in a transmembrane helix, making turn formation less favorable. Consistent with this, Ser and Thr are rather frequently found in transmembrane helices, in contrast to Asn and Gln (von Heijne, 1992). Finally, Trp is known to have the strongest preference for the lipid-water interface of all the amino acid residues (Wimley & White, 1996), which may explain its somewhat higher turn propensity compared to, e.g. Phe and Tyr. Although the data reported in Figure 1(c) and Table 1 are largely consistent with a two-tier system where residues have either a high or a low turn propensity, it may be possible to provide a better discrimination between different residues in the transition region between high and low turn propensity by inserting pairs of residues in the middle of the model transmembrane segment; such studies are in progress. The turn propensity scale derived here deviates signi®cantly from the turn propensities observed in globular proteins (Figure 3, top panel). Thus, while the charged and highly polar amino acids all have high turn propensities in the transmembrane helix context, this is not the case in globular proteins. Ser, in contrast, has a rather high turn propensity in globular proteins but not in transmembrane helices. Pro and Gly are turn promoters in both contexts. The correlation between the turn propensity scale and the so-called interface hydrophobicity scale (Wimley & White, 1996) is not very strong (Figure 3, middle panel), whereas the correlation with helical propensities in n-butanol (Liu & Deber, 1998) is somewhat better (Figure 3, bottom panel). Good correlations are also obtained with some statistically de®ned hydrophobicity scales (von Heijne, 1992; results not shown). In summary, we have measured the turn propensities for all the 20 naturally occurring amino acids placed in the middle of a poly(Leu) segment that is long enough to form either a single or two closely spaced transmembrane helices. Most of our results can be explained by hydrophobicity: all hydrophobic residues prefer the membrane environment over the membrane-water interface region (Wimley & White, 1996), and have low turn propensities. Conversely, the charged and highly polar residues induce turn formation in order to avoid the membrane interior. Thus, intrinsic conformational preferences become largely irrelevant in the context of a transmembrane helix, as observed previously in peptide studies of a versus b-structure formation in water, detergent, and lipid vesicle environments (Deber & Li, 1995; Li & Deber, 1994). In the context of the microsomal membrane, Pro behaves as a strongly polar residue, presumably because its inclusion in a transmembrane helix necessitates the disruption of at 144 least one hydrogen bond. Perhaps the most surprising result is that Gly has such a high turn propensity, since it neither has a polar side-chain, nor disrupts backbone hydrogen bonds when in a Membrane Protein Turn Propensities helix. Apparently, its exceptional conformation ¯exibility suf®ces to make the turn conformation preferred over the intact transmembrane helix. A possible mechanism for turn formation in a transmembrane helix during its insertion into the membrane of the endoplasmic reticulum has been suggested previously (Nilsson & von Heijne, 1998). Finally, we anticipate that the turn propensity scale presented here will improve our ability to distinguish between cases of a single long and two closely spaced transmembrane helices when predicting membrane protein topology from amino acid sequence information. Materials and Methods Enzymes and chemicals Unless otherwise stated, all enzymes were from Promega. T7 DNA polymerase, [35S]Met, ribonucleotides, deoxyribonucleotides, dideoxyribonucleotides, and the cap analog m7G(50 )ppp(50 )G were from Amersham-Pharmacia (Uppsala, Sweden). Plasmid pGEM1, DTT, bovine serum albumin (BSA), SP6 RNA polymerase, RNasin and rabbit reticulocyte lysate were from Promega. Spermidine was from Sigma. Oligonucleotides were from Cybergene (Stockholm, Sweden). DNA manipulations Figure 3. Turn propensities in transmembrane helices are different from turn propensities in globular proteins. Top, the turn propensity scale from Table 1 is plotted against a typical turn propensity scale for globular proteins (Williams et al., 1987); middle, the interface hydrophobicity scale described by Wimley & White (1996); bottom, a scale of helical propensities in n-butanol (Liu & Deber, 1998). Correlation coef®cients are indicated in the respective panels. For cloning into and expression from the pGEM1 plasmid, the 50 end of the lep gene was modi®ed: ®rst, by the introduction of an XbaI site and; second, by changing the context 50 to the initiator ATG codon to a ``Kozak consensus'' sequence (Johansson et al., 1993; Kozak, 1989). Replacement of the H2 region in Lep was performed by ®rst introducing BclI and NdeI restriction sites in codons 59 and 80 ¯anking the H2 region, and then replacing the BclI-NdeI fragment by the appropriate doublestranded oligonucleotides. Residues 59-80 in H2 were replaced by poly(Leu) sequences of the general design LIK4L21XL7VL10Q3P, where X is one of the 20 naturally occurring amino acids; the Val residue results from the inclusion of a SpeI restriction site. The (5-46) and P2 constructs were made by deleting, respectively, residues 5-46 and 2-78 in Lep. Site-speci®c mutagenesis used to add BclI and NdeI restriction sites at the 30 and 50 ends of H2 in Lep and to introduce Asn-Ser-Thr acceptor sites for N-linked glycosylation was performed according to the Kunkel method (Geisselsoder et al., 1987; Kunkel, 1987) . Glycosylation acceptor sites were designed as described (Nilsson et al., 1994), i.e. by replacing three appropriately positioned codons with codons for the acceptor tripeptide Asn-Ser-Thr. For the L22 ! X substitutions, the QuickChange2 site-directed mutagenesis kit from Stratagene was used. Some of the primers were designed with a degenerate base in the second position of the codon for X in order to get more than one mutant from the primer pair. All mutants were con®rmed by DNA sequencing of plasmid using T7 DNA polymerase. Expression in vitro The constructs in pGEM1 were transcribed by SP6 RNA polymerase for one hour at 37 C. The transcription mixture was as follows: 1-5 mg DNA template, 5 ml of 10 SP6 H-buffer (400 mM Hepes-KOH (pH 7.4), Membrane Protein Turn Propensities 60 mM Mg acetate, 20 mM spermidine-HCl), 5 ml of 1 mg/ml BSA, 5 ml of 10 mM m7G(50 )ppp(50 )G, 5 ml of 50 mM DTT, 5 ml of rNTP mix (10 mM ATP, 10 mM CTP, 10 mM UTP, 5 mM GTP), 18.5 ml of H2O, 1.5 ml of RNase inhibitor (50 units), 0.5 ml of SP6 RNA polymerase (20 units). Translation was performed in reticulocyte lysate in the presence and absence of dog pancreas microsomes (LiljestroÈm & Garoff, 1991) . Sodium carbonate extraction of microsomes was carried out as described (Sakaguchi et al., 1987) . Translation products were analyzed by SDS-PAGE and gels were quanti®ed on a Fuji BAS1000 phosphoimager using the MacBAS 2.31 software. The glycosylation ef®ciency of a given mutant was calculated as the quotient between the intensity of the glycosylated band divided by the summed intensities of the glycosylated and non-glycosylated bands. Acknowledgments This work was supported by grants from the Swedish Cancer Foundation, the Swedish Natural and Technical Sciences Research Councils, and the GoÈran Gustafsson Foundation to G.v.H. References Bowie, J. U. (1997). Helix packing in membrane proteins. J. Mol. Biol. 272, 780-789. Creighton, C. (1993). Proteins: Structures and Molecular Properties, 2nd edit., W. H. Freeman & Co., New York. Deber, C. M. & Li, S. C. (1995). Peptides in membranes: helicity and hydrophobicity. Biopolymers, 37, 295318. Fujiki, Y., Hubbard, A. L., Fowler, S. & Lazarow, P. B. (1982). Isolation of intracellular membranes by means of sodium carbonate treatment. J. Cell Biol. 93, 97-102. Geisselsoder, J., Witney, F. & Yuckenberg, P. (1987). Ef®cient site-directed in vitro mutagenesis. BioTechniques, 5, 786-791. Gray, T. M. & Matthews, B. W. (1984). Intrahelical hydrogen bonding of serine, threonine and cysteine residues within alpha-helices and its relevance to membrane-bound proteins. J. Mol. Biol. 175, 75-81. Johansson, M., Nilsson, I. & von Heijne, G. (1993). Positively charged amino acids placed next to a signal sequence block protein translocation more ef®- 145 ciently in Escherichia coli than in mammalian microsomes. Mol. Gen. Genet. 239, 251-256. Kozak, M. (1989). Context effects and inef®cient initiation at non-AUG codons in eucaryotic cell-free translation systems. Mol. Cell. Biol. 9, 5073-5080. Kunkel, T. A. (1987). Rapid and ef®cient site-speci®c mutagenesis without phenotypic selection. Methods Enzymol. 154, 367-382. Li, S.-C. & Deber, C. M. (1994). A measure of helical propensity for amino acids in membrane environments. Nature Struct. Biol. 1, 368-373. LiljestroÈm, P. & Garoff, H. (1991). Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J. Virol. 65, 147-154. Liu, L. P. & Deber, C. M. (1998). Uncoupling hydrophobicity and helicity in transmembrane segments. aHelical propensities of the amino acids in non-polar environments. J. Biol. Chem. 273, 23645-23648. Nilsson, I. & von Heijne, G. (1993). Determination of the distance between the oligosaccharyltransferase active site and the endoplasmic reticulum membrane. J. Biol. Chem. 268, 5798-5801. Nilsson, I. & von Heijne, G. (1998). Breaking the camel's back: proline-induced turns in a model transmembrane helix. J. Mol. Biol. 284, 1185-1189. Nilsson, I., Whitley, P. & von Heijne, G. (1994). The C-terminal ends of internal signal and signal-anchor sequences are positioned differently in the ER translocase. J. Cell Biol. 126, 1127-1132. Sakaguchi, M., Mihara, K. & Sato, R. (1987). A short amino-terminal segment of microsomal cytochrome P-450 functions both as an insertion signal and as a stop-transfer sequence. EMBO J. 6, 2425-2431. von Heijne, G. (1987). Sequence Analysis in Molecular Biology: Treasure Trove or Trivial Pursuit?, Academic Press, San Diego. von Heijne, G. (1992). Membrane protein structure prediction - hydrophobicity analysis and the positiveinside rule. J. Mol. Biol. 225, 487-494. Williams, R., Chang, A., Juretic, D. & Loughran, S. (1987). Secondary structure predictions and medium range interactions. Biochim. Biophys. Acta, 916, 200204. Wimley, W. C. & White, S. H. (1996). Experimentally determined hydrophobicity scale for proteins at membrane interfaces. Nature Struct. Biol. 3, 842-848. Wolfe, P. B., Wickner, W. & Goodman, J. M. (1983). Sequence of the leader peptidase gene of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J. Biol. Chem. 258, 12073-12080. Edited by F. E. Cohen (Received 16 December 1998; received in revised form 22 February 1999; accepted 22 February 1999) II doi:10.1006/jmbi.2000.3933 available online at http://www.idealibrary.com on J. Mol. Biol. (2000) 301, 191–197 Formation of Cytoplasmic Turns between Two Closely Spaced Transmembrane Helices During Membrane Protein Integration into the ER Membrane Annika SaÈaÈf, Marika Hermansson and Gunnar von Heijne* Department of Biochemistry Stockholm University, S-106 91 Stockholm, Sweden The helical hairpin, two closely spaced transmembrane helices separated by a short turn, is a recurring structural element in integral membrane proteins, and may serve as a compact unit that inserts into the membrane en bloc. Previously, we have determined the propensities of the 20 natural amino acids, when present in the middle of a long hydrophobic stretch, to induce the formation of a helical hairpin with a lumenally exposed turn during membrane protein assembly into the endoplasmic reticulum membrane. Here, we present results from a similar set of measurements, but with the turn placed on the cytoplasmic side of the membrane. We ®nd that a signi®cantly higher number of turn-promoting residues need to be present to induce a cytoplasmic turn compared to a lumenal turn, and that, in contrast to the lumenal turn, the positively charged residues Arg and Lys are the strongest turn-promoters in cytoplasmic turns. These results suggest that the process of turn formation between transmembrane helices is different for lumenal and cytoplasmic turns. # 2000 Academic Press *Corresponding author Keywords: membrane protein; topology; glycosylation; transmembrane helix Introduction Transmembrane a-helices in integral membrane proteins often form ``helical hairpins'', i.e. closely spaced pairs of helices separated by a short turn. In principle, the short turn could be located either on the cytoplasmic or extra-cytoplasmic side of the membrane, although the term ``helical hairpin'' was ®rst used to describe helices connected by an extra-cytoplasmic turn (Engelman & Steitz, 1981). Given the current models for co-translational insertion of transmembrane helices into the membrane of the endoplasmic reticulum (ER) of eukaryotic cells (Do et al., 1996; Hamman et al., 1998; Liao et al., 1997; Mothes et al., 1997), it may well be that the ``rules'' for forming helical hairpins with cytoplasmic and extra-cytoplasmic turns are different. We recently measured the propensities of all the 20 naturally occurring amino acids to induce helical hairpin formation in a long hydrophobic stretch, and were able to propose a turn propensity scale for transmembrane helices (Monne et al., Abbreviations used: ER, endoplasmic reticulum; Lep, leader peptidase. E-mail address of the corresponding author: [email protected] 0022-2836/00/010191±7 $35.00/0 1999a,b; Nilsson & von Heijne, 1998). In those studies, the turn was always placed on the lumenal (extra-cytoplasmic) side of the ER membrane. We now report similar measurements of turn propensities, but with the turn placed on the cytoplasmic side of the membrane. We ®nd that the rank order is similar to that found for lumenal turns, but with weaker turn-promoting abilities. Thus, two or more turn-promoting residues are needed to fully induce a cytoplasmic turn, whereas a single residue with high turn propensity is enough to induce a lumenal turn, suggesting that it may be generally more dif®cult to induce the formation of helical hairpins with cytoplasmic turns. Results A model system for measuring cytoplasmic turn-propensities in transmembrane helices inserted into the ER membrane Leader peptidase (Lep) is a well-characterized inner membrane protein from Escherichia coli containing two transmembrane helices (H1 and H2) and a large C-terminal domain (P2) (Figure 1). Lep inserts into the inner E. coli membrane and into microsomes with the P1 loop facing the cytoplasm # 2000 Academic Press 192 Cytoplasmic Turns in Membrane