 |
INTRODUCTION |
Phosphoinositide 3-kinase
(PI3K)1 was originally
isolated from bovine brain as a heterodimeric complex consisting of an
85-kDa regulatory subunit and a 110-kDa catalytic subunit. Since then, additional purification strategies and reverse transcriptase-polymerase chain reaction-based approaches have identified seven other mammalian PI3K catalytic subunits. On the basis of structure and in
vitro activity, these PI3K isozymes have been divided into three
classes termed I, II, and III (1). Class I further subdivides into class IA and class IB. Class IA contains p110
, p110
, and p110
. Each of these PI3K subunits bind a p85-like adaptor that facilitates their translocation to phosphotyrosine-containing signaling complexes. Class IB contains the G-protein-activated enzyme p110
. This PI3K isozyme does not bind a p85 adaptor but instead a molecule termed p101
whose function is currently unclear. In vitro, these PI3K enzymes are all able to utilize phosphatidylinositol (PtdIns), PtdIns(4)P, and PtdIns(4,5)P2 as substrate. However, it has
been suggested that PtdIns(3,4,5)P3 is their principle
product in situ (2, 3).
Class II PI3K enzymes include PI3K-C2, PI3K-C2, and PI3K-C2.
They are distinguished from other PI3K isozymes by the presence of two
tandem domains at their carboxyl terminus. The first is termed a phox
homology (PX) domain and despite its presence in many eukaryotic
signaling molecules, the function of this domain remains unclear (4).
The second is the C2 domain which is a phospholipid-binding module that
can confer a Ca2+ sensitivity (5). The function of either
domain in the PI3K enzymes is unknown. In vitro, all class
II PI3K isozymes phosphorylate PtdIns and PtdIns(4)P but their spectrum
of 3-phosphoinositide production in vivo remains to be
determined. However, PI3K-C2 plays a signaling role downstream of
both the receptor for the chemokine monocyte chemotactic peptide
(MCP-1) (6) and the insulin receptor (7). Both PI3K-C2 and
PI3K-C2 are downstream signaling targets of activated epidermal
growth factor and platelet-derived growth factor
receptors.2 In platelets,
PI3K-C2 is activated in response to stimulation of integrin
receptors by fibronectin (9). Interestingly, although both PI3K-C2
and PI3K-C2 share a wide tissue distribution, expression of
PI3K-C2 is restricted primarily to hepatocytes (10, 11). Class III
PI3K contains only a single enzyme termed Vps34p. In yeast, this PI3K
isozyme regulates vacuolar trafficking through its generation of
PtdIns(3)P. Although a mammalian homologue of Vps34p has also been
characterized, most of what we understand about its function has been
extrapolated from work in yeast (12, 13).
The 3-phosphoinositides produced by PI3K activity play a pivotal role
as second messenger molecules in a diverse array of physiological
events. These include cell proliferation, the prevention of apoptosis,
vesicle trafficking, migration and motility, receptor internalization,
and glucose transport (3). In quiescent cultures, only
phosphatidylinositol 3-phosphate (PtdIns(3)P) is readily detected.
However, following ligand stimulation, the intracellular concentration
of PtdIns(3,4,5)P3 transiently increases within seconds
followed by a more sustained accumulation of PtdIns(3,4)P2 (2). Given that 3-phosphoinositides direct an array of molecules to
different positions within the cell in response to different stimuli,
some degree of spatial and temporal regulation of 3-phosphoinositide production must exist. Several PI3K isozymes may be found within any
single cell type and although not all PI3K enzymes can produce each
3-phosphoinositide product there is often overlap in the distribution
of PI3K isozymes within the same class. Although much effort has been
applied to identifying distinct mechanisms of activation, little has
been done to examine the subcellular localization of the PI3K isozymes.
In eukaryotic cells, the formation and transport of vesicles from both
the plasma membrane and trans-Golgi network (TGN) to endosomes involves
the assembly of clathrin-coated vesicles (CCVs) (14, 15). While CCVs
derived from the plasma membrane and TGN both contain clathrin as a
major component, the two types of vesicles can be distinguished by
their assembly or adaptor protein (AP). Thus, AP-1 is present on
vesicles derived from the Golgi apparatus while AP-2 is found on
vesicles produced from the plasma membrane. Both AP-1 and AP-2 are
heterotetrameric complexes. AP-1 is composed of and ' adaptin
(approximately 100 kDa), a 47- and 19-kDa protein. In contrast, AP-2
contains and adaptin (approximately 100 kDa), a 50- and 17-kDa
protein (16, 17). Polymerization of clathrin with AP complexes in the
cell is associated with phosphoinositide interactions and an array of
receptors and regulatory molecules such as the GTPase ADP-ribosylation
factor-1 (18). Tyrosine- and dileucine-based motifs represent two
sorting signals contained within the cytoplasmic portion of several
receptors and transmembrane glycoproteins. Such motifs mediate
interactions with either the µ2 subunit of AP-2 or the 1 subunit
of AP-1 to select proteins for inclusion into the vesicle and
facilitate their transfer from one part of the cell to another (19,
20). A long outstanding question remains as to how the adaptor
complexes can specifically target selected intracellular membranes.
