(Received for publication, August 19, 1996, and in revised form, November 1, 1996)
From the Section on Molecular Neurobiology of Glia, Laboratory of Cellular and Molecular Neurophysiology, NICHD, National Institutes of Health, Bethesda, Maryland 20892
We have isolated and analyzed the structure of
the gene grik5 (lutamate eceptor
onotropic ainate ), encoding the
rat kainate receptor subunit KA2. Six overlapping DNA fragments
containing the entire grik5 gene were identified in a rat
genomic library. grik5 is a unique gene composed of 20 exons that together span over 54 kilobases (kb). Reporter gene analysis
demonstrated that 2 kb of grik5 5-flanking sequence
confers tissue-specific expression on a chloramphenicol
acetyltransferase gene in vitro. We show that (i) the first
intron of grik5 (3.4 kb) inhibited transcription of the
chloramphenicol acetyltransferase gene driven by the 2-kb grik5 5
-flanking region; (ii) the negative regulatory
element was located within 500 bp of the 3
-end of intron 1, and this 500-bp fragment selectively bound nuclear proteins isolated from neural
and nonneural cells; (iii) the effect of the negative regulatory element on grik5 transcription was orientation- and
distance-independent; and (iv) a 24-nucleotide sequence
(CTTTCTGTGGCCTCTGACCTTTCC) was identified as the binding site for
nuclear proteins within the 500-bp fragment, as determined by
footprinting and gel shift assays. We conclude that an intronic element
that displays features of a silencer modulates grik5
transcription.
Glutamate is a ubiquitous excitatory neurotransmitter in the
mammalian central nervous system (1). Glutamate receptors (GluRs)1 have been classified into
N-methyl-D-aspartate,
(R,S)--amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid (AMPA), and kainate subtypes, according to agonist binding selectivity (2). Cloning of glutamate receptor subunit cDNAs has
revealed that the molecular diversity of the gene families encoding the
various receptor subunits is responsible for the pharmacological and
functional heterogeneity of glutamate receptors in the brain (3, 4).
Analysis of genomic DNA also identified RNA editing and alternative
splicing as molecular mechanisms that increase functional diversity of
GluR subunits (for review, see Refs. 3 and 4).
Two kainate-preferring GluR subunit gene families have been described, GluR 5-7 and KA1-2 (3, 4). The kainate-preferring subunit KA2 (5) is encoded by the gene grik5 (lutamate eceptor onotropic ainate ), which has been mapped to chromosomes 1, 7, and 19 in rat, mouse, and human, respectively (6). The subunit KA2 is closely related to KA1 (5, 7), displaying 68% identity in amino acid sequence. Like KA1, KA2 does not form functional homomeric ion channels but interacts with GluR5 and GluR6 to form heteromeric, agonist-sensitive channels (5). The grik5 gene is widely expressed in the adult brain and is especially abundant in the cerebral cortex, pyriform cortex, caudate-putamen, hippocampal complex, medial habenula, and granule cell layer of the cerebellum (5, 8). Little is known about the function and physiological role of the KA2 subunit, or how the grik5 gene is regulated during brain development.
Analysis of the 5-untranslated region revealed that grik5
is a TATA-less gene, whose transcription can initiate at several sites
within a 500-bp region flanking the receptor subunit coding sequence
(9). The DNA elements required for tissue-specific expression of
grik5 in vitro and in vivo reside within 2 kb of the 5
-flanking region (9). We have now isolated the entire grik5 gene, determined its exon-intron organization, and
identified a negative regulatory region within the first intron, which
inhibits reporter gene expression driven by the 5
-flanking region of
grik5. We also established that, within this negative
regulatory region, nuclear proteins bind to a 24-nucleotide sequence.
The negative regulatory element within the first intron exerts its
inhibitory effect in a distance- and orientation-independent manner and
therefore displays features of a silencer.
Isolation and Characterization of grik5 Genomic Clones
A Harlan Sprague Dawley rat genomic library in DASH
(Stratagene, La Jolla, CA) was screened for kainate receptor subunit grik5 gene sequences, using restriction fragments of the rat
KA2 cDNA as probes (5). A 0.6-kb SmaI/ClaI
restriction fragment (cDNA positions +1 to +647) was used to
identify the 5
-region. Two restriction fragments, 1.2-kb
ClaI/SmaI and 0.9-kb SmaI (cDNA positions +648 to +1837 and +1838 to +2736, respectively), derived from
middle portions of KA2 cDNA, were used to isolate recombinant phage
carrying exons for the middle region and the transmembrane regions.
Recombinant phage covering the 3
-end part of the grik5 gene
were isolated using a 1.0-kb EcoRI fragment (cDNA
position +2677 to +3662) derived from the 3
-end coding and noncoding
sequences of KA2 cDNA. All probes were labeled with
[
-32P]dCTP (DuPont NEN) using a random primer labeling
kit (Stratagene). Overlap of clones was determined by Southern blot
analysis of
DNA. Single or combinatorial BamHI,
EcoRI, and SmaI digests were blotted onto Nytran
membrane (Schleicher & Schuell, Keene, NH) and hybridized with the
probes used for the phage isolation. Restriction fragments of genomic
clones were subcloned in pBluescript SK
(Stratagene) and
sequenced using a Sequenase II kit (U. S. Biochemical Corp.). Some of
the sequencing was performed by Lark Sequencing Technologies, Inc.
