(Received for publication, February 13, 1997, and in revised form, March 17, 1997)
From the Department of Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta T6G 2H7, Canada
CD45 is an abundant, highly glycosylated
transmembrane protein-tyrosine phosphatase expressed on hematopoietic
cells. Herein we demonstrate that two proteins of 116 kDa and 80 kDa
copurify with CD45 from mouse T cells. Microsequence analysis of the
116-kDa protein revealed high similarity to an incomplete human open
reading frame that has been suggested to correspond to the catalytic
-subunit of glucosidase II. We determined the nucleotide sequence of
the mouse cDNA and observed that it encodes a protein product
nearly identical to its human homologue and shares an active site
consensus sequence with Family 31 glucosidases. Amino acid sequencing
of the 80-kDa protein, followed by molecular cloning, revealed high homology to human and bovine cDNAs postulated to encode the
-subunit of glucosidase II. Antisera developed to the mouse
-subunit allowed us to demonstrate that the interaction between CD45
and glucosidase II can be reconstituted in vitro in an
endoglycosidase H-sensitive manner. The strong interaction between
glucosidase II and CD45 may provide a paradigm for investigating novel
aspects of the biology of these proteins.
CD45 is a high molecular mass (~180-~220 kDa) transmembrane protein-tyrosine phosphatase (PTP)1 expressed in abundance on all cells of hematopoietic lineage (1). Although CD45 is encoded by a single gene, alternative splicing of the mRNA allows for the generation of at least eight distinct isoforms of the molecule based on variable usage of exons 4-6, which encode a region of the amino-terminal ectodomain. Of particular interest, and experimental utility, is the observation that these isoforms are expressed in a cell type-specific and differentiation stage-specific manner. The variable exon repertoire expressed by different isoforms is known to greatly influence the charge properties of CD45 since the region encoded by these exons contains multiple sites for O-linked glycosylation (2). CD45 also possesses 11-18 putative N-linked glycosylation sites, which are subject to cell-specific controls, conferring further microheterogeneity upon CD45 (3). It has been estimated that approximately one third of the total molecular weight of mature CD45 is contributed by carbohydrates, which vary qualitatively and quantitatively depending on the particular cell and isoform(s) expressed (3).
The cytoplasmic domain of CD45 is identical among all isoforms and encodes two tandem PTP domains, at least one of which possesses intrinsic activity (1). A series of studies in CD45-deficient cell lines (1) and, more recently, in CD45 gene-disrupted mice (4, 5) have revealed that CD45 functions as a key regulator of maturation and activation pathways in lymphocytes. One mechanism through which CD45 appears to function is by regulating the activity of various members of the Src family of protein-tyrosine kinases (1). Despite the advances that have been made, questions remain regarding many aspects of the biology of CD45. For example, it is not known whether certain, as yet unidentified, factors are responsible for regulating the PTP activity of CD45, possibly by directing the subcellular localization of the phosphatase. It is also not clear whether, in addition to its ability to regulate Src family kinases, CD45 performs additional functions in hematopoietic cells distinct from, and possibly more general than, its ability to regulate cellular activation pathways.
One means through which insight can potentially be gained into the regulation and function of CD45 is to identify molecules with which it physically interacts. Previously, we described a 116-kDa glycoprotein of unknown identity that specifically associates with CD45 in a wide variety of hematopoietic cells (6). We have now employed affinity chromatography to purify this protein along with an additional CD45-associated protein of 80 kDa. Amino acid microsequence analysis of these proteins, followed by molecular cloning, has revealed that they are the mouse homologues of two proteins that copurify with glucosidase II activity from rat liver microsomes (7). The characterization of the association of these molecules with CD45 may provide insight into novel aspects of the biology of this complex glycoprotein.
The mouse T-lymphoma
cell lines SAKRTLS 12.1 (SAKR) and BW5147 (BW), CD45-negative variants
of these lines (SAKR/CD45 and BW/CD45
), and
a revertant of BW/CD45
expressing a truncated form of
CD45 (BW/rev) (8) were obtained from Dr. R. Hyman (The Salk Institute,
La Jolla, CA) and maintained as described previously (6). Monoclonal
antibodies I3/2.3 (9) and M1/9.32 HL2 (American Type Culture
Collection), recognizing pan-specific determinants in the CD45
ectodomain, and monoclonal antibody M1/42.3.9.8 (ATCC), directed
against class I major histocompatibility complex (MHC) molecules, were
purified and directly coupled to cyanogen-activated Sepharose 4B
(6).
