Mannose 6-Phosphate Receptors and Lysosomal Enzyme Targeting

The targeting of lysosomal enzymes from their site of synthesis in the rough endoplasmic reticulum (RER) to their final destination in lysosomes is a multi-step process requiring a series of interactions between cellular components and protein and carbohydrate recognition signals present on the lysosomal enzymes (1-6). These proteins share a common pathway with secretory proteins and membrane proteins during the early stages of their biosynthesis. All three classes of proteins contain a hydrophobic signal sequence that allows for their synthesis on membrane-bound polysomes in the RER and translocation into t.he lumen of this organelle. During this process the lysosomal enzymes as well as many secretory and membrane proteins are co-translationally glycosylated at selected asparagine residues. Following cleavage of the signal sequence and initial proc-essing of asparagine-linked oligosaccharides, the proteins move by vesicular transport from the RER to the Golgi apparatus where they undergo a variety of post-translational modifications and are segre- gated from one another for targeting to their final destinations (7). A key step in the sorting process is the generation of phospho- mannosyl residues on the lysosomal enzymes. The phosphorylating enzyme recognizes a protein determinant shared by lysosomal en- zymes, thereby selectively marking this class of proteins for subse-quent segregation. The phosphomannosyl residues serve as high affinity ligands for binding to mannose 6-phosphate receptors (MPRs)

The targeting of lysosomal enzymes from their site of synthesis in the rough endoplasmic reticulum (RER) to their final destination in lysosomes is a multi-step process requiring a series of interactions between cellular components and protein and carbohydrate recognition signals present on the lysosomal enzymes (1)(2)(3)(4)(5)(6). These proteins share a common pathway with secretory proteins and membrane proteins during the early stages of their biosynthesis. All three classes of proteins contain a hydrophobic signal sequence that allows for their synthesis on membrane-bound polysomes in the RER and translocation into t.he lumen of this organelle. During this process the lysosomal enzymes as well as many secretory and membrane proteins are co-translationally glycosylated a t selected asparagine residues. Following cleavage of the signal sequence and initial processing of asparagine-linked oligosaccharides, the proteins move by vesicular transport from the RER to the Golgi apparatus where they undergo a variety of post-translational modifications and are segregated from one another for targeting to their final destinations (7).
A key step in the sorting process is the generation of phosphomannosyl residues on the lysosomal enzymes. The phosphorylating enzyme recognizes a protein determinant shared by lysosomal enzymes, thereby selectively marking this class of proteins for subsequent segregation. The phosphomannosyl residues serve as high affinity ligands for binding to mannose 6-phosphate receptors (MPRs) in the Golgi. In this way the lysosomal enzymes are physically separated from proteins destined for secretion. The ligand-receptor complex then exits the Golgi via a coated vesicle and is delivered to a prelysosomal acidified compartment where dissociation of the ligand occurs. The released lysosomal enzyme is packaged into a lysosome while the receptor either returns to the Golgi to repeat the process or moves to the plasma membrane where it functions to internalize exogenous lysosomal enzymes. Recent work has focused on the MPRs. Two different MPRs have been identified, characterized, and their cDNAs cloned. The routing of the receptors has been studied, and the determinants on the receptor that mediate trafficking are beginning to be defined. Some of the cellular components that interact with the receptors as they move from one compartment to the next have been identified. This review will summarize our current understanding of the MPRs and their biological functions.

