|
|
|
|||||||
|
|||||||||
(Received for publication, February 8, 1995; and in revised form, April 27, 1995) From the
Antisera raised against detergent-extracted membrane fractions
from the human intestinal epithelial cell line Caco-2 were used to
screen a human colon cDNA library in a bacteriophage expression vector.
This led to the identification, molecular cloning, and sequencing of a
novel plasma membrane protein (p137) which was present in approximately
equal amounts on the basolateral and apical surfaces of the cell. The
pattern of extraction of p137 from membranes by Triton X-114 and its
release from membranes after incubation with
phosphatidylinositol-specific phospholipase C were consistent with it
being a glycosylphosphatidylinositol-anchored membrane protein. Using
antibodies raised against bacterial fusion proteins, it was shown that
p137 was present on the cell surface as a reducible homodimer of 137
kDa subunits. There was constitutive release of p137 into the culture
medium as a non-reducible 280-kDa entity. Pulse-chase experiments
showed that newly synthesized p137 appeared at the basolateral side of
a Caco-2 cell layer before appearing at the apical domain.
Domain-specific surface biotinylation of Caco-2 cells at 4 °C,
followed by chasing at 37 °C, demonstrated that p137 is capable of
transcytosing in both directions across Caco-2 cells. The unusual
plasma membrane domain distribution of this
glycosylphosphatidylinositol-linked protein and its transcytosis
characteristics demonstrate the existence of a previously
uncharacterized apical to basolateral transcytotic pathway in Caco-2
cells.
The plasma membrane of polarized epithelial cells consists of
apical and basolateral domains of unique protein and lipid composition
(reviewed in (1) ). Plasma membrane polarity is achieved by
sorting newly synthesised membrane constituents, followed by targeting
to the correct membrane domain. Sorting can occur either in the trans-Golgi network (TGN) ( Transcytosis not only contributes to the
development and maintenance of membrane polarity, but also provides an
endocytic route whereby specific membrane proteins and bound ligands
internalized at one side of a polarized cell may be transported and
delivered to the other. The best characterized transcytotic route is
that taken by the polymeric immunoglobulin receptor (pIgR) from the
basolateral to the apical domain of a variety of polarized epithelial
cells(5, 6) . In liver, salivary gland and mammary
tissue transcytosis of the pIgR results in release of the ectodomain of
the receptor together with pIgA into bile, saliva, and colostrum,
respectively, and makes a major contribution to the immune system in
these secretions (reviewed in (7) ). Sequence motifs in the
cytoplasmic tail of the pIgR required for transcytosis have been
defined (5, 9, 10) as well as the endocytic
compartments through which it travels (8) and the proteolytic
events occurring when it arrives at the apical cell surface and is
released(11) . In contrast, much less is known about the apical
to basolateral transcytosis of any single membrane protein. Among the
best described apical to basolateral transcytotic routes are those of
the neonatal rat gut immunoglobulin receptor, FcRn (12) and of
FcRII-B2 transfected into MDCK cells(13) . Despite the
availability of much sequence and structural information, in neither
case have the targeting motifs in the cytoplasmic tail of the receptor
been well defined. Transcytosis of cobalamin has also been investigated
in adult gut and the human colon adenocarcinoma-derived cell line
Caco-2,(14) , but this involves an indirect route whereby
cobalamin internalized from the apical surface complexed with intrinsic
factor is transferred to endogenously synthesized transcobalamin II
prior to secretion from the basolateral side. Such apical to
basolateral transcytic routes in enterocytes lining the gut are of
particular interest since, if better defined, they could be exploited
for efficient drug delivery(15, 16) . In the
present study we attempted to identify and clone endogenous membrane
proteins which would be candidates for apical to basolateral
transcytosis in Caco-2 cells. We have previously shown that some
antisera raised against plasma membrane fractions from Caco-2 cells
react with subsets of proteins common to the apical and basolateral
domains(17) . Such proteins are candidates for apical to
basolateral transcytosis, since if internalized from the apical surface
they would not necessarily carry signals allowing sorting back to the
apical domain. To clone these proteins we screened an appropriate cDNA
library in a bacterial expression vector and sorted the clones using
antibodies affinity purified on individual expressed fusion proteins.
This approach has been successfully used previously to identify Golgi
membrane proteins when starting with an antiserum raised against Golgi
membranes(18) . In the present study we have identified and
cloned a novel membrane protein present in both the apical and
basolateral plasma membrane domains of Caco-2 cells, and show that it
is transcytosed in both directions.
A human colon carcinoma
Figure 1:
Immunoblots of apical (A) and
basolateral (B) membranes of Caco-2 cells using affinity
purified antibody 660. 5-µg membrane samples were run on 7%
SDS-PAGE in the presence or absence of 10 mM DTT, blotted, and
incubated with affinity purified antibody 660 at 1:500 and bands
visualized by ECL. The lower molecular weight bands visible in the left-hand lane appear to be proteolytic degradation products
of p137, the amounts of which increased considerably after repeated
cycles of freezing and thawing of membrane
fractions.
