(Received for publication, November 27, 1996, and in revised form, January 23, 1997)
From PSD-95/SAP90 is a member of membrane-associated
guanylate kinases localized at postsynaptic density (PSD) in neuronal
cells. Membrane-associated guanylate kinases are a family of signaling molecules expressed at various submembrane domains which have the PDZ
(DHR) domains, the SH3 domain, and the guanylate kinase domain.
PSD-95/SAP90 interacts with
N-methyl-D-aspartate receptors 2A/B,
Shaker-type potassium channels, and brain nitric oxide synthase through
the PDZ (DHR) domains and clusters these molecules at synaptic
junctions. However, neither the function of the SH3 domain or the
guanylate kinase domain of PSD-95/SAP90, nor the protein interacting
with these domains has been identified. We have isolated here a novel
protein family consisting of at least four members which specifically
interact with PSD-95/SAP90 and its related proteins through the
guanylate kinase domain, and named these proteins SAPAPs
(P90/SD-95-ssociated
roteins). SAPAPs are specifically expressed in
neuronal cells and enriched in the PSD fraction. SAPAPs
induce the enrichment of PSD-95/SAP90 to the plasma membrane in
transfected cells. Thus, SAPAPs may have a potential activity to
maintain the structure of PSD by concentrating its components to the
membrane area.
The PSD1 is a dense thickening of
postsynaptic submembranous cytoskeleton observed in electron
microscopy. Since PSD has a characteristic structure and is contiguous
to the presynaptic active zone, where neurotransmitter release occurs,
PSD is proposed to have several functions, such as the stabilization of
synaptic junctions, the concentration and regulation of
neurotransmitter receptors, and the induction of the transcriptions in
response to the synaptic transmission (reviewed in Ref. 1). Many
studies have revealed several components of PSD, including fodrin (2), tubulin (3), actin (4), calmodulin (5),
Ca2+/calmodulin-dependent protein kinase II
(6), and PSD-95/SAP90 (7-9). However, there still remain many
components to be identified and the mechanism how the individual
components are assembled to form PSD is unknown.
PSD-95/SAP90, one of the components of PSD, has a characteristic
molecular structure composed of three PDZ (DHR) domains, one SH3
domain, and one guanylate kinase domain. Three isoforms, SAP97,
PSD-93/chapsyn, and SAP102, have been reported and all isoforms keep
the same molecular structure (10-14). Recently, many studies have
revealed the function of the PDZ (DHR) domains. The PDZ (DHR) domain is
responsible for protein-protein interactions and identified in various
proteins (reviewed in Ref. 15). PSD-95/SAP90 and its isoforms interact
with NMDA receptors, Shaker-type potassium channels, and brain nitric
oxide synthase through the PDZ (DHR) domains to induce the clustering
of these molecules at PSD (12-14, 16-20). Thus, PSD-95/SAP90 is
important for the concentration of receptors and channels at PSD.
Another line of evidence for the importance of PSD-95/SAP90 at
synaptic junctions comes from the recent findings about MAGUKs (reviewed in Ref. 21). MAGUKs are a family of proteins expressed at
various submembrane domains. They include Drosophila
discs-large tumor suppressor gene (dlg-A) (22), nematode
lin-2 (23), palmitoylated erythrocyte membrane protein (p55)
(24), ZO-1 (25, 26), ZO-2 (27), CASK (28), PSD-95/SAP90 (7, 8), and the
isoforms of PSD-95/SAP90 (10-14). All the members of MAGUKs keep the
same molecular structures as that of PSD-95/SAP90. They have the PDZ (DHR), the SH3, and the guanylate kinase domains. Based on the characteristic molecular structure, MAGUKs are considered to play important roles in maintaining the structures of submembrane domains and to be involved in signaling at these domains. Indeed, genetic evidence indicates that MAGUKs are essential for maintenance of the
structures of cell junctions. The product of dlg-A is
expressed at separate junctions and neuromuscular junctions, and the
mutations of this gene lead to neoplastic overgrowth of imaginal discs
and morphological changes of synaptic bouton structures (22, 29). Lin-2 expressed at Pn.p cells in nematode is involved in the
localization of let-23, a receptor for lin-3, and
essential for vulval induction (23). In the context of the general
concept about MAGUKs, PSD-95/SAP90 is assumed to play pivotal roles in
signaling at synaptic junctions, not only in the clustering of
receptors, which may be mediated by domains other than the PDZ (DHR)
domains.
