(Received for publication, September 16, 1996, and in revised form, April 15, 1997)
From the Laboratory of Cell Signaling, National Heart, Lung, and
Blood Institute, National Institutes of Health, Bethesda, Maryland
20892-0320 and the Department of
Pharmacological Sciences, State University of New York,
Stony Brook, New York 11794-8651
In the accompanying paper (Chung, J.-K., Sekiya, F., Kang, H.-S., Lee, C., Han, J.-S., Kim, S. R., Bae, Y. S., Morris, A. J., and Rhee, S. G. (1997) J. Biol. Chem. 272, 15980-15985), synaptojanin is identified as a protein that inhibits phospholipase D (PLD) activity stimulated by ADP-ribosylation factor and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2). Here, the purification from rat brain cytosol of another PLD-inhibitory protein that is immunologically distinct from synaptojanin is described, and this protein is identified as clathrin assembly protein 3 (AP3) by peptide sequencing and immunoblot analysis. AP3 binds both inositol hexakisphosphate and preassembled clathrin cages with high affinity. However, neither inositol hexakisphosphate binding nor clathrin cage binding affected the ability of AP3 to inhibit PLD. AP3 also binds to PI(4,5)P2 with low affinity. But the PI(4,5)P2 binding was not responsible for PLD inhibition, because the potency and efficacy of AP3 as an inhibitor of PLD were similar in the absence and presence of PI(4,5)P2. A bacterially expressed fusion protein, glutathione S-transferase-AP3 (GST-AP3), also inhibited PLD with a potency equal to that of brain AP3. The inhibitory effect of AP3 appeared to be the result of direct interaction between AP3 and PLD because PLD bound GST-AP3 in an in vitro binding assay. Using GST fusion proteins containing various AP3 sequences, we found that the sequence extending from residues Pro-290 to Lys-320 of AP3 is critical for both inhibition of and binding to PLD. The fact that AP3 is a synapse-specific protein indicates that the AP3-dependent inhibition of PLD might play a regulatory role that is restricted to the rapid cycling of synaptic vesicles.
We recently showed that rat brain cytosol contains proteins that inhibit the activity of partially purified brain membrane phospholipase D (PLD)1 stimulated by ADP-ribosylation factor (ARF) and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) (1). Sequential chromatography of the brain cytosolic preparation yielded four inhibitory fractions, which exhibited differential sensitivity to heat. The most heat-labile inhibitor (inhibitor IA) was purified to yield a 150-kDa protein, which, in the accompanying paper (2), we showed to be synaptojanin, a nerve terminal-enriched protein with inositol polyphosphate phosphatase (IPPase) activity, and which inhibits PLD by hydrolyzing PI(4,5)P2.
We now describe the purification of PLD inhibitor IB, which has an apparent molecular mass of 165 kDa and is less susceptible to heat treatment than synaptojanin. Sequences of tryptic peptides of inhibitor IB suggest that it is identical to clathrin assembly protein 3 (AP3). We further show that AP3 binds to PLD.
The sources of rat brains, bovine brain
phosphatidylcholine (PC), phosphatidylethanolamine (PE), GTPS,
PI(4,5)P2,
[choline-methyl-3H]dipalmitoylphosphatidylcholine
((pam)2PC) (50 Ci/mmol),
[2-palmitoyl-9,10-3H](pam)2PC (89 Ci/mmol), and n-octyl-
-D-glucopyranoside were as described previously (2). Phytic acid (inositol hexakisphosphate, or
IP6) and monoclonal antibodies to AP3 (clone AP180-I) were obtained from Sigma. PLD, ARF, human PLD1 (hPLD1), phospholipase C-
1
(PLC-
1), PLC-
1, and PLC-
1 were purified as described (2). Purified clathrin triskelia was a gift from Drs. Evan Eisenberg and
Ruo-Fan Jiang (National Institutes of Health). Rabbit antiserum specific to hPLD1 was as described (3).
Procedures for assaying the activities of PLD and PLC were as described in the accompanying paper (2).