Post-nuclear extracts from 5 × 107 cells resuspended in ice-cold lysis buffer (0.5% Nonidet P-40, 150 mM NaCl, 10 mM Tris, pH 7.6) were prepared as described (6) and incubated either for 1-2 h with antibody-coupled Sepharose 4B beads that had previously been blocked with 4% bovine serum albumin or for 30-60 min with 5 µl of polyclonal antiserum. Rabbit antiserum was subsequently absorbed to Protein A-Sepharose beads (Pharmacia Biotech Inc.) for 1-2 h. Immunoprecipitates were washed three to five times in high salt wash buffer (lysis buffer containing 0.5 M NaCl) followed by one or two washes in lysis buffer. Proteins were released by boiling in reducing sample buffer and subjected to electrophoretic analysis.
Polyacrylamide Gel Electrophoresis, Protein Staining, and ImmunoblottingProteins were resolved on 7.5% gels by the Laemmli method (10) and either stained directly with Coomassie Blue R-250 or transferred to Immobilon-P (Millipore) as described (6). Transferred proteins were stained with India Ink or subjected to Western blot analysis using the indicated antisera followed by Protein A-HRP (Pierce) and detected by enhanced chemiluminescence detection reagents (DuPont NEN). Apparent molecular weights of the CD45-associated proteins were estimated by comparison to the mobilities of known standard proteins (Bio-Rad).
Purification of CD45-associated ProteinsPost-nuclear
extracts from 1-2 × 1010 viable SAKR or
SAKR/CD45 cells were prepared as described above, passed
through 0.45-µm filters (Costar), and applied overnight to an
affinity column containing a 10-ml bed volume of Sepharose-I3/2 or
-M1/42. The column was washed with 200-300 ml of high salt (0.5 M NaCl) wash buffer followed by approximately 150 ml of
lysis buffer. Elution buffer (0.5% DOC, 20 mM Tris, pH
7.6) was applied, and 1-ml fractions were collected, saving 40 µl
from each for electrophoretic analysis.
The I3/2 immunoaffinity column fractions positive for 80-kDa and 116-kDa CD45-associated proteins were pooled and concentrated using Mr 30,000 cut-off Ultrafree-MC Filters (Millipore), separated on preparative gels, transferred to Immobilon-P, and stained. Protein bands were excised for amino-terminal sequence analysis or digested with 30 mg/ml cyanogen bromide in 70% formic acid prior to solid phase microsequence analysis by the Alberta Peptide Institute (Edmonton, AB, Canada).
Cloning of Glucosidase IIBased on
microsequence analysis of the p116 protein, degenerate oligonucleotides
were synthesized. These were screened for their ability to amplify DNA
from random-primed, reverse-transcribed SAKR poly(A)+ RNA.
Two primers, 5-tt(ct)ga(ag)ca(ct)ca(ag)(ac)g(ag)gc-3
(a sense
oligonucleotide corresponding to internal peptide sequence FEHQRA), and
5
-at(agct)gg(ag)tc(ag)tc(ct)ttcat-3
(an antisense oligonucleotide
corresponding to internal peptide sequence MKDDPI), were found to
amplify a 1.9-kb fragment. The PCR amplicon was cloned into the pCRII
vector (Invitrogen), and partial sequencing confirmed the presence of
nucleotide sequence in agreement with our amino acid microsequence
data. The insert was excised, labeled, and used to screen a Uni-ZAP XR
oligo(d)T-primed mouse EL4 T-cell cDNA expression library
(Stratagene). The longest clone was found to be incomplete after
sequencing. To isolate the 5
end of the cDNA, two sequential
rounds of PCR-RACE were performed with the Marathon cDNA
Amplification kit (CLONTECH) using SAKR cDNA as a template. The specific antisense PCR primers were designed to span
nucleotides 1586-1609 (first round) and 231-260 (second round) and
were used in conjunction with the adapter primer (AP1) supplied with
the kit. Products of approximately 1.6 kb (first round) and 0.3 kb
(second round) were TA cloned directly from the reaction mix. Positive
clones containing sequences overlapping those of the incomplete
cDNA were fully sequenced in both directions. Finally, the entire
open reading frame of p116 was PCR-amplified using a primer pair
spanning nucleotides 1-26 and 3065-3091, cloned into pCRII, and
sequenced.
The entire open
reading frame of human homologue 80K-H (11) was cloned by RT-PCR from
the A431 human epidermal carcinoma cell line using primers spanning
nucleotides 129-149 and 1718-1739. This amplicon was radiolabeled and
used to screen the EL4 cDNA library as described above. Five
positive clones were rescued, each with inserts of different sizes
corresponding to the 3 end of mouse glucosidase II
-subunit
(GII
) mRNA, as revealed by partial sequencing. The
longest of these, clone 80-5 (1.75 kb), was completely sequenced in
both directions. The remaining 0.6-kb 5
end of the cDNA was cloned
by PCR-RACE as outlined above, using a gene-specific primer spanning
bases 603-636. The entire open reading frame was sequenced following
RT-PCR amplification from SAKR cDNA using the primer pair 70-97
and 1681-1707.
