* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download ref. #27 of the TIBS article
Genetic code wikipedia , lookup
Membrane potential wikipedia , lookup
Lipid bilayer wikipedia , lookup
Bottromycin wikipedia , lookup
Magnesium transporter wikipedia , lookup
Protein adsorption wikipedia , lookup
Ribosomally synthesized and post-translationally modified peptides wikipedia , lookup
Theories of general anaesthetic action wikipedia , lookup
Model lipid bilayer wikipedia , lookup
List of types of proteins wikipedia , lookup
G protein–coupled receptor wikipedia , lookup
NADH:ubiquinone oxidoreductase (H+-translocating) wikipedia , lookup
Protein–protein interaction wikipedia , lookup
Mechanosensitive channels wikipedia , lookup
Proteolysis wikipedia , lookup
Two-hybrid screening wikipedia , lookup
Protein domain wikipedia , lookup
Endomembrane system wikipedia , lookup
Cell membrane wikipedia , lookup
SNARE (protein) wikipedia , lookup
Western blot wikipedia , lookup
Protein structure prediction wikipedia , lookup
Cell-penetrating peptide wikipedia , lookup
Article No. mb982217 J. Mol. Biol. (1998) 284, 1165±1175 Proline-induced Disruption of a Transmembrane a -Helix in its Natural Environment IngMarie Nilsson1,2, Annika SaÈaÈf1, Paul Whitley3, Guro Gafvelin3 Cecilia Waller1 and Gunnar von Heijne1* 1 Department of Biochemistry Stockholm University, S-106 91 Stockholm, Sweden 2 Department of Biosciences Karolinska Institute, NOVUM S-141 57 Huddinge, Sweden 3 Deptartment of Laboratory Medicine, Divsion of Clinical Immunology, Karolinska Hospital, S-171 76 Stockholm Sweden a-Helix formation in globular proteins has been studied both theoretically and experimentally for decades, while a lack of both high-resolution structures and suitable experimental techniques has hampered the study of helices in membrane proteins. We have developed a new experimental approach, glycosylation mapping, where the active site of the lumenally exposed endoplasmic reticulum enzyme oligosaccharyl transferase is used as a point of reference against which the position of a transmembrane segment in the membrane can be measured. Here, we report an initial analysis of the helix-breaking properties of proline residues inserted in a transmembrane helix. We ®nd that proline residues can break a transmembrane helix, but only when inserted near the end, and only when the helix is suf®ciently long. The glycosylation mapping technique may be generally useful for determining the position of transmembrane helices in the membrane. # 1998 Academic Press *Corresponding author Keywords: membrane protein; protein structure; glycosylation; transmembrane helix; proline Introduction Integral membrane proteins are thought to come in two basic varieties, built, respectively, on a helix bundle and a b-barrel architecture (von Heijne, 1996). The helix bundle design with its long, hydrophobic transmembrane a-helices (TMHs) is particularly simple, yet detailed studies of the conformational properties of individual amino acid residues in such transmembrane segments and of the effects of different residues on the segments position in the lipid bilayer are dif®cult, since neither X-ray crystallography nor NMR are easily applied to membrane proteins. Previous studies of helix formation in a non-aqueous environment have mainly been carried out with puri®ed peptides dissolved in detergents or incorporated into liposomes (Deber & Goto, 1996; Li & Deber, 1994; Li et al., 1996; Liu et al., 1996; Papavoine et al., 1997; Shon et al., 1991; Stopar et al., 1996). We have developed an alternative approach Abbreviations used: TMHs, transmembrane a-helices; ER, endoplasmic reticulum; OST, oligosaccharyl transferase; MGD, minimum glycosylation distance; Lep, leader peptidase. E-mail address of the corresponding author: [email protected] 0022±2836/98/491165±11 $30.00/0 where the active site of the lumenally exposed endoplasmic reticulum (ER) enzyme oligosaccharyl transferase (OST) is used as a point of reference against which positional alterations in a TMH embedded in the ER membrane can be measured with high precision. As a ®rst application of this technique, we have investigated the conformational effects of proline residues placed in different positions in a poly-Leu TMH. A calibration of the method using the TMH from the H-subunit of the Rhadobacter sphaeroides photosynthetic reaction center makes it possible to deduce the position of other TMHs relative to the lipid bilayer from glycosylation mapping data. Results Glycosylation mapping The basic idea behind the glycosylation mapping technique is depicted in Figure 1(a). As shown by (Nilsson & von Heijne (1993), the OST can transfer a glycosyl moiety to an acceptor Asn residue in a nascent membrane protein only when the Asn-X(Thr/Ser) acceptor site is placed a minimum number of residues away from either end of a TMH, the ``minimum glycosylation distance'' (MGD). N and C-terminal MGD values (MGDN, MGDC) can # 1998 Academic Press 1166 Glycosylation Mapping of Transmembrane Helices Figure 1. Glycosylation mapping of constructs derived from leader peptidase (Lep) (Wolfe et al., 1983). (a) Potential Asn-Ser-Thr acceptor sites for N-linked glycosylation (Y glycosylated acceptor site, = non-glycosylated; acceptor site) are introduced in different positions relative to the H1 and H2 transmembrane segments, as well as to a segment (H3) inserted into the P2 domain by site-directed mutagenesis. Proteins are expressed in vitro in the presence of dog pancreas microsomes, and the degree of glycosylation is determined as a function of d, the number of residues between a chosen reference residue at the end of the hydrophobic segment and the acceptor Asn. The minimal glycosylation distance (MGD) is the number of residues required for half-maximal glycosylation. (b) The gels show the results obtained from in vitro translation in the absence (ÿ) and presence () of rough microsomes (RM) of