N-Linked Oligosaccharides on the Low Density Lipoprotein Receptor Homolog SorLA/LR11 Are Modified with Terminal GalNAc-4-SO4 in Kidney and Brain*

Sorting protein-related receptor (SorLA/LR11) is a highly conserved mosaic receptor that is expressed by cells in a number of different tissues including principal cells of the collecting ducts in the kidney and neurons in the central and peripheral nervous systems. SorLA/LR11 has features that indicate it serves as a sorting receptor shuttling between the plasma membrane, endosomes, and the Golgi. We have found that a fraction of SorLA/LR11 that is synthesized in the kidney and the brain bears N-linked oligosaccharides that are modified with terminal β1,4-linked GalNAc-4-SO4. Oligosaccharides located in the vacuolar sorting (Vps) 10p domain (Vps10p domain) are modified with β1,4-linked GalNAc when the Vps10p domain is expressed in cells along with either of two recently cloned protein-specific β1,4GalNAc-transferases, GalNAcTIII and GalNAcTIV. Either of two sequences with basic amino acids located within the Vps10p domain is able to mediate recognition by these β1,4GalNAc-transferases. The highly specific modification of oligosaccharides in the Vps10p domain of SorLA/LR11 with terminal GalNAc-4-SO4 suggests that this unusual modification may modulate the interaction of SorLA/LR11 with proteins and influence their trafficking.

Sorting protein-related receptor (SorLA), 2 also known as LR11, is a highly conserved mosaic, type 1 receptor. The luminal/extracellular domain consists of a short NH 2 -terminal sequence preceding a consensus site for furin cleavage, a vacuolar sorting (Vps) 10p domain, 5 F/YWTD/␤-propeller modules, an epidermal growth factor precursor sequence, 11 low density lipoprotein receptor (LDLR) class A (LDLR-A) repeats, and 6 fibronectin type III repeats (1,2). The single transmembrane domain is followed by a 54-amino acid cytosolic domain containing a motif that mediates interaction with GGA (Golgilocalized, ␥-ear containing, ARF binding) proteins involved in Golgi-endosome sorting (3). The presence of these varied domains places SorLA/LR11 in a number of different families and underscores its ability to interact with a range of different ligands. A Vps10p domain is also present in the receptors sortilin-1 (4) and SorCS (5). The Vps10p domain of SorLA/ LR11 mediates the endogenous binding of the propeptide released from the N terminus of SorLA/LR11 by furin cleavage. The peptides neurotensin and hydra head activator also bind to the Vps10p domain (6). The YWTD repeats, epidermal growth factor repeat, and the LDLR class A repeats are characteristic of LDLR family members (7) and likely mediate apolipoprotein E and receptor-associated protein binding by SorLA/LR11 (6). In addition, the 6 fibronectin type III repeats are characteristic of a number of neural adhesion molecules (2).
The peptide portion of SorLA/LR11 consists of 2215 amino acids with a calculated molecular weight of 247,000. The receptor migrates with M r of over 300,000 when examined by SDS-PAGE indicating that a large fraction of the 20 potential Asn-glycosylation sites present on the luminal domain are utilized. We have determined that a fraction of SorLA/LR11 produced in the kidney and brain bears N-linked oligosaccharides that terminate with ␤1,4-linked GalNAc-4-SO 4 . We first described N-linked oligosaccharides terminating with ␤1,4-linked GalNAc-4-SO 4 on the glycoprotein hormone lutropin (8). A limited number of other glycoproteins including thyrotropin (8 -11), proopiomelanocortin (12)(13)(14), tenascin-R (15,16), carbonic anhydrase VI (17), Tamm-Horsfall glycoprotein (18,19), and sialoadhesin (20) also bear these novel sulfated oligosaccharides. Terminal GalNAc-4-SO 4 on N-linked oligosaccharides is recognized by the Mannose/GalNAc-4-SO 4 receptor, expressed by hepatic endothelial cells of mammals, which mediates the clearance of glycoproteins bearing these structures from the blood (21)(22)(23)(24). The cysteine-rich domain, a region located near the amino terminus of the mannose/GalNAc-4-SO 4 -receptor, accounts for GalNAc-4-SO 4 binding (25). We have used a chimeric protein termed Cys-Fc, which consists of the cysteinerich domain and the constant region of human IgG1, to identify tissues and specific cells that express glycoproteins bearing terminal GalNAc-4-SO 4 and to isolate these glycoproteins by affinity chromatography (15,16,25). In this report we provide evidence that SorLA/LR11 is among the small family of glycoproteins that are modified with N-linked oligosaccharides that terminate with ␤1,4-linked GalNAc-4-SO 4 .
