The role of the mature part of secretory proteins in translocation across the plasma membrane and in regulation of their synthesis in Escherichia coli

Biochimie (1990) 72, 157- 167 @ SociCte de Chimie biologique / Elsevier, Paris

157

The role of the mature part of secretory proteins in translocation across the plasma membrane and in regulation of their synt ia coli S MacIntyre*,

U Henning

Max-Planck-Institutfiir Biologie, D-7400 Tiibingen, FRG (Received 10 November 1989; accepted 4 December

1989)

Summary - Presently available data are reviewed which concern the role of the mature parts of secretory precursor proteins in translocation across the plasma membrane of Escherichia coli. The following conclusions can be drawn; i) signals, acting in a positive fashion and required for translocation do not appear to exist in the mature polypeptides; ii) a number of features have been identified which either affect the efficiency of translocation or cause export incompatibility. These are: (Y)protein folding prior to translocation; 0) restrictions regarding the structure of N-terminus; y) presence of lipophilic anchors; S) too low a size of the precursor. Efficiency of translocation is also enhanced by binding of chaperonins (SecB, trigger factor, GroEL) to precursors. Binding sites for chaperonins appear to exist within the mature parts of the precursors but the nature of these sites has remained rather mysterious. Mutant periplasmic proteins with a block in release from the plasma membrane have been described, the mechanism of this block is not known. The mature parts of secretory proteins can also be involved in the regulation of their synthesis. It appears that exported proteins are already recognized as such before they are channelled into the export pathway and that their synthesis can be feed-back inhibited at the translational level. protein export /export incompatibility / chaperonins/ sorting control

Introduction The vast majority of both prokaryotic and eukaryotic secretory proteins are synthesized with an N-terminal signal or leader peptide, which is removed during or following translocation to release the mature polypeptide. If a cleavable signal peptide is present, it is an absolute prerequisite for translocation; its deletion abolishes export. The structure and role of the signal peptide has been studied extensively (for reviews, see [l-6]). There is no common consensus sequence among different signal peptides, but they all share the same structural composition: an N-terminal positive charge, a hydrophobic core and a cleavage area. In Escherichia coli, signal peptide mediated export of secretory proteins involves, as a general rule, the cellular SecA / SecY machinery and to varying extents one of the chaperonins, eg, SecB [2, 71 and, in this process, the signal peptide appears to have several, although as yet undefined, functions. In contrast, the role of the mature polypeptide in translocation has remained a question of debate since the existence and importance of the N-terminal signal were first recognized [5, 8- 151. The central question here is whether the mature part of the precursor plays any active role in export or whether it is simply translocated across the membrane as an inactive passenger. As many secretory proteins *Correspondence

and reprints

use a common export machinery and signal peptides are not specific for their own mature polypeptide [16-181, any active role of the mature polypeptide should be reflected in some type of conunon sequence or structural information. In this article, we review this possibility together with the alternative, ie that the mature polypeptide plays no active role but at the same time must possess no sequence or structural property which might inhibit export. This review is limited to the role of the mature polypeptide in signal peptide mediated translocation across the plasma membrane of E coli. Therefore, we have excluded from our discussions: signal peptide independent mechanisms of translocation such as that for haemolysin [19] or flagellin [20]; translocation across the outer membrane (eg [19,21], recent review: [ala]) and insertion into the plasma membrane [22]. However, because the mechanisms of translocation across the plasma membrane of E coli and the endoplasmic reticulum (ER) of eukaryotes appear to be fundamentally very similar and, therefore, results from studies in one system are also frequently relevant to the other, where appropriate we have included information derived from studies in eukaryotes. Finally, since it has become evident that the mature parts of secretory proteins can play a role in the regulation of their synthesis, we include a chapter reviewing this very interesting phenomenon.

158

S Maclntyre,U Henning

Export signals within the mature sequence? An important question in determining the role of the mature polypeptide in export is: does this part of the protein possess any unique sequence(s) which is absolutely required for translocation across the plasma membrane? From studies involving analyses of deletion mutants of 2 different outer membrane proteins, OmpA [14, 23-25] and PhoE [26-28], the answer to this question must be no. In these experiments, a series of internal overlapping deletions were constructed within the ompA and phoE genes, and, to avoid the type of erroneous results which are known to occur with abnormal proteins following cell fractionation [1, 29], products were located by electron microscopic examination of immuno-gold labelled samples. Although taken together these deletions spanned the entire mature polypeptide in the case of pro-OmpA and all but the first two N-terminal residues of mature PhoE, all products were successfully translocated across the inner membrane, processed by leader peptidase and localized to either the periplasm or the outer membrane. As the OmpA protein is thought to be assembled into the outer membrane in a series of 4 B-loop structures, encompassing residues 1-170 of the 325 residue [30] protein [31,32], it had also been considered that an "export signal" may be represented by a repeating secondary structure rather than by a specific amino acid sequence, but following deletion of the entire membrane region plus part of the periplasmic domain (OmpA A 1-229) the remaining part of the carboxy terminus was still localized to the periplasm [24]. It could therefore be concluded that no essential export signal exists within the mature part of these proteins and that the leader sequence can, in theory, mediate export of any part of the mature protein. From earlier studies with pro-LamB-/3-galactosidase hybrids, it had been proposed that an essential export signal ties between residues 27 and 39 of the mature part of LamB [33]. More recent analyses of numerous LamB deletion mutants, which together possessed over-lapping deletions covering the region from residues 1-93 of the mature polypeptide, have demonstrated reasonable, if somewhat variable, rates of translocation (determined by leader peptidase cleavage) in all mutants [34]. Thus, there can be no absolute requirement for this region of LamB for export. Also, deletions have been described in the mature portion of several other exported proteins including maltose-binding protein (MalE) [35-38], phosphodiesterase GlpQ [39], and/3-1actamase [4, 40] and thus far none have led to abortive translocation. Two additional observations lend further credence to the conclusion that no essential positive export signal exists within the mature polypeptides of secretory proteins. Firstly, unlike the plethora of point mutations or

deletions described within the leader peptide, which severely inhibit or abolish t~anslocation (reviewed in [2, 3]), this type of mutation has only been rarely observed within the mature polypeptide (see below). Secondly, one would not expect to find this type of sequence / export signal in no,l-secretory proteins and yet at least 2 non-exported polypeptides, a T4 phage tail fibre fragment [42] and chicken triosephosphate isomeras~. [41] were exported to the periplasm in E coli when fused in the first case directly to the OmpA signal sequence and in the second case to the fl-lactamase signal plus a few N-terminal residues of mature/3lactamase. Also, g-globulin was translocated across the ER membrane when fused to the signal peptide plus 5 residues of mature/3-1actamase [11]. The first 19 Nterminal residues of mature B-lactamase are known to be non-essential for export [41]. In summary, all the evidence together rather strongly argues against the existence of positively acting signals, within the mature part of secretory proteins, which are required for translocation of the polypeptide across the plasma membrane.

Factors affecting the efficiency of translocation or leading to export incompatibility The apparent absence of unique export signals within the mature parts of secretory proteins does not mean that the nature of this part of a precursor is unimportant for translocation. In recent years a number of structural features have been recognized which, when present within the mature moiety, either reduce the efficiency of translocation or render the precursor more or less IIlUUIIIp~I.LIUIU WIL[I UXpU[L.

