Cell, Vol. 63, 261-291,
October
19, 1990, Copyright
0 1990 by Cell Press
Generation of a Lysosomal Enzyme Targeting Signal in the Secretory Protein Pepsinogen Thomas J. Baranski,” Phyllis L. Faust:t and Stuart Kornfeld’ Departments of Internal Medicine and Biological Chemistry Washington University School of Medicine St. Louis, Missouri 63110 l
Summary Lysosomal enzymes contain a common protein determinant that is recognized by UDP-GlcNAc:lysosomal enzyme N-acetylglucosamine-1-phosphotransferase, the initial enzyme in the formation of mannose 6-phosphate residues. To identify this protein determinant, we constructed chimeric molecules between two asparty1 proteases: cathepsin D, a lysosomal enzyme, and pepsinogen, a secretory protein. When expressed in Xenopus oocytes, the oligosaccharides of cathepsin D were efficiently phosphorylated, whereas the oligosaccharides of a glycosylated form of pepsinogen were not phosphorylated. The combined substitution of two noncontinuous sequences of cathepsin D (lysine 203 and amino acids 265-292) into the analogous positions of glycopepsinogen resulted in phosphorylation of the oligosaccharides of the expressed chimerlc molecule. These two sequences are in direct apposition on the surface of the molecule, indicating that amino acids from different regions come together in three-dimensional space to form this recognition domain. Other regions of cathepsin D were identified that may be components of a more extensive recognition marker. Introduction In many cell types, the acquisition of mannose &phosphate (Man-6-P) residues by newly synthesized acid hydrolases is required for their efficient targeting to lysosomes. These phosphomannosyl residues are synthesized by the concerted action of two enzymes (for reviews, see Kornfeld, 1986; von Figura and Hasilik, 1986). First, UDP-N-acetylglucosamine (UDP-GlcNAc):lysosomal enzyme N-acetylglucosamine-1-phosphotransferase (abbreviated phosphotransferase) transfers N-acetylglucosamine l-phosphate from UDP-GlcNAc to selected mannose residues on the lysosomal enzymes to give rise to phosphodiester intermediates. Then, the N-acetylglucosamine residues are removed by N-acetylglucosamine-l-phosphodiester-a-N-acetylglucosaminidase to generate Man-6-P monoesters, which serve as high affinity ligands for the binding of the hydrolases to Man-6-P receptors in the Golgi and subsequent translocation to the lysosome. The phosphotransferase is the critical enzyme in this 7 Present address: Department School of Medicine, St. Louis,
of Pathology, Washington Missouri 63110.
University
sorting system. It selectively phosphorylates the high mannose oligosaccharides of lysosomal enzymes over similar high mannose oligosaccharides of nonlysosomal glycoproteins (Reitman and Kornfeld, 1961). In addition, deglycosylated lysosomal enzymes have been shown to be specific and potent inhibitors of the phosphorylation of intact lysosomal enzymes (Lang et al., 1964). This has led to the proposal that phosphotransferase recognizes a protein domain that is common to all lysosomal enzymes but absent in nonlysosomal glycoproteins. The identification of this recognition domain has been elusive for several reasons. First, the domain does not appear to be a simple linear sequence of amino acids, since the numerous lysosomal enzymes that have been cloned to date do not share any significant sequence similarity. Furthermore, the use of standard biochemical approaches to identify the recognition marker have been unsuccessful, because procedures that alter the conformational integrity of the lysosomal enzyme, such as proteolytic fragmentation, also disrupt the recognition domain (Lang et al., 1964). To gain insight into the nature of the phosphotransferase recognition domain, we constructed chimeric proteins between human pepsinogen, a secretory protein, and human cathepsin D, a lysosomal hydrolase. The goal was to determine which residues of cathepsin D sequence could generate a functional phosphotransferase recognition marker when substituted into pepsinogen. Cathepsin D and pepsinogen were chosen for this study because they are both members of the aspartyl protease family and are 45% identical in amino acid sequence. Seven members of the family, including porcine pepsinogen, have been crystallized and their structures determined (for review, see Davies, 1990). These studies have revealed that the overall