Cell, Vol. 62, 1-2, July 13, 1990, Copyright
0 1990 by Cell Press
Salvage Receptors: Two of a Kind? Graham Warren Imperial Cancer Research Lincoln’s Inn Fields London WC2A 3PX England
Fund
The lumen of the endoplasmic reticulum (ER) is densely packed with resident proteins. Many of these are involved in protein assembly (Rothman, 1989); they bind nascent polypeptides that have crossed the ER membrane and catalyze their folding. Since signals are not needed for assembled proteins to leave the ER (Wieland et al., 1987) one might imagine that any soluble protein in the lumen would be free to leave. But the resident assembly proteins, perhaps as a consequence of their function, are soluble proteins that do not leave the ER. They are retained by a C-terminal tetrapeptide salvage signal, which in mammalian cells has the sequence -KDEL. When this salvage signal is removed, the truncated protein is slowly secreted; when transplanted onto a protein that is normally secreted, that protein is retained within the ER (Munro and Pelham, 1987). Several lines of evidence argue for a catalytic mechanism (Pelham, 1989; Warren, 1987) whereby resident soluble proteins occasionally escape and are then salvaged by a receptor that recognizes the C-terminal tetrapeptide and returns them to the ER (see figure). The Salvage Pathway A key prediction is that these resident ER proteins must temporarily leave the ER and therefore be exposed to post-ER enzymes. Some evidence for this was obtained in mammalian cells (Pelham, 1988) but more convincing evidence has now been provided using the yeast Saccharomyces cerevisiae (Dean and Pelham, 1990). Prepro-a factor, a secretory protein, was fused to -HDEL, the preferred salvage signal in this yeast. The fused protein was retained within the cell, but some was modified by the addition of al-6 mannose linkages. This modification characteristic of the early Golgi did not occur in a mutant (sec78) defective in ER-to-Golgi transport at the restrictive temperature (Schekman, 1985). This showed that the protein had to leave the ER in order to be modified. Fractionation experiments localized the modified protein to the ER, so it must have returned to the ER following modifi-
Endoplasmic
Reticulum
Salvage Compartment
Minireview
The Yeast Salvage Receptor The next step was to identify the salvage receptor. S. cerevisiae will only grow on sucrose if it secretes the enzyme invertase. Transplanting -HDEL onto this protein prevents secretion and hence growth on sucrose. Mutants lacking a functional salvage receptor should then grow on sucrose. Several mutants have been characterized (Pelham et al., 1988) and one, erd;! (ER retention defective), disrupts specifically the salvage pathway. Pelham and his colleagues have now shown that the ERDP gene encodes the yeast salvage receptor (Lewis et al., 1990; Semenza et al., 1990). From the model in the figure, one would expect that overexpression of a protein fused to -HDEL could saturate the receptor. Cells overexpressing prepro-a factor fused to -HDEL were found to secrete one of their resident ER proteins. If this is the consequence of too few receptors, then overexpressing the receptor should reverse this process; the ERDP protein expressed from a multicopy plasmid did just that. In addition, the early Golgi modifications to the fusion protein were also suppressed, suggesting that the protein was being removed either from an earlier compartment or before the Golgi enzymes had time to act. If the ERDP gene encodes the receptor, it must recognize -HDEL. In the absence of binding data, Pelham and colleagues have exploited the fact that different salvage signals are used by different organisms. They found the salvage mechanism of a closely related yeast, Kluyveromyces lactis, recognizes both -HDEL and -DDEL, whereas S. cerevisiae is unable to salvage proteins terminating with -DDEL. However, expression of the K. lactis homolog of ERDP in S. cerevisiae changes the specificity of the salvage mechanism to that of K. lactis. Furthermore, overexpression increased the efficiency of salvage but not its specificity. The Mammalian Salvage Receptor The corresponding mammalian receptor has been identified by Vaux and his colleagues (1990) using a completely different approach. They raised monoclonal antibodies to the C-terminal salvage signals on the assumption that some of the resultant antibodies would recognize -KDEL in the same way as the salvage receptor. In other words, the antibody combining site would mimic the receptor (or at least a part of it). Antibodies to this combining site (antiidiotypic antibodies) should then recognize the salvage
Golgl Apparatus
Schematic
View of the Salvage
Pathway
Proteins leave the ER and enter the salvage compartment, where resident ER proteins are returned to the ER by salvage receptors. The morphological boundaries of the salvage compartment are not yet known, but may include the intermediate compartment (LippincottSchwartz et al., 1990) and the early Golgi stack.