Cys-Fc-Agarose Preparation-Versalinx technology (Calbiochem) was used to covalently attach Cys-Fc to agarose. Versalinx amine modifying reagent was incubated with Cys-Fc at a molar ratio of 10:1. The phenyldiboronic acid-modified Cys-Fc was then incubated with Versalinx SHA-agarose. The conjugate contained 10 -20 mg of Cys-Fc per ml of agarose.
Western Blot Analyses-Samples were analyzed by SDS-PAGE using 3-8% NuPAGE Tris acetate polyacrylamide gels (Invitrogen). Molecular weight standards, Precision Plus Protein Dual Color Standards (Bio-Rad), were included to determine the apparent mobilities of proteins of interest. Following electrophoretic transfer to PVDF membranes (Millipore, Billerica, MA) the membranes were blocked with 1% casein (Hammersten grade, BDH Chemicals) and then analyzed using Super Signal West Femto Maximum Sensitivity Substrate (Pierce) for detection of HRP-labeled probes. Cys-Fc-HRP was prepared by conjugating HRP to Cys-Fc (25) using the EZ-link Activated Peroxidase Antibody Labeling Kit (Pierce). Incubations with Cys-Fc-HRP were carried out between 24 and 72 h at 4°C. Rabbit antibody directed against the fibronectin domain of SorLA/ LR11 (26) was provided by Dr. Chica Shaller (Universitat Hamburg, Hamburg, Germany). Constructs containing the Vps10p domain of SorLA/LR11 with a V5 epitope at the carboxyl terminus were detected using HRP-anti-V5 (Invitrogen). Mouse anti-Tenascin-R/Janusin (catalog number 610629) was purchased from BD Biosciences.
Purification of Cys-Fc Reactive Glycoproteins from Kidney-The medullas from 50 mouse kidneys were excised, frozen in liquid nitrogen, and fractured into a powder using a Bio-Pulverizer (BioSpec Products, Bartlesville, OK). The powder was suspended in ice-cold tissue protein extraction reagent (TPER) (Pierce Chemicals) containing EDTA-free Complete protease inhibitor (Roche, Nutley, NJ), 5 ml/g of powder, using a Polytron (Brinkmann). The extract was sedimented at 10,000 ϫ g for 5 min and the supernatant collected. The supernatant was incubated with the anion exchanger Vivapure-D (Vivasciences, Carlsbad, CA). Bound proteins were eluted with a step gradient consisting of 0.1, 0.5, 1.0, 1.5, and 2.0 M NaCl in TPER. Cys-Fc reactive proteins were eluted in the 0.5 and 1.0 M NaCl fractions as determined by Western blot analysis following separation on NuPAGE gels. The fractions containing Cys-Fc reactive protein were pooled, dialyzed against TPER, and then incubated with 150 l of Cys-Fc-agarose at 4°C for 48 h. The Cys-Fc-agarose was collected by brief sedimentation and washed 4 times with 3000 l of ice-cold TPER for 30 min. The bound proteins were eluted by incubation 3 times with 300 l of ice-cold TPER con-FIGURE 1. Immunostaining of mouse kidney with Cys-Fc. Cryosections, 10 m, were immunostained using Cys-Fc as described under "Experimental Procedures" and photographed at different magnifications. Immunoreactive material is largely confined to collecting ducts in the medulla and papilla (A and B). At higher magnification (C) the immunoreactive material is located in principal cells. taining 1 mM GalNAc-4-SO 4 for 30 min. The fractions were analyzed by Western blot on 3-8% Tris acetate NuPAGE gels as shown in Fig. 2.
MS/MS Peptide Sequence Analysis-A protein corresponding to the Cys-Fc reactive material seen by Western blot was visualized with Colloidal Coomassie Blue G-250 and excised from the acrylamide gel for sequence analysis. Proteolytic digestion with trypsin, separation of peptides by microcapillary reverse-phase HPLC, nano-electrospray tandem mass spectrometry on a Finnigan LCQ DECA XP quadrapole ion trap mass spectrometer, and correlation of the fragments with known sequences using the Sequest algorithm were performed by the Harvard Microchemistry Facility, Harvard, University, Cambridge, MA. A total of 33 distinct peptides were identified that corresponded to the major component mouse SorLA/ LR11 were identified.
