(Received for publication, July 16, 1996, and in revised form, November 11, 1996)
From the Department of Medical Biochemistry,
University of Aarhus, 8000 Aarhus C, Denmark and ¶ The John F. Kennedy Institute, DK-2600 Glostrup, Denmark
Receptor-associated protein (RAP) is an endoplasmic reticulum/Golgi protein involved in the processing of receptors of the low density lipoprotein receptor family. A ~95-kDa membrane glycoprotein, designated gp95/sortilin, was purified from human brain extracts by RAP affinity chromatography and cloned in a human cDNA library. The gene maps to chromosome 1p and encodes an 833-amino acid type I receptor containing an N-terminal furin cleavage site immediately preceding the N terminus determined in the purified protein. Gp95/sortilin is expressed in several tissues including brain, spinal cord, and testis. Gp95/sortilin is not related to the low density lipoprotein receptor family but shows intriguing homologies to established sorting receptors: a 140-amino acid lumenal segment of sortilin representing a hitherto unrecognized type of extracellular module shows extensive homology to corresponding segments in each of the two lumenal domains of yeast Vps10p, and the extreme C terminus of the cytoplasmic tail of sortilin contains the casein kinase phosphorylation consensus site and an adjacent dileucine sorting motif that mediate assembly protein-1 binding and lysosomal sorting of the mannose-6-phosphate receptors. Expression of a chimeric receptor containing the cytoplasmic tail of gp95/sortilin demonstrates evidence that the tail conveys colocalization with the cation-independent mannose6-phosphate receptor in endosomes and the Golgi compartment.
Sorting of newly synthesized lysosomal enzymes from the Golgi compartment to late endosomes in eukaryotic cells is a sophisticated transport process involving specific sorting receptors in the trans-Golgi network. In mammals, the 46- and 275-kDa mannose-6-phosphate (M6P)1 receptors are the known sorting receptors that bind to phosphorylated mannose residues in lysosomal hydrolases (1). In yeast, a M6P-independent sorting pathway has been demonstrated by identification of the vacuolar protein-sorting 10 protein (Vps10p) (2) and a highly homologous protein encoded by the yeast VTH2 gene (3). Both are capable of targeting yeast carboxypeptidase Y to lysosomes (2, 3). Mammalian counterparts to these sorting receptors have so far not been identified. However, studies of I-cell disease patients suggest that mammals may sort lysosomal enzymes by alternative mechanisms (4-9).
The 40-kDa endoplasmic reticulum/Golgi receptor-associated protein (RAP) assists folding and processing of the cysteine-rich low density lipoprotein (LDL) receptor class A repeats in receptors of the LDL receptor family (10-13). In addition to the high affinity binding to the LDL receptor family proteins and the newly identified LDL receptor type A repeat containing receptor sorLA/LR11 (14, 15), RAP binds calmodulin and is phosphorylated by calmodulin-dependent kinase II and casein kinase II (16). Recently, independent observations have shown the binding of RAP to an approximately 100-kDa protein expressed in osteosarcoma (17) and Chinese hamster ovary cells (18).
In the present study we have identified, purified, and characterized a ~95-kDa protein by RAP affinity chromatography of membrane protein extracts from human brain. Cloning of the encoding gene by cDNA library screening disclosed a type I receptor structure and striking sequence homologies with yeast Vps10p and the cation-dependent (CD) and cation-independent (CI) M6P receptors. Transfection analyses of the novel receptor, which represents the first known RAP binding receptor not related to the LDL receptor family, disclosed that the cytoplasmic tail conveys localization to the Golgi and CI-M6P receptor-rich vesicles.
Human cadaver brain tissue samples were, with permission, obtained from autopsy cases 12-24 h postmortem, and crude membrane fractions were generated, solubilized in 1% CHAPS (Boehringer, Mannheim, Germany) dissolved in 2 mM CaCl2, 1 mM MgCl2, 10 mM Hepes, and 140 mM NaCl, pH 7.8 (buffer A) and applied to RAP affinity chromatography (16). After washing, protein retained on the column was eluted at pH 4.0 in phosphate-buffered saline containing 10 mM EDTA and 0.25% CHAPS. The eluted fractions (1 ml) were neutralized by Tris-base, analyzed by SDS-PAGE, and finally pooled and concentrated using a centricon-10 from Amicon (Beverly, MA).