receptor. Two monoclonal antibodies to different salvage signals both recognize the same protein. If this protein is the receptor, then it should be found in a compartment between the ER and the Golgi apparatus. lmmunofluorescenca studies show quite clearly that the protein is juxtanuclear, partially overlapping the Golgi pattern, but quite distinct from the pattern for the ER. Treatment with brefeldin A redistributes the Golgi stack to the ER (Lippincott-Schwartz et al., 1990) but does not alter the distribution of this protein. Removal of the drug leads to the transient appearance of the protein in the ER, as would be expected of a salvage receptor. The mammalian receptor has been tested directly for its ability to bind to -KDEL. The hybridomas secreting the anti-idiotypic antibodies protect themselves against saturation of their own receptor by secreting large quantities of a soluble form that mops up the antibody. This soluble form binds only to -KDEL salvage signals, though the efficiency is low (as expected) and the binding rather variable. The dissociation constant is in the micromolar range, the same range as the concentration of salvage signals in the ER lumen.
Comprr/son of the 7ko Receptors The receptor should be an integral membrane protein exposed to both the lumen (so as to bind the salvage signal) and the cytoplasm (to provide recognition for the mechanism that transports it back to the ER). Both the yeast and mammalian receptors are integral membrane proteins, because they resist removal by high pH and span the bilayer. In the case of the yeast receptor, this could be inferred only from the sequence, which predicts a very hydrophobic protein with three to five membrane-spanning segments. There, however, the similarity ends. The yeast receptors have a molecular size of 26 kd; the mammalian, 72 kd. It is difficult to see why the mammalian protein is nearly three times bigger than the yeast if the only functional difference is recognition of -KDEL instead of -HDEL and/or -DDEL. The mammalian protein is also isolated as disulfide-linked oligomers, which must surely be important for the protein’s function. Yet the S. cerevisiae receptor does not contain any cysteine residues. Resolution of these differences must await the sequence of the mammalian protein; but if no homology with the yeast receptor is found, what might be the reason? One intriguing possibility is that both proteins are needed to form the salvage receptor, the mammalian protein providing recognition of the last three amino acids of XDEL, and the ERD2 protein providing the recognition of X. The Recepfor Cycle A critical requirement for the salvage receptor is that it bind the tetrapeptide in the salvage compartment and release it in the ER. The necessary difference in binding affinity could be provided by a difference between the two lumens, e.g., in theconcentration of calcium ions. Notably, divalent cations are required for binding -KDEL to the mammalian receptor (Vaux et al., 1990) and calcium ionophores stimulate secretion of ER proteins (Booth and Koch, 1989). There is, however, an additional possibility based on the
likelihood that receptors will not pass easily beyond the salvage compartment and so will accumulate there. This would result in the formation of a patch of receptors, an ideal substrate for the mechanism that returns the receptors to the ER. Because the surface area of the ER is many times greater than that of the salvage compartment, delivery will be followed by breakup of the patch as it is diluted into the ER membrane. If one imagines that a receptor binds the salvage signal only as part of a patch, then binding will occur in the salvage compartment, and breakup of the patch in the ER will automatically lead to release of the salvaged proteins. Any receptors leaking from the salvage compartment should accumulate in subsequent Golgi compartments, allowing salvage from later stages of the secretory pathway. Identification of the receptor allows this and other hypotheses to be tested, but the first priority must be to determine whether the mammalian and yeast receptors are really two of a kind. References Booth, C., and Koch, G. L. (1989). Cell 59, 729-737. Dean, N., and Pelham, f-l. R. B. (1990). J. Cell Biol., in press. Lewis, M. J., Sweet, D., and Pelham, H. Ft. B. (1990). Cell x359-1383.
Lippincott-Schwarz, J., Donaldson,
67,
J. G., Schweizer, A., Berger, E., Hauri, H.-P, Yuan, L. C., and Klausner, R. D. (1990). Cell 60, 821-836. Munro, S., and Pelham, H. R. B. (1987). Cell 48, 899-907 Pelham, H. R. B. (1988). EMBO J. z 913-918. Pelham, H. Ft. 8. (1989). Annu. Rev. Cell Biol. 5, l-23. Pelham. H. R. B., Hardwick, K. 0.. and Lewis, M. J. (1989). EMBO J. Z 1757-1782. Rothman, J. E. (1989). Cell 59, 591601. Schekman, R. (1985). Annu. Rev. Cell Biol. 1, 115-143. Semenza, J. C., Hardwick, K. G., Dean, N., and Pelham, H. R. B. (1990). Cell 67, 1349-1357. Vaux, D., Twze, J., and Fuller, S. (1990). Nature 345, 495-502. Warren, G. (1987). Nature 32i: 17-18. Wieland, F. T, Gleason, M. L., Serafini, T A., and Rothman, J. E. (1987). Cell 50, 289-300.