constructs with the n 23 poly-Leu transmembrane segment replacing the H2 segment in Lep and with the acceptor-site Asn located at d 9, 10, 11, and 12 residues (counting from the ®rst Gln after the hydrophobic stretch). Filled and empty circles indicate the non-glycosylated and glycosylated forms of the protein, respectively. The MGD value (de®ned as the value of d for which the glycosylation ef®ciency is 40%, i.e. one-half of the maximal glycosylation ef®ciency which is 80%) is determined from the graph by interpolation and is shown above the sequence of the transmembrane segment. be measured with high precision using in vitro translation in the presence of dog pancreas microsomes of mutant proteins with differently placed glycosylation acceptor sites (Figure 1(b)). Changes in the position of a TMH relative to the membrane (or, more accurately, relative to the active site of the membrane-bound OST enzyme) caused by mutations in the hydrophobic stretch should be re¯ected in proportional changes in the MGDN and MGDC values, thus providing a sensitive assay for studying the conformational role of different amino acids in TMHs. In the studies reported here, we have replaced the ®rst (H1) and second (H2) transmembrane domains in the well-characterized Escherichia coli inner membrane protein leader peptidase (Lep; Figure 1(a) left-hand panel), with either poly-Leu based model transmembrane segments or with the N-terminal TMH from the R. sphaeroides photosyn- thetic reaction center. We have also inserted polyLeu sequences in the middle of the lumenal P2 domain (right-hand panel). Although of bacterial origin, Lep integrates ef®ciently into dog pancreas microsomes with the same topology as in E. coli (Johansson et al., 1993; Nilsson & von Heijne, 1993). To facilitate the analysis, a set of Lep-encoding plasmids differing only in the position of the glycosylation acceptor site Asn-Ser-Thr upstream of H1 and H3 or downstream of H2 was constructed, and MGD values were determined for each transmembrane segment by cloning it into the appropriate plasmids and determining the number of residues between the N or C-terminal end of the hydrophobic stretch and the acceptor site required for half-maximal glycosylation. Since proline residues are known to reduce the ef®ciency of glycosylation when present either immediately upstream, Glycosylation Mapping of Transmembrane Helices 1167 in the middle, or immediately downstream of an Asn-X-Thr/Ser acceptor site (Mellquist et al., 1998; Shakineshleman et al., 1996; our unpublished observations), all such Pro residues in Lep were changed to Gln in the engineered acceptor sites. MGD is a function of the length of the transmembrane segment We have shown that MGDC values vary with the length of the hydrophobic transmembrane segment (Nilsson et al., 1994). To quantify this effect, H2 was replaced by a stretch of n leucine residues and one valine side ¯anked by four N-terminal lysine residues and a C-terminal Gln-Gln-Gln-Pro segment, and MGDC (counting from the ®rst ¯anking Gln residue) was determined for different n-values. As seen in Figure 2(a), there is a roughly linear correlation between n and MGDC for 12 < n < 23 whereas MGDC 12.5 for n 4 12 and 9.5 for n 5 23. Similarly, the variation of MGDN with the length of the hydrophobic region was determined. A stretch of n leucine residues (11 4 n 4 29) and one valine residue, ¯anked by an N-terminal SerGln-Gln-Gln segment and four C-terminal lysine residues was inserted in place of the N-terminal transmembrane segment (H1) or in the middle of the large P2 domain in Lep (H3). In both cases, a roughly linear dependence of MGDN on n was found for the entire range of n-values tested (Figure 2(a)), although there is a slight upwards curvature for the shortest H3-constructs. As was noted (Nilsson & von Heijne, 1993), MGDN values are consistently three to ®ve residues larger than the corresponding MGDC values, suggesting that the OST active site is oriented with the Asn-binding pocket closer to the membrane surface than the Ser/Thr-binding pocket. Since Lep uses the SRP/Sec61-translocon pathway for membrane assembly when expressed in the microsomal in vitro system (Mothes et al., 1997), we also wanted to test the relation between MGD and n for a protein that does not use this pathway. Synaptobrevin is known to insert a hydrophobic C-terminal segment into the microsomal membrane by an as yet poorly understood mechanism that does not involve the SRP/Sec61 machinery (Kutay et al., 1995), and it can be glycosylated on its lumenal, C-terminal tail even when the single transmembrane domain is replaced by poly-Leu stretches (Whitley et al., 1996a). We thus determined MGDC vaules for four different constructs derived from human synaptobrevin 2 with poly-Leu transmembrane segments (n 11, 14, 17 and 20); since synaptobrevin is not very ef®ciently targeted to microsomes, all quanti®cations were done with carbonate-extracted membrane pellets. The results for the n 17 constructs are shown in Figure 2(b). Although the MGDC values were found to be increased by about 1.5 residues compared to the H2 data (Figure 2(a)), their dependence on n was virtually identical to that seen for Figure 2. MGD values vary with the length of the hydrophobic segment. (a) MGDN and MGDC values are shown as a function of the number of leucine residues (n) in poly-Leu transmembrane segments of the general sequence LISQQQLnVKKKKHM (H1, H3) or PGLIKKKKLnVQQQP (H2). MGD-values are counted from the ®rst Gln residue before (MGDN) or after (MGDC) the hydrophobic stretch. Filled circles, MGDN-values for H1-poly-Leu segments; open squares, MGDC-values for H2-poly-Leu segments; open circles, MGDN-values for H3-poly-Leu segments; ®lled squares, MGDC-values for poly-Leu segments in the C-terminal tail of synaptobrevin. For synaptobrevin, MGDC-values were based on 30% glycosylation since the maximal glycosylation ef®ciency is only 60±65%. (b) Glycosylation