Numerous studies suggest that one or more 3-phosphoinositides may
provide at least part of the required specificity. However, it remains
unclear which PI3K isozymes are involved in the control of vesicle
trafficking. In this study we investigate the subcellular localization
of PI3K-C2 with the goal of determining whether this PI3K could play
a role in regulating adaptor protein function.
| |
EXPERIMENTAL PROCEDURES |
Cell Culture--
Stock cultures of HEK293 and COS-1 cells were
subcultured in 90-mm Nunc dishes using Dulbecco's modified Eagle's
medium supplemented with 10% fetal bovine serum and 100 units/ml
penicillin and 100 µg/ml streptomycin. Cultures were incubated
in a humidified atmosphere of 10% CO2, 90% air at
37 °C for 3 days. After this time, cells were switched to
Dulbecco's modified Eagle's medium containing 0.5% fetal bovine
serum. The cultures were confluent and quiescent 24-48 h later.
Cultures of U937 cells were passaged every 3-4 days in RPMI containing
5% fetal bovine serum and penicillin/streptomycin.
Subcellular Fractionation--
Cells were disrupted by 20 strokes of a small Dounce homogenizer on ice and 10 passages through a
25-gauge needle in 1 ml of Tris/sucrose buffer (0.25 M
sucrose, 20 mM Tris-HCl, pH 7.4, 1 mM
Na3VO4, 1 mM phenylmethylsulfonyl
fluoride, 10 µg/ml leupeptin, 200 KIU/ml aprotonin, 1 µg/ml
antipain, 1 µg/ml pepstatin, and 400 µg/ml benzamidine). Intact
cells and nuclei were removed by centrifugation at 3,000 × g for 10 min. The supernatant was then centrifuged at
18,000 × g for 20 min to yield a PM/TGN-enriched fraction and then at 350,000 × g for 30 min. The
resultant pellet was referred to as the LDM fraction and the
supernatant as cytosol (21). All centrifugations were performed at
4 °C. Crude enrichment of PM/TGN and LDM from cytosol was verified
by Western blotting for M6PR, TGN-46 the human homologue of the
trans-Golgi resident protein TGN-38 and measurements of lactate
dehydrogenase activity.
Immunoprecipitation--
Quiescent cultures were washed twice
with PBS and lysed at 4 °C with either Triton X-100 lysis buffer (10 mM Tris-HCl, 5 mM EDTA, 150 mM
NaCl, 30 mM sodium pyrophosphate, 50 mM sodium
fluoride, 100 µM Na3VO4, 1%
Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin, 1 µg/ml pepstatin, 200 KIU/ml aprotonin, and 1 µg/ml antipain, pH 7.6) or Nonidet P-40 lysis buffer (20 mM Tris-HCl, 2 mM EDTA, 137 mM
NaCl, 1 mM Na3VO4, 1% Nonidet
P-40, 1 mM phenylmethylsulfonyl fluoride, and 10 µg/ml
each of leupeptin, chymostatin, and antipain, 15% glycerol, pH 7.5).
Cell lysates were clarified by centrifugation at 15,000 × g for 10 min and pre-cleared for 1 h at 4 °C. The
supernatants were then transferred to fresh tubes and proteins were
immunoprecipitated upon addition of antibodies and protein G-agarose.
Immunoprecipitates were either washed three times with lysis buffer and
extracted with sample buffer (200 mM Tris-HCl, 6% SDS, 2 mM EDTA, 4% 2-mercaptoethanol, 10% glycerol, pH 6.8) or
washed twice with lysis buffer and three times with kinase buffer for
use in phosphoinositide kinase assays.
Western Blotting--
Following extraction, proteins were
fractionated by SDS-PAGE. After transfer to polyvinylidene difluoride
membranes these were blocked with 5% non-fat dried milk in PBS, pH
7.4. Anti-PI3K-C2 antisera (1:1000), anti-p85 (1:500), or
anti-Glu (1:700) antibody was added in PBS that contained 3% milk,
0.05% Tween 20 for 2-4 h. Following a 1-2 h incubation with
horseradish peroxidase-labeled second antibody (1:5000), immunoreactive
proteins were visualized using enhanced chemiluminescence (ECL,
Amersham Pharmacia Biotech).
Phosphoinositide Kinase Assay--
Preparations of purified
CCVs, immunoprecipitated Glu-tagged PI3K-C2 or p110 were
incubated with [-32P]ATP and PtdIns at 30 °C for
either 10 min (CCV preparations) or 30 min (PI3K-C2 and p110).