Importance of conformation Using MalE, Randall and Hardy [43] were able to show a correlation between export compatibility and the absence of a stable tertiary structure. The precursor of MalE, like MalE itself, can fold into a stable conformation which is resistant to proteolytic degradation. Thus, using sensitivity to proteolysis as a monitor to follow the folding rates in vivo they could show that the rapid loss of export competence of a mutant with a defective signal peptide closely paralleled formation of a stable conformation. The wild type precursor, in contrast, only very gradually lost both its protease sensitivity and export competence. These studies complemented those from Eilers and Schatz [44] which had demonstrated that import of cytosolic murine dihydrofolate reductase (DHFR) into mitochondria was abolished if the conformation of the reductase was stabilized by the inhibitor methotrexate. Taken together, these observations have led to the generally accepted hypothesis that flexibility of structure is a prerequisite

Protein translocation across the E coli plasma membrane

for the translocation step, whether across the ER in eukaryotes, the plasma membrane in bacteria, or for import into mitochondria. Thus, a feasible explanation has been provided for the occasional occurrence of mutations, within the mature polypeptide, which are capable of weakly suppressing defects in the signal peptide. Three examples of this type of mutation include mutations within MalE at positions 19 (Gly to Val) [13], 19 (Gly to Cys) [45] and 283 (Tyr to Asp) [46] respectively. In the latter case, it has been shown directly that the mutation results in a decrease of the refolding kinetics of purified, denatured preparations of both the original double mutant and the mature polypeptide [47]. Thus, with this mutation the precursor remains competent longer and has more time to overcome the defect in the signal peptide. Export competent and incompetent states, almost certainly involving some sort of eonformational change, have also been functionally defined, in vitro, for the O m p A protein [48]. This will be taken up in the chapter on "Chaperonins". Maintenance of precursors in an export competent form, as described above for wild-type pro-MalE, is achieved via interaction with a component (SecB) of the cellular export machinery (see "Chaperonins" below for detailed discussion). However, although in the absence of such factors, the precursors studied thus far, all lose competence for export very rapidly, the signal peptide itself clearly affects the folding kinetics of several periplasmic proteins. It has been shown for both Male and ribose binding protein that following denaturation and dilution from the denaturant, purified precursors refolded at a rate (in seconds) approximately 3-fold slower than the respective mature proteins [49] and similar results have been reported for the precursor of/3-1actamase [50]. Does this reflect an essential interaction between the signal peptide and the mature polypeptide? Randall et al [5, 49] have considered the possibility that such a transient interaction may lead to formation or exposure of some basic structural entity which might then be recognized by a component of the export machinery. In the event that this does occur, it must be a very non-specific interaction as signal peptides can not only apparently export any part of OmpA or PhoE longer than about 50 residues (see "Lower size limit of precursors"), but can also be exchanged between different secretory [16-18], and even some non-secretory, proteins [41, 42] and still mediate export. We would, therefore, not consider this type of interaction as representative of an inherent signal for secretion within the mature polypeptide, but more as a passive feature which a wide range of polypeptides might possess, provided they were free from the type of export-incompatible sequences described below. It should be noted, however, that any essential interaction of the signal peptide and mature poly-

159

peptide, whether it be to temporarily destabilize the precursor or to form some particular basic structure, could explain the observed differences in efficiency of export of deletion mutants [15, 34] or of 2 different hybrid proteins both with the same signal peptide [51]. Role of the structure of the N-terminus

As has already been pointed out [15, 34], the sequence of the N-terminus of the mature polypeptide is likely to have some effect on the efficiency of signal peptide function and thus on export. For optimal activity, the signal must presumably be accessible and free to interact as is decreed by its function(s). In addition, it must be cleaved by leader peptidase [52] (signal peptidase I [53]) to release the mature protein following translocation. The recognition site for leader peptidase cleavage is located at the C-terminus of the signal peptide [4], but there is also evidence to suggest that the first few residues following the cleavage site may contribute to a secondary structure requirement for processing. Deletion of Pro at position 2 or of Pro2 and Ser3 from an O m p A signal peptide-Staphylococcus nuclease hybrid resulted in a progressive decrease in the probability of forming a turn in the area around the cleavage site with a concomitant decrease in the rate of processing [53]. A similar requirement for a turn structure around the cleavage site of lipoprotein has been noted [54]. Whether these effects were directly on peptidase cleavage of the translocated species, on translocation itself, or on both has not as yet been shown. In conclusion to a survey which revealed that the C-termini of bacterial signal peptides are virtually all devoid of charged residues and that the N-termini of the mature proteins, for the most part, possess either no charged residues or acidic residues, yon Heijne [55] proposed that a positive net charge difference between the N- and C-termini of the signal peptide may be essential for export. The strong inhibitory effect of basic amino acids at the N-terminus of the mature polypeptide on transloca:Lion has since been experimentally proven with precursors of alkaline phosphatase [56], fllactamase [17], a chimera of the 2 outer membrane proteins lipoprotein (Lpp) and OmpF [17], and OmpA (MacIntyre et al, in press), following introduction of basic residues after the cleavage site by site-directed mutageneses. Also, in 2 independent, second site, suppressor mutations of a MalE mutant with a defective signal sequence, Lys at position 1 had been converted to an uncharged residue (Thr, Ash) [45]. This phenomenon also explains earlier reports of poor secretion of some hybrid proteins; for example, a /3-1actamase-triosephosphate isomerase hybrid in which the isomerase with two N-terminal basic residues was fused directly to the/3-1actamase signal [41, 57] and a PhoA-Staphylo-

160

S Maclntyre, U Henning

coccus nuclease fusion protein [16] which pf~ssessed an Arg residue between the signal peptide and mature orotein. Although it would thus appear that for successhd export, location of basic residues after the cleavage site has to be avoided, a few exported proteins of E coil do, in fact, possess a posi'tively charged residue at position 1 (MalE [45], PhoA [56] and binding proteins for ribose [58l and sulphate [59]) or close to the N-terminus of the mature protein (OmpA [30] and histidine binding protein [60]). Any inhibitory effect of these residues appears to be at least partly compensated for by a reduction in the net charge due to nearby, apparently precisely located, acidic residues [56] and possibly also by the presence of a more efficient signal peptide (MacIntyre et al, in press), for example one with a hydrophobic core uninterrupted by the helix breaking residues, Pro and Gly [6, 61, 62]. Studies with the Lpp-OmpF chimeric protein [17] indicated that mutants with an inhibitory positive charge after the cleavage site probably reach the membrane but cannot be translocated. The mechanism of inhibition, however, is not yet understood and indeed cannot be understood due to the lack of knowledge of the way in which the signal peptide mediates translocation. Yet, it is hard not to draw parallels between inhibition of translocation by such Nterminal basic residues and the "positive-inside" rule for the topological arrangement of integral membrane proteins which observes that, whereas the cytosolic loops tend to have a higher than average concentration of Arg and Lys, periplasmic loops are relatively devoid of basic residues [63, 64]. Both observations would appear to infer difficulties in translocation of basic re£idue.g cln_gelv f n l ln w i n o hvdrnnhnhlt~ ~anaat=nc,~e