of synaptobrevin-derived constructs with a transmembrane segment of the composition . . . KKKKLnVQQQP . . . and glycosylation acceptor sites placed 11, 13, and 15 residues downstream of the hydrophbobic stretch (counting from the ®rst ¯anking Gln). Microsomes were subjected to a carbonate wash procedure to remove soluble and extrinsic membrane proteins before loading onto the gel. Open and ®lled crcles indicate the non-glycosylated and glycosylated forms of the protein, respectively. the Lep-constructs. The MGDN,C f(n) relationships thus appear to be independent of the mode of membrane assembly, and presumably re¯ect interactions between the transmembrane segments and the lipid bilayer rather than translocon-speci®c interactions. In all cases, an increase in n by four residues leads to a reduction in MGD by about one residue. A possible interpretation of this effect is presented in the Discussion; for now we simply 1168 Glycosylation Mapping of Transmembrane Helices take the MGDN,C f(n) relationships as calibration curves (see below). Helix-breaking effects of C-terminal proline residues Figure 3. Proline residues disrupt transmembrane helices. (a) MGDC values (counting from Gln ÿ 1) for H2 transmembrane segments with n 8, 11, 14, 20, 23, and 29 leucine residues and with proline residue replacement mutations in the indicated positions. The no P value is for a segment without proline reidues (cf. Figure 2(a)), and 23L P6 indicates the MGD-value for a 23L double mutant with proline residues in positions 3 and 6. Position ÿ1 is the Gln residue immediately C-terminal to the hydrophobic segment, and position 1 is the Val residue at the C-terminal end of this segment (i.e. proline residue positions are counted in the C to N-terminal direction). The position in the membrane of the n 23 (Leu5 ! Pro) mutant relative to the n 18 construct is shown below the graph (see the text). To study the conformational effects of proline residues introduced near the C-terminal end of a TMH, a number of residues throughout the H2 n 23 poly-Leu transmembrane stretch were individually substituted by proline, and MGDC was determined for each mutant. As shown in Figure 3(a), the exchange of Gln ÿ 1 or Val1 for Pro had little effect on MGDC, whereas the Leu2 ! Pro substitution led to a substantial decrease in MGDC from 9.7 to 8.5 residues. The effect on MGDC persisted six residues into the hydrophobic segment with a minimum value of MGDC 7.2 for the Leu5 ! Pro substitution, but then quickly disappeared as the proline was moved beyond Leu6. A ®rst interpretation of these results is that the stretch of residues between the end of the hydrophobic segment and the OST active site has a ¯exible, extended conformation that is not affected by the introduction of a helix-breaking proline (mutants Gln ÿ 1 ! Pro, Val1 ! Pro). The Leu2 ! Pro substitution leads to termination of the TMH at the proline, thus pushing one residue out of the membrane to become part of the ¯exible segment. When the proline is inserted deep into the transmembrane segment (beyond Leu6), there is little or no effect on MGDC and hence the proline does not break the helix. The interpretation is somewhat more complicated for the Leu3 ! Pro to Leu6 ! Pro mutants, since MGDC varies with n for transmembrane segments with n 4 23 (Figure 2(a)). Thus, if the helix in the Leu5 ! Pro mutant were to end at Leu6, this would correspond to a TMH with n 18. From Figure 2(a), the n 18 construct has MGDC 11.5 residues (counting from Gln ÿ 1). The Leu5 ! Pro mutant has MGDC 7.2 residues (or 12.2 counting from Pro5), suggesting that the helix indeed terminates around Pro5, as illustrated in Figure 3(a). As a further test, we made a double mutant with Pro both in positions 3 and 6 (denoted 23LP6 in Figure 3(a)); in this case, we found MGDC 5.1 (or MGDC 11.1 counting from Pro6). From Residues assumed to be in a helical conformation are shown in upper case, those in a more ¯exible, extended conformation in lower case. (b) MGDN values (counting from Gln ÿ 1) for H1 and H3 transmembrane segments with n 23 leucine residues and with proline replacement mutations in the indicated positions. The no P value is for a segment without proline (cf. Figure 2(a)). Position ÿ1 is the Gln residue immediately N-terminal to the hydrophobic segment, and position 1 is the Leu residue at the N-terminal end of this segment (i.e. proline residue positions are counted in the N to C-terminal direction). Glycosylation Mapping of Transmembrane Helices Figure 2(a), if the transmembrane segment ends at Leu7 in this case (i.e. n 17), we should have MGDC 11.7. It is thus apparent that the entire 23residue long TMH is either beginning to re-form in the Leu6 ! Pro mutant, or that the short hydrophobic stretch between Leu5 and Val1 somehow dips back into the membrane; in either case, the introduction of the additional Pro3 mutation obviously pushes the whole stretch towards the lumen. In summary, we conclude that a proline inserted into the C-terminal 1.5 turns of the n 23 TMH has a strongly disrupting effect, whereas more deeply buried proline residues do not break the helix. MGDC-values were also measured for proline substitutions in H2 transmembrane segments composed of one valine and n 8, 11, 14, 20, and 29 leucine residues, (Figure 3(a)); i.e. starting from the shortest poly-Leu stretch known to function as an ef®cient signal-anchor sequence in the microsomal system (Sakaguchi et al., 1992). In all cases except n 8 and 11, the MGDC-values dropped by about one residue in the Val1 ! Pro mutant and by about two residues in the Leu2 ! Pro mutant, consistent with a disruption of the TMH. Interestingly, Leu ! Pro substitutions deeper into the hydrophobic stretch were more disruptive in the longer segments: the Leu3 ! Pro mutation had only a weak disruptive effect for n 14, whereas a signi®cant disruptive effect was apparent even for the Leu9 ! Pro substitution in the