Receptor Structure
The first MPR to be characterized was a membrane-associated glycoprotein with an apparent M, of 215,000. This receptor binds ligand independent of divalent cations. The other MPR is also a membrane-associated glycoprotein, but it has an apparent M, of 46,000 and requires divalent cations for optimal ligand binding. Both receptors show similar, but not identical, binding specificities toward phosphorylated oligosaccharides (8,9). Based on their divalent cation requirements, we refer to the larger receptor as the cationindependent (CI) MPR and the smaller receptor as the cationdependent (CD) MPR. The cloning of cDNAs for the CI-MPR from bovine (lo), human (11,12), and rat sources (13) and cDNAs for the CD-MPR from bovine (14) and human sources (15) has provided insights into the relationship between these two different proteins.
Sequence analyses, combined with proteolysis experiments of the receptor in membranes (16,17), indicate that the bovine CI-MPR precursor consists of a 44-residue amino-terminal signal sequence, a 2269-residue extracytoplasmic domain, a single 23-residue transmembrane region, and a 163-residue carboxyl-terminal cytoplasmic domain (Fig. l ) . T h e extracytoplasmic domain has 19 potential glycosylation sites, and a t least two are utilized. Therefore, the size of the mature glycoprotein is likely to be between 275 and 300 kDa. The extracytoplasmic domain has a highly repetitive structure consisting of 15 contiguous units that have a n average length of 147 amino acids. When the repeating units are compared to each other, numerous sequence identities are seen with the percent of identical residues ranging from 16 to 38%. In addition, there are numerous regions of conservatively substituted amino acids and a characteristic spacing of cysteine residues. The receptor is known to be phosphorylated (18,19), and analysis of the cytoplasmic domain reveals sequences that are potential substrates for various protein kinases including protein kinase C and casein kinases I and I1 (13).
The predicted structure of the bovine CD-MPR consists of a 28residue amino-terminal signal sequence, a 159-residue extracytoplasmic domain, a single 25-residue transmembrane region, and a 67residue carboxyl-terminal cytoplasmic domain ( Fig. 1). This receptor has five potential asparagine-linked glycosylation sites, four of which are utilized (14).
A comparison of the sequences of the two receptors reveals that they are related. The entire extracytoplasmic domain of the CD-MPR is similar to each of the repeating units of the CI-MPR, with sequence identity ranging from 14 to 28%. Thus, the CD-MPR is almost as similar to the different repeating units of the CI-MPR as the repeating units are to each other. This similarity suggests that the two receptors share a common ancestry and that the CI-MPR arose from duplication of a single ancestral sequence. In contrast to these homologies, there are no significant primary sequence similarities between their signal sequences, transmembrane regions, or their cytoplasmic domains. However, the cytoplasmic domains of both receptors contain potential serine phosphorylation sites and clusters of acidic residues that are also found on other recycling receptors (20).
Chemical cross-linking experiments indicate that the CD-MPR is a dimer in the membrane (21,22) and either a dimer or a tetramer in solution (23). The quaternary structure of the GI-MPR has not been analyzed in great detail, but hydrodynamic measurements are consistent with it being a monomer (24) while chemical cross-linking experiments indicate that it may be an oligomer (21). Ligand binding studies have revealed that the CD-MPR binds 1 mol of the monovalent ligand Man-6-P and 0.5 mol of a divalent phosphorylated oligosaccharide/monomeric subunit (9). Therefore each functional dimer would have two Man-6-P binding sites, both of which can be occupied by a single oligosaccharide containing two Man-6-P residues. Evidence that each polypeptide monomer can fold into an independent ligand binding unit has been obtained by demonstrating that a truncated form of the bovine CD-MPR, which behaves as a soluble monomer in solution, is capable of binding Man-6-P (22). The CI-MPR, on the other hand, binds 2 mol of Man-6-P or 1 mol of a divalent phosphorylated oligosaccharide/monomer (8). This may indicate that only two of the 15 repeating segments of this receptor function in the binding of Man-6-P. While the identity of the binding segments has not yet been established, two different proteolytic fragments of the CI-MPR encompassing repeating units 1-3 and 7-10 have been shown to bind a phosphorylated lysosomal enzyme.' Thus the two functional Man-6-P binding domains are probably contained within these regions of the receptor.

Role of the Receptors in Sorting and Endocytosis
Two complementary experimental approaches indicate that the CI-MPR functions both in the sorting of newly synthesized lysosomal enzymes and in endocytosis of extracellular phosphorylated lysosomal enzymes. First, cultured cells that either lack endogenous CI-MPR (25) or are depleted of CI-MPR by treatment with anti-CI-MPR antiserum (26,27) secrete -70% of their newly synthesized lysosomal enzymes and do not endocytose extracellular phosphorylated lysosomal enzymes. Second, the defective sorting and endocytosis pheno-' B. Wevtlund and S. Kornfeld type of the cells that lack endogenous CI-MPR can be corrected by transfection with a CI-MPR cDNA (28, 29). The residual sorting found in the CI-MPR defective cells appears to be mediated by the CD-MPR. Supporting this, when such cells are treated with an anti-CD-MPR antiserum, secretion of lysosomal enzymes is increased (27). These results indicate that both receptors function in lysosomal enzyme sorting, with the CI-MPR handling most of the traffic. One clear difference between the two receptors relates to the endocytosis of extracellular lysosomal enzymes. While both receptors cycle to the plasma membrane (30), only the CI-MPR is capable of binding and internalizing lysosomal enzymes (27). This difference, however, does not resolve the fundamental question of why two separate MPRs exist.