(Houston, TX).
Cell Cultures and Transient Transfections
Cells of the rat glial cell line CG-4 were cultured as described previously (10, 11) and plated on poly-D-ornithine-coated plates (0.1 mg/ml). HeLa and NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (both from Life Technologies, Inc.). PC12 cells were grown in Dulbecco's modified Eagle's medium with 10% horse serum and 5% fetal bovine serum. Differentiated PC12 cells were grown on collagen-coated plates and treated with nerve growth factor (50 ng/ml) every other day for 5 days.
All expression plasmid DNAs used for transfections were purified with
the Qiagen DNA purification kit (Qiagen Inc., Chatsworth, CA). A LacZ
gene driven by the RNA polymerase II gene promoter (pPolIIplacF.gal)
was simultaneously transfected with all the grik5-CAT
constructs, to correct for possible variations in transfection efficiency. CAT constructs (10 µg) and pPolIIplacF.
gal (2 µg) were transiently transfected using LipofectAMINE reagent (Life Technologies) according to the manufacturer's directions. Cells were
transfected for 8 h in serum-free Dulbecco's modified Eagle's medium and then returned to serum-containing medium except for PC12
cells, which were transfected for 20 h in Opti-MEM I medium (Life
Technologies). All cells were harvested 38 h after transfection and assayed for CAT and
-galactosidase activities as described (12).
Levels of CAT activity were determined on a PhosphorImager (Molecular
Dynamics, Inc., Sunnyvale, CA). All transfection experiments were
repeated 3-6 times, and averages (± S.E.) of the results are
presented.
Construction of Plasmids
All restriction enzymes and DNA modifying enzymes were purchased from New England Biolabs (Beverly, MA).
CAT Constructs
2-CATThe 2-kb 5-flanking sequence of the grik5
gene was first cloned into pBluescriptSK
by inserting the
2-kb EcoRI-BamHI fragment from the 5
-end of
5
into the EcoRI-BamHI site of
pBluescriptSK
. The 2-kb fragment was then excised by
HindIII and XbaI digestion and cloned into the
HindIII/XbaI site of pCAT-basic (Promega, Madison, WI).
The 4.3-kb BamHI DNA fragment of
grik5 5-flanking sequence was isolated from the genomic
A11 clone (Fig. 1A) and subcloned into pBluescript
SK
. A clone with a 3
5
orientation was selected,
and the DNA was digested with HindIII/XbaI. The
resulting 4.3-kb fragment (HindIII site at the 5
-end and
XbaI site at 3
-end) was cloned into the pCAT-basic vector
to form the 4.3-CAT construct.
5.4-CAT
This construct contains 2 kb of 5-flanking
sequence and the 3.4-kb first intron of grik5. It was
prepared from two genomic fragments joined by multiple subcloning
steps. First, a 2.3-kb XbaI/HindIII fragment from
A11 was subcloned into LITMUS-28 vector (New England Biolabs) to
form LITMUS-2.3 kb. A 3.1-kb HindIII/MvnI
A11
fragment was subcloned into pBluescript SK
at
HindIII/SmaI sites and excised by
HindIII/SpeI digestion. This 3.1-kb
HindIII/SpeI fragment with a SpeI site
at its 3
end was then ligated with LITMUS-2.3 kb plasmid DNA at
HindIII/SpeI sites to obtain a 5.4-kb fragment
linked by HindIII resulting in the LITMUS-5.4 kb construct.
Finally, the 5.4-kb grik5 fragment was isolated by
SalI and SpeI digestion and cloned into
SalI/XbaI sites of the pCAT-basic vector to
generate the 5.4-CAT construct.
This construct containing 4.3 kb of
grik5 5-flanking sequences and the 3.4-kb first intron was
prepared similarly to the 5.4-CAT construct. A 4.6-kb
XbaI/HindIII genomic DNA fragment was subcloned
into the LITMUS-28 vector and then digested by
HindIII/SpeI and ligated with a 3.1-kb
HindIII/SpeI fragment to form LITMUS-7.7 kb
construct. This 7.7-kb fragment was isolated by
SalI/SpeI digestion and cloned into pCAT-basic
vector at SalI/XbaI sites to generate the 7.7-CAT
construct.
Fusion of grik5 First Intron with the Chick -Actin
Promoter
A 494-bp HindIII DNA fragment containing the chicken
-actin promoter region was cloned into the HindIII site
of the pCAT-basic vector to form the
-actin-CAT construct. To clone
intron 1 downstream of the
-actin promoter in the CAT vector, the
3.4-kb SpeI DNA fragment of intron 1 was cloned into the
XbaI site of
-actin CAT to generate the
-actin-I-CAT.
Fusion of grik5 First Intron with the SV40-CAT Transcriptional Unit
grik5 intron 1 was cloned downstream of the SV40-CAT
transcriptional unit (p-CAT control vector; Promega), containing SV40 promoter and enhancer sequences. The 3.4-kb SpeI DNA
fragment of grik5 intron 1 was cloned in the XbaI
site of p-CAT control, 3 of the SV40 enhancer. A clone with a 5
3
orientation of the intron was selected to generate the
pSV-CAT-intron construct.