Whether this is a reflection of an interaction of positively charged residues with acidic phospholipids [56], too high an energy threshold for translocation of basic residues against the membrane potential (positive, out and negative, in) [17] or interference with an interaction of the signal peptide with some receptor of the export machinery remains to be seen. Interestingly, and possibly an indication of some difference in export mechanism, the presence of basic residues around the cleavage site is far from rare in eukaryotic secretory proteins [55]. Lipophilic anchors

Another restriction on the primary sequence of mature secretory polypeptides is that they contain no long internal stretch of hydrophobic residues which might interfere with translocation by anchoring the polypeptide in the inner membrane. Anchor sequences are central topogenic components of plasma membrane proteins and function by halting translocation and embedding the polypeptide in the membrane with the

N-terminus, out and the C-terminus, in [65]. They are classically composed of a stretch of 20 or more, mainly hydrophobic, uncharged residues from which Asn, Gin, His and Trp are either excluded or are located at the extremities [4]. Hydrophobicity is the main feature of these sequences; an artificial anchor composed of 4 repeats of Leu-Ala-Leu-Val acts as efficiently as the natural anchor of coliphage fl gene III protein [66]. The assembly pathway of this type of inner membrane protein obviously shares properties with that of exported proteins as insertion of long stretches of hydrophobic residues within normally exported proteins leads to anchorage of the protein in the inner membrane in the same orientation. This has been demonstrated with an outer membrane lipoprotein construct, possessing, in addition to the wild type N-terminal signal, an internalized signal peptide which functioned as an anchor [67], as well as with an O m p A mutant possessing an artificial internal stretch of 16 lipophilic residues [68]. Notably, neither 8 nor 12 internal lipophilic residues were sufficient to anchor O m p A or coliphage gene III protein in the inner membrane. Also, neither an internalized OmpF signal peptide [69] nor an internalized OmpA signal peptide (M Klose, unpublished observations) appeared to be hydrophobic enough to act as classic anchors; neither stopped translocation of the respective protein. In addition to the degree of hydrophobicity, several other features can affect the ability of a hydrophobic domain to function as an anchor. Position within the polypeptide has been shown to affect anchor function in 2 different ways. Firstly, at the lower threshold of hydrophobicity, C-terminal lipophilic domains anchor more efficiently than internal ~/¢111~ LlOII,

IIJJ.

~,,llg;K,U l l l l O t l ~ / , I ~ Y l g ; l l l i D LIII~ IJll~;~l~lll~,lg; OIL O[11 Jt'~"

terminal signal peptide, an internal hydrophobic domain may act as a signal if it is either located close enough to the N-termimts or if it represents a more efficient signal, ie if it functions co-translationally and the N-terminal signal only l~St-translationally [67]. Also, the importance of the charge distribution around signal and anchor sequences has been well defined. Efficient signal function requires N-terminal positive charges [71] and is, in fact, inhibited by a high C-terminal positive charge (see above), while anchors are frequently followed by a series of basic residues which are thought to stabilize interaction of the hydrophobic domain with the membrane [4]. As anchor sequences tend to be more hydrophobic than signals, and the efficiency of a signal sequence increases with hydrophobicity [61, 62], an additional role of the basic residues following anchor sequences may be to prevent these domains initiating translocation of the C-terminus. In accordance with these activities of internal, hydrophobic domains, and in spite of the fact that classic membrane proteins are assembled via one or more membrane-spanning o~ helices [4, 32], no outer

Protein translocation across the E coil plasma membrane

membrane protein sequenced thus far possesses an internal domain hydrophobic enough to function as an anchor [68, 72-74]. Instead, it would appear that integral outer membrane proteins have evolved with the potential for a high level of g-structure to avoid anchorage in one membrane, while retaining the ability to assemble into a second membrane, for example, in the form of an amphipathic g-barrel as has been suggested for OmpA [32].

Lower size limit of precursors No upper size limit has yet been defined for a secretory protein. A clear cut lower size limit, however, has been described for translocation of fragments of OmpA across the bacterial cytoplasmic membrane [75], and of prelysozyme [76] and preprolactin [77] into canine microsomes. The precursor to OmpA contains 346 residues. Starting with an internally deleted gene (ompA A 1-229) encoding a 116 residue precursor, ompA fragments encoding precursors of 123, 88, 72 and 68 residues were constructed. Whereas, the latter 2 remained unprocessed and totally cytosolie, the 123, 116 and 88 residue constructs were processed and localized to the periplasm. Similar results were obtained in the eukaryotic system. Prelysozyme fragments of 102 and 74 residues and preprolactin fragments of 132 and 87 residues, but neither a 51 residue prelysozyme fragment nor a 55 residue preprolac'~n fragment were translocated in vitro into microsomes. The explanation given [75, 78] for this observation is that, in the eukaryotic system, binding of signal recognition particle (SRP) to the signal peptide, an essential step prior t 0 traneln~aflan

n~nre;

i~ r~nh: a e

long ~e t h e

n2e~ent

chain is still ribosome bound [77, 79] and ii) optimally with nascent chains of between about 70 and 140 residues [77], ie as the ribosome covers about 40 residues [80, 81], shortly after the signal emerges from the ribosome. Thus, with precursors smaller than about 70 residues, no SRP-signal interaction, and thus no import into microsomes, can occur as translocation is completed and the ribosome complex dissociated prior to exposure of the signal. It is possible that a similar early requirement for interaction of signal peptides with a component of the export machinery also exists in bacteria. Indeed the results with OmpA would appear to support this. No functional bacterial analog of any of the SRP polypeptides [82] has yet been conclusively identified, although several potential candidates exist including SecB [83], trigger factor [84], (SecA?) and an uncharacterized gene product (termed 48 kD protein [85] or FFH [86]) which has recently been shown to bear extensive sequence homology to the 54 kD SRP component. Although it may appear so, there is no contradiction between the above observations and the secretion of

161

polypeptides with very small precursors such as prepromellitin (70 residues) [87] and frog prepropeptide GLa (64 residues) [88]. Detailed studies with prepromellitin [78, 87, 89] indicate that these polypeptides are imported into microsomes by a different or at least divergent pathway which is SRP/docking protein independent and which imposes constraints on the primary structure of the mature polypeptide as well as on the signal sequence. In prepromellitin, a cluster of negatively charged residues within the pro-region must be counterbalanced by a group of C-terminal basic residues. Extension of the mature portion of prepromellitin by fusion to DHFR rendered the system SRP-docking protein dependent but independent of the charge distribution within promellitin. Prepromellitin and one of the prepromellitin-DHFR hybrids behaved in an analogous way in E coil in that export of the hybrid, but not of prepromellitin itself, was dependent on the cellular SecA / SecY machinery [89]. Export of prepromellitin and the prepromeUitin-DHFR hybrid, therefore, bears similarities to the insertion of M13 coat protein and a coat protein-OmpA hybrid into the inner membrane [90-92]. Together they appear to be indicative of a certain pattern, ie that some small polypeptides can bypass the central export machinery (SecA / SeeY) but are strongly dependent on the structure of the mature polypeptide, while longer polypeptides lacking such structural information (either exported proteins or long periplasmic segments of inner membrane proteins [931) require the SecA/SecY pathway.