Analytic Separation of Cys-Fc Reactive Glycoproteins from Kidney-Following identification of the major Cys-Fc reactive band as SorLA/LR11 analytic separations were carried out. Medullas from 10 mouse kidneys were excised, minced, placed in 300 l of ice-cold TPER containing EDTA-free Complete protease inhibitor, and ground using a PlusOne Sample Grinding Kit (Amersham Biosciences). The extract was diluted with an additional 700 l of TPER containing EDTA-free Complete protease inhibitor and incubated for 1 h on ice with mixing. The sample was sedimented a 10,000 ϫ g for 10 min. The supernatant was incubated with 50 l of Cys-Fc-agarose in a Mini-Spin column (product number 69705, Pierce) at 4°C for 5 days. The unbound fraction was obtained by spinning the column at 50 ϫ g for 1 min. The Cys-Fc-agarose was then wash 2 times with 100 l of ice-cold TPER followed by collection by sedimentation at 50 ϫ g. The column was then plugged and the bound material eluted by incubation with cold TPER containing 1 mM GalNAc-4-SO 4 for 2 h followed by collection by sedimentation at 50 ϫ g. This was repeated a total of 4 times. Each of the fractions from the Cys-Fc-agarose column was analyzed by Western blotting using Cys-Fc and anti-SorLA/LR11 as shown in Fig. 3.
Expression of SorLA/LR11 Constructs-Human SorLA/LR11 cDNA in the expression plasmid pcDNA3.1/Zeoϩ (6) was kindly provided by Dr. Claus M. Petersen, Department of Medical Biochemistry, University of Aarhus, Denmark. The aminoterminal region containing the Vps10p domain was amplified and ligated into pcDNA3.1V5HisTOPO (Invitrogen) using primers 5Ј-GATGGCGACACGGAGCAGC-3Ј and 5Ј-GATC-TTCCGATAGCCTCTC-3Ј. Mutations were introduced into the Vps10p domain of pcDNA3.1hSorLA/Vps10p(V5His) by PCR-mediated overlap extension using the following primers: 1) deletion of KPLRRKR, 5Ј-CGCGCGGACGAGAGCGCT-GCCCTGCAGCCCGAGCCCATC-3Ј and 5Ј-CAGGGCA-GCGCTCTCGTCCGCGCGGCTCGCCCCCCTGG-3Ј; 2) replacement of KTVFKRR with KTVFQQQ, 5Ј-GTTTGC-TGGG ACACAAGACTGTTTTCCAGCAGCAGACCCCC-FIGURE 2. Affinity purification of Cys-Fc reactive glycoproteins from kidney. The medulla and papilla of the kidney were solublized with TPER. Proteins in the extract were bound to an anion exchange resin, Vivapure D, and eluted with NaCl. The fractions containing Cys-Fc reactive material were pooled and incubated with Cys-Fc immobilized on phenyldiboronic acid-SHA-agarose. After washing with TPER the bound proteins were eluted from the immobilized Cys-Fc using TPER containing 1 mM GalNAc-4-SO 4 . Aliquots of the extract prior to incubation with Cys-Fc-agarose (lane T, 0.3% of extract) and after incubation with Cys-Fc-agarose (lane marked UB, 0.3% of unbound fraction) as well as the wash fractions (lanes marked W1, W2, W3, and W4, 10.8% of each fraction) and GalNAc-4-SO 4 eluted fractions (lanes marked E1, E2, and E3 (10.8% of each fraction)) were analyzed by SDS-PAGE and Western blot analysis using Cys-Fc. The location of the major Cys-Fc reactive band is indicated by the arrow. Molecular weight standards are indicated by the arrowheads.

TABLE 1 Peptides identified from SorLA/LR11 by the MS/MS-peptide sequence analysis and comparison with known sequences using the Sequest algorithm
The sequence of each peptide identified from SorLA/LR11 and the amino acids represented by that sequence are shown.

