Ovalbumin Utilizes an NHz-terminal Signal Sequence*

In synchronized translation experiments in the wheat germ and reticulocyte lysate systems, ovalbumin (385 amino acids) was glycosylated by and segregated in dog pancreatic microsomes only if microsomes were added before the nascent ovalbumin polypeptide contained less than 150 amino acids. This would place the “signal” sequence of ovalbumin prior to residue 150, in contrast to a previous report. synthesized

In synchronized translation experiments in the wheat germ and reticulocyte lysate systems, ovalbumin (385 amino acids) was glycosylated by and segregated in dog pancreatic microsomes only if microsomes were added before the nascent ovalbumin polypeptide contained less than 150 amino acids. This would place the "signal" sequence of ovalbumin prior to residue 150, in contrast to a previous report.
In the study of the biosynthesis of secretory and membrane proteins, it is desirable to ascertain whether insertion of the newly synthesized protein occurs in a co-translational manner. This determination can be made by means of a synchronized translation experiment. In such a study, initiation of protein synthesis is allowed for a brief period, after which initiation is specifically inhibited, and the ribosomes undergoing translation of the messenger are allowed to traverse the message synchronously. The addition of dog pancreatic microsomes at various stages of growth of the nascent chain, followed by completion of the protein in the presence of the membranes, will show when such membranes must be present if insertion is to occur at all. The sequence mediating insertion into membranes, the "signal sequence," must have been synthesized prior to the time when addition of microsomes no longer brings about insertion.
This type of experiment has been performed with several proteins, including membrane proteins like vesicular stomatitis virus G and Sindbis PE2, pancreatic secretory proteins, and the peculiar case of ovalbumin (1-4). In all save the last case, insertion is prevented if more than about 100 amino acid residues are synthesized before the addition of membranes. This constitutes good evidence for the existence of an NH2terminal signal sequence which directs co-translational insertion; this sequence must act rapidly after its emergence from the ribosome or else become ineffective. The exception, ovalbumin, seemed not to require membranes until over half the protein was synthesized. This and other results suggested that ovalbumin utilizes an internal signal sequence (4).
Unlike other secretory proteins, there is no removal of an NHZ-terminal leader peptide from ovalbumin upon its segregation in the endoplasmic reticulum (5). Althougb no NH2terminal residues are removed, nascent ovalbumin competes with other secretory proteins (which have cleaved leader peptides) for insertion into microsomes in vitro (6). This * This research was supported by Grants AI-08814, AM 15322, and AM 27375 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
suggests that, despite the failure to remove the NHZ-terminal portion, microsomal segregation of ovalbumin and other proteins was accomplished by the same mechanism. The implication, then, that this mechanism can accommodate an internal signal sequence made confiiation of the presence of such a sequence in ovalbumin a vital matter.

EXPERIMENTAL PROCEDURES
Gel Electrophoresis and Autoradiography-SDS'-gel electrophoresis analysis was performed by the method of Laemmli (7), except that samples were prepared with a modified sample buffer (2% w/v) SDS, 10% (v/v) glycerol, 20 m~ dithiothreitol, 0.7 M 2-mercaptoethanol, 0.01% bromphenol blue dye, and 80 mM Tris-HC1 buffer, pH 6.8). Fluorography was performed by the procedure of Laskey and Mills (8). Autoradiography was performed with Kodak SB-5 film. Exposures were scanned on a microdensitometer (Joyce-Loebl and Co.) using a wedge with a full scale deflection of 1.16 optical density units. Such exposures are within the linear response region of the film and the microdensitometer.
Preparation of VSV Messenger RNA-BHK cells (9) were harvested 5 h after infection by 10 plaque-forming units/cell of the Glasgow isolate of VSV. The extraction of mRNA was as described in Ref. 9, except that 50 pg/ml of Proteinase K was added to the lysed cell suspension after the DNase I treatment, and the solution was incubated an additional 10 min at 37 "C. Typically, the RNA was not purified by oligo(dT) columns and was used in cell-free translations at concentrations of 300-1000 pg/ml.
Cell-free Translation of Messenger RNA-Reagents were prepared from sterile water and stored frozen at -20 "C for up to 1 month before discarding. ATP stock solution was neutralized with ammonium hydroxide before use. Hemin was prepared as a 4 mg/ml stock solution in ethylene glycol. Amino acids (5 rn each) were made up as a stock without cysteine and methionine; 5 mM cysteine was made up separately. To some reactions, dog pancreatic microsomes, prepared by the method of Shields and Blobel (IO), were added at 3-6 AZm units/ml. Wheat germ extract was prepared as described by Roberts and Paterson (11). Reticulocyte lysate, prepared from rabbits made anemic with acetylphenylhydrazine (12), was treated with micrococcal nuclease to destroy endogenous messenger RNA, as described by Pelham and Jackson (13). VSV messenger RNA was used as described above. Ovalbumin messenger RNA (oligo(dT)-cellulosepurified), the &t of Drs. R. Meek and R. D. Palmiter of the University of Washington, was used at 5 pg/ml. At all times, autoclaved or disposable utensils were used.
Endoglycosidase H Digestion of Reactions-Translation reactions were boiled 5 min in 4 volumes of 2% SDS and 2% 2-mercaptoethanol. These were made 0.25 M in sodium citrate buffer, pH 6.0,0.01% NaN3, and 1 mM in phenylmethylsulfonyl fluoride, diluting the boiled sample 2-fold. Each sample was split and incubated for 18 h at 37 "C with 0.1 volume of the same buffer or with 0.1 volume of endoglycosidase H (30 pg/ml), a gift from Dr. P. W. Robbins of MIT, essentially as described by Zilberstein et al. (16). Reactions were precipitated with acetone, and the pellets were lyophilized and dissolved in 60 pl of sample buffer, with heating at 100 "C for 5 min, in preparation for SDS-gel electrophoresis and fluorography.
Protease Treatment of Dog Pancreatic Microsomes-Cell-free translations which were to be analyzed for the degree of protection of the translated products from protease degradation were treated with 100 pg/ml of emetine at 25 OC for 5 min to inhibit further protein synthesis. In some cases, 3 mM tetracaine-HCl, pH 7, was also added during this period. This anesthetic apparently improves the efficiency of protease protection of dog pancreatic microsomes (3).
Protease digestion was at 0 "C for 60 min. For VSV G protein, trypsin, treated with tosylphenylalanyl chloromethyl ketone, at 1 mg/ ml was used, while for ovalbumin, Proteinase K (Merck) at 200 pg/ml was used. Trypsin was inhibited by the addition of excess soybean trypsin inhibitor, while Proteinase K was inhibited with 2 RIM phenylmethylsulfonyl fluoride. Digested and inhibited samples were made up in sample buffer and heated 5 min at 100 "C prior to storage at -70 "C.