SDS-PAGE, Two-dimensional Gel Electrophoresis, Deglycosylation, and BlottingSDS-PAGE was performed in a Laemmli system using a
4% stacking gel and a 4-16% gradient separation gel. Two-dimensional
gel electrophoresis was done as described (16). Deglycosylation was
performed using peptide-N-glycosidase-F (19). Molecular weight markers were from Sigma or Novex (San Diego,
CA). For ligand/immunoblotting cells, membrane preparations and whole
tissue samples were lysed for 10 min at 4 °C in 1% Triton X-100 or
Triton X-114, 10 µg/ml aprotinin, 2 mM
phenylmethylsulfonyl fluoride, and 0.1 M Tris/HCl (pH 8).
After centrifugation (5 min, 104 × g, 4 °C),
the pellet was discarded and supernatants were stored at 20 °C.
Ligand blotting was performed as described using 5 × 104 cpm/ml of 125I-RAP or
125I-urokinase in complex with plasminogen activator
inhibitor-1. Iodinated ligands were prepared as described previously
(20).
For immunoblotting, Immobilon membranes were blocked in 2% Tween 20 buffer (pH 7.4) and incubated with antibodies in 0.2% defatted milk powder, 0.05% Tween 20, and buffer A (pH 7.4).
Amino Acid SequencingPooled fractions of RAP affinity-purified proteins were resolved by SDS-PAGE and transferred to Immobilon filters by electroblotting in 50 mM Tris-base, 50 mM boric acid buffer. Membrane-bound proteins were stained using 0.1% amido black in 45% methanol, 9% acetic acid, and the respective bands were cut from the membrane, incubated (15 min) in water containing 0.2% polyvinylpyrrolidone, washed twice in water, and cleaved in situ by overnight incubation at 37 °C in 200 µl digestion buffer (ammonium bicarbonate, 0.01% polyvinylpyrrolidone, 0.1 M Ca2+) supplemented with 1 µg of trypsin (modified trypsin, sequencing grade; Promega). The digestion mixture (supernatant) was recovered and acidified by addition of 5-10 µl of trifluoroacetic acid, and the membrane was finally washed twice in distilled water. The washing solutions were added to the digestion mixture, and the peptide separation was performed on an applied Biosystems microbore high pressure liquid chromatography (model 172) with a 500-µl injection loop. The digests were fractionated on a 2 × 100-mm, 5 µm Nucleosil 18 reverse phase column. Peptides were eluted in a linear gradient of acetonitrile and 0.1% trifluoroacetic acid at a flow rate of 100 µl/min and detected at 214 nm. Amino acid sequencing of selected peptides and N-terminal sequencing of undigested protein was performed in a 477 gas-phase sequenator connected on line to a 120A PTH-analyzer. The sequenator was operated according to the manufacturer's instructions with previously described modifications (21).
cDNA Cloning and SequencingA degenerated
oligodeoxyribonucleotide probe
(5-GA(A/S)TT(C/T)GGIATGGCIAT(A/C/T)GG-3
), derived from the peptide
sequence EFGMAIG and containing deoxyinosine at the most ambiguous
positions, was used to screen a Jurkat cDNA library established in
the ZAP Express vector (Stratagene). Following 5
-end labeling with
[
-32P]ATP, the probe was used to screen four sets of
replicate nylon filters (22 × 22 cm, Amersham), each containing
1.25 × 105 plaque-forming units. The filters were
hybridized at 50 °C using the tetramethylammonium chloride technique
(22, 23). Three positive clones were identified, purified, and rescued
into the pBK-CMV vector according to the manual from Stratagene.
Further analysis of the purified clones was carried out by dot and
Southern blotting using the original cloning probe and three
additional, fully degenerated, oligodeoxyribonucleotide probes derived
from peptides PGEDEE, DFVAKL, and EEHLTT. Two of the clones reacted with all four probes, and one of these, containing an insert of 4.5 kb,
was selected for sequence analysis. The insert was recovered by
EcoRI cleavage yielding two fragments. For enzyme
restriction analysis, the fragments were subcloned into the M13BM20/21
vector, and single-stranded DNA was sequenced by dideoxy sequencing
using a T7 sequencing kit (Pharmacia). The partial sequences were used to construct 17mer oligonucleotides allowing consecutive sequencing of
both strands. As the 500-base pair upstream G/C-rich sequence contained
several compressions, this stretch was sequenced using a T7 inosine
sequencing kit (U. S. Biochemical Corp.).