Proteins Figure 1. Model proteins used in this study. Wild-type Lep (I) has two transmembrane segments (H1, H2), a cytoplasmic loop (P1), and a large extra-cytoplasmic domain (P2). To study turn-induction in a long hydrophobic transmembrane segment (H3), a 40 residue long poly(Leu) stretch was inserted in the middle of the P2 domain, and the glycosylation status of two Asn-X-Thr glycosylation acceptor sites ¯anking H3 was determined after in vitro translation in the presence of dog pancreas microsomes (glycosylated and non-glycosylated glycosylation acceptor sites are indicated by Y and Y, respectively). Constructs where H3 forms a single transmembrane segment (II) will be glycosylated on only one site, whereas those where H3 forms a helical hairpin (III) will be glycosylated on both sites. Turn-induction propensities were measured for all the 20 natural amino acids by introducing them either singly or as a pair in the middle of H3 (arrows). and the N terminus as well as the P2 domain in the extra-cytoplasmic space (Johansson et al., 1993; Wolfe et al., 1983). For the studies reported here, an extra transmembrane segment (H3) composed of a 40 residue long poly(Leu) stretch of the composition . . . E225-T-S-(L)40-R-S-V233 . . . (the numbers refer to residues in Lep) was introduced into the periplasmic P2 domain (Figure 1). This poly(Leu) stretch is long enough to form either a single transmembrane segment or a helical hairpin (Monne et al., 1999b). To study helical hairpin formation in H3, one or a pair of each of the 20 natural amino acids were introduced in the middle of the poly(Leu) stretch, and their ability to induce a cytoplasmic turn was assessed. For the analysis of the constructs, we used a standard in vitro translation system supplemented with dog pancreas microsomes (LiljestroÈm & Garoff, 1991). Since N-linked glycosylation is a reliable marker for translocation of a domain to the lumenal side of the microsomal membrane (Nilsson & von Heijne, 1993), we introduced AsnX-Thr glycosylation acceptor sites in positions 96 and 258 ¯anking the poly(Leu) segment. As shown in Figure 1, if H3 spans the membrane once only one glycosylation-acceptor site (in position 96) will be exposed to the lumen and glycosylated, whereas if H3 forms a helical hairpin with a cytoplasmic loop, both glycosylation acceptor sites will be exposed to the lumenal side of the membrane and hence glycosylated. Singly and doubly glycosylated molecules are easily distinguished by SDSPAGE. As an example, results for a series of constructs with zero to three proline residues replacing Leu20 in the middle of the H3 segment are shown in Figure 2. Glycosylation of constructs with 40 leucine residues in H3 (lanes 1 and 2), with 1-3 proline residues in the middle of H3 (lanes 3-5), and with a stretch of six lysine residues in the middle of H3 (lane 6). In lane 1, no rough microsomes were added (-RM), whereas rough microsomes were present during translation in lanes 2-6. Molecules with zero, one, and two modi®ed glycosylation sites are indicated. The fraction of molecules with two modi®ed glycosylation sites is given below each lane. Cytoplasmic Turns in Membrane Proteins 193 Figure 3. Ef®ciencies of cytoplasmic turn-induction for the 20 natural amino acids measured as the fraction of molecules with two modi®ed glycosylation sites obtained after in vitro translation in the presence of rough microsomes. Results for constructs with a single residue placed in the middle of H3 are shown by open bars, and for constructs with a pair of residues in the middle of H3 by ®lled bars. As shown in the cartoons, formation of a cytoplasmic turn in H3 results in doubly glycosylated molecules. Arrows indicate the position of the introduced residues. Figure 2, together with a construct with six lysine residues replacing Leu19-Leu22 in the middle of H3 (6K) that serves as a positive control for a fully induced, doubly glycosylated helical hairpin structure. A single Pro residue induces a helical hairpin in only 36 % of the molecules, and three Pro residues are needed to reach the same level of doubly glycosylated molecules, as seen for the 6K construct. This is in contrast to our previous studies of lumenal turns, where one Pro residue was suf®cent to induce full helical hairpin formation (Nilsson & von Heijne, 1998). Propensities of the 20 natural amino acids to induce the formation of a helical hairpin with a cytoplasmic loop Lep-derived constructs with the 40 residue long poly(Leu) H3 stretch or with one or a pair of each of the natural amino acid placed in the middle of H3, were expressed in vitro in the presence of dog pancreas microsomes. The fraction of glycosylated molecules with both acceptor sites modi®ed was taken as a measure of turn propensity. For all constructs, the fraction of non-glycosylated molecules was approximately the same (10-20 %), indicating that targeting and membrane insertion per se was not affected by the mutations in H3. As shown in Figure 3 (open bars), none of the singly inserted residues was able quantitatively to promote the formation of a cytoplasmic turn, although charged and polar amino acids all gave some degree of helical hairpin formation. Pairs of polar or charged residues had a higher turn-indu- cing effect, though in most cases not reaching full helical hairpin formation (®lled bars; note that the maximum level of glycosylation reached in our system is 80 %). Only the 2 K and 2R constructs reached the same degree of helical hairpin formation as the 6 K construct. All the hydrophobic residues plus Thr, Ser, Cys, Gly, Trp, and Tyr had low or no turn-inducing effects even when present in tandem. A comparison with the previously measured propensities to induce a lumenal turn (Monne et al., 1999a,b) is shown in Figure 4. Since the lumenal turn propensities were measured by placing the 40 residue long poly(Leu) stretch in the H2 position of Lep (c.f. Figure 5), low levels of glycosylation of the P2 domain correspond to ef®cient turn formation in this case. It is clear that turn propensities for both single and pairs of polar and charged residues are generally lower on the cytoplasmic side of the membrane. Moreover, although the rank order of turn propensities are similar for lumenal and cytoplasmic turns, the positively charged residues Arg and Lys have the highest cytoplasmic turn propensities while Pro and Asn are the strongest turn-inducers on the lumenal side. As a control to ensure that differences in the residues ¯anking the H2 segment used in our previous studies and the H3 segment used in this study do not compromise these comparisons, the H3 segment from the 1D, 1G, 1P, 1Q, 1S, and 1W mutants (see Figure 3) plus their ¯anking regions were introduced in place of a segment encompassing H2 and its ¯anking regions as shown in Figure 5 (top). 194 Cytoplasmic Turns in Membrane Proteins Figure 4. Comparison between ef®ciencies of helical hairpin formation with a lumenal turn (y-axis) and a cytoplasmic turn (x-axis). Lumenal turn propensities for single (left panel) and pairs (right panel) of residues were measured (Monne et al., 1999a,b) by placing a 40 residue long poly (Leu) stretch with the compo sitions . . . K4L21XL7VL10QQQP . . . or . . . K4L20X2L7VL10QQQP . . . (where X denotes one of the 20 natural amino acids) in the H2 position in Lep (c.f. Figure 5). Ef®cient formation of a lumenal turn in H2 results in low levels of glycosylation (i.e. high percentage of nonglycosylated molecules), whereas formation of a cytoplasmic turn in H3 results in high levels of doubly glycosylated molecules, as indicated by the cartoons. The diagonal lines represent equal ef®ciencies of lumenal and cytoplasmic turn formation (the maximum level of glycosylation in our system is 80-90 %). Note that not all pairs were tested in the H2 construct (right panel). From Figure 4, we expected that these residues should induce helical hairpin formation more ef®ciently on the lumenal side than on the cytoplasmic side. Indeed, as shown in Figure 5 (bottom), all six residues induced much higher degrees of helical hairpin formation when the H3 region was transplanted into the H2 context. Thus, differences in the regions ¯anking the poly(Leu) stretches in the H2 and H3 contexts are not responsible for the observed differences between lumenal and cytoplasmic turn-induction. We conclude that, although the effects of the different residues on helical hairpin formation on the lumenal versus cytoplasmic side of the microsomal membrane are qualitatively similar, a signi®cantly smaller number of turn-promoting residues are needed to induce a lumenal turn compared to a cytoplasmic turn between two closely spaced transmembrane helices. Discussion With the study reported here, we can now provide a rather complete picture of the turn-inducing propensities of all the 20 natural amino acids when present in the middle of a long hydrophobic transmembrane stretch inserted into the ER membrane. Not surprisingly, hydrophobic amino acids do not induce turns, charged residues are strong turn-promoters, and most polar residues are in-between. Among the non-charged residues, Pro and Asn, which are strong turn-promoters in globular proteins, have the highest turn propensities. This general picture holds both for turns on the cytoplasmic and lumenal sides of the ER membrane. However, there are some notable differences between turn induction on the two sides. Thus, a higher number of turn-promoting residues are required to induce a turn on the cytoplasmic side (Figure 4). Moreover, Arg and Lys are the two strongest turn-promoting residues on the cytoplasmic side, but not on the lumenal side (where Pro and Asn are stronger). Possibly, the relatively high cytoplasmic turn propensities for Arg and Lys is related to the ``positive inside'' rule, i.e. to the general tendency of positively charged amino acids to be enriched in cytoplasmic loops in integral membrane proteins (von Heijne & Gavel, 1988; Wallin & von Heijne, 1998). These differences between cytoplasmic and lumenal turns may be a re¯ection of the underlying membrane-insertion event. Current models for membrane protein insertion into the ER generally posit that the transmembrane segments enter the translocon complex one by one, with the ribosome bound on the top of the translocation channel (Liao et al., 1997; Mothes et al., 1997). Within the context of this model, the situation will thus be different during the formation of a helical hairpin with a lumenal loop compared to a helical hairpin with a cytoplasmic loop (Figure 6). In the former case, a hydrophobic stretch presumably goes through a bent, helical hairpin-like state even if it ends up as a single transmembrane segment, whereas in the latter situation it will not bend at all during the insertion process unless turn-promoting residues are present. It thus seems reasonable that helical hairpin formation will require a smaller number of turn-promoting residues when the turn is on the lumenal side. Materials and Methods Enzymes and chemicals Unless otherwise stated, all enzymes were from Promega (Madison, WI, USA). T7 DNA polymerase, BclI, 195 Cytoplasmic Turns in Membrane Proteins cyte lysate were from Promega. Oligonucleotides were from Cybergene (Stockholm, Sweden). DNA manipulations Figure 5. Residues ¯anking the 40 residue long poly (Leu) stretches in the model proteins used previously for measuring lumenal turn propensities and in this work for measuring cytoplasmic turn propensities are not responsible for the observed differences in lumenal versus cytoplasmic turn induction. For the experiments shown, the entire region between a BclI site in lep codon 59 (three codons upstream of the poly(Leu) H2 segment) in the 1P mutant (construct I) used by Monne et al., 1999a and the C terminus was replaced by a region from the 1D, 1G, 1P, 1Q, 1S, and 1W mutants constructed in this work (see Figures 1 and 2) starting 22 residues upstream of the H3 segment (construct II). As shown in the lower panel, all six residues induced much higher degrees of helical hairpin formation when the H3 region was transplanted into the H2 context than seen in Figure 3. Note that helical hairpin formation in H2 results in high levels of non-glycosylated molecules (open bars) whereas in H3 it corresponds to high levels of doubly glycosylated molecules (®lled bars). The arrows in the upper panel indicate the position of the six residues in the two series of constructs. [35S]Met, ribonucleotides, deoxyribonucleotides, dideoxyribonucleotides, and the cap analog m7G(50 )ppp(50 )G were from Amersham-Pharmacia (Uppsala, Sweden). Plasmid pGEM1, transcription buffer and rabbit reticulo- All constructs were made from a recently described Lep construct carried in the pING1 vector containing a lep gene with SpeI and BglII restriction sites introduced in codons 226-227 and 231-232 (SaÈaÈf et al., 1998). Introduction of the H3 region into the periplasmic P2 domain of Lep was performed by replacing the SpeI-BglII fragment by a double-stranded oligonucleotide encoding a stretch of 40 leucine residues ¯anked by SpeI and BamHI (compatible with BglII) restriction sites, yielding an H3 region with the sequence . . . E225-T-S-(L)40-R-S-V233 . . . (numbers refer to the wild-type Lep sequence). The lep gene was further modi®ed by an Asn214 ! Gln mutation (to remove a potential glycosylation site present in the wild-type Lep sequence) and by mutations converting residues 96-98 and 258-260 to, respectively, Asn-Ser-Thr and Asn-Ala-Thr (thus introducing two new potential glycosylation sites ¯anking the H3 segment). The modi®ed lep gene was re-cloned into a pGEM1derived vector for the in vitro expression experiments. During this cloning step, the 50 end of the lep gene was modi®ed such that it acquired a 50 XbaI site and ``Kozak consensus'' sequence (Kozak, 1992), as described (Nilsson & von Heijne, 1993). For all L ! X and LL ! XX mutations (where X is any of the 20 natural amino acids), Leu20 in the poly (Leu) stretch was replaced by X or XX using the QuickChange site-directed mutagenesis kit from Stratagene. The 6 K mutation was made by replacing the Leu19Leu22 stretch by six lysine residues. To study turn formation in the H3 segment (including its ¯anking regions) when it is placed in the H2 position, the entire region between a BclI site in lep codon 59 (three codons upstream of the poly(Leu) H2 segment) in the 1P mutant used in (Monne et al., 1999a) and a SmaI site located downstream of the termination codon in lep was replaced by a region from the 1P mutant constructed in this work starting 22 residues upstream of the H3 segment and ending at the SmaI site at the 30 end of the lep gene. Expression in vitro The constructs in pGEM1 were transcribed by SP6 RNA polymerase for one hour at 37 C. The transcription mixture was as follows: 1-5 mg of DNA template, 5 ml of 10 SP6 H-buffer (400 mM Hepes-KOH (pH 7.4), 60 mM magnesium acetate, 20 mM spermidine-HCl), 5 ml of BSA (1 mg/ml), 5 ml of m7G(50 )ppp(50 )G (10 mM), 5 ml of DTT (50 mM), 5 ml of rNTP mix (10 mM ATP, 10 mM CTP, 10 mM UTP, 5 mM GTP), 18.5 ml of water, 1.5 ml of RNase inhibitor (50 units), 0.5 ml of SP6 RNA polymerase (20 units). Translation was performed in reticulocyte lysate in the presence and absence of dog pancreas microsomes (LiljestroÈm & Garoff, 1991). Translation products were analyzed by SDS-PAGE and gels were quantitated on a Fuji BAS1000 phosphoimager using the MacBAS 2.31 software. Glycosylation ef®ciencies were calculated as the quotient between the intensity of the band corresponding to the relevant glycosylated protein species divided by the summed intensities of all glycosylated and non-glycosylated species. 