In contrast to the accumulation of information about the PDZ (DHR)
domains, our knowledge about the SH3 domain and the guanylate kinase
domain of PSD-95/SAP90 is limited. The SH3 domain is also responsible
for protein-protein interactions and identified in a wide variety of
proteins. However, the molecules interacting with the SH3 domain of
PSD-95/SAP90 have not been identified. The guanylate kinase domain, is
similar to an enzyme guanylate kinase, has the binding activity for GMP
and GDP, but does not show kinase activity (30). The mutations in this
domain of dlg-A cause the abnormalities of imaginal discs
(22). Thus, this domain should also have some essential role.
To clarify the function of PSD-95/SAP90 and to identify novel
components of PSD, we have tried to identify by use of the yeast two-hybrid method the molecules which interact with the region of
PSD-95/SAP90 containing the SH3 domain and the guanylate kinase domain,
and are localized at PSD. We have obtained novel proteins which are
homologous to each other. We named them SAPAPs
(P90/SD-95-ssociated roteins). SAPAPs function to induce the translocation of
PSD-95/SAP90 from the cytosol to the plasma membrane and may be a new
family of proteins involved in signaling at PSD.
The
bait vector, pBTM116 PSD-95-3, was constructed by subcloning the insert
encoding amino acid residues 430-724 of PSD-95/SAP90 into
EcoRI/SalI sites of pBTM116 (31). The yeast two-hybrid library constructed from adult rat brain cDNA was screened using pBTM116 PSD95-3 as a bait, as described previously (28).
To obtain full-length clones, rat brain
cDNA libraries in Prokaryote and
eukaryote expression vectors, bait constructs, and prey constructs were
constructed in pGexKG (32), pMalC2 (New England Biolabs), pCMV5 (a gift
of Dr. David W. Russell), pBTM116, and pVP16-3 using standard molecular
biology methods (33). pVP16-3 was constructed from pVP16 (a gift from
Dr. Stanley Hollenberg) as described previously (34). Various GST
fusion constructs and maltose-binding protein fusion constructs
contained the following amino acid residues: pGex PSD-95-1 and pMal
PSD-95-1, 1-724 of rat PSD-95/SAP90; pGex PSD-95-2 and pMal PSD-95-2,
1-431 of rat PSD-95/SAP90; pGex PSD-95-4, 430-724 of rat
PSD-95/SAP90; and pMal PSD-95-6, 534-724 of rat PSD-95/SAP90. Various
bait constructs contained the following amino acids: pBTM116 SAP97-5,
587-910 of rat SAP97; pBTM116 PSRP1-3, 526-835 of rat PSRP1; pBTM116
Dlg-1, 601-960 of dlg-A; pBTM116 p55-1, 163-466 of human
p55; pBTM116 CASK-11, 601-909 of rat CASK; pBTM116 ZO-1-1, 504-909 of
mouse ZO-1; pBTM116 PSD-95-9, 534-724 of rat PSD-95/SAP90; pBTM116
SAP97-4, 726-910 of rat SAP97; pBTM116 PSRP1-2, 645-835 of rat PSRP1;
pBTM116 Dlg-2, 770-960 of dlg-A; pBTM116 CASK-10, 711-909
of rat CASK; and pBTM116 PSD-95-11, 411-495 of rat PSD-95/SAP90. pVP16
SAPAP1-12 contained the full-length of SAPAP1. pCMV PSD-95-1 contained
the full-length of PSD-95/SAP90 and pCMV PSD-95-2 contained amino acid
residues 1-431 of PSD-95/SAP90. pCMV Myc vector was constructed by
ligating the oligonucleotides,
aattgccccccaacatggagcagaagcttatcagcgaggaggacctg/aattccaggtcctcctcgctgataagcttctgctccatgttggggggc, into the EcoRI site of pCMV5. pCMV Myc SAPAP1 was
constructed by ligating the polymerase chain reaction fragment encoding
the full-length of SAPAP1 into MluI/SalI sites of
pCMV Myc, so that SAPAP1 was expressed with amino acids MEQKLISEEDL as
the N-terminal tag. The partial cDNAs of rat SAP97,
dlg-A, and p55 were obtained by polymerase chain reaction
and the sequences were confirmed. GST fusion proteins and MBP fusion
proteins were expressed and purified as described previously (28).
COS cells were transfected using the
DEAE-dextran method as described previously (34). 293 cells were
transfected with the calcium phosphate method using mammalian
transfection kit (Stratagene).
COS cells transfected with pCMV Myc SAPAP1
were sonicated in lysis buffer (20 mM Hepes/NaOH, pH 8.0, 150 mM NaCl, 1 mM EDTA, 25 mM
n-octyl- The interactions of
PSD-95/SAP90 and SAPAPs were evaluated by use of Biacore biosensor
technology (Pharmacia Biosensor) based on the basic principles and
detection methods as described previously (35). The antibodies against
GST were covalently fixed on a CM5 research grade sensor chip by the
amine coupling kit (Pharmacia) and then the Escherichia coli
lysates containing approximately 500 pmol of GST-1201-3, GST-1305-2,
and GST were passed on the chip to gain the elevation of about 3800 RU.