A fragment corresponding to amino acid
residues 126-272 of GII was amplified by RT-PCR and
cloned into the pGEX-4T-3 expression vector (Pharmacia). Recombinant
GST-fusion protein from lysates of
isopropyl-1-thio-
-D-galactopyranoside-induced cultures
(Boehringer Mannheim) was recovered using glutathione Sepharose-4B
matrix (Pharmacia). Following extensive dialysis against
phosphate-buffered saline, the fusion protein was injected into a
rabbit via intramuscular and subcutaneous routes. The antiserum thus
obtained was designated anti-80.1. The same strategy was employed to
generate a second antiserum, designated anti-80.2, against a GST-fusion
protein spanning COOH-terminal amino acid residues 444-521 of
GII
. In both cases, Coomassie Blue staining of the
injected material confirmed that it was highly pure.
I3/2
immunoprecipitates were prepared from SAKR cells. Beads were washed
twice with elution buffer (to remove CD45-associated proteins) and
twice with lysis buffer, and then incubated for 1-2 h with 1 ml of
detergent-solubilized extract from 5 × 107
SAKR/CD45 cells. Immune complexes were washed and
processed for SDS-PAGE.
CD45 was immunoprecipitated from
SAKR cells and stripped of associated proteins as described for the
reconstitution assays. CD45 beads were incubated for 16 h on a
rotator at 37 °C in a 60-µl reaction mix containing 0.5 units of
endoglycosidase F (Boehringer Mannheim) in 20 mM EDTA, 0.2 M sodium acetate, pH 6.0, or 10 milliunits of
endo--N-acetylglucosaminidase H (Boehringer Mannheim) in
0.2 M sodium acetate, pH 6.0. As a control, mock-treated
samples were incubated in an equivalent volume of 20 mM
EDTA, 0.2 M sodium acetate, pH 6.0. Digests were washed
twice in high salt wash buffer and twice in regular lysis buffer.
Reconstitution assays were then performed as outlined above.
In our initial
experiments to identify proteins that physically interact with CD45, we
observed a protein with an apparent molecular mass of 116 kDa (p116)
that was present in immunoprecipitates from CD45-positive T cells but
absent from parallel immunoprecipitates from CD45-negative variants of
these cells (6). Also visible on some of our gels, although frequently
obscured by nonspecific protein bands, was a protein with an apparent
mass of approximately 80 kDa (p80, Fig. 1). Association
of these proteins with CD45 was stable in 0.5% Nonidet P-40 but not in
0.5% deoxycholate (DOC) (Fig. 1). Having failed to identify these
proteins using antibodies to known molecules, we purified p116 and p80
by exploiting the differential detergent stability of the association.
SAKR mouse T-lymphoma cells (>1010) were lysed in buffer
containing 0.5% Nonidet P-40 and passed through an immunoaffinity
column containing the same pan-specific anti-CD45 antibody (I3/2)
employed in our immunoprecipitation assays. After extensive washing in
high salt (0.5 M NaCl) buffer, fractions were eluted in
0.5% DOC. SDS-PAGE analysis of these fractions by total protein
staining revealed two major components migrating at 116 and 80 kDa and
additional minor species (Fig. 2A). To
control for nonspecific protein associations, lysates from an
equivalent number of cells from a CD45-negative variant of SAKR were
processed in an identical manner. Analysis of column eluates from this
experiment revealed only faintly staining minor species, indicating
that the 80-kDa and 116-kDa proteins are physically complexed with CD45
(Fig. 2B). As a further control, we confirmed that p80 and
p116 could not be copurified using an isotype-matched anti-class I MHC
molecule (M1/42) immunoaffinity column (Fig. 2C). Thus,
using an affinity chromatography strategy, we succeeded in copurifying
two proteins of unknown identity by virtue of their stable association
with CD45.