Figure 2:
Properties of p137. a, a Caco-2
apical membrane fraction was treated with 0.1 M Na
Figure 3:
Summary of partial cDNA clones encoding
p137 and of antisera raised against pGEX proteins. The initial clone
isolated from the human colon cDNA library in
Figure 4:
Nucleotide and derived amino acid sequence
of p137. The numbering of the amino acids is based on the position of
the first predicted methionine residue. The underlined sequence represents the moderately hydrophobic region which may
signal GPI addition. The two broken underlines indicate the
regions which, relative to each other, show 23% identity of amino
acids. Amino acids typed in bold represent potential sites for
attachment of N-linked oligosaccharides.
An
ORF of 649 amino acids was evident within the nucleotide sequence,
commencing with an initiating methionine codon 202 bases from the 5`
end of the contiguous cDNA sequence flanked by a Kozak consensus
sequence(40) . It should be noted that this region of sequence
agreed exactly with that of the first of the expressed sequence tags
described above providing independent confirmation of its accuracy.
Analysis of the derived primary amino acid sequence showed it to be
enriched overall in glutamate (9.6%) and glutamine (10.9%). There are
two regions enriched in proline and glutamine residues, which show 23%
identity to one another (amino acid residues 310-370 and
431-488; broken underline in Fig. 4) using
similarity investigation program. The derived protein sequence contains
three potential N-glycosylation sites, NXS/T (where X cannot be P). A single moderately hydrophobic region
(residues 583-613; underlined in Fig. 4),
occurred toward the carboxyl terminus. This moderately hydrophobic
region is interrupted by polar residues and two charged His residues
and does not conform to a conventional transmembrane region. Analysis
of this region and surrounding residues based on the sequences of known
GPI-anchored membrane proteins(41, 42) , suggests the
possibility of a GPI anchor cleavage/attachment
site,`` The predicted molecular
mass of the derived amino acid sequence without post-translational
modification was 72,751 Da, considerably smaller than the observed
molecular mass of p137 on SDS-PAGE. The possibility that the protein
was running anomalously was suggested from the observed size of the
glutathione transferase fusion proteins obtained after subcloning
inserts C and 15 into pGEX-1N. Each of these fusion proteins appeared
to have an apparent molecular weight on SDS-PAGE approximately double
that predicted from the encoded sequence (data not shown). To resolve
this issue, a cDNA construct covering nucleotides 1-2252 was
prepared and expressed after transcription and translation in
reticulocyte lysate. On SDS-PAGE, the molecular mass of the protein
produced was close to 137 kDa, considerably higher than predicted from
the derived ORF (data not shown). Immunoblotting of the translation
product showed that it reacted with affinity purified antiserum 660
(data not shown). To show that initiation did not start upstream of the
predicted initiating methionine, a cDNA construct missing bases
1-128 of the predicted 5`-UTR was transcribed and translated in
the reticulocyte lysate. This also resulted in the production of a
protein of the same size (
Figure 5:
In vitro transcription and
translation of DNA encoding for p137. pBluescript containing bases
129-2252 of the full-length contiguous cDNA encoding p137 was
transcribed and translated as described under ``Experimental
Procedures.'' Approximately 1 µg of mRNA was
translated per lane. Lane 1, Caco-2 apical plasma membranes; lane 2, translated product, blotted with affinity purified
antibody 660. Lanes 3-7, autoradiographs of
[
Northern blot analysis
showed that transcripts for p137 were present in all the tissues
examined and in each case, except testis, two transcripts of 3.4 and
2.7 kb were observed (Fig. 6). In testis, two additional
transcripts of 5.3 and 2.0 kb were also present.
Figure 6:
Northern blot of various human tissues.
The Northern blot contained
Figure 7:
Shedding of p137 into the medium. a, [
The
post-synthetic delivery of p137 to the cell surface was examined by
pulse-chase labeling followed by preparation of apical and basolateral
plasma membrane fractions and immunoprecipitation(17) , using
affinity purified antibody 660. In addition, the tissue culture media
in the apical and basolateral chambers present during the chase were
also immunoprecipitated. It was found that newly synthesized p137 was
first detected as a 280-kDa entity in the basolateral culture medium
within 1 h of the start of the chase (Fig. 8). p137 could just
be detected in the apical plasma membrane domain within 1 h of the
start of the chase. By 2 h from the start of the chase fairly equal
amounts of radiolabeled p137 were present in both apical and
basolateral membranes, but there was still a considerable excess of the
non-reducible 280-kDa protein in the basolateral culture medium
compared with the apical culture medium (Fig. 8).
