Cell, Vol. 64, 13-15, January
11, 1991, Copyright
0 1991 by Cell Press
Letter to the Editor
Naming a Targeting Signal A staggering variety of potential fates awaits a protein emerging from a ribosome. The presence (or absence) of certain amino acid sequences in a nascent protein constrains the set of metabolic fates that actually befall it by making some fates more probable than others. These sequences, known as targeting signals, can determine the spatial destination of a protein, its folding, posttranslational modifications, and metabolic stability. The understanding of targeting signals has grown enormously over the last fifteen years (reviewed by Blobel, 1980; Wold, 1981; Creighton, 1984; Walter and Lingappa, 1987; Randall et al., 1987; Pfeffer and Rothman, 1987; Burgess and Kelly, 1987; Verner and Schatz, 1988; Gierasch, 1989; Hart1 and Neupert, 1990). Unfortunately, these advances have not been accompanied by a clarification and simplification of terminology. For instance, signals that enable a protein to translocate across the plane of a membrane are called signal sequences, topogenic sequences, targeting sequences, transit sequences, presequences, translocation sequences, or leader sequences. Ad hoc, often semantically cumbersome terms are also used to denote other targeting signals. Yet most of the transformations that proteins undergo can be grouped into five distinct classes of transitions: l
l
l
l
l
Translocation of a protein across the plane of a membrane separating two compartments. Transfer of a protein between compartments without crossing the plane of a membrane (vesicle-mediated transport). Hydrolysis of some or all of the peptide bonds in a protein. Chemical modifications of a protein that do not involve hydrolysis of peptide bonds. Conformational modifications of a protein, including its folding and oligomerization.
I propose the terms transferon, comperfon, degron, and modon for signals that specify each of the first four fates, respectively (Table 1). Examples of the new terminology are presented in Table 2. The distinction between transferons and compartons is illustrated in the figure. (Terminology for the fifth class of transitions has been left for braver people to deal with.) The known complexities of targeting signals are many and continue to accrue. For instance, the distinction between transferons (Table 1) and stop-transfer signals that halt the translocation process is ambiguous for those stop-transfer signals that can act as transferons when positioned within other sequence contexts (Saier et al., 1989). A transferon may or may not be accompanied by a distinct degron that specifies removal of the transferon by a membrane-attached protease. Nonremovable transferons can serve as membrane anchors, sharing this func-
tion with stop-transfer signals (von Heijne, 1988). Transitions of one class often immediately precede, follow, or depend on transitions of another class in the same protein. Amino acid sequences specifying a targeting signal are not always contiguous, are often highly degenerate (Kaiser and Botstein, 1990; Bachmair and Varshavsky, 1989) and may even reside within different subunits of an oligomeric protein (Johnson et al., 1990). The new terminology can accomodate these and other complexities of targeting signals because the proposed terms denote fundamental distinctions and are functional rather than sequence-, structure-, or mechanism-based terms. For example, an X-comparton (Table 1) is defined as a signal that confers retention of a protein in a compartment X, irrespective of whether the retention is achieved
Table 1, Definitions
of Targeting
Signals
Class
Property Conferred
Traneferon
Translocation from a compartment X to a compartment Y (X/Y-transferon) across the real or “virtual” plane of a membranes
Cornpatton
Retention in a compartment X (X-comparton) or transfer from a compartment X to a compartment Y (WY-comparton)without crossing the plane of a membrane
Degron
Metabolic instability of some or all of the peptide bonds in a protein
Modon
Chemical
on a Protein
modification
other than proteolvsis
B When the first compartment is cytoeol, a shorter name (Y-transferon) can be used. The inclusion of a virtual plane expands the definition of transferon to include cases such as nuclear pore-mediated transfer, in which a protein is transported through an aqueous channel across a plane defined by two or more juxtaposed but noncontiguous membranes (Figure 16).
The Distinction
between Transferons
and Compartons
(A and B) Transferons confer on a protein (wavy lines) the ability to translocate through a distinct aqueous channel across the real (A) or “virtual” (8) plane of a membrane (thick lines). Proteinaceous aqueous channels (rectangles) either directly span a membrane (A) or are embedded at points of convergence between juxtaposed but noncontiguous membranes (B). Translocation from and into the nucleus through the nuclear pores is an example of the latter kind. (C) Compartons specify vesicle-mediated transitions in which proteins (dots) move between compartments without crossing the plane of a membrane.