secondary and tertiary structures of these bilobed molecules are remarkably conserved from the fungal to the mammalian aspartyl proteases. Therefore, it seemed reasonable that regions of cathepsin D sequence could be substituted into the backbone of pepsinogen while maintaining their native conformation. Our experiments have defined two noncontinuous sequences of cathepsin D that, when substituted into pepsinogen, generate a recognition domain for phosphotransferase. Furthermore, a single lysine residue in one of the regions has been identified as a critical component of the recognition marker. Results Glycopepsinogen Is Not Phosphorylated Prior to making chimeric molecules between pepsinogen and cathepsin D, it was necessary to establish that pepsinogen lacks the protein domain recognized by the phosphotransferase. Because human pepsinogen is not glycosylated, it was possible that it contained an occult phosphotransferase recognition domain. To address this issue, potential N-linked glycosylation
Cell 282
-40
-20
-30
Figure 1. Alignment of Amino Acid Sequences of Human Cathepsin D and Human Pepsinogen The positions of the restriction enzyme sites used to prepare the constructs are indicated. The numbers of the residues are designated with respect to cathepsin D. N-linked glycosylation sites are indicated by the letters CHO. Negative numbers indicate the propeptide region. The active site aspartyl residues are marked by a dot. The identical amino acids between cathepsin D and pepsinogen are boxed, and dashes indicate insertion of gaps at positions at which there are no homologous residues.
CD PG
CD PG
CD PG
100
110
CD PG
CD PG
CD PG
CD PG 240
250
GC~EAIVDTGTSLNVGPVDEVRELQKA )~]LT~TSPSANI~SD
CD PG
260
POUII
270
280
CD PG
CD PG
BaPPEI CD PG
330
340
P[LWILGDVFI]GR[Y~[D~~F~EA~RL ELWILGDVFIRQYFTVFDRANNQVGLAPVA--
sites were inserted by site-directed mutagenesis into the pepsinogen cDNA at the identical positions as in cathepsin D (Figure 1). Three mutant pepsinogens were created: first, mPep1, which contains a single glycosylation site introduced at position 70 (cathepsin D numbering [Faust et al., 19851 is used throughout) by changing threonine 70 and glutamic acid 72 to an asparagine and threonine, respectively; second, mPep2, which contains a single glycosylation site introduced at position 199 by changing proline 199 to asparagine; and third, mPepl2, which contains both of the above glycosylation sites introduced into pepsinogen. Xenopus oocytes were injected with mRNA generated by in vitro transcription of the various cDNAs, and the fate of the expressed proteins was determined. After incubation of the injected oocytes in [35S]methionine-labeling medium for 70 hr (note that incubation past 20 hr constitutes a chase in this system [Colman, 19843) 51%-59% of the various mPep proteins were secreted into the medium. To demonstrate that the N-linked glycosylation
sites were efficiently utilized, aliquots of the immunoprecipitates were treated with endo H, which specifically removes high mannose-type oligosaccharides. The cell-as sociated material for mPep1, mPep2, and mPep12 showed increased mobility on SDS-PAGE, indicating that the proteins were glycosylated and that the oligosaccharides were of the high mannose type (Figure 2, lanes 2, 8, and 10). The observed M, 1500 decrease for mBp1 and mPep2 is consistent with the removal of one high mannose unit, and the M, 3000 decrease for mPepl2 is consistent with the removal of two high mannose units. The secreted material, however, was completely resistant to endo H digestion in all three instances, indicating that the oligosaccharides had been processed to complex-type units (Figure 2, lanes 4, 8, and 12). This would account for the decreased mobility of the secreted glycopepsinogens on SDS-PAGE. Since previous studies have established that only high mannose-type oligosaccharides on lysosomal enzymes are phosphorylated, the presence of complextype oligosaccharides in the secreted proteins indicated
Phosphotransferase 263
Recognition
Marker
oocytes, and the oligosaccharides were analyzed. As summarized in Table 1, the glycopepsinogen in the medium contained 99% complex-type oligosaccharides and the oocyte material contained 80% high mannose units and 20% complex-type units. No phosphorylated species were detected. Based on the total [3H]mannose incorporated into oligosaccharides, we estimate that glycopepsinogen contained