n 29 construct. Leu ! Pro substitutions had negligible effects on MGDC for n 8. The observation that proline residues have no effect on MGDC for n 8 and only strongly affect MGDC for the n 14 segment in positions Val1 and Leu2 but not Leu3 further suggests that a Ê long) ``core helix'' of about 13 residues (i.e. 20 A is always present, irrespective of amino acid comÊ position. This ®ts nicely with the roughly 20 A wide central hydrophobic core evident in the crystal stuctures of the photosynthetic reaction center and mitochondrial cytochrome c oxidase (Wallin et al., 1997), where disruption of the helical conformation would be very costly (Popot & Engelman, 1990). It also ®ts well with a recent molecular dynamics simulation of a transmembrane poly-Ala helix embedded in a DMPC bilayer, where the central 12 residues were found to form a very stable a-helix (Shen et al., 1997). Helix-breaking effects of N-terminal proline residues We also tested the effect on MGDN of a proline residue inserted near the N-terminal end of a transmembrane segment. A stretch of 23 leucine residues and one valine, ¯anked by an N-terminal Ser-Gln-Gln-Gln segment and four C-terminal lysine residues was inserted in place of the N-terminal transmembrane segment (H1) in Lep (Figure 1(a)). As shown in Figure 3(b), the Gln ÿ 1 ! Pro substitution reduced MGDN by 1169 about two residues, suggesting that the helix normally extends almost to the N-terminal end of the Ser-Gln-Gln-Gln segment: conversion of four helical residues to an extended conformation will increase the length spanned by about Ê , which corresponds to a 4 (3.3 ÿ 1.5) 7.2 A movement of the glycosylation acceptor site by about 8/3.3 2.2 residues relative to the OST active site; also, Ser is a common N-cap and Gln a common N3 residue in helices in globular proteins (Harper & Rose, 1993). The MGDN value was reduced by another 1.6 residues in the Leu1 ! Pro mutant (which effectively shortens the hydrophobic stretch by one residue), and then increased as the proline was moved further into the hydrophobic stretch. Again, a detailed interpretation requires that the variation in MGDN with the length of a transmembrane segment is taken into account. As shown in Figure 2(a), MGDN increases by about 0.25 residues for every leucine residue removed from the transmembrane segment. If the Leu1 ! Pro mutation were to lead to termination of the helix at the proline residue, the expected increase in MGDN compared to the Gln ÿ 1 ! Pro mutant would thus be 1.25 residues, in reasonable agreement with the observed value. For the Leu3 ! Pro and Leu6 ! Pro mutations, the observed values would suggest that the helix ends near Leu2 in both cases. This is consistent with the obsevation that proline resudes are often found in the three most N-terminal positions in a-helices both in globular and integral membrane proteins but not further in (Richardson & Richardson, 1988; Wallin et al., 1997). Thus, N-terminal proline residues also seem to cause a break in the n 23 TMH when present in the ®rst one to two helical turns. Similar MGDN values were obtained when proline residue insertions were made in a 23-residues long poly-leucine transmembrane stretch (H3) placed in the middle of the large C-terminal domain, Figure 3(b), demonstrating that the location within the protein does not much in¯uence the results. Helix-breaking effects of N-terminal proline residues assayed in E. coli Since the immediate environment of a TMH in the ER membrane at the time of glycosylation is dif®cult to characterize in detail (see Discussion), we also carried out protease-protection experiments in E. coli. To this end, we expressed a Lepderived construct where H1 has been replaced by a 20-residues long (Ala, Leu) stretch (Whitley et al., 1994) and where H2 has the composition L7ML15, together with two further mutants in H2 with the substitutions Leu5 ! Pro and Leu10 ! Pro (counting from the N-terminal end of the hydrophobic stretch), and probed the location of the N-terminal end of H2 relative to the inner membrane by proteinase K digestion of spheroplasts (Figure 4). In this case, there is little doubt that the TMH is 1170 Figure 4. Disruption of a poly-Leu transmembrane helix by proline residues assayed in E. coli. Lepderived constructs with an ``inverted'' topology, H1 replaced by a 20-residues long (Ala, Leu) stretch (Whitley et al., 1994), and H2 replaced either by the sequence . . . N3L7ML15K4 . . . (lanes 1, 4), by this sequence with a Leu5 ! Pro replacement (counting from the N-terminal leucine; lanes 2, 5), or a Leu10 ! Pro replacement (lanes 3, 6) were expressed in E. coli and labelled for one minute by [35S]Met. Cells were converted to spheroplasts, digested with proteinase K, lysed, and subjected to immunoprecipitation with a Lep antibody. The protease-protected fragments (lanes 4-6) represent the H2-P2 domain. The faint lower Mr band in lanes 1-3 probably represents an endogenous proteolysis product cleaved in the periplasmic loop. embedded in the lipid bilayer when the experiment is performed. All three Lep constructs expose a short, protease-sensitive loop between H1 and H2 to the periplasm, and proteinase K digestion will produce a protected fragment composed of the H2P2 domain (Whitley et al., 1994). Signi®cantly, the protease-protected fragment in the Leu5 ! Pro mutant was considerably smaller than in the construct without proline residues or in the Leu10 ! Pro mutant (Figure 4; lanes 4-6), indicating that the N-terminal part of H2 becomes more accessible to protease added from the periplasmic side when the proline is placed in the ®rst two turns of the helix. Although much less precise than the glycosylation mapping measurements, the protease accessibility assay thus yields similar results, strengthening the assumption that the TMH is at least partly embedded in the lipid bilayer when glycosylation takes place. The photosynthetic reaction center H-subunit