The 5
fragment of the selected gp95/sortilin cDNA, spanning from the
XbaI site of the pBK-CMV vector to the SmaI site
at position 3590, was cloned into the eukaryotic expression vector pMT21 (24). A cDNA encoding a chimeric receptor was constructed by
generating a 3
HindIII site and 5
XhoI site in
the gp95/sortilin cDNA encoding the cytoplasmic tail (aa 748-800)
and inserting this sequence into the previously described plasmid
vector (25) (kindly donated by Dr. Susan E. LaFlamme) encoding the
extracellular and the transmembrane domain of the interleukin 2 (IL-2)
receptor (Tac/CD25). Transfection of the pMT21-gp95/sortilin, the
chimeric receptor construct, and control plasmid into COS-1 cells was
carried out using LipofectAMINE (Life Technologies, Inc.) as
transfection agent (26). 48 h after transfection, the cells
cultured in Petri dishes were washed and lysed in 1% Triton X-100 for
ligand and Western blotting. For confocal and fluorescence microscopy,
transfected cells were washed in 10 mM phosphate, 150 mM NaCl, pH 7.3 (buffer B), fixed in buffer B containing
3.7% formaldehyde (10 min), and permeabilized in the same solution
with 1% Triton X-100 (1 min). Fixed cells were washed in 0.05% Tween
20 buffer B, and incubations with primary and secondary antibodies were
carried out in antibody diluent (DAKO, Denmark). Nuclei were stained by
incubation with 4
,6-diamino-2-phenylindole dihydrochloride hydrate
(DAPI; Sigma). For surface labeling, the cells were
incubated at 4 °C with the primary antibody prior to fixation in
formaldehyde. Horseradish peroxidase-, fluorescein isothiocyanate-, and
rhodamine-conjugated anti-rabbit/anti-mouse antibodies were from DAKO.
Rabbit anti-gp95/sortilin peptide anti-serum was generated (Neosystem,
Strasburg, France) using a synthetic peptide (SAPGEDEEGGRVRDFVAKLA)
based on the N-terminal sequence obtained from purified gp95/sortilin.
Mouse anti-CD25 was from Boehringer. Metabolic labeling of transfected cells was done overnight in cysteine- and methionine-free medium supplemented with L-[35S]methionine and
L-[35S]cysteine (Pro-mix; Amersham) and 10%
full medium. Labeled cells were lysed in 1% Triton X-114, Tris/HCl (pH
8.0), and following phase separation, proteins were precipitated from
the fluid phase (in 0.1% CHAPS buffer A) using RAP immobilized on
Sepharose beads and uncoated beads for control. Precipitated protein
was analyzed by reducing SDS-PAGE, and diphenyloxazole-fluorographed
gels were exposed at
70 °C.
Northern blotting was performed using tissue blots 7760-1, 7759-1, and 7767-1 (Clontech) containing 2 µg of poly(A)+ purified RNA from different human tissues. Hybridizations were carried out with a random-primed probe including 1581 nucleotides (an EcoRI fragment) of the sortilin clone. Washing was done at high stringency conditions according to standard procedure, and results were analyzed by autoradiography.
Chromosome MappingFluorescence in situ hybridization using the biotin-labeled 4.5-kb clone with corresponding DAPI banding and measurement of the relative distance from the short arm telomere to the signals (FLpter value) was performed as described previously (27).
A ~95-kDa RAP binding protein (designated
gp95/sortilin) was identified by ligand blotting of human brain tissue
lysates (Fig. 1A) and purified by RAP
affinity chromatography. Fig. 1B presents a non-reducing
SDS-PAGE profile of the proteins eluting from the RAP column and shows
that in addition to high molecular weight proteins (>200,000),
including the 515-kDa -chain of LDL receptor-related protein, a
single protein of the expected molecular mass of ~95 kDa was released
from the column. As determined by ligand blotting, both the high
molecular weight proteins and gp95/sortilin bound RAP (not
demonstrated). Fig. 2A shows that
125I-RAP binding to gp95/sortilin was inhibited by heparin
and completely abolished in the presence of 100 nM
unlabeled RAP or (not shown) 10 mM EDTA. Furthermore,
binding of 125I-RAP was not inhibited by urokinase in
complex with plasminogen activator inhibitor type-1 (200 nM), and 125I-labeled urokinase-inhibitor
complex did not bind, verifying that the RAP binding protein was
different from the very low density lipoprotein receptor. To ensure
that the observed RAP binding was accounted for by a single protein,
electroeluted gp95/sortilin was subjected to reducing SDS-PAGE and
two-dimensional gel electrophoresis. As demonstrated in Fig.