The CI-MPR and the IGF-11 Receptor Are the Same Protein
When Morgan et al. (11) cloned and sequenced the human insulinlike growth factor I1 (IGF-11) receptor, they made the surprising MPR. The identity of the CI-MPR and the IGF-I1 receptor has been discovery that its sequence corresponds to that of the bovine CIconfirmed by biochemical studies which show that this protein can bind phosphomannosyl residues and IGF-11, a nonglycosylated polypeptide hormone, simultaneously (31,32). However, each ligand may influence the binding of the other ligand to the receptor (13,33,34). The stoichiometry of IGF-I1 binding to the CI-MPR has been determined to be 1 mol of IGF-II/polypeptide chain (31). In contrast, the CD-MPR does not bind IGF-I1 (31,33). The biological significance of the finding that the CI-MPR and the IGF-I1 receptor are the same protein is still uncertain. The CI-MPR is known to bind and internalize IGF-I1 at the cell surface, resulting in the degradation of this ligand (35). In this manner it may serve to clear IGF-I1 from the circulation. One critical question is whether IGF-I1 binding to the CI-MPR results in signal transduction. This has been difficult to determine since IGF-I1 also binds to the IGF-I receptor, a member of the tyrosine kinase family of receptors known to transmit signals across the plasma membrane. There are several reports suggesting that IGF-I1 mediates a response through its own receptor (5) including one in which IGF-I1 promoted cell proliferation in cells that lack the IGF-I receptor but contain the CI-MPR (36). In addition, IGF-I1 stimulated production of inositol trisphosphate in membranes from renal proximal tubular cells, and this effect was potentiated by Man-6-P (37). In these instances the CI-MPR may be transducing a signal upon IGF-I1 binding, perhaps via coupling to a pertussis toxin-sensitive G protein (38, 39). Another possibility is that IGF-I1 may modulate the trafficking of lysosomal enzymes. This could be important for processes such as tissue and bone remodeling during development where it is necessary to secrete lysosomal en-zymes. In these instances, IGF-I1 might act by inhibiting lysosomal enzyme binding to cell surface receptors thereby preventing the recapture of secreted lysosomal enzymes (34) or by altering the intracellular trafficking and distribution of the CI-MPR, possibly resulting in less efficient sorting in the Golgi (39). Interestingly, the chicken CI-MPR does not bind either human or chicken IGF-I1 (40). Nevertheless, chicken fibroblasts are highly responsive to IGF-11, possibly via binding to the IGF-I receptor (41).

Other Growth Factors Bind to the MPR
Recently a number of secreted glycoproteins has been shown to contain the Man-6-P recognition marker (42-47). Presumably these ligands are secreted because they have a low affinity for the MPR and do not effectively compete with the bulk of lysosomal enzymes at the intracellular sorting site (48). Three of these proteins have been identified as lysosomal enzymes that are secreted only under special circumstances (42-44). Another is porcine thyroglobulin which is secreted by thyroid follicle cells and then recaptured for degradation in lysosomes (45). The other proteins appear to be growth factors with no known lysosomal enzyme activity. Proliferin is a prolactin-related protein postulated to be an autocrine growth factor while transforming growth factor-81 precursor is the proform of a hormone that has multiple effects on cell growth and differentiation (46,47). Both of these proteins can bind to the CI-MPR at the cell surface via their Man-6-P moieties. Their internalization could result in activation in endosomes or degradation in lysosomes. These findings indicate that the phosphomannosyl recognition system may have a broader biologic role than previously recognized.