Purification of PLD Inhibitor IBInhibitor IB was partially
purified from the 0-35% ammonium sulfate precipitate of rat brain
cytosol by sequential chromatography on DEAE-Sephacel, DEAE-5PW,
phenyl-5PW, and heparin-5PW columns as described (1). Fractions
corresponding to the major peak (5 mg of protein) from the last column
were dialyzed against buffer A (10 mM Mes-NaOH (pH 6.5), 1 mM EDTA, and leupeptin and aprotinin (each at 1 µg/ml))
and then applied to a Mono-S 5/5 column that had been equilibrated with
the same buffer. Proteins were eluted at 1 ml/min with a linear
gradient of 0-400 mM NaCl in buffer A for 60 min (Fig.
1A), and fractions (1.0 ml) were collected and assayed for
PLD-inhibitory activity. The peak fractions (fractions 25-28) were
pooled and concentrated. Three-fourths of the resulting sample was
divided into aliquots and stored at 70 °C. The remaining one-fourth (260 µg of protein in 50 µl) was applied to Superose 12 PC 3.2/30 column (3.2 × 300 mm, Pharmacia Biotech Inc.) and chromatography was performed with a Pharmacia-LKB SMART System equipped
with a µ separation unit and a µ precision pump. Proteins were
eluted at a flow rate of 40 µl/min with a solution containing 20 mM Tris-HCl (pH 7.5) and 150 mM NaCl (Fig.
1B). Fractions of 40 µl were collected, assayed for
PLD-inhibitory activity, and analyzed by SDS-polyacrylamide gel
electrophoresis (PAGE).
Peptide Sequence Analysis
Purified PLD inhibitor IB (500 µg) from the Superose 12 PC 3.2/30 column was denatured and reduced
by treatment with 50 mM Tris-HCl (pH 8.0), 6 M
guanidine hydrochloride, and 2 mM dithiothreitol. The
sulfhydryl groups were labeled with 2-nitro-5-thiobenzoate by adding
Ellman's reagent (5,5-dithiobis-(2-nitrobenzoic acid)) to a final
concentration of 10 mM, and the resulting conjugated protein was precipitated with 10% (w/v) trichloroacetic acid. The
pellet was washed with cold acetone, suspended in 50 mM
Tris-HCl (pH 8.0), and digested overnight at 37 °C with trypsin. The
generated peptides were applied to a C18 column (4.6 × 250 mm; Vydac) that had been equilibrated with 0.1% trifluoroacetic
acid, and were eluted at a flow rate of 1 ml/min with a 60-ml linear
gradient of 0-60% (v/v) acetonitrile in 0.1% trifluoroacetic acid.
Peptides were detected by measuring absorbance at 215 nm.
An expression vector (pGEX-3X-F1-20
(AS15)) that encodes a fusion protein consisting of
glutathione S-transferase (GST) and full-length
(915-residue) bovine AP3 has been described (4). Truncated DNAs
corresponding to the NH2-terminal 289 amino acid residues
(AP3-(1-289)), NH2-terminal 320 residues (AP3-(1-320)), and COOH-terminal 595 residues (AP3-(321-915)) were prepared by polymerase chain reaction using the plasmid pGEX-3X-F1-20
(AS15
) as the template. The 5
primer
(AAAAAAGCGGCCGCAAATGTCGGGCCAAACGCTC) for AP3-(1-289) and AP3-(1-320)
contained a NotI site. The 3
primers for AP3-(1-289)
(AAAAAAGCGGCCGCTTATTTCTTTCCTTCTAGGG) and AP3-(1-320)
(AAAAAAGCGGCCGCTTATTTAGCTGGTGTAGAATTAGG) also contained a
NotI site. For AP3-(321-915), a BamHI
site-containing 5
primer (AAAAAAGGATCCACTATCGACACGTCCCCAC) and an
EcoRI site-containing 3
primer
(AAAAAAGAATTCTTACAAGAAATCCTTGATG) were used.
