(Received for publication, December 20, 1996)
From the Division of Cellular Biochemistry,
Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam,
The Netherlands, the
Department of Anatomy, Tohoku University
School of Medicine, Sendai 980, Japan, and the ** Department of
Biochemistry, Institute of Medical Science, University of Tokyo, Tokyo
108, Japan
Diacylglycerol kinase (DGK) attenuates levels of
second messenger diacylglycerol in cells and produces another
(putative) messenger, phosphatidic acid. We have previously purified a
110-kDa DGK from rat brain (Kato, M., and Takenawa, T. (1990)
J. Biol. Chem. 265, 794-800). Here we report the
cDNA cloning from human brain and retina cDNA libraries. The
cDNA encodes a novel DGK isotype, termed DGK, of 941 amino acids
with an apparent molecular mass of 110 kDa. DGK
contains a
C-terminal putative catalytic domain, which is present in all
eukaryotic DGKs. In contrast to other DGK isotypes, DGK
contains
three cysteine-rich domains instead of two. The third cysteine-rich
domain is most homologous to the second one in other DGK isotypes. This
particular sequence homology extends C-terminally beyond the typical
cysteine/histidine core structure and is DGK-specific. DGK
furthermore contains various domains for protein-protein interaction,
such as a proline- and glycine-rich domain with a putative SH3
domain-binding site and a pleckstrin homology domain with an
overlapping Ras-associating domain. DGK
is expressed in the brain
and, to a lesser extent, in the small intestine, duodenum, and liver.
In situ hybridization of DGK
mRNA in adult rat brain
reveals high expression in the cerebellar cortex and hippocampus.
DGK
activity in COS cell lysates is optimal toward diacylglycerols
containing an unsaturated fatty acid at the sn-2
position.
Diacylglycerol kinase (DGK1; EC
2.7.1.107) plays an important role in signal transduction (1-3). It
phosphorylates the second messenger diacylglycerol (DG) to phosphatidic
acid (PA) and is therefore thought to attenuate the activation of
protein kinase C (PKC), for which DG is a physiological activator (4).
In addition, the product of DGK (PA) may play a second messenger role
as well (3, 5, 6). PA has been shown to activate a number of enzymes
involved in signal transduction, including PKC- (7), unidentified
protein kinases (8, 9), phospholipase C-
1 (10), and
polyphosphoinositide kinases (11, 12), in vitro.
Furthermore, PA binds to and may regulate the translocation and
subsequent activation of the protein kinase Raf-1 (13) and the
protein-tyrosine phosphatase PTP1C (14).
Various isotypes of mammalian DGK have been identified and show their
own remarkably cell-specific expression patterns among a wide variety
of cell types. The first group of highly homologous isozymes that has
been cloned, DGK (15-17), DGK
(18), and DGK
(19, 20) (also
named DGK-I, -II, and -III, respectively), has an apparent molecular
mass in the 80-90-kDa range. These DGKs are characterized by a
conserved N-terminal domain of unknown function and EF-hands that bind
Ca2+ (2). A C-terminal (putative) catalytic domain and two
cysteine-rich (or zinc-finger) domains (CRDs) are common to all DGKs
cloned thus far. A recently characterized 64-kDa isotype, DGK
, lacks EF-hands, but is otherwise structurally similar to the first group of
DGKs and is highly selective for arachidonate-containing substrates (21). Two other recently cloned isozymes, DGK
and DGK
(130-140 kDa), contain a pleckstrin homology domain (PH domain) near the N
terminus (22, 23). Finally, DGK
(also named DGK-IV; 104 kDa) is
characterized by four tandem ankyrin-like repeats near the C terminus
(24, 25), a nuclear targeting motif (25), and a region that is
homologous to the phosphorylation site of the MARCKS
(yristoylated lanine-ich
inase ubstrate) protein (24).
The sequence of DGK
closely resembles that of the eye-specific DGK2
from Drosophila encoded by the RgdA gene (26). A
RgdA mutation inactivates DGK2 and causes retinal
degeneration, illustrating the essential role of DGK in neurologic
functions.