Intron Deletion Constructs
The 5.4-CAT plasmid was used to make all intron deletion
constructs (Fig. 4). Plasmids SacI-CAT and
2-I-R-SacI-CAT were made by digesting 5.4-CAT with
SacI, religating, and selecting clones with the 2.0-kb
SacI fragment deleted (
SacI-CAT) or reversed (2-I-R-SacI-CAT). BglII/SacI deletions
were obtained by complete digestion of 5.4-CAT DNA with
BglII and partial digestion with SacI and then
blunt-ended with Klenow and ligated. The
BglII/SacI-CAT has a 1.3-kb deletion, whereas
SacI/BglII-CAT has a 0.7-kb deletion. To
construct plasmid
HindIII/SacI, the plasmid
SacI-CAT DNA was digested with HindIII and
SacI. Two of the three resulting DNA fragments
(SacI/HindIII, 4.3 kb; HindIII, 2.3 kb) were isolated, blunt-ended with Klenow, and ligated. The clone with
the correct orientation was selected. This plasmid was designed to
delete intron 1 except for the splicing sites. The
HindIII/AccI deletion was made by digestion of
AccI-CAT with AccI and EcoR V. The
AccI/EcoRV fragment was isolated, blunt-ended,
and ligated with the blunt-ended HindIII fragment. The other
deletion constructs were generated by digestion with selected
restriction enzymes, blunt-ending with Klenow, and religating the ends.
The resulting plasmids have corresponding restriction enzyme fragment
deletions (Fig. 4). All deletion junctions were analyzed by sequencing
to confirm that correct deletions were obtained.
LacZ Constructs
A promoterless LacZ vector, galscript-ES4, containing the
nuclear localization signal, was used to construct LacZ expression plasmids (Fig. 3C). 2-LacZ and 4.3-LacZ constructs were
prepared by separately cloning the 2-kb
NotI/BamHI and 4.3-kb BamHI fragments, both excised from the 5-flanking region of grik5
A11,
into a galscript-ES4 vector. 5.4-LacZ and 7.7-LacZ constructs were
obtained as follows: LITMUS-5.4 kb and LITMUS-7.7 kb plasmids were
digested with NotI and SpeI, and the 5.4- and
7.7-kb DNA fragments were isolated. The galscript-ES4 vector DNA was
digested with NotI/XbaI and ligated with 5.4- or
7.7-kb NotI/SpeI DNA fragments to form 5.4-LacZ
or 7.7-LacZ constructs, respectively.
Reverse Transcription-Polymerase Chain Reaction (RT-PCR)
Plasmids pCAT-basic, 2-CAT, 4.3-CAT, 5.4-CAT, and 7.7-CAT were used to transfect CG-4 cells as described above. Total RNA was isolated from transfected cells by RNAzol (Tel-Test, Inc., Friendswood, TX) and treated with RNase-free DNase (Ambion, Inc., Austin, TX). For each RT reaction, 2 µg of total RNA isolated as outlined above, was combined with 100 pmol of random hexamers (Life Technologies) and denatured at 80 °C for 10 min. The reactions were then assembled by adding 1 × PCR buffer, 1.0 mM MgCl2, 333 µM each dNTP, 10 mM dithiothreitol, 100 mg/ml acetylated BSA, 1 unit/ml RNasin, and 200 units of Superscript II RT in a volume of 30 µl. Reactions were incubated at 25 °C for 10 min and then at 42 °C for 50 min. After heat inactivation, 30 units of RNase H was added, and incubation continued at 37 °C for 20 min.
For PCR reactions, 10 µl of each RT reaction was added to a tube
containing 1 × PCR buffer, 50 pmol of sense primer
(5-TTGCTTTTCCACCTGTCC-3
, located at the 3
-end of the 5
noncoding
region of grik5), 50 pmol of either one of the two antisense
primers (located in the SV40 small T antigen coding region; Fig. 3) AS1
(5
-CAGCTTTTTCCTTTGTGGTG-3
) or AS2 (5
-ATGTTTCAGGTTCAGGGGG-3
), 200 µM each dNTP, and 2.5 units of AmpliTaq in a volume of
100 µl. The PCR cycling profile was as follows: 95 °C for 5 min
and then 30 cycles of 94 °C for 1 min, 56 °C for 1 min, and
72 °C for 1 min with a 5-s time extension added per cycle and a
final extension at 72 °C for 7 min. Products were separated on 1%
agarose gels and visualized by ethidium bromide staining.
DNA Mobility Shift Assays
The 1.3-kb BglII/SacI DNA fragment from
grik5 intron 1 was subcloned into LITMUS-28 vector. The
resulting plasmid was grown in a methylase-deficient E. coli
strain GM2163 (dam-, dcm-; New England Biolabs). The 1.3-kb
BglII/SacI DNA fragment was purified and digested
with Sau3AI. The resulting restriction fragments of the
Sau3AI digestion are shown in Fig. 6A. Fragments
were purified and end-labeled with [-32P]dCTP using
Klenow fragment. Nuclear protein extracts from CG-4 and HeLa cells were
prepared according to Dignam et al. (13). The labeled probes
were used in gel shift assays with or without nuclear protein extracts.