Low efficiency of translocation of non-secretory proteins Finally, there is another type of export incompatibility which is probably characterized by the classic "lethal jammer" phenotype of secretory protein-B-galactosidase fusions [29, 94] possessing N-terminal fragments of the precursors of the secretory polypeptides. Production of such hybrids results in toxicity together with an accumulation of precursors of a number of other exported proteins. Also, the presence of a mature membrane-spanning species of the hybid (secretory fragment, out; B-galactosidase, in) has been directly demonstrated by accessibility to trypsin located in the periplasm [95-97]. Thus, the "jammer" phenotype has been attributed to saturation of the export machinery as export is initiated by the secretory protein but halts due to an inability to translocate the B-galactosidase moiety. That the export incompatibility of/3-galactosidase is due to some inhibitory property of the polypeptide itself rather than to the absence of an essential sequence (see above, "Export signals within the mature sequence?") is substantiated by the following 2 facts. Firstly, it is now clear that most of B-galacto-

162

S Maclntyre, U Henning

sidase can be transiocated, albeit very inefficiently, across the bacterial plasma membrane. Successful export of an OmpA-/3-galactosidase hybrid has been achieved at low levels of syntl'.esis, whereas, the hybrid precursor accumulated in the cytosol if the level of synthesis was increased [95]. Also, using alkaline phosphatase activity as a marker, :,t was estimated that 9% of a tripartite fusion protein, composed of the Nterminal 211 residues of pro-MalE followed by 90% of the /3-galactosidase 0olypeptide and ending in an almost complete alkadne phosphatase, was completely translocated across the plasma membrane [98]. Secondly, using the same tripartite hybrid system, it was shown that although fragments of/3-galactosidase still interfered with export, a higher fraction of the total hybrid population could be translocated across the membrane [98]. An even more significant conclusion from this latter study is that structural features within/3-galactosidase, which inhibit export, are not confined to one site but are distributed throughout the polypeptide. In view of the apparent conformational constraints on exported proteins [43], a likely explanation for the difficulty in translocation of /3-galactosidase is rapid folding of domains into stable tertiary structures. Interestingly, at least 2 other cytosolic proteins (mouse DHFR and the Klenow fragment of E coil DNA polymerase I) have proved equally difficult to export. Only 7% of a MalE-Klenow fragment hybrid was recovered in the periplasm [99], while OmpA-DHFR fusions were exported when synthesized at low levels but caused "jamming" at higher levels of synthesis [95]. Also, prepromellitin-DHFR hybrids, expressed in E coil, had a processing half life of about 45 min [89]. An ~nr~21inn

Lmplio2ti,~n ~¢ thic " that r~niA f•lAin.

into a stable conformation may be a necessity for retaining cytosolic proteius within the cytoplasm. A remarkably high percent of random, chromosomallyderived sequences, coding for stretches of hydrophobic amino acids, are capable of functioning as N-terminal signal sequences [100]. In addition, internal hydrophobic sequences can function as signals. This has been demonstrated both with such a polypeptide lacking an N-terminal signal peptide [68] and with one possessing a less efficient N-terminal signal, ie one of the posttranslational mode [67]. Remarkably, in the first case the internal hydrophobic domain functioned with close to 100% efficiency as a signal even though it was located in the middle of the protein (165 residues from the Nterminus). However, internal signal activity appears to be inhibited by extensive folding of the N-terminal region, at least in the eukaryotic system [101, 102]. Thus, it is possible that rapid folding of cytosolic proteins is essential to mask internal hydrophobic sequences and to prevent them from functioning as signals. If this is true, there may then be only certain classes of cytosolic proteins which can be efficiently

exported. These may include polypeptides which, by themselves, cannot fold into their final conformation a n d / o r which maintain an intermediate conformation by interaction with a chaperonin (see below). An example of this may be the T4 phage tail 13bre fragment which can be very efficiently exported w|~en fused to OmpA [421.

Release from the plasma membrane Release from the outer surface of the plasma membrane represents the final step in translocation. With the exception of a couple of mutants with defective signal sequences [103, 10t], presence of the signal peptide maintains the precursor in a translocated but membrane bound form [52]. Thus, cleavage of the signal is an a priori step to release from the membrane. Following cleavage, release is accompanied by a change in conformation of the mature polypeptide, at least of several periplasmic proteins. Deletion mutants of/3-1actamase [40], MalE [35, 36, 105] and GIpQ [39, 106] which are translocated and processed but which remain membrane-bound have been described. Unlike the wild-type proteins, the MalE and fl-laetamase mutants remain protease-sensitive. That a similar conformational change occurs in the normal export pathway of /3-1actamase was demonstrated by identification of a protease-sensitive membrane-bound form of the wild type protein in experiments performed at 15oC [107]. /~-lactamase possesses an internal disulphide bond. This is required for stability, but is not essential for either release or correct folding of the enzyme [50, 108]. Release of/3-1actamase was blocked, however, in mutants in which either of the Cys residues involved in the disulphide bond had been substituted with a Tyr residue [109]. It was proposed that release was prevented, as the polypeptide could not fold into the correct conformation due to the presence of the bulky Tyr residue. It is unknown, however, why and how such mutants remain bound to the plasma membrane.

Chaperonins Chaperonins were originally defined as proteins which promote correct assembly of oligomeric proteins [110, 111]. The term has now also been used for proteins which stabilize precursors of secretory polypeptides in an export-compatible form [112]. In E coil, 3 such proteins have been studied, SecB, trigger factor and GroEL. What represents their target? Using a pioneering selection procedure [113] which yielded mutants in genes coding for components of the export machinery (eg [114]), the secB gene was discovered [115, 116]. SecB is an oligomeric [117, 118],

Protein translocation across the E coil plasma membrane

soluble cytosolic [37, 83] protein consisting of rather acidic 155 residue, subunits [i19]. In-depth studies have been performed with pro-MalE [37, 38, 83, 117, 118, 120, 121,122], the folding behavior of which has been summarized above. From both in rive and in vi~o studies it has become clear that SecB retards folding of this precursor into an export-incompatible form; todate, however, there is no consensus concerning the recognition site(s) for SecB. A series of C-, N-terminally or internally deleted Male proteins, which were all, in addition, exportdefective due to the absence of a signal peptide or to a mutation therein, were expressed from a multicopy plasmid. One set interfered with the export of other secretory proteins, another did not. Interference was apparently due to titration of SecB which could be overcome by increasing secB gene dosage [37]. The deletions defined an area between residues 150 and 186 of the 370 residues [123] protein as a region required for SecB binding. However, the signal peptide proved not to be without influence on the action of SecB. Rather surprisingly, small deletions within the hydrophobic core rendered the precursor competent for cotranslational export in secB-null mutants [122] (such mutants are able to grow in minimal medium, [116]), while in such a strain, export of the wild type precursor was completely post-translational [120j. A different approach, using MalE-PhoA fusion proteins, located 1 hind,_'ng region within the first third of the polypeptide and, possibly, a ~econd one in a MalE fragment up to residue 230 [1211 . Finally, using an in vitro system, it was concluded that SecB binds to the sign.:l peptide of pro-MalE [83]. Based on the interference phenomenon or on results obtained with secB-nuii mutants, it was concluded that efficient export of the precm'sors of MalE, OmpA, LamB, O m p F and the periplasmie oligopeptide binding protein requires SecB, while that of the precursors of PhoA, ribose binding protein, TEM/3-1actamase and the Braun lipoprotein does not [37, 116, 124, 125]. However, a protein, consisting of the signal peptide plus 11 N-terminal residues of OmpF fused to residue 8 of the lipoprotein did require SecB for optimal export, altheugh the export defect observed in a secB-null background was much less severe than that of the complete pro-OmpF [124]. It is clear from these data that, although SeeB is a well characterized component of the export machinery, it remains unknown i~s to what it recognizes on a target protein and how it functions mechanistically. The interference data suggest that only a relatively short stretch of amino acid residues suffice to bind SecB. Since these results, however, are somewhat at variance with those obtained with the MalE-PhoA fusion proteins and completely at variance with those obtained with a n / n vitro system [83], and since so far there has been no mention of a consensus sequence present in SecB-

