Signal recognition particle-dependent insertion of coronavirus E1, an intracellular membrane glycoprotein.

The membrane insertion of the E1 protein of a coronavirus, mouse hepatitis virus A59, was studied in a wheat germ cell-free translation system. E1 is a transmembrane protein spanning the lipid bilayer several times. It is synthesized without a cleavable signal sequence, localized intracellularly, and not transported to the cell surface. It thus represents a model intracellular protein. We found that the synthesis of E1 is specifically and stably blocked by the addition of signal recognition particle to the wheat germ system. Subsequent addition of salt-extracted pancreatic microsomes resulted in the full release of this arrest as well as the completion and the correct membrane integration of E1. Such signal recognition particle-induced arrests failed to produce shorter peptides of a defined length. Addition of signal recognition particle to a synchronized translation at any time during the synthesis of about the first two thirds of E1 (150 amino acids) blocked further translation, suggesting that the most C-terminal of the three internal hydrophobic domains of E1 could function as its signal sequence.

The membrane insertion of the E l protein of a coronavirus, mouse hepatitis virus A59, was studied in a wheat germ cell-free translation system. E l is a transmembrane protein spanning the lipid bilayer several times. It is synthesized without a cleavable signal sequence, localized intracellularly, and not transported to the cell surface. It thus represents a model intracellular protein. We found that the synthesis of E l is specifically and stably blocked by the addition of signal recognition particle to the wheat germ system. Subsequent addition of salt-extracted pancreatic microsomes resulted in the full release of this arrest as well as the completion and the correct membrane integration of E l . Such signal recognition particle-induced arrests failed to produce shorter peptides of a defined length. Addition of signal recognition particle to a synchronized translation at any time during the synthesis of about the first two thirds of E l (150 amino acids) blocked further translation, suggesting that the most C-terminal of the three internal hydrophobic domains of E l could function as its signal sequence.
Signal recognition particle (SRP') has been shown to be required for the proper membrane insertion or translocation of a number of membrane and secretory proteins (1,2). It is generally accepted that SRP binds to the signal peptide as it emerges from the ribosome, specifically arresting translation (2). The blocked translation complex migrates to the membrane of the rough endoplasmic reticulum, where an interaction with docking protein (also called SRP receptor) causes the release of the translation block; the nascent polypeptide is then translocated across the endoplasmic reticulum membrane concomitant with its continued synthesis (3,4). SRP has been shown to be involved in the recognition of secretory (3,4), lysosomal (5) as well as plasma membrane (6,7) proteins.
Several recent attempts have been made to determine if this sequence of events is also a characteristic of proteins which remain in an intracellular location, principally the endoplasmic reticulum. Ca2'-ATPase of rabbit sarcoplasmic * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Supported at the European Molecular Biology Laboratory by a short-term fellowship from the European Molecular Biology Organisation.
The abbreviations used are: SRP, signal recognition particle; MHV, mouse hepatitis virus; RMK, salt-extracted, nuclease-treated rough microsomes; SDS, sodium dodecyl sulfate. reticulum (S), baby hamster kidney cell hydroxymethylglutaryl-CoA reductase (9) and rabbit liver cytochrome P-450 (10) have all been examined with regards to the role of SRP in their membrane insertion. However, a clear picture has yet to emerge. It was initially postulated (9) that endoplasmic reticulum proteins lack a cleavable signal sequence. Recently, Rosenfeld et al. (11) have shown that the endoplasmic reticulum-specific ribophorins possess a transient signal sequence, cleaved cotranslationally. Whether SRP was required for mediating this insertion was not studied. Of the remaining proteins, all were found to require SRP for integration, yet an SRP-mediated translation arrest was observed only for cytochrome P-450 (10). In this case, an arrested peptide, indicative of a translation block (3), could not be observed, possibly due to lack of an antibody capable of recognizing such a species (10).
Mouse hepatitis virus A59, a coronavirus, has the unusual property of acquiring its envelope by budding at membranes inside the cell (12,13) rather than at the plasma membrane. This behavior is associated with the 0-glycosylated viral glycoprotein E l , which, after its synthesis, is confined to membranes of the endoplasmic reticulum and, perhaps, the Golgi region (13, 14) and only ever reaches the cell surface as part of the mature virion. The fact that the restricted localization of E l is an intrinsic property of the protein itself was further demonstrated both by microinjection of specific mRNA2 as well as by expression of the E l cDNA2v3 in cultured cells. In both cases, the protein was detected solely in intracellular membranes of ER and possibly Golgi. Thus, the E l molecule represents a model intracellular membrane protein.
Previously, we described the cotranslational assembly of E l into dog pancreatic microsomal vesicles in uitro (15) and showed both that it lacks a cleavable signal sequence and that it spans the membrane several times, a feature confirmed by determining the primary structure of the molecule (16). Interestingly, the N-terminal amino acid sequence does not resemble the signal sequences of either eukaryotic or prokaryotic secretory or membrane proteins (17). There are, however, three stretches of about 20 uninterrupted hydrophobic or uncharged amino acids at positions 27-46,57-77, and 82-103 respectively (16).4 It was therefore of interest to see if E l needs SRP for its insertion into membranes and whether or not such putative "internal" signals participate in this event. Additionally, choosing E l has the advantage in that purified preparations of E l mRNA can be obtained, eliminating the need to immunoprecipitate either complete or incomplete products of cell-free translations.
In this study, we demonstrate that SRP is required for the correct insertion of El into pancreatic microsomes in a wheat germ cell-free system. SRP brings about a stable arrest of translation which can subsequently be released by the addition of salt-extracted rough microsomes. Significant is the fact that the SRP arrest can still be induced relatively late during the synthesis of El, suggesting an internal signal sequence. These arrests did not generate a specific blocked peptide of uniform length.