The polymerase chain reaction products were purified, digested with
NotI or BamHI and EcoRI, and ligated
into pGEX-4T1 (Pharmacia) to construct the respective GST fusion
protein expression vector. Escherichia coli DH5 cells
were transfected with each of the expression vectors. The vectors
obtained were retransfected into E. coli BL21 cells. The
colonies containing each vector were grown at 37 °C. Expression of
the fusion proteins was then induced for 6 h at 30 °C with 0.1 mM isopropyl-
-D-thiogalactopyranoside. The
cells were sonicated in phosphate-buffered saline (PBS) and centrifuged
at 5,000 × g for 15 min. The resulting supernatant was
mixed with a 50% (w/v) slurry of glutathione-Sepharose 4B (Pharmacia)
in PBS and incubated at room temperature for 30 min. After
centrifugation at 5,000 × g for 15 min, the resulting
pellet was washed with 10 volumes of PBS containing protease inhibitors including N-tosyl-L-phenylalanine chloromethyl
ketone (0.5 µg/ml), N
-p-tosyl-L-lysine
chloromethyl ketone (0.5 µg/ml), leupeptin (1 µg/ml), and aprotinin
(1 µg/ml). Proteins were eluted from the Sepharose beads by
incubation at room temperature for 30 min with one volume of 10 mM reduced glutathione. SDS-PAGE analysis of the eluted
proteins revealed a major protein band (>90%) with the predicted
molecular size for each of the truncated GST-AP3 fusion proteins.
However, the preparation of the full-length GST-AP3 showed a number of
smaller bands in addition to the major band of 190 kDa for GST-AP3 plus
several smaller bands. This probably reflects the extreme
susceptibility of AP3 to proteolytic degradation (5, 6). Further
purification of GST-AP3 on a Superose 12 PC 3.2/30 column improved the
purity of the 190-kDa protein preparation to 70%. However, storage at
70 °C and subsequent thawing resulted in significant
degradation.
Phospholipid vesicles containing either PE and PC in a molar ratio of 16:1 or PE, PI(4,5)P2 and PC in a molar ratio of 16:1.4:1 were added to polycarbonate tubes (7 by 20 mm, Beckman) containing AP3 (inhibitor IB fraction from Mono-S column). The final reaction mixture contained 3.4 µM PC in a total volume of 30 µl of 20 mM Hepes-NaOH (pH 7.5). After incubation at 37 °C for 30 min and subsequently on ice for 5 min, the tubes were centrifuged at 100,000 × g for 15 min at 4 °C. The resulting supernatant was saved, and the pellet was rinsed once with 30 µl of 20 mM Tris-HCl (pH 7.5) and subsequently resuspended in 30 µl of the same buffer. Both supernatant and pellet fractions were subjected to SDS-PAGE on an 8% gel and subsequent immunoblot analysis with monoclonal antibodies to AP3.
Measurement of ProteinsThe amounts of AP3, GST-AP3,
GST-AP3-(1-289), GST-AP3-(1-320), GST-AP3-(321-915), and GST were
determined spectrophotometrically with extinction coefficients
(280) of 39,970, 80,650, 60,100, 60,100, 65,590, and
40,680, respectively (7).
Clathrin triskelia were assembled into cages by dialyzing overnight against isolation buffer (0.1 M Mes (pH 6.5), 1 mM EGTA, 0.5 mM MgCl2, and 3 mM CaCl2) as described (8). The assembled cages were precipitated by ultracentrifugation at 100,000 × g for 20 min at 4 °C, washed three times with 500 µl of the isolation buffer, and suspended in 150 µl of the buffer. The clathrin cage solution was centrifuged at 13,000 × g for 5 min to remove nonspecific aggregates. Complexes of AP3 with clathrin cage were formed by incubating preassembled clathrin cages at 0.45 µM (corresponding to the molarity of the triskelia) and 4.5 µM AP3 in 400 µl of isolation buffer without CaCl2 for 1 h on ice with rocking. The clathrin cage-AP3 complex was precipitated by ultracentrifugation at 100,000 × g for 20 min at 4 °C and was washed three times with 500 µl of the isolation buffer without CaCl2. The 1:1 stoichiometry of clathrin triskelia and AP3 in the complex (5) was confirmed by SDS-PAGE, followed by Coomassie Blue staining and Western blot analysis using antibodies to AP3.
Binding of GST Fusion Proteins to PLD1GST or GST fusion
proteins containing AP3-(1-289), AP3-(1-320), or AP3-(321-915) were
incubated with hPLD1 for 30 min on ice. The final reaction mixture
contained 1 µM GST or GST fusion protein and partial
purified hPLD in 1.5 µg of total protein in a total volume of 150 µl of binding buffer (50 mM Tris-HCl (pH 8.0), 10 mM KCl, 1 mM dithiothreitol, and 0.7%
n-octyl--D-glucopyranoside). Ten microliters
of 50% (w/v) slurry of glutathione-Sepharose 4B was added to the
mixture and incubated for 5 min on ice with rocking. After
centrifugation at 12,000 × g for 2 min, the resulting
supernatants were transferred and the pellets were washed three times
with 1 ml each of the binding buffer. Both the pellet and supernatant fractions were subjected to electrophoresis on an 8%
SDS-polyacrylamide gel, and hPLD1 was visualized by alkaline
phosphatase-immunoblotting using rabbit antibodies to hPLD1 (3).