Microsequence Analysis and cDNA Cloning of p116 Identifies It as the Putative Catalytic Subunit of Glucosidase II
Having
purified p116 and p80, we next subjected these proteins to
amino-terminal microsequence analysis and utilized this sequence
information to search the BLAST data base (12). Sequences from
independently prepared samples of p116 revealed high homology to an
incomplete human open reading frame (ORF), GenBank accession number
D42041[GenBank], that had been randomly isolated from the immature myeloid cell
line KG1 (Table I, samples 1 and 2). This ORF predicts a
putative signal sequence upstream of our amino-terminal sequence but is
lacking an ATG start codon. Seeking further evidence that p116 is the
mouse homologue of this molecule, we obtained internal amino acid
sequence information from peptide fragments generated by cyanogen
bromide digestion. This internal sequence data closely matched the
protein sequence predicted by the human ORF (Table I, samples 3-7)
with the exception of one peptide fragment (sample 4) that could only
be partially aligned with the human protein. Intriguingly, the D42041[GenBank]
ORF recently has had a function ascribed it as the catalytic ()
subunit of glucosidase II (GII
), an
-1,3 ER lumenal
glucosidase, based on extensive microsequence analysis of a 110-kDa rat
protein with glucosidase activity and deletion of a homologous gene in
Saccharomyces cerevisiae (7).
|
Given that a full-length cDNA for GII had not been
isolated in mammalian cells, and to confirm the identity of p116 as the
mouse homologue of this molecule, we sought to clone and sequence the
corresponding cDNA from mouse T cells. Several degenerate oligonucleotide primers were synthesized based on our peptide sequence
data. Two of these amplified a 1.9-kb RT-PCR fragment from SAKR
cDNA. Cloning and partial sequencing of this amplicon revealed high
homology to D42041[GenBank]. Probing of an oligo(dT)-primed mouse T-cell
cDNA library with this PCR product led to the isolation of two
clones, each 2.9 kb in size, which appeared, from restriction analysis,
to be derived from the same mRNA. One of these was sequenced in
entirety and found to encode an ORF highly homologous to human D42021[GenBank],
containing a stop codon and downstream polyadenylation signal, but
lacking over 1 kb of 5
sequence. Two rounds of PCR-RACE led to the
isolation of a cDNA fragment that contained 11 bases in the
5
-untranslated region. Juxtaposing the first ATG triplet is a motif
that conforms well with the consensus sequence for initiation of
translation in vertebrates (13). The predicted mass of the nascent
product encoded by the longest ORF (109 kDa) and the fact that our
amino acid sequence information matches that predicted by the
nucleotide sequence at all but a single residue (Table I) strongly
argue that the cDNA sequence we obtained corresponds to that of the
p116 protein that we co-purified with CD45.
Shown in Fig. 3 in alignment with human ORF D42041[GenBank], the
product of the cDNA contains a long, hydrophobic leader peptide that appears to be cleaved at Ala-32, based on our amino-terminal sequence data and amino-terminal sequence analysis of the rat homologue
(7). An additional 22 amino acids is predicted in the mouse protein
(positions 187-208), explaining why our amino acid sequence of this
region (Table I) only partially matched the sequence of the human
protein. Consistent with our previous biochemical analysis of p116
suggesting limited N-linked glycosylation (6) and analysis
by others of an apparently identical protein possessing
N-glycans of the high mannose type (7, 14-17), a single
N-linked glycosylation consensus sequence is present in the
mouse protein. Notable also is a conspicuous 21-amino acid hydrophobic
stretch (Fig. 3, underlined), which, based on biochemical data from a number of sources, does not appear to function as a
transmembrane domain (7, 17, 18). Given that the protein product of our
mouse cDNA shares 90% identity with human ORF D42041[GenBank], which has
recently been shown to have glucosidase II catalytic function (7), we
conclude that the cDNA we have cloned encodes the -subunit of
mouse glucosidase II, a protein that stably associates with CD45 in T
cells.
Glucosidase II
That the large subunit of glucosidase II, alone,
accounts for the catalytic activity of the molecule is suggested by a
number of studies (7, 16-18). Data base analysis of mouse and human GII protein sequences revealed regions of considerable
homology to Family 31 glycosyl hydrolases. Contained within this family are a number of ancestrally related gene products, including the intestinal sucrase-isomaltase complex and lysosomal
-glucosidase (19, 20). Shown in Fig. 3 is a portion of the
carboxyl-terminal sequence of human lysosomal
-glucosidase (H-LAG)
(21) aligned with mouse and human GII
. A significant
degree of homology exists between the aligned regions of these
molecules, with 36% sequence identity between H-LAG and mouse
GII
. In contrast to this, the remaining amino- and
carboxyl-terminal portions of H-LAG (amino acids 1-477 and 777-952,
respectively) cannot be aligned with GII
sequences. Of
particular interest is the conservation of a number of residues in the
region surrounding Asp-518 in H-LAG (Fig. 3, residues 514-521).
Asp-518 has been identified by mutagenesis as the catalytic nucleophile
of H-LAG, while Trp-516 has also been shown to be essential for
catalytic function (22). These residues, together with flanking
sequences, define the catalytic consensus sequence for Family 31 glycosyl hydrolases: (G/F)-(L/I/V/M)-W-X-D-M-N-E
(19, 21-26). As can been seen from the alignments in Fig. 3, human
GII
possesses the entire catalytic consensus sequence of
acid
-glucosidase, while the mouse sequence predicts a conservative
substitution of Tyr for Phe at position 560. Although all of these
enzymes are
-glucosidases, their specific activities are distinct.