Figure 8:
The
post-synthetic transport of p137 in Caco-2 cells. Caco-2 cells grown on
filters for 7 days were pulse-chased with
[
Figure 9:
Domain specific biotinylation and
transcytosis. Caco-2 monolayers grown on filter supports were
biotinylated either on the apical or basolateral surface and analyzed
for the release of p137 into the media as described under
``Experimental Procedures.'' Medium from three
filters was used for tracks 1-8 on the gel and 25% of
the cells from three filters for tracks 9-12. a, Caco-2 monolayers were biotinylated on the apical (A, lanes 1, 2, 5, 6, 9, and 10) or basolateral (B, lanes
3, 4, 7, 8, 11, and 12) surface. p137 dimer was detected in the membranes after
sample preparation in the presence (lanes 9 and 11)
or absence (lanes 10 and 12) of DTT. Shedding of
biotinylated p137 dimer was examined after incubation at 4 °C for
24 h (lanes 1-4) or 37 °C for 24 h (lanes
5-8). Apical (A) medium was immunoprecipitated for lanes 2, 4, 6, and 8 and
basolateral (B) medium for lanes 1, 3, 5, and 7. b, p137 shed from the same surface
domain that was biotinylated is expressed as a percentage of the total
p137 that was originally biotinylated (
In the present study we have developed a strategy to identify
and clone membrane proteins capable of transcytosis in polarized
epithelial cells. This strategy was based on the availability of
antisera to plasma membrane fractions which recognized a subset of
common proteins present on opposite cell surface domains. The antisera
were used to identify and sort clones from a cDNA library in a
bacterial expression vector, in much the same way as has previously
been used in the molecular cloning of membrane proteins in the Golgi
complex(18, 43) . The only general strategy previously
available for identifying transcytosed membrane proteins involved the
use of a ricin-resistant MDCK cell line(44) , in which it was
subsequently shown that there was a sorting defect which resulted in
abnormal surface distribution of at least one of the identified
transcytosed proteins(45) . The present strategy avoids this
difficulty since it does not require a mutant cell line. The
membrane protein, p137, identified and characterized in the present
study is a surprising example of a transcytosed protein in that it
appears to be attached to the plasma membrane by a GPI anchor. In some
polarized epithelial cell lines such as MDCK, it is well established
that newly synthesized GPI-anchored membrane proteins are co-sorted in
the TGN, together with glycosphingolipids, before delivery to the
apical surface(46) . Indeed the GPI anchor has been identified
as an apical targeting signal in these cells(3, 47) ,
although the situation is complicated by it acting as a basolateral
targeting signal in Fischer rat thyroid epithelial cells(48) .
In yet other cells, all apically targeted newly synthesized membrane
proteins are first delivered to basolateral cell surface(2) .
This is true in hepatocytes where it has been shown that even
GPI-anchored proteins such as 5`-nucleotidase first appear at the blood
sinusoidal surface before transcytosis to the bile canalicular
surface(49) . In Caco-2 cells both biosynthetic delivery
pathways to the apical surface exist(50, 51) , and
there is evidence for the presence of some GPI-anchored proteins such
as alkaline phosphatase on the basolateral as well as the apical cell
surface domain(17, 51) . In the present study it was
found that newly synthesized p137 was clearly present on the
basolateral side of the cell in the culture fluid as what appeared to
be a 280-kDa dimer at early time points when it was only just possible
to detect it at the apical surface. The simplest explanation of this
data is that newly synthesized p137 is first delivered basolaterally
from where it may be shed or transcytosed to the apical surface. The
derived amino acid sequence of p137 was in agreement with the
biochemical data in suggesting that the protein may be GPI anchored to
the membrane with the predicted anchor attachment site at the Ala
residue 565. Although all endogenous GPI-anchored membrane protein
precursors cloned to date possess a cleavable amino-terminal
hydrophobic signal sequence, such a signal sequence is not absolutely
required for translocation across the endoplasmic reticulum
phospholipid bilayer(52) . The predicted open reading frame for
p137 encodes no such signal, yet in vitro transcription/translation experiments have shown that the protein
can be incorporated into microsomes. At the cell surface p137 is
present as a disulfide bonded dimer. It is most likely that an
additional covalent bond must also be formed between the subunits since
p137 is constitutively released into the culture medium as a
non-reducible 280 kDa entity, presumably a non-reducible dimer. There
are many examples of GPI-linked proteins which are secreted including
decay accelerating factor(53) , carcinoembryonic
antigen(54) , GP2(55) , and
melanotransferrin(56) . Within the GPI attachment domain of
p137, there are a number of sites that are potential substrates for GPI
anchor-degrading enzymes including PI-PLC and phospholipase D. Although
p137 was shown to be sensitive to PI-PLC, there is no evidence that
this enzyme has an extracellular location. It is not yet clear which
enzymic mechanism results in shedding of p137. The endocytosis and
transcytosis of GPI-anchored proteins is not well understood.
GPI-anchored proteins interact poorly with clathrin-coated pits, due to
the absence of the necessary cytoplasmic tail, and for this reason were
thought to be endocytosed poorly or not at all. Recently it has been
suggested that they may be clustered in caveolae, flask-shaped
invaginations characterized by the presence of the integral membrane
protein, caveolin(57) , although there is some dispute as to
whether caveolae act as a route of internalization from the cell
surface(58) . In Caco-2 cells there have been no reports of
caveolae, and we found no evidence for caveolin being present by
immunoblotting membrane fractions (data not shown). Irrespective of the
route of endocytic entry, the present experiments have demonstrated
that p137 may be transcytosed in both the basolateral to apical and
reverse directions. While the former direction is consistent with the
route of delivery of some apical plasma membrane proteins in Caco-2
cells, the latter suggests the existence of a novel apical to
basolateral transcytosis route. The initial lag and slow time course of
apical to basolateral transcytosis of p137 is similar to the
transcytosis of cobalamin(14) , although this molecule has to
be transferred intracellularly from one carrier protein (intrinsic
factor) to another (transcobalamin II). A small amount of apical to
basolateral transcytosis of nerve growth factor occurs in rat ileum (59) and non-selective transcytosis of epidermal growth factor
in this direction has been reported in MDCK cells(60) . Some
apical to basolateral transcytosis of ricin has also been observed in
Caco-2 cells(61) , although the toxicity of the lectin
precludes experiments lasting longer than 2-3 h. It is not clear
whether the apical to basolateral transcytosis of p137 in Caco-2 cells
is non-selective, for example as a result of failing to be correctly
recycled from a common subapical early endosome(4) , or whether
it identifies a new selective route. The fact that so much apically
labeled p137 can be transcytosed (
The
nucleotide sequence(s) reported in this paper has been submitted to the
GenBank[GenBank]and ID HSGPI137.