Cell 14
Table 2. Examples of Proposed and Current Terminologies Proposed Name of a Signal
Some Names Currently in Use
Function
ER-transferon
Signal sequence, leader sequence
Translocation from cytosol into endoplasmic reticulum
M-transferon
Matrix targeting sequence, mitochondrial presequence
Translocation from cytosol into mitochondrial matrix (compartment enclosed by the inner membrane). See footnote to Table 1
MAIMS-transferon
Intermembrane space targeting sequence
Translocation from mitochondrial matrix (M) into intermembrane space (IMS) between the inner and outer membranes. Proteins destined to reside in IMS contain both an M-transferon and an MIIMS-transferon, which act sequentially (Hart1 and Neupert, 1990)
Nu-transferon
Nuclear targeting sequence, nuclear translocation signal
Translocation from cytosol into nucleus. See Figure 1B and footnote to Table 1
Eflcomparton
ER retention signal
Retention in endoplasmic reticulum (Pelham, 1969)
GIL-comparton
Lysosome transfer signal
Transfer from trens-Golgi to lysosomes (see text)
Xa-degron
Factor Xa recognition sequence
Conversion of prothrombin into thrombin by Factor Xa, one of the proteases in the blood clotting cascade (Nagai and Thbgersen, 1967)
N-degron
Amino-terminal degradation signal
N-degron is manifested as the N-end rule, which relates the metabolic stability of a protein to the identity of its amino-terminal residue (Johnson et al., 1990)
Nlaa-modon
Signal for conjugation of an amino acid to the amino terminus of a protein
Generation of N-degron (see text)
ManGP-modon
Golgi-mediated generation of Man6P residues on a protein (see text)
by preventing the protein’s transfer or by a compartmentspecific return of the transferred protein (Pelham, 1989; Warren, 1990). This terminology is therefore inherently flexible. Its other advantages are brevity, logical consistency, and uniformity of designations within a class of transitions. The proposed terms can also be used with other biopolymers; for instance, degrons and modons occur in both RNA and DNA. When applied to proteins, the signals of Table 1 are specified exclusively by amino acid sequences. This constraint bypasses the difficulty of deciding, for example, whether a comparton is defined by amino acid residues of a comparton-bearing protein or by the protein’s carbohydrate moieties as well. More generally, if an amino acid sequence specifies a modon, and the resultant chemical modification of a protein is required for the function of its other signals, one still defines these signals exclusively in terms of amino acid sequences-by including sequences that specify the relevant modon as well. (Amino acid sequences that specify a modon encompass both the modification site(s] in a protein and sequences required for the modon’s recognition.) Example: upon arrival from the endoplasmic reticulum (ER) to a compartment close to but apparently distinct from the cis-Golgi, future lysosomal enzymes bearing the mannose-bphosphate (ManGP)-modon (Table 2) are recognized by a transferase that adds N-acetylglucosaminyl phosphate (GlcNAc-P) to a-1,24inked mannose residues present on the N-linked oligosaccharides that have been added to these proteins in the ER. Later, in the Golgi, another enzyme converts the modified mannose residues into the Man8P residues. Since the presence of sterically accessible Man8P residues in a Golgi-located protein is
apparently sufficient for its delivery to lysosomes (Korn-, feld and Mellman, 1989), amino acid sequences comprising this type of a G/L-comparton (Table 2) may be identical to those specifying the ManGP-modon. As with other “downstream“ signals, implementation of the GIL-cornparton is conditional upon the completion of several preceding transitions involving the same protein, in particular those that require the ER-transferon (Table 2) and modons for ER-specific glycosylation. In the absence of additional instructions, the default secretory pathway transports a protein that has entered the ER to the outside of the cell. A default secretion signal is therefore simply the ER-transferon (Table 2). To avoid secretion by the default pathway, a protein that bears the ER-transferon should also contain a comparton, for example, the