RESULTS
Cys-Fc Reactive Material Is Located in the Medulla and Papilla of the Mouse Kidney-We previously reported that both protein-specific ␤1,4GalNAc-transferase activity and GalNAc-4-sulfotransferase activity can be detected in a number of tissues including the pituitary, salivary glands, brain, and kidney (27). Because the Cys-rich domain of the Man/GalNAc-4-SO 4 receptor is highly specific for terminal sulfated sugars such as ␤1,4-linked GalNAc-4-SO 4 (25,28,29), we utilized a chimeric protein consisting of the Cys-rich domain and the Fc domain of human IgG, Cys-Fc, to immunostain tissue sections for the presence of glycoproteins bearing terminal ␤1,4-linked Gal-NAc-4-SO 4 (15,25). Cells comprising collecting ducts in the medulla and papilla of the mouse kidney are intensely stained (Fig. 1). The immunoreactivity seen in the kidney in combination with the presence of protein-specific ␤1,4GalNAc-transferase activity and GalNAc-4-sulfotransferase activity indicate that one or more of the glycoproteins produced by these cells are potentially modified with terminal ␤1,4-linked GalNAc-4-SO 4 on their N-or O-linked oligosaccharides.
Proteins were solubilized from the cortex and medulla of mouse kidneys using TPER and examined by Western blotting following separation by SDS-PAGE and electrophoretic transfer to PVDF membranes. A glycoprotein with M r Ͼ 250,000 is the predominant species that reacts with the Cys-Fc chimera and is enriched in the extract of the medulla as compared with the cortex (not shown). This protein was bound by the anion exchanger Vivapure D and eluted at a NaCl concentration between 0.5 and 1.0 M. When the material eluted from the anion exchanger was incubated with Cys-Fc-agarose, the Cys-Fc reactive band was removed from the extract (Fig. 2, lane T versus UB). Consistent with the modest affinity we have observed for glycoproteins with a limited number of GalNAc-4-SO 4 termini (28), the Cys-Fc reactive material was present in the fractions eluted with 150 mM NaCl (Fig. 2, W1-W4) and in the fractions eluted with 1 mM GalNAc-4-SO 4 (Fig. 2, E1-E3). Staining with Coomassie Blue revealed the presence of a protein band at the position of the Cys-Fc reactive material eluted with 1 mM Gal-NAc-4-SO 4 . The fractions eluted with NaCl contained a large number of additional bands when stained with Coomassie Blue (not shown). We therefore utilized the GalNAc-4-SO 4 eluted material for MS/MS analysis.
MS/MS analysis of the protein eluted with GalNAc-4-SO 4 following proteolytic digestion with trypsin and separation of the resulting peptides by microcapillary HPLC yielded 33 peptides that corresponded to the glycoprotein SorLA/LR11 when analyzed by Sequest (Table 1). A non-erythroid form of spectrin was also identified in this sample, however, because spectrin is not glycosylated this was not pursued further. SorLA/LR11 is a member of the LDL-receptor family that is expressed in collecting ducts of the medulla and papilla of murine kidney (31).

SorLA/LR11 Produced in Kidney and Brain Is Bound by Immobilized
Cys-Fc-An extract of mouse kidney was prepared using TPER and Cys-Fc reactive glycoproteins were enriched by affinity chromatography as described above. As seen during the larger scale purification, the major Cys-Fc reactive glycoprotein is present in both the NaCl-eluted wash fractions and the GalNAc-4-    (Fig. 3A). Western blot analysis with an antibody directed at the fibronectin domain of SorLA/LR11 (26) reacted with the same glycoprotein species (Fig. 3B); however, the amount of immunoreactive SorLA/LR11 in the unbound (UB) fraction was not significantly reduced as compared with the total extract (T). Taken together this suggests that a subset of the SorLA/LR11 molecules produced in the kidney is modified with terminal GalNAc-4-SO 4 .
SorLA/LR11 is abundantly expressed by neurons in the brain (2, 4, 32, 33). Furthermore, GalNAc-4-sulfotransferase and protein-specific ␤1,4GalNAc-transferase activities are also expressed in the brain (27), and Cys-Fc reactive material is particularly prominent in the cerebellum and hippocampus (15,16). We recently reported that the extracellular matrix protein tenascin-R is modified with terminal GalNAc-4-SO 4 and can be bound to immobilized Cys-Fc (15). As can be seen in Fig. 4, a fraction of SorLA/LR11 (A and B, solid arrow), like tenascin-R (A and C, open arrows), is bound by immobilized Cys-Fc and eluted with GalNAc-4-SO 4 . This suggests that a subset of the SorLA/LR11 produced in the brain, as in the kidney, bears N-linked oligosaccharides that are modified with terminal ␤1,4-linked GalNAc-4-SO 4 .

