Target specificity of neuronal RNA-binding protein, Mel-N1: direct binding to the 3
' untranslated region of its own mRNA
Target specificity of neuronal RNA-binding protein, Mel-N1: direct binding to the 3 ' untranslated region of its own mRNA
Ryoichi
Abe
,
Koichi
Yamamoto
and
Hiroshi
Sakamoto*
Department of Biology, Faculty of Science, Kobe University, 1-1 Rokkodaicho,
Nadaku
, Kobe 657,
Japan
Received March 11, 1996;
Revised and Accepted April 22, 1996
EMBL accession no. U29088
ABSTRACT
We have identified cDNAs encoding Mel-N1, the mouse homologue of a human nervous system-specific RNA-binding protein, Hel-N1. Two major mRNA transcripts of Mel-N1 were detected predominantly in the adult mouse
brain by Northern blot analysis. To gain insight into the RNA binding
specificity of Mel-N1, we performed iterative
in vitro
RNA selection. The resulting
in vitro
selected RNAs were found to contain AU-rich sequences as well as a GAAA motif in the majority of clones. By means
of
in vitro
binding assays we demonstrate that this GAAA sequence appears to significantly
affect the Mel-N1 RNA-binding efficiency. Our studies further reveal that Mel-N1 can bind to its own 3
'
untranslated region (3
'
UTR) as well as to the
c-fos
3
'
UTR, and is localized predominantly in the cytoplasmic region in cells,
suggesting that post-transcriptional autoregulation of Mel-N1 gene expression occurs
in vivo
.
INTRODUCTION
In eukaryotes, RNA-binding proteins have been found to play important roles in various
aspects of post-transcription regulation of gene expression (
1
,
2
). One well-characterized group of RNA-binding proteins that have been shown to be involved in many RNA
processing events, such as pre-mRNA splicing and polyadenylation, are those that contain an RNA
recognition motif (RRM) (
3
-
12
). The RRM consists of ~90 amino acids, is structurally characterized by the presence of two [alpha] helices and four [beta] strands (
9
,
13
,
14
), and functions as an RNA-binding domain (RBD) (
10
,
15
-
19
).
Recently, a group of nervous system-specific RNA binding proteins with RRM-type RBDs have been identified. Elav is a neuron-specific protein, of which deficiency causes abnormal neural development and leads to embryonic lethality in
Drosophila
melanogaster
(
20
-
22
). It is also required for maintenance of the nervous system in adult flies (
23
,
24
). Rbp9 is another fly Elav-like protein which is also restricted to the nervous system (
25
). HuC and HuD, two human proteins which were initially identified as antigens
recognized by autoimmune Hu antibodies of patients with paraneoplastic
neurologic disorders (
26
,
27
) are related to Hel-N1, a human Elav homologue which can also be recognized by Hu antibodies (
28
). These proteins share extensive similarity to Elav and are believed to be
members of the mammalian Elav homologue family.
The restricted expression of the Hu proteins and their rodent counterparts in
the nervous system, suggest that like Elav, they may play important roles in
neuronal cell regulation (
29
-
31
). Hel-N1 and HuD have been shown to bind to specific sequences within the 3' untranslated regions (3'UTRs) of mRNAs which encode cell proliferation regulatory
elements (
28
,
30
,
32
). The 3'UTRs of these mRNAs contain an AU-rich element (ARE) which is characterized by the presence of an
AUUUA pentamer and are generally AU-rich. Since AREs have been demonstrated to influence mRNA stability (
33
-
38
), Hu proteins may regulate the expression of particular mRNAs by altering their
stability, thereby contributing to neuronal differentiation and/or maintenance.
In this study, we report the cloning and sequence determination of a cDNA
encoding a mouse Hu protein, Mel-N1, which is homologous to human Hel-N1. Using a Mel-N1 fusion protein, we performed iterative
in vitro
RNA selection analysis and show that Mel-N1 binds to RNAs containing ARE-like motifs, and that this binding may be further modulated by
specific sequences downstream of the ARE-like motif. Moreover, we propose that autoregulated expression of Mel-N1 may occur via binding to the 3'UTR of its own mRNA.