196 Cytoplasmic Turns in Membrane Proteins Figure 6. Model for helical hairpin formation with a lumenal turn (middle), and a cytoplasmic turn (bottom) during protein insertion into the ER membrane (adapted from Mothes et al., 1997). In the top panel, there is no turn-inducing residue in the poly(Leu) stretch (assumed to be in the H2 position), and only a single transmembrane segment is formed. Formation of a lumenal turn by a turn-inducing residue (*) in the middle of H2 is facilitated by the looped conformation assumed by H2 upon entering the translocation channel (middle panel). Formation of a cytoplasmic turn in H3 is less ef®cient, since H3 does not need to assume a looped conformation during membrane insertion (bottom panel). The opening and closing of the translocation channel during membrane protein integration as envisaged by Liao et al. (1997) is indicated. Acknowledgments This work was supported by grants from the Swedish Cancer Foundation and the Swedish Natural and Technical Sciences Research Councils (to G.vH.). Dog pancreas microsomes were a kind gift from Dr M. Sakaguchi, Fukuoka. References Do, H., Falcone, D., Lin, J., Andrews, D. W. & Johnson, A. E. (1996). The cotranslational integration of membrane proteins into the phospholipid bilayer is a multistep process. Cell, 85, 369-378. Engelman, D. M. & Steitz, T. A. (1981). The spontaneous insertion of proteins into and across membranes: the helical hairpin hypothesis. Cell, 23, 411-422. Hamman, B. D., Hendershot, L. M. & Johnson, A. E. (1998). BiP maintains the permeability barrier of the ER membrane by sealing the lumenal end of the translocon pore before and early in translocation. Cell, 92, 747-758. Johansson, M., Nilsson, I. & von Heijne, G. (1993). Positively charged amino acids placed next to a signal sequence block protein translocation more ef®- ciently in Escherichia coli than in mammalian microsomes. Mol. Gen. Genet. 239, 251-256. Kozak, M. (1992). Regulation of translation in eukaryotic systems. Annu. Rev. Cell Biol. 8, 197-225. Liao, S., Lin, J., Do, H. & Johnson, A. (1997). Both lumenal and cytosolic gating of the aqueous ER translocon pore are regulated from inside the ribosome during membrane protein integration. Cell, 90, 3141. LiljestroÈm, P. & Garoff, H. (1991). Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J. Virol. 65, 147-154. MonneÂ, M., Hermansson, M. & von Heijne, G. (1999a). A turn propensity scale for transmembrane helices. J. Mol. Biol. 288, 141-145. MonneÂ, M., Nilsson, I., Elofsson, A. & von Heijne, G. (1999b). Turns in transmembrane helices: determination of the minimal length of a ``helical hairpin'' and derivation of a ®ne-grained turn propensity scale. J. Mol. Biol. 293, 807-814. Mothes, W., Heinrich, S., Graf, R., Nilsson, I., von Heijne, G., Brunner, J. & Rapoport, T. (1997). Molecular mechanisms of membrane protein integration into the endoplasmic reticulum. Cell, 89, 523-533. Cytoplasmic Turns in Membrane Proteins Nilsson, I. & von Heijne, G. (1993). Determination of the distance between the oligosaccharyltransferase active site and the endoplasmic reticulum membrane. J. Biol. Chem. 268, 5798-5801. Nilsson, I. & von Heijne, G. (1998). Breaking the camel's back: proline-induced turns in a model transmembrane helix. J. Mol. Biol. 284, 1185-1189. SaÈaÈf, A., Wallin, E. & von Heijne, G. (1998). Stop-transfer function of pseudo-random amino acid segments during translocation across prokaryotic and eukaryotic membranes. Eur. J. Biochem. 251, 821-829. 197 von Heijne, G. & Gavel, Y. (1988). Topogenic signals in integral membrane proteins. Eur. J. Biochem. 174, 671-678. Wallin, E. & von Heijne, G. (1998). Genome-wide analysis of integral membrane proteins from eubacterial, archaean, and eukaryotic organisms. Protein Sci. 7, 1029-1038. Wolfe, P. B., Wickner, W. & Goodman, J. M. (1983). Sequence of the leader peptidase gene of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J. Biol. Chem. 258, 12073-12080. Edited by F. Cohen (Received 16 February 2000; received in revised form 24 May 2000; accepted 30 May 2000) III doi:10.1006/jmbi.2001.5108 available online at http://www.idealibrary.com on J. Mol. Biol. (2001) 313, 1171–1179 Formation of Helical Hairpins during Membrane Protein Integration into the Endoplasmic Reticulum Membrane. Role of the N and C-terminal Flanking Regions Marika Hermansson, Magnus Monne and Gunnar von Heijne* Department of Biochemistry and Biophysics, Stockholm University, S-106 91 Stockholm, Sweden The helical hairpin, two closely spaced transmembrane helices separated by a short turn, is a common structural element in integral membrane proteins. Previous studies on the sequence determinants of helical hairpin formation have focussed on the role of polar and charged residues placed centrally in a long stretch of hydrophobic residues, and have yielded a ``propensity scale'' for the relative ef®ciency with which different residues promote the formation of helical hairpins. In this study, we shift our attention to the role of charged residues ¯anking the hydrophobic stretch. Clusters of charged residues are known to hinder membrane translocation, and thus ¯anking charged residues may conceivably force a long hydrophobic segment to form a helical hairpin even if there are no or only weakly turn-promoting residues in the hydrophobic stretch. We indeed ®nd that Lys and, more surprisingly, Asp residues strongly affect helical hairpin formation when placed next to a poly-Leu-based transmembrane segment. We also ®nd that a cluster of four consecutive Lys residues can affect the ef®ciency of helical hairpin formation even when placed 30 residues downstream of the hydrophobic stretch. These observations have interesting implications for the way we picture membrane protein topogenesis within the context of the endoplasmic reticulum (ER) translocon. # 2001 Academic Press *Corresponding author Keywords: membrane protein; protein structure; glycosylation; transmembrane helix; helical hairpin Introduction The helical hairpin, i.e. two closely spaced hydrophobic transmembrane helices separated by a short turn,1 is a common structural element in integral membrane proteins and is thought to serve as an important ``topogenic element'' during membrane protein assembly.2 In previous studies, we have shown that the ef®ciency of formation of helical hairpins during protein insertion into the endoplasmic reticulum (ER) membrane depends both on the overall length of the hydrophobic segment and on the identity of the central, potentially turn-forming residues.3 ± 6 Thus, charged, polar, and ``classical'' helix-breaking residues are good turn-formers and effectively induce helical hairpin formation when placed near the middle of a suf®E-mail address of the corresponding author: [email protected] 0022-2836/01/051171±9 $35.00/0 ciently long hydrophobic stretch, whereas centrally placed apolar residues do not induce helical hairpin formation but rather cause the hydrophobic stretch to insert as a single, long transmembrane helix that spans the membrane only once. While the basic sequence determinants behind the formation of helical hairpins are thus beginning to be mapped out, one aspect that has so far not been addressed is the possible role of the residues that ¯ank the hydrophobic stretch. As charged residues are known to hinder membrane translocation when present in certain sequence contexts, and thus to act as powerful topological determinants in both prokaryotic and eukaryotic membrane proteins,2 we considered the possibility that charged residues in the regions ¯anking a long hydrophobic stretch may speci®cally affect helical hairpin formation. We now report that both Lys and Asp residues strongly affect helical hairpin formation when present close to a poly-Leu hydro# 2001 Academic Press 1172 phobic stretch, the formation of a helical hairpin with a lumenally oriented turn is promoted by C-terminally ¯anking charged residues, whereas the formation of a helical hairpin with a cytoplasmically oriented turn is prevented by such ¯anking residues. We also ®nd that a cluster of four consecutive Lys residues can affect the ef®ciency of helical hairpin formation even when placed 30 residues downstream of the hydrophobic stretch. These ®ndings show that a helical hairpin consists of a long hydrophobic stretch with a central tight turn, and a rather extended region of ¯anking residues must be included in the de®nition of this basic ``folding unit'' in integral membrane proteins. Results Model proteins and topology assay As in our previous studies of helical hairpin formation in transmembrane helices, we have used Helical Hairpins in Membrane Proteins the well-characterized Escherichia coli inner membrane protein leader peptidase (Lep) as a model protein. Lep consists of two transmembrane segments (H1 and H2) connected by a short cytoplasmic loop (P1) and followed by a large C-terminal periplasmic domain (P2). When expressed in vitro in the presence of dog pancreas microsomes, Lep adopts the same membrane topology as in its natural environment in the inner membrane of E. coli,7 i.e. with the N and C termini on the lumenal side,8 Figure 1(a) (left). To study the formation of helical hairpins with the short turn oriented towards the lumenal side of the ER membrane, H2 was substituted by poly-Leu stretches of the general composition ...P58-LI K4L29VL10Q3P-E82... (superscript numbers refer to residues in Lep , subscripts indicate the number of consecutive residues of a given kind), cf. Monne et al.4,5 We have shown that this stretch forms a single transmembrane segment when the model protein is integrated in vitro into dog pancreas Figure 1. Lys and Asp residues at the C-terminal end of H2 promote the formation of helical hairpins with a lumenal turn. (a) The model protein used in this study. The H2 transmembrane segment in Lep (white) was replaced with three different poly-Leu-based segments of the composition K4L29VL10Q3P, K4L21WL7VL10Q3P, and K4L17PL7VL5Q3P, and one to four Lys or Asp residues were introduced at the C-terminal end of the poly-Leu stretch. A glycosylation acceptor site was placed 20 residues downstream of H2 (counting from the ®rst Gln after the hydrophobic stretch). Depending on the lumenal or cytoplasmic localization of the P2 domain, the glycosylation acceptor site will either be modi®ed (®lled Y) or not (open Y). (b) L17PL7VL5-derived constructs with different numbers of C-terminally ¯anking Lys residues were translated in vitro in the absence (ÿ) and presence () of rough microsomes (RM) and analyzed by SDS-PAGE. Black and white dots indicate the glycosylated and non-glycosylated forms of the proteins, respectively. (c) Quanti®cation of the full set of data. The percentage glycosylation was calculated as 100 I/(I Iÿ), where I (Iÿ) is the intensity of the glycosylated (non-glycosylated) band. Helical Hairpins in Membrane Proteins microsomes, but that the introduction of ``turnpromoting'' residues such as Pro, Asn, Arg, or Asp near the middle of the poly-Leu stretch leads to the formation of a helical hairpin.5 As an easily scored marker for the lumenal or cytoplasmic localization of the P2 domain, an N-glycosylation site (Asn-SerThr) was introduced 20 amino acid residues downstream of H2; in constructs where the poly-Leu stretch spans the membrane only once, this site will be glycosylated by the lumenally disposed oligosaccharyl transferase enzyme (Figure 1(a), left), while it will not be modi®ed in constructs where the poly-Leu stretch has been mutated to form a helical hairpin (Figure 1(a), right). We have shown, using alkaline extraction, that poly-Leu-based helical hairpins are properly assembled into the ER membrane, ruling out the possibility that low levels of glycosylation are caused by a failure in membrane integration rather than helical hairpin formation.9 Turn formation can thus be assessed by in vitro transcription/translation of the relevant constructs in the presence of microsomes followed by quanti®cation of the ef®ciency of glycosylation of the engineered N-glycosylation site. A similar strategy was used to study the formation of helical hairpins with the turn oriented towards the cytoplasmic side of the ER membrane (see Figure 4(a)).6 In this case, an extra transmembrane segment (H3) composed of a 40 residue poly-Leu stretch of the composition ...E225-TSL40RSV233... was introduced into the periplasmic P2 domain. This poly-Leu stretch is long enough to form either a single transmembrane segment or a helical hairpin, depending on the identity of residues placed near the middle of the hydrophobic stretch. The topology of the poly-Leu stretch can be determined easily by analyzing the glycosylation status of two strategically placed N-glycosylation acceptor sites, one upstream and one downstream of the hydrophobic stretch. Only the ®rst site will be modi®ed in constructs where the poly-Leu stretch spans the membrane once (Figure 1(a), left), whereas both will be modi®ed when a helical hairpin is formed (Figure 4(a), right). C-terminally flanking Lys and Asp residues promote the formation of helical hairpins with a lumenally oriented turn As a ®rst test of the effect of ¯anking charged residues on helical hairpin formation, we studied three constructs where the H2 transmembrane segment in Lep was replaced by poly-Leu-based segments of the composition K4L29VL10Q3P, K4L21WL7VL10Q3P, and K4L17PL7VL5Q3P. As shown previously,5 the ®rst, uniformly hydrophobic stretch forms a single transmembrane segment, while the two other constructs both have a mixed topology with a helical hairpin formed in approximately 50 % of the molecules. The overall length of the