Here we report the molecular cloning of a new isozyme, named DGK,
with several unique and interesting structural features. It contains
three instead of two CRDs and a PH domain at a different location than
in DGK
and DGK
. The first half of this PH domain overlaps a
recently identified putative Ras-binding site, the so-called
Ras-associating domain (RA domain) (27). DGK
furthermore contains a
proline- and glycine-rich domain at the N terminus, but lacks EF-hands.
Each of these conserved domains is typically contained in signaling
molecules and is thought to mediate lipid-protein or protein-protein
interaction in signal-transducing complexes. The structural properties
of this novel DGK
and its tissue distribution are presented here in
the context of what is known about other DGK isotypes.
Radiolabeled nucleotides, Hybond-N nylon membranes, and enhanced chemiluminescence (ECL) reagents were from Amersham Corp. Acrylamide was from Serva. ATP, restriction enzymes, T4 DNA ligase, and Klenow enzyme were from Boehringer Mannheim. Taq polymerase was from Life Technologies, Inc. Lipids were from Sigma. T7 DNA polymerase and oligonucleotide primers were from Pharmacia Biotech Inc.
cDNA Cloning and SequencingThe peptides
ATPVQVDGEPWIQAPGH and EIRLQVEQQEVELPSIEGL were obtained from purified
110-kDa DGK protein (see "Results") through digestion with lysyl
endopeptidase and fractionation by C18 reverse-phase high
pressure liquid chromatography with a 0-60% gradient of acetonitrile in 0.1% trifluoroacetic acid. From these peptides, we derived the
respective synthetic oligonucleotides
GCCACCCCTGTGCAGGTGGATGG(A/G)GAGCCCTGGATCCAGGCCCCTGGCCAC and
GAGATCAGATTACAGGTGGAACAGCAGGAGGTGGAGTTACCCTCGATAGAGGGCTTA as
probes to screen a
gt10 random-primed cDNA library from
rat brain according to standard protocols (28). From the resulting 2-kb
positive clone, a 400-bp PstI fragment was used for further screening of human cDNA libraries (see "Results"). For
sequencing, increasing nested deletions were made using the
double-stranded nested deletion kit (Pharmacia). Nucleotide sequencing
was done by the dideoxy chain termination method (29) as well as
automatically on a Model 373 DNA Sequencer (ABI Advanced
Biotechnologies, Inc.). Sequence data were analyzed using Genetics
Computer Group software (30).
Total
RNA was isolated from rat tissues using the LiCl method (28). RNA (10 µg) was incubated with random hexamers (50 µM final
concentration) and annealed for 5 min at 65 °C, followed by 10 min
at room temperature. cDNA synthesis was performed for 1 h at
37 °C in a 20-µl reaction mixture containing 50 mM
KCl, 10 mM Tris-HCl (pH 8.3), 1.5 mM
MgCl2, 0.01% gelatin, 1 mM dNTPs, 2 units/ml
RNase inhibitor, and 10 units of reverse transcriptase (SuperScript,
Life Technologies, Inc.). After inactivation (5 min at 90 °C), the
reaction mixture was diluted 5-fold, and 5 µl of this diluted mixture
was used for PCR amplification. PCR was done in the presence of two rat
DGK-specific primers or two glyceraldehyde-phosphate
dehydrogenase-specific primers as a control. PCR was carried out for 25 cycles as follows: 94 °C for 1 min, 55 °C for 1 min, and 72 °C
for 1 min, followed by a final elongation step for 10 min at 72 °C.
One-tenth of the PCR product was subsequently separated on an agarose
gel, transferred to Hybond-N nylon membrane, and hybridized with a rat
cDNA probe.
A mouse anti-DGK monoclonal antibody was made
against an Escherichia coli cell-expressed,
affinity-purified glutathione S-transferase fusion protein
of a C-terminal portion (part of the catalytic domain) of rat DGK
.
To this end, a 1.7-kb PstI-EcoRI fragment of the
rat cDNA clone was subcloned into pGEX3X vector.