MATERIALS AND METHODS
Screening and sequencing
A newborn mouse brain cDNA library (Stratagene) was screened according to
previously described methods (
39
). Positive recombinant [lambda] ZAPII phages were subjected to plasmid rescue using helper phage and
plasmid DNA was obtained by conventional methods (
39
). Sequencing was performed using Sequenase version 2.0 kit (United States
Biochemicals).
Northern blot analysis
A mouse MTN Blot (Clontech) was used for Northern blot analysis. The RNA from
each tissue was checked both qualitatively and quantitatively with a human
glyceraldehyde 3-phosphate dehydrogenase (G3PDH) cDNA (Clontech) control probe.
Hybridization was performed as described under high stringency conditions (
39
). A Fuji BAS-2000 imaging analyzer and standard autoradiography were used to analyze
the hybridization patterns.
a
The first and last four nucleotides in the sequences are derived from the
primers used in PCR amplification step. ARE-like sequences are bold-faced. The GAAA sequence observed with 18 of 20 clones are
underlined. All clones contained <25 nt in the randomized region, possibly due to unknown error in the PCR
amplification step.
Construction of Mel-N1 expression plasmids
DNA containing the Mel-N1 coding region was PCR-amplified using the following synthetic primers: 5'-TCG GAT CCA TGG AAA CAC AAC TGT C-3' and 5'-CCG AAT TCG AGC TCA TTA GGC
TTT G-3'. The amplified fragment was cloned into the
Eco
RI and
Bam
HI sites of a bacterial expression vector, pGEX-2T (Pharmacia). The resultant plasmid, pGEX-Mel-N1, was transformed into
Escherichia
coli
XL1-blue, GST-Mel-N1 fusion protein was induced with 1 mM IPTG for 4 h and
affinity-purified by glutathione-Sepharose. pEFT7-Mel-N1 was made by inserting the entire Mel-N1 coding region into the
Bam
HI site of pEF-BOS-T7, a derivative of pEF-BOS (
40
,
41
) kindly provided by M. Ohno of Kyoto University.
In vitro
selection and RNA binding assay
In vitro
selection using GST-Mel-N1 was performed as described previously (
17
) with the following modifications. During the last two rounds of selection, the
KCl concentration was raised to 350 mM in the binding and washing buffers, and
the final round of washing buffer contained 0.5 M urea.
In vitro
RNA synthesis was performed as described previously (
17
). RNA binding reaction mixture [containing ~0.3 [mu]g GST-Mel-N1, labeled RNA (2 * 10
4
c.p.m.), 1 [mu]g yeast RNA and 55 U RNase inhibitor in the RNA binding buffer] was
incubated at 20oC for 20 min followed by UV light irradiation and RNase digestion and then
analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) as described previously (
46
). The efficiency of label transfer to GST-Mel-N1 with each selected RNA was calculated by densitometry using a
Fuji BAS 2000 Image Analyzer. The GAAA sequence of Mel-N1-selected clone 27 was changed to CCCC sequence by PCR-based mutagenesis. Sequence confirmation of the resulting
clone (mu27) however, revealed an additional single nucleotide deletion just
upstream of the target sequence. FosA and FosB RNAs were synthesized using the
following DNAs. FosA1 (sense): 5'-AAT TCG TTT TTA TTG TGT TTT CAA TTT ATT TAT TAA G-3', FosA2 (anti-sense): 5'-GAT CCT TAA TAA ATA AAT TGA AAA CAC
AAT AAA AAC G-3', FosB1 (sense): 5'-AAT TCA TAT TTA TAT TTT TAT TTT ATT TTT TTC TAG-3', FosB2 (anti-sense): 5'-GAT CCT AGA AAA AAA
TAA AAT AAA AAT ATA AAT ATG-3'. Two pairs of sense and anti-sense DNAs (FosA1-FosA2 and FosB1-FosB2) were annealed and introduced into
Eco
RI and
Bam
HI sites of pBluescriptII SK
+
. The resultant plasmids were digested with
Xba
I and used as templates for
in vitro
transcription with T7 RNA polymerase.