0 1990 by Cell Press
Salvage Receptors: Two of a Kind? Graham Warren Imperial Cancer Research Lincoln’s Inn Fields London WC2A 3PX England
Fund
The lumen of the endoplasmic reticulum (ER) is densely packed with resident proteins. Many of these are involved in protein assembly (Rothman, 1989); they bind nascent polypeptides that have crossed the ER membrane and catalyze their folding. Since signals are not needed for assembled proteins to leave the ER (Wieland et al., 1987) one might imagine that any soluble protein in the lumen would be free to leave. But the resident assembly proteins, perhaps as a consequence of their function, are soluble proteins that do not leave the ER. They are retained by a C-terminal tetrapeptide salvage signal, which in mammalian cells has the sequence -KDEL. When this salvage signal is removed, the truncated protein is slowly secreted; when transplanted onto a protein that is normally secreted, that protein is retained within the ER (Munro and Pelham, 1987). Several lines of evidence argue for a catalytic mechanism (Pelham, 1989; Warren, 1987) whereby resident soluble proteins occasionally escape and are then salvaged by a receptor that recognizes the C-terminal tetrapeptide and returns them to the ER (see figure). The Salvage Pathway A key prediction is that these resident ER proteins must temporarily leave the ER and therefore be exposed to post-ER enzymes. Some evidence for this was obtained in mammalian cells (Pelham, 1988) but more convincing evidence has now been provided using the yeast Saccharomyces cerevisiae (Dean and Pelham, 1990). Prepro-a factor, a secretory protein, was fused to -HDEL, the preferred salvage signal in this yeast. The fused protein was retained within the cell, but some was modified by the addition of al-6 mannose linkages. This modification characteristic of the early Golgi did not occur in a mutant (sec78) defective in ER-to-Golgi transport at the restrictive temperature (Schekman, 1985). This showed that the protein had to leave the ER in order to be modified. Fractionation experiments localized the modified protein to the ER, so it must have returned to the ER following modifi-
Endoplasmic
Reticulum
Salvage Compartment
Minireview
The Yeast Salvage Receptor The next step was to identify the salvage receptor. S. cerevisiae will only grow on sucrose if it secretes the enzyme invertase. Transplanting -HDEL onto this protein prevents secretion and hence growth on sucrose. Mutants lacking a functional salvage receptor should then grow on sucrose. Several mutants have been characterized (Pelham et al., 1988) and one, erd;! (ER retention defective), disrupts specifically the salvage pathway. Pelham and his colleagues have now shown that the ERDP gene encodes the yeast salvage receptor (Lewis et al., 1990; Semenza et al., 1990). From the model in the figure, one would expect that overexpression of a protein fused to -HDEL could saturate the receptor. Cells overexpressing prepro-a factor fused to -HDEL were found to secrete one of their resident ER proteins. If this is the consequence of too few receptors, then overexpressing the receptor should reverse this process; the ERDP protein expressed from a multicopy plasmid did just that. In addition, the early Golgi modifications to the fusion protein were also suppressed, suggesting that the protein was being removed either from an earlier compartment or before the Golgi enzymes had time to act. If the ERDP gene encodes the receptor, it must recognize -HDEL. In the absence of binding data, Pelham and colleagues have exploited the fact that different salvage signals are used by different organisms. They found the salvage mechanism of a closely related yeast, Kluyveromyces lactis, recognizes both -HDEL and -DDEL, whereas S. cerevisiae is unable to salvage proteins terminating with -DDEL. However, expression of the K. lactis homolog of ERDP in S. cerevisiae changes the specificity of the salvage mechanism to that of K. lactis. Furthermore, overexpression increased the efficiency of salvage but not its specificity. The Mammalian Salvage Receptor The corresponding mammalian receptor has been identified by Vaux and his colleagues (1990) using a completely different approach. They raised monoclonal antibodies to the C-terminal salvage signals on the assumption that some of the resultant antibodies would recognize -KDEL in the same way as the salvage receptor. In other words, the antibody combining site would mimic the receptor (or at least a part of it). Antibodies to this combining site (antiidiotypic antibodies) should then recognize the salvage
Golgl Apparatus
Schematic
View of the Salvage
Pathway
Proteins leave the ER and enter the salvage compartment, where resident ER proteins are returned to the ER by salvage receptors. The morphological boundaries of the salvage compartment are not yet known, but may include the intermediate compartment (LippincottSchwartz et al., 1990) and the early Golgi stack.