The reactions were carried out in a total volume of 20 µl of a
binding buffer containing 25 mM HEPES, pH 7.5, 60 mM KCl, 10% glycerol, 5 mM MgCl2,
1 mM dithiothreitol, 1 mM EDTA, 50 µg/ml
poly(dI-dC) (Pharmacia Biotech Inc.). The binding reactions were
incubated on ice for 60 min and directly loaded onto a 4%
nondenaturing polyacrylamide gel in 0.5 × TBE buffer (1 × TBE: 90 mM Tris borate, pH 8.3, 1 mM EDTA) and
electrophoresed at 10 V/cm for 3-4 h at 4 °C. The gels were dried
and autoradiographed on x-ray film (Eastman Kodak Co.).
Oligonucleotides were synthesized on an Applied Biosystems DNA
synthesizer (Perkin-Elmer). Oligonucleotides ON1
(5-CGGCGTCTTTACTTTTGCGG-3
), ON2 (5
-GTCAGATACTTTTTCTTTTCCTTCCGTT-3
),
and ON3 (5
-CCTGTCCTTTCTGTGGCCTCTGACCTTTCCTTGTTCCAGA-3
) are
derived from different regions of intron 1, corresponding to nucleotide
positions 2779-2798, 2805-2832, and 3202-3241, respectively (Fig.
8A). The complementary oligonucleotide pairs were annealed and labeled using [
-32P]ATP (DuPont NEN). The
double-stranded oligonucleotides were then gel-purified and used for
gel shift assays, as described above.
DNase I Footprinting
Four pairs of primers were made to PCR-amplify four overlapping
DNA fragments (153-220-bp size range) spanning the entire Sau3AI/SacI region of intron 1, corresponding to
nucleotide positions 2660-2879, 2778-2963, 2938-3090, and
3040-3244, respectively (Fig. 7A). The PCR products were
cloned into pCRTM 2.1 vector using the TA cloning kit
(Invitrogen, San Diego, CA), excised by cutting with
HindIII/XbaI (these sites flank the
BamHI-EcoRV insertion sites), and treated with
CIP. The fragments were then labeled at both ends using
[-32P]ATP (DuPont NEN) and T4 polynucleotide kinase
(New England Biolabs). A singly end-labeled probe was then generated by
digesting the label off one end with EcoRV (upper strand
labeled), or with BamHI (lower strand labeled). The labeled
probes were gel-purified. For DNase I footprinting, nuclear extracts
(5-20 µg) were incubated on ice for 15 min in 50 µl, containing 25 mM HEPES, 60 mM KCl, 5 mM
MgCl2, 1 mM dithiothreitol, 1 mM
EDTA, 10% glycerol, 40 µg/ml poly(dI-dC), pH 7.5. The singly
32P-end-labeled probes were then added, and the samples
were incubated on ice for 30 min. Fifty microliters of
Ca2+/Mg2+ solution (5 mM
CaCl2, 10 mM MgCl2, at room
temperature) were added to the reactions. Diluted RQ1 RNase-free DNase
(Promega) was added to the samples at 0.25 units/reaction and incubated at room temperature for 1 min. The reactions were terminated by adding
90 µl of prewarmed (37 °C) stop solution (200 mM NaCl, 30 mM EDTA, 1% SDS, 100 µg/ml yeast tRNA) and extracted
with phenol/chloroform/isoamyl alcohol. The reaction products were
precipitated and analyzed on 6% sequencing gels. Dideoxy sequencing
reactions were performed with a specific upper strand primer
(initiating at the HindIII site) and a specific lower strand
primer (initiating at the XbaI site) to serve as size
standards.
Localization of the nuclear factor binding
site by DNase I footprinting. DNase I footprinting reactions were
performed by using nuclear protein fractions prepared from CG-4 and
HeLa cells. A, unidirectionally labeled probes were
generated for the upper and lower strand by 5-end labeling of four DNA
overlapping fragments (size range: 153-220 nucleotides) obtained by
PCR amplification of the Sau3AI/SacI region of
intron 1. B, DNase I footprinting of the 205-bp PCR product.
The footprinting reactions of upper and lower strands were performed
with (lanes 2-4, 6-8,
11-13, and 15-17) or
without (lanes 5 and14) nuclear proteins (see "Experimental Procedures"), and products were separated on sequencing gels. Dideoxy
sequencing reactions were performed using specific upper strand
(left panel) and lower strand (right panel)
primers. Protected regions are marked, and the corresponding sequence
positions are indicated. C, nucleotide sequence of the
24-nucleotide protected region (nucleotides 3207-3231), as determined
by DNase I footprinting.
Four
different rat KA2 cDNA restriction fragments were used as probes to
screen a rat DASH genomic library (see "Experimental Procedures"). Six overlapping recombinant
clones were isolated, which contained the entire grik5 gene (Fig.