Receptor Trafficking
Lysosomal enzymes can be targeted to the lysosome by either one of two pathways: a direct intracellular route ("biosynthetic" pathway) or an endocytic pathway, with the former being the major pathway (Fig. 2). In the biosynthetic pathway, the formation of the active phosphomannosyl monoester on lysosomal enzymes occurs in the cis (early) Golgi compartment (1,49). This raised the possibility that lysosomal enzymes might bind to the CI-MPR in the early Golgi and either pass through the Golgi as a complex or exit the Golgi at this point. Indeed, data consistent with the latter possibility have come from immunocytochemical studies demonstrating that in some, but not all, cells the CI-MPR is concentrated in the cis Golgi with very low levels in the trans (late) Golgi (50). However, sorting in most cells has been postulated to occur in a late Golgi compartment, based on the following observations. A number of lysosomal enzymes have been shown to contain terminally processed oligosaccharides (51,52), The enzymes are then packaged into lysosomes while the receptors cycle back to the Golgi or to the plasma membrane. In addition to the pathways shown, it is likely that MPRs also cycle to the plasma membrane from the Golgi (via leak into secretory vesicles) and from late endosomes.
indicating that these enzymes have traversed the entire Golgi complex since the glycosyltransferases responsible for terminal glycosylation reside in the trans Golgi elements (53). In addition, studies of the kinetics of receptor trafficking have demonstrated that the MPRs return to the last Golgi compartment, the trans Golgi network, much more frequently than they cycle to the early Golgi compartments (30, 54). Furthermore, CI-MPRs and lysosomal enzymes have been localized to clathrin-coated vesicles in the region of the trans Golgi network (55,56). Taken together, these data indicate that lysosomal enzymes are sorted from other classes of proteins in the trans Golgi network (7). Immunolocalization studies and biochemical analyses reveal very low or undetectable amounts of the CI-MPR in lysosomes while, in contrast, a significant amount of the receptor is found in endosomal structures (18,55,(57)(58)(59). This has led to the concept that Golgiderived vesicles containing lysosomal enzyme-receptor complexes are delivered to acidic prelysosomal/endosomal compartments rather than to lysosomes (57-60). The low pH of the endosomal compartment would cause the complex to dissociate and the released lysosomal enzymes could be packaged into lysosomes while the CI-MPRs could recycle out of this compartment. (The variation in the pH of the different compartments assures the proper vectorial transport of ligand. The receptor binds ligand at neutral pH and discharges ligand at acidic pH; the Golgi is near neutrality while the endosome is acidic. Consequently, the receptor will bind lysosomal enzymes in the Golgi and discharge them in endosomes.) Griffiths et al. (57) have described an acidic late endosomal structure in normal rat kidney cells which may serve as a site where lysosomal enzymes dissociate from the CI-MPRs. This compartment is located in the vicinity of the Golgi complex, contains lysosomal enzymes as well as a lysosomal membrane protein (lgp120), and is enriched in CI-MPRs. This structure is distinct from the trans Golgi network and early endosomes. Studies by Geuze et al. (58) also have provided evidence that the CI-MPR is segregated from lgp120 and presumably from lysosomal enzymes in prelysosomal/endosomal structures. However, it is currently unknown if the newly synthesized lysosomal enzymes are delivered to early endosomes, late endosomes, or both types of endosomes. The finding of CI-MPRs in early endosomes does not resolve this issue since these molecules may be derived from the plasma membrane.
Extracellular lysosomal enzymes may also be delivered to the lysosome via the endocytic pathway. A small proportion of lysosomal enzymes is typically secreted by cells (1,2). Some of these enzymes may bind to cell surface CI-MPRs and be internalized via clathrincoated pits and vesicles (55, 61). The internalized acid hydrolases enter acidified endosomal compartments where they dissociate from the CI-MPRs and are delivered to lysosomes. Measurements of the number and half-life of MPRs and the rate of ligand internalization indicate that the MPRs are reutilized and can undergo many rounds of ligand delivery (62). In addition, studies using antibodies (26,60,63) or galactosyltransferase (30) to label receptors on the cell surface indicate that all of the MPRs in the cell are in rapid equilibrium. This suggests that there is only one pool of receptor and that a single CI-MPR functions in both the endocytic and biosynthetic pathways (60).