Recombinant SorLA/LR11 Is Modified with Terminal Gal-NAc-4-SO 4 When Expressed in HEK-293 Cells-HEK-293 cells but not CHO cells are able to add terminal GalNAc-4-SO 4 to
Asn-linked oligosaccharides on glycoproteins containing the appropriate peptide recognition determinant, such as the glycoprotein hormone LH (34,35). A fraction of recombinant human SorLA/LR11 synthesized in HEK-293 cells endogenously expressing ␤1,4-GalNAc-transferase and GalNAc-4sulfotransferase activities required for GalNAc-4-SO 4 addition was bound by immobilized Cys-Fc and eluted with GalNAc-4-SO 4 (Fig. 5, lower panel). In contrast, none of the recombinant SorLA/LR11 synthesized in CHO cells, which do not express the required transferases, was bound by immobilized Cys-Fc (Fig. 5, upper panel). The fraction of recombinant SorLA/LR11 produced in HEK-293 cells that was bound by immobilized WFA, which binds terminal ␤1,4-linked GalNAc (36, 37), was higher than the fraction of SorLA/LR11 bound by immobilized Cys-Fc (not shown). This indicates that much of the GalNAc added to the recombinant SorLA/LR11 is not further modified with sulfate. Expression of recombinant GalNAc-4-ST1 (38) along with recombinant SorLA/LR11 in HEK-293 cells reduces the amount of SorLA/LR11 bound by immobilized WFA and increases the amount bound by Cys-Fc (not shown). Thus the level of endogenous GalNAc-4-sulfotransferase is not sufficient   W2, and W3), and 40% of three fractions eluted with 50 mM GalNAc (E1, E2, and E3) were analyzed by SDS-PAGE. Intact SorLA/LR11 (solid arrow) was identified using rabbit anti-SorLA 6.7 directed against a fibronectin type III domain. Additional species seen in the elution fractions reflect components in the elution buffer that are nonspecifically labeled.
to modify all of the terminal ␤1,4-linked GalNAc present on recombinant SorLA/LR11 when the latter is expressed in HEK-293 cells.

The Vps10p Domain of SorLA/ LR11 Contains Recognition Determinants That Are Required for GalNAc Addition to its N-Linked
Oligosaccharides-The amino-terminal 755 amino acids of SorLA/ LR11 contain the Vps10p domain and five potential N-linked glycosylation sites (see Fig. 7). Replacement of the carboxyl-terminal 1460 amino acids of SorLA/LR11 with a V5 eptiope followed by six His residues yields a glycoprotein, SorLA/ LR11-V5His, that is secreted into the medium. When expressed in parent Flp-In CHO cells that do not contain any ␤1,4GalNAcT activity, SorLA/LR11-V5His is not bound by either immobilized Cys-Fc (not shown) or WFA-agarose (Fig. 8, Wt  CHO)). In contrast, SorLA/LR11-V5His expressed in HEK-293 cells containing both ␤1,4GalNAcT and GalNAc-4-sulfotransferase activity is bound by both immobilized Cys-Fc (not shown) and WFA-agarose (Fig. 8, Wt 293T). Thus, the Vps10p domain contains N-linked oligosaccharides and one or more recognition determinants that are capable of mediating the selective addition of ␤1,4-linked GalNAc by the protein-specific ␤1,4-GalNAcTs that are expressed in HEK-293T cells. These ␤1,4-GalNAcTs modify the oligosaccharides on glycoproteins such as LH (21,35), tissue factor pathway inhibitor (34), and protein C (42). In the case of the ␣-subunit of the glycoprotein hormones, we have shown that basic amino acids in the sequence Pro-Leu-Arg-Ser-Lys-Lys are components of the recognition determinant utilized by the ␤1,4-GalNAcTs in the pituitary (43,44). Three distinct clusters of basic amino acids are present within the Vps10p domain of SorLA/LR11 (see Fig. 7), each of which could serve a recognition determinant for GalNAc addition. The potential recognition sequences were deleted or mutated in the SorLA/LR11-V5His construct as illustrated in Fig. 7, and the constructs were expressed in HEK-293T cells. Only those mutants, Mu-3 and Mu-5, in which the furin cleavage site, KPL-RRKR, was deleted and the sequence KTVFKRR was changed to KTVFQQQ were completely devoid of GalNAc following expression in HEK-293T cells (Fig. 8). Thus, either KPLRRKR or KTVFKRR but not RRTRGYRK are able to mediate recognition by the protein-specific ␤1,4-GalNAcTs in HEK-293T cells. The same experiments were performed with Flp-In CHO cells expressing ␤1,4-GalNAcTIII or ␤1,4-GalNAcTIV (not shown) with the same results as summarized in Table 2, indicating that these two transferases display similar specificities with respect to GalNAc addition to SorLA/LR11.