hydrophobic segment (31 residues) in the K4L17PL7VL5Q3P construct is close to the 1173 minimum length required for helical hairpin formation.5 One to four positively charged Lys residues were inserted immediately downstream of the three hydrophobic stretches (replacing one to four residues of the Q3P sequence: KQ2P, K2QP, K3P, K4) and helical hairpin formation was assayed by determining the fraction of glycosylated molecules, Figure 1(b). For all three constructs, the fraction of glycosylated molecules decreases with the number of C-terminally ¯anking Lys residues, Figure 1(c) (left panel). Two extra lysine residues are suf®cient to ensure fully ef®cient helical hairpin formation in the L17PL7VL5 construct, and a signi®cant degree of helical hairpin formation is evident even in the uniformly hydrophobic L29VL10 construct when four lysine residues are added. Since the difference in degree of helical hairpin formation between the two ``hairpin-prone'' and the uniformly hydrophobic segments was maximal for three added lysine residues, we compared helical hairpin formation in the absence and presence of three C-terminally ¯anking Lys residues for an additional series of constructs with the general composition K4L21XL7VL10K3P (where X is any of the 20 natural amino acids). The X residues were chosen on the basis of our earlier results with the K4L21XL7VL10Q3P constructs4,5 to re¯ect the relative ef®ciency of helical hairpin formation induced by hydrophobic and mildly polar amino acids. As seen in Figure 2, the effect of adding three C-terminal lysine residues is minor for the hydrophobic residues, but is quite dramatic for Trp, Ser, Cys, and Thr (compare the white and black bars). As a comparison, Figure 2 also shows previously obtained results5 for the same poly-Leu stretch lacking ¯anking charged residues, but with a pair of residues inserted in the middle of the stretch (gray bars). From this comparison, it is clear that the effect of three ¯anking lysine residues is similar to that seen when a second polar X residue is introduced near the middle of the hydrophobic stretch. We also tested the effects on helical hairpin formation by negatively charged Asp residues. Again, the K4L29VL10Q3P and K4L21WL7VL10Q3P constructs were used, and the Q3P sequence was progressively replaced by one to four aspartic acid residues, Figure 1(c) (right panel). While overall the effects were similar to those seen for Lys (cf. Figure 1(c), left panel), Asp appears to promote helical hairpin formation somewhat more ef®ciently than Lys for the uniformly hydrophobic K4L29VL10Q3P construct, while the differences seen for the K4L21WL7VL10Q3P construct are less conspicuous. Finally, we compared the effects of N and Cterminally ¯anking Asp and Lys residues on helical hairpin formation in a X4L29VL10Z4 construct (X, Z K or D), where four consecutive Lys or Asp residues were inserted in all four combinations immediately upstream and downstream of the hydrophobic stretch, Figure 3. Quanti®cation of the degree of glycosylation of the various con- 1174 Helical Hairpins in Membrane Proteins Figure 2. Helical-hairpin formation in H2-segments of the general composition K4L21XL7VL10K3P (black bars), where X is any of the 20 natural amino acids. Results from our earlier turn propensity study,5 where H2 was substituted by a stretch of the design K4L21XL7VL10Q3P (white bars) or K4L20X2L7VL10Q3P (gray bars) are included for comparison. structs revealed that the identity of the N-terminally ¯anking charged residues made no difference, ruling out an effect dependent on, e.g. charge-pairing between the N and C-terminal ¯anking regions. To make sure that the low levels of glycosylation seen for the these constructs were not caused by a failure of the helical hairpin to insert into the membrane, we expressed a truncated form of the K4L29VL10D4 construct where residues 5-46 (including the H1 transmembrane segment) were deleted, thus leaving the L29VL10 stretch as the only potential membrane-spanning segment. As expected, the molecules were not glycosylated and were associated with the membrane pellet after alkaline extraction of the Figure 3. N-terminally ¯anking residues do not affect helical hairpin formation in H2 (white). Four Lys or four Asp residues were introduced simultaneously at both ends of an H2-stretch of the composition L29VL10, and the degree of glycosylation was determined for each combination of K4 and D4 ¯anking segments. 1175 Helical Hairpins in Membrane Proteins microsomes,10 thus demonstrating ef®cient membrane integration (data not shown). We conclude that the ef®ciency of formation of helical hairpins with a lumenally oriented turn is strongly affected by the presence of C-terminally (but not N-terminally) ¯anking charged residues in our model protein, that a helical hairpin can be induced even in a uniformly hydrophobic segment provided that the number of C-terminally ¯anking charged residues is suf®ciently high, and that the ef®ciency of helical hairpin formation can be increased signi®cantly by ¯anking charged residues in constructs such as K4L17PL7VL5Q3P, where the overall length of the hydrophobic segment is close to the minimum length required for the formation of a helical hairpin. Furthermore, both positively and negatively charged C-terminally ¯anking residues promote helical hairpin formation in this context, i.e. the sign of the charge has little effect. It should be noted, though, that the lumenal or cytoplasmic location of the N-terminal end of the poly-Leu segment may be ®xed by the topology of the N-terminal part of the protein (i.e. the H1-P1 part), and thus that the effects of N-terminally ¯anking residues may well be different in constructs where the poly-Leu stretch is at the N terminus. C-terminally flanking Lys and Asp residues inhibit the formation of helical hairpins with a cytoplasmically oriented turn We have shown that helical hairpins with a cytoplasmically oriented turn can be induced in a long poly-Leu stretch by the introduction of pairs or triplets of polar or charged residues near the middle of the hydrophobic stretch.6 To study the effect of C-terminally ¯anking charged residues in this context, we used a construct with a poly-Leu-based stretch of the composition TSL19P2L19RS placed in the middle of the P2 domain, and the topology of this stretch (a single transmembrane span or a helical hairpin) was determined by assessing the glycosylation status of two N-glycosylation acceptor sites ¯anking the hydrophobic stretch, Figure 4(a). The L19P2L19 segment was chosen because it is poised near the threshold for helical hairpin formation and has an intermediate level of doubly glycosylated molecules (56 %). Again, from one to four Lys or Asp residues were introduced by replacing one to four residues immediately C-terminally to the L19 P2L19 stretch, and the fraction of doubly versus singly glycosylated molecules was determined, Figure 4(b). As seen in Figure 4(c), the degree of helical hairpin formation decreased with increasing numbers of charged residues in this case, and the effect of Lys was somewhat stronger than that of Asp. We conclude that C-terminally ¯anking charged residues inhibit the formation of helical hairpins with a cytoplasmically oriented turn. The effect on helical hairpin formation of C-terminally flanking Lys and Asp residues decreases with the distance from the hydrophobic stretch Finally, we studied the effect of increasing the separation between the charged residues (four Lys or four Asp) and the hydrophobic stretch for both ``lumenal turn'' (K4L29VL10Q3P construct) and ``cytoplasmic turn'' (TSL19P2L19RS construct) helical hairpins. As shown in Figure 5, the K4 segment promoted the formation of a helical hairpin with a lumenally oriented turn even when placed 32 residues downstream of the hydrophobic stretch (left panel), while the effect of a D4 segment decreased much more rapidly with the separation distance. Similarly, the K4 segment inhibited the formation of a helical hairpin with a cytoplasmically oriented turn up to larger separation distances than did the D4 segment, although the effect drops off more rapidly in this case (right panel). Thus, the ef®ciency of helical hairpin formation can depend on sequence determinants located quite far away from the hydrophobic transmembrane segment. Discussion The helical hairpin appears to be a basic folding unit in multi-spanning integral membrane proteins, and is thus central to our understanding of membrane protein topology. Based on analyses of model membrane proteins speci®cally designed to facilitate the study of helical hairpin formation during membrane protein insertion into the ER membrane, we previously quanti®ed the relative propensities of the 20 natural amino acids to induce the formation of helical hairpins when placed in the middle of a long poly-Leu stretch, both for helical hairpins with lumenally and cytoplasmically oriented turns.4 ± 6,9 We have also determined the minimal length of the hydrophobic segment required for ef®cient helical hairpin formation.5 Membrane protein topology is, however, not determined by only the hydrophobic transmembrane helices themselves. Charged residues in short loops connecting the hydrophobic segments are potent topogenic determinants; the best-known effect in this regard is codi®ed in the so-called positive inside rule, which states that positively charged loops tend to remain on the cytoplasmic side of the membrane.11,12 Based on this premise, we hypothesized that the tendency for helical hairpin formation in a long hydrophobic stretch would depend on the number of charged residues in the immediate ¯anking regions, in addition to the characteristics of the hydrophobic stretch itself. In this study, we have thus asked whether helical hairpin formation can be in¯uenced by ¯anking positively and negatively charged residues (Lys and Asp). We ®nd that N-terminally ¯anking residues have no effect on helical hairpin formation in our model protein (possibly because the lumenal 1176 Helical Hairpins in Membrane Proteins Figure 4. Lys and Asp residues at the C-terminal end of H2 inhibit the formation of helical hairpins with a cytoplasmic turn. (a) A 40 residue poly-Leu stretch (H3, white) was inserted into the middle of the P2 domain, and the glycosylation status of two Asn-X-Thr glycosylation acceptor sites ¯anking H3 was determined after in vitro translation in the presence of dog pancreas microsomes (glycosylated and non-glycosylated glycosylation acceptor sites are indicated, respectively, by ®lled Y and open Y). Constructs where H3 forms a single transmembrane segment (left) will be glycosylated on only one site, whereas those where H3 forms a helical hairpin (right) will be glycosylated on both sites. (b) TSL19P2L19RS-derived constructs with different numbers of C-terminally ¯anking Asp residues were translated in vitro in the absence (ÿ) and presence () of rough microsomes (RM) and analyzed by SDS-PAGE. Molecules with zero, one, and two modi®ed glycosylation sites are indicated by one white dot, one black dot, and two black dots, respectively. (c) Quanti®cation of the full set of data for C-terminally ¯anking Lys (white dots) and Asp (black dots) residues. The percentage of doubly glycosylated molecules was calculated as the quotient between the intensity of the doubly glycosylated band divided by the summed intensities of the doubly and singly glycosylated bands. or cytoplasmic location of the N terminus of the particular hydrophobic stretch studied here is determined by the topology of the preceding part of the model protein), whereas charged residues of either sign placed C-terminally to the hydrophobic stretch generally promote the cytoplasmic location of the C-terminal end of the hydrophobic stretch. Thus, charged C-terminally ¯anking residues promote helical hairpin formation in constructs where the turn is oriented towards the lumenal side of the ER membrane, and weaken the tendency to form helical hairpins in constructs where the turn is oriented towards the cytoplasmic side. Interestingly, ¯anking positively and negatively charged residues have similar effects on helical hairpin formation when located close to the hydrophobic stretch (Figures 1 and 4), suggesting that they somehow ``sense'' the sidedness of the ER membrane during the co-translational membrane assembly of the model protein. Given the high hydrophobicity of the poly-Leu stretch used in these studies, it is possible that the hydrophobic stretch partitions into the lipid environment almost immediately upon entering the protein-conducting translocon channel in the ER membrane,13,14 thus facilitating interactions between the charged ¯anking residues and the lipid head groups. The surprising observation that an effect on the formation of a helical hairpin with a lumenal turn of a stretch of four Lys residues can be detected even at quite large separation distances (30 residues; Figure 5) between the hydrophobic stretch and the charge cluster further shows that a considerable length of nascent chain can impact the process of hairpin for- 1177 Helical Hairpins in Membrane Proteins Figure 5. The effect on helical hairpin formation of C-terminally ¯anking Lys and Asp residues decreases with the distance from the hydrophobic stretch. Four Lys (white dots) or Asp (black dots) residues were placed at different distances C-terminally to the H2 L29VL10 stretch (left panel) or the H3 L19 P2L19 stretch (right panel). The percentage singly (left panel) or doubly (right panel) glycosylated molecules were calculated as for Figures 1 and 4, respectively. mation. This is somewhat reminiscent of the socalled charge block on protein translocation observed when a string of positively charged residues is placed downstream of a signal-anchor sequence15,16 and indicates that a fairly sizeable portion of the nascent chain (the 40 residue hydrophobic stretch and at least 30 additional residues) may be present within the translocon when the helical hairpin forms. An alternative, though perhaps less likely, possibility is that a block of charged residues located in the ribosomal tunnel can trigger a change in the ribosome-translocon interaction that facilitates the helical hairpin formation, cf. Liao et al.17 The stronger effect on helical hairpin formation seen for positively charged Lys residues compared to the negatively charged Asp residues, both for helical hairpins with a cytoplasmic turn (Figure 4), and for helical hairpins with a lumenal turn at large separation distances (Figure 5), is in accord with the positive inside rule. From this point of view, it is surprising that the effects on helical hairpin