163

dependent proteins, binding of SecB probably involves more than amino acid sequence specificity (see below). Concerning its mode of action, it is often more or less tacitly assumed that SecB keeps a target polypeptide in an "unfolded state" (implying a structure similar to that of a polypeptide denatured by a chaotropic agent). There is no evidence for this. Also, because protein folding represents an exceedingly fast process, we feel that maintenance of a completely unfolded polypeptide is rather problematic to visualize. It would appear more likely that, in a precursor-SecB complex, the precursor possesses a defined, at a minimum, secondary structure (it should be noted that, as processing of pro-MalE occurs partially co- and partially post-translationally [12@ in such a complex the precursor need not be complete). Does the area(s) of MalE, which binds SecB, play an active role in the export process, that is to say, are there export signals within Male [121,127], and consequently, within the other proteins exhibiting SeeB-modulated export? Almost certainly not in the sense defined above (see first section) because export of none of the proteins affected in secB-null mutants is blocked completely. In this context, a comparison of the rates of export of a pro-MalE, missing residues 150 through 186, with the wild type protein, both in a secB and a secB + background, would be interesting and telling. The question remains as to whether SecB induces the formation of an export-compatible conformation or only maintains the polypeptide longer as a folding intermediate which it transiently assumes even in the absence of SecB. Since pro-OmpA can fold spontaneously into a form competent for export (see below), the latter mode of action appears more likely. Trigger factor was discovered [128] as a soluble protein which stabilized an export-competent form of pro-OmpA synthesized in vitro. Purified pro-OmpA could not be translocated into membrane vesicles in an in vitro system. It could be made translocationcompetent by denaturing it in 8 M urea, followed by dilution into buffer containing 0.8 M urea. Although remaining soluble, this form of pro-OmpA rapidly lost competence unless trigger factor was present [48]. Trigger factor is a monomeric protein (Mr 63 000) [48] which forms a i:1 complex with pro-OmpA and which can also be isolated from ribosomes where it associates with the 50S subunits known to contain the exit site for nascent polypeptides [129]. It has been shown that trigger factor and SecB (and notably SecB more efficiently) will bind not only to pro-OmpA but also to O m p A in vitro, but that neither, under the same conditions of denaturation-renaturation, will bind to several non-secretory proteins [112]. This indicates firstly, that the signal peptide is not a target, at least not the only target, for the 2 "chaperonins" and, secondly, that the 2 factors may be specific for exported

164

S Maclntyre,U Henning

proteins. It seems puzzling, however, that pro-OmpA should interact with SecB or trigger factor. An answer to the question of the true physiological role of trigger factor as well as a definition of target specificities awaits the isolation of mutants with a defective gene for this factor. Finally, another protein capable of associating with proteins which have not yet assumed their final conformarion should be mentioned briefly, although it appears somewhat doubtful as to whether it is designed for a role fike SecB or trigger factor. The cytosolic GroEL protein represents a member of the classical chaperonins [110]; it belongs to the family of heat shock proteins and exhibits ATPase activity (such an activity has not been reported for SecB or trigger factor). It is known that GroEL (product of the mopA gene: morphogenesis of phages) is required for the assembly of many phages [130]. It can also form a complex with newly synthesized chloramphenicol acetyltransferase [131]. It will, however, also chaperone secretory proteins prior to the translocation step. Thus, it could associate with a newly synthesized pro-/3lactamase which had not yet undergone-• oxidation to form the intrachain disulfide bond [132], dissociation of the complex required ATP hydrolysis [131] and the GroEL-bound precursor in the complex remained competent for membrane translocation in an in vitro system [131]. Similarly, purified GroEL was shown to stabilize pro-OmpA for membrane translocation and, like SecB or trigger factor, to form a complex with OmpA [112]. Thus, this protein can recognize a number of very different proteins which have not yet arrived at their final conformation. This may be a more general property of chaperonins; as far as we know it has not yet been shown if SecB is entirely specific for secretory proteins. Taking all tile data into account, the mechanism of recognition of target proteins by the chaperouins, a very interesting question indeed, has remained obscure m a strange way; it appears that different systems have yielded different answers. A rather surprising result that may give a clue as to why this has happened has recently been reported for BiP [133], a eukaryotic polypeptide belonging to the family of the heat shock proteins (bu'~ expressed constitutively), which is present in the lumen of the endoplasmatic reticulum and, among other activities, is involved in the assembly of immunoglobulins [134, 135]. Like GroEL, BiP possesses ATPase activity. It ¢,as found that BiP could bind short hydrophilic (8-25 residues) peptides, and that binding caused hydrolysis of ATP which was followed by release of the peptide. Of 9 peptides tested, one was inactive and the others bound with rather different Km values (between 12 and 900/zM). Remarkably, the basis of the peptide specificity could not be deduced from the amino acid sequences of these

peptides. If target recognition by the E coli chaperonins is similar to this, it would perhaps be understandable why different approaches to locate their binding sites gave different answers. Folding of MalE-PhoA fusion proteins may well differ from that of the internally deleted MalE proteins. The results leading to the conclusion that SecB binds to the signal peptide of proMalE were obtained with yet another ligand, a Cterminally truncated MalE ([38, 83, 118]; from the 396 residue pro-MalE, 347 were present, followed by 7 non-MalE residues) which might possess yet another conformation. In each case, then, a recognition site, and quite possibly more than one, of the type observed with the small peptides could be buried or exposed in a framework of conformational differences too subtle to detect with presently available methods. Competition experiments using peptides derived from secretory proteins might provide an answer. In any event, since complex formation between OmpA and SecB or trigger factor or GroEL certainly does not appear to be an artefact [112], binding site(s) for the chaperonins must reside in the mature parts of exported proteins. This view would also be consistent with the fact that a pro-OmpF-lipoprotein fusion polypeptide was much less dependent, for export, on SecB than the complete pro-OmpF (see above).