GST-1201-3 contained amino acid residues 315-569 of SAPAP1. GST-1305-2
contained amino acid residues 480-577 of SAPAP2. The chip was
equilibrated with phosphate-buffered saline and various doses of
various purified MBPs in 200 µl of phosphate-buffered saline were
injected across the immobilized surfaces at a flow rate of 20 µl/min.
The binding activities (in RU) were measured as the difference between
the baseline value determined 10 s prior to the sample injection
and the measurements taken at the indicated time points. All
experiments were performed at 25 °C. Data were analyzed using
BIAevaluation program 2.1 (Pharmacia).
Polyclonal antibodies were raised against the
GST fusion proteins of SAPAP1 (anti-SAPAP1), -2 (anti-SAPAP2), -3 (anti-SAPAP3), and -4 (anti-SAPAP4) using rabbits. The antigens used
for each antibody contained the following amino acid residues:
anti-SAPAP1, 589-689 of SAPAP1; anti-SAPAP2, 569-676 of SAPAP2;
anti-SAPAP3, 560-654 of SAPAP3; and anti-SAPAP4, 592-688 of
SAPAP4. The amino acid residues used for the antibodies were
specific for each SAPAP and each antibody was confirmed to be specific
for each SAPAP. The antibody against PSD-95/SAP90 was raised against
the GST fusion protein containing the full-length of PSD-95/SAP90 using
rabbits (anti-PSD-95) or mice (monoclonal anti-PSD-95). The antibody
against NMDA receptors 2A/B and various second antibodies were
purchased from Chemicon. The monoclonal antibody against the Myc tag,
QE10, was obtained from American Type Culture Collection.
Embryos
were obtained from Wistar rats on the gestation day 20. Hippocampi were
isolated, dissociated, plated on poly-L-lysine-coated glass
coverslips (Matsunami), and cultured in minimal essential medium with
10% horse serum. After 4 days, the medium was replaced with minimal
essential medium supplemented with N2 supplement (36), 1 mg/ml
ovalbumin, 1 mM pyruvate, and 5 mM cytosine
arabinoside. Immunocytochemistry of hippocampal cells and 293 cells
were performed using a confocal imaging system (Bio-Rad MRC1024).
293 cells
transfected with various pCMV constructs were sonicated in lysis buffer
(20 mM Hepes/NaOH pH 7.4, 10 µM
(4-amidinophenyl)-methanesulfonyl fluoride, 10 mM
leupeptin, and 5 mg/liter aprotinin) using 300 µl/10-cm dish. The
lysates were centrifuged at 100,000 × g for 30 min at
4 °C. Comparable amounts of cytosol and membrane fractions were
analyzed through SDS-PAGE and immunoblotting with either anti-PSD-95
antibody or anti-SAPAP1 antibody.
Subcellular fractionation of fresh
rat brains was performed as described previously (37, 38). SDS-PAGE,
immunoblotting, and protein determination were performed using standard
procedures as described previously (28). RNA blots were purchased from CLONTECH and hybridized with uniformly labeled DNA
probes according to the manufacturer's protocol.
By
use of the yeast two-hybrid method, we first searched for the proteins
which interacted with the region of PSD-95/SAP90 containing both the
SH3 domain and the guanylate kinase domain. 1.4 × 106
yeast transformants with a rat brain cDNA library were screened. Nine positive clones were obtained and six clones of them were independent. Four clones (pPrey 1201, pPrey1305, pPrey1310, and pPrey1377) contained similar but not the same sequences, which showed
no homology to known proteins. The remaining two clones (pPrey1294 and
pPrey1314) did not show any homology to known proteins. We focused on
pPrey1201, pPrey1305, pPrey1310, and pPrey1377, since the preliminary
Northern blot analysis using the inserts from these prey clones showed
strong signals only in brain and named the corresponding proteins
SAPAP1 to -4.