We previously showed that chromatography of rat brain cytosol on a DEAE-Sephacel column yielded three partially resolved peaks when fractions were assayed for the ability to inhibit PLD activated by ARF and PI(4,5)P2 (1). Fractions corresponding to the first peak were further purified by high performance liquid chromatography on DEAE-5PW and phenyl-5PW columns, the latter of which yielded two well separated peaks, termed IA and IB. Chromatography of peak fractions containing inhibitor IB on a heparin-5PW column yielded one major and two minor inhibitor peaks (1). We have now subjected fractions corresponding to the major peak to sequential chromatography on a Mono-S column (Fig. 1A) and a Superose 12 gel filtration column (Fig. 1B). These procedures yielded single peak of PLD-inhibitory activity corresponding to an estimated molecular size of 370 kDa. Analysis of the peak fractions by SDS-PAGE revealed three broad bands with apparent molecular masses of 165, 155, and 135 kDa, as well as several minor bands (Fig. 1C). Repeating the purification procedure revealed that the relative abundance of the three proteins varied (as the relative intensity of the 165-kDa band decreased, those of the 155- and 135-kDa bands increased) and that the inhibitory activity correlated with none of the three bands alone but rather with the sum of all three. These results suggest that the three proteins are related and components of the PLD inhibitor IB.
The inhibitor IB preparation from the gel filtration column was
digested with trypsin, and the resulting peptides were isolated on a
C18 column (Fig. 2). Four of the purified
peptides were sequenced and all four sequences matched exactly those of
rat AP3 (9, 10); peptides 1-4 corresponded to AP3 residues 150-157,
789-795, 113-128, and 872-886, respectively. Immunoblot analysis of
inhibitor IB from the Mono-S column with a monoclonal antibody to AP3
detected not only the protein with 165-kDa protein but also the 155- and 135-kDa proteins (see Fig. 4), suggesting that the three proteins correspond to AP3 and its proteolytic fragments. The PLD inhibitor IB
is therefore subsequently referred to as AP3. It was previously shown
that AP3 migrates anomalously on SDS-PAGE because the high density of
acidic residues in the central domain interferes with SDS binding. Its
apparent molecular mass ranges from 155 to 185 kDa depending on the
conditions, whereas its molecular mass predicted from the corresponding
cDNA is 91.4 kDa (9, 10). Purified AP3 appears as multiple bands on
SDS-PAGE because it is highly susceptible to proteolytic attack (5, 6).
In addition, AP3 is phosphorylated (9, 11) and glycosylated (11), both
of which may cause broadening of AP3 bands on SDS-PAGE.
To exclude the possibility that a contaminating protein accounted for
the inhibition of PLD activity, we compared the effects of GST-AP3 and
purified rat brain AP3 (from the final gel filtration column) on hPLD1
activity. Bacterially expressed GST-AP3 was purified with the use of
glutathione-Sepharose beads and gel filtration chromatography. Both AP3
(from rat brain) and GST-AP3 inhibited hPLD1 activity stimulated by ARF
and PI(4,5)P2 in a concentration-dependent manner and exhibited similar potencies. GST alone had no effect on
hPLD1 activity (Fig. 3).
AP3 binds inositol polyphosphates with high affinity (7, 12). Because many inositol polyphosphate-binding proteins also bind PI(4,5)P2, it was possible that AP3 inhibited PLD activity by masking PI(4,5)P2. We therefore evaluated the ability of AP3 to bind PI(4,5)P2 with the use of lipid vesicles of identical composition to those used for the PLD assay. The amount of AP3 that associated with the PI(4,5)P2-containing vesicles was markedly greater than that associated with lipid vesicles containing only PE and PC (Fig. 4). When the amount of AP3 added to lipid vesicles was reduced by one-half, the ratio of the amount of AP3 associated with the pellet to that present in the supernatant was not affected (data not shown), suggesting that the number of binding sites was not a limiting factor. When the pellets were washed more than once, the amount of AP3 associated with the vesicles devoid of PI(4,5)P2 decreased to virtually undetectable levels, and the amount of AP3 associated with the PI(4,5)P2-containing vesicles was also reduced proportionally (data not shown). These results suggest that AP3 interacts with PI(4,5)P2 with low affinity.