The formation of radiolabeled 3- and 4-phosphoinositides was determined
upon separation by thin layer chromatography and autoradiography as
described previously (22).
Preparation of PI3K-C2 Deletion Mutants--
The PX-C2 domain
and C2 domain deletion mutants were prepared by a polymerase chain
reaction based strategy. The sense oligonucleotide, ccttcagggttaccagaacttac, annealed across the BstEII
restriction site at nucleotide 4022. The antisense oligonucleotide,
ccggaattcttagatgggctcatcattagaagg, terminated the PI3K-C2 cDNA
at nucleotide 4224 and incorporated a 3' EcoRI restriction
site. The antisense oligonucleotide, ccggaattcttaagacctagctatcccttcagc, terminated the PI3K-C2 cDNA at nucleotide 4638. Sense and
antisense primers were used in a polymerase chain reaction using 5'
Glu-tagged PI3K-C2-pBluescript SK plasmid as template. The 222- and
636-base pair products were purified and digested with
BstEII and EcoRI enzymes together with 5'
Glu-tagged PI3K-C2-pBluescript SK plasmid. Once re-purified, the
inserts were ligated using T4 ligase. After sequencing, cDNA
representing 5' Glu-tagged PI3K-C2, the 
PX-C2 and C2 domain
deletion mutants were subcloned into pShooter (pEF/cyto, Invitrogen)
using NcoI/XhoI. These constructs were used for
expression of recombinant protein in mammalian cells.
Mammalian Cell Transfections--
Cultures of HEK 293 cells
(60% confluent) were transfected with 5' Glu-tagged
PI3K-C2-pShooter, PX-C2 deleted (PX-C2), and C2 domain deleted
(C2) 5' Glu-tagged PI3K-C2-pShooter plasmid (5 µg/dish) using
LipofectAMINE (Life Technologies) in accordance with the supplied
protocol. Cells were used for experimental purposes, 48 h
following transfection.
Immunostaining of Cells--
Cells (1 × 105)
were plated onto 13-mm glass coverslips and cultured overnight with
10% fetal bovine serum in Dulbecco's modified Eagle's medium at
37 °C in a humidified atmosphere containing 10% CO2,
90% air. The following day, cells were washed twice with PBS and fixed
with 3.7% formaldehyde in PBS for 30 min at room temperature. After 6 washes with PBS, cells were permeabilized with 0.2% Triton X-100 in
PBS and sites of nonspecific binding were blocked using PBS containing
5% non-fat dried milk, 0.05% Tween 20. All antibodies were added in
PBS containing 0.5% non-fat dried milk, 0.05% Tween 20 which was also
used for all washes. Cells were incubated with primary antibody for 90 min and after washing, FITC and TRITC second antibody was added at
1:400 final dilution for 60 min. After 6 washes and a final wash in
H2O, the coverslips were inverted onto anti-fade mount and
examined using fluorescence microscopy (Olympus). Images were captured
using a cooled CCD camera (Phototronics) linked to a PC running Image ProPlus software (Media Cybernetics).
Antisera--
Anti-PI3K-C2 antisera was produced as described
previously to a NH2-terminal fragment of the PI3K-C2
protein (6). This was affinity purified using recombinant protein and
eluted with 0.1 M glycine into Tris, pH 8. Monoclonal
anti- adaptin AP-1 (clone 100/3) and anti- adaptin AP-2 (clone
100/2) were purchased from Sigma. Monoclonal anti-Lamp-1 was a generous
gift from Prof. C. Hopkins, Medical Research Council Laboratory for
Molecular Cell Biology, University College, London. Rabbit anti-TGN-46
was a gift from Dr. G. Banting, University of Bristol, sheep
anti-TGN-46 was a gift from Dr. S. Ponnambalam, University of Dundee.
Anti-M6PR antisera were a gift from Professor P. Luzio, University of Cambridge.
Enrichment of Clathrin-coated Vesicles--
Bovine brain CCVs
were prepared using discontinuous sucrose density gradient
centrifugation steps as described previously (16). These partially
purified preparations were further fractionated on a
sucrose-D2O step gradient (23). The molecular mass of the clathrin triskelion was approximated at 650 kDa. Quantification was
performed by comparison of immunoblots with a 10-fold range of
recombinant PI3K-C2 protein standards using methods previously described (24, 25).
| |
RESULTS |
PI3K-C2 Is Constitutively Localized to Phospholipid
Membranes--
Confluent and quiescent cultures of COS-1 and U937
cells were scraped into hypotonic Tris/sucrose buffer and their plasma membranes disrupted on ice using sequential Dounce homogenization and
passage through a narrow gauge needle. Differential centrifugation of
successive supernatants generated two membrane-enriched fractions and
cytosol. Western blot analysis demonstrated that endogenous PI3K-C2
protein present in both cell types was enriched in the plasma
membrane/trans-Golgi network (PM/TGN) and low density microsomal fractions (LDM) (Fig. 1). In comparison,
little of this PI3K isoform was present in the cytosolic fraction
(Cyt). Western blotting of the M6PR and TGN-46 confirmed the efficiency
of the enrichment protocol. Each of these marker proteins predominated
in the LDM and PM/TGN membrane fractions, respectively (Fig.