ERcomparton, the GIL-comparton (Table 2) or a comparton that causes transfer of the protein to storage vesicles of a regulated secretory pathway (Burgess and Kelly, 1987). Proteins recognized as abnormal (misfolded or misassembled) after their translocation into the ER also fail to be secreted by the default pathway. Such proteins are selectively degraded either in the ER or after their diversion (via the Golgi) into lysosomes; they may also be spared but retained in the ER through an associatioa with a resident ER protein such as BiP that bears the ERcomparton (Table 2) (Hurtley and Helenius, 1989; Klausner and Sitia, 1990). A targeting signal may function either differently or not at all in a different in vivo setting. For instance, the N/Argmodon, which causes the conjugation of arginine to the amino terminus of a protein (Table 2) is a single aminoterminal residue, Asp or Glu (and Cys in some cells) (Gonda et al., 1989). Implementation of the N/Arg-modon
Letter to the Editor 15
is required to generate a specific degradation signal in a modon-bearing protein. This signal (N-degron; Table 2) is manifested as the N-end rule, which relates the metabolic stability of a protein to the identity of its amino-terminal residue (Johnson et al., 1990). The NIArg-modon is active in eukaryotes, which contain Arg-tRNA-protein transferase, but inactive in bacteria, which lack this enzyme. A different, N/Phe,Leu-modon, is active in bacteria. This modon causes the conjugation of either Phe or Leu to the amino terminus of a modon-bearing protein @offer, 1980). Thus, specification of a targeting signal implies a reference to an appropriate in vivo environment, and to a relevant genetic background as well. In a discourse that employs the new terms, one could start by introducing the full name of a signal, for instance, the mitochrondrial matrix transferon, then specify and use its acronym, M-transferon (Table 2). When the names for different signals share first letters, one could assign a single-letter acronym to some of these signals (e.g., N-degron) and use multiple-letter acronyms for the other signals (e.g., Nu-transferon). Table 2 lists nonoverlapping acronyms for some of the targeting signals; other such acronyms could evolve gradually, by consensus. I thank R. Baker, B. Bartel, L. Gierasch, G. von Heijne, M. Hochstrasser, K. Lewis, H. Pelham, T. Shrader, and one of the reviewers for comments, and B. Doran for secretarial assistance. Alexander Varehaveky Department of Biology Massachusetts Institute of Technology Cambridge, Massachusetts 02139 References Bachmair, A., and Varshavsky,
A. (1969). Cell 56, 1019-1032.
Blobel, G. (1960). Proc. Natl. Acad. Sci. USA 77 1496-1500. Burgess, T. L., and Kelly, R. B. (1967). Annu. Rev. Cell Biol. 3, 243-293. Creighton,
T. E. (1964). Proteins (New York: W. H. Freeman and Co.).
Gierasch,
L. M. (1969). Biochemistry
28, 923-930.
Gonda, D. K., Bachmair, A., Wiinning, I., Tobias, J. W., Lane, W. S., and Varshavsky, A. (1969). J. Biol. Chem. 264, 16700-16712. Hart& F. U., and Neupert,
W. (1990). Science 247 930-936.
Hurtley, S. M., and Helenius, 277-307. Johnson, 267-291.
A. (1969). Annu.
E. S., Gonda, D. K., and Varshavsky,
Rev. Cell Biol. 5,
A. (1990). Nature 346,
Kaiser, C. A., and Botstein, D. (1990). Mol. Cell. Biol. 70, 3163-3173. Klausner, R. D., and Sitia. R. (1990). Cell 62, 611-614. Kornfeld, S., and Mellman, I. (1969). Annu. Rev. Cell Biol. 5, 463-525. Nagai, K., and ThOgersen, H. C. (1967). Meth. Enzymol.
153461-465.
Pelham, H. R. B. (1969). Annu. Rev. Cell Biol. 5, l-23. Pfeffer, S. R.. and Rothman, 629-652.
J. E. (1967). Annu. Rev. Biochem.
56,
Randall, L. L., Hardy, S. J. S., and Thorn, J. R. (1967). Annu. Rev. Microbial. 47, 507-541. Saier, M. H., Werner, f? M., and Miiller, M. (1969). Microbial. 333-366.
Rev. 53,
Soffer, R. L. (1960). In Transfer RNA: Biological Aspects, D. Soll, J. Abelson, and I? Schimmel, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 493-505. Verner, K., and Schatz, G. (1966). Science 247, 1307-1313.
von Heijne. G. (1966). Biochim. Biophys. Acta 947: 307-333. Walter, P, and Lingappa, V. R. (1967). Annu. Rev. Cell Biol. 2,499-516. Warren, G. (1990). Cell 62, l-2. Weld, F. (1961). Annu. Rev. B&hem.