The AP3-dependent inhibition of hPLD1, which is stimulated
by ARF and PI(4,5)P2, was measured with substrate vesicles
containing PE and
[choline-methyl-3H](pam)2PC and
either containing or not containing PI(4,5)P2. The
ARF-dependent activity of hPLD1 in the absence of
PI(4,5)P2 is ~5% of that in its presence, so the amount
of hPLD1 used to measure activity in the absence of
PI(4,5)P2 was 20 times that used in presence. AP3 inhibited
the generation of [3H]choline with similar efficacies and
potencies in the presence and absence of PI(4,5)P2 (Fig.
5A). To confirm this result, we measured the
production of phosphatidylethanol (PEtOH), as well as that of
phosphatidic acid (PA), from
[2-palmitoyl-9,10-3H](pam)2PC in
the presence of ethanol. The production of both PEtOH and PA was
inhibited by 43 nM AP3 to similar extents in the absence
and presence of PI(4,5)P2 (Fig. 5B). These
results suggest that the inhibition of hPLD1 by AP3 is not attributable to the weak interaction of AP3 with PI(4,5)P2.
We next investigated the effects of AP3 on the activities of PLC-1,
PLC-
1, and PLC-
1 with substrate vesicles identical to those used
for the PLD assay, with the exception that the 3H-labeled
lipid was PI(4,5)P2 rather than PC. At a concentration (62 nM) at which PLD activity was nearly maximally inhibited, AP3 did not substantially affect the activities of the various PLC
isoforms (data not shown). These data support the notion that the weak
binding of AP3 to PI(4,5)P2-containing vesicles does not
reduce the effective concentration of this phospholipid on the vesicle
surface.
The concentration-dependent inhibition of hPLD1 by AP3 was measured at three different concentrations of ARF and PC. The inhibition profiles were virtually identical when the concentrations of ARF and PC were reduced by up to 95 and 75%, respectively, suggesting that inhibition does not result from competition of AP3 with either ARF or PC for binding to PLD (data not shown).
AP3 binds IP6 with a dissociation constant of 0.24-1.2
µM, and IP6 inhibits the interaction of AP3
with clathrin (7, 12). At a concentration of 15 µM,
however, IP6 had no effect on the inhibition of PLD by AP3
measured in the absence or presence of PI(4,5)P2 (data not
shown). AP3 also binds clathrin cages with a high affinity (8, 13).
However, the capacity of the cage-bound AP3 to inhibit PLD was similar
to that of free AP3 (Fig. 6). Clathrin cages alone did
not affect PLD significantly. These results suggest that the region of
the AP3 that interacts with clathrin cages differs from that
responsible for interaction with PLD and that the inhibitory function
of AP3 is not affected by its cage binding.
Previous studies demonstrated that AP3 is readily cleaved by trypsin
into a 33-kDa (often a mixture of 30- and 33-kDa)
NH2-terminal and a 58-kDa COOH-terminal fragment (6, 8).
The clathrin assembly activity is associated with the 58-kDa fragment,
whereas the 33-kDa fragment contains a high affinity binding site for IP6 (8). To localize the region responsible for the PLD
inhibition, AP3 was subjected to limited digestion by trypsin, followed
by separation on a gel filtration column. Two protein peaks were observed (Fig. 7A). The SDS-PAGE indicated
that the first peak contained several proteins of higher apparent
molecular mass (130-180-kDa species), whereas the second peak was
mainly composed of the partially resolved 30- and 33-kDa fragments
(Fig. 7B). The 30- and 33-kDa fragments as well as the
58-kDa fragment, which is known to migrate abnormally on
SDS-polyacrylamide gel (seen as an apparent 100-kDa species) and known
to be poorly stained by Coomassie Blue because of its high content of
acidic residues, were also detected in the first peak by immunoblot
with antibodies to AP3 (data not shown), suggesting that under
non-denaturing conditions parts of the 30- and 33-kDa fragments remain
associated with the the acidic 58-kDa fragment. The proteins in both
the first and second peaks inhibited hPLD1, and the inhibitory activity
of the second peak correlated better with the elution profile of the
33-kDa protein than with that of the 30-kDa protein. The 30- and 33-kDa proteins were eluted from excised gel bands, and the sequences of their
NH2-terminal 12 residues were determined; both sequences were the same as the NH2-terminal sequence of AP3. These
results suggest that the NH2-terminal 33-kDa contains the
region required for PLD inhibition and its COOH-terminal 3-kDa region
might be critical for interaction with hPLD1.