2, upper and middle
panel). Separation of membranes from cytosol was demonstrated by
lactate dehydrogenase assay (Fig. 2, lower panel). Our
results demonstrate that PI3K-C2 is constitutively associated with
phospholipid membranes.
View larger version (46K):
[in this window]
[in a new window]
|
Fig. 1.
Enrichment of PI3K-C2
on phospholipid-containing membranes. Confluent and
quiescent cultures of COS-1 (Panel A) and U937 (Panel
B) cells were homogenized in isotonic buffer as described under
"Experimental Procedures." Differential centrifugation of the
homogenate (Homog) resulted in a plasma membrane/trans-Golgi
network-enriched fraction (PM/TGN), a low density microsomal
enriched fraction (LDM), and cytosol (Cyt). These
were extracted with Lamelli sample buffer, fractionated by SDS-PAGE,
transferred onto polyvinylidene difluoride membrane, and Western
blotted with anti-PI3K-C2 antisera. Immunoreactive proteins were
detected using horseradish peroxidase-labeled secondary antisera and
visualized by ECL.
|
|
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 2.
Characterization of the subcellular
fractions. Membrane fractions were prepared from HEK293 cells as
described under "Experimental Procedures." The PM/TGN, LDM, and
cytosol fractions were extracted, fractionated by SDS-PAGE, and Western
blotted with polyclonal antisera to either the M6PR and TGN-46.
Aliquots of each fraction were also examined for lactase dehydrogenase
(LDH) activity.
|
|
Clathrin-coated Vesicles Contain the PI3K-C2 Enzyme--
There
is good evidence to demonstrate that phosphoinositides, particularly
3-phosphoinositides, bind to AP-2 adaptors (26, 27). To evaluate the
possibility that PI3K-C2 might be present in CCV, these were
prepared from bovine brain using established procedures. Fractions
containing equivalent protein loadings were then analyzed by Western
blotting (Fig. 3). Little PI3K-C2
immunoreactivity was detected in the low speed pellet and essentially
none in the high speed supernatant, however, differential sucrose and
D2O gradient centrifugation of the high-speed pellet
yielded fractions of progressively purified CCVs. On a per microgram
basis, these were dramatically increased in PI3K-C2 content,
quantitatively in parallel to the enrichment of clathrin.
Interestingly, the class I PI3K adaptor p85 was also detectable in
these vesicle preparations confirming an earlier report (28), however,
only an extremely small proportion of the total cellular complement of
this enzyme was present in CCVs and no evidence of any enrichment was
apparent. Finally, quantification of PI3K-C2 in these CCV preparations suggests that this PI3K isozyme represents approximately 0.3% of total CCV protein. This corresponds to approximately 1 molecule of PI3K-C2/CCV.
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 3.
PI3K-C2 but not the
class I PI3K adaptor p85 is highly enriched in clathrin-coated
vesicles. Clathrin-coated vesicles were purified from bovine
brain. Equivalent amounts of the indicated fractions (10 µg) were
separated by SDS-PAGE and immunoblotted with anti-PI3K-C2
(upper panel) or anti-p85 (lower panel)
antibody. LSS, 10,000 × g supernatant;
HSS, 245,000 × g supernatant;
grad-1, crude coated vesicle fraction pooled from the first
discontinuous sucrose gradient; grad-2, coated vesicle
fraction pooled from the second discontinuous sucrose gradient;
CCV, purified clathrin-coated vesicles collected after
centrifugation of the fraction grad-2 through the 8% sucrose cushion
in D2O.
|
|
In a complementary series of experiments, the formation of both
PtdIns(3)P and PtdIns(4)P was detected in purified CCV fractions upon
their incubation with [-32P]ATP and PtdIns (Fig.
4, middle panel). These
findings are consistent with the results presented in Fig. 3.
Pretreatment of the CCV preparations with wortmannin revealed a
dose-dependent inhibition of PtdIns(3)P production.
However, little inhibition of enzyme activity was achieved using 10 nM wortmannin. PI3K activity was still evident using 100 nM inhibitor and only abolished using 1 µM
wortmannin. This dose-effect relationship was similar to that obtained
using isolated PI3K-C2 (Fig. 4, upper panel) but markedly
contrasts with that using recombinant class I PI3K catalytic subunit
p110 (Fig. 4, lower panel). These observations provide functional data to support the presence of the PI3K-C2 enzyme to CCV
preparations.
View larger version (40K):
[in this window]
[in a new window]
|
Fig. 4.