2B, reduction did not significantly alter the mobility of
gp95/sortilin and did not give rise to additional protein bands.
Results where similar when using two-dimensional gels (not shown),
which presented gp95/sortilin as a streak of different pI species (pI
4.5-5.0) in accordance with differences in glycosylation.
peptide-N-glycosidase-F treatment indicated that glycosyl
groups account for about 10% of the mass of the native protein (Fig.
2B).
N-terminal Sequencing and Antibody Generation
N-terminal
sequencing of purified gp95/sortilin produced a single 20-amino acid
sequence (Ser45-Ala64, Fig. 4) that showed no
homology to any known protein sequence in the EMBL or Genbank data
bases. A synthetic peptide, derived from the sequence obtained by the
N-terminal sequencing, was used for generation of rabbit antibodies. As
demonstrated in Fig. 3, the anti-peptide antibody
reacted with gp95/sortilin but not with any of the other proteins
eluting from the RAP affinity column (panel B). Moreover,
the antibody bound to electroeluted gp95/sortilin (Fig. 3A,
shown to bind RAP in Fig. 2A) and, like
125I-RAP, to a ~95-kDa protein in crude lysates of brain
membranes (Fig. 3C). Thus, the combined findings identify
gp95/sortilin as a hitherto undescribed RAP-binding membrane
glycoprotein.
Sequence Analysis of gp95/Sortilin
To obtain information of
the primary sequence of gp95/sortilin, N-terminal sequences of tryptic
fragments isolated by reverse phase chromatography were determined. 90 residues dispersed over 11 fragments were identified, and four
sequences were selected for construction of oligonucleotides. A
cDNA clone containing a 4.5-kb insert hybridizing with all four
oligonucleotides was isolated from a human Jurkat cDNA library and
sequenced. The sequence presents a single open reading frame encoding a
33-aa signal peptide and a 800-aa protein (Fig. 4) with
a predicted mass of 88.8 kDa and a pI of 5.7. The amino acid sequences
obtained from the tryptic fragments were all identified in the
translated sequence (underlined residues in Fig. 4). As seen
in Fig. 4, the translated protein has a typical type I receptor
structure with two hydrophobic regions. The N terminus (aa 33 to
1)
constitutes a potential endoplasmic reticulum import signal sequence
with a hydrophobic core and a predicted cleavage site at
Gln1 (28). The second hydrophobic stretch (aa 759-780) in
the C-terminal region constitutes the putative transmembrane segment.
These predictions suggest that the synthesized protein consists of a
lumenal domain of 726 amino acids with 6 potential N-linked
glycosylation sites and a short cytoplasmic tail of 53 residues.
Interestingly, the peptide sequence
Ser45-Ala64, which was determined as the N
terminus of the affinity-purified protein, is not the N terminus of the
protein sequence deduced from the cDNA sequence. Instead, this
sequence is C-terminal to the RWRR (aa 41-44) sequence that abides by
the basic consensus motif
Arg
4-Xaa
3-Arg
2/Lys
2-Arg
1-Xaa+1
for cleavage (between positions
1 and +1) by the trans-Golgi proteinase furin (29). Consequently, these data suggest that gp95/sortilin is synthesized as a precursor, which is processed into a
5-kDa peptide and a 95-kDa membrane protein.
The encoding gp95/sortilin gene (designated sort1) was mapped to the proximal part of the short arm of chromosome 1 as determined by the fluorescence in situ hybridization technique (not shown). At least one signal was displayed in 30 of 50 analyzed methaphases with 46 of the 200 chromatids (23%) being labeled. The FLpter value was 0.41 ± 0.04, corresponding to a localization at 1p13.1-p21.3.
Sequence AlignmentsComparison of the protein sequence with known sequences of the Genbank and EMBL data bases revealed no similarity to the LDL receptor but striking sequence homologies with the 210-kDa yeast receptor for carboxypeptidase Y sorting and with the M6P receptors, in particular the CI-M6P receptor.
The 1371-aa lumenal domain of Vps10p consists of two approximately
685-aa domains of 20% identity. Alignment of the entire lumenal domain
of the 95-kDa protein with Vps10p shows a similar degree of identity to
both Vps10p domains. Moreover, a particular high degree of homology is
seen between the three cysteine-rich regions comprised by Vps10p aa
562-693, Vps10p aa 1225-1355, and gp95/sortilin aa 567-709.