Transfections and immunofluorescence
HeLa and COS7 cells were plated 24 h before transfection at ~30% confluence on circular 15 mm coverglasses. Transfection was performed
using Lipofectoamine (Gibco BRL). Twelve hours after transfection, cells were
fixed, incubated with anti-T7 tag mouse monoclonal antibody (Novagen) and FITC-labelled anti-mouse immunoglobulin secondary antibody (DACO, F479), and
examined under a Zeiss confocal laser microscope. Anti-T7 tag and anti-mouse immunoglobulin antibodies were used at dilutions of 1:1000 and
1:200 respectively. Cells were double-stained with DAPI to localize nuclei.
RESULTS
Cloning and tissue-specific expression of Mel-N1
During cDNA screening for mouse homologues of HuD from a brain cDNA library, we
obtained a fragment which showed homology to the third RRM of Hel-N1 (
31
). Using that cDNA fragment as a probe, we screened another mouse brain cDNA
library for longer clones. From ~5 * 10
5
recombinants, two positive clones were isolated and sequenced. The longest
insert had an open reading frame encoding a 360 amino-acid protein containing three RBDs (Fig.
1
). The deduced protein showed 99% identity to the human Hel-N1 and was designated Mel-N1. Interestingly, the Mel-N1 cDNA contains a long AU-rich 3'UTR of ~1 kb. The AU content (66.2%) of the 3'UTR is significantly higher than that
of the 5'UTR (26.6%) and the protein coding region (53.4%).
In vitro
selection of RNAs with affinity for Mel-N1
To determine the RNA binding ability of Mel-N1, we constructed a GST fusion protein to take advantage of iterative
in vitro
ligand selection from pools of random RNAs (
15
,
17
,
42
). After three rounds of binding and washing steps at moderate stringency,
followed by two rounds at high stringency, RNAs bound by GST-Mel-N1 were reverse-transcribed and 20 cDNA clones were subjected to sequence
analysis (Table
1
). Of these, 15 clones comprised a single sequence with a motif resembling the
core AUUUA element of ARE. Three other clones contained slightly different
sequences which were also comprised of ARE-like elements. A tetranucleotide of GAAA was found downstream of the AU-rich region in all of the above 18 clones. The sequences of the
remaining two clones were more varied, both contained an AUUUG followed by 7-11 consecutive polyuridines headed by an A residue. No GAAA sequence was
found in either of these two clones. Enrichment of sequences displaying such
similar features suggest that these motifs may represent the target sequences
recognized by Mel-N1.
In vitro
RNA binding of Mel-N1
To confirm the results obtained from the
in vitro
selection further, we selected four RNAs as representative Mel-N1-selected RNAs (clones 22-2, 23-1, 27 and 28-1) and directly tested Mel-N1 RNA binding ability by means of UV-crosslinking assays (Fig.
3
). We found that GST-Mel-N1 was able to specifically bind to all of the RNAs. To evaluate the
effect of the GAAA tetranucleotide on Mel-N1 RNA binding, we altered the sequence within clone 27 to CCCC by
in vitro
mutagenesis and checked the binding affinity of the mutated RNA (27m). Mel-N1 binding to the mutant RNA was reduced by ~3-fold as compared with that of the wild-type RNA, indicating that the GAAA sequence may exert a
positive influence on Mel-N1 binding.