Functional PI3K-C2
activity in clathrin coated vesicles. Purified CCV were
incubated with [-32P]ATP and exogenous PtdIns. The
radiolabeled phosphoinositides produced were separated by thin layer
chromatography as described previously (22). Formation of PtdIns(3)P by
CCV preparations was readily detectable under these conditions
(arrow). Wortmannin pretreatment attenuated this lipid
kinase activity in CCV preparations at concentrations higher than those
required to inhibit the activity of p110 (lower panel)
but similar to those needed to inhibit the activity of purified
PI3K-C2 (upper panel).
|
|
PI3K-C2 Colocalizes with AP-1 Complexes at the TGN--
There
have been few reports detailing the subcellular distribution of PI3K
isozymes by immunofluorescence. The reasons for this appear to be the
lack of suitable antisera combined with low levels of the endogenous
protein. Alternate strategies such as the expression of epitope-tagged
recombinant protein and its localization with monoclonal antibody
suffers from a possible criticism. Namely, that protein distribution
may become aberrant upon high levels of protein expression and possible
limitation of endogenous binding partners. This is especially relevant
to both the class I and III PI3K catalytic subunits. Consequently, we
used an affinity purified polyclonal antibody raised against the
NH2 terminus of PI3K-C2 to determine the distribution of the endogenous enzyme.
Immunostaining of Triton X-100 permeabilized HEK293 cells with
anti-PI3K-C2 antisera revealed a punctate distribution of immunoreactivity across the entire cell with a distinctive and characteristic enrichment in a perinuclear region (Fig.
5, panels A, C, and
E). Dual labeling was performed with monoclonal anti- adaptin, anti- adaptin, and anti-Lamp-1 antibodies in an effort to
clarify the nature of these subcellular compartments. Anti--adaptin also produced a punctate staining pattern (Fig. 5, panel B)
similar to that observed for PI3K-C2, however, there was little
evidence of overlap in these cells. In contast, immunostaining for
-adaptin produced a far more localized, perinuclear staining pattern
(29) (Fig. 5, panel D). Some AP-1 adaptor containing CCVs
were also scattered around the cytoplasm and close to the plasma
membrane. However, it was clear that at the perinuclear region, there
was marked overlap between the staining patterns of -adaptin and PI3K-C2. Although there was evidence of co-localization, this overlap was not absolute.
View larger version (57K):
[in this window]
[in a new window]
|
Fig. 5.
Co-localization of PI3K-C2
with the AP-1 adaptor -adaptin.
Formaldehyde-fixed HEK-293 cells were processed for dual label
immunofluorescence microscopy using the affinity purified
anti-PI3K-C2 antiserum (Panels A, C, and E)
and mouse monoclonal antibodies directed against either the subunit
of AP-2, -adaptin (Panel B), the subunit of AP-1,
-adaptin (Panel D), or the lysosomal
glycoprotein marker Lamp-1 (Panel F). Rabbit and mouse
primary antibodies were detected with corresponding secondary IgG
labeled with FITC (PI3K-C2) or TRITC (-adaptin, -adaptin, and
Lamp-1). Arrowheads illustrate an example of the overlap
between PI3K-C2 and -adaptin immunoreactivity (Panels
C and D).
|
|
Lamp-1 is a heavily glycosylated protein present in lysosomal membranes
(30). Our immunostaining also revealed a localized perinuclear
distribution of Lamp-1 (Fig. 5, panel F) but despite a
limited degree of overlap, its pattern was quite distinct from that of
PI3K-C2 (Fig. 5, panel E). Even at the TGN, lysosomal membrane proteins such as Lamp-1 and Lamp-2 are packaged into vesicles
that are distinct from those that carry AP-1, M6PR, and vacuolar
hydroxylases (31). Interestingly, the expression of epitope-tagged rat
PI3K-C2 in primate COS cells suggested that this class II PI3K
isozyme also has a perinuclear distribution, however, it was also shown
to be present on the plasma membrane and nuclear membranes.
Unfortunately, in that study, no attempt was made to characterize the
nature of the perinuclear compartment (10).
The overlap in PI3K-C2 and -adaptin immunostaining was further
examined in COS cells (Fig. 6,
upper panels). The merged image clearly shows that the
distribution of the two proteins strikingly overlaps at a perinuclear
region. The presence of PI3K-C2 at the TGN was confirmed by
co-staining HEK-293 cells with anti-TGN-46 antisera (Fig. 6,
lower panels). TGN-46 is the human homologue of TGN-38, a
well characterized transmembrane protein predominantly resident in the
TGN (32).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 6.