To determine the whole structures of
SAPAPs, rat brain cDNA libraries were screened to obtain
overlapping cDNA clones for each SAPAP using the insert from the
corresponding prey clone as each probe. The sequence contexts of the
putative initiator methionines of SAPAP1, -3, and -4 agreed well with
the consensus sequence for initiator codons and were preceded by
in-frame stop codons (39). The clone of SAPAP2 had another methionine
at 237 bases upstream to the putative initiator codon. We identified the putative initiator methionine in the cDNA of SAPAP2 by
comparison with those of other SAPAPs. The coding regions of SAPAPs
showed 41-52% homology at the amino acid level, while 5
To confirm whether the clones of SAPAPs were full-length clones, we
constructed eukaryotic expression vectors using these clones and
checked the expressions of SAPAPs in COS cells. The polyclonal
antibodies were raised against SAPAP1, -2, -3, and -4 using peptides
specific for each SAPAP. The antibodies against SAPAP1, -2, and -4 recognized the proteins with Mr of about 140,000 and the antibody against SAPAP3 recognized the protein with a Mr of about 120,000 in COS cells transfected
with each pCMV construct (data not shown). The sizes of the proteins
detected in the transfected COS cells were different from the
calculated sizes based on the sequences. However, these antibodies also
recognized the proteins with Mr of 140,000 or
120,000 in brain homogenates (data not shown). Therefore, clones of
SAPAP1 to -4 are concluded to be the full-length clones of SAPAPs, and
SAPAPs may show abnormal mobilities on SDS-PAGE due to
post-translational modifications.
We examined by the
yeast two-hybrid method whether SAPAPs interact with MAGUKs other than
PSD-95/SAP90. Various bait constructs which contained the SH3 domain
and the guanylate kinase domain of various MAGUKs were prepared and
tested for interactions with original prey clones (Table
I). All prey clones of SAPAPs (pPrey1201, pPrey1305, pPrey1310, and pPrey1377) interacted with bait
constructs containing both the SH3 domain and the guanylate kinase
domain of not only PSD-95/SAP90, but also SAP97, PSD-95/SAP90-related protein 1 (PSRP1), and dlg-A. The clones did not interact
with the bait constructs containing the SH3 domain and the guanylate kinase domain of p55, CASK, or ZO-1. PSRP1 is a novel isoform of
PSD-95/SAP90, which is considered to be an alternative splicing isoform
of SAP1022. Of these MAGUKs, PSD-95/SAP90, SAP97, PSRP1, and
dlg-A form a closely related family different from p55,
CASK, and ZO-1. All SAPAPs interacted specifically with members of this
closely related family.
Table I.
Interaction of SAPAPs with various MAGUKs
Takai Biotimer Project,
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
Note added in Proof
REFERENCES
Yeast Two-hybrid Screening and -Galactosidase Assays
-Galactosidase assays were performed as described previously
(28).
ZAP II (Stratagene) were screened using the
inserts from pPrey1201, pPrey1305, pPrey1310, and pPrey1377 as
described previously (28). For pPrey1201, 1 × 106
clones were screened and 25 positive clones were obtained. For pPrey1305, 2 × 106 clones were screened and 10 positive clones were obtained. For pPrey1310, 1 × 106
clones were screened and 60 positive clones were obtained. For pPrey1377, 1 × 106 clones were screened and 23 positive clones were obtained. Two at least overlapping clones for each
SAPAP were analyzed. DNA sequencing was performed by the dideoxy
nucleotide termination method using an ABI373 DNA sequencer.
-D-glucoside, 1 mM
phenylmethylsulfonyl fluoride, 10 mM leupeptin, and 5 mg/liter aprotinin) using 500 µl per of each 10-cm dish. The lysates
were centrifuged at 100,000 × g for 30 min at 4 °C.
1 ml of the supernatant was incubated with 30 pmol of various GST
fusion proteins immobilized on the glutathione-Sepharose 4B beads for
2 h at 4 °C. The beads were collected by centrifugation, washed
with lysis buffer four times, and analyzed through SDS-PAGE and
immunoblotting with the monoclonal antibody against the Myc tag.
Isolation of SAPAPs Through the Yeast Two-hybrid Screening
and 3
non-coding regions did not show significant homology (Fig.
1, A and B). SAPAP1 to -4 showed
calculated Mr values of 110,172, 110,134, 106,123, and 108,678, and consisted of 992, 980, 977, and 992 amino
acids, respectively. The amino acid sequences of SAPAP1 to -4 predicted hydrophilic proteins without a transmembrane region or a signal sequence. All SAPAPs had proline-rich domains in the middle region and
the C-terminal region. Data bank searches revealed no significant homology to known proteins. pPrey1201, pPrey1305, pPrey1310, and pPrey1377 contained amino acid residues 315-556 of SAPAP1, 480-577 of
SAPAP2, 463-605 of SAPAP3, and 370-617 of SAPAP4, respectively. Based
on the homology conserved among the peptides from these four pPrey
clones, the domain of SAPAPs essential for the interaction with the
region of PSD-95/SAP90 used as the bait can be tentatively concluded to
be located in the region of about 80 amino acid residues in the middle
portion (Fig. 1A).