receptor. Two monoclonal antibodies to different salvage signals both recognize the same protein. If this protein is the receptor, then it should be found in a compartment between the ER and the Golgi apparatus. lmmunofluorescenca studies show quite clearly that the protein is juxtanuclear, partially overlapping the Golgi pattern, but quite distinct from the pattern for the ER. Treatment with brefeldin A redistributes the Golgi stack to the ER (Lippincott-Schwartz et al., 1990) but does not alter the distribution of this protein. Removal of the drug leads to the transient appearance of the protein in the ER, as would be expected of a salvage receptor. The mammalian receptor has been tested directly for its ability to bind to -KDEL. The hybridomas secreting the anti-idiotypic antibodies protect themselves against saturation of their own receptor by secreting large quantities of a soluble form that mops up the antibody. This soluble form binds only to -KDEL salvage signals, though the efficiency is low (as expected) and the binding rather variable. The dissociation constant is in the micromolar range, the same range as the concentration of salvage signals in the ER lumen.
Comprr/son of the 7ko Receptors The receptor should be an integral membrane protein exposed to both the lumen (so as to bind the salvage signal) and the cytoplasm (to provide recognition for the mechanism that transports it back to the ER). Both the yeast and mammalian receptors are integral membrane proteins, because they resist removal by high pH and span the bilayer. In the case of the yeast receptor, this could be inferred only from the sequence, which predicts a very hydrophobic protein with three to five membrane-spanning segments. There, however, the similarity ends. The yeast receptors have a molecular size of 26 kd; the mammalian, 72 kd. It is difficult to see why the mammalian protein is nearly three times bigger than the yeast if the only functional difference is recognition of -KDEL instead of -HDEL and/or -DDEL. The mammalian protein is also isolated as disulfide-linked oligomers, which must surely be important for the protein’s function. Yet the S. cerevisiae receptor does not contain any cysteine residues. Resolution of these differences must await the sequence of the mammalian protein; but if no homology with the yeast receptor is found, what might be the reason? One intriguing possibility is that both proteins are needed to form the salvage receptor, the mammalian protein providing recognition of the last three amino acids of XDEL, and the ERD2 protein providing the recognition of X. The Recepfor Cycle A critical requirement for the salvage receptor is that it bind the tetrapeptide in the salvage compartment and release it in the ER. The necessary difference in binding affinity could be provided by a difference between the two lumens, e.g., in theconcentration of calcium ions. Notably, divalent cations are required for binding -KDEL to the mammalian receptor (Vaux et al., 1990) and calcium ionophores stimulate secretion of ER proteins (Booth and Koch, 1989). There is, however, an additional possibility based on the
likelihood that receptors will not pass easily beyond the salvage compartment and so will accumulate there. This would result in the formation of a patch of receptors, an ideal substrate for the mechanism that returns the receptors to the ER. Because the surface area of the ER is many times greater than that of the salvage compartment, delivery will be followed by breakup of the patch as it is diluted into the ER membrane. If one imagines that a receptor binds the salvage signal only as part of a patch, then binding will occur in the salvage compartment, and breakup of the patch in the ER will automatically lead to release of the salvaged proteins. Any receptors leaking from the salvage compartment should accumulate in subsequent Golgi compartments, allowing salvage from later stages of the secretory pathway. Identification of the receptor allows this and other hypotheses to be tested, but the first priority must be to determine whether the mammalian and yeast receptors are really two of a kind. References Booth, C., and Koch, G. L. (1989). Cell 59, 729-737. Dean, N., and Pelham, f-l. R. B. (1990). J. Cell Biol., in press. Lewis, M. J., Sweet, D., and Pelham, H. Ft. B. (1990). Cell x359-1383.
Lippincott-Schwarz, J., Donaldson,
67,
J. G., Schweizer, A., Berger, E., Hauri, H.-P, Yuan, L. C., and Klausner, R. D. (1990). Cell 60, 821-836. Munro, S., and Pelham, H. R. B. (1987). Cell 48, 899-907 Pelham, H. R. B. (1988). EMBO J. z 913-918. Pelham, H. Ft. 8. (1989). Annu. Rev. Cell Biol. 5, l-23. Pelham. H. R. B., Hardwick, K. 0.. and Lewis, M. J. (1989). EMBO J. Z 1757-1782. Rothman, J. E. (1989). Cell 59, 591601. Schekman, R. (1985). Annu. Rev. Cell Biol. 1, 115-143. Semenza, J. C., Hardwick, K. G., Dean, N., and Pelham, H. R. B. (1990). Cell 67, 1349-1357. Vaux, D., Twze, J., and Fuller, S. (1990). Nature 345, 495-502. Warren, G. (1987). Nature 32i: 17-18. Wieland, F. T, Gleason, M. L., Serafini, T A., and Rothman, J. E. (1987). Cell 50, 289-300.