PI3K-C2 is present
at the TGN. COS and HEK293 cells were stained with affinity
purified polyclonal anti-PI3K-C2 antiserum and either monoclonal
anti--adaptin antibody or goat anti-TGN-46 polyclonal antisera
respectively. Anti-PI3K-C2 antisera was visualized with
TRITC-labeled anti-rabbit IgG. Anti--adaptin was detected with
FITC-labeled anti-mouse IgG while anti-TGN-46 was visualized using
FITC-labeled donkey anti-goat IgG. Positions of protein co-localization
are revealed by the isignal.
|
|
The Association of PI3K-C2 with Phospholipid Membranes Is
Independent of Both Its C2 and PX Domains--
To examine if either
the carboxyl-terminal PX domain or C2 domain that are unique to the
class II PI3K enzymes were responsible for the observed membrane
localization of PI3K-C2, two deletion mutants were produced. The
first termed PX-C2 PI3K-C2 (encoding amino acid residues 1-1408)
lacked both domains while C2 PI3K-C2 (encoding amino acid
residues 1-1546) lacked only the latter domain. Each deletion was
confirmed by cDNA sequence analysis and following protein
expression each mutant could be distinguished by their migration on
SDS-PAGE (data not shown).
Recombinant 5' Glu-tagged PI3K-C2 and the deletion mutants were
expressed in HEK293 cells that were used to determine their subcellular
localization either by immunofluorescence analysis (Fig.
7A) or differential
centrifugation and Western blotting of the homogenates (Fig.
7B). In each experiment, recombinant PI3K-C2 was detected
with a monoclonal anti-Glu tag antibody. Immunofluorescence staining
showed that recombinant PI3K-C2 also had a punctate staining pattern
with an enrichment of the enzyme in a perinuclear region.
Interestingly, deletion of the PX and/or C2 domain had little effect
upon the distribution of this PI3K isozyme. Western blotting of crude
membrane fractions revealed that like the endogenous enzyme,
recombinant PI3K-C2 and each COOH-terminal deletion mutant were also
present in the PM/TGN and LDM fractions (Fig. 7B). These
results further suggest that the association of PI3K-C2 with
phospholipid membranes is independent of any adaptor or binding
proteins. If it were, a rapid saturation of the available sites is
likely to have occurred, resulting in an increase of cytosolic
PI3K-C2 relative to membrane fractions.
View larger version (51K):
[in this window]
[in a new window]
|
Fig. 7.
Localization of recombinant
PI3K-C2, PX-C2,
and C2 domain deletion mutants. HEK 293 cells were transiently transfected with a cDNA plasmid containing
the mammalian expression vector pShooter and either full-length 5'
Glu-tagged PI3K-C2 or deletion mutants 5' Glu-tagged PI3K-C2
1-1408 (PX-C2 domains) and 5' Glu-tagged PI3K-C2 1-1546 (C2
domain). After 48 h, cultures were used for immunofluorescence
staining with anti-Glu tag antibody and visualized with FITC-labeled
IgG (upper panels). In addition, cells were homogenized in
Tris/sucrose buffer and centrifuged to yield two membrane-enriched
fractions containing plasma membrane/trans-Golgi network
(PM/TGN), low density microsomes (LDM), and
cytosol (Cyt). After SDS-PAGE, protein was Western blotted
with anti-Glu tag antibody and visualized by ECL.
|
|
The fungal metabolite brefeldin A causes release of -adaptin from
its position at the TGN (33). This results in the redistribution of
AP-1-containing CCVs but it does not affect the Golgi stack. Treatment
of HEK293 cells with brefeldin A produced a dramatic reorganization of
both PI3K-C2 (Fig. 8, panels
A and C) and -adaptin immunoreactivity (Fig. 8,
panels B and D) from the perinuclear region into
the cytoplasm. We conclude that the localization of both PI3K-C2 and
-adaptin at the TGN is dependent upon Arf-1 GTPase activity.
View larger version (71K):
[in this window]
[in a new window]
|
Fig. 8.
Brefeldin A disrupts the association of
PI3K-C2 with the TGN. HEK293 cells grown
on coverslips were incubated in the absence (Panels A and
B) or presence (Panels C and D) of
brefeldin A (5 mg/ml) for 30 min. After treatment, the cultures were
washed with PBS, fixed with formaldehyde, and permeabilized with Triton
X-100. The cells were then incubated with affinity purified
anti-PI3K-C2 antiserum (Panels A and C) and a
mouse monoclonal antibody directed against -adaptin (Panel
B and D). Rabbit and mouse primary antibodies were
detected with corresponding secondary IgG labeled with FITC
(PI3K-C2) or TRITC (-adaptin).
|
|
| |
DISCUSSION |
One of the most striking findings we report here is that the
PI3K-C2 enzyme is constitutively associated with phospholipid membranes in mammalian cells. We demonstrate that this PI3K isozyme is
enriched upon purification of CCVs and although it is widely distributed across the cytoplasm with a punctate distribution it is
clearly concentrated at the TGN.
Our observations contrast with the previously reported subcellular
distribution of the class IA PI3K isozymes which, in quiescent cells,
are largely present within the cytoplasm as a heterodimeric complex and
translocate to the plasma membrane upon ligand stimulation (28, 34).