Fig. 1.
Amino acid sequences of SAPAPs.
A, alignment of SAPAPs. Sequences are shown in single-letter
amino acid code and numbered on the right. Residues
conserved among all SAPAPs are shown on a black background.
Residues conserved among three SAPAPs are shown on a gray
background. The potential interacting domain with PSD-95/SAP90 is
underlined. B, SAPAPs' sequence homologies. The numbers represent the percentages of the amino acid
homologies between the indicated SAPAPs.
[View Larger Version of this Image (91K GIF file)]
-galactosidase activity (unit/mg protein n = 3 ± S.D.) in the yeast
lysates was measured in triplicate. Data represent the mean values with
±S.D.
Yeast clones cotransfected with the prey pVP16 and the bait pBTM116
vectors in the indicated combinations were selected on the selection
plates and grown in selection medium. pBTM116 PSD-95-3, pBTM116
SAP97-5, pBTM116 PSRP1-3, pBTM116 dlg-1, pBTM116 p55-1, pBTM116
CASK-11, and pBTM116 ZO-1-1 contained both the SH3 domain and the
guanylate kinase domain of various MAGUKs. pBTM116 PSD-95-9, pBTM116
SAP97-4, pBTM116 PSRP1-2, pBTM116 dlg-2, and pBTM116 CASK-10 contained the guanylate kinase domain of various MAGUKs. pBTM116 PSD-95-11 contained the SH3 domain of PSD-95/SAP90. pVP16 SAPAP1-12 contained the full length of SAPAP1. The
-galactosidase activity (unit/mg protein n = 3 ± S.D.) in the yeast
lysates was measured in triplicate. Data represent the mean values with
±S.D.
Control (pVP16)
SAPAP1 (pPrey1201)
SAPAP2
(pPrey1305)
SAPAP3 (pPrey1310)
SAPAP4 (pPrey1377)
Full-length
of SAPAP1 (pVP16 SAPAP1-12)
Control
(pBTM116)
NDa
ND
ND
ND
ND
ND
Constructs containing the SH3 and guanylate kinase domains
PSD-95/SAP90
(pBTM116
PSD-95-3)
ND
(5.9
± 0.1) × 103
(1.7 ± 0.1) ×103
(2.4
± 0.1) ×102
(1.5
± 0.2) ×103
(6.9 ± 0.4) × 10
SAP97
(pBTM116 SAP97-5)
ND
(3.7
± 0.4) ×102
(1.4 ± 0.1) ×103
(4.3
± 0.1) ×102
(2.2
± 0.1) ×103
NTb
PSRP1(SAP102)
(pBTM116 PSRP1-3)
(2.1
± 0.3) ×102
(3.4 ± 0.1) ×103
(3.3
± 0.2) ×103
(7.5 ± 0.3) ×102
(2.1
± 0.3) ×103
NT
dlg-A
(pBTM116
dlg-1)
ND
(4.3
± 0.6) ×102
(1.2 ± 0.1) ×103
(5.1
± 0.1) ×102
(3.1
± 0.1) ×103
NT
p55
(pBTM116
p55-1)
ND
ND
ND
ND
ND
NT
CASK
(pBTM116
CASK-11)
ND
ND
ND
ND
ND
NT
ZO-1
(pBTM116
ZO-1-1)
ND
ND
ND
ND
ND
NT
Constructs containing the guanylate kinase domain
PSD-95/SAP90
(pBTM116
PSD-95-9)
ND
(4.9
± 0.1) ×102
(1.2 ± 0.1) ×103
(3.4
± 0.3) ×102
(3.3 ± 0.2) ×103
(3.2
± 0.4) ×10
SAP-97
(pBTM116
SAP97-4)
ND
(5.8
± 0.3) ×102
(8.9 ± 0.8) ×102
(4.4
± 0.1) ×102
(4.1
± 0.2) ×103
NT
PSRP1(SAP102)
(pBTM116
PSRP1-2)
ND
(7.7
± 0.7) ×102
(1.3 ± 0.1) ×103
(4.0
± 0.1) ×102
(3.8
± 0.4) ×103
NT
Dlg-A
(pBTM116
dlg-2)
ND
(5.2
± 0.1) ×102
(1.6 ± 0.1) ×103
(3.8
± 0.1) ×102
(3.3
± 0.1) ×102
NT
CASK
(pBTM116
CASK-10)
ND
ND
ND
ND
ND
NT
A construct containing the SH3 domain
PSD-95/SAP90
(pBTM116
PSD-95-11)
ND
ND
ND
ND
ND
ND
a
ND, not detectable.
b
NT, not tested.