50, 763-614.
11, 1991, Copyright
0 1991 by Cell Press
Letter to the Editor
Naming a Targeting Signal A staggering variety of potential fates awaits a protein emerging from a ribosome. The presence (or absence) of certain amino acid sequences in a nascent protein constrains the set of metabolic fates that actually befall it by making some fates more probable than others. These sequences, known as targeting signals, can determine the spatial destination of a protein, its folding, posttranslational modifications, and metabolic stability. The understanding of targeting signals has grown enormously over the last fifteen years (reviewed by Blobel, 1980; Wold, 1981; Creighton, 1984; Walter and Lingappa, 1987; Randall et al., 1987; Pfeffer and Rothman, 1987; Burgess and Kelly, 1987; Verner and Schatz, 1988; Gierasch, 1989; Hart1 and Neupert, 1990). Unfortunately, these advances have not been accompanied by a clarification and simplification of terminology. For instance, signals that enable a protein to translocate across the plane of a membrane are called signal sequences, topogenic sequences, targeting sequences, transit sequences, presequences, translocation sequences, or leader sequences. Ad hoc, often semantically cumbersome terms are also used to denote other targeting signals. Yet most of the transformations that proteins undergo can be grouped into five distinct classes of transitions: l
l
l
l
l
Translocation of a protein across the plane of a membrane separating two compartments. Transfer of a protein between compartments without crossing the plane of a membrane (vesicle-mediated transport). Hydrolysis of some or all of the peptide bonds in a protein. Chemical modifications of a protein that do not involve hydrolysis of peptide bonds. Conformational modifications of a protein, including its folding and oligomerization.
I propose the terms transferon, comperfon, degron, and modon for signals that specify each of the first four fates, respectively (Table 1). Examples of the new terminology are presented in Table 2. The distinction between transferons and compartons is illustrated in the figure. (Terminology for the fifth class of transitions has been left for braver people to deal with.) The known complexities of targeting signals are many and continue to accrue. For instance, the distinction between transferons (Table 1) and stop-transfer signals that halt the translocation process is ambiguous for those stop-transfer signals that can act as transferons when positioned within other sequence contexts (Saier et al., 1989). A transferon may or may not be accompanied by a distinct degron that specifies removal of the transferon by a membrane-attached protease. Nonremovable transferons can serve as membrane anchors, sharing this func-
tion with stop-transfer signals (von Heijne, 1988). Transitions of one class often immediately precede, follow, or depend on transitions of another class in the same protein. Amino acid sequences specifying a targeting signal are not always contiguous, are often highly degenerate (Kaiser and Botstein, 1990; Bachmair and Varshavsky, 1989) and may even reside within different subunits of an oligomeric protein (Johnson et al., 1990). The new terminology can accomodate these and other complexities of targeting signals because the proposed terms denote fundamental distinctions and are functional rather than sequence-, structure-, or mechanism-based terms. For example, an X-comparton (Table 1) is defined as a signal that confers retention of a protein in a compartment X, irrespective of whether the retention is achieved
Table 1, Definitions
of Targeting
Signals
Class
Property Conferred
Traneferon
Translocation from a compartment X to a compartment Y (X/Y-transferon) across the real or “virtual” plane of a membranes
Cornpatton
Retention in a compartment X (X-comparton) or transfer from a compartment X to a compartment Y (WY-comparton)without crossing the plane of a membrane
Degron
Metabolic instability of some or all of the peptide bonds in a protein
Modon
Chemical
on a Protein
modification
other than proteolvsis
B When the first compartment is cytoeol, a shorter name (Y-transferon) can be used. The inclusion of a virtual plane expands the definition of transferon to include cases such as nuclear pore-mediated transfer, in which a protein is transported through an aqueous channel across a plane defined by two or more juxtaposed but noncontiguous membranes (Figure 16).
The Distinction
between Transferons
and Compartons
(A and B) Transferons confer on a protein (wavy lines) the ability to translocate through a distinct aqueous channel across the real (A) or “virtual” (8) plane of a membrane (thick lines). Proteinaceous aqueous channels (rectangles) either directly span a membrane (A) or are embedded at points of convergence between juxtaposed but noncontiguous membranes (B). Translocation from and into the nucleus through the nuclear pores is an example of the latter kind. (C) Compartons specify vesicle-mediated transitions in which proteins (dots) move between compartments without crossing the plane of a membrane.