EXPERIMENTAL PROCEDURES
Preparation of mRNA-Poly(A)+ RNA was prepared from MHV-A59-infected Sac-cells as described (18). mRNA enriched for E l (RNAG) was purified by agarose gel electrophoresis as described (18) and then bound to oligo(dT)-cellulose, eluted, and precipitated with ethanol.
In Vivo Labeling of El-Monolayers of Sac-cells were infected with MHV-A59 at multiplicities of 60 and pulse-labeled at 6.5 h postinfection with 120 pCi of [%]methionine/ml (19). Cells were then chilled, washed with Dulbecco's phosphate-buffered saline, and allowed to swell for 10 min in 10 mM Tris-C1 (pH 8.0), 2 mM MgC12, 40 pg/ml phenylmethylsulfonyl fluoride. Cell breakage was facilitated by pipetting up and down. Homogenates were centrifuged for 5 min at 1300 X g to remove nuclei and debris, and the supernatants were treated as described below.
Proteolysis-Samples of 1300 X g supernatants or the products of in vitro translations were treated for 15 min at 37 "C with proteinase K (Merck, Darmstadt, West Germany) at a final concentration of 1 mg/ml. The reaction was stopped by addition of phenylmethylsulfonyl fluoride to a concentration of 40 pg/ml. Samples were placed on ice for 10 min, and proteins were precipitated in 10% trichloroacetic acid prior to SDS-polyacrylamide gel electrophoresis.
In Vitro Translation-The preparation of wheat germ extract, dog pancreas microsomes, SRP, and salt-extracted, nuclease-treated rough microsomes (RMK) was described previously (4). Details of each translation are presented in the figure legends. SRP units are those defined by Walter and Blobel (20). Typically, translation in a wheat germ extract was performed with 0.2 pg of mRNA/25-pl translation. Synchronization was achieved by adding 7-methylguanosine 5'-monophosphate (P-L Biochemicals) (21) to a final concentration of 4 mM after 2 min of translation. Aliquots were removed at various times and further treated as stated in the figure legends. Samples were analyzed by SDS-polyacrylamide gel electrophoresis, fluorography, and densitometric scanning as previously described   and RMK were treated with proteinase K, which caused the disappearance of nudeocapsid protein, and a reduction in molecular mass of E l (Fig. lA, lane 4 ) to a 22,000-dalton form indistinguishable from that found by protease treatment of microsomes from virally infected cells (Fig. 1B). RMK in the absence of SRP caused neither a block of E l synthesis (Fig.  lA, lune 5 ) nor a conversion to protease resistance (Fig. l A ,  lune 6). Thus, SRP causes a specific block of E l synthesis and is required for correct assembly of the protein into the microsomal membrane.

SRP Requirement for
Stability of the SRP-dependent Translation Block-A recent study indicated that the inability of SRP to induce a translation arrest in the case of Ca2+-ATPase might be due to the lack of a cleavable signal sequence. In such cases, SRP would affect the rate of synthesis instead of producing a stable block (8). To characterize further the specific arrest of E l translation by SRP, the stability of the translation block was measured. E l RNA was translated in vitro in a synchronized system (as described under "Experimental Procedures") to prevent reinitiation of synthesis. In this study, we initially allowed translation to proceed in the presence of SRP for 20 min. At this point, only about one quarter of the total possible E l which can be synthesized in the absence of SRP has been completed (compare Fig. 2, lane 2 to lane I ) . When this translation was allowed t o continue for a total of 40 min, only a small additional quantity of E l appeared (lane 3). This inhibition was completely rescued by the addition of RMK after 20 min (lane 4 ) . Thus, rather than causing a general inhibition of E l synthesis, SRP brings about a stable, and completely reversible, arrest of translation.
T i n e of Addition of SRP-An examination of the primary structure of E l (16) indicates that its N terminus does not possess the hydrophobic properties characteristic of signal sequences (17). Instead, three hydrophobic stretches can be found in the region between amino acids located 27-46, 57-77, and 82-103 from the N terminus. The use of timed SRP addition should enable one to determine if the hydrophobic sequences distal to the N terminus can also function as insertion signals. In principle, the further away from the N terminus the signal resides, the longer the period during which SRP should be able to be added to the translation and still bring about an arrest. Such an experiment is detailed in Fig. 3. Completed E l (228 amino acids) first appears between 18 and 22 min of translation (Fig. 3, A and B). SRP was added to parallel synchronized translations at various times, and these were allowed to continue for a total of 30 min. It can be seen in Fig. 3, C and L), that addition of SRP as late as 17 min after the start of translation still results in a half-maximal inhibition of E l synthesis (arrow TI).
Since the rate of elongation appears to be linear (see Fig.  3A), one can calculate that at 16-18 min approximately two thirds of the molecule, or roughly 150 amino acids of the total 228, have been polymerized. The effectiveness of such a late addition of SRP suggests that even the most C-terminal stretch of hydrophobic amino acids (82-103) could serve as a signal sequence.
It is clear from Fig. 3C that bands were not observed that could represent a blocked peptide. This was corroborated by identical studies carried out using gel systems capable of even better resolution of such smaller peptides (data not shown).