DISCUSSION
The addition of terminal ␤1,4-linked GalNAc-4-SO 4 to N-linked oligosaccharides is a highly regulated and specific FIGURE 7. Schematic of SorLA/LR11. The locations of the VPs10p domain, the LDL-receptor type A and type B repeats, the epidermal growth factor precursor sequence, the fibronectin type III repeats, and the transmembrane domain are illustrated schematically. The Vps10p domain contains 6 potential Asn-linked glycosylation sites. The furin cleavage site () and two other sequences contain basic residues that could potentially act as recognition determinants for the ␤1,4GalNAc-transferases. These were either deleted as indicated by the dashed lines or basic residues replaced with glutamine residues (Q) in the combinations shown for Mu1-Mu6. The ϩ and Ϫ indicate those constructs that were or were not modified with GalNAc, respectively. process. We have shown that expression of a glycoprotein containing a peptide sequence that is recognized by a protein-specific ␤1,4-GalNAcT, a protein-specific ␤1,4-GalNAcT, and a GalNAc-4-sulfotransferase are all required for terminal Gal-NAc-4-SO 4 addition (45). In the case of the glycoprotein hormones, we have determined that the peptide recognition determinant, protein-specific ␤1,4-GalNAcT activity, and GalNAc-4-sulfotransferase activity have been conserved throughout vertebrate evolution. As a consequence glycoprotein hormones of all vertebrates from fish to man bear this unique sulfated carbohydrate structure (45,46). The terminal GalNAc-4-SO 4 is recognized by the Man/GalNAc-4-SO 4 receptor in hepatic endothelial cells. This receptor controls the circulatory half-life of LH following its secretion into the blood (21)(22)(23)(24)(25)47). Thus, this unique sulfated carbohydrate structure has important functional consequences for glycoproteins that are secreted into the blood.
We have sought to identify additional glycoproteins that bear this uncommon sulfated carbohydrate structure, to determine whether terminal GalNAc-4-SO 4 moieties also have functional consequences for these glycoproteins. Cys-Fc immunostains of histologic sections from different tissues have revealed specific structures and cells containing glycoproteins that potentially terminate with ␤1,4-linked GalNAc-4-SO 4 . The presence of protein-specific ␤1,4-GalNAcT (40,41) and GalNAc-4-sulfotransferase (38, 48 -50) activity and/or mRNA in these same locations indicates that one or more glycoproteins bearing terminal GalNAc-4-SO 4 are present. SorLA/LR11, like sialoadhesin and CD45 that are also reported to bear terminal GalNAc-4-SO 4 (20), is an integral membrane protein rather than a circulating hormone such as LH that also bears terminal Gal-NAc-4-SO 4 . It is, therefore, highly unlikely that the SorLA/ LR11 bearing GalNAc-4-SO 4 from either the kidney or the brain will ever encounter the Man/GalNAc-4-SO 4 receptor expressed by hepatic endothelial cells. Yet, SorLA/LR11 synthesized by two unrelated cell types, principal cells of the kidney (31) and neurons in the central nervous system (1,2), is in both instances modified with terminal ␤1,4-linked GalNAc-4-SO 4 . It is therefore likely that the terminal ␤1,4-linked GalNAc-4-SO 4 present on SorLA/LR11 will prove to have important functional consequences.