transmembrane helix In order to calibrate the glycosylation mapping results against a TMH with known 3D structure, we inserted the N-terminal TMH (plus some ¯ank- Glycosylation Mapping of Transmembrane Helices ing residues) from the H-subunit of the R. sphaeroides photosynthetic reaction center (Allen et al., 1987; Rees et al., 1989) in place of the H1 or H2 transmembrane segment. Judging from the X-ray structure of the reaction center (Figure 5(a)), the H-subunit helix does not interact extensively with the other transmembrane helices and extends roughly between Leu12 and Thr33, which are probably located in the lipid headgroup regions (Wallin et al., 1997). As illustrated in Figure 5(b), the N-terminal MGDN was found to be 13.2 residues (counting from Asp11) and the C-terminal MGDC was found to be 10.1 residues (counting from Glu34). To assess the effect of proline insertions on MGDC, we changed Tyr30 to Pro; for this mutant we found an MGDC value of only 4.6 residues (8.6 residues counting from Pro30), i.e., a decrease of 5.5 residues. The helix-breaking effect of the proline is thus even stronger in this context, possibly because the residues C-terminal to the proline residue are less hydrophobic in the H-subunit helix than in the poly-Leu helices and thus more easily pushed out of the membrane. In order to provide an independent check on the relation between MGD and the length of the transmembrane segment (Figure 2(a)), two further mutations were made in the H-subunt helix placed in the H1 location: one insertion and one deletion, each of four hydrophobic residues. The MGDN values were found to be 12.1 and 13.7 residues, respectively (data not shown), corresponding to a decrease in MGDN of 0.8 residues for every four residues added to the transmembrane segment. This is in reasonable agreement with the value found above for the H1 poly-Leu segment (decrease of one residue for every four added; Figure 2(a)), and suggests that the results obtained with the poly-Leu model sequences are representative also for natural transmembrane segments. The results for the H-subunit helix allow the n 20 poly-Leu helix to be positioned relative to the H-subunit helix (which contains a 20-residue long hydrophobic stretch between Leu12 and Leu31; Figure 5(b) (bottom). The MGDN-vaule for the H1 n 20 construct is 16.3, but, as shown above, this value should be reduced by about two to correct for the assumed N-terminal extension of the helix through the Ser-Gln-Gln-Gln segment. This predicts that the N-terminal end of the n 20 poly-Leu stretch is located at about the same position in the membrane as Ala13 in the H-subunit helix. The MGDC vaule for the n 20 construct is 11.1 residues, roughly placing the valine at the end of the poly-Leu stretch in an equivalent position to Leu31 in the H-subunit helix. The location of the poly-Leu transmembrane segment relative to the H-subunit helix deduced from the glycosylation mapping data is thus as expected from their hydrophobicity pro®les. 1171 Glycosylation Mapping of Transmembrane Helices Discussion We have developed a new method, glycosylation mapping, that makes it possible to study the position of transmembrane helices in a natural membrane. Our results do not directly address the question of the molecular environment of the TMH at the time of glycosylation, although they are consistent with a predominantly non-polar milieu. Since glycosylation near the MGDN of a transmembrane hydrophobic stretch is observed even when truncated mRNA transcripts lacking a stop-codon are used (Whitley et al., 1996b), it is clear that the modi®cation can take place within the context of the ER translocon, which is known to include both the Sec61 translocation complex and the oligosaccharyl transferase (GoÈrlich & Rapoport, 1993; Rapoport et al., 1996). The recent demonstration that hydrophobic transmembrane segments in general (and the Lep H1 segment in particular) become exposed to lipids very soon after entering the translocation channel (Martoglio et al., 1995; Mothes et al., 1997) nevertheless suggests that the MGD-values measured here pertain to a situation where the TMH is in a predominantly lipidic environment. This conclusion is further supported by the observation that the variation in MGD vaules with the length of the poly-Leu transmembrane stretch is essentially the same, irrespective of whether or not membrane assembly proceeds through the SRP/Sec61-pathway (Figure 2(a)). A ®nal argument in favor of a lipidic environment is the observation that Leu ! Pro mutations have similar effects on the position of a transmembrane segment in the membrane as probed by glycosylation mapping in the ER membrane or by protease trimming in the inner membrane of E. coli (Figures 3 and 4). In a ®rst detailed application of the glycosylation mapping approach, we have found that proline residues break a poly-Leu TMH when inserted as far as one to two helical turns from the ends of the hydrophobic stretch. In contrast to what has been found for globular proteins (Barlow & Thornton, 1988), proline residues appear not to have any gross conformational effects when placed more centrally in the TMH, as also observed in prolinecontaining transmembrane helices in proteins of known 3D structure such as the photosynthetic reaction center, bacteriorhodopsin, and cytochrome c oxidase (Iwata et al., 1995; Sansom, 1992; Tsukihara et al., 1996; von Heijne, 1991; Wallin Figure 5. MGD-determination for the R. sphaeroides reaction center H-subunit transmembrane helix. (a) Location of the H-subunit transmembrane helix in the reaction center. Leu12 and Thr33 are shown as CPK models. Co-ordinates are from the PDB ®le 2RCR (Grigorieff et al., 1996) and the plot was made using MOLSCRIPT (Kraulis, 1991). (b) Glycosylation ef®ciency for acceptor sites located at different distances d from the H-subunit helix placed in the H2 position and for the indicated Tyr30 ! Pro mutant (counting from