Numerous mechanisms have been proposed to activate the catalytic
activity of class IA PI3K enzymes including phosphotyrosine-stimulated conformational change and phospholipid substrate presentation (35, 36).
Clearly, if PI3K-C2 is present close to its phospholipid substrate
then this PI3K isozyme is either constitutively active or its enzymatic
activity is tightly regulated by an as yet unknown mechanism. We
consider the former hypothesis unlikely since transient transfection of
either recombinant PI3K-C2 or PI3K-C2 into COS cells did not
alter the pattern of 3-phosphorylated lipid production in these cells
(data not shown and Ref. 37). In contrast to both the class I and the
class III enzymes, no binding partner has been identified for the class
II PI3Ks. Given the comparatively high molecular mass of these enzymes
it is possible that these PI3K enzymes contain an autoregulatory domain
to modulate their catalytic activity. Arcaro et al. (37) has
suggested that the C2 domain of PI3K-C2 may act as a negative
regulator of its catalytic activity by sequestering lipid substrate,
however, further work is required before this hypothesis can be
extrapolated to other class II PI3K enzymes.
Our biochemical localization of PI3K-C2 to phospholipid membranes
and CCVs (Figs. 1-4) was confirmed by immunofluorescence. Although
PI3K-C2 immunoreactivity was distributed across the cytoplasm in a
punctate manner it was strikingly localized at the TGN in close
proximity to -adaptin (Figs. 5 and 6) (17, 29). Interestingly,
although there was considerable overlap between the staining patterns
of these two proteins, they did not fully co-localize. However,
brefeldin A treatment did disrupt the distribution of both proteins
(Fig. 8) suggesting that they share a common mechanism of localization.
Brefeldin A is a fungal metabolite originally found to produce extended
tubular structures from the Golgi apparatus and redistribute Golgi
components to the endoplasmic reticulum. In addition, it induces
the formation of membranous tubules at both the TGN and endosomes (38,
39). These effects are produced by its inhibition of
ADP-ribosylation factor and AP-1 binding to the TGN and COP binding
to the cis-Golgi (40).
A modulatory role for 3-phosphoinositides in the control of vesicular
traffic has been clearly established. At the plasma membrane these
phosphoinositides are required for transport of internalized receptor
to the lysosome (41, 42) whereas at the TGN, they regulate vesicle
traffic to endosomes (43, 44). In contrast, less is understood about
how 3-phosphoinositides produce these effects and which PI3K enzymes
are responsible for their production. Over the last 2-3 years the
protein domains that mediate the action of 3-phosphoinositides have
been elucidated. PtdIns(3)P binds a double zinc finger motif termed the
FYVE domain present in an assortment of molecules that regulate various
aspects of vesicle trafficking (45-47). Both PtdIns(3,4)P2
and PtdIns(3,4,5)P3 bind pleckstrin homology domains
present in a number of signaling molecules (48, 49). In this way,
3-phosphoinositides orchestrate the translocation of molecules to
signaling complexes at phospholipid membranes and in some cases
directly regulate their enzyme activity. The binding of the µ2 chain
of purified AP-2 complexes with proteins is also enhanced by both
PtdIns(3,4)P2 and PtdIns(3,4,5)P3 (50). Interestingly, the interaction between dileucine-containing peptides and the 1 subunit of AP-1 was inhibited by the same
3-phosphoinositides (19). In this way, the products of PI3K activity
could ensure the specific interaction of transmembrane proteins with
one adaptor complex but not another. In addition to binding clathrin
(51), synaptotagmin (52), Grb2, and the EGF receptor-binding protein Eps15 (53), the chain of AP-2 has also been shown to bind inositol
phosphates and 3-phosphoinositides (27). The physiological relevance of
the last observation has been demonstrated in a recent study in which
the phosphoinositide binding ability of the AP-2 subunit was
abrogated. AP-2 complexes containing the mutant subunit assembled
properly but were not incorporated into coated pits (54). Similarly,
arrestin 3 lacking a functional phosphoinositide-binding site was
incapable of effectively mediating the concentration of activated
2-adrenergic receptor in coated pits (55). Consequently, 3-phosphoinositides appear to regulate two of the four subunits that
form the AP-2 adaptor. It will be interesting to see if additional AP-1
subunits also bind these phospholipids.