Volume 270,
Number 35,
Issue of September 01, pp. 20717-20723, 1995
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
)prior to separate
vesicular transport to the apical and basolateral domains or, after
initial delivery to the basolateral surface, by transcytosis of
selected proteins to the apical domain(2) . The extent to which
one or other sorting site is used is dependent on cell type and the
individual protein(3) . Maintenance of membrane polarity is
further aided by recycling most endocytosed membrane proteins to the
domain from which they are internalized, despite the existence of
common endosomal compartments which can be accessed from both sides of
the cell(4) .
Cell Culture, Radiolabeling, and Isolation of Membrane
Fractions
Caco-2 cells were grown as described previously (17) and were used between passage number 90 and 110, either 7
days (filters) or 3 days (flasks) after subculture. For steady-state
radiolabeling of proteins, cells in 75-cm
flasks were
rinsed in PBS (145 mM NaCl, 7.5 mM NaHPO
, 2.5 mM NaHPO, pH 7.4) and then incubated
overnight at 37 °C with 1 mCi of
[S]methionine (1205 Ci/mmol; DuPont NEN) in 5 ml
of methionine-free Dulbecco's modified Eagle's medium (Life
Technologies, Inc.) containing 10% dialyzed fetal calf serum.
Pulse-chase radiolabeling of proteins and the preparation of apical and
basolateral membrane fractions from filter-grown cells were as
described previously(17) . For steady-state radiolabeling of
GPI anchors, cells in 75-cm
flasks were rinsed in PBS and
then incubated overnight at 37 °C with 0.125 mCi of myo-[2-
H]inositol (16.5Ci/mmol; Amersham
International, Amersham, Bucks, United Kingdom) in 5 ml of
inositol-free Dulbecco's modified Eagle's medium containing
10% dialyzed fetal calf serum.Immunoprecipitation, SDS-PAGE, and
Immunoblotting
Rabbit polyclonal antisera to Caco-2 membrane
fractions (anti-apical antibody 630; anti-basolateral antibody
620)(17) , and antibodies against bacterial fusion proteins (18, 19, 20, 21) were raised and
affinity purified as described previously. A mouse monoclonal antibody
to caveolin (22-kDa substrate; clone 2283) (22) was from
Affiniti Research Products Ltd., Nottingham, U.K. and was used at
1:1000 dilution in immunoblots. Immunoprecipitation of whole cells,
membrane fractions, and tissue culture media after
[S]methionine or
[
H]inositol labeling were as
described(17) , except that affinity purified antibodies were
used. When necessary, samples were subjected to 7 or 10%
SDS-PAGE(23) . To identify
[S]methionine-labeled proteins gels were exposed
for 1 week at -70 °C against Kodak X-Omat film (Eastman Kodak
Co.) and to identify [
H]inositol-labeled proteins
gels were exposed for 3 days against a tritium detecting image plate
and analyzed on a Fujix BAS 2000 bio-imaging analyzer (Fuji Photo Film
Co., Japan). Immunoblotting of non-radioactive proteins was carried out
after SDS-PAGE essentially according to Burnette(24) .Cell Surface Biotinylation
Domain-specific
biotinylation followed by immunoprecipitation, SDS-PAGE, and
visualization by horseradish peroxidase-streptavidin using the Amersham
ECL blotting system were carried out according to Neame and
Isacke(25) . To study the transcytosis of p137, domain-specific
biotinylation of Caco-2 monolayers was carried out at 4 °C, and
then the cells were incubated at 37 °C for different periods of
time. In control experiments, the cells were kept at 4 °C. The
basolateral and apical media were collected, immunoprecipitated and
analyzed by SDS-PAGE followed by visualization with horseradish
peroxidase-streptavidin.Phase Partioning in Triton X-114, Sodium Carbonate
Extraction, and Treatment with PI-PLC, Pronase, and
N-Glycanase
Cells labeled overnight with
[S]methionine were cooled on ice prior to
harvesting in PBS and then solubilized in preclouded 1% Triton X-114,
and the soluble supernatant was subjected to phase partitioning
according to Bordier(26) . The insoluble pellet, detergent, and
aqueous phases were immunoprecipitated as described above. The nature
of membrane association was also assessed by the addition of sodium
carbonate (0.1 M at pH 11.5), which strips peripherally bound
proteins from membranes. Purified PI-PLC was kindly provided by Dr. M.