To further delineate the AP3 sequence involved in hPLD1 inhibition, we
prepared GST fusion proteins containing the NH2-terminal 289 amino acid residues (AP3-(1-289)), NH2-terminal 320 residues (AP3-(1-320)), or COOH-terminal 595 residues (AP3-(321-915))
and tested their abilities to inhibit hPLD1. GST-AP3-(1-320) was
nearly as potent as AP3, whereas GST-AP3-(321-915) and
GST-AP3-(1-289) had almost no effect (Fig. 8). It
appears, therefore, that NH2-terminal 320 residues but not
the COOH-terminal 576 residues of AP3 are important for interaction
with hPLD1 and that the sequence extending from residues Pro-290 to
Lys-320 of AP3 is critical for the interaction.
The GST fusion proteins were also tested for the ability to bind hPLD1
by incubation with hPLD1, followed by precipitation with
glutathione-Sepharose 4B beads and immunoblot analysis of hPLD1 in
precipitate and supernatant fractions (Fig. 9). The
ability of GST-AP3 and GST-AP3-(1-320) coupled with the inability of
GST-AP3-(321-915), GST-AP3-(1-289), and GST to retain hPLD1 in the
precipitate support the notion that the sequence extending from
residues Pro-290 to Lys-320 of AP3 is involved in interaction with
hPLD1. GST-AP3 partially retained hPLD1 and the ratio of hPLD1 in
precipitate and supernatant varied (data not shown). This was probably
because the GST-AP3 preparations used in the experiment were
proteolyzed, to different extents, to AP3 fragments that were detached
from the GST moiety but able to bind hPLD1.
AP3 was independently discovered in a variety of contexts and has been termed AP 180 (14), AP3 (15), NP185 (16), and F1-20 (9). It binds to clathrin triskelia and promotes their assembly into a homogeneous population of 60-70-nm coats (5, 14). Clathrin-coated vesicles are dynamic organelles that participate in intracellular membrane trafficking, including the biogenesis and recycling of synaptic vesicles. In addition to AP3, clathrin-coated vesicles contain one or more other assembly proteins, including AP1, AP2, and auxilin (17, 18). AP1 and AP2 are tetramers and widely distributed, whereas AP3 and auxilin are monomers and present only in neuronal cells and tissues. In bovine brain, AP3 is localized predominantly in cytosol, with only ~30% associated with membranes (19).
AP3 binds IP6 with high affinity (7, 12). Like many other inositol polyphosphate-binding proteins, AP3 also binds PI(4,5)P2, as judged from the observation that AP3 associated with lipid vesicles containing PI(4,5)P2 but not with those vesicles lacking PI(4,5)P2. However, the interaction of AP3 with PI (4,5)P2 is of low affinity and was not responsible for inhibition of PLD. Furthermore, at a concentration that maximally inhibits PLD, AP3 had no effect on the PI(4,5)P2-hydrolyzing activities of three PLC isozymes assayed with substrate vesicles of lipid composition identical to that of those used for the PLD assay. This result suggests that the amount of PI(4,5)P2 that is bound to AP3 on the surface of substrate vesicles is negligible and that the inhibition of PLD by AP3 is not due to nonspecific interaction of AP3 with lipid vesicles.
AP3 purified from brain and bacterially expressed GST-AP3 were equally potent inhibitors of PLD. In an inhibition assay utilizing a partial trypsin digestion of AP3, the NH2-terminal 33-kDa fragment appeared to inhibit hPLD1 while the NH2-terminal 30-kDa fragment did not. Furthermore, the GST fusion protein comprising the NH2-terminal 320 residues of AP3 was able to inhibit hPLD1, whereas the GST fusion protein comprising the COOH-terminal 595 residues was ineffective. We were able to further localize the inhibitory site to 31 amino acid residues extending from the Pro-290 to Lys-320 of AP3 on the basis of the ability of GST-AP3-(1-320) and the inability of GST-AP3-(1-289) to inhibit hPLD1. Direct binding of AP3 to hPLD1 and the essential role of the 31 residues in the binding were also demonstrated by the fact that GST-AP3 and GST-AP3-(1-320) retain hPLD1 during sedimentation while GST, GST-AP3-(1-289), and GST-AP3-(321-915) do not. However, the AP3 binding does not appear to interfere with the binding of ARF or PC to hPLD1, as indicated by the observation that the inhibition by AP3 was not affected by changes in the concentration of ARF or PC.