Fresh frozen blocks of
brain from adult male rats were sectioned at 30-µm thickness on a
cryostat. The sections were mounted on silane-coated glass slides and
immersed in 4% paraformaldehyde and 0.1 M phosphate buffer
(pH 7.2) for 20 min, followed by acetylation in 0.25% acetic anhydride
in 0.1 M triethanolamine. The slides were prehybridized in
a fluid containing 50% formamide, 4 × SSC, 1 × Denhardt's
solution, 1% sarcosyl, 0.1 M sodium phosphate buffer (pH
7.2), 100 mM dithiothreitol, and 200 µg/ml heat-denatured salmon sperm DNA for 2 h at room temperature. Hybridization was in
the same solution containing 10% dextran sulfate and 1 × 106 cpm/slide of 35S-dATP-labeled cDNA
probe at 42 °C for 16 h in a moist chamber. The probe
corresponded to the 3-noncoding region of the rat DGK
sequence
(HindIII-EcoRI fragment, 800 bp). After
hybridization, the slides were sequentially rinsed in 2 × SSC and
0.1% sarcosyl at 45 °C for 30 min and three times in 0.1 × SSC and 0.1% sarcosyl at 45 °C for 40 min each and dehydrated in 70 and 100% ethanol. After exposure to Hyperfilm-
max (Amersham Corp.)
for 3 weeks, the sections were dipped in NTB2 emulsion (Eastman Kodak
Co.) and exposed for 2 months.
Cell lysates were made by brief sonication in a
medium containing 0.25 M sucrose, 50 mM Hepes
(pH 7.4), and protease inhibitors, followed by centrifugation (10 min
at 15,000 × g). Lysate samples of 30 µl were
incubated in an assay mixture (total of 120 µl) containing 50 mM Tris-HCl (pH 7.5), 10 mM NaF, 10 mM MgCl2, 1 mM dithiothreitol, 1 mM phosphatidylserine, 1 mM deoxycholate, and 1 mM DG or ceramide as substrate. The reaction (10 min at 30 °C) was started by the addition of [-32P]ATP (2 mM, 3 µCi) and terminated by thoroughly mixing 100 µl of assay mixture with 3.7 ml of methanol, chloroform, and 2 M NaCl (2:1:0.7, v/v/v). After the addition of 1 ml of
chloroform and 1 ml of 2 M NaCl and phase separation,
radiolabeled reaction product (PA or ceramide 1-phosphate) in the
chloroform phase was separated by TLC (Silica Gel 60 plates, Merck)
with ethyl acetate/isooctane/acetic acid/H2O (13:2:3:10,
v/v/v/v) as developing solvent (two runs), scraped off, and quantitated
by liquid scintillation counting.
COS-7 cells were transfected using the calcium phosphate precipitation method (31). SDS-polyacrylamide gel electrophoresis was performed with 8% acrylamide gels (32). Protein concentration was determined by the Bradford assay (33).
We have previously purified a
110-kDa DGK protein from rat brain cytosol (34). To clone the
corresponding cDNA, 100 µg of this protein was applied to a
preparative SDS-polyacrylamide gel and transferred to nitrocellulose
membrane. The 110-kDa region was cut out and digested with lysyl
endopeptidase, and some of the generated peptides were sequenced (see
"Experimental Procedures"). Based on the amino acid sequences,
oligonucleotide probes were designed to screen a gt10 rat brain
cDNA library. A 2-kb positive clone was isolated. Sequence analysis
showed a single open reading frame of 965 bp containing a stop codon,
but no start codon. Three of the sequenced peptides were localized, and
comparison with other DGKs revealed that this clone encoded part of the
putative catalytic domain. We used a 400-bp 5
-PstI fragment
of this 2-kb rat cDNA to screen a human fetal brain cDNA
library. Four positive clones of 1.2, 3, 5, and 6 kb were isolated
(Fig. 1). From restriction enzyme analysis, it appeared
that they represented four overlapping cDNAs. Sequencing of these
human clones revealed an open reading frame of 2646 bp encoding 882 amino acids with a sequence 90% identical to that of the rat homologue
(data not shown). Since the start codon was lacking, we continued by
screening a human retina cDNA library using a 1.7-kb
EcoRI-KpnI fragment from the partial human DGK
cDNA as a probe (Fig. 1). An additional 1.4-kb cDNA clone was
isolated that overlapped the already obtained sequence, extending to
the 5
-end and containing a putative ATG start codon. The composite
full-length cDNA and its deduced amino acid sequence are shown in
Fig. 2.