Alignment of the three sequences (Fig. 5A)
shows a pronounced similarity of these regions and an almost identical
spacing of ten cysteines. The structural homology between these "ten
cysteine consensus" (10CC) regions strongly indicates that they
represent a hitherto unrecognized type of extracellular module.
The homology with the 275-kDa CI-M6P receptor (Fig. 5B) is
seen in the cytoplasmic tail of gp95/sortilin, which in the C terminus contains a sequence (HDDSDEDLL) identical with the phoshorylation site
and sorting signal in the C terminus of the CI-M6P receptor and very
similar (Fig. 5B) to the corresponding stretch in the CD-M6P
receptor. Fig. 6 is a schematic presentation of the
structural similarities between gp95/sortilin, yeast Vps10p, and the
CI-M6P receptor.
Transient Expression of gp95/Sortilin and Chimeric Receptors
The sequence encoding gp95/sortilin was cleaved from
the cloned cDNA and inserted into a pMT21 vector for transfection
of COS-1 cells. Fig. 7A shows an
anti-gp95/sortilin peptide antibody staining demonstrating a high level
of gp95/sortilin expression in transfected COS-1 cells. The expression
of gp95/sortilin was confirmed by Western blotting performed on lysates
of transfected cells (Fig. 7B) and RAP affinity
precipitation of a ~95-kDa expression product from transfected
biolabeled cells (Fig. 7C). Confocal lasermicroscopy of
transiently expressed full-length protein in COS-1 cells demonstrated
localization of gp95/sortilin in endoplasmic reticulum and, in
particular, in the Golgi compartment. Staining was not detectable on
the cell surface (not shown).
To further determine the localization conveyed by the C-terminal
g995/sortilin tail, a chimeric receptor comprising the lumenal and
transmembrane domains of the IL-2 receptor (Tac/CD25) and the
cytoplasmic tail of gp95/sortilin was expressed in COS-1 cells. Fig.
8 demonstrates a predominant localization of the
chimeric receptor in the Golgi compartment (panel a) and
demonstrates in addition that it exhibits a pronounced colocalization
with the CI-M6P in Golgi as well in endosome-like vesicles
(panels b-d).
Gp95/Sortilin mRNA Transcripts in Human Tissues
The
occurrence of gp95/sortilin mRNA in vivo was estimated
by Northern blotting using total mRNA isolated from various human tissues. As demonstrated in Fig. 9, the probe hybridized
to two species of mRNA, e.g. one transcript of about 8 kb and another of about 3.5 kb. The 8-kb transcript appeared
particularly abundant, but both species were expressed at high levels
in brain, spinal cord, heart and skeletal muscle, thyroid, placenta,
and testis. In contrast, samples from lymphoid organs and cells and
from kidney, colon, and liver contained only minor amounts of specific
mRNA.
The ~95-kDa protein, designated gp95/sortilin, was purified by RAP affinity chromatography and represents the first RAP binding receptor not related to the LDL receptor family. gp95/sortilin shows no homology to the LDL receptor family receptors but displays some striking structural similarities to the recognized sorting receptors, yeast Vps10p and the CI-M6P receptor, suggesting it to be a novel candidate sorting receptor. Moreover, the homology between gp95/sortilin and Vps10p defines a previously unrecognized cysteine-rich structural protein module.
gp95/sortilin purified from human brain represents a truncated form
cleaved at the furin cleavage site after the RWRR sequence in the
N-terminal part (aa 41-44) of the receptor. The furin-specific Arg-Xaa-Arg/Lys-Arg cleavage site is found in a number of growth factor
precursors and in some growth factor proreceptors including those for
the insulin receptor, insulin-like growth factor 1 receptor, and
hepatocyte growth factor (29). Also, several reports indicate that
furin is responsible for the generation of bioactive proteins such as
transforming growth factor
(30) and the insulin receptor (31).
Future studies should determine if full-length gp95/sortilin represents
an inactive precursor form and if the 5-kDa peptide thought to be
generated by the furin-mediated cleavage has any biological function.
However, the functional relevance of the furin cleavage site seems
supported by the fact that SorLA, another RAP binding receptor, which
we recently identified (asterisk in Fig. 1), also carries a
Vps10p homologous domain truncated at an N-terminal furin site
(14).
gp95/sortilin exhibits two interesting homologies. The most extensive concerns the lumenal region of sortilin, which aligns with each of the two lumenal domains in yeast Vps10p and with similar domains in the yeast VTH2 product (3) and in Sorla-1 (14)/LR11 (15). The homology between the domains is particularly strong within a ~140-aa stretch (Fig. 5A), and the conserved positions of 10 cyteine residues in these segments strongly indicate that they represent a hitherto unrecognized type of lumenal module.