1A). The restriction map of each DNA insert
was determined by electrophoretic analysis of restriction fragments
(Fig. 1A). To determine the exon-intron organization, the
size, and exact location of the grik5 gene introns, Southern
blot analysis was performed on the clones to identify restriction
fragments containing coding sequence. These were subcloned into
pBluescript SK
and sequenced. Restriction mapping and
sequence analysis indicated that grik5 is distributed over
54 kb of genomic sequence. As shown in Fig. 1B, the gene is
composed of 19 translated exons and one untranslated exon, ranging in
size from 54 bp (exon 7) to 944 bp (exon 20). The size of introns
varies between 0.1 and 12 kb. All of the exon-intron boundaries were
determined by DNA sequencing and conform to consensus splice
donor/splice acceptor sequences (Fig. 1C; Ref. 14). Exon 20 (944 bp) of grik5 contains 423 bp of coding and 521 bp of
3
-untranslated region (a TGA translational stop codon is present 424 nucleotides downstream of the intron-exon boundary). The total length
of the grik5 coding region (3.78 kb) deduced from the
genomic sequence is consistent with the 4-kb mRNA identified by
Northern blot analysis of tissue and cultured neural cell
poly(A+) RNAs (Ref. 9; data not shown). Based on the
previously cloned rat cDNA sequence (5), polyadenylation of
grik5 mRNA occurs 23 nucleotides downstream of a
canonical signal sequence (AATAAA) in exon 20 (nucleotides 917-922
from the 5
-end of exon 20).
grik5 is a TATA-less gene, whose
transcription can initiate at several sites within 500 bp upstream of
the exon1/intron 1 boundary (9). It was previously found that 2 kb of
grik5 5-flanking region confers tissue-specific expression
on the CAT reporter gene both in vitro and in
vivo (9). grik5-CAT reporter gene constructs were
tested by transient transfection assays in neural (CG-4 and PC12) and
nonneural (HeLa and NIH3T3) cells. Either 2 or 4.3 kb of
grik5 5
-flanking region were able to selectively direct CAT
gene expression in neural cells (Fig. 2,
B-D). In nonneural cells, CAT activity of both
grik5 gene constructs was below that of promoterless
pCAT-basic. As shown in Fig. 1B, 3.4-kb intron 1 interrupts
the grik5 5
-untranslated region. To examine if the presence
of intron 1 altered the expression of the CAT gene driven by the
5
-flanking region of grik5, we tested reporter gene
constructs that contained the two different segments of
grik5 5
-flanking region, either with the intron (5.4-CAT
and 7.7-CAT) or without it (2-CAT and 4.3-CAT) (Fig. 2A).
The results showed that in CG-4 and PC12 cells the activities of the
constructs containing intron 1 (5.4-CAT and 7.7-CAT) were reduced
approximately 3-fold when compared with the activities of the
constructs without the intron (2-CAT and 4.3-CAT) (Fig. 2, B
and D). Thus, the presence of intron 1 inhibited expression
of the CAT reporter gene driven by the grik5 promoter.
To exclude the possibility that grik5 promoter activity was
diminished by intron 1 as a result of aberrant splicing (15), two sets
of experiments were performed. First, RNA was isolated from CG-4 cells
transfected with different grik5-CAT constructs. RT-PCR was
carried out with two different sets of specific primers corresponding
to sequences in the grik5 2-kb 5-flanking region (sense, S
primer) and in the SV40 small T antigen sequence (antisense, AS
primers; Fig. 3A). If aberrant splicing
occurred, a different size of RT-PCR products would be observed in
cells transfected with grik5 constructs containing intron 1 than with constructs that did not contain it. On the other hand, if
intron 1 were spliced out of transcripts properly, the length of RT-PCR
products would be independent of its presence in the DNA constructs.
Fig. 3B shows that single bands with the expected sizes of
1.4 and 1.7 kb were produced with primers S and AS1 and primers S and
AS2, respectively. Identical results were obtained from RNAs of cells transfected by grik5 constructs with or without intron 1 (compare 5.4-CAT with 2-CAT and 7.7-CAT with 4.3-CAT), indicating that aberrant splicing likely did not occur. This conclusion is further supported by transient transfection experiments performed with the
5.4-CAT
HindIII/SacI construct (Fig.
4A), which only contained the 5
splice donor
and 3
splice acceptor sequences of intron 1 and displayed reporter
gene activity identical to 2-CAT (Fig. 4B).
In a second set of experiments, the same DNA fragments from
grik5 were subcloned into a different promoterless vector,
galscript-ES4, to form LacZ reporter gene constructs (see
"Experimental Procedures"). The results in Fig. 3C
indicate that intron 1 inhibited the -galactosidase expression by a
factor of 3 when 2-LacZ and 4.3-LacZ were compared with 5.4-LacZ and
7.7-LacZ, respectively. Since intron 1 conferred inhibition independent
of the sequence of the transcriptional unit (i.e. CAT
versus LacZ), we conclude that aberrant splicing does not
contribute to the inhibition of reporter gene expression observed in
grik5 constructs containing intron 1.
If the inhibitory effect of intron 1 on grik5
reporter expression was due to the presence of a negative
cis-acting element, elimination of this element should
abolish the inhibitory effect. To explore this possibility, a series of
intron 1 deletion constructs were made in order to define the essential
negative regulatory region. All constructs were generated by internal
deletions of intron 1 in the 5.4-CAT construct (Fig. 4A).
Both the 5 splice donor and 3
splice acceptor sequences were
maintained in all constructs (Fig. 4A), which were tested by
transient transfection assays in CG-4 cells.