The predicted primary structure and characterization of the products of limited proteolysis suggest a three-domain organization for AP3: a central 42-kDa domain flanked by 33-kDa NH2-terminal and 16-kDa COOH-terminal domains (6, 8, 10). The 33-kDa NH2-terminal domain of AP3 binds to clathrin triskelia but neither promotes the assembly of nor binds to preassembled clathrin cages, whereas the 58-kDa COOH-terminal region (the 42-kDa central domain plus the 16-kDa COOH-terminal domain) binds clathrin, induces assembly of a homogeneous population of clathrin cages in a manner similar to that of the full-length protein, and binds to preassembled clathrin cages (8). AP3 also binds IP6 with high affinity through the 33-kDa NH2-terminal domain, and IP6 binding inhibits clathrin assembly (7, 12). Our data show that IP6 binding does not affect inhibition of PLD by AP3. Our data also show that AP3 molecules that are associated with clathrin cages through their 58-kDa COOH-terminal region can inhibit PLD via their 33-kDa NH2-terminal region.
AP3 shares no significant sequence similarities with other known clathrin-binding proteins, such as the subunits of AP1 and AP2, or auxilin (9, 10). Murphy et al.(11) therefore suggested that AP3 performs an additional cellular function that distinguishes it from other clathrin assembly proteins. Although AP3 is a high affinity receptor for inositol polyphosphates, interaction of the latter with AP2 also modulates its clathrin assembly activity (20-22). Inhibition of PLD might therefore be the additional function that is specific to AP3. Immunoblot analysis with antibodies to AP1 or AP2 suggests that none of the PLD-inhibitory fractions from column chromatography of rat brain proteins contained AP1 or AP2 (data not shown).
Although the presence of PLD in synaptic vesicles has not been established, ARF-activated PLD is highly enriched in Golgi membranes (23) and the production of PA by PLD is a key event in the formation of coatomer-coated vesicles (24). Under certain conditions, increased concentrations of PA and PI(4,5)P2 are sufficient to promote the binding of coatomer to membranes (24). The generation of PA and that of PI(4,5)P2 are related events; PI(4,5)P2 stimulates PA production by activating PLD, and the PA thus produced enhances the synthesis of PI(4,5)P2 by activating PI(4)P 5-kinase, constituting a positive feedback loop for the generation of these two lipid messenger molecules (25, 26).
In the accompanying paper (2), we show that synaptojanin inhibits PLD by hydrolyzing PI(4,5)P2. Synaptojanin is a nerve terminal-specific IPPase and has been implicated in the trafficking of synaptic vesicles (27). Golgi membranes also contain an IPPase known as OCRL (28), which we showed is capable of inhibiting PLD. We therefore proposed that the hydrolysis of PI(4,5)P2, and consequent inhibition of PLD, by IPPases such as synaptojanin and OCRL might constitute a common mechanism for interrupting the positive feedback loop during membrane invagination and act as a signal for uncoating (2). The present study suggests that direct inhibition of PLD by AP3 without the involvement of PI(4,5)P2 might constitute an additional mechanism to halt the positive feedback loop and to signal uncoating. Because AP3 is a synapse-specific protein (29, 30), such a mechanism would be specific to synaptic vesicles.
The synaptic vesicle cycle at the nerve terminal shares many features with membrane trafficking in nonneuronal cells (31). However, synaptic transmission requires rapid transitions through each step of the cycle (32). For example, the membranes of empty synaptic vesicles are taken up within 1-5 s after exocytosis (33), which is much faster than most endocytotic events. This special requirement for speed might necessitate the additional, AP3-dependent mechanism for inhibition of PLD.
We thank E. M. Lafer for the E. coli strain expressing GST-AP3, Evan Eisenberg and Ruo-Fan Jiang for clathrin, and Rodney Levine and Michael Nauman for peptide sequencing.