Cell 14
Table 2. Examples of Proposed and Current Terminologies Proposed Name of a Signal
Some Names Currently in Use
Function
ER-transferon
Signal sequence, leader sequence
Translocation from cytosol into endoplasmic reticulum
M-transferon
Matrix targeting sequence, mitochondrial presequence
Translocation from cytosol into mitochondrial matrix (compartment enclosed by the inner membrane). See footnote to Table 1
MAIMS-transferon
Intermembrane space targeting sequence
Translocation from mitochondrial matrix (M) into intermembrane space (IMS) between the inner and outer membranes. Proteins destined to reside in IMS contain both an M-transferon and an MIIMS-transferon, which act sequentially (Hart1 and Neupert, 1990)
Nu-transferon
Nuclear targeting sequence, nuclear translocation signal
Translocation from cytosol into nucleus. See Figure 1B and footnote to Table 1
Eflcomparton
ER retention signal
Retention in endoplasmic reticulum (Pelham, 1969)
GIL-comparton
Lysosome transfer signal
Transfer from trens-Golgi to lysosomes (see text)
Xa-degron
Factor Xa recognition sequence
Conversion of prothrombin into thrombin by Factor Xa, one of the proteases in the blood clotting cascade (Nagai and Thbgersen, 1967)
N-degron
Amino-terminal degradation signal
N-degron is manifested as the N-end rule, which relates the metabolic stability of a protein to the identity of its amino-terminal residue (Johnson et al., 1990)
Nlaa-modon
Signal for conjugation of an amino acid to the amino terminus of a protein
Generation of N-degron (see text)
ManGP-modon
Golgi-mediated generation of Man6P residues on a protein (see text)
by preventing the protein’s transfer or by a compartmentspecific return of the transferred protein (Pelham, 1989; Warren, 1990). This terminology is therefore inherently flexible. Its other advantages are brevity, logical consistency, and uniformity of designations within a class of transitions. The proposed terms can also be used with other biopolymers; for instance, degrons and modons occur in both RNA and DNA. When applied to proteins, the signals of Table 1 are specified exclusively by amino acid sequences. This constraint bypasses the difficulty of deciding, for example, whether a comparton is defined by amino acid residues of a comparton-bearing protein or by the protein’s carbohydrate moieties as well. More generally, if an amino acid sequence specifies a modon, and the resultant chemical modification of a protein is required for the function of its other signals, one still defines these signals exclusively in terms of amino acid sequences-by including sequences that specify the relevant modon as well. (Amino acid sequences that specify a modon encompass both the modification site(s] in a protein and sequences required for the modon’s recognition.) Example: upon arrival from the endoplasmic reticulum (ER) to a compartment close to but apparently distinct from the cis-Golgi, future lysosomal enzymes bearing the mannose-bphosphate (ManGP)-modon (Table 2) are recognized by a transferase that adds N-acetylglucosaminyl phosphate (GlcNAc-P) to a-1,24inked mannose residues present on the N-linked oligosaccharides that have been added to these proteins in the ER. Later, in the Golgi, another enzyme converts the modified mannose residues into the Man8P residues. Since the presence of sterically accessible Man8P residues in a Golgi-located protein is
apparently sufficient for its delivery to lysosomes (Korn-, feld and Mellman, 1989), amino acid sequences comprising this type of a G/L-comparton (Table 2) may be identical to those specifying the ManGP-modon. As with other “downstream“ signals, implementation of the GIL-cornparton is conditional upon the completion of several preceding transitions involving the same protein, in particular those that require the ER-transferon (Table 2) and modons for ER-specific glycosylation. In the absence of additional instructions, the default secretory pathway transports a protein that has entered the ER to the outside of the cell. A default secretion signal is therefore simply the ER-transferon (Table 2). To avoid secretion by the default pathway, a protein that bears the ER-transferon should also contain a comparton, for example, the