Lysosomal
-glucosidase is an acid hydrolyze that functions to cleave
-1,4 and
-1,6 linkages in glycogen, maltose, and isomaltose (27, 28), while glucosidase II has been shown to act upon
-1,3 glucosidic linkages present on immature ER glycoproteins, although it also possesses
-1,4 activity (29). Moreover, the pH optima of these enzymes appear to be mutually exclusive (15, 28). Of note, Family 31 sucrase and isomaltase possess substrate specificities and catalytic
consensus sequences similar to those of lysosomal
-glucosidase (21,
23), but exhibit neutral pH optima (30). Taken together, these
observations imply that the conserved catalytic consensus sequence in
GII
, shared among a number of apparently ancestrally
related genes encoding functionally divergent
-glucosidases, represents the active site of the molecule, while other non-conserved regions may direct the specificity of the enzyme.
Microsequence information derived from the amino terminus of
purified p80, in addition to limited sequence data from an internal cyanogen bromide-digested peptide fragment (Table II),
revealed a close relationship to a human cDNA that was originally
cloned as a potential PKC substrate, termed 80K-H (11). More recently, this protein has been proposed to function as a second () subunit of
glucosidase II (GII
), based primarily on the observation that it copurifies with the
-subunit from rat microsomes and is
resistant to biochemical separation from this larger subunit (7). The
high degree of amino acid conservation between the amino termini of
mouse p80 and human 80K-H (Table II) led us to attempt to isolate the
mouse cDNA by cross-species hybridization. A RT-PCR amplicon
corresponding to the entire coding region of human 80K-H was used to
screen a mouse T-cell cDNA library for homologous sequences. Five
independent clones were rescued from the library, and their inserts
were completely or partially sequenced. All appeared to encode an
identical protein exhibiting high homology to 80K-H, although all were
missing varying amounts of 5
sequence. PCR-RACE was employed to
isolate the 5
end of the cDNA, and PCR cloning of the entire ORF
was carried out in SAKR cells.
|
The protein sequence deduced from our cloning experiments is presented
in Fig. 4. Significantly, the mouse cDNA product
possesses sequences identical to those obtained from microsequence
analysis of p80 (Table II). Data base analysis of the complete coding
region revealed striking homology to the human 80K-H protein and an
unpublished bovine sequence (GenBank accession number U49178[GenBank]), also shown in Fig. 4, but no additional significant primary sequence homologies to other known proteins. The mouse cDNA encodes a highly acidic 59-kDa nascent protein, which lacks a transmembrane domain but
possesses a hydrophobic signal sequence that, based upon amino-terminal sequencing of the mature protein, is cleaved at position 14 in mouse
(Table II) and at the corresponding site in other species (7, 11, 31,
32). The mouse protein shares 86% amino acid identity with its human
homologue and 82% identity with the bovine product, with conservation
of a number of motifs of potential functional significance. Shared by
all species is a carboxyl-terminal His-Asp-Glu-Leu (HDEL) sequence,
which may serve as an ER-retention/retrieval signal by virtue of its
ability to interact with a specialized membrane receptor (33-35). A
region containing tandem glutamic acid repeats is also present in all
species; such negatively charged domains are common among ER proteins
and may function to allow these proteins to coordinate
Ca2+, possibly as an additional ER retention mechanism
(36-38). The unusual charge properties of this region may contribute
to the anomalously slow migration of this molecule by SDS-PAGE, as has been observed for similar proteins (11). In addition to the possible
Ca2+ binding activity of the acidic stretch, two putative
EF-hand motifs are present, which may confer high affinity
Ca2+ binding capacity (39). Finally, it is noteworthy that
the distribution of cysteine residues, clustered near the amino- and
carboxyl termini of the molecule, is entirely conserved between
species, in agreement with evidence for the presence of intrachain
disulfide linkages in the rat homologue (7). In summary, the second
cDNA we have cloned encodes a product that is highly conserved on
an evolutionary basis and possesses several hallmarks of an ER protein.
Our ability to copurify this molecule with the catalytic subunit of
glucosidase II and the major glycoprotein CD45 is in agreement with a
study in the rat system proposing that it represents the strongly
associated -subunit of mouse glucosidase II, which may function, in
part, to retain the catalytic subunit in proximity to the membrane of the endoplasmic reticulum (7).