To determine which region of PSD-95/SAP90 is necessary for interaction with SAPAPs, we tested the interactions of the original prey clones and bait constructs containing various regions of PSD-95/SAP90 and its related proteins. The prey clones of SAPAPs interacted with the bait constructs containing only the guanylate kinase domains of PSD-95/SAP90, SAP97, PSRP1, and dlg-A, but not with the bait constructs containing the SH3 domain of PSD-95/SAP90 (Table I). Therefore, the guanylate kinase domains of PSD-95/SAP90 and its related proteins are sufficient for interaction with SAPAPs. We also tested the interactions between the bait construct containing the SH3 domain or the guanylate kinase domain of PSD-95/SAP90 and the prey construct containing full-length SAPAP1. The bait construct containing the SH3 domain did not interact even with the prey construct containing full-length SAPAP1 in contrast to the bait construct containing the guanylate kinase domain (Table I). This result suggests that the SH3 domain of PSD-95/SAP90 is not involved in the interaction with SAPAP1.
In Vitro Interaction of SAPAPs with PSD-95/SAP90To examine
the interaction of PSD-95/SAP90 and SAPAPs in vitro, the
extracts of COS cells expressing SAPAP1 with the Myc tag at the N
terminus were incubated with various GST fusion proteins fixed on
glutathione-Sepharose 4B beads. The bound proteins were analyzed by
SDS-PAGE followed by immunoblotting with the antibody against the Myc
tag. SAPAP1 was bound to GST-PSD-95-1 containing full-length
PSD-95/SAP90 and GST-PSD-95-4 containing both the SH3 domain and the
guanylate kinase domain (Fig. 2). SAPAP1 was not bound
to either GST or GST-PSD-95-2 containing three PDZ (DHR) domains (Fig.
2).
To confirm whether SAPAPs directly interact with PSD-95/SAP90, Biacore
biosensor technology was used. GST fusion proteins containing the
potential PSD-95/SAP90 interacting region of SAPAPs or GST itself were
fixed on the Biacore sensor chip through the antibody against GST. MBP
fusion protein containing the guanylate kinase domain of PSD-95/SAP90
was flowed on the chip and the affinities for the GST fusion constructs
were measured (Fig. 3). The Kd value
of PSD-95/SAP90 for the GST fusion protein of SAPAP1 was calculated to
be 2.1 × 107 M from the association
(ka = 2.7 × 103
M
1 s
1) and dissociation
(kd = 5.6 × 10
4
s
1) constants. The Kd value of
PSD-95/SAP90 for the GST fusion protein of SAPAP2 was calculated to be
2.3 × 10
7 M from the association
(ka = 2.6 × 103
M
1 s
1) and dissociation
(kd = 5.9 × 10
4
s
1) constants. The MBP fusion protein of SNAP-25A did not
interact with SAPAPs (data not shown). Neither the MBP fusion protein
containing three PDZ (DHR) domains nor the MBP fusion protein
containing the SH3 domain interacted with SAPAPs (data not shown).
Specific Expression of SAPAPs in Brain
Northern blot analysis
using the coding region of SAPAP1 showed 6.4-kb major and 5.2-kb minor
hybridizing mRNAs in rat brain, possibly because of differential
polyadenylation. In rat testis, mRNA with a smaller size was weakly
observed. The significance of this weak signal was unknown. Other rat
tissues tested, including heart, spleen, lung, liver, skeletal muscle,
and kidney, did not show any detectable signal (Fig.
4A). Northern blot analysis using the coding
region of SAPAP2 showed a 8.0-kb major hybridizing mRNA and three
weak smaller signals, only in brain (Fig. 4B). Northern blot
analyses of SAPAP3 and -4 showed 5.0-kb and 4.2-kb hybridizing
mRNAs only in brain, respectively (data not shown). In Western blot
analysis using the polyclonal antibody against SAPAP1, the signal was
detected only in rat brain (Fig. 4C). Western blot analyses
using antibodies against SAPAP2, -3, and -4 showed similar results
(data not shown).
Specific Expression of SAPAPs in Neurons
To determine whether
SAPAPs are neuronal proteins, we stained these proteins in primary
cultured hippocampal neurons from rat embryo using antibodies. The
antibody against SAPAP1 stained in the dendrites, the cell bodies, and
the nuclei of hippocampal neurons (Fig. 5A).
When the antibody was preincubated with an excess amount of GST fusion
protein of SAPAP1, the signals detected in the dendrites and cell
bodies disappeared, while the signals in the nuclei remained (Fig.