Low (Columbia University, New York) and used as described
previously(17) . The presence of the CRD, an epitope appearing
on some GPI-linked proteins after PI-PLC treatment, was examined by
immunoblotting with a rabbit polyclonal anti-CRD antibody (RP189 raised
against PI-PLC solubilized pig membrane dipeptidase), kindly provided
by Dr. N. Hooper (Department of Biochemistry, University of Leeds,
Leeds, U.K.) and used according to Zamze et al.(27) .
cDNA Library Screening and DNA Sequencing
All DNA
manipulations were carried out according to Sambrook et al.(28) . A human colon cDNA library constructed in gt11
was purchased from Clontech (Palo Alto, CA) and used for immunological
screening (21) with antisera 620 and 630 mixed equally, at a
dilution of 1:100. Following affinity purification of the mixed
antisera on individual fusion proteins (18, 20, 21) and subsequent immunoblotting of
apical and basolateral membrane fractions, one clone (clone C),
containing a cDNA insert (insert C) encoding a protein (p137) found in
both membrane fractions, was selected for further analysis. Insert C
(1076 bp) was
P-labeled by random priming and used to
screen a human colon 5'STRETCH
gt11 library (Clontech) in
order to obtain additional cDNA sequence. Positive cDNA clones from
these screens were assessed for the presence of additional sequence by
polymerase chain reaction using appropriately designed primers within
the insert sequence and the vector(21) . Using these procedures
a second clone was isolated containing a cDNA insert (insert 15; 1885
bp) which overlapped insert C with an extra 596 bp at the 5` end and
214 bp at the 3` end, and a third clone containing a cDNA insert
(insert 40; 2200 bp) which overlapped with the 3` end of insert 15 by
1029bp, and confirmed the identity of the stop codon. To find the 5`
end of the predicted ORF, advantage was taken of the internal EcoRI site 239 bp downstream of the 5` end of insert 15. The
239-bp EcoRI fragment was isolated, radiolabeled, and used to
rescreen the human colon 5` STRETCH cDNA
gt11 library, identifying
a clone containing an insert (insert 9; 1615 bp) with a predicted start
codon. Cloning of a partial EcoRI digest of clone 9 allowed
sequencing across the internal EcoRI site in the cDNA insert.
The final DNA sequence was determined throughout from both strands
after subcloning the inserts into pBluescript KS± (Stratagene
Cloning Systems, San Diego, CA), constructing appropriate nested
deletions, and sequencing by the dideoxy chain termination method (29) using Sequenase version 2.0 with Universal and Reverse
primers (United States Biochemical Corp.).
In Vitro Transcription and Translation
pBluescript
clones containing bases 1-2252 and 129-2252 of the full-length
contiguous cDNA encoding p137 were linearized with BamHI and
mRNA transcribed from the T7 promoter and translated using an
mCAP RNA capping kit (Stratagene) and nuclease-treated
rabbit reticulocyte lysate translation kit (Promega) according to the
manufacturer's instructions. Incorporation into microsomes was
assessed by inclusion of canine pancreatic microsomal membranes
(Promega) in the translation mix as described by the manufacturer.
Proteins were judged to be translocated into microsomes if protected
from incubation with proteinase K, 0.1 mg/ml, 1 h, 4 °C. Addition
of 1% Triton X-100 resulted in complete proteolysis. Proteinase K was
inactivated by addition of 1 mM phenylmethylsulfonyl fluoride
prior to analysis by SDS-PAGE.
Northern Blotting
A human Northern blot II
(Clontech) with tracks containing 2 µg of poly(A) RNA from each of colon, small intestine, testis, ovary, prostate,
pancreas, spleen, and peripheral blood leukocytes was probed according
to the manufacturer's instructions, using a
P-random
primed, labeled 1647 bp EcoRI fragment of insert 15 and
washing at increasingly high stringency.
Computer Analysis of DNA Sequence
The cDNA
sequence was analyzed using the computer programmes of
Staden(30) . Hydropathy analysis was performed in accordance
with Kyte and Doolittle(31) , with a window size of 11 amino
acids and a plot interval of 3 amino acids. To identify an ORF within
the DNA sequence, use was made of nucleotide interpretation program for
the detection of uneven base frequencies, with a window size of
11(30) . Secondary structure predictions were made using the
Robson prediction program in nucleotide interpretation program (32) and similarity investigation program(33) .
gt11 cDNA library (
1.5
10
independent recombinants) was screened with a
mixture of two antisera (coded 620 and 630), raised against isolated
apical and basolateral plasma membrane fractions prepared from Caco-2
cells. Seventeen positive plaques were picked and plaque-purified.
Antibodies were affinity purified from the original antisera on
individual gt11 fusion proteins and used to immunoblot Caco-2
plasma membrane fractions and to sort clones into epitope groups. In
this way, clones encoding one apical plasma membrane protein (80 kDa;
identified by three affinity purified antibodies), one basolateral
membrane protein (64 kDa; identified by two affinity purified
antibodies), and two common proteins (105 and 137 kDa; each recognized
by one affinity purified antibody) were identified (data not shown).
Antibodies affinity purified on 10 of the original 17 positive clones
did not give clear, unique immunoblotting signals with membrane
fractions.