Sequence Analysis
The nucleotide sequence contains an
ATG/methionine start codon (CGGGAG) that conforms
reasonably to the Kozak consensus sequence (CCA/GCCG)
for efficient initiation of translation (35). Downstream of this
initiation codon is a single open reading frame of 2823 bp. The encoded
protein, named DGK, is a new member of the DGK family. It has a
calculated molecular mass of 101.3 kDa, close to the apparent size of
110 kDa found for the purified protein (34).
Fig. 3A schematically shows the types and
positions of conserved domains in DGK compared with a few other
representative DGKs. DGK
has a putative catalytic domain similar to
all other known DGK isotypes, but does not have the N-terminal
conserved domain and EF-hand (Ca2+-binding) motifs that are
present in the classical isozymes DGK
, -
, and -
. DGK
has a
proline- and glycine-rich region at the N terminus. It also contains a
recently identified putative Ras-binding site, the so-called RA domain
(Val397-Val488)
(27),2 which partially overlaps a PH domain
(Lys399-Arg564). Fig. 3B shows the
amino acid sequence of the PH domain of DGK
aligned with those of a
few other proteins including DGK
(22) as well as with the published
PH consensus sequence based on 71 different PH domains (36). Seven of
the eight most conserved, almost completely identical (hydrophobic)
residues (marked by asterisks in Fig. 3B) are
indeed present in the DGK
PH domain. The eighth one, a tryptophan in
block 6B, is replaced in DGK
by a similar residue,
tyrosine (Fig. 3B). The overall identity/similarity of
residues in the six blocks of the PH domain of DGK
to those in the
consensus sequence amounts to 23/46%, respectively.
A unique structural feature of DGK is the presence of three CRDs
(zinc fingers), starting at amino acids 60, 121, and 183, respectively
(Figs. 2 and 3C). All other known DGK isotypes contain only
two CRDs. Close inspection of the individual CRDs reveals that the
third CRD (CRD3) of DGK
is most homologous to the second one (CRD2)
of DGK
and DGK
(Fig. 3C) and other known DGK isotypes. The amino acid sequence of this domain, apart from its typical core
structure
(HXWX10CX2CX14CX2CX4HX2CX7C),
contains an additional 13 conserved amino acids, 11 of which (marked by
asterisks in Fig. 3C) are specifically present in
the CRD2 domains of all known DGK isotypes (15-26). Interestingly, six
of these conserved residues extend the core structure in the C-terminal
direction (Fig. 3, A and C). Klauck et
al. (23) already described that the end of CRD2 in previously
cloned DGKs is defined by GX7PP, but we note
that there is actually much more homology and DGK specificity in CRD2
(or DGK
CRD3) (Fig. 3C). CRD1 and CRD2 of DGK
, on the other hand, do not at all show such DGK specificity. Although they have
additional conserved amino acids next to the cysteine/histidine core
structure, they share most of these conserved residues with other
CRD-containing molecules, such as PKC (39) and Raf-1 (40). The homology
between DGK
CRD2 and CRD3, apart from the typical cysteine/histidine
core structure, is relatively low. CRD2 shows higher homology to
CRD1.
Another conspicuous and novel feature of DGK is the proline- and
glycine-rich region near the N terminus (Figs. 2 and 3A). It
contains 11 glycines (34%) and 9 prolines (28%) over a stretch of 32 amino acids. The first part of this region contains a
pXPXXP motif (amino acids 18-23), typical for
SH3 domain-binding sites (41), and is followed by a proline-glycine
tandem repeat (amino acids 37-45), the function of which is undefined.