Since it is widely considered that the class IA PI3K enzymes are
intimately involved in signal transduction at the plasma membrane they
are unlikely to regulate protein trafficking from the TGN. Because of
the seminal work on yeast Vps34, the mammalian homologue of this class
III PI3K is considered to be the principle mediator of vesicle
transport from the TGN to lysosomes, the organelle in mammalian cells
that is functionally equivalent to the yeast vacuole. Transport of
newly synthesized lysosomal enzymes from the TGN can be inhibited by
the PI3K inhibitor wortmannin (43, 56). However, these effects are
observed over a wide range of concentrations (IC50 10-100
nM) depending upon the cell type chosen suggesting the
involvement of PI3K isozymes in addition to the wortmannin-sensitive
human Vps34p homologue (13). Furthermore, direct extrapolation from the
yeast model to mammalian cells may be somewhat simplistic since in
yeast, Vps34p is the only PI3K and PtdIns3P their principal
3-phospholipid product. Because of their greater complexity, mammalian
cells may have evolved and recruited additional PI3K isoforms to
produce a larger spectrum of 3-phosphoinositide products to broaden the
number of target sites available for an individual phospholipid vesicle.
The class II PI3K isozymes are distinguished by the presence of two
adjacent domains in their carboxyl terminus. The C2 or CalB domain that
was originally identified in protein kinase C (57) appears to have
diverged in evolutionary terms into
Ca2+-dependent and Ca2+-independent
phospholipid binding forms (5). The C2 domain present on each of the
class II PI3K enzymes binds phospholipids in a
Ca2+-independent manner (37, 58). Even less is known about
the functional role of the PX domain but it has been proposed to
mediate protein-protein interactions (4).
PX domains are often associated with events that involve the actin
cytoskeleton, GTP-binding proteins, and phospholipid membranes (59).
Somewhat surprisingly, our data demonstrates that neither the PX domain
nor the C2 domain are responsible for the interaction of PI3K-C2
with membranes (Fig. 7). Consequently, the mechanism responsible for
their localization remains to be established. Despite the extended
NH2 terminus of the class II PI3K enzymes we have yet to
identify any additional phospholipid-binding domains within this
region. Similar observations were recently made with the PI3P-binding
FYVE domain. Deletion of this domain from both Vac1p and Hrs did not
alter the membrane localization of either protein (45, 46). Although
the membrane localization of PI3K-C2 could be achieved via a
protein-protein interaction, immunoprecipitation of PI3K-C2 from
cell lysates has not yet revealed any proteins which
co-immunoprecipitate in a high stoichiometry.
Identification of PI3K isozymes responsible for driving vesicular
traffic from the mammalian TGN has proved somewhat elusive. An
apparently novel PI3K appears to act synergistically with the phospholipid transfer protein PITP and is present in a complex termed
p62(cplx) (60). This PI3K was characterized as a 100-kDa protein that
associates with TGN-38 (61). However, we had previously demonstrated a
Vps34p-like enzyme activity in association with human PITP (62).
Another study has isolated a 250-kDa PI3K in association with TGN-46
the human homologue of TGN-38 (32). It remains to be established if
different PI3K isozymes fulfil the same functional role in rodent and
man or if indeed two distinct PI3K isozymes are involved.
As we begin to appreciate the role PI3K activity plays in vesicle
trafficking, it is becoming ever clearer that the study of signal
transduction and membrane trafficking can no longer remain mutually
exclusive disciplines. The ability to constantly select and
redistribute receptors through coated pits and the endosomal network
allows cells to respond quickly to alterations in their external
environment. Consequently, these events are highly dynamic and require
tight regulatory control. The internalization of epidermal growth
factor receptor (EGFR) into a clathrin coat by an AP-2-mediated
mechanism provides a good example (63). In addition to receptor
down-regulation, internalization of receptor subunits also serves to
modulate receptor tyrosine kinase signaling pathways. Using an
internalization defective cell line it was shown that EGFR-mediated
tyrosine phosphorylation and mitogen-activated protein kinase
activation were both attenuated in internalization defective cells
(64). This suggests that certain steps of the signaling cascade can
only be effected within specific intracellular compartments. The
adaptor proteins appear to co-ordinate this specificity since in
addition to EGFR·AP-2 complex formation at the plasma membrane, EGFR
also interacts with AP-1 in large perinuclear endosomes (63).
Many questions remain regarding the function of PI3K-C2 in the
context of adaptor complex function. Since PI3K-C2 is present on
CCVs and resides in the TGN close to the AP-1 adaptor, this suggests
that its production of PtdIns(3)P could drive vesicle formation and
traffic them from the TGN to endosomes. PI3K-C2 may contribute to
the formation of CCVs at receptors such as the M6PR (65) or other
membrane-associated protein complexes that selectively bind APs (8).
Previously, we have shown that PI3K-C2 is a downstream target for a
growing number of receptors located in the plasma membrane. It is
therefore possible that PI3K-C2 is involved in some aspect of
receptor mediated signaling even after its internalization. In this way
it may continue to disseminate the signaling cascade within the
interior of the cell in a coordinated and localized manner. This
proposition warrants further experimental analysis.
Is Concentrated
in the Trans-Golgi Network and Present in Clathrin-coated Vesicles*
§,
TOP
ABSTRACT
INTRODUCTION