formation exerted by negatively charged residues located close to the hydrophobic segment are very similar or even stronger than those seen for positively charged residues (Figure 1). In previous studies, negatively charged residues have been found to have only weak effects on the translocation of N-terminal tails across the membrane,18,19 and on the translocation of C-terminal domains when placed downstream of a hydrophobic signalanchor sequence.20 We have no good explanation for their much stronger effects on helical hairpin formation. In summary, our results show that ¯anking residues need to be included in the de®nition of the helical hairpin when viewed as a fundamental folding unit in integral membrane proteins. As a ®rst approximation, membrane proteins may thus be pictured as being composed of single transmembrane helices (and their immediate ¯anking segments) spaced far apart in the sequence and of helical hairpins (again including their immediate ¯anking segments). This view rests on the assumption that the topology of a (long or short) hydro- phobic segment can be in¯uenced directly only by the part of the nascent chain that is present within the ribosome-translocon channel at the time when the segment integrates into the lipid bilayer, and thus that topology is, to an important extent, determined locally. The topological determinants for single transmembrane helices are quite well understood,21 and our recent work on helical hairpin formation has now shed some light on this second kind of topogenic element. Finally, we note that the observations reported here may be relevant for improving current methods of membrane protein topology prediction,22 since these methods do not distinguish between single transmembrane helices and helical hairpins. Materials and Methods Enzymes and chemicals Unless stated otherwise, all enzymes were from Promega (Madison, WI, USA). Phage T7 DNA polymerase, BclI, [35S]Met, ribonucleotides, deoxyribonucleotides, dideoxyribonucleotides, and the cap analog m7G(50 )ppp(50 )G were from Amersham-Pharmacia (Uppsala, Sweden). Plasmid pGEM1, transcription buffer and rabbit reticulocyte lysate were from Promega. Oligonucleotides were from Cybergene (Stockholm, Sweden). DNA manipulations For cloning into and expression from the pGEM1 plasmid, the 50 end of the lep gene was modi®ed, ®rst, by the introduction of an XbaI site and, second, by changing the context 50 to the initiator ATG codon to a ``Kozak consensus'' sequence.23 Thus, the 50 region of the gene was modi®ed to: ...ATAACCCTCTAGAGCCACCATGGCGAAT... XbaI site and initiator codon underlined. Replacement of the H2 region in Lep was performed as described,9 i.e. by ®rst introducing BclI and NdeI restriction sites in codons 59 and 80 ¯anking the H2 region and then replacing the BclI-NdeI fragment by the appropriate double-stranded oligonucleotides. Sitespeci®c mutagenesis used to add BclI and NdeI restriction sites at the 30 and 50 ends of H2 in Lep and to introduce an Asn±Ser-Thr acceptor site for N-linked glycosylation was performed according to the method of 1178 Kunkel.24,25 The glycosylation acceptor site was designed as described,26 i.e. by replacing three codons positioned 20 codons downstream of H2 with codons for the acceptor tripeptide Asn-Ser-Thr. In all constructs, the naturally occurring glycosylation site at Asn214 in Lep was removed by an Asn214 ! Gln mutation. Residues 59-81 in H2 were replaced by a poly-Leu sequence of the design LIK4L29VL10Q3P (subscripts indicate the number of consecutive residues) for the 40 residue poly-Leu construct. In the 31 residue poly-Leu constructs, four leucine residues were deleted from the N terminus and ®ve leucine residues from the C terminus of the 40 residue poly-Leu stretch by PCR mutagenesis. Introduction of the H3 region into the periplasmic P2 domain of Lep was performed as described,6 i.e. by using a pING1 vector containing a lep gene with SpeI and BglII restriction sites introduced into codons 226-227 and 231-232, and then replacing the SpeI-BglII fragment by a double-stranded oligonucleotide encoding the sequence ...E225-TSL19P2L19RS-V233... (superscript numbers refer to the wild-type Lep sequence) ¯anked by SpeI and BamHI (compatible with BglII) restriction sites. The lep gene was further modi®ed by an Asn214 ! Gln mutation (to remove a potential glycosylation site present in the wild-type Lep sequence) and by mutations converting residues 96-98 and 258-260 to, respectively, Asn-Ser-Thr and Asn-Ala-Thr (thus introducing two new potential glycosylation sites ¯anking the H3 segment). The modi®ed lep gene was re-cloned into the pGEM1derived vector described above for the in vitro expression experiments. The QuickChange site-directed mutagenesis kit (Stratagene) was used for the introduction of ¯anking charged Lys and Asp residues and all L ! X substitutions. Asp residues ¯anking the H2 poly-Leu stretches on the N-terminal side were introduced by replacing the K4 segment by a D4 segment; Lys and Asp residues ¯anking the H2 poly-Leu stretches on the C-terminal side were introduced by replacing the Q3P segment (KQ2P, K2QP, K3P, K4, and the same for D); and Lys and Asp residues ¯anking the H3 poly-Leu stretch on the Cterminal side were introduced by replacing one to four residues immediately following the poly-Leu stretch. All mutants were con®rmed by DNA sequencing using T7 DNA polymerase. Expression in vitro The constructs in pGEM1 were transcribed by SP6 RNA polymerase for one hour at 37 C. The transcription mixture was as follows: 1-5 mg of DNA template, 5 ml of 10 SP6 H-buffer (400 mM Hepes-KOH (pH 7.4), 60 mM magnesium acetate, 20 mM spermidine-HCl), 5 ml of 1 mg/ml BSA, 5 ml of 10 mM m7G(50 )ppp(50 )G, 5 ml of 50 mM DTT, 5 ml of rNTP mix (10 mM ATP, 10 mM CTP, 10 mM UTP, 5 mM GTP), 18.5 ml of water, 1.5 ml of 33 units/ml RNase inhibitor, 0.5 ml of 40 units/ ml SP6 RNA polymerase. Translation of 1 ml of mRNA was performed as described27 at 30 C for one hour in 9 ml of nucelase-treated reticulocyte lysate, 1 ml of 40 units/ml RNase inhibitor, 1 ml of 15 mCi/ml [35S]Met, 1 ml of amino acids mix (1 mM of each amino acid except Met), 1 ml of mRNA, and 1 ml of 2 units/ml dog pancreas microsomes (one unit is de®ned as the amount of microsomes required for 50 % translocation of in vitro synthesized preprolactin). Translation products were analyzed by SDS-PAGE and gels were quanti®ed on a Fuji FLA-3000 phosphoimager using the Fuji Image Helical Hairpins in Membrane Proteins Reader 8.1j software. The glycosylation ef®ciency of a given mutant was calculated as the quotient between the intensity of the glycosylated band divided by the summed intensities of the glycosylated and non-glycosylated bands for the H2-based constructs, and as the quotient between the intensity of the doubly glycosylated band divided by the summed intensities of the doubly and singly glycosylated bands for the H3-based constructs. Acknowledgments This work was supported by grants from the Swedish Cancer Foundation and the Swedish Research Council to G.v.H. Dog pancreas microsomes were a kind gift from Dr M. Sakaguchi, Fukuoka. References 1. Engelman, D. M. & Steitz, T. A. (1981). The spontaneous insertion of proteins into and across membranes: the helical hairpin hypothesis. Cell, 23, 411-422. 2. von Heijne, G. (2000). Recent advances in the understanding of membrane protein assembly and structure. Quart. Rev. Biophys. 32, 285-307. 3. Nilsson, I. & von Heijne, G. (1998). Breaking the camel's back: proline-induced turns in a model transmembrane helix. J. Mol. Biol. 284, 1185-1189. 4. MonneÂ, M., Hermansson, M. & von Heijne, G. (1999). A turn propensity scale for transmembrane helices. J. Mol. Biol. 288, 141-145. 5. MonneÂ, M., Nilsson, I., Elofsson, A. & von Heijne, G. (1999). Turns in transmembrane helices: determination of the minimal length of a `'helical hairpin'' and derivation of a ®ne-grained turn propensity scale. J. Mol. Biol. 293, 807-814. 6. SaÈaÈf, A., Hermansson, M. & von Heijne, G. (2000). Formation of cytoplasmic turns between two closely spaced transmembrane helices during membrane protein integration into the ER membrane. J. Mol. Biol. 301, 191-197. 7. Wolfe, P. B., Wickner, W. & Goodman, J. M. (1983). Sequence of the leader peptidase gene of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J. Biol. Chem. 258, 12073-12080. 8. Nilsson, I. & von Heijne, G. (1993). Determination of the distance between the oligosaccharyltransferase active site and the endoplasmic reticulum membrane. J. Biol. Chem. 268, 5798-5801. 9. Nilsson, I., SaÈaÈf, A., Whitley, P., Gafvelin, G., Waller, C. & von Heijne, G. (1998). Proline-induced disruption of a transmembrane a-helix in its natural environment. J. Mol. Biol. 284, 1165-1175. 10. Fujiki, Y., Hubbard, A. L., Fowler, S. & Lazarow, P. B. (1982). Isolation of intracellular membranes by means of sodium carbonate treatment. J. Cell. Biol. 93, 97-102. 11. von Heijne, G. & Gavel, Y. (1988). Topogenic signals in integral membrane proteins. Eur. J. Biochem. 174, 671-678. 12. von Heijne, G. (1986). The distribution of positively charged residues in bacterial inner membrane proteins correlates with the trans-membrane topology. EMBO J. 5, 3021-3027. 1179 Helical Hairpins in Membrane Proteins 13. Heinrich, S., Mothes, W., Brunner, J. & Rapoport, T. (2000). The Sec61p complex mediates the integration of a membrane protein by allowing lipid partitioning of the transmembrane domain. Cell, 102, 233244. 14. Mothes, W., Heinrich, S., Graf, R., Nilsson, I., von Heijne, G., Brunner, J. & Rapoport, T. (1997). Molecular mechanisms of membrane protein integration into the endoplasmic reticulum. Cell, 89, 523-533. 15. Andersson, H. & von Heijne, G. (1991). A 30-residue-long ``Export Initiation Domain'' adjacent to the signal sequence is critical for protein translocation across the inner membrane of Escherichia coli. Proc. Natl Acad. Sci. USA, 88, 9751-9754. 16. Johansson, M., Nilsson, I. & von Heijne, G. (1993). Positively charged amino acids placed next to a signal sequence block protein translocation more ef®ciently in Escherichia coli than in mammalian microsomes. Mol. Gen. Genet. 239, 251-256. 17. Liao, S., Lin, J., Do, H. & Johnson, A. (1997). Both lumenal and cytosolic gating of the aqueous ER translocon pore are regulated from inside the ribosome during membrane protein integration. Cell, 90, 31-41. 18. Kiefer, D., Hu, X. T., Dalbey, R. & Kuhn, A. (1997). Negatively charged amino acid residues play an active role in orienting the Sec-independent Pf3 coat protein in the Escherichia coli inner membrane. EMBO J. 16, 2197-2204. 19. Nilsson, I. M. & von Heijne, G. (1990). Fine-tuning the topology of a polytopic membrane protein. role 20. 21. 22. 23. 24. 25. 26. 27. of positively and negatively charged residues. Cell, 62, 1135-1141. Laws, J. K. & Dalbey, R. E. (1989). Positive charges in the cytoplasmic domain of Escherichia coli leader peptidase prevent an apolar domain from functioning as a signal. EMBO J. 8, 2095-99. Goder, V. & Spiess, M. (2001). Topogenesis of membrane proteins: determinants and dynamics. FEBS Letters, 504, 87-93. MoÈller, S., Croning, M. & Apweiler, R. (2001). Evaluations of methods for the predictive evaluation of membrane spanning regions. Bioinformatics, 17, 646653. Kozak, M. (1992). Regulation of translation in eukaryotic systems. Annu. Rev. Cell Biol. 8, 197-225. Kunkel, T. A. (1987). Rapid and ef®cient site-speci®c mutagenesis without phenotypic selection. Methods Enzymol. 154, 367-382. Geisselsoder, J., Witney, F. & Yuckenberg, P. (1987). Ef®cient site-directed in vitro mutagenesis. BioTechniques, 5, 786-791. Nilsson, I., Whitley, P. & von Heijne, G. (1994). The C-terminal ends of internal signal and signal-anchor sequences are positioned differently in the ER translocase. J. Cell Biol. 126, 1127-1132. LiljestroÈm, P. & Garoff, H. (1991). Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J. Virol. 65, 147-154. Edited by F. Cohen (Received 4 June 2001; received in revised form 24 September 2001; accepted 24 September 2001) IV doi:10.1016/j.jmb.2003.10.019 J. Mol. Biol. (2003) 334, 803–809 Inter-helical Hydrogen Bond Formation During Membrane Protein Integration into the ER Membrane Marika Hermansson and Gunnar von Heijne* Department of Biochemistry and Biophysics, Arrhenius Laboratory, Stockholm University, SE-106 91 Stockholm, Sweden Recent work has shown that efficient di- or trimerization of hydrophobic transmembrane helices in detergent micelles or lipid bilayers can be driven by inter-helix hydrogen bonding involving polar residues such as Asn or Asp. Using in vitro translation in the presence of rough microsomes of a model integral membrane protein, we now show that the formation of so-called helical hairpins, two tightly spaced transmembrane helices connected by a short loop, can likewise be promoted by the introduction of Asn-Asn or Asp-Asp pairs in a long transmembrane hydrophobic segment. These observations suggest that inter-helix hydrogen bonds can form within the context of the Sec61 translocon in the endoplasmic reticulum, implying that hydrophobic segments in a nascent polypeptide chain in transit through the Sec61 channel have immediate access to a non-aqueous subcompartment within the translocon. q 2003 Elsevier Ltd. All rights reserved. *Corresponding author Keywords: membrane protein; protein structure; hydrogen bond; transmembrane helix; helical hairpin Introduction In eukaryotic cells, most integral membrane proteins insert into and fold in the membrane of the endoplasmic reticulum (ER) from which they may then be transported further along the secretory pathway or to the nuclear envelope. The initial recognition of the hydrophobic segments that form the transmembrane a-helices takes place during polypeptide transfer through the Sec61 translocon complex in the ER membrane.1,2 It appears that the transmembrane helices can either be recognized and transferred laterally into the surrounding lipid bilayer one by one, or in pairs,3 – 6 or even as higher-order multimers,7 though these early steps on the folding pathway are only poorly understood at present. Recent biochemical studies of model peptides in detergent micelles or liposomes have shed some light on the basic physical chemistry of helix – helix interactions in a non-polar environment. Tight close-packing between helices mediated by the so-called GxxxG-motif, or inter-helical hydrogen bond formation between pairs of polar residues such as Asn and Asp, or more complex Abbreviations used: ER, endoplasmic reticulum; Lep, leader peptidase; RM, rough microsomes. E-mail address of the corresponding author: [email protected] motifs involving Thr or Ser8 – 16 have been shown to drive efficient dimer formation in these systems. These interactions have been confirmed by twohybrid assays in Escherichia coli where the formation of a dimeric protein complex in the inner membrane ultimately results in the activation of a reporter gene.17,18 It is thus quite clear that hydrogen bonding can drive helix – helix interactions in biological membranes; what is not clear, however, is if inter-helix hydrogen bonds can form already within the context of the translocon. Inspired by previous observations that suggest that transmembrane helices may encounter a bilayer-like environment almost immediately upon entering the translocon channel,2,19 we have carried out a detailed analysis of the possible influence of Asn-Asn and Asp-Asp pairs on the formation of so-called helical hairpins (i.e. two closely spaced transmembrane helices connected by a short loop) during co-translational membrane protein insertion into the ER. We have already reported that helical hairpin formation is facilitated by the presence of charged, polar, and weakly hydrophobic residues in the short connecting loop,20 – 22 and by charged residues placed immediately downstream of a long, uniformly hydrophobic segment.23 Here, we show that Asn-Asn or Asp-Asp pairs can promote the formation of helical hairpins in a strictly position-specific manner. The observed pattern of hairpin-promoting pairs 0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. 