The activity of 2-CAT was defined as 100%. Deletion of 2,936 nucleotides within intron 1 (construct
HindIII/SacI) completely eliminated the
inhibition on reporter gene activity (Fig. 4B). Furthermore,
when a 2-kb SacI fragment was deleted from intron 1 in the
5.4-CAT construct (
SacI), reporter gene activity was completely recovered (97% of 2-CAT level; Fig. 4B). This
result indicated that the inhibitory region resides within the
SacI fragment. The position of the inhibitory region was
further defined by deletion mapping of the SacI fragment
(Fig. 4A). The activities of constructs containing deletions
in the 5
portion of intron 1 (
HindIII/BglII,
SacI/BglII, and
KpnI/BglII) were still significantly
inhibited. On the other hand, a deletion in the 3
-end region of intron
1 (
BglII/SacI) relieved the inhibition (83%
of 2-CAT). Finally, deletion of the smaller
BglII/AccI or AccI fragments did not
modify the inhibitory activity of intron 1. Therefore, we concluded
that one or more negative regulatory elements are located at the 3
-end of intron 1, within the AccI/SacI fragment
(nucleotides 2625-3245 of intron 1). To further support this
conclusion, in a construct that only contained the
AccI/SacI region of intron 1 (
HindIII/AccI; Fig. 4A), CAT
activity was reduced to the same extent as if the the entire intron
were present (Fig. 4B).
Deletion analysis indicated that the
negative regulatory region of intron 1 is located within the
SacI fragment (Fig. 4). To test if the inhibitory effect of
intron 1 was orientation-dependent, we made constructs with
the SacI fragment in a reverse (3
5
) orientation (Fig.
5A). Results obtained from transiently
transfected CG-4 cells showed that there was no significant difference
in CAT activities between the constructs containing the SacI
fragment cloned in a reverse (2-I-R-SacI-CAT) or forward
orientation (2-I-F-CAT; Fig. 5A). These results indicate
that the inhibitory effect of intron 1 is independent of its
orientation relative to the site of transcription initiation.
To test if the inhibitory effect of intron 1 was position-dependent, we also made constructs in which intron 1 was cloned upstream of the grik5 promoter. Results indicated that when intron 1 was placed upstream of the grik5 promoter (I-F-2-CAT) its inhibitory effect was weaker but still significant (Fig. 5A). A similar construct containing the SacI fragment cloned in a reverse orientation (I-R-SacI-2-CAT) also reduced promoter activity (Fig. 5A). These results indicate that the negative regulatory element in intron 1 is orientation- and distance-independent and acts in a relatively position-independent manner.
Since intron 1 of the grik5 gene inhibited expression of its
own promoter, we sought to examine whether the negative regulatory sequences could also regulate expression of heterologous promoters. As
shown in Fig. 5B, intron 1 cloned downstream of the chicken -actin promoter in a 5
3
orientation (actin-I-CAT) strongly inhibited
-actin expression in transfected CG-4, HeLa, and NIH3T3 cells. We also analyzed whether intron 1 was able to limit SV40 promoter activity in the presence of SV40 enhancer, which confers high
reporter gene expression in many types of eukaryotic cells. CG-4 cells
were transfected with a plasmid containing grik5 intron 1 placed 3
of the SV40 promoter-CAT-SV40 enhancer construct
(pCAT-control). The resulting CAT activity of the SV40 construct
containing grik5 intron 1 was 55 ± 10% of
pCAT-control (three separate transfection experiments; data not shown).
In conclusion, grik5 intron 1 significantly inhibited
heterologous promoters.
The transcriptional analysis of intron 1 in CG-4 cells (Fig. 4) indicated that deletion of the 500-bp AccI/SacI fragment in intron 1 resulted in loss of inhibitory activity, whereas its presence was sufficient to inhibit grik5 expression to a similar extent of the entire intron. To begin to identify the factors responsible for this inhibition, we first asked whether protein binding sites were present within this 500-bp DNA fragment.
Gel shift assays were then performed with individual
Sau3AI-digested DNA fragments (Fig.
6A). As shown in Fig. 6B, the DNA fragments A and B were not shifted by CG-4 nuclear proteins, but both
fragments C and D were complexed and up-shifted. Because fragment C
comprised both B and D and fragment B was not shifted, it can be
concluded that fragment D contained nuclear protein binding sites. This
500-bp fragment formed complexes and was shifted with nuclear proteins
from both CG4 and HeLa cells (Fig. 6C), and the shift was
competed by unlabeled specific DNA fragment D (Fig. 6C). In
conclusion, the results from the band shift assays were consistent with
the intron deletion analysis (Fig. 4) and indicated that nuclear
factors specifically bound to the 500-bp region at the 3-end of intron
1 might inhibit grik5 promoter activity.
The nuclear factor binding site in intron 1 was defined by
DNase I footprinting experiments. The deletion analysis (Fig. 4) and
the gel shift experiments (Fig. 6) indicated that the binding site was
contained within the Sau3AI/SacI region of intron
1; therefore, four overlapping PCR-amplified DNA fragments spanning the
whole Sau3AI/SacI region were obtained (Fig.