Feedback inhibition of synthesis of exported proteins It has long been known that regulatory circuits exist which allow the cell to sense the occupancy of the outer membrane by protein. For example, in cells carrying 2 copies of the ompA gene no gene dosage effect was observed [136]; induction of synthesis of the proteins involved in iron uptake (notably the outer membrane protein FeuB) caused a reduction of synthesis of the OmpF protein [137]. Also, we routinely observed that mutants missing OmpA possess more OmpC plus OmpF than the parent, and vice versa. The same situation was reported for the pair Lamb and OmpC [138]. More recently, Click et al [139] observed that, when the ompC gene (present on a multicopy plasmid and under the control of the tac promoter) was induced, the resulting overexpression of OmpC was accompanied by a rapid and nearly complete cessation of synthesis of OmpA and LamB, while production of the periplasmic MalE protein remained unaffected. There was no accamulation of precursors of any of these outer membrane proteins, thus their translocation across the plasma membrane was not inhibited. The authors found that the mechanism of inhibition operated at the level of translation; the level of ompA mRNA was far from sufficiently reduced to explain the disappearance of de novo synthesis of OmpA, nor was the expression

Protein translocation across the E coli plasma membrane

of the LamB-/J-galactosidase fusion protein reduced (first 2 residues of the LamB signal peptide followed by most of the glycosidase). A phenotypically similar situation was encountered with an OmpC mutant, OmpCtd, missing Val-300 and Gly-301 of the 346 residue [140] protein. Expression of the corresponding gene caused a decrease in synthesis of the wild type outer membrane proteins OmpC, OmpA, OmpF and Lc without any precursor accumulation [141]. Since there was little effect of synthesis of OmpCtd on the expression of an ompC-lacZ protein or operon fusion gene, inhibition was again assumed not to act at the transcriptional level. Interruption of the area encoding the hydrophobic core of the pro-ompCtd gene by linker mutagenesis, prevented export of the corresponding precursor and led to a relief of the OmpCtd phenotype, ie the OmpCtd induced inhibition was subsequent to the block imposed by the defective signal. It was not shown whether the inhibitory action of OmpC overproduction also disappeared when entry of the precursor into the export pathway was hindered, but this was almost certainly so. Overproduction of OmpA had an effect mirroring that of OmpC, ie it caused a much reduced level of synthesis of OmpC and OmpF, and in this case, interruption of the signal peptide abolished the inhibition although masswe amounts of the protein accumulated (R Freudl, U Henning, unpublished). The target protein need not enter the export pathway to be subject to the translational control exerted by OmpC over-production. The effect of this over-production has been studied using several proLamB proteins with mutant signal peptides [142]. One of these missed most of the hydrophobic core, in another the 6th residue of the 25 residue signal was fused to residue 148 of the mature, 421 residue [143] polypeptide. Hence, the 2 mutant precursors could not be channelled into the export pathway and yet, their synthesis was still inhibited by ompC over-expression. Remarkably, the Lamb fragments or their mRNAs •were apparently already recognized in the cytosol as belonging to or coding for, respectively, an outer membrane protein. Somewhat earlier Hengge-Aronis and Boos [39,106] described a very similar phenomenon. A C-terminally truncated periplasmic protein, GIpQ', derived from the phosphodiesterase GIpQ, was translocated across the plasma membrane but not released into the periplasm. Concomitantly, synthesis of other periplasmic proteins, including MalE, but not of outer membrane proteins, was inhibited although none of the MalE precursor accumulated. As with OmpCtd, the inhibitory effect apparently acted at the level of translation and was abolished in the presence of a second mutation (unidentified) within GIpQ' which led to cytosolie products. Expression of glpQ' in cells producing a pro-

165

MalE with a defective signal peptide also inhibited synthesis of the mutant pro-MalE; the authors suggested that this indicated that this mutant precursor abortively entered the export pathway; however, in view of the results obtained with the signal sequence mutants of LamB (see above) this may not be so. Finally, they demonstrated that the translational inhibition caused by GlpQ' was dominant over the accumulation of proMalE following growth, at the non-permissive temperature, of cells carrying a temperature sensitive secA allele; ie, GlpQ' acted before or at the level of SecA recognition. A detailed discussion of the possible mechanisms underlying the regulatory controls exerted by GlpQ', OmpCtd or overexpression of OmpC is outside the scope of this review and the reader is referred to the relevant publications. Taken together and assuming that the mechanisms yielding all these results are basically the same, they suggest a most remarkable possibility, namely, that secretory proteins are recognized as belonging to the outer membrane or the periplasm even before translocation begins. It is not known what the primary target is for this recognition, the protein, the mRNA or both. Other cases of an apparent coupling of translation to export (which is by no means obligatory), involving mutant signal peptides or the synthesis of a mutant (ts) SecA protein have been described [144, 145, 146, 147]. It is not known if they are mechanistically related to those summarized above. If so, the following case might point to the protein as the target for inhibition of synthesis. In a strain carrying a secA amber mutant which is suppressible at 30oC but not at 42°C, growth at the non-permissive temperature caused accumulation of pro-MalE and a much reduced rate of synthesis of this precursor. Introduction of a secB-nuil mutation eliminated this inhibition of synthesis [143]. Perhaps, therefore, the complex SecB-prc'-MalE or the pro-MalE conformation held by SecB is required for the expression of inhibition. It would be interesting to see how the levels of synthesis of the partially deleted MalE proteins which bind and do not bind SecB [37] compare in a secA background. Whatever the mechanisms are leading to the types of inhibition described here, it is clear that secretory proteins with defects in their mature parts can cause translational inhibition of other secretory proteins in a class-specific way. Finally, it is rather intriguing that the expression of glpQ' or ompCtd inhibits the synthesis of several but not all periplasmie or outer membrane proteins, respectively. Also, periplasmic and outer membrane proteins appear to exist which are exported via a SecAindependent pathway [114, 148]. It remains to be seen whether the latter and the former polypeptides are the same, how they are secreted and how their synthesis is regulated.

166

S Maclntyre, U Henning

I Randall LL, Hardy SIS, Thorn JR (1987) Ann Rev Microbiol 41,507-541 2 Benson SA, Hall MN, Silhavy TJ (1985) Ann Rev Biochem 54, 101-134 3 Pollitt S, inouye M (1987) In: Bacterial Outer Membranes as Model Systems (Inouye M, ed) Wiley, New York, 117-140 4 voa Heijne G (1985) Curr Top Membr and Transp 24, 151-179 5 Randall LL, Hardy SIS (1989) Science 243, 1156-1159 6 Gierash LM (1989) Biochemistry 28, 923-930 7 Saier MH, Wemer PK, Mfiller M (1989) Microbiol Rev 53, 333-366 8 Milstein C, Brownlee GG, Harrison TM, Mathews MB (1972) Nature New Bio1239, 117-120 9 Blobel G, Dobberstein B (1975) J Cell Bio167, 835-851 10 Wickner W (1979) Ann Rev Biochem 48, 23-45 11 Lingappa VR, Chaide J, ¥ost CS, Hedgpeth S (1984) Proc Nail Acad Sci USA 81,456-460 12 Moreno F, Fowler AV, Hall M, Silhavy TJ, Zabin I, Schwartz M (1980) Nature (London) 286, 356-359 13 Bankiatis VA, Rasmussen BA, Bassford PJ (1984) Cell 37, 243-252 14 Freudl R, Henning U (1988) J Mol Bio1203,517-519 15 Ferenci T, Silhavy TJ (1987) J Bacterio1169, 5339-5342 16 Liss LR, Johnson BL, Oliver DB (1985) J Bacteriol 164, 925-928 17 Y amane K, b/fizuslfima S (1988) J Biol Chem 263,19690-19696 18 Takahara M, Hibler DW, Barr PJ, Gedt JA, Inouye M (1985) J Biol Chem 260, 2670-2674 19 Holland B, Wang R, S~ror SI, Blight M (1989) In: Society for General Microbiology Syrup 44 (Banmberg S, Hunter I, Rhodes M~ eds) Cambridge University Press, 219-254 20 Kuwaj.:.ma G, Kawagishi I, Homma M, Asaka Jl, Kondo E, Macnab RM (1989) Proc Nall Acad Sci USA 86, 4953-4957 21 Pohlner J, Halter R, Beyreuther K, Meyer TF (1987) Nature (London) 325,458-462 21a Pugsley AlP, (1988) In: Protein Transfer and Organclle Biogenesis (Das RA, Robbins PW, eds) Academic Press, San