804 is consistent with the formation of inter-helical hydrogen bonds. If this interpretation is correct, it provides yet another argument for the contention that transmembrane helices are formed within a non-aqueous subcompartment in the Sec61 translocon. Results Model protein and topology assay As in our previous studies of helical hairpin formation in membrane proteins, we have used the well-characterized E. coli inner membrane protein leader peptidase (Lep) as a model protein. Lep consists of two transmembrane segments (H1 and H2) connected by a short cytoplasmic loop (P1) and a large C-terminal periplasmic domain (P2). When expressed in vitro in the presence of dog pancreas rough microsomes (RM), Lep adopts the same membrane topology as in its natural environment in the inner membrane of E. coli,24 i.e. with the N terminus and the large C-terminal P2 domain on the lumenal side,25 Figure 1(a) (left). For the studies reported here, H2 was substituted by a poly-Leu stretch of the general composition …P58-LIK4L29VL10Q3P-E82… (superscript numerals refer to residues in Lep, subscript numerals indicate the number of consecutive residues of a given kind).20,22 The segment has four Nterminal lysine residues intended to anchor this end to the cytoplasmic side of the membrane, a 40 residue stretch composed of 39 leucine and one valine (the latter resulting from an engineered Hydrogen Bonds in Membrane Proteins restriction site) that is long enough to span the membrane either as a single or as two closely spaced transmembrane helices, and four uncharged polar residues (Q3P) demarcating the C-terminal end of the hydrophobic stretch. We have shown previously that this stretch forms a single transmembrane segment when the model protein is integrated in vitro into RMs,26 but that the introduction of “turn-promoting” residues such as Pro, Asn, Arg, or Asp near the middle of the poly-Leu stretch leads to the formation of a helical hairpin.20 As an easily scored marker for the lumenal or cytoplasmic localization of the P2 domain, an Nglycosylation site (Asn-Ser-Thr) was introduced 20 amino acid residues downstream of the poly-Leu stretch. In constructs where the poly-Leu stretch spans the membrane only once, this site will be glycosylated by the lumenally disposed oligosaccharyl transferase enzyme (Figure 1(a), left), while it will not be modified in constructs where the poly-Leu stretch has been mutated to form a helical hairpin (Figure 1(b), right).27 Turn formation can thus be assayed by in vitro transcription/translation of the relevant constructs in the presence of RM followed by quantification of the efficiency of glycosylation of the engineered N-glycosylation site. Typical gels for three constructs described below (D8, D33, and D8-D33) are shown in Figure 1(b); in these constructs, the poly-Leu segment in D8-D33 inserts almost exclusively as a helical hairpin (very little glycosylation), while essentially no helical hairpin formation is seen for D8 (almost complete glycosylation). The D33 construct has a Figure 1. (a) The Lep model protein. Wild-type Lep has two transmembrane segments (H1, H2), and inserts into dog pancreas rough microsomes in vitro with the Nlum –Clum orientation shown on the left. In the studies reported here, the H2 segment has been replaced by a poly-Leu-based sequence (LIK4L29VL10Q3P) into which one or two Asn or Asp residues have been inserted. Constructs where H2 spans the membrane once are glycosylated by the lumenally disposed oligosaccharyl transferase enzyme on a unique Asn-Ser-Thr glycosylation acceptor site (Y) in the P2 domain (left); those where the H2 segment forms a helical hairpin are not (right). In this study, one or a pair of Asn or Asp residues has been inserted in the areas in the H2 segment encircled on the right-hand model. (b) The D8, D33, and D8-D33 constructs were translated in vitro in the absence (2 ) and in the presence (þ) of dog pancreas rough microsomes (RM). Unglycosylated and glycosylated molecules are indicated by the white dot and the black dot, respectively. 805 Hydrogen Bonds in Membrane Proteins mixed topology with about half the molecules being glycosylated. Effects on helical hairpin formation of single Asn and Asp residues in a (39L 1 V) transmembrane segment To test the effect on helical hairpin formation when an Asn or Asp residue is placed in different positions in a poly-Leu stretch, we introduced Asn or Asp in positions 8, 10, 12 and 31 – 39 in the (39L þ V) segment, and helical hairpin formation was assayed by determining the fraction of nonglycosylated molecules ( f ng) for each construct. None of the single Asn-mutations induce efficient formation of a helical hairpin (Figure 2(a)). We always observe a background level of f ng , 0.2, most likely a result of somewhat inefficient targeting and glycosylation in the in vitro system. The only positions in which an Asn leads to a slight increase in f ng are 12 and 37. We have already shown that an Asn-residue placed in position 22 near the middle of the (39L þ V) segment induces almost complete formation of a helical hairpin ( f ng ¼ 0.85),22 and it is therefore not very surprising that we see a slight effect with the N12 mutation. It is less obvious why the N37 mutation should have a comparable effect (but see Discussion). For constructs with a single Asp in the (39L þ V) stretch, the picture is the same as for Asn for positions 8, 10, and 12 (Figure 2(b)). The Asp mutations near the C-terminal end of the segment have significantly higher f ng values than the corresponding Asn mutations, with the D37 mutation having the strongest effect. It thus appears that an Asp residue near the C-terminal end of the (39L þ V) stretch induces a significant level of helical hairpin formation, in agreement with earlier studies where Asp residues immediately flanking the C-terminal end of the (39L þ V) stretch were shown to induce a helical hairpin conformation.23 We have shown that a single Pro residue placed 10– 12 residues from either end of the (39L þ V) stretch induces a significant amount of the helical hairpin conformation,20 suggesting that the higher f ng values seen for the more centrally placed D12, D32, D33, and (possibly) D34 mutations might be caused by the strong hairpin-inducing effect demonstrated earlier for a centrally placed Asp residue.20 Position-specific effects on helical hairpin formation by Asn-Asn and Asp-Asp pairs Figure 2. Degree of helical hairpin formation measured as the fraction of non-glycosylated molecules ( f ng) observed after in vitro translation in the presence of RM for single Asn (a) and single Asp (b) mutations in the (39L þ V) H2 segment. The position of the Asn or Asp residue is counted from the N-terminal end of the (39L þ V) segment. All measurements are mean values of two or three independent experiments; in general, repeat f ng-value measurements differ by no more than ^0.05 from the mean value. We next examined the tendencies of Asn-Asn and Asp-Asp pairs to mediate helical hairpin formation by introducing such pairs in the (39L þ V) stretch using the single-residue constructs reported above for comparison. After an initial screening of a small number of Asn-Asn pairs, we focused mainly on the N8 position, and made a complete set of pairs with the second Asn in positions 32 – 39. For the N10 and N12 pairs, we made constructs with the second Asn in positions 33, 35, 37, 38 (for N12), and 39. Strikingly, there is an almost twofold increase in f ng for the N8 –N33, N8 –N37, and N12 –37 constructs over the corresponding single-Asn constructs, while all the other Asn-Asn constructs have f ng values similar to those of the corresponding single-Asn constructs. The same set of constructs but with Asp instead of Asn were tested (Figure 3(b)). The D8-D33 and D8-D37 constructs have , 1.5-fold higher f ng values than seen for the corresponding individual mutations. In addition, the D10 – D35, D10 – D37, and D10 –D39 constructs, and all constructs that include D12, have higher f ng values than those of the corresponding single-Asp constructs. As a control to ensure that the poly-Leu segments in even the poorly glycosylated constructs were integrated properly into the microsomal membrane, we made a final series of constructs where the H1 transmembrane segment and a part 806 Figure 3. Degree of helical hairpin formation measured as the fraction of non-glycosylated molecules ( f ng) observed after in vitro translation in the presence of RM for Asn-Asn (a) and Asp-Asp (b) pairs in the (39L þ V) H2 segment. The positions of the Asn or Asp residues are counted from the N-terminal end of the (39L þ V) segment. of P1 (residues 5 – 46) were deleted, leaving the poly-Leu stretch as the only hydrophobic segment in the protein. These constructs were expressed in the presence of RM, and their membrane integration was tested by alkaline extraction of the membranes.28,29 Two typical examples are shown in Figure 4. The D33(DH1) construct is partially glycosylated when expressed in the presence of RM, and both the glycosylated and the non-glycosylated molecules remain in the membrane pellet (lanes 3 and 4). Likewise, even though the D8D33(DH1) construct is glycosylated only weakly, the bulk of the molecules remain in the membrane pellet when the protein is expressed in the presence of RM. Similar results were obtained for the other DH1 constructs tested (D33(DH1), D37(DH1), D8-D35(DH1), and D8-D37(DH1); data not shown). Discussion In earlier work on helical hairpin formation, we have shown that single polar or charged residues Hydrogen Bonds in Membrane Proteins introduced in the middle of a sufficiently long poly-Leu transmembrane segment can induce an almost complete conformational transition from a long single-spanning transmembrane segment to a helical hairpin.20 – 22,26,30 We have further shown that charged residues placed immediately downstream of the poly-Leu transmembrane segment can induce the formation of helical hairpins even when no polar or charged residue is present within the hydrophobic stretch itself.23 Inspired by the observation that hydrogen bonding between Asn or Asp residues in transmembrane helices in detergent micelles, model membranes, or the E. coli inner membrane can drive oligomerization,12,13,31 we have now tested whether pairs of Asn or Asp residues, i.e. residues shown to mediate strong dimerization in these studies, can promote the formation of intramolecular helical hairpins during membrane protein assembly into the ER membrane, i.e. within the context of the Sec61 translocon. As shown in Figures 2 and 3, Asn-Asn and AspAsp pairs placed in a model (39L þ V) transmembrane segment can increase the efficiency of helical hairpin formation by 1.5– 2-fold over that seen for the corresponding single Asn or Asp residues. Even more strikingly, the effect is highly positionspecific. For both Asn and Asp, pairs in positions 8 –33 and 8 –37 in the 40 residue long hydrophobic segment induce the formation of a helical hairpin, as does the N12 – N37 pair and all pairs with Asp in positions 10 or 12 combined with a second Asp in positions 33, 35, 37 or 39. The Asn and Asp 8 – 33 and 8 –37 pairs are especially interesting, as they are clearly positionspecific (a shift of either of the two residues by one or two positions abolishes the effect) and further involve residues that are both one to two turns from the ends of the (39L þ V) stretch. Moreover, if one assumes a helical hairpin conformation for the (39L þ V) stretch, residues 33 and 37 will be located one turn apart on the same face of the Cterminal helix of the hairpin. These pairs of residues would thus be in a perfect position to stabilize the helical hairpin conformation by interresidue hydrogen bonding. For Asn, the only other pair that has an increased level of helical hairpin formation compared to the two corresponding single-Asn constructs is N12 – N37. Presumably, N12 is near the middle of the first helix in the hairpin, while N37 is close to the C-terminal end of the second helix, making a direct interaction a less likely explanation in this case. Since both N12 and N37 have a slightly higher level of helical hairpin conformation than the other single-Asn constructs, there may be an additive effect of the two Asn residues that reflects, e.g. interactions with the translocon rather than inter-residue hydrogen bonding. Further studies will be necessary to resolve this point. The Asp-Asp pairs are less informative, since the single-Asp mutations, in general, have a stronger effect than the single-Asn mutations (Figure 2(b)). 