DISCUSSION
We have characterized the involvement of SRP in the membrane insertion of El, a model intracellular membrane protein. In the case of most plasma membrane and secretory proteins, SRP functions by bringing about an arrest of translation which can be relieved by the presence of rough microsomes (3). Our results indicate that insertion of E l is carried out in the same manner. We have shown that SRP causes a specific (Fig. 1A) and quite stable (Fig. 2) arrest of synthesis of E l in vitro. Addition of salt-washed microsomal membranes causes a release of the translation block (Fig. lA, lane 3) which is apparently quantitative (Fig. 2). The resulting protein is assembled into the membrane, in a form which is indistinguishable, as judged by protease resistance, from the protein produced in viuo (Fig. l B , lane 2; Fig. lA, lane 4 ) .
As we used mRNA which encoded virtually only El, it was thought that the signal could be defined by identifying the arrested peptide produced by the addition of SRP. We have consistently failed, however, to observe such a species. In these translations, arrested peptides of a defined length, if present, would have been visible.
A very recent report has shown a similar insertion scheme for cytochrome P-450 (10). In this case, no blocked peptide could be found by immunoprecipitation of the cytochrome synthesized in vitro. This finding is strengthened by our results in which immunoprecipitation is eliminated through the use of a specific E l mRNA. One must conclude, therefore, that the translation arrest, although stabIe, is relatively imprecise, such that specific bands cannot be seen on a gel. Last, the results of this study as well as those of others (10) demonstrate that no correlation exists between the lack of a cleavable signal sequence and the failure to observe translation arrest by SRP in vitro (9).
Although we prove that E l is treated like most membrane and secretory proteins, studies using timed SRP addition to synchronous translations suggest possibilities that warrant further attention. Apparently SRP can exert its effect on E l synthesis even when it is added quite late in translation (Fig.   3, C and D); approximately 150 amino acids of a total of 228 can be polymerized before translations can no longer be arrested. Based on known measurements of arrested N-terminal peptides (3, 4), one assumes that a block occurs when the signal sequence emerges from the large ribosomal subunit. This would entail having about 40 amino acids in addition to the signal sequence in order to span the ribosome. Thus, in the case of El, a block could occur earliest at a chain length of 85-90 amino acids (approximately 10,000 daltons) if the signal sequence was represented by the first hydrophobic domain (amino acids 27-46). Similarly, the size of the arrested nascent chain would correspond to 115-120 amino acids (12,500 daltons) and 140-150 amino acids (16,000 daltons) in the case of the second and third domains, respectively. On the basis of our finding that SRP still brings about an arrest when two thirds of the protein (approximately 17,000 daltons) is completed, we interpret these results to mean that even the most C-terminal hydrophobic domain can interact with SRP resulting in this arrest. We cannot completely rule out the possibility that the most N-terminal hydrDphobic domain is the sole signal and is still being recognized by SRP at a point as late in translation when two thirds of E l is completed. In light of the fact that the signal sequence of most secretory proteins is inaccessible to cleavage by signal peptidase late in their translation (23), it seems unlikely that such signals would still be accessible to the much larger SRP.
These findings substantiate the notion that a considerable fraction of a protein can be synthesized and still become inserted into the membrane. The insertion of E l into microsomes late in its synthesis has been documented previously using similar synchronized translation in the reticulocyte of Coronavirus E l lysate system (15). The kinetics of insertion in this system paralleled those of the SRP arrests seen here, implying that the same signals are functioning in both cases.
Of the three possible signals of El, all are internal, i.e. not Nor C-terminal. Several proteins having uncleaved N-terminal signals have been shown to become inserted into the membrane in an SRP-dependent fashion (8)(9)(10). E l represents the first case of a protein with an implicit internal signal which has been shown to require SRP. This result points out that SRP is also capable of recognizing this type of sequence and that such sequences can be functional in more C-terminal locations.