Signals for Receptor Sorting and Endocytosis
One model for trafficking of the CI-MPR is that the receptor contains multiple signals that mediate its departure from and arrival at three different destinations: the Golgi, plasma membrane, and endosomes. The cloning of the cDNA encoding the CI-MPR has provided a tool that is proving to be useful in the characterization of these signals. One approach that has been taken is to transfect CI-MPR-deficient cells with either normal CI-MPR cDNA or cDNAs containing mutations in the cytoplasic domain and then assay for lysosomal enzyme sorting and endocytosis (29, 64). The normal receptor functioned well in the sorting of newly synthesized lysosomal enzymes and in the endocytosis of exogenous lysosomal enzymes (28,29). Mutant receptors with either 40 or 89 residues deleted from the carboxyl terminus of the 163-amino acid cytoplasmic tail mediated endocytosis as well as the wild-type receptor but were significantly impaired in sorting (29). This result shows that one aspect of receptor trafficking (sorting) can be perturbed without, affecting the other aspect (endocytosis) and suggests that the signals for return to or departure from the Golgi are disrupted in these mutants. When the cytoplasmic tail was truncated to contain 7 or 20 residues, the resultant receptors accumulated on the cell surface and were thus defective in both endocytosis and sorting. These receptors are pre-sumably unable to enter clathrin-coated pits efficiently. A mutant receptor containing a cytoplasmic tail with alanine residues at positions 24 and 26 instead of the normal tyrosine residues was found to be defective in endocytosis, thereby demonstrating that these two tyrosine residues are required for rapid internalization of the CI-MPR.
The importance of tyrosine residues in the cytoplasmic tails of proteins involved in receptor-mediated endocytosis is emerging as a general theme (65). The first evidence for this came from studies on the LDL receptor (66). When a single tyrosine was changed to other charged or uncharged aliphatic residues, the resultant mutant LDL receptor was unable to enter coated pits efficiently. In contrast, substitution of the tyrosine with other aromatic amino acids resulted in a functional receptor, thus ruling out tyrosine phosphorylation as the signal for internalization. The introduction of a proline residue next to the tyrosine significantly decreased the efficiency of internalization, suggesting that the environment surrounding the tyrosine residue plays a role in the internalization of the LDL receptor. This is also corroborated by studies on a mutant influenza virus hemagglutinin. The introduction of a single tyrosine residue into the cytoplasmic tail of the hemagglutinin caused this protein, which is normally excluded from coated pits, to be rapidly internalized via clathrin-coated pits (67). The location of the tyrosine residue in this short (10-amino acid) cytoplasmic domain is critical since placement of tyrosines at two other positions did not result in rapid endocytosis of hemagglutinin. Inspection of the sequences of the recycling receptors in the region of their tyrosine residues reveals that while the surrounding residues are quite hydrophilic, no simple pattern of sequence identity is apparent. Further work is needed to define what constitutes the internalization signal.
How may specific amino acid sequences in the cytoplasmic domain of the CI-MPR mediate its routing? One possibility is that cellular components recognize these signals and direct the receptor to its proper destination. Recent studies have begun to identify these components. Clathrin-coated pits in the plasma membrane contain a complex of 100.50.16-kDa polypeptides (65). These polypeptides, which have been termed "adaptors," bind clathrin and also interact with endocytic receptors (68). 69) have shown that the plasma membrane-derived adaptor proteins interact with the cytoplasmic tails of the CI-MPR, LDL receptor, and the poly-Ig receptor, but not with a mutant CI-MPR cytoplasmic tail that lacks the two tyrosine residues at positions 24 and 26. These results indicate that tyrosine residues in the cytoplasmic tails of these endocytic receptors are necessary for their interaction with the plasma membrane-derived adaptor proteins and suggest that this interaction is what mediates the entry of selected receptors into clathrin-coated pits. A different set of adaptor polypeptides (100.47. 19-kDa complex) has been isolated from clathrin-coated pits in the Golgi region (65). The Golgi-derived adaptor proteins bind to the cytoplasmic tail of the CI-MPR but not to that of the LDL receptor (68,69). Furthermore, the mutant CI-MPR tail that lacked tyrosines still interacted with the Golgi-derived adaptor proteins. Thus, both classes of adaptor proteins are likely to be involved in the routing of the CI-MPR, with one set directing departure from the plasma membrane and the other set directing departure from the Golgi.
The identification of these two adaptor protein complexes, each of which is located at a unique site along the receptor's targeting pathway and interacts with distinct signals on the receptor, provides a first step toward understanding the mechanism by which the CI-MPR is routed. In addition, the selective recycling of the CI-MPR to the trans Golgi network has been reconstituted in vitro (70). The properties of receptor movement in this cell-free system indicate a vesicular transport mechanism. The ability to reconstitute in vitro this portion of the CI-MPRs intracellular pathway should aid in the identification of the cellular components involved in this targeting process. Furthermore, the recent isolation of endosomal structures enriched in MPRs should be useful in the characterization of other molecules involved in routing the receptors (71).

Potential Regulators o f Receptor Trafficking
Several studies have analyzed the effect of ligand binding on CI-MPR trafficking. Alterations in the steady state distribution of the CI-MPR have been found in some (72,73) but not all (74, 75) studies where lysosomal enzyme synthesis is inhibited or where the dissociation of receptor-ligand complexes is prevented. On the other hand, studies of the kinetics of receptor movement under these conditions have shown that receptor occupancy does not have a significant effect Minireview:  Enzyme Targeting on the rate of receptor movement (30, 54, 76). These latter results indicate that the CI-MPR shuttles constitutively between the cell surface and intracellular compartments. However, the presence or absence of ligand might induce small changes in the rates of receptor movement to various compartments, thereby affecting the steady state distribution of receptor without having much of a n effect on the overall rate of receptor movement. In some specialized cell types, insulin and other growth factors induce a dramatic redistribution of the CI-MPR and the glucose transporter from endosomal compartments to the cell surface (77, 78, and op. cit.). This redistribution is associated with a decrease in the phosphorylation state of the plasma membrane-associated CI-MPRs (78). It is not known if this change in phosphorylation is the cause or a consequence of the redistribution.
This can now be tested directly by mutating the potential phosphorylation sites on the receptor and determining if the response to growth factors is maintained.

Conclusions
The cloning of the MPRs has provided new insight into the structure and function of the receptors, yet many questions remain.