Glu34; note that the MGDC value for the latter is determined by extrapolation from the d 5 construct). The position in the membrane relative to the H-subunit transmembrane helix of the n 20 poly-Leu transmembrane helix is shown at the bottom. Residues assumed to be in a helical conformation are shown in upper case, those in a more ¯exible, extended conformation in lower case. 1172 et al., 1997). The helix-disrupting effect is only seen when the hydrophobic segment is longer than 13 residues, suggesting that a core helix of this length will always be present in any transmembrane segment, irrespective of amino acid sequence. By calibrating our measurements against the TMH from the H-subunit of the photosynthetic reaction center, we have found that a model transmembrane segment composed of one valine and 20 leucine residues is located in a position in the membrane equivalent to that occupied by the Ala13-Leu31 segment in the reaction center H-subunit (as shown in the accompanying paper; Monne et al., 1998), very similar results are obtained with the M13 coat protein TMH which also has a known position in the bilayer). This not only reinforces our structural interpretation of the data, but also suggests that glycosylation mapping may be used as a general method for determining the position of transmembrane helices in the membrane. It has recently been shown that analysis of a series of glycosylation acceptor sites placed throughout a loop between two transmembrane helices can give a rough idea of where the helix ends are located (Popov et al., 1997); with the increased precision obtainable from a comparison to the H-subunit and phage M13 coat protein helices (see Monne et al., 1998), this kind of information may provide important constraints for model building exercises. What, ®nally, does the striking correlation between the MGD-values and the length of the hydrophobic transmembrane segment mean in structural terms? As is clear from Figure 2(a) and from the equivalent deletion/insertion experiment on the H-subunit helix, the relation MGD 0.25n (where n is the length of the hydrophobic segment) holds over an extended range of n-values irrespective of whether MGDN or MGDC is measured, and irrespective of whether the protein is inserted into the membrane via the SRP/Sec61 pathway or not. A possible explanation is based on the model shown in Figure 6, which assumes that a helix comprised of only 13 hydrophobic residues is Ê wide hydrophobic core enough to span the 20 A of the lipid bilayer (cf. Figure 3(a)), whereas helices composed of up to 25 hydrophobic residues can still be accomodated within a reasonably hydroÊ thickness phobic bilayer environment of 35 ± 40 A (cf. (Wallin et al., 1997). As illustrated in the Figure, this model predicts a relation of the form MGD 0.75/3.3n 0.23n, which is reasonably close to the observed one. For very long helices that would extend into the aqueous phase if perpendicular to the bilayer, tilting would be a way to bury more of the hydrophobic surface area. If this happens, MGD would increase much more slowly, if at all, with n, as seen for the H2 transmembrane segment in Figure 2(a). However, for the H1 and H3 transmembrane segments, the linear decrease in the MGDN-values seems to continue well beyond n 23; at present, we have no good explanation for this discrepancy. Glycosylation Mapping of Transmembrane Helices Figure 6. Model for the relation between MGD-values and the length of a hydrophobic transmembrane helix. For short segments, it is assumed that a perpendicular orientation has the lowest free energy. An increase in length by one hydrophobic residue extends the helix by Ê ; if the locations of the two helix ends are symme1.5 A Ê closer to the trically affected, each end moves 0.75 A membrane surface. To reduce the MGD-value by one residue, the linker segment (assumed to be in a ¯exible, extended conformation) between the transmembrane Ê, helix and the OST active site needs to move 3 ± 3.5 A i.e. MGD 0.75/3.3n 0.23n. Y, glycosylated acceptor site; non-glycosylated acceptor site. In summary, the glycosylation mapping technique has allowed us to make a detailed and comprehensive analysis of the helix-breaking effects of proline residues in transmembrane helices of different lengths (a similar analysis of the effects of charged residues is presented in the accompanying paper). The approach also opens up a possibility to deduce the position in the membrane of transmembrane segments of arbitrary amino acid sequence relative to ``standards'' such as the reaction center H-subunit TMH. Since glycosylation is in most cases a co-translational event, the technique cannot be applied to fully folded, multi-spanning membrane proteins; however, the position in the membrane of individual transmembrane helices is in most cases unlikely to change dramatically upon folding (Popot & Engelman, 1990). 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, BSA, Sp6 RNA polymerase, RNasin and rabbit reticulocyte lysate were from Promega. Spermidine was from Sigma. Oligonucleotides were from Kebo Lab (Stockholm, Sweden). 1173 Glycosylation Mapping of Transmembrane Helices 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 (Kozak, 1989). Thus, the 50 region of the gene was modi®ed to: . . . ATAACCCTCTAGAGCCACCATGGCGAAT . . . (XbaI site and initiator codon underlined). Replacement of the H2 segment 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 with the appropriate doublestranded oligonucleotides. Residues 59-81 in H2 were replaced by residues Val4-Glu38 from the R. sphaeroides reaction center H-subunit or by poly-Leu sequences of the general design PGLIKKKKLnVQQQP, where the valine residue at the end of the poly-Leu stretch was included to obtain a SpeI restriction site in this position. The H1 segment was replaced by ®rst introducing BclI and NdeI restriction sites in codons 4 and 22, and then replacing the BclI-NdeI fragment with the appropriate double-stranded oligonucleotides encoding either residues 4-38 from the reaction center H-subunit or poly-Leu segments of the general design LISQQQLnVKKKKHM. The H3 segment was replaced by ®rst introducing BclI and NdeI restriction sites in codons 213 and 221, and then replacing the BclI-NdeI fragment by the appropriate double-stranded oligonucleotides encoding poly-Leu segments of the general design LISQQQLnVKKKKHM. 