807 Hydrogen Bonds in Membrane Proteins Figure 4. Alkaline extraction of the D33(DH1) and D8-D33(DH1) constructs expressed in vitro in the absence (2) and presence (þ ) of RM. Unglycosylated and glycosylated molecules are indicated by the white dot and the black dot, respectively. S, supernatant; P, membrane pellet. It is thus difficult to discern an obvious pattern in the Asp-Asp pair constructs (Figure 3(b)). Nevertheless, it is interesting to note that helical hairpin formation is clearly increased in the 8– 33 and 8 – 37 pairs also for Asp. If our interpretation that the 8 –33 and 8 –37 Asn and Asp pairs indeed stabilize the helical hairpin conformation by hydrogen bonding is correct, it implies, first, that the formation of the helical hairpin must take place in a non-aqueous environment within the translocon, since stable inter-residue hydrogen bonding would not be expected in an aqueous environment; previous results from our laboratory also point in this direction.21 Second, it further implies that the two halves of the helical hairpin formed in the (39L þ V) segment cannot rotate freely relative to one another, as only positions on one face of the two helices (8 – 33 and 8 – 37 but not 8– 35 or 10 –33) are implicated. It is not immediately obvious why these faces are special; perhaps the polar residues flanking the hydrophobic stretch help orient the helices relative to the translocon. In any case, our results suggest that very specific helix –helix interactions can be formed within the context of the ER translocon and that such interactions can have a dramatic effect on membrane protein topology. Materials and Methods Enzymes and chemicals Unless stated otherwise, all enzymes were from Promega (Madison, WI, USA). [35S]Met, ribonucleotides, deoxyribonucleotides, dideoxyribonucleotides, and the cap analog m7G(50 )ppp(50 )G were from Amersham-Pharmacia (Uppsala, Sweden). Plasmid pGEM1, DTT, transcription buffer and rabbit reticulocyte lysate were from Promega. Spermidine was from Sigma. Oligonucleotides were from Cybergene (Stockholm, Sweden). DNA manipulations For cloning into and expression from the pGEM1 plasmid, the 50 end of the lep gene was modified, first, by the introduction of an Xba I site and, second, by changing the context 50 to the initiator ATG codon to a “Kozak consensus” sequence.32 Thus, the 50 region of the gene was modified to: …ATAACCCTCTAGAGCCACCATGGCGAAT… (Xba I site and initiator codon underlined). Replacement of the H2 region in Lep was performed as described,27 i.e. by first introducing Bcl I and Nde I restriction sites in codons 59 and 80 flanking the H2 region and then replacing the Bcl I-Nde I fragment by the appropriate double-stranded oligonucleotides. Sitespecific mutagenesis used to add Bcl I and Nde I restriction sites at the 30 and 50 ends of H2 in Lep and to introduce an Asn-Ser-Thr acceptor site for N-linked glycosylation was performed according to the method of Kunkel.33,34 The glycosylation acceptor site was designed as described,35 i.e. by replacing three codons positioned 20 codons downstream of H2 with codons for the acceptor tripeptide Asn-Ser-Thr. In all constructs, the naturally occurring glycosylation site at Asn214 in Lep was removed by an Asn214 ! Gln mutation. Residues 59 – 81 in H2 were replaced by a (39L þ V) sequence of the design LIK4L29VL10Q3P (subscripts indicate the number of consecutive residues). The QuickChange site-directed mutagenesis kit (Stratagene) was used for the introduction of Asn or Asp residues in position 8, 10, and 12 and in positions 33 – 39 of the (39L þ V) stretch. All mutants were confirmed by DNA sequencing of plasmids using ABI PRISMw BigDyee Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems). Expression in vitro The constructs in pGEM1 were transcribed by SP6 RNA polymerase for one hour at 37 8C. The transcription mixture was as follows: 1– 5 mg of DNA template, 5 ml of 10 £ SP6 H-buffer (400 mM Hepes– KOH (pH 7.4), 60 mM magnesium acetate, 20 mM spermidine– HCl), 5 ml of BSA (1 mg/ml), 5 ml of 10 mM m7G(50 )ppp(50 )G, 808 5 ml of 50 mM DTT, 5 ml of rNTP mix (10 mM ATP, 10 mM CTP, 10 mM UTP, 5 mM GTP), 18.5 ml of water, 1.5 ml of RNase inhibitor (33 units/ml), 0.5 ml of SP6 RNA polymerase (40 units/ml). Translation of 1 ml of mRNA was performed as described36 at 30 8C for one hour in 9 ml of nuclease-treated reticulocyte lysate, 1 ml of RNase inhibitor (40 units/ml), 1 ml of [35S]Met (15 mCi/ml), 1 ml of amino acids mix (1 mM each amino acid except Met), 1 ml of mRNA, and 1 ml of dog pancreas microsomes (2 units/ml; one unit is defined as the amount of microsomes required for 50% translocation of in vitro synthesized preprolactin). Translation products were analyzed by SDS-PAGE and gel bands were quantified on a Fuji FLA-3000 phosphoimager using the Fuji Image Reader 8.1j software. The glycosylation efficiency of a given mutant was calculated as the intensity of the glycosylated band divided by the summed intensities of the glycosylated and non-glycosylated bands. Sodium carbonate extraction of microsomes was carried out as described.29 Hydrogen Bonds in Membrane Proteins 10. 11. 12. 13. 14. Acknowledgements This work was supported by grants from the Swedish Cancer Foundation, the Swedish Research Council, and the Margareta and Marcus Wallenberg Foundation to G.v.H. Dog pancreas microsomes were a kind gift from Dr M. Sakaguchi, Fukuoka. References 1. Goder, V. & Spiess, M. (2001). Topogenesis of membrane proteins: determinants and dynamics. FEBS Letters, 504, 87 – 93. 2. von Heijne, G. (2003). Membrane protein assembly in vivo. Advan. Protein Chem. 63, 1 – 18. 3. Oliver, J., Jungnickel, B., Görlich, D., Rapoport, T. & High, S. (1995). The Sec61 complex is essential for the insertion of proteins into the membrane of the endoplasmic reticulum. FEBS Letters, 362, 126– 130. 4. Do, H., Falcone, D., Lin, J., Andrews, D. W. & Johnson, A. E. (1996). The cotransational integration of membrane proteins into the phospholipid bilayer is a multistep process. Cell, 85, 369– 378. 5. Heinrich, S., Mothes, W., Brunner, J. & Rapoport, T. (2000). The Sec61p complex mediates the integration of a membrane protein by allowing lipid partitioning of the transmembrane domain. Cell, 102, 233– 244. 6. Kanki, T., Young, M. T., Sakaguchi, M., Hamasaki, N. & Tanner, M. J. A. (2003). The N-terminal region of the transmembrane domain of human erythrocyte band 3—residues critical for membrane insertion and transport activity. J. Biol. Chem. 278, 5564– 5573. 7. Borel, A. C. & Simon, S. M. (1996). Biogenesis of polytopic membrane proteins: membrane segments assemble with translocation channels prior to membrane integration. Cell, 85, 379– 389. 8. Russ, W. P. & Engelman, D. M. (2000). The GxxxG motif: a framework for transmembrane helix– helix association. J. Mol. Biol. 296, 911 – 919. 9. Zhou, F. X., Cocco, M. J., Russ, W. P., Brunger, A. T. & Engelman, D. M. (2000). Interhelical hydrogen bond- 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. ing drives strong interactions in membrane proteins. Nature Struct. Biol. 7, 154– 160. Senes, A., Gerstein, M. & Engelman, D. M. (2000). Statistical analysis of amino acid patterns in transmembrane helices: the GxxxG motif occurs frequently and in association with beta-branched residues at neighboring positions. J. Mol. Biol. 296, 921– 936. Zhou, F. X., Merianos, H. J., Brunger, A. T. & Engelman, D. M. (2001). Polar residues drive association of polyleucine transmembrane helices. Proc. Natl Acad. Sci. USA, 98, 2250– 2255. Choma, C., Gratkowski, H., Lear, J. D. & DeGrado, W. F. (2000). Asparagine-mediated self-association of a model transmembrane helix. Nature Struct. Biol. 7, 161– 166. Gratkowski, H., Lear, J. D. & DeGrado, W. F. (2001). Polar side chains drive the association of model transmembrane peptides. Proc. Natl Acad. Sci. USA, 98, 880– 885. Li, R., Mitra, N., Gratkowski, H., Vilaire, G., Litvinov, R., Nagasami, C. et al. (2003). Activation of integrin aIIbb3 by modulation of transmembrane helix associations. Science, 300, 795– 798. Dawson, J. P., Melnyk, R. A., Deber, C. M. & Engelman, D. M. (2003). Sequence context strongly modulates association of polar residues in transmembrane helices. J. Mol. Biol. 331, 255– 262. Dawson, J., Weinger, J. & Engelman, D. (2002). Motifs of serine and threonine can drive association of transmembrane helices. J. Mol. Biol. 316, 799– 805. Langosch, D., Brosig, B., Kolmar, H. & Fritz, H. J. (1996). Dimerisation of the glycophorin A transmembrane segment in membranes probed with the ToxR transcription activator. J. Mol. Biol. 263, 525–530. Russ, W. P. & Engelman, D. M. (1999). TOXCAT: a measure of transmembrane helix association in a biological membrane. Proc. Natl Acad. Sci. USA, 96, 863– 868. Mothes, W., Heinrich, S., Graf, R., Nilsson, I., von Heijne, G., Brunner, J. & Rapoport, T. (1997). Molecular mechanisms of membrane protein integration into the endoplasmic reticulum. Cell, 89, 523–533. Monné, M., Nilsson, I., Elofsson, A. & von Heijne, G. (1999). Turns in transmembrane helices: determination of the minimal length of a “helical hairpin” and derivation of a fine-grained turn propensity scale. J. Mol. Biol. 293, 807– 814. Nilsson, I., Johnson, A. E. & von Heijne, G. (2003). How hydrophobic is alanine? J. Biol. Chem. 278, 29389– 29393. Monné, M., Hermansson, M. & von Heijne, G. (1999). A turn propensity scale for transmembrane helices. J. Mol. Biol. 288, 141– 145. Hermansson, M., Monné, M. & von Heijne, G. (2001). Formation of helical hairpins during membrane protein integration into the endoplasmic reticulum membrane. Role of the N and C-terminal flanking regions. J. Mol. Biol. 313, 1171– 1179. Wolfe, P. B., Wickner, W. & Goodman, J. M. (1983). Sequence of the leader peptidase gene of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J. Biol. Chem. 258, 12073– 12080. Nilsson, I. & von Heijne, G. (1993). Determination of the distance between the oligosaccharyltransferase active site and the endoplasmic reticulum membrane. J. Biol. Chem. 268, 5798– 5801. Nilsson, I. & von Heijne, G. (1998). Breaking the 809 Hydrogen Bonds in Membrane Proteins 27. 28. 29. 30. 31. camel’s back: proline-induced turns in a model transmembrane helix. J. Mol. Biol. 284, 1185–1189. Nilsson, I., Sääf, A., Whitley, P., Gafvelin, G., Waller, C. & von Heijne, G. (1998). Proline-induced disruption of a transmembrane a-helix in its natural environment. J. Mol. Biol. 284, 1165– 1175. Fujiki, Y., Hubbard, A. L., Fowler, S. & Lazarow, P. B. (1982). Isolation of intracellular membranes by means of sodium carbonate treatment. J. Cell Biol. 93, 97 –102. Sakaguchi, M., Mihara, K. & Sato, R. (1987). A short amino-terminal segment of microsomal cytochrome P-450 functions both as an insertion signal and as a stop-transfer sequence. EMBO J. 6, 2425– 2431. Sääf, A., Hermansson, M. & von Heijne, G. (2000). Formation of cytoplasmic turns between two closely spaced transmembrane helices during membrane protein integration into the ER membrane. J. Mol. Biol. 301, 191– 197. Howard, K. P., Lear, J. D. & DeGrado, W. F. (2002). 32. 33. 34. 35. 36. Sequence determinants of the energetics of folding of a transmembrane four-helix-bundle protein. Proc. Natl Acad. Sci. USA, 99, 8568– 8572. Kozak, M. (1992). Regulation of translation in eukaryotic systems. Annu. Rev. Cell Biol. 8, 197– 225. Kunkel, T. A. (1987). Rapid and efficient site-specific mutagenesis without phenotypic selection. Methods Enzymol. 154, 367– 382. Geisselsoder, J., Witney, F. & Yuckenberg, P. (1987). Efficient site-directed in vitro mutagenesis. BioTechniques, 5, 786– 791. Nilsson, I., Whitley, P. & von Heijne, G. (1994). The C-terminal ends of internal signal and signal-anchor sequences are positioned differently in the ER translocase. J. Cell Biol. 126, 1127– 1132. Liljeström, P. & Garoff, H. (1991). Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J. Virol. 65, 147– 154. Edited by F. E. Cohen (Received 18 June 2003; accepted 8 October 2003) V Confronting fusion protein-based membrane protein topology mapping with reality: the Escherichia coli ClC chloride channel Marika Cassel and Gunnar von Heijne Department of Biochemistry and Biophysics Stockholm University SE-106 91 Stockholm, Sweden Corresponding author: GvH Phone: Int+46-8-16 25 90 Fax: Int+46-8-15 36 79 E-mail: [email protected] Keywords: membrane protein, ion channel, YadQ, topology, protein structure Abstract The topology of bacterial inner membrane proteins is commonly determined using topology reporters such as PhoA and GFP fused to a series of C-terminally truncated versions of the protein in question. Here, we report a detailed topology mapping of the Escherichia coli inner membrane chloride channel YadQ. Since the 3D structure of YadQ is known, our results provide a critical test of the reporter fusion approach and offer new insights into the YadQ folding pathway. -2- Introduction With the number of high-resolution membrane protein structures still hovering around 100 and increasing only slowly, topology modeling will remain an important aspect of membrane protein biochemistry and bioinformatics for years to come. Theoretical topology prediction methods are being continually improved but are still far from perfect (Chen et al. 2002). On the experimental side, reporter fusion techniques are by far the most widely used for determining the topology of bacterial inner membrane proteins (Manoil 1991), and can also be used in yeast (Kim et al. 2003). Obviously, once a high-resolution structure becomes available reporter fusions would seem to be of little value, and we do not know of a single case where an experimental topology analysis has been done post facto. Here, we report the results of a detailed, reporter fusion-based topology mapping of the E. coli ClC chloride channel, a protein for which an X-ray structure was reported in 2002 (Dutzler et al. 2002). Using the structure as a guide, we have made a set of optimally placed reporter fusions, giving us the possibility to confront a best-case reporter fusion scenario with the hard facts of a high-resolution structure. We find that the reporter fusion data accurately reflects the topology of 10 of the 14 transmembrane helices, but misses two short helical hairpins on the periphery of the protein. Given the rather weak hydrophobicity of these hairpins and their peripheral locations, we suggest that the reporter fusion results reflect not only the final topology of the protein but also pin-point parts of the molecule that may insert into the membrane at a late stage in the folding process. -3- Results The ClC chloride channel The amino acid sequence, topology, and 3D structure of the E. coli ClC chloride channel YadQ are shown in Fig. 1. The protein is a homodimer, with an ion pore in each of the monomers. There are 14 transmembrane helices in the monomer, some of which are very long, some very short, and some highly inclined relative to the membrane normal. Given this rather complicated structure of the transmembrane helix bundle, we reasoned that YadQ would be particularly challenging for topology mapping techniques and that the results of a topology analysis may help to elucidate the propensity of the different transmembrane helices to insert into the membrane. Topology reporter analysis The most widely used