RESULTS
Most of the ovalbumin synthesized in the wheat germ cellfree system containing dog pancreatic microsomes was in two Comdetion slowly migrating glycosylated forms, OV1 and OV2 (Fig. 1,  lune a). Upon treatment with endoglycosidase H these convert to a form that co-migrates with the unglycosylated form made by cell-free systems without microsomes, OV,, (Fig. 1, lanes e and f). OVI and OV2 resist degradation by Proteinase K, unlike OVO, unless detergent is present during digestion (Fig. 1, lanes b, c, and g). OVO In parallel with the experiment of Figs. 2 and 3, VSV messenger RNA was translated in the wheat germ system under identical conditions of synchronization. In lanes a through h, translation was in the absence of microsomes. To measure the degree of completion of G protein, samples were removed into SDS gel sample buffer at 2, 4.5, 7, 12, 17, 22,32, and 62 min after RNA addition. In lanes i throughp, microsomes were added at 2, 4.5, 7, 12, 17, 22, 32, and 62 min after RNA addition. The reactions were then incubated for the remainder contain at least one high mannose oligosaccharide, and are segregated in the microsomal vesicles. OVo is not segregated within the microsomes. As judged by its resistance to endoglycosidase H (data not shown). OVO is unglycosylated. Palmiter et al. (17) have obtained similar results on the OVO, OV1, and Ova forms of ovalbumin.
The synchronized membrane addition experiment was performed first in the wheat germ system using inhibitors that inhibit initiation to 99% or better (1.5 rn 7-methylguanosine 5"phosphate) (1). The degree of completion of ovalbumin was monitored by delivery of aliquots of reaction at various times directly into SDS gel sample buffer and analyzing by SDS-gel electrophoresis and fluorography (Fig. 2). The timing of trans-h i j k l m n o p of a 65-min period from the time of inhibitor addition. All of the reactions were then treated with trypsin. The G species in lanes i, g, and h is unglycosylated. The slower migrating C species in lanes i through k contain two Asn-linked high mannose oligosaccharides, has lost the 16 NH2-terminal amino acids (the "signal sequence") and has had removed, by the trypsin, the 30 COOH-terminal amino acids that are accessible to the cytoplasm (1). The labels on the left denote the principal VSV proteins. location of ovalbumin was monitored by adding microsomes to aliquots of the synchronized reaction at various times, completing read-off of the polypeptide, and analyzing both the degree of formation of glycosylated ovalbumin, and its protection from Proteinase K (Fig. 3).
Ovalbumin was completed after about 17 min of translation, yielding a rate of elongation of about 22 amino acid residues/ min. The amount of ovalbumin completed in the system did not increase upon longer incubation (Fig. 2), demonstrating that the inhibition of initiation had been effective. In the same system, VSV G was completed after 27 min, yielding a rate of elongation of 21 residues/min (Fig. 4). Ovalbumin was protected from protease as long as membranes were added before were synthesized if membranes were added prior to the 7-min time, whereas OVo, the unglycosylated form, was synthesized if membranes were added after this time (Fig. 3). VSV G was protected if membranes were present before 5 min had elapsed, corresponding to a synthesis of about 100 amino acid residues (Figs. 4 and 6B). These results demonstrate that the sequence which directs the translocation of ovalbumin occurs prior to about residue 150. This is much more in keeping with an NHp-terminal signal sequence for ovalbumin.
As the experiment seeming to show that ovalbumin has an internal signal sequence (4) was performed in the reticulocyte lysate system, rather than the wheat germ system, it could be argued that the latter system gives anomalous results. We therefore repeated the synchrony experiment in the reticulocyte lysate system, using 1.5 p edeine and 4.5 mM 7-methylguanosine 5"phosphate to inhibit initiation of protein synthesis. This combination of inhibitors, fist used by Scheele and co-workers (3), inhibited initiation greater than 99% (data not shown). Synthesis of ovalbumin was completed after about 46 min (Figs. 5 and 6 0 giving a rate of elongation of 8.5 amino acid residues/min. Glycosylation and protection from protease protection from protease (D) in the reticulocyte lysate system. For all panels, the times of translation a t which 50% completion or Protection were observed were taken as the mean values for these Parameters. Arrows indicate the times of addition of inhibition of initiators Of protein synthesis.
were prevented if microsomes were added after about 16 min of translation (Figs. 5 and 6D) placing the signal sequence prior to residue 140. This result agrees well with the result obtained in the wheat germ system. Again, synchronized translation of VSV glycoprotein in the synchronized reticulocyte lysate system placed the VSV G signal sequence prior to residue 100, although the rate of elongation, 13 amino acids/ min, was significantly faster than for ovalbumin (Fig. 6, C and