The expression of
DGK in rat tissues was determined at the mRNA level by reverse
transcriptase PCR. Fig. 4 shows that DGK
is expressed
in the brain and, to a lesser extent, in the small intestine, duodenum,
and liver.
In situ hybridization histochemical analysis in adult rat
brain reveals that mRNA for DGK is expressed more or less in all the gray matter regions, but not in white matter (Fig.
5A). Expression is most intense in the
cerebellar cortex and hippocampus, while moderate expression is seen in
the olfactory bulb neuronal layers and brain stem nuclei. In the
cerebellar cortex, the hybridization signals are deposited equally in
both the Purkinje cell somata and the granule cells at the same
intensity (Fig. 5, B and C). In control
experiments in which brain sections were hybridized with the plasmid
vector of an appropriate length, no significant hybridization signals
were detected in any brain sections (data not shown).
Expression of DGK
To test whether the cloned DGK cDNA indeed
encodes a DGK, the protein was transiently expressed in COS cells by
cDNA transfection. Proper expression was confirmed by Western
blotting using a DGK
-directed monoclonal antibody (Fig.
6). The antibody detected a protein of the correct size
(110 kDa), indicating that the putative ATG start codon present in the
DGK
cDNA is indeed used as a translation start codon. The
antibody cross-reacted with overexpressed 86-kDa DGK
that was used
as a positive control. A lysate of vector-transfected cells (control)
showed no immunoreactivity (Fig. 6).
DGK activity was subsequently assayed in the respective COS cell
lysates. Table I shows that the expressed DGK is
indeed active as a DGK. Similar to DGK
, the new DGK
exhibits the
highest activity toward 1,2-dioleoyl-sn-glycerol and
1-stearoyl-2-arachidonoyl-sn-glycerol. In contrast,
ceramide, 1,3-dioleoyl-sn-glycerol, and monoacylglycerol are
relatively poor substrates.
|
With the molecular cloning of DGK, we add a new member to the
growing family of DGKs (15-26). DGK
is structurally different from
the previously characterized DGK isotypes and contains unique features,
such as three (instead of two) CRDs, a proline- and glycine-rich
region, and a recently identified RA domain (27). No other DGK isotype,
except a putative DGK in Caenorhabditis elegans (42),
contains such an RA domain (27). This implies that DGK
is a
potential new effector of one or more of the Ras-like small GTP-binding
proteins. The RA domain is located within a PH domain. DGK
(22) and
DGK
(23) also possess a PH domain, but it is differently (centrally)
located in the molecule and is without an RA domain. The finding that a
Ras-binding site coincides with another functional domain is not
unique. In Raf-1, for example, a Ras-binding site (different from the
RA domain) has been detected in its CRD (43). Analogous to this Raf-1
CRD, the PH/RA region in DGK
might play a multifunctional role,
i.e. binding to a Ras-like small G protein and to another,
PH domain-binding molecule. In general, PH domains are believed to
function in protein-protein interactions among cytoskeletal and other
signaling molecules (36). For example, they may interact with
-complexes of heterotrimeric G proteins (44) and with protein
kinase C (45). In addition, PH domains can bind polyphosphoinositides
(46-49). All these molecules, involved in signaling, are located in or
at the membrane and may thus potentially be involved in proper membrane
relocation and subsequent activation of DGK
. Enzymatic activity may
be further regulated through phosphorylation, e.g. by PKC,
as we have previously found for DGK
in vivo (50). DGK
contains six potential PKC phosphorylation sites (51) at amino acids
240, 260, 294, 311, 317, and 363 in an alanine- and glycine-rich region
between CRD3 and the PH domain. It would be interesting to investigate
whether DGK
activity in cells is regulated by phosphorylation
through interaction of its PH domain with an activated PKC, as found
for Bruton tyrosine kinase (45).