technique for topology mapping of bacterial inner membrane proteins is based on reporter protein fusions to C-terminally truncated versions of the inner membrane protein in question. We have used two well-established topology reporter proteins, green fluorescent protein (GFP) and alkaline phosphatase (PhoA), to map the topology of YadQ. GFP folds and becomes fluorescent only when localized in the cytoplasm, not when located in the periplasm (Feilmeier et al. 2000; Drew et al. 2002). PhoA behaves in the opposite way, and is enzymatically active when located in the periplasm but not when in the cytoplasm (Manoil and Beckwith 1986). A useful rule of thumb when making topology reporter fusions is to place the fusion joints near the C-terminal ends of the loops between the transmembrane helices (Boyd et al. 1993). With the YadC structure in hand, we could choose optimal fusion joint as shown in Fig. 1A. Every loop in the structure was targeted; the final list of fusions that were constructed is given in Table 1. GFP and PhoA activities were measured as described previously (Rapp et al. 2004), and are also given in Table 1. In all -4- cases, there is a good correlation between the activities of the two reporters: fusions with a high GFP activity invariably have a low PhoA activity, and vice versa. Reporter fusions vs. 3D structure The reporter fusion results are mapped onto the 3D structure in Fig. 2A and on the corresponding topology map in Fig. 1B. In general, the activities of the GFP and PhoA fusions correspond well with the 3D structure, but there are two obvious exceptions where the fusion data is not in accord with the 3D structure: the E/F and P/Q loops (the E-F and P-Q helical hairpins are highlighted in the Figure). Although these loops are oriented towards the periplasm in the 3D structure, the GFP and PhoA fusions consistently report cytoplasmic localizations. From a topology mapping point of view, the best current approach is to constrain theoretical prediction methods such as TMHMM or HMMTOP by experimental data of the kind generated here (Melén et al. 2003). Using the full set of reporter fusion results as constraints for TMHMM, we obtain the topology model shown in Fig. 2B (top). Compared to the unconstrained TMHMM prediction, Fig. 2B (middle), the ends of the transmembrane helices are much better defined in the constrained model. The unconstrained model misses helix F, fuses helices G-H as well as O-P into single helices, and incorrectly predicts the C-terminus to be periplasmic. The fully constrained prediction, on the other hand, misses the E-F and P-Q hairpins. In a more realistic setting where the 3D structure of the target protein is not known, one generally resorts to hydrophobicity plots to guide the selection of reporter fusion sites. Such a plot is shown in Fig. 2C, with the reporter fusions analyzed above indicated by filled and unfilled circles. A conservative guess as to which fusions sites would have been chosen based only on the hydrophobicity plot is shown by the arrows. Using only the latter as constraints for TMHMM results in the topology model shown in -5- Fig. 2B (bottom). In this model, both the E-F and G-H hairpins are missed, the O and P helices are merged, and the Q helix is missed. Discussion Among the known helix bundle membrane protein 3D structures, the chloride channel YadQ stands out in terms of length variation among the transmembrane helices and modes of helix-helix packing interactions (Dutzler et al. 2002). It would thus appear to be a particularly challenging protein for topology mapping by reporter fusions, and might pin-point particular kinds of shortcomings of this widely used approach. Seen from a different perspective, the behavior of topology reporters fused to internal loops in the protein might also reveal differences in membrane insertion ability between different transmembrane helices and thereby help illuminate steps in the membrane insertion pathway. For these reasons, we have undertaken a detailed topology mapping of YadQ, using complementary GFP and PhoA reporter fusions. The fusions sites were chosen based on the 3D structure: all loops between helices (both transmembrane helices and surface helices) were targeted, and the fusion sites were placed at the C-terminal end of each loop, Table 1. The results of the reporter fusion analysis are mapped onto the 3D structure in Fig. 2A, and are also summarized in the topology model presented in Fig. 1B. Despite the unruly 3D structure, all reporter fusions but two report the correct location of the corresponding loop, attesting to the general reliability of the reporter fusion technique. The fusions even identify one a loop between helices G and H that is not possible to detect from the hydrophobicity plot shown in Fig. 2C. The two loops that are not identified by the reporter fusions are located between the E-F and P-Q helices. They are both oriented towards the periplasm in the 3D -6- structure, but the reporter fusions made immediately downstream of the E and P helices are cytoplasmic. In contrast to all the other Nin-Cout transmembrane helices in YadQ, the E and P helices are thus unable to promote translocation of the reporters to the periplasm. Interestingly, the E helix (RWWRVLPVKFFGGLGTLGG) has a rather low average hydrophobicity, and the short P helix (APLTGIILVLEMT) is not even discernible on the hydrophobicity plot, Fig. 2C, suggesting the possibility that the E-F and P-Q segments insert into the membrane not as individual helices but as pre-formed “helical hairpins” (Engelman and Steitz 1981). The E-F and P-Q hairpins are highlighted in Fig. 2A: they are both located on the perimeter of the YadQ monomer, and it does not seem unreasonable to assume that they can fold into pre-formed pockets in the structure only after most or all of the other helices have assembled. This would be analogous to the suggested late insertion of the so-called pore helices in certain potassium channels (Umigai et al. 2003), except that the E-F and P-Q hairpins are lipid-exposed to a much greater degree than the potassium channel pore helices. Since the P-Q hairpin is at the monomer-monomer interface, dimer formation would only be possible after the proper assembly of this hairpin in the monomer. Finally, we note that when TMHMM is constrained with a more realistic set of reporter fusions – one that it could conceivably have been constructed without knowledge of the 3D structure – only 4 of the 7 helical hairpins are correctly predicted, Fig. 2B (bottom). Although the hydrophobicity plot (Fig. 2C) might have led one to suspect additional transmembrane helices in the region corresponding to helices E-G and perhaps P-Q, it would have been very difficult to obtain strong evidence for their existence using only reporter fusions. Possibly, complementing techniques such as Cyslabelling (van Geest and Lolkema 2000) might have helped, but it is nevertheless clear that it is going to be very difficult to avoid artifactual results in cases where a transmembrane helix can only insert across the membrane as part of a helical hairpin. -7- Given that YadQ appears to be a rather extreme case compared to most other known membrane protein structures, this may not be too much of a problem, however. -8- Materials and methods DNA techniques. The cloning of the E. coli yadQ into phoA and gfp fusion vectors was done as described (Rapp et al. 2004; Daley et al. 2005). Primers were obtained from Cybergene (Stockholm) or MWG (Ebersberg, FRG). All constructs were checked by plasmid sequencing (BM labbet, Furulund, Sweden). Experimental determination of C-terminal location. Constructs encoding PhoA fusions were transformed into E. coli CC118 (Lee and Manoil 1994) and those encoding GFP fusions were transformed into E. coli BL21(DE3)pLysS. The PhoA and GFP assays were performed as described (Rapp et al. 2004; Daley et al. 2005). The assays were repeated 3 times for each construct and the results were averaged. Acknowledgements The work was supported by grants from the Swedish Research Council, the Marianne and Marcus Wallenberg Foundation, the Swedish Foundation for Strategic Research, and the Swedish Cancer Foundation to GvH. We thank Karin Melén for help with the TMHMM predictions. -9- References Boyd, D., Traxler, B., and Beckwith, J. 1993. Analysis of the topology of a membrane protein by using a minimum number of alkaline phosphatase fusions. J Bacteriol 175: 553-556. Chen, C.P., Kernytsky, A., and Rost, B. 2002. Transmembrane helix predictions revisited. Protein Sci 11: 2774-2791. Claros, M.G., and von Heijne, G. 1994. TopPred II: An improved software for membrane protein structure prediction. CABIOS 10: 685-686. Daley, D.O., Rapp, M., Granseth, E., Melén, K., Drew, D., and von Heijne, G. 2005. Global topology analysis of the Escherichia coli inner membrane proteome. Science 308: 1321-1323. Drew, D., Sjöstrand, D., Nilsson, J., Urbig, T., Chin, C.N., de Gier, J.W., and von Heijne, G. 2002. Rapid topology mapping of Escherichia coli inner-membrane proteins by prediction and PhoA/GFP fusion analysis. Proc Natl Acad Sci USA 99: 2690-2695. Dutzler, R., Campbell, E.B., Cadene, M., Chait, B.T., and MacKinnon, R. 2002. X-ray structure of a ClC chloride channel at 3.0 Å reveals the molecular basis of anion selectivity. Nature 415: 287-294. Engelman, D.M., and Steitz, T.A. 1981. The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell 23: 411-422. Engelman, D.M., Steitz, T.A., and Goldman, A. 1986. Identifying nonpolar transbilayer helices in amino acid sequences of membrane proteins. Annu Rev Biophys Biophys Chem 15: 321-353. Feilmeier, B.J., Iseminger, G., Schroeder, D., Webber, H., and Phillips, G.J. 2000. Green fluorescent protein functions as a reporter for protein localization in Escherichia coli. J Bacteriol 182: 4068-4076. Kim, H., Melén, K., and von Heijne, G. 2003. Topology models for 37 Saccharomyces cerevisiae membrane proteins based on C-terminal reporter fusions and prediction. J Biol Chem 278: 10208-10213. Lee, E., and Manoil, C. 1994. Mutations eliminating the protein export function of a membrane-spanning sequence. J Biol Chem 269: 28822-28828. - 10 - Manoil, C. 1991. Analysis of membrane protein topology using alkaline phosphatase and ß-galactosidase gene fusions. Methods in Cell Biol. 34: 61-75. Manoil, C., and Beckwith, J. 1986. A genetic approach to analyzing membrane protein topology. Science 233: 1403-1408. Melén, K., Krogh, A., and von Heijne, G. 2003. Reliability measures for membrane protein topology prediction algorithms. J Mol Biol 327: 735-744. Rapp, M., Drew, D.E., Daley, D.O., Nilsson, J., Carvalho, T., Melén, K., de Gier, J.W., and von Heijne, G. 2004. Experimentally based topology models for E. coli inner membrane proteins. Prot Sci 13: 937-945. Umigai, N., Sato, Y., Mizutani, A., Utsumi, T., Sakaguchi, M., and Uozumi, N. 2003. Topogenesis of two transmembrane type K+ channels, Kir 2.1 and KcsA. J Biol Chem 278: 40373-40384. van Geest, M., and Lolkema, J.S. 2000. Membrane topology and insertion of membrane proteins: Search for topogenic signals. Microbiol. Mol. Biol. Rev. 64: 13-33. von Heijne, G. 1992. Membrane protein structure prediction - Hydrophobicity analysis and the positive-inside rule. J Mol Biol 225: 487-494. - 11 - Table 1 Reporter fusion sites and activities. PhoA and GFP activities (mean of 3 experiments) were normalized to the most active reporter fusion. Reporter fusions were classified as - (little or no activity), + (medium activity), and ++ (high activity). Fusion site <PhoA> <GFP> PhoA class GFP class 73 1.00 0.01 ++ - 107 0.00 0.78 - ++ 122 0.00 1.00 - ++ 146 0.01 0.44 - ++ 170 0.00 0.21 - ++ 192 0.50 0.01 ++ - 213 0.00 0.07 - + 250 0.15 0.04 + - 288 0.00 0.00 - - 308 0.36 0.01 ++ - 318 0.45 0.01 ++ - 328 0.60 0.00 ++ - 357 0.01 0.17 - + 384 0.20 0.03 ++ - 403 0.00 0.15 - + 420 0.00 0.05 - + 442 0.00 0.15 - + 458 0.00 0.22 - ++ 473 0.00 0.22 - ++ - 12 - Figure legends Figure 1 The ClC chloride channel protein YadQ. (A) Amino acid sequence. The 14 transmembrane helices are underlined. Boxed residues show the position of the last residue in the YadQ part of each fusion. (B) Topology derived from the 3D structure. Transmembrane helices are indicated by upper case letters, other helices by lower case letters. Qualitative reporter fusion activities are shown as (-, +, ++; upper sign = PhoA, lower sign = GFP). (C) 3D structure of the YadQ homodimer (PDB-code 1KPK). The two subunits are shaded in gray and white. The periplasmic side is up. Figure 2 Topology mapping results. (A) Reporter fusion activities mapped onto the 3D structure of the YadQ homodimer (left subunit). White CPK residues indicate fusion sites with high PhoA and low GFP activity; black CPK residues indicate sites with low PhoA and high GFP activity. Arrows point to the reporter fusion sites in the E/F and P/Q loops. The E-F and P-Q helical hairpins are shown as cartoons (right subunit). (B) TMHMM predictions. The unconstrained TMHMM prediction is shown on top, the fully constrained prediction in the middle, and the model obtained when using the fusions indicated by arrows in panel C as constraints is shown on the bottom. The transmembrane helices identified in the 3D structure are indicated by thin black lines. (C) Hydrophobicity plot produced by TopPred (von Heijne 1992; Claros and von Heijne 1994) using the GES hydrophobicity scale (Engelman et al. 1986) and default parameters. The helices identified in the 3D structure are indicated (transmembrane helices in upper case). The reporter fusion sites are indicated by white (high PhoA, low GFP activity) or black (low PhoA, high GFP activity) circles. Arrows point to reporter - 13 - fusion sites that might have been used had the 3D structure not been known (see text). The dashed lines indicate the default cutoff hydrophobicity used by TopPred for predicting “putative” and “certain” transmembrane helices. - 14 - E F H I K N O Q Figure 1A ALLAASI R APLTGIILVLEMTDNW Q LILPMIITGLGATLLAQFTGG K PLWSAILARTLAKQE A EQLARSKAASASEN T P F N LIPIATAGN F SMGMLVFIFVARVITTLLCFSSGAPGGI F APMLALGTVLGTAFGMVAVELFPQWH L EAGTFAIAGMG M LIDVGKLSDAPL N TLWLWLILGIIFGIFGPIFNKWVLGMQDLLHRVHGGN I TKWVLMGGAIGGLCGLLGF V APATSGGG J GRMVLDIFRLK G DEARHTLLATGAAAGLAAAFN A PLAGILFIIEEMRPQFRWTL I SIKAVFIGVIMSTIMWRIFNHEVA G TVAFLCSAVLAMFGWFLVRKWAPEAGG S GIPEIEGALEDQRP V RWWRVLPVKFFGGLGTLGGGMVL G REGPTVQIGGNI C MKTDTPSLETPQAARLRRRQLIRQLLERDKTPLAILFMAAVVGTLVGLAAVAFDKGVAWLQNQRMGALVHTA D NWPLLL B a B ++ - d ++ Figure 1B E ++ C ++ F ++ G ++ - + I J cytoplasm H + - periplasm - K M + ++ ++ - l N ++ - + O P + ++ + r Q Figure 1C Figure 2A P-Q E-F Figure 2B Figure 2C a B C d E F G H I J K l M N O P Q r