.
To summarize, in both the wheat germ and reticulocyte lysate synchronized translation systems, microsomes had to be added before 150 amino acid residues of ovalbumin had been synthesized if glycosylation and segregation were to occur. This would place the ovalbumin signal sequence within the fist 150 amino acids, rather than around residue 250 as was suggested by Lingappa et al. (4). In agreement with these results, investigators in Palmiter's laboratory (18)' have found that nascent chains of ovalbumin as short as 50 amino acid are bound to and inserted into microsomes.

DISCUSSION
The experiments on ovalbumin, performed by Lingappa et al. (4), centered on the isolation of a tryptic peptide of ovalbumin which, when present at concentrations of several milligrams per milliliter, was a competitive inhibitor of the translocation of in vitro synthesized prolactin into dog pancreatic microsomes. This peptide was localized to residues 229 to 276 in ovalbumin, a protein of 385 total residues. On the basis of this competition, it was suggested that this peptide acted as an internal signal sequence for insertion of nascent ovalbumin into microsomal membranes. The evidence that this peptide was the actual signal sequence for ovalbumin was provided by a synchronized membrane addition experiment as outlined above, using the reticulocyte lysate system with 1 ~L M pactamycin to inhibit initiation. In this experiment, ovalbumin was no longer glycosylated if membranes were added after about 5.5 min, in a system where the protein was completed in 8.5 min. This placed the signal sequence prior to residue 250 of ovalbumin, and was interpreted as evidence for an insertion signal at that point. By contrast, our results, using both wheat germ and reticulocyte cell-free systems, position the "signal" sequence prior to residue 150, and exclude any signal function for an internal set of amino acids in the COOHterminal two-thirds of the protein.
Exactly why Lingappa and co-workers obtained for ovalbumin a critical point for insertion of 250 amino acids is not clear. One problem was the choice of inhibitor, pactamycin. The synchrony experiment depends on a virtually complete shut-off of initiation, without inhibition of elongation, in order to give accurate results. This is because the time allowed to complete the polypeptide is much longer than the time window allowed for polypeptide chain initiation. For example, if the initiation window is 5% of the completion period, 95% inhibition of initiation will allow as much initiation after the inhibitor is added as before. This nonspecific initiation generates an artifactually longer time before membranes no longer can effect insertion. For accurate results, therefore, initiation must be inhibited 99% or better. 1 ,ug/ml of pactamycin inhibits initiation less than the 99% required, and at R. D. Palmiter, personal communication.
higher concentration also slows the elongation rate, making the time needed for completion even longer (19).
Another possibility is suggested by our own results in the reticulocyte lysate system. Ovalbumin was observed by us to take longer to complete in this system than VSV glycoprotein, although VSV glycoprotein (570 amino acids) is larger than ovalbumin (385 amino acids). If Lingappa et al. (4) misconstrued the completion time of ovalbumin to have been shorter than it actually was, an artifactually high elongation rate would be obtained which would skew their results. As the completion data were not shown, however, this can only be a speculation. Nonetheless, this result indicates that elongation rates for one protein should not be applied to different proteins translated in the same system, at least for the reticulocyte lysate system.
The actual location of the signal sequence of ovalbumin is not determined by our results, except that it is NHa-terminal to residue 150. A combination of our results with those of Meek et al. (18)3 would seem to place the time of the sequence's action as between the times when residues 50 and 150 are synthesized. Allowing for 40 residues residing within the ribosome (3), it is clear that the nascent chain of ovalbumin interacts with membranes almost as soon as it emerges from the ribosome. The sequence thus constitutes an actual NH2-terminal signal sequence, which need not be cleaved from the completed polypeptide in order to effect segregation.