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-Thr-Ser acceptor sites for N-linked glycosylation was performed according to the method of Kunkel (Geisselsoder et al., 1987; Kunkel, 1985). Glycosylation acceptor sites were designed as described (Nilsson & von Heijne, 1993; Nilsson et al., 1994), i.e. by replacing three appropriately positioned codons upstream of H1 and H3 or downstream of H2 with codons for the acceptor tri-peptide Asn-Ser-Thr. In the glycosylation construct Asn84-Ser85-Thr86 and Asn88-Ser89-Thr90 (numbering corresponding to the Lep wild-type sequence), the ¯anking Pro residues were changed to Gln since proline residues were found to reduce the ef®ciency of glycosylation, cf. (Gavel & von Heijne (1990). To make the constructs with d 5, 6, and 7 (counting from Glu34) for the Tyr30 ! Pro mutant in the H-subunit transmembrane helix, Ser38, Arg37, and Met36 in the H-subunit segment (see Figure 5(b)) were progressively deleted from the d 8 construct. To make glycosylation sites six and ®ve residues downstream of Val1, the SpeI-NdeI fragment was replaced by double-stranded oligonucleotides, deleting one or two glutamine residues. All mutants were con®rmed by DNA sequencing of plasmid or single-stranded M13 DNA using T7 DNA polymerase. The synaptobrevin series of constructs were made as detailed by (Whitley et al., 1996a). The natural transmembrane segment M95MIILGVICAIILIIIIAYV was replaced with M95MIKKKKLnVQQQPYV. For expression and protease-protection experiments in E. coli, the TM2 segment in construct 2 K/TM1/0 K/ TM2 (Whitley et al., 1994) between the NheI and KpnI sites was replaced by double-stranded oligonucleotides encoding, respectively, the sequences N3L7ML15K4, N3L4PL2ML15K4, and N3L7MLPL13K4. Expression in vitro Synthesis of mRNA from pGEM1 by SP6 RNA polymerase and translation in reticulocyte lysate in the presence of dog pancreas microsomes was performed as described (LiljestroÈm & Garoff, 1991). For the synaptobrevin constructs, the microsomes were subjected to a sodium carbonate wash procedure (Sakaguchi et al., 1987) before being loaded onto the gel. Proteins were analyzed by SDS-PAGE and gels were quanti®ed on a Fuji BAS1000 phosphoimager using the MacBAS 2.1 software. The extent of glycosylation 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. In general, the glycosylation ef®ciency varies by no more than 5% between different experiments, and the precision in the MGD determinations is 0.2 residues. Expression in E. coli Experiments were performed in E. coli strain MC1061 (lacX74, araD139, (ara, leu)7697, galU, galK, hsr, hsm, strA; Dalbey & Wickner, 1986). Constructs were expressed from the pING1 plasmid (Johnston et al., 1985) by induction with arabinose. MC1061 cells transformed with the pING1 vector carrying the relevant constructs under control of the arabinose promoter were grown at 37 C in M9 minimal medium supplemented with 100 mg/ml ampicillin, 0.5% (w/v) fructose, 100 mg/ml thiamine, and all amino acids (50 mg/ml each) except methionine. An overnight culture was diluted 1:25 in fresh medium, shaken for 3.5 hours at 37 C, induced with arabinose (0.2% (w/v)) for ®ve minutes, labeled with [35S]methionine (75 mCi/ml) for one minute, and put on ice. Cells were spun at 14,000 rpm for two minutes, resuspended in ice-cold buffer (40% (w/v) sucrose, 33 mM Tris, pH 8.0), and incubated with lysozyme (5 mg/ml) and 1 mM EDTA for 15 minutes on ice. Aliquots of the cell suspension were incubated 60 minutes on ice, either with no additions, or with the addition of 400 mg/ml proteinase K. After addition of PhMeSO2F (PMSF), samples were acid-precipitated with trichloroacetic acid (TCA, 10% (v/v) ®nal concn), resuspended in 10 mM Tris/2% SDS, immunoprecipitated with antisera to Lep, washed, and analyzed by SDS-PAGE. Gels were scanned in a FUJIX Bas 1000 phosphoimager. 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 Allen, J., Feher, G., Yeates, T., Komiya, H. & Rees, D. (1987). Structure of the reaction center from Rhodobacter sphaeroides R-26: the protein subunits. Proc. Natl Acad. Sci. USA, 84, 6162± 6166. Barlow, D. J. & Thornton, J. M. (1988). Helix geometry in proteins. J. Mol. Biol. 201, 601± 619. Dalbey, R. E. & Wickner, W. (1986). The role of the polar, carboxyl-terminal domain of Escherichia coli 1174 leader peptidase in its translocation across the plasma membrane. J. Biol. Chem. 261, 13844± 13849. Deber, C. M. & Goto, N. K. (1996). Folding proteins into membranes. Nature Struct. Biol. 3, 815± 818. Gavel, Y. & von Heijne, G. (1990). Sequence differences between glycosylated and non-glycosylated Asn-XThr/Ser acceptor sites ± implications for protein engineering. Protein Eng. 3, 433±442. Geisselsoder, J., Witney, F. & Yuckenberg, P. (1987). Ef®cient site-directed in vitro mutagenesis. BioTechniques, 5, 786± 791. GoÈrlich, D. & Rapoport, T. A. (1993). Protein translocation into proteoliposomes reconstituted from puri®ed components of the endoplasmic reticulum membrane. Cell, 75, 615± 630. Grigorieff, N., Ceska, T. A., Downing, K. H., Baldwin, J. M. & Henderson, R. (1996). Electron-crystallographic re®nement of the structure of bacteriorhodopsin. J. Mol. Biol. 259, 393± 421. Harper, E. T. & Rose, G. D. (1993). Helix stop signals in proteins and peptides: the capping box. Biochemistry, 32, 7605± 7609. Iwata, S., Ostermeier, C., Ludwig, B. & Michel, H. Ê resolution of cytochrome c (1995). Structure at 2.8 A oxidase from Paracoccus denitri®cans. Nature, 376, 660± 669. 