ERcomparton, the GIL-comparton (Table 2) or a comparton that causes transfer of the protein to storage vesicles of a regulated secretory pathway (Burgess and Kelly, 1987). Proteins recognized as abnormal (misfolded or misassembled) after their translocation into the ER also fail to be secreted by the default pathway. Such proteins are selectively degraded either in the ER or after their diversion (via the Golgi) into lysosomes; they may also be spared but retained in the ER through an associatioa with a resident ER protein such as BiP that bears the ERcomparton (Table 2) (Hurtley and Helenius, 1989; Klausner and Sitia, 1990). A targeting signal may function either differently or not at all in a different in vivo setting. For instance, the N/Argmodon, which causes the conjugation of arginine to the amino terminus of a protein (Table 2) is a single aminoterminal residue, Asp or Glu (and Cys in some cells) (Gonda et al., 1989). Implementation of the N/Arg-modon
Letter to the Editor 15
is required to generate a specific degradation signal in a modon-bearing protein. This signal (N-degron; Table 2) is manifested as the N-end rule, which relates the metabolic stability of a protein to the identity of its amino-terminal residue (Johnson et al., 1990). The NIArg-modon is active in eukaryotes, which contain Arg-tRNA-protein transferase, but inactive in bacteria, which lack this enzyme. A different, N/Phe,Leu-modon, is active in bacteria. This modon causes the conjugation of either Phe or Leu to the amino terminus of a modon-bearing protein @offer, 1980). Thus, specification of a targeting signal implies a reference to an appropriate in vivo environment, and to a relevant genetic background as well. In a discourse that employs the new terms, one could start by introducing the full name of a signal, for instance, the mitochrondrial matrix transferon, then specify and use its acronym, M-transferon (Table 2). When the names for different signals share first letters, one could assign a single-letter acronym to some of these signals (e.g., N-degron) and use multiple-letter acronyms for the other signals (e.g., Nu-transferon). Table 2 lists nonoverlapping acronyms for some of the targeting signals; other such acronyms could evolve gradually, by consensus. I thank R. Baker, B. Bartel, L. Gierasch, G. von Heijne, M. Hochstrasser, K. Lewis, H. Pelham, T. Shrader, and one of the reviewers for comments, and B. Doran for secretarial assistance. Alexander Varehaveky Department of Biology Massachusetts Institute of Technology Cambridge, Massachusetts 02139 References Bachmair, A., and Varshavsky,
A. (1969). Cell 56, 1019-1032.
Blobel, G. (1960). Proc. Natl. Acad. Sci. USA 77 1496-1500. Burgess, T. L., and Kelly, R. B. (1967). Annu. Rev. Cell Biol. 3, 243-293. Creighton,
T. E. (1964). Proteins (New York: W. H. Freeman and Co.).
Gierasch,
L. M. (1969). Biochemistry
28, 923-930.
Gonda, D. K., Bachmair, A., Wiinning, I., Tobias, J. W., Lane, W. S., and Varshavsky, A. (1969). J. Biol. Chem. 264, 16700-16712. Hart& F. U., and Neupert,
W. (1990). Science 247 930-936.
Hurtley, S. M., and Helenius, 277-307. Johnson, 267-291.
A. (1969). Annu.
E. S., Gonda, D. K., and Varshavsky,
Rev. Cell Biol. 5,
A. (1990). Nature 346,
Kaiser, C. A., and Botstein, D. (1990). Mol. Cell. Biol. 70, 3163-3173. Klausner, R. D., and Sitia. R. (1990). Cell 62, 611-614. Kornfeld, S., and Mellman, I. (1969). Annu. Rev. Cell Biol. 5, 463-525. Nagai, K., and ThOgersen, H. C. (1967). Meth. Enzymol.
153461-465.
Pelham, H. R. B. (1969). Annu. Rev. Cell Biol. 5, l-23. Pfeffer, S. R.. and Rothman, 629-652.
J. E. (1967). Annu. Rev. Biochem.
56,
Randall, L. L., Hardy, S. J. S., and Thorn, J. R. (1967). Annu. Rev. Microbial. 47, 507-541. Saier, M. H., Werner, f? M., and Miiller, M. (1969). Microbial. 333-366.
Rev. 53,
Soffer, R. L. (1960). In Transfer RNA: Biological Aspects, D. Soll, J. Abelson, and I? Schimmel, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 493-505. Verner, K., and Schatz, G. (1966). Science 247, 1307-1313.
von Heijne. G. (1966). Biochim. Biophys. Acta 947: 307-333. Walter, P, and Lingappa, V. R. (1967). Annu. Rev. Cell Biol. 2,499-516. Warren, G. (1990). Cell 62, l-2. Weld, F. (1961). Annu. Rev. B&hem.
50, 763-614.