October
19, 1990, Copyright
0 1990 by Cell Press
Generation of a Lysosomal Enzyme Targeting Signal in the Secretory Protein Pepsinogen Thomas J. Baranski,” Phyllis L. Faust:t and Stuart Kornfeld’ Departments of Internal Medicine and Biological Chemistry Washington University School of Medicine St. Louis, Missouri 63110 l
Summary Lysosomal enzymes contain a common protein determinant that is recognized by UDP-GlcNAc:lysosomal enzyme N-acetylglucosamine-1-phosphotransferase, the initial enzyme in the formation of mannose 6-phosphate residues. To identify this protein determinant, we constructed chimeric molecules between two asparty1 proteases: cathepsin D, a lysosomal enzyme, and pepsinogen, a secretory protein. When expressed in Xenopus oocytes, the oligosaccharides of cathepsin D were efficiently phosphorylated, whereas the oligosaccharides of a glycosylated form of pepsinogen were not phosphorylated. The combined substitution of two noncontinuous sequences of cathepsin D (lysine 203 and amino acids 265-292) into the analogous positions of glycopepsinogen resulted in phosphorylation of the oligosaccharides of the expressed chimerlc molecule. These two sequences are in direct apposition on the surface of the molecule, indicating that amino acids from different regions come together in three-dimensional space to form this recognition domain. Other regions of cathepsin D were identified that may be components of a more extensive recognition marker. Introduction In many cell types, the acquisition of mannose &phosphate (Man-6-P) residues by newly synthesized acid hydrolases is required for their efficient targeting to lysosomes. These phosphomannosyl residues are synthesized by the concerted action of two enzymes (for reviews, see Kornfeld, 1986; von Figura and Hasilik, 1986). First, UDP-N-acetylglucosamine (UDP-GlcNAc):lysosomal enzyme N-acetylglucosamine-1-phosphotransferase (abbreviated phosphotransferase) transfers N-acetylglucosamine l-phosphate from UDP-GlcNAc to selected mannose residues on the lysosomal enzymes to give rise to phosphodiester intermediates. Then, the N-acetylglucosamine residues are removed by N-acetylglucosamine-l-phosphodiester-a-N-acetylglucosaminidase to generate Man-6-P monoesters, which serve as high affinity ligands for the binding of the hydrolases to Man-6-P receptors in the Golgi and subsequent translocation to the lysosome. The phosphotransferase is the critical enzyme in this 7 Present address: Department School of Medicine, St. Louis,
of Pathology, Washington Missouri 63110.
University
sorting system. It selectively phosphorylates the high mannose oligosaccharides of lysosomal enzymes over similar high mannose oligosaccharides of nonlysosomal glycoproteins (Reitman and Kornfeld, 1961). In addition, deglycosylated lysosomal enzymes have been shown to be specific and potent inhibitors of the phosphorylation of intact lysosomal enzymes (Lang et al., 1964). This has led to the proposal that phosphotransferase recognizes a protein domain that is common to all lysosomal enzymes but absent in nonlysosomal glycoproteins. The identification of this recognition domain has been elusive for several reasons. First, the domain does not appear to be a simple linear sequence of amino acids, since the numerous lysosomal enzymes that have been cloned to date do not share any significant sequence similarity. Furthermore, the use of standard biochemical approaches to identify the recognition marker have been unsuccessful, because procedures that alter the conformational integrity of the lysosomal enzyme, such as proteolytic fragmentation, also disrupt the recognition domain (Lang et al., 1964). To gain insight into the nature of the phosphotransferase recognition domain, we constructed chimeric proteins between human pepsinogen, a secretory protein, and human cathepsin D, a lysosomal hydrolase. The goal was to determine which residues of cathepsin D sequence could generate a functional phosphotransferase recognition marker when substituted into pepsinogen. Cathepsin D and pepsinogen were chosen for this study because they are both members of the aspartyl protease family and are 45% identical in amino acid sequence. Seven members of the family, including porcine pepsinogen, have been crystallized and their structures determined (for review, see Davies, 1990). These studies have revealed that the overall secondary and