5B). Therefore, signals detected in dendrites and cell
bodies were considered to be specific. This staining pattern was
similar to that of PSD-95/SAP90 (data not shown). The glia cells were
not stained with the antibody against SAPAP1 (data not shown).
Essentially the same results were obtained using the antibodies against
SAPAP2, -3, and -4 (data not shown).
Enrichment of SAPAPs in PSD
Western blot analysis of the
subcellular fractions of rat brain indicated that proteins detected
with antibody against SAPAP1 were enriched in the synaptosome fraction
(P2). The subfractionations of synaptosomes into the synaptic vesicle
fraction (S3), the lysed synaptosomal membrane fraction (P3), and the
synaptosomal membrane fraction (SPM) showed that SAPAP1 was enriched in
SPM, in which PSD-95/SAP90 and NMDA receptors 2A/B were also enriched.
PSD was prepared by Triton X-100 extraction of SPM. SAPAP1 was further enriched in PSD (Fig. 6). Western blot analyses using
antibodies against SAPAP2, -3, and -4 showed essentially the same
results (data not shown).
Translocation of PSD-95/SAP90 from the Cytosol to the Plasma Membrane Depending on the Presence of SAPAPs
In the last set of
experiments, we expressed SAPAPs and PSD-95/SAP90 in various
combinations in 293 cells to examine whether SAPAPs interacted with
PSD-95/SAP90 in intact cells. When PSD-95/SAP90 was expressed in 293 cells, PSD-95/SAP90 was located in the cytosol (Fig.
7A). On the other hand, SAPAP1 was located at
the plasma membrane in 293 cells (Fig. 7B). Coexpression of
PSD-95/SAP90 with SAPAP1 caused the translocation of PSD-95/SAP90 from
the cytosol to the plasma membrane where SAPAP1 was located (Fig. 7C). However, when the pCMV construct encoding only three
PDZ (DHR) domains of PSD-95/SAP90 (pCMV PSD-95-2) was used, the
distribution of the product of this construct was not affected by
SAPAP1 (Fig. 7D). Other SAPAPs showed essentially the same
effect on the translocation of PSD-95/SAP90 (data not shown).
The translocation of PSD-95/SAP90 depending on the presence of SAPAPs
was also verified through subfractionations of transfected 293 cells as
described above. SAPAP1 was recovered in the membrane fraction of the
transfected 293 cells, both when it was expressed alone and when it was
coexpressed with PSD-95/SAP90 (Fig. 8A). PSD-95/SAP90 was mainly recovered in the cytosol fraction in the transfected 293 cells, when it was expressed alone, but was recovered in the membrane fraction, when it was coexpressed with SAPAP1 (Fig.
8B). The protein containing only the PDZ (DHR) domains of PSD-95/SAP90 was recovered in the cytosol fraction, even when coexpressed with SAPAP1 (Fig. 8C). Similar results were
obtained using other SAPAPs than SAPAP1 (data not shown).
In this article, we have reported four novel proteins which interact with the guanylate kinase domain of PSD-95/SAP90. We have named these proteins SAPAPs. We have raised polyclonal antibodies against each SAPAP and confirmed the expressions of SAPAPs in COS cells. The Mr of SAPAPs detected in COS cells transfected with pCMV constructs containing the coding regions of the clones of SAPAPs are the same as those in brain. Therefore, these clones are considered to be full-length clones.
We have concluded that SAPAPs actually interact with PSD-95/SAP90 based
on the following lines of evidence: 1) the recombinant constructs of
SAPAPs interact with PSD-95/SAP90 in vitro with the
Kd values of about 2 × 107
M, these values are in a reasonable range; 2) SAPAPs are
expressed in neurons and highly enriched in the PSD fraction in which
PSD-95/SAP90 are enriched; 3) SAPAPs also show interactions with
PSD-95/SAP90 in 293 cells and induce the translocation of PSD-95/SAP90
from the cytosol to the plasma membrane; and 4) in the preliminary yeast two-hybrid screenings for SAPAPs-associated proteins,
PSD-95/SAP90, SAP97, and PSD-93/chapsyn are obtained
reproducibly.2
SAPAPs are about 50% homologous to each other at the amino acid level. The 200 amino acids in the C-terminal region are especially well conserved, although the conserved regions are also observed in the N-terminal region and in the middle region. SAPAPs show no homology to known proteins and the sequences do not suggest the functions of SAPAPs. Since the original prey clones obtained from the yeast two-hybrid screening contain various partial sequences of SAPAPs, we can tentatively map the PSD-95/SAP90-interacting region of SAPAPs. This region is located in the middle region and encompasses about 80 amino acid residues, which do not contain any known motif. The potentially interesting domains are the proline-rich domains conserved in the middle (PPPI/VPP) and C-terminal regions (PPPV/DP and PPPI/VP). The proline-rich domains are generally considered to interact with the SH3 domain. Although the proline-rich domains in the middle regions of SAPAP1, -2, and -4 have the consensus motif of the sequences of the proteins which interact with the SH3 domain of Grb2, the surrounding amino acid contexts of other proline-rich domains do not completely fit to the criteria of the proline-rich domains reported to interact with the SH3 domain (40, 41). It is at present unknown whether these domains function for the interaction with other molecules.