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. Johnston, S., Lee, J. H. & Ray, D. S. (1985). High-level expression of M13 gene II protein from an inducible polycistronic messenger RNA. Gene, 34, 137± 145. 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. Kraulis, P. J. (1991). MOLSCRIPT: A program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946± 950. Kunkel, T. A. (1985). Rapid and ef®cient site-speci®c mutagenesis without phenotypic selection. Proc. Natl Acad. Sci. USA, 82, 488± 492. Kutay, U., Ahnerthilger, G., Hartmann, E., Wiedenmann, B. & Rapoport, T. A. (1995). Transport route for synaptobrevin via a novel pathway of insertion into the endoplasmic reticulum membrane. EMBO J. 14, 217±223. Li, S.-C. & Deber, C. M. (1994). A measure of helical propensity for amino acids in membrane environments. Nature Struct. Biol. 1, 368± 373. Li, S.-C., Goto, N. K., Williams, K. A. & Deber, C. M. (1996). a-Helical but not b-sheet, propensity of proline is determined by peptide environment. Proc. Natl Acad. Sci. USA, 93, 6676± 6681. 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., Li, S. C., Goto, N. K. & Deber, C. M. (1996). Threshold hydrophobicity dictates helical conformations of peptides in membrane environments. Biopolymers, 39, 465± 470. Martoglio, B., Hofmann, M. W., Brunner, J. & Dobberstein, B. (1995). The protein-conducting channel in the membrane of the endoplasmic reticulum is open laterally toward the lipid bilayer. Cell, 81, 207± 214. Glycosylation Mapping of Transmembrane Helices Mellquist, J. L., Kasturi, L., Spitalnik, S. L. & ShakinEshleman, S. H. (1998). The amino acid following an Asn-X-Ser/Thr sequon is an important determinant of N-linked core glycosylation ef®ciency. Biochemistry, 37, 6833± 6837. MonneÂ, M., Nilsson, I., Elmhed, N., Johansson, M. & von Heijne, G. (1998). Positively and negatively charged residues have different effects on th position in the membrane of a model transmembrane helix. J. Mol. Biol.. 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. 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., 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. Papavoine, C., Remerowski, M., Horstink, L., Konings, R., Hilbers, C. & van de Ven, F. (1997). Backbone dynamics of the major coat protein of bacteriophage M13 in detergent micelles by 15N nucelar magnetic resonance relaxation measurements using the model-free approach and reduced spectral density mapping. Biochemistry, 36, 4015± 4026. Popot, J. L. & Engelman, D. M. (1990). Membrane protein folding and oligomerization ± the 2-stage model. Biochemistry, 29, 4031± 4037. Popov, M., Tam, L. Y., Li, J. & Reithmeier, R. A. F. (1997). Mapping the ends of transmembrane segments in a polytopic membrane protein ± scanning N-glycosylation mutagenesis of extracytosolic loops in the anion exchanger, Band 3. J. Biol. Chem. 272, 18325± 18332. Rapoport, T. A., Jungnickel, B. & Kutay, U. (1996). Protein transport across the eukaryotic endoplasmic reticulum and bacterial inner membranes. Annu. Rev. Biochem. 65, 271±303. Rees, D. C., Komiya, H., Yeates, T. O., Allen, J. P. & Feher, G. (1989). The bacterial photosynthetic reaction center as a model for membrane proteins. Annu. Rev. Biochem. 58, 607± 633. Richardson, J. S. & Richardson, D. C. (1988). Amino acid preferences for speci®c locations at the ends of a-helices. Science, 240, 1648± 1652. 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. Sakaguchi, M., Tomiyoshi, R., Kuroiwa, T., Mihara, K. & Omura, T. (1992). Functions of signal and signalanchor sequences are determined by the balance between the hydrophobic segment and the N-terminal charge. Proc. Natl Acad. Sci. USA, 89, 16 ±19. Sansom, M. S. P. (1992). Proline residues in transmembrane helices of channel and transport proteins: a molecular modelling study. Protein Eng. 5, 53 ± 60. Shakineshleman, S. H., Spitalnik, S. L. & Kasturi, L. (1996). The amino acid at the X position of an AsnX-Ser sequon is an important determinant of N-linked core-glycosylation ef®ciency. J. Biol. Chem. 271, 6363± 6366. Shen, L., Bassolino, D. & Stouch, T. (1997). Transmembrane helix structure, dynamics, and interactions: Glycosylation Mapping of Transmembrane Helices multi-nanosecond molecular dynamics simulations. Biophys. J. 73, 3 ± 20. Shon, K. J., Kim, Y. G., Colnago, L. A. & Opella, S. J. (1991). NMR studies of the structure and dynamics of membrane-bound bacteriophage-P¯ coat protein. Science, 252, 1303± 1304. Stopar, D., Spruijt, R. B., Wolfs, C. & Hemminga, M. A. (1996). Local dynamics of the M13 major coat protein in different membrane-mimicking systems. Biochemistry, 35, 15467± 15473. Tsukihara, T., Aoyama, H., Yamashita, E., Tomizaki, T., Yamaguchi, H., Shinzawa-Itoh, K., Nakashima, R., Yaono, R. & Yoshikawa, S. (1996). The whole stucture of the 13-subunit oxidized cytochrome c oxiÊ . Science, 272, 1136± 1144. dase at 2.8 A von Heijne, G. (1991). Proline kinks in transmembrane a-helices. J. Mol. Biol. 218, 499± 503. von Heijne, G. (1996). Principles of membrane protein assembly and structure. Prog. Biophys. Mol. Biol. 66, 113± 139. 1175 Wallin, E., Tsukihara, T., Yoshikawa, S., von Heijne, G. & Elofsson, A. (1997). Architecture of helix bundle membrane proteins: an analysis of cytochrome c oxidase from bovine mitochondria. Protein Sci. 6, 808± 815. Whitley, P., Nilsson, I. & von Heijne, G. (1994). De novo design of integral membrane proteins. Nature Struct. Biol. 1, 858±862. Whitley, P., Grahn, E., Kutay, U., Rapoport, T. & von Heijne, G. (1996a). A 12 residues long poly-leucine tail is suf®cient to anchor syntaptobrevin to the ER membrane. J. Biol. Chem. 271, 7583± 7586. Whitley, P., Nilsson, I. M. & von Heijne, G. (1996b). A nascent secretory protein may traverse the ribosome/ER translocase complex as an extended chain. J. Biol. Chem. 271, 6241± 6244. 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 15 May 1998; received in revised form 1 September 1998; accepted 1 September 1998)