tertiary structures of these bilobed molecules are remarkably conserved from the fungal to the mammalian aspartyl proteases. Therefore, it seemed reasonable that regions of cathepsin D sequence could be substituted into the backbone of pepsinogen while maintaining their native conformation. Our experiments have defined two noncontinuous sequences of cathepsin D that, when substituted into pepsinogen, generate a recognition domain for phosphotransferase. Furthermore, a single lysine residue in one of the regions has been identified as a critical component of the recognition marker. Results Glycopepsinogen Is Not Phosphorylated Prior to making chimeric molecules between pepsinogen and cathepsin D, it was necessary to establish that pepsinogen lacks the protein domain recognized by the phosphotransferase. Because human pepsinogen is not glycosylated, it was possible that it contained an occult phosphotransferase recognition domain. To address this issue, potential N-linked glycosylation
Cell 282
-40
-20
-30
Figure 1. Alignment of Amino Acid Sequences of Human Cathepsin D and Human Pepsinogen The positions of the restriction enzyme sites used to prepare the constructs are indicated. The numbers of the residues are designated with respect to cathepsin D. N-linked glycosylation sites are indicated by the letters CHO. Negative numbers indicate the propeptide region. The active site aspartyl residues are marked by a dot. The identical amino acids between cathepsin D and pepsinogen are boxed, and dashes indicate insertion of gaps at positions at which there are no homologous residues.
CD PG
CD PG
CD PG
100
110
CD PG
CD PG
CD PG
CD PG 240
250
GC~EAIVDTGTSLNVGPVDEVRELQKA )~]LT~TSPSANI~SD
CD PG
260
POUII
270
280
CD PG
CD PG
BaPPEI CD PG
330
340
P[LWILGDVFI]GR[Y~[D~~F~EA~RL ELWILGDVFIRQYFTVFDRANNQVGLAPVA--
sites were inserted by site-directed mutagenesis into the pepsinogen cDNA at the identical positions as in cathepsin D (Figure 1). Three mutant pepsinogens were created: first, mPep1, which contains a single glycosylation site introduced at position 70 (cathepsin D numbering [Faust et al., 19851 is used throughout) by changing threonine 70 and glutamic acid 72 to an asparagine and threonine, respectively; second, mPep2, which contains a single glycosylation site introduced at position 199 by changing proline 199 to asparagine; and third, mPepl2, which contains both of the above glycosylation sites introduced into pepsinogen. Xenopus oocytes were injected with mRNA generated by in vitro transcription of the various cDNAs, and the fate of the expressed proteins was determined. After incubation of the injected oocytes in [35S]methionine-labeling medium for 70 hr (note that incubation past 20 hr constitutes a chase in this system [Colman, 19843) 51%-59% of the various mPep proteins were secreted into the medium. To demonstrate that the N-linked glycosylation
sites were efficiently utilized, aliquots of the immunoprecipitates were treated with endo H, which specifically removes high mannose-type oligosaccharides. The cell-as sociated material for mPep1, mPep2, and mPep12 showed increased mobility on SDS-PAGE, indicating that the proteins were glycosylated and that the oligosaccharides were of the high mannose type (Figure 2, lanes 2, 8, and 10). The observed M, 1500 decrease for mBp1 and mPep2 is consistent with the removal of one high mannose unit, and the M, 3000 decrease for mPepl2 is consistent with the removal of two high mannose units. The secreted material, however, was completely resistant to endo H digestion in all three instances, indicating that the oligosaccharides had been processed to complex-type units (Figure 2, lanes 4, 8, and 12). This would account for the decreased mobility of the secreted glycopepsinogens on SDS-PAGE. Since previous studies have established that only high mannose-type oligosaccharides on lysosomal enzymes are phosphorylated, the presence of complextype oligosaccharides in the secreted proteins indicated
Phosphotransferase 263
Recognition
Marker
oocytes, and the oligosaccharides were analyzed. As summarized in Table 1, the glycopepsinogen in the medium contained 99% complex-type oligosaccharides and the oocyte material contained 80% high mannose units and 20% complex-type units. No phosphorylated species were detected. Based on the total [3H]mannose incorporated into oligosaccharides, we estimate that glycopepsinogen contained