38 Watanabe M, Blobel G (1989) Proc Nail Acad Sci USA 86, 2248-2252 39 Hengge R, Boos W (1985) J Bacteriol 162,972-978 40 Koshland D, Botstein D (1982) Cell 30, 893-902 41 Pliickthun A, Knowles JR (1987) J Biol Chem 262, 3951-3957 42 MacIntyre S, Freudl R, Degen M, Hindennach I, Henning U (1987) J Biol Chem 262, 8416-8422 43 Randall LL, Hardy JS (1986) Cell 46, 921.-928 44 Eilers M, Schatz G (1986) Nature (London) 322,228-232 45 Ryan JP, Duncan MC, Bankiatis VA, Bassfor~ PJ (1986) J Biol Chem 261, 3389-3395 46 Cover WH, Ryan JP, Bassford PJ, Walsh I,'~,, Bollinger J, Randall LL (1987) J Bacteriol 169, 1794-1800 47 Liu G, Topping TB, Cover WH, Randall LL (1988) J Biol Chem 263, 14790-14793 48 Crooke E, Brundage L, Rice M, Wickner W (1988) EMBO J 7, 1831-1835 49 Park S, Liu G, Topping TB, Cover WH, Randall LL (1988) Science 239, 1033-1035 50 Laminet AA, Pliickthun A (1989) EMBO J 8, 1469-1477 51 Lehnhardt S, Pollitt S, Inouye M (1987) J Biol Chem 262, 1716-1719 52 Dalbey RE, Wickner W (1985) J Biol Chem 260, 15925-15931 53 Duffaud G, Inouye M (1988) J Biol Chem 263, 10224-10228 54 Inouye S, Duffaud G, Inouye M (1986) J Biol Chem 261, 10970-10975 55 yon Heijne G (1986) J Mol Biol 192, 287-290 56 Li P, Beckwith J, Inouye H (1988) Proc Nail Acad Sci USA 85, 7685-7689 57 Kadonaga JT, Gautier AE, Straus DR, Charles MDE, Knowies JR (1984) J Biol Chem 259, 2149-2154 58 Groarke JM, Mahoney WC, Hope JN, Furlong CE, Robb FT, Zalkin H, Hermodson MA (1983) J Biol Chem 258, 12952-12956 59 Hellinga HW, Evans PR (1985) EurJ Biochem 149, 363-376 60 Higgins CF, Ames GFL (1981) Proc Nati Aced Sci USA 78, 6038-6042 61 Kendall DA, Bock SC, Kaiser El' (1986) Nature 321,706-708 62 Freudl R, MacIntyre S, DegenM, Henning U (1988) J Biol

22 Wickner W (1988) Biochemistry 27, 1081-1086 23 Freudl R, Schwarz H, loose M, Mowa NR, Henning U (1985) EMBO J 4, 3593-3598 24 Freudl R, Schwarz H, Degen M, Henning U (1987) J Bacterioi 169, 66-71 25 loose M, Schwarz H, Madntyre S, Freudl R, Eschbach ML, Henning U (1988) J Biol Chem 263, 13291-13296 26 Bosch D, Leunissen J, Verbakel J, de Jong M, van Erp H, Tommassen J (1986) J Mol Bio1189, 449-455 27 Bosch D, Voohnut W, Tommassen J (1988) J Biol Chem 263, 9952-9957 28 Bosch D, Scholten M, Verhagen C, Tommassen J (1989) Mot Gen Genet 216, 144-148 29 Tommassen J, Leunissen J, van Damme Jongsten M, Overduin P (1985) EMBO J 4, 1041-1047 30 Chen R, Schmidmayr W, Kramer C, Chen-Schmeisser U, Henning U (19~) Proc Nail Acad Sci USA 77, 4592-4596 31 Morona R, lOose M, Henning U (1984) J Bacteriol 159, 570-578 32 J~mig F (1989) in: Prediction of Protein Str~cture and the Principles of Pro-ein Conformation (Fasman GD, ed) Plenum, New York, 707-717 33 Benson SA, Bremer E, Silhavy TJ (1984) Proc Nail Acad Sci USA 81, 3830-3834 34 Rasmussen BA, Silhavy TJ (1987) Genes Dev 1,185-196 35 Ito K, Beckwith JR (1981) Cell25, 143-150 36 Duplay P, Szraelcman S, Bedouelle H, Hofnung M (1987) J Mol Bio1194, 6{~-673 37 Collier DN, Bankaitis VA, Weiss JB, Bassford PJ (1988) Cell 53, 273-283

63 yon Heijne G (1986) EMBO J 5, 3021-3027 yon Heijne G (1989) Nature (London) 341. 456-458 65 Blobel G (1980) Proc Nail Acad Sci USA 77, 1496-1500 66 Davis NG, Model P (1985) Cell41, 607-614 67 Coleman J, Inukal M, Inouye M (1985) Cell 43, 351-360 68 Maelntyre S, Freudl R, Eschbach ML, Hen,ning U (1988) J Biol Chem 263, 19053-19059 69 Shinkay A, Yamada H, Mizuno T, Mizushima S (1989) J BIOchem 106, 323-330 70 Paterson RG, Lamb RA (1987) Cell 48, 44! -452 71 Vlasuk GP, Inouye S, lto H, Itakura K, Inouye M (1983) J Biol Chem 258, 7141-7148 72 Homma H, Kobayashi T, Chiba N, Karasawa K, Mizushima H, Kudo I, Inoue K, lkeda H, Sekiguchi M, Nojima S (1984)JBiochem 96, 1655-1664 73 Krone WJA, Stegehnis F, Koningstein G, vail Doom C, Roosendaal B, de Graf FK, Oudega B (1985) FEMS Micro Letts 26, 153-161 74 Heller K, Kadner RJ (1985) J Bacteriol 161,904-908 75 Freudl R, Schwarz H, Degen M, Henning U (1989) J Mol Biol 205,771-775 76 Ibrahimi IM, Cutler D, Stueber D, Bujard H (1986) EurJBiochem 155,571-576 77 Siegel V, Walter P (1988) EMBO J 7, 17~9-1775 78 Mtiller G, Zimmermann R (1987) E M B O J 6, 2099-2107 79 Wiedmann M, Kurzchalia TV, Bielka H, Rapoport TA (1987) J Cell Biol 104, 201-208 80 Blobel G, Sabatini DD (1970) Y Cell Bio145, 130-145 81 Bernabeu C, Lake JA (1982) Proc Nail Acad Sci USA 79, 3111-3115