Identification and Properties of p137
The cDNA inserts
(insert C) encoding part of the 137-kDa membrane protein (p137) was
subcloned into pGEX-1N(19) , and an antiserum (coded 803)
prepared against, and immunoaffinity purified on, the resulting
gluthathione S-transferase fusion protein. Affinity purifed
antiserum 803 reacted with p137 on immunoblots of Caco-2 cell plasma
membrane fractions in agreement with the data from blots with
antibodies from antisera 620/630 affinity purified on the clone C
fusion protein (data not shown). A subsequently prepared, affinity
purified antiserum (coded 660), raised against a pGEX-1N fusion protein
encoded by part of an overlapping clone (the large EcoRI
fragment of insert 15), also reacted with p137 in membrane fractions,
but was of higher avidity than antiserum 803 and therefore was used in
subsequent experiments. The reaction of affinity purified antiserum 660
with p137 on immunoblots of reduced samples of both apical and
basolateral membrane fractions from Caco-2 cells is shown in Fig. 1. The immunoblots suggested that the concentration of p137
in each membrane fraction was approximately equal. This distribution of
p137 could not be accounted for by cross-contamination of the two
membrane fractions, since this had previously been estimated as <10%
by analysis of marker enzymes(17) . When non-reduced samples of
membrane fractions were immunoblotted with antiserum 660, a major band
of 280 kDa was observed in each membrane fraction suggesting that p137
normally exists in the plasma membrane as a disulfide bonded dimer (Fig. 1). p137 was found to be in the pellet following
NaCO treatment and centrifugation, showing that
it was neither a soluble nor peripheral membrane protein (Fig. 2a). Extraction of
[S]methionine-labeled Caco-2 cell membranes with
Triton X-114, followed by temperature-induced phase separation (26) and immunoprecipitation with antibody 660, gave anomolous
results. p137 was found to partition mostly into the insoluble pellet
formed at 0 °C and the aqueous phase formed at 30 °C with
negligible amounts in the detergent phase (Fig. 2b).
Such anomalous distribution, upon Triton X-114 phase separation, is
characteristic of some GPI-anchored proteins(34, 35) .
Further evidence for the presence of a GPI anchor on p137 was obtained
by incubating [
S]methionine-labeled Caco-2 cell
membranes with PI-PLC for 1 h at 37 °C, which resulted in the
release of most of the protein into a soluble fraction (Fig. 2c). No change in molecular weight on SDS-PAGE
was seen after removal of the acylglycerol moeity, but this is a
frequently observed phenomenon(36) . The small amount of p137
seen in the soluble phase after control incubation in the absence of
exogenous PI-PLC (Fig. 2c) may be explained by the
presence of endogenous PI-PLC or other enzymes cleaving p137 from the
membrane (see also below). This may also explain the observation of
some p137 in the aqueous phase after Triton X-114 extraction and phase
separation. No reaction with the anti-CRD antibody (27) was
observed after PI-PLC treatment in the present experiments. However,
the antiserum used does not react with all PI-PLC-cleaved membrane
proteins(37) . Confirmation of the presence of a GPI anchor was
obtained by [
H]inositol labeling followed by
immunoprecipitation (Fig. 2d). Neither incubation of
Caco-2 cells with tunicamycin prior to
[S]methionine labeling nor treatment of membrane
fractions with Pronase under the conditions of PI-PLC treatment or N-glycanase had any apparent effect on the molecular mass of
p137 determined by SDS-PAGE (data not shown).
CO as described under
``Experimental Procedures,'' the resulting soluble (S) and insoluble (P) phases subjected to 7%
SDS-PAGE, blotted, and incubated with affinity purified antibody 660.
p137 was visualized by ECL (basolateral membranes gave the same
result). In b and c, Caco-2 cells were labeled with
[S]methionine and after treatment as described
below were immunoprecipitated with affinity purified antibody 660
followed by SDS-PAGE and autoradiography. b, cells were
subjected to differential partitioning in Triton X-114 to produce an
insoluble pellet (P), soluble (S), and detergent-rich (D) fractions; c, membranes were incubated in the
presence (+) or absence(-) of bacterial PI-PLC, and
separated into soluble (S) and insoluble (P)
fractions. In d, Caco-2 cells were labeled with
[
S]methionine (lanes 1 and 3)
or [
H]inositol (lanes 2 and 4)
and after immunoprecipitation of solubilized cells (lanes 1 and 2) or culture medium (lanes 3 and 4) with affinity purified antibody 660, were subjected to
SDS-PAGE followed by analysis on a Fujix BAS 2000 bio-imaging analyzer.
All samples were treated with DTT prior to SDS-PAGE. Only the portion
of the gel containing p137 is shown in each
case.
Cloning and Sequencing of p137
Starting with the
initially identified partial cDNA clone (insert C), a strategy was
developed to obtain cDNA clones encoding the complete sequence of p137
( Fig. 3and ``Experimental Procedures''). DNA
sequencing of overlapping clones (Fig. 3) resulted in the
contiguous cDNA sequence and deduced amino acid sequence shown in Fig. 4. Comparison of the nucleotide sequence with the EMBL,
GenBank and OWL data bases, showed it to be novel, although seven
regions showed identity with human expressed sequence tags to unknown
genes recorded in the EMBL data base. These expressed sequence tags
were as follows: accession no. T24120, ID HS1207, 100% identity to
nucleotides 125-471; accession no. M85418, ID
HSXTO1933(38) , 100% identity to nucleotides 389-603 when
reversed and complemented; accession no. T23613, ID HS6137, ()100% identity to nucleotides 1258-1591 when reversed
and complemented; accession no. T32790, ID HS79013, ()97%
identity to nucleotides 1326-1587; accession no. T10644, ID
T10644 (39) , 94% identity to nucleotides 2456-2818;
accession no. T31552, ID HS55215, 97% identity to
nucleotides 3017-3268; accession no. T 10313, ID HS313, 95% identity to nucleotides 3034-3268. In addition, a
mouse-expressed sequence tag, accession no. Z36353, ID MM49 ()showed 80% identity to nucleotides 2604-2848.
gt11 was clone C.