The attempt to identify the SAPAP-interacting region of PSD-95/SAP90 has first been performed using the yeast two-hybrid method. The original prey clones of SAPAPs interact with the bait construct containing the guanylate kinase domain of PSD-95/SAP90 alone, and not the bait construct containing the SH3 domain. Thus, the guanylate kinase domain is sufficient for the interaction. The bait construct containing the SH3 domain alone does not interact with the prey clone containing the full-length of SAPAP1, either. The expression of this bait construct is confirmed, since PSD-95/SAP90 itself is obtained from the yeast two-hybrid screening using this bait construct.2 Furthermore, since no SAPAPs are obtained from the yeast two-hybrid screening using the SH3 domain of PSD-95/SAP90 as a bait,2 SAPAPs are not considered to be the interacting proteins with the SH3 domain of PSD-95/SAP90. SAPAP1 expressed in COS cells do not interact with the GST fusion protein containing only three PDZ (DHR) domains of PSD-95/SAP90. SAPAPs do not interact with the product of the construct containing only three PDZ (DHR) domains in the transfected 293 cells either. Therefore, the guanylate kinase domain of PSD-95/SAP90 is concluded to be the interacting domain with SAPAPs.
SAPAPs are expressed in brain and highly enriched in the PSD fraction. However, in primary cultured rat hippocampal cells, the specific immunoreactivities detected by the antibodies are almost uniformly distributed in the cell bodies and the dendrites of neurons, and do not show high enrichment at synaptic junctions. This pattern is almost the same as that observed for PSD-95/SAP90. The reason why the enrichment of SAPAPs and PSD-95/SAP90 at synaptic junctions is not observed in primary cultured rat hippocampal cells is not clear. The establishment of the intact synaptic junctions may require some conditions, such as close contacts and continuous stimuli between neurons, which are not supplied in the primary cultures of hippocampal cells.
Although the functions of SAPAPs are unknown, they may function as anchoring molecules to facilitate the interactions of PSD-95/SAP90 and the molecules located at the plasma membrane, since SAPAPs induce the enrichment of PSD-95/SAP90 at the plasma membrane in 293 cells. However, preliminary experiments suggest that SAPAPs have no effect on the clustering of NMDA receptor 2A by PSD-95/SAP90, although further studies are necessary. As discussed above, not all the proline-rich domains of SAPAPs have the consensus motif for the interacting domains with the SH3 domains. However, there is a possibility that some of the proline-rich domains of SAPAPs may interact with unidentified molecules and SAPAPs may work as an adaptor between PSD-95/SAP90 and these molecules.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U67137[GenBank]-67140 for SAPAPs and U53367[GenBank] for PSRP1.
We thank Dr. D. W. Russell (University of Texas Southwestern Medical Center at Dallas) for pCMV vector, Dr. S. Nakanishi (Kyoto University, Kyoto, Japan) for cDNA of NMDA receptor 2A, Dr. S. Tsukita (Kyoto University, Kyoto, Japan) for cDNA of mouse ZO-1, Dr. S. Hollenberg (University of Washington, Seattle, WA) for the materials of the yeast two-hybrid method, and Dr. H. Nakanishi (Takai Biotimer Project, ERATO) for pCMV Myc vector. Some constructs used in this study were produced by Y. H. in the laboratory of Dr. T. C. Südhof (University of Texas Southwestern Medical Center at Dallas), who generously permitted use of these constructs.
After the submission of this work, two groups have independently identified the molecules interacting with the guanylate kinase domain of PSD-95/SAP90 (Kim, E., Naishitt, S., Hsueh, Y.-P., Rao, A., Rothschild, A., Craig, A. M., and Sheng, M. (1997) J. Cell Biol. 136, 669-678; Sato, K., Yanai, H., Senda, T., Kohu, K., Nakamura, T., Okumura, N., Matsumine, A., Kobayashi, S., Toyoshima, K., and Akiyama, T. personal communication). GKAP, the protein reported by Kim et al. is an alternative splicing isoform of SAPAP1. DAP1a and DAP2 identified by Sato, K. et al. are the same as SAPAP1 and SAPAP2, respectively. DAP1b is the same as GKAP.