7A) and subcloned into pCRTM 2.1 vector. Unilaterally labeled fragments were generated by 5-end
labeling of the upper or lower strand and analyzed by DNase I
footprinting (Fig. 7B). Footprinting reactions were
performed with increasing amounts of CG-4 and HeLa cell nuclear
extracts, and products were separated on sequencing gels. One protected region of 24 or 25 nucleotides could be recognized in both the coding
and noncoding strands of the 3040-3244 fragment, with both CG-4 and
HeLa nuclear extracts (Fig. 7B). This region corresponds to
nucleotide positions 3206-3231, as shown in Fig. 7C. No
protected sequences could be found in any of the other three fragments
(2660-2879, 2778-2963, and 2938-3090; data not shown).
An oligonucleotide probe, ON3
(5-CCTGTCCTTTCTGTGGCCTCTGACCTTTCCTTGTTCCAGA-3
; Fig.
8A), corresponding to the protected region shown in Fig. 7, was tested for its ability to bind CG-4 and HeLa cell
nuclear factors by gel retardation assays (Fig. 8). Binding to ON3
resulted in the formation of an easily detectable retarded complex
(Fig. 8B). Identical results were obtained by using an ON3-related 27-bp oligonucleotide (5
-TCCTTTCTGTGGCCTCTGACCTTTCCT-3
). Conversely, the oligonucleotides ON1 and ON2, corresponding to sequences contained within the 2778-2963 fragment, did not form any
detectable complex with cell nuclear factors (Fig. 8B). This finding is consistent with DNase I footprinting experiments that showed
no protected sequences within the 2778-2963 DNA fragment (data not
shown). The specificity of the retarded complex formed by ON3 and
nuclear proteins was also tested by competition experiments, using a
molar excess of either unlabeled ON3 or an unrelated competitor (Fig.
8, B and C). The complex was specifically
competed by the addition of a molar excess of ON3 but not by the
unrelated competitor (Fig. 8, B and C). This
complex is therefore formed by the specific interaction of the DNA
sequence CTTTCTGTGGCCTCTGACCTTTCC with a nuclear factor(s). Similar
results were also obtained with nuclear extracts isolated from HeLa
cells (Fig. 8C).
grik5 is the first kainate receptor subunit gene for which the structure, with respect to exon-intron organization, has been determined. grik5 contains more exons than the rat gluR-B and mouse nr2C (17 and 15 exons, respectively; Refs. 17 and 18), but fewer than the rat nr1 gene (22 exons; Ref. 19). The exon-intron organization of grik5 is different from that of gluR-B, nr1, and nr2C. In particular, the grik5 intron positions are not conserved, compared with the other GluR subunit genes cloned so far, except for the three introns that separate the TMIII and TMIV containing exons (17-19). A unique feature of the rat grik5 gene is that the TMIII coding sequence is split by an intron between exons 15 and 16. The TMI and TMII regions are encoded in two separate exons in the kainate receptor subunit genes rat grik5 and mouse gluR-5 and -6 but are both encoded in a single exon in the AMPA receptor genes gluR-A to -D (20). In some genes, introns separate functional or structural domains of the proteins encoded by exons, suggesting that introns may accelerate evolution of proteins with different functional properties (21). Therefore, differences in gene intron/exon structure may reflect the evolutionary origin as well as the distinct functional properties of different GluR classes. The cloning and characterization of other kainate receptor subunit genes, in particular KA1, will confirm a close evolutionary origin of the corresponding subunits.
Previous studies have demonstrated that kainate and AMPA receptor subunit genes can exist in alternative spliced isoforms (for reviews, see Refs. 3 and 4). Additionally, differences were found between genomic and mRNA nucleotide sequences of GluR2-4, -5, and -6 in codons contained within the TMI and TMII regions as well as in an extracellular loop regulating receptor desensitization (22, 23). These differences in one nucleotide are due to RNA editing and significantly alter the functional properties of the corresponding GluR channels (22, 23). Our analysis of the nucleotide sequence of grik5 exons offers no evidence for alternative splicing or RNA editing, when compared with the rat and mouse KA2 cDNAs (5, 24). A rigorous proof will require more thorough analysis of RNA species expressed in different cell types at distinct developmental stages. Our findings with the grik5 gene, however, raise the possibility that in view of its physiological relevance (23) RNA editing is limited only to GluR genes that encode subunits capable of forming functional homomeric channels.
We have previously demonstrated that DNA elements residing within 2 kb
upstream of the 5-end of intron 1 are necessary for tissue-specific
expression of grik5 in the brain (9). This was determined by
reporter gene analysis in transgenic mice as well as in the glial cell
line CG-4. We now have used CG-4 and PC12 cells (the latter as a
neuronal model) and found that inclusion of an additional 2.3 kb of
sequence further upstream of the previously characterized 5
-flanking
region did not modify reporter gene expression in neural and nonneural
cells (Fig. 2). These results indicate that, at least in
vitro, this distal sequence is not involved in regulating
grik5 transcription and tissue-specific expression.