Subsequently, overlapping clones 15, 40, and 9 were isolated as
described under ``Experimental Procedures.''
Positions of the clones relative to the final contiguous sequence (see Fig. 4) are indicated by double-headed arrows.
Fragments of cDNA were subcloned into pGEX-1N to prepare fusion
proteins for immunization, and the predicted ORF sequences to which
antisera were raised are indicated by arrows labeled with the
numbers given to the resulting antisera (660, 803, and 1002). The open bar represents the 5`-UTR (1-202 bp); the closed bar represents the predicted ORF (203-2148 bp);
and the stippled bar represents the 3`-UTR (2149-3268
bp).
indicates the
position of the proposed GPI cleavage/attachment site. The stop codon
limiting the open reading frame is indicated by an asterisk.
''; at the Ala residue 565. Taken together with
secondary structure predictions this sequence analysis suggests that
the protein can be divided into three structural domains; residues
1-275 containing several potential
-helices, 275-469
containing the proline/glutamine-enriched repeats, and 469-601
containing the potential GPI anchor site.137 kDa), which could be immunoblotted
with antibody 660 (Fig. 5). Translation in the reticulocyte
lysate in the presence of dog pancreatic microsomes showed that the
newly synthesized protein was incorporated into the microsomes such
that protease treatment only fully destroyed the protein when detergent
was present (compare Fig. 5, lanes 6 and 7).
In this experiment, a positive control,
-lactamase, was
translocated into microsomes and processed from a 31.5-kDa precursor to
the 28.9-kDa mature form as expected (data not shown), demonstrating
the functional integrity of the microsomes.
S]methionine-labeled translation products. Lane 3, no mRNA; lane 4, mRNA encoding p137; lane
5, mRNA encoding p137 in the presence of microsomes (M); lane 6, mRNA encoding p137 plus microsomes followed by
incubation in 0.1 mg/ml proteinase K (PK) for 1 h, 4 °C; lane 7, as for lane 6 but in the presence of 1%
Triton X-100. Only the portion of the gel containing p137 is
shown.
2 µg of poly(A)
RNA/lane from eight different tissues. The blot was probed with
P-random primed labeled fragment of insert 15 (bases
605-2252 of the contiguous sequence shown in Fig. 4).
Biosynthesis of p137
The discovery of a
GPI-anchored protein present in approximately equal amounts on the
apical and basolateral surfaces of Caco-2 cells raised questions about
its biosynthesis and route of delivery to each cell surface domain.
Prior to performing a pulse-chase experiment, the extent to which p137
was released into the culture medium was determined, since if there was
substantial release this would have to be taken into account when
considering delivery of newly synthesized protein to the cell surface.
Culture medium from flask-grown Caco-2 cells radiolabeled overnight
with [S]methionine was collected and shown to
contain p137 by immunoprecipitation with affinity purified antibody 660 (Fig. 7a). The released protein was recovered as a 280
kDa entity, which could not be reduced by treatment with
dithiothreitol, in contrast to the 280-kDa dimer found on the cell
surface (Fig. 7a). The 280-kDa released protein also
contained inositol (Fig. 2d, lanes 3 and 4). Biosynthetic labeling experiments showed the
[
S]methionine/[
H]inositol
ratio to be the same in the membrane-bound and released forms of the
protein (Fig. 2d and data not shown).
Immunoprecipitation of both plasma membrane and released forms of p137
was also carried out with a further antibody (coded 1002, Fig. 3) prepared against, and immunoaffinity purified on, a
glutathione transferase fusion protein containing the fragment of p137
encoded by HindIII-EcoRI fragment (nucleotides
263-605). This showed that both amino-terminal and
carboxyl-terminal epitopes were present in the immunoprecipitable
protein (Fig. 7b), providing additional evidence that
the predicted ORF in the contiguous cDNA sequence encodes p137.
S]methionine-labeled, flask grown
Caco-2 cells (lanes 1 and 2) and surrounding media (lanes 3 and 4) were immunoprecipitated with affinity
purified antibody 660 and analyzed by 7% SDS-PAGE and autoradiography
after sample preparation in the absence (lanes 1 and 3) or presence (lanes 2 and 4) of DTT. b, [
S]methionine-labeled, flask grown
Caco-2 cells (lanes 1 and 2) and surrounding media (lanes 3 and 4) were immunoprecipitated with either
affinity purified antibodies 1002 (lanes 1 and 3) or
660 (lanes 2 and 4) and analyzed by 7% SDS-PAGE and
autoradiography after sample preparation in the presence of
DTT.
S]methionine, subcellularly fractionated, and
membranes and chase media were immunoprecipitated with antisera 660. a, time course of appearance of p137 in apical (A)
and basolateral (B) membranes; b, time course of
appearance of secreted p137 in apical (A) and basolateral (B) chase media.