Based on the mutations we have introduced into the Vps10p domain of SorLA/LR11, there are two clusters of basic amino acid residues that are able to act as recognition determinants for the ␤1,4GalNAcT activities present in HEK-293T cells and for both recombinant ␤1,4GalNAcTIII and ␤1,4GalNAcTIV when expressed in CHO cells. We have found that ␤1,4GalNAcTIII and ␤1,4GalNAcTIV are also both able to modify the N-linked oligosaccharides on the glycoprotein hormone ␣ subunit, 3   Aliquots of the medium, the medium following incubation with WFA-agarose, the wash fractions, and the bound material eluted with GalNAc were analyzed by Western blot using antibody directed against the V5 epitope following SDS-PAGE and electrophoretic transfer to PVDF. Equal percentages of material were analyzed from the medium, unbound fraction, and eluted (bound) fractions except for 1 where 4-fold more of the bound fraction was analyzed than the total or unbound fraction and 2 where 10-fold more of the bound fraction was analyzed that the total or unbound fraction.

TABLE 2
The basic residues in the sequences KTVFKRR are necessary and sufficient for the addition of ␤1,4-linked GalNAc to the N-linked oligosaccharides in the Vps10p domain of SorLA/LR11 The Vps10p domain of SorLA/LR11 was expressed in HEK-293T cells, CHO/Gal-NAcTIII, and CHO/GalNAcTIV cells. The medium was collected and incubated with WFA-agarose. After washing the WFA-agarose, bound material was eluted with 50 mM GalNAc. The amount of the Vps10p domain present in the medium, unbound fraction, washes, and eluted fractions were determined by Western blotting with anti-V5 following SDS-PAGE and electrophoretic transfer to PVDF. Forms of the Vps10p domain that were bound are indicated with a plus (ϩ), those that were not bound with a minus (Ϫ). Insufficient levels of Mu-2 were released into the medium for analysis. The function of SorLA/LR11 in either the kidney or the brain has not yet been established. SorLA/LR11 is able to mediate endocytosis of ligands such as apoE; however, only 10% of the receptor is expressed at the cell surface (6). The cytosolic domain of SorLA/LR11 contains a motif that results in binding to proteins GGA-1 and GGA-2 (Golgilocalized, ␥-ear containing, ARF binding). GGA proteins are involved in Golgi-endosome sorting and SorLA/LR11 is found predominantly on intracellular vesicular membranes (3). Thus, SorLA/LR11 is thought to play a role in intracellular trafficking.

HEK-293T CHO/GalNAcTIII CHO/GalNAcTIV
Recent studies have shown that the luminal domain of SorLA/LR11 binds amyloid precursor protein and reduces the rate of amyloid precursor protein cleavage to amyloid ␤-peptide (A␤). Increased levels of A␤ are present in the brains of mice that do not express SorLA/LR11 (51). Furthermore, reduced levels of SorLA/LR11 are found in brains of Alzheimer disease patients (52). Similarly, it is possible that SorLA/LR11 in the principal cells of the kidney influences the intracellular trafficking of intracellular shuttling compartments that contain aquaporin2. The vasopressin-mediated trafficking of aqua-porin2 between intracellular compartments and the plasma membrane is important for the maintenance of water balance (31,53). The presence or absence of terminal GalNAc-4-SO 4 has the potential to modulate binding of a number of different proteins with SorLA/LR11 and thereby alter their intracellular trafficking. This would only be the case for the subset of SorLA/ LR11 that is modified with terminal GalNAc-4-SO 4 . At present we do not know what determines the proportion of SorLA/ LR11 that is modified with GalNAc-4-SO 4 . Not all cells that express SorLA/LR11 necessarily also express the ␤1,4GalNAcT and GalNAc-4-sulfotransferase activities required for GalNAc-4-SO 4 addition. Alternatively, the extent of modification within individual cells may be regulated, producing two distinct populations of SorLA/LR11 that differ in the patterns of terminal glycosylation.
SorLA/LR11 is one of a small number of integral membrane glycoproteins, including sialoadhesin and CD45 (20), selectively modified with terminal, ␤1,4-linked GalNAc-4-SO 4 in vivo. This is true in at least two distinct cell types, distal collecting tubules in the kidney and neurons in specific regions of the central nervous system. Because the addition of ␤1,4-linked GalNAc-4-SO 4 is protein-specific and highly regulated and only a fraction of SorLA/LR11 is modified with this structures, the presence of this sulfated carbohydrate structure may be critical for some aspects of SorLA/LR11 function. Determining the functional significance of these sulfated carbohydrates on SorLA/LR11 may provide insight into the function of this complex receptor.