References

t'/,*~ '3~.~t ,L.~lac~*vJ ~ . u J ,

"~L4A__'IAO .r.v--r .z,-r7

Protein translocation across the E coli plasma membrane 82 Walter P, Lingappa VR (1986) Ann Rev Cell Biol 2,499-516 83 Watanabe M, Blobel G (1989) Cell 58, 695-705 84 Crooke E, Guthrie B, Lecker S, Lill R, Wickner W (1988) Cell 54, 1003-1011 85 R6misch K, Webb J, Herz J, Prehn S, Frank R, Vingron M, Dobberstein B (1989) Nature (London) 340, 478-482 86 Bernstein HD, Poritz MA, Strub K, Hoben PJ, Brenner S, Walter P (1989) Nature (London) 340, 482-486 87 Zimmermann R, Mollay C (1986) J Biol Chem 261, 12889-12895 88 Schlenstadt G, Zimmcrmann R (1987) EMBO J 6, 699-703 89 Cobet WNE, MoUay C, Mfiller G, Zimmermann R (1989) J Biol Chem 264, 10169-10176 90 Kutm A, Kreil G, Wickner W (1986) EMBO J 5, 3681-3685 91 Kuhn A, Wickner W, Kreil G (1986) Nature 322, 335-339 92 Kuhn A (1988) Eur J Biochem 177, 267-271 93 Gebert JF, Overhoff B, Manson MD, Boos W (1988) J Biol Chem 263, 16652-16660 94 Bassford PJ, Silhavy TJ, Beckwith JR (1979) J Bacteriol 139, 19-31 95 Freudl R, Schwarz H, Kramps S, Hindennach I, Henning U (1988) J Biol Chem 263, 17084-17091 96 Rasmussen BA, Bankiatis VA, Bassford PJ (1984) J Bacteriol 160, 612-617 97 Tommassen J, de Kroon T (1987) FEBS Lett 221,226-230 98 Lee C, Li P, Inouye H, Brickman ER, Beckwith J (1989) JBacterio1171, 4609-4616 99 Bedouelle H, Duplay P (1988) EurJ Biochem 171,541-549 100 Kaiser CA, Preuss D, Grisafi P, Botstein D (1987) Science 235, 312-317 101 Spiess M, Lodish I ~ (1986) Cell 44, 177-185 102 Finidori J, Rizzolo L, Gonzalez A, Kreibich G, Adesnik M, Sabatini DD (1987) J Cell Biol 104, 1705-1714 103 Fikes JD, Bassford PJ (1987) J Bacterio1169, 2352-2359 104 Kadonaga JT, Plfickthun A, Knowles JR (1985) J Biol Chem 260, 16192-16199 105 Duplay P, Hofnung M (1988) J Bacterio! 170, 4445-4450 106 Hengge-Aronis R, Boss W (1986) J Bacteriol 167, 462-466 107 Minsky A, Summers RG, Knowles JR (1986) Proc Natl Acad Sci USA 83~ 4180-4184 108 Schultz SC, Dalbadie-McFurland G, Neitzei J J, Richards JH (1987) Proteins 2, 290-297 109 FiRs R, Reuveny Z, van Amsterdam J, Mulholland J, Botstein D (1987) Proc Natl Acad Sci USA 84, 8540-8543 110 Hemmingsen SM, Woolford C, van der Vies SM, Tilly K, Denis DT, Georgopoulos CP, Hendrix RW, Ellis RJ (1988) Nature (London) 333, 330-334 111 Oestermann J, Horwich AL, Neupert W, Hartl FU (1989) Nature (London) 341,125-130 112 Lecker S, Lill R, Ziegelhoffer T, Georgopoulos C, Bassford PJ, Kumamoto C, Wickner W (1989) EMBO J 8, 2703-2709 113 Oliver D, Kumamoto C, Quinlan M, Beckwith J (1982) Ann Microbiol (Paris) 133, 105-110 114 Oliver DB, Beckwith J (1981) Cell 25,765-772

167

115 Kumamoto CA, Beckwith J (1983) J Bacteriol 154,253-260 116 Kumamoto CA, Beckwith J (1985) J Bacteriol 163, 267-274 117 Weiss JB, Ray PH, Bassford PJ (1988) Proc Natl Acad Sci USA 85, 8978-8982 118 Watanabe M, BIobel G (1989) Proc Natl Acad Sci USA 86, 2728-2732 119 Kumamoto CA, Nault AK (1989) Gene 75, 167-175 120 Kumamoto CA, Gannon PM (1988) J Biol Chem 263, 11554-11558 121 Gannon PM, Li P, Kumamoto CA (1989) J Bacteriol 171, 813-818 122 Collier DN, Bassford PJ (1989) J Bacteriol 171, 4640-4647 123 Duplay P, Beduelle H, Fowler A, Zabin l, Saurin W, Hofnung M (1984) J Biol Chem 259, 10606-10613 124 Watanabe T, Hayashi S, Wu HC (1988) J Bacteriol 170, 4001-4007 125 Bankaitis VA, Bassford PJ (1984) J Bioi (.hem 259, 12193-12200 126 Josefsson LG, Randall LL (1981) Cell 25,151-157 127 Bankaitis VA, Ryan JP, Rasmussen BA, Bassford PJ (1985) Curr Top Membr Transp 24, 105-150 128 Crooke E, Wickner W (1987) Proc Natl Acad Sci USA 84, 5216-5220 129 Lill R, Crooke E. Guthrie B, Wickner W (1988) Cell 54. 1013-1018 130 Neidhardt FC. Van Bogelen RA, Vaughn V (1984) Ann Rev Genet 18, 295-329 131 Bochkareva ES, Lissin MN. Girshovieh AS (1988) Nature (London) 336, 254-257 132 Pollit S, Zalkin H (1983) J Bacterio1153, 27-32 133 Flynn GC, Chappell TG, Rothman JE (1989) Science 245, 385-390 134 Hendershot LM, Bole DG, Kohler G. Kearney JF (1987) J Cell Biol 104, 761-767 135 Bole D(3, Hendershot LM. Kearney JF (1986) J Cell Bio1102. 1558- 1566 136 Datta DB, Krfime~ C, Helming U (1976) J Bacteeiol 128. 834-841 137 Boyd A, Holland IB (1979) Cell 18, 287-296 !38 l~ieArit-hn C Fraliek ]A tIQR2~I/ kartpr/al 14q_ 156-160 139 Click EM, McDonald GA, Schnaitman CA (1988) J Bacteriol 170, 2005-2011 140 Mizuno T, Chou MY, Inouye M (1983) J Biol Chem 258, 6932-6940 141 Catron KM, Schnaitman CA (1987) J Bacterio1169 , 4327-4334 142 Click EM, Schnaitman CA (1989) J Bacteriol 171,616-619 143 Cl6ment JM, Hofnung M (1981) Cell 27, 507-514 144 lnouye S, Soberon X, Franceschini K, Nakamura K, ltakura K, Inouy¢ M (1982) Proc Nad Acad Sci USA 79. 3438-3441 145 Hall MN, Gabay J, Schwartz M (1983) EMBO J 2, 15-19 146 Kumamoto CA, Oliver DB. Beckwith J (1984) Nature (London) 308, 863-864 147 Oliver DB, Beckwith J (1982) Cell 30, 311-319 148 Liss LR, Oliver DB (1986) J Biol Chem 261. 2299-2303 .

.

.

.

.

.

.

.

.

.

.

]

.

.

.

.

.

.

.

.

.

.

.

X I

--

~

1

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.