The Association between CD45 and Glucosidase II Can Be Reconstituted in Vitro
Antisera to two distinct regions of
GII were generated by injecting recombinantly expressed
GST-fusion proteins into rabbits. Both sera (anti-80.1 and anti-80.2),
but not preimmune sera from these rabbits, were capable of blotting and
immunoprecipitating the 80-kDa protein from SAKR cells or I3/2 column
fractions (Fig. 5A and data not shown). In
addition, GII
immunoprecipitated with one antibody could
be immunoblotted with the other antibody, and vice versa
(data not shown). As an additional control, we confirmed that
immunoreactive protein is absent from immunoprecipitates prepared using
M1/42 antibody (Fig. 5B). A second antibody to CD45 that is
known to co-precipitate the 116-kDa (
-subunit) protein (6) was also
found to co-precipitate GII
(Fig. 5B). These results convincingly argue that the cDNA clone we obtained
representing GII
encodes the same molecule that we had
purified by virtue of its ability to form a stable complex with the
-subunit and CD45.
We previously reported that the interaction between CD45 and the
116-kDa (-subunit) protein could be reconstituted in
vitro by mixing affinity purified CD45 that had been stripped of
associated molecules with lysates from CD45-negative cells (6). Since the 80-kDa region of these protein-stained gels was obscured by nonspecific bands, we repeated these experiments and assayed for in vitro association of the
-subunit with CD45 using our
specific antisera. As anticipated, we found that washing CD45
immunoprecipitates with 0.5% DOC led to release of the 80-kDa
associated protein (Fig. 6, lane 3). By
incubating CD45-positive immune complexes that had been washed with DOC
with lysates from SAKR/CD45
cells, we were able to
reconstitute the association between CD45 and GII
(Fig.
6, lane 4). The association could not be reconstituted if
CD45-negative beads were incubated with the cell lysates (Fig. 6,
lane 5). In conjunction with our previous analysis of the
116-kDa protein, and consistent with the success of our purification
strategy (Fig. 2A), these data indicate that the interaction
between CD45 and glucosidase II is stable and of high affinity.
Binding of Glucosidase II to CD45 Is Dependent upon Endoglycosidase H-sensitive Oligosaccharide Linkages
The apparent strength and
high stoichiometry of association between glucosidase II and CD45, and
the resistance of this association to disruption in 0.5 M
NaCl, were surprising. Although CD45 is an abundant and highly
glycosylated protein in T cells, we failed to copurify with it
detectable levels of any other carbohydrate-processing enzymes or ER
chaperone molecules (Fig. 2A). Moreover, we failed to
copurify glucosidase II with another cell-surface glycoprotein, the
class I MHC molecule (Fig. 2C). We thus sought to
investigate the structural requirements mediating the association of
these proteins. To first eliminate the possibility that the CD45
cytoplasmic domain, which is not exposed to the ER lumen, is required
for these proteins to associate, we confirmed that the association between GII and CD45 can occur in a revertant of a
CD45-negative cell (8) that expresses reduced levels of a truncated
form of CD45 lacking most of the cytoplasmic domain (Fig.
7A). This result is in accordance with our
previous observation that the 116-kDa
-subunit also associates with
this mutant form of CD45 (6). We next asked whether immature
N-linked oligosaccharides of the high mannose, Endo
H-sensitive type are necessary for the interaction of these proteins.
We incubated CD45 that had been washed with DOC to strip it of
associated proteins with either Endo H, which releases high mannose
type linkages, Endo F, which releases all N-linkages, or
digestion buffer alone. We then assayed for the ability of
GII
from CD45-negative lysates to become associated with
deglycosylated CD45. As is clear from the data presented in Fig.
7B, our ability to reconstitute the association between
GII
and CD45 was dependent upon the presence of
N-linked carbohydrates, specifically those of the high
mannose, Endo H-sensitive type. This result indicates that immature
triantennary carbohydrate structures present on CD45 are necessary for
the association with glucosidase II; based on the known specificity of
glucosidase II, we speculate that terminal
-1,3 glucose linkages are
specifically required. We cannot, however, formally exclude the
possibility that other binding determinants on CD45 may interact with
GII
and GII
.
Glucosidase II is a neutral exo-glucohydrolase that localizes
primarily to the ER and transitional elements (40, 41) and is involved
in the processing of N-linked triantennary core glycans following their co-translational transfer to nascent polypeptide chains. Glucosidase II acts in sequence after glucosidase I has cleaved
the terminal, -1,2-linked glucose (Glc) from protein-conjugated Glc3(mannose)9(N-acetylglucosamine)2
to hydrolyze the inner two
-1,3-linked Glc units (42). The activity
of glucosidase II in the ER is counteracted by that of a
UDP-Glc:glycoprotein glucosyltransferase, which can add back a single
Glc linkage to
(mannose)9(N-acetylglucosamine)2 (7). The opposing activities of these two enzymes has been accounted
for in a "quality-control" model of ER processing, whereby incompletely folded proteins must present a monoglucose moiety to be
recognized by the folding chaperone calnexin (43-46). This scheme
ascribes a dual function to glucosidase II. By removing the first
-1,3-linked Glc, glucosidase II allows nascent polypeptides to
interact with calnexin, and by removing the innermost Glc, it regulates
the dissociation of folded proteins from calnexin and allows for their
egress from the ER.