Transcriptional regulatory elements of eukaryotic genes are found not
only in their 5
upstream region, but also within their introns
(25-34). These sequences can act as transcriptional enhancers (27, 28,
30, 33) or can be required for appropriate tissue- or cell-specific
expression (25, 26, 34). Intragenic negative regulatory elements were
initially described in yeast as cis-acting elements that
repress transcription, were able to function in either orientation, and
were relatively independent of position (35). Similar DNA elements have
also been characterized in other eukaryotic genes (25, 26, 31, 36-38).
In the present study, we provide evidence that the first intron of the
neural gene grik5 negatively regulates transcription without
altering the tissue-specific expression pattern of the gene in
vitro. In transiently transfected CG-4 and PC12 neural cells,
reporter gene activity was significantly reduced in CAT constructs
containing intron 1. Deletion analysis of intron 1 (Fig. 4), combined
with gel retardation assays (Fig. 6), indicated that the DNA element responsible for inhibiting grik5 transcription resided
within a 500-bp Sau3AI fragment.
The inhibitory effect of intron 1 on grik5 transcription was
orientation-independent, as shown by experiments in which the SacI intronic fragment was cloned in a reverse orientation
in the intron 1-containing grik5-CAT constructs (Fig. 5).
Also, intron 1 exerted its inhibitory effect in a distance-independent
fashion, as demonstrated by the significant inhibition of reporter gene expression found with the HindIII/BglII and
SacI/BglII constructs (Fig. 4). Finally, the
effect of intron 1 was relatively position-independent, because partial
inhibition was observed when the intron was placed upstream of the
grik5 promoter. In view of its orientation independence and
distance independence from the transcription initiation site, the
negative regulatory element of intron 1 displays the features of a
transcriptional silencer (29). The presence of an intronic silencer is
not unprecedented for neural genes (25, 26), but to our knowledge
grik5 represents the first case of a neurotransmitter receptor gene.
In agreement with the deletion analysis experiments in transiently
transfected cells, DNase I footprinting and gel shift assays demonstrated that nuclear factors bind to a 24-nucleotide sequence (5-CTTTCTGTGGCCTCTGACCTTTCC-3
) within the region of intron 1 (500-bp
Sau3AI fragment) responsible for the inhibition of
grik5 transcription. Comparison of this 24-nucleotide region
with published sequences in data bases revealed no obvious homologies
to known motifs. Mutation experiments are in progress to determine the sequence of the minimal intronic element involved in the binding to
nuclear proteins and in the inhibitory effect on grik5
transcription.
The identity of the nuclear factors binding to the silencing element of
intron 1 remains to be determined, but it appears that their expression
occurs in cells of neural and nonneural origin. Similar findings were
reported with nuclear proteins isolated from different cell types that
bind to the first intron of the human 1(I) collagen gene (31), or to
the immunoglobulin heavy chain enhancer (39). Our experiments in HeLa
and 3T3 cells transiently transfected with grik5 intron 1 linked to a heterologous
-actin or SV40 promoter are consistent with
the idea that the nuclear factors expressed in nonneural cells are
capable of binding to one or more intragenic regulatory elements and
may thereby inhibit gene transcription.
The effect of intron 1 on the production of reporter gene mRNA may occur through any of several mechanisms. The intron might contain sequences that block transcriptional elongation, target the primary transcript for rapid turnover, or prevent efficient processing. Alternatively, the silencer in this intron may interact with promoter and upstream elements through DNA-binding proteins. This rather distant interaction could occur through the formation of DNA loops (40). It can therefore be hypothesized that protein binding to the intronic silencing element may directly interfere with formation of the transcriptional complex and/or with the binding of trans-acting factors to the promoter region. Based on this hypothesis, it could be speculated that the inhibitory potential of the intronic region would depend on the levels of relevant DNA binding proteins in the cell. These could be different in distinct neural cell types and therefore determine the cell-specific expression pattern of the GluR gene grik5 that occurs in several brain regions (8). Alternatively, higher levels of intron-binding proteins might contribute to maintain the grik5 RNA levels at a plateau in the postnatal rat brain (41). To address these questions more directly, we are currently investigating the effects of intron 1 on grik5 expression in transgenic mice during development.
In conclusion, our results indicate that expression of the kainate receptor subunit gene grik5 is modulated by an intronic element that displays features typical of a silencer. These findings represent the first demonstration that, similar to the other neural genes encoding the glial fibrillary acidic protein and the cell adhesion molecule Ng-CAM (25, 26), promoter activity of a GluR gene can be modulated by intragenic DNA elements. This will provide further impetus to identify negative regulatory elements in introns of other neurotransmitter receptor genes.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U81010[GenBank].
We thank Montse Molne for help during the
first phase of this project and for generating the 5 clone and the
grik5-2.CAT construct. We are grateful to Xiaoqing Yuan and
Jiamin Zhou for assistance. We thank Arther Sands for the gift of
plasmid
-actin-CAT, Kathleen Mahon and Maria Morasso for the
galscript-ES4 plasmid, and Grant Mac Gregor for the pPolIIplacF.
gal
plasmid. We thank Michael Whalin and Gordon Guroff for PC12 cells. We
are particularly indebted to Gretchen Gibney for discussion and for
critically reading the manuscript. We thank Steve Scherer and Albert
Dobi for technical advice and Sharmila Banjeree Basu, Michael Brenner, and Steve Scherer for critically reading the manuscript.