The presence of three CRDs is unique to DGK. All other known DGK
isotypes contain only two CRDs. In fact, no other protein has been
described to contain three such domains. Most PKC isotypes, except the
atypical ones, have two CRDs (39), while various other signaling
proteins such as Raf isotypes (40, 43), Vav (52), and
n-chimaerin (53) have only one CRD. While CRDs in these
proteins are generally thought to mediate protein-protein interaction
as well as to bind to acidic phospholipids in membranes (43, 54), those
in the classical and "new" PKCs are also known to bind phorbol
ester and DG (54). However, we found that DGK
and DGK
do not bind
phorbol esters (data not shown).
All CRDs typically contain the Cys6/His2 core
structure with defined spacing, which is important for binding of zinc
in a tetrahedral geometry (55). Between these core residues, however, the primary sequence (inter-Cys sequence) varies according to the type
of protein. In DGK, the inter-Cys sequence of CRD3 is homologous to
that of CRD2 in other DGKs, but quite distinct from CRD1 domains and
DGK
CRD2, which are more homologous to CRDs of PKC (39) and Raf-1
(40). Most interesting, within the CRD2/DGK
CRD3 group, the sequence
homology extends 12 amino acids C-terminally beyond the last core
cysteine. Together with additional conserved amino acids within the
inter-Cys sequence, it makes this structure very typical and specific
for DGKs and different from CRDs of PKC (39) and Raf-1 (40). While the
function of the three CRDs in DGK
or the two CRDs in DGKs in general
is unknown, we found that deletion of the CRDs in DGK
inactivates
the enzyme.3 It is therefore tempting to
speculate that the CRDs, particularly the "extended" most
C-terminal one, are somehow involved in DG binding and/or presentation
to the catalytic domain.
The proline- and glycine-rich region near the N terminus of DGK is
intriguing. It includes a pXPXXP motif, which is
characteristic of SH3 domain-binding proteins (41). We found this motif
also in DGK
(21) and DGK
(24) (likewise near the N terminus), but
not in other DGKs. Whether DGK
, -
, and -
actually bind to SH3
domains of other proteins remains to be demonstrated. A little
downstream from this motif, DGK
also contains a conspicuous proline-glycine tandem repeat. Data base screening revealed one other
protein, a DNA-binding regulatory factor, RFX5 (56), with such a motif.
The precise function of this (PG)n domain is unknown. It may
induce a loop or turn in the three-dimensional structure of DGK
,
which is important for proper folding of the molecule, or it might be
involved in protein-protein interaction.
The (putative) catalytic domain of DGK is homologous (62%) to that
in other DGKs and is contiguous, as in all DGKs except DGK
and
DGK
(22, 23), where it is separated into two subdomains (Fig.
3A). Although a protein kinase ATP-binding motif was found in the catalytic domain of several DGK isotypes, such a motif is not
present in DGK
and DGK
(22). In fact, based on mutational analysis of DGK
, we have previously argued that the consensus sequence for ATP-binding sites in protein kinases does not apply to
DGKs (57).
The substrate specificity of DGK is not much different from that of
most other DGK isotypes. The activity of DGK
is optimal toward DG
with an unsaturated fatty acid at the sn-2 position, but is
not specific for arachidonoyldiacylglycerol, as has been found for DGKs
in the testis (21, 58).
DGK shows a narrow tissue distribution, as do other DGK isotypes
except DGK
(24). Expression of DGK
mRNA is highest in the
brain, as was also found for DGK
, -
, -
, and -
, but not for
DGK
(skeletal muscle) (22) and DGK
(testis) (21). In situ hybridization revealed the highest DGK
mRNA levels in
the cerebellum and hippocampus. Compared with DGK
(DGK-III), which is also dominantly expressed in the cerebellum, most strongly in
Purkinje cells (20), DGK
shows a wider expression in gray matter and
a more equal expression in Purkinje and granule cells (Fig. 5).
Very little is known about the detailed function of DGK in signal
transduction. The substantial sequence (domain) diversity among DGK
family members, however, suggests their physiological importance and
their unique functions in distinct signaling pathways. The finding of
several distinct structural and functional domains in DGK may help
to discover in which signaling pathway(s) this isozyme operates.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) L38707[GenBank].