The design of lengthy purification protocols allowed several groups to isolate glucosidase II and characterize its activity (15, 16, 18, 29, 47). However, the molecular identity of this enzyme was not revealed until recently, when Trombetta et al. reported the results of exhaustive microsequence analysis of two proteins that copurify as a heterooligomer with glucosidase II activity from rat liver microsomes (7). We now report the cloning of two homologous mouse cDNAs that encode proteins that copurify with the transmembrane PTP, CD45, in T cells.
The larger of these proteins (116 kDa in mouse, 110 kDa in rat) is a
molecule that we partially characterized as a tyrosine-phosphorylated transmembrane glycoprotein that physically associates with CD45 in a
wide variety of hematopoietic cell lines (6). Our present work reveals
that the cDNA sequence encodes a nascent protein with a mass of 109 kDa that shares 90% sequence identity with the product of an
uncharacterized human ORF (GenBank accession number D42041[GenBank]) that lacks
an initiation codon and 5-flanking sequences. An internal region near
the carboxyl termini of these mouse and human proteins contains
considerable sequence identity to Family 31 glycosyl hydrolases,
including a highly conserved catalytic consensus sequence. This
observation, combined with the finding that disruption of a gene
homologous to our cDNA in S. cerevisiae resulted in the
complete loss of glucosidase II activity (7), suggests that the protein
encoded by this cDNA is the catalytic unit of glucosidase II
(GII
). We have also been able to detect glucosidase II
activity in CD45 and GII
but not in class I MHC
immunoprecipitates (data not shown); however, in the absence of an
antibody to the catalytic subunit, it is difficult to determine if the
activity that we observe is significant.
The mouse cDNA encoding the smaller protein that we copurified with
CD45 and that was copurified with glucosidase II activity by Trombetta
et al. (80 kDa in mouse, 90 kDa in rat) shares high sequence
identity with a previously cloned human protein, termed 80K-H (11), and
an unpublished bovine sequence (U49178[GenBank]). The human cDNA was
originally isolated in an attempt to clone what appeared to be two
major PKC substrates of 80 kDa, termed 80K-H (for "high") and 80K-L
(for "low") based on their gel mobilities (48). The clone isolated
as 80K-H, however, proved to be a poor substrate for this kinase (48),
in contrast to the 80K-L protein, which was revealed to be the
biochemically similar, but genetically distinct, myristylated
alanine-rich protein kinase C substrate (48-50). More recently, the
80K-H protein was identified as one of several proteins in the 90-kDa
molecular mass range to become tyrosine-phosphorylated in response to
basic fibroblast growth factor (32). In conflict with these data
implicating 80K-H in intracellular signal transduction cascades is
information suggesting a role for this molecule as a cell-surface
receptor or co-receptor on cells of the monocyte/macrophage lineage,
conferring an ability to bind proteins modified by advanced glycation
end products, a heterogeneous series of complex end-products resulting
from the non-enzymatic reaction of free-amino groups of proteins with reducing sugars such as Glc (51). Added to this is the report from
Trombetta et al. identifying 80K-H as the -subunit of
glucosidase II, based on the finding that GII
and
GII
exist as a heterooligomer and are refractory to
biochemical separation (7).
Although it is not yet possible to completely resolve these diverse observations, a number of clues to the localization of this molecule emerge from analysis of its cDNA sequence. Highly revealing is the fact that 80K-H possesses a hydrophobic signal sequence that is cleaved at the same position in the mouse protein as in the human and rat proteins identified in the studies described above (7, 11, 31, 32). Since the mature protein lacks a transmembrane domain, this implies that the protein initially enters the secretory pathway. Also present in 80K-H, and conserved among mouse, human, and bovine, is a carboxyl-terminal HDEL tetrapeptide. This sequence has been shown to interact with a membrane receptor that functions to continually retrieve from the Golgi and return to the ER proteins containing specific carboxyl-terminal tetrapeptides (33). Moreover, mutation of the HDEL tetrapeptide found in ER proteins has been shown to favor the secretory pathway (34, 35). 80K-H has in common with other ER proteins a highly acidic region that may bind Ca2+ at low affinity, forming a "zipper" structure linking ER proteins in a loose matrix (36-38). 80K-H also possesses a pair of 12-amino acid motifs that may form two interacting helix-loop-helix EF-hand structures (39). Such motifs have been shown to coordinate Ca2+ in other ER proteins, and may allow for conformational regulation by Ca2+ levels or confer a buffering capacity on the protein (34, 35). Finally, 80K-H possesses a polar distribution of conserved Cys residues and possesses intrachain disulfide linkages (7). Taken together, these observations are consistent with 80K-H residing in the ER lumen.