Transcytosis and Shedding of p137
To investigate
transcytosis of p137, filter-grown Caco-2 cells were cell surface
biotinylated in a domain-specific manner, essentially according to
Neame and Isacke (25) . First, both the apical and basolateral
surface of Caco-2 cells were biotinylated, the membrane proteins
immunoprecipitated, separated by SDS-PAGE, and biotinylated p137
detected using horseradish peroxidase-streptavidin (Fig. 9a). This confirmed that p137 could be
biotinylated and demonstrated its presence on both surface domains of
the cell monolayer by a technique other than subcellular fractionation.
The cell surface biotinylated form of p137 was detected as a 280 kDa
entity which could not be reduced by the addition of dithiothreitol.
This biotinylated protein was released into the culture medium from the
cell surface when cells were incubated at 37 °C. Transcytosis of
p137 was studied by immunoprecipitating the culture medium from the
opposite chamber to the cell surface that had been biotinylated and
visualizing the biotinylated proteins by SDS-PAGE followed by
horseradish peroxidase-steptavidin detection. The time courses of
shedding of biotinylated p137 (Fig. 9b) and
transcytosis in the basolateral to apical and apical to basolateral
directions (Fig. 9c) were followed by measuring the
appearance of the 280-kDa biotinylated protein in the appropriate
chambers. It was found that basolateral to apical transcytosis was
detectable by 3 h and proceeded at a steady rate for 24 h by which time
50% of basolaterally labeled p137 had been transcytosed compared
with
25% shed from the basolateral surface. Apical to basolateral
transcytosis could not be detected until after 4 h, but by 24 h
25% of apically labeled p137 had been transcytosed and
20%
shed from the apical surface.
, Ato A,
apical;
, B to B, basolateral). c, p137
transcytosed from one membrane domain to the other was quantified by
expressing the amount secreted from the opposite domain as a percentage
of the total p137 that was originally biotinylated ( A to
B, apical to basolateral;
, B to A, basolateral to
apical). Data from representative experiments are
shown.
25% in 24 h), is consistent with
the latter. Further biochemical and morphological experiments will be
required to characterize this novel membrane traffic route.
))))
We thank Dr. Nigel Hooper for the gift of the
anti-CRD antibody, Dr. Martin Low for the gift of PI-PLC, Sally Gray
for some sequencing plus a pGEX construct, and Dr. C. G. Tate for help
with several molecular biology techniques.
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Solomon, Y. Xu, B. Wang, M. D. David, P. Schubert, D. Kennedy, and J. W. Schrader Distinct Structural Features ofCaprin-1 Mediate Its Interaction with G3BP-1 and Its Induction of Phosphorylation of Eukaryotic Translation Initiation Factor 2{alpha}, Entry to Cytoplasmic Stress Granules, and Selective Interaction with a Subset of mRNAs Mol. Cell. Biol., March 15, 2007; 27(6): 2324 - 2342. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Shiina, K. Shinkura, and M. Tokunaga A Novel RNA-Binding Protein in Neuronal RNA Granules: Regulatory Machinery for Local Translation J. Neurosci., April 27, 2005; 25(17): 4420 - 4434. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. C. Katsafanas and B. Moss Vaccinia Virus Intermediate Stage Transcription Is Complemented by Ras-GTPase-activating Protein SH3 Domain-binding Protein (G3BP) and Cytoplasmic Activation/Proliferation-associated Protein (p137) Individually or as a Heterodimer J. Biol. Chem., December 10, 2004; 279(50): 52210 - 52217. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Grill, G. M. Wilson, K.-X. Zhang, B. Wang, R. Doyonnas, M. Quadroni, and J. W. Schrader Activation/Division of Lymphocytes Results in Increased Levels of Cytoplasmic Activation/Proliferation-Associated Protein-1: Prototype of a New Family of Proteins J. Immunol., February 15, 2004; 172(4): 2389 - 2400. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. L. TUMA and A. L. HUBBARD Transcytosis: Crossing Cellular Barriers Physiol Rev, July 1, 2003; 83(3): 871 - 932. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. W. Shupp, M. Jett, and C. H. Pontzer Identification of a Transcytosis Epitope on Staphylococcal Enterotoxins Infect. Immun., April 1, 2002; 70(4): 2178 - 2186. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. McCarthy, Y Yoong, and N. Simister Bidirectional transcytosis of IgG by the rat neonatal Fc receptor expressed in a rat kidney cell line: a system to study protein transport across epithelia J. Cell Sci., January 4, 2000; 113(7): 1277 - 1285. [Abstract] [PDF]
|
|
|
|
|
|
J. Ellis, M Craxton, J. Yates, and J Kendrick-Jones Aberrant intracellular targeting and cell cycle-dependent phosphorylation of emerin contribute to the Emery-Dreifuss muscular dystrophy phenotype J. Cell Sci., January 3, 1998; 111(6): 781 - 792. [Abstract] [PDF]
|
|
|
|
 |
|
 A. R. A. Hamad, P. Marrack, and J. W. Kappler Transcytosis of Staphylococcal Superantigen Toxins J. Exp. Med., April 21, 1997; 185(8): 1447 - 1454. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Molecular and Cellular Proteomics |
| Journal of Lipid Research | ASBMB Today |