Homology to Functional Cytoplasmic Motifs in the M6P ReceptorsThe second type of homology is found in the cytoplasmic tail of gp95/sortilin, which in the extreme C terminus contains a nine-residue stretch (HDDSDEDLL) also found in the extreme C terminus of the CI-M6P receptor and similar to the EESEERDDHLL segment in the CD-M6P receptor. This nanopeptide is potentially important in gp95/sortilin as it comprises two functional sites, e.g. a casein kinase II phosphorylation site and an adjacent dileucine motif, that are known to be major functional sites for sorting of the M6P receptors between the trans-Golgi network (TGN) and late endosomes (32, 33). In fact, it is conceivable that the potential sorting of gp95/sortilin is much similar to that of the CI-M6P receptor as suggested by the confocal microscopy of the intracellular localization of the transiently expressed full-length protein and of chimeric receptors containing the gp95/sortilin tail (Fig. 8).
M6P receptor-mediated sorting involves trafficking by clathrin-coated vesicles derived from both Golgi and the plasma membrane. In contrast to plasma membrane vesicles, which are associated with type 2 assembly proteins (AP-2), formation of Golgi vesicles depends on the recruitment of AP-1 assembly proteins. In the CD-M6P receptor, the casein kinase site is essential to the formation of clathrin-coated vesicles, the initial step in TGN sorting, while the dileucine as such relates to downstream sorting events (32, 34). The di-leucine motif, in some cases adjacent to a potential kinase site, is a well known sorting motif in many membrane receptors, but in contrast to similar phosphorylation sites in the cytoplasmic tails of other receptors (35-37), the casein kinase II site in the CD-M6P receptor, and very likely in the CI-M6P receptor, mediates membrane binding of type-1 assembly proteins (AP-1) and conditions TGN vesicle formation (34). It therefore seems that the AP-1 binding of the M6P receptors relates not just to the kinase site per se but also to its unique position at the C terminus in M6P receptors. Thus, the structural identity/homology between gp95/sortilin and the M6P receptors and in particular the identical C-terminal localization of the stretch in all three receptors may reflect that gp95/sortilin is engaged in protein sorting from TGN and possibly even engaged in AP assembly.
Other Sorting Motifs in the Cytoplasmic TailIn addition to
the dileucine motif and the potential casein kinase phosphorylation
site, gp95/sortilin contains at least two candidate sorting signals.
(i) The sequence YSVL, e.g. aa 14-17 in the cytoplasmic
tail, abides by the consensus motif YXXZ, where Z
stands for either a bulky hydrophobic or an aromatic residue (35, 38).
This type of motif mediates rapid internalization and sorting of
several membrane proteins including the CD-M6P receptor and the CD3
(38, 39), and in the CI-M6P receptor, the sequence YSKV (aa 26-29 of
the cytoplasmic domain) is known to be decisive for rapid
internalization (40-42). Interestingly, a similar aromatic
residue-based sorting signal, the tetrapeptide YSSL, conveys sorting in
yeast Vps10p (3). (ii) The sequence FLVHRY (aa 9-14 in the cytoplasmic
tail of gp95/sortilin), immediately adjacent to the transmembrane
domain and partly overlapping the tetrapeptide YSVL, constitutes
another candidate internalization/sorting signal in accordance with the
overall motif (F/Y)XXXX(F/Y), which has been deduced from
active sites in both M6P receptor and LDL receptor family receptors (1,
35, 38, 39).
In summary, we have cloned, sequenced, and expressed a previously unknown receptor purified from human brain by RAP affinity chromatography. Strong homologies to known eukaryotic lysosomal sorting receptors and the putative sorting signals in its C-terminal tail strongly suggest gp95/sortilin as a receptor involved in endosomal protein sorting. In agreement with this suggestion, expression experiments demonstrated that the tail conveys colocalization with the CI-M6P receptor in both Golgi and endosome-like vesicles. Finally, the identification of this protein is of structural interest, since the established homology between gp95/sortilin, sorLA/LR11, and two separate domains in the yeast Vps10p defines a novel mammalian acidic protein module with conserved positioning of 10 cysteines. Further studies are in progress to investigate the ligand binding properties of this domain.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) X98248[GenBank].
The expert technical help by Nina Jørgensen, Helle Nielsen, Kate Nielsen, and Winnie Pedersen is greatly acknowledged.