Our biochemical data fit best with the suggestion by Trombetta
et al. that 80K-H represents the -subunit of glucosidase
II. We have shown that the 80-kDa protein co-purifies with
GII
and a major cellular glycoprotein. Moreover, we have
found that the 80-kDa protein can associate with glucosidase II
activity in the absence of detectable CD45 (data not shown). Although
GII
does not appear to be necessary for
GII
catalytic function (7, 16-18), it may be capable of
modulating the activity of the holoenzyme. Additionally, it may
function to retain the
-subunit in association with the lumenal face
of the ER membrane. If correct, this could explain why an 80-90-kDa
protein was consistently observed to be present in GII
preparations after exhaustive purification steps (7, 16, 18) and also
help reconcile the ability of GII
to partition to the
membrane fraction as a non-integral protein (6, 18). It is important to
note, however, that GII
(80K-H) has been purified in the
apparent absence of the larger catalytic subunit (11, 31). Our data do
not exclude the possibility that 80K-H may adhere to a cell-surface
protein(s) in certain cell types, where it may contribute to the
binding of proteins modified by advanced glycation end products. The
basis for detecting an admittedly small subpopulation of
tyrosine-phosphorylated 80K-H in a previous study, however, is unclear
(32); we have been unable to detect tyrosine phosphorylation of this
protein in response T-cell receptor-stimulation of CTL clone cells
(data not shown).
An important question arising from our study is why are we able to
copurify glucosidase II with CD45. Our evidence that the interaction
between glucosidase II and CD45 is dependent on high mannose
triantennary core oligosaccharide linkages supports the hypothesis that
glucosidase II interacts with immature CD45 that presumably contains
terminal -1,3 Glc linkages. The simplest explanation for our results
is that GII
and GII
interact with CD45 in
the same way that they do with all immature glycoproteins; however,
because CD45 is a highly expressed, multivalent ligand, we are able to
detect the association. Nevertheless, our observations that we are
unable to copurify glucosidase II with class I MHC molecules or detect
the association of these ER proteins with class I molecules by blotting
or enzymatic assay, suggest that there may be something special about
the way in which glucosidase II interacts with CD45. Moreover, it is
notable that we failed to copurify detectable levels of any other
enzyme or receptor that recognizes oligosaccharides present on CD45, or
any other ER chaperone molecule that interacts with CD45.
It is also possible that our data may relate to novel aspects of CD45
maturation or function. That maturation of CD45 is highly unconventional in some cell types is revealed by a report demonstrating the presence of high mannose triantennary oligosaccharide linkages on
cell-surface CD45 derived from immature thymocytes (52). It has further
been noted that in a resting T-cell line approximately 30% of total
cellular CD45 is retained in the cis-Golgi (53). Within
minutes of activating this cell with anti-Thy-1, this intracellular pool of CD45 was found to disperse, partitioning to an insoluble intracellular fraction. It is tempting to speculate that a novel function of glucosidase II may be to target and/or retain a portion of
newly synthesized CD45, possibly in a transitional ER compartment. It
is not clear whether the PTP activity of the cytoplasmic domain of CD45
may be required to perform a function in this particular subcellular
microenvironment; however, it is interesting that the cytosolic
phosphatase, PTP1B, has been shown to localize to the endoplasmic
reticulum (54). In addition, tyrosine kinase and phosphatase inhibitors
can block vesicular transport (55). Also intriguing, particularly in
view of the fact that GII may bind Ca2+, is
a report detailing defects in regulation of intracellular Ca2+ by cells deficient in CD45 (56) and another indicating
that re-distribution of intracellular pools of CD45 correlates with changes in the ability of T cells to liberate Ca2+ from
internal stores (57). Further investigation of the interactions between
CD45 and glucosidase II may provide novel insights into the functions
of these molecules.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U92793[GenBank] and U92794[GenBank].
We thank Dr. Robert Hyman for the
CD45+ and CD45 BW5147 and SAKRTLS 12.1 cell
lines and Yew Hon Lai for assistance with the DNA hybridizations. We
also acknowledge the excellent sequencing work of the Alberta Peptide
Institute and the University of Alberta, Department of Biochemistry DNA
Sequencing Facility and extend our gratitude to Dr. Kevin Kane for his
critical review of the manuscript.