Mel-N1 binding to the
c-fos
ARE
To confirm previous observations that the human homologue of Mel-N1 binds to the
c-fos
3'UTR and to determine the precise binding site within the 3'UTR, we examined Mel-N1 binding to two kinds of RNAs, FosA and FosB, which are
derived from the mouse
c-fos
3'UTR ARE and whose sequences are highly conserved between mice and humans
(Fig.
4
and see
43
,
44
). As the presence of ARE core sequences predicted, both FosA and FosB RNAs were
bound specifically by Mel-N1. As judged from the intensity of the crosslinked bands, FosA seems to
be bound ~2-fold more efficiently by than FosB, possibly due to the fact that
FosA contains two overlapping ARE core repeats. These results demonstrate that
Mel-N1 binds to the
c-fos
ARE and suggest that the ARE contains at least two Mel-N1 binding sites.
Autogenous binding of Mel-N1 to its 3
'
UTR
As mentioned above, the Mel-N1 3'UTR is rich in A and U residues. Close inspection revealed that
there are six copies of the ARE core (AU
3
A) and 12 of related sequences (AU
5
A, AU
6
A, AU
3
G, AU
4
G), within the 3'UTR (see Fig.
1
). In addition, 11 copies of the GAAA motif are present in the 3'UTR, most of which are located near the ARE and ARE-like sequences. Since these types of sequences were selected
in vitro
and bound by Mel-N1, we examined whether Mel-N1 could bind to the 3'UTR of its own mRNA (Fig.
5
) by dividing the Mel-N1 cDNA into three segments and synthesizing the corresponding RNAs. The
results show that indeed Mel-N1 could bind the UTR-A and UTR-B RNAs derived from its 3'UTR. No significant binding was observed with ORF RNA
derived from the coding region.
Cytoplasmic localization of Mel-N1
To elucidate the cellular localization pattern of Mel-N1, we designed an expression plasmid, pEFT7-Mel-N1, which is driven by an EF-1[alpha] promoter (
40
) with a T7 epitope tag sequence fused upstream to the full-length Mel-N1 coding region. This expression vector has been shown to give
efficient yields of tagged protein in tissue culture cells (
41
). pEFT7-Mel-N1 was transiently transfected into HeLa and COS7 cells, and the
cells harvested 48 h after transfection. These cell extracts were subjected to
immunoblot analysis with a monoclonal antibody against the T7 tag. A discrete
band of a size corresponding to the tagged Mel-N1 was seen in extracts from both cell lines, indicating that the tagged
Mel-N1 protein is efficiently produced in cells (data not shown).
Cells were transfected with pEFT7-Mel-N1, incubated with antibody against the tag and secondary FITC-labelled antibody, and observed under confocal laser
microscope. In 70% of HeLa and 30% of COS7 cells which express Mel-N1, fluorescence could be observed only in the cytoplasmic region,
complementary to the DAPI-stained region (Fig.
6
). In the remaining Mel-N1 positive cells, which expressed much higher levels of Mel-N1, the fluorescence extended into the nuclei (data not shown). This
appears to be the result of excessive Mel-N1 overexpression and may not reflect the physiological situation. Our
results indicate that Mel-N1 localization occurs primarily in the cytoplasm.
DISCUSSION
The human HuD and Hel-N1 are among a new family of RNA binding proteins which are expressed in
the nervous system and have been demonstrated to possess specific RNA binding
ability
in vitro
(
28
,
32
). HuD binds to the ARE within the 3'UTR of c-
fos
mRNA, and Hel-N1 to the 3'UTR of c-
fos
, c-
myc
and granulocyte-macrophage colony-stimulating factor (GM-CSF) mRNAs. Although the precise sites on the mRNAs bound by
Hel-N1 have yet to be determined,
in vitro
RNA selection analysis by Gao
et al.
(
45
) suggested that Hel-N1 and its alternative isoform Hel-N2 recognize specific sequences resembling the ARE core, AUUUA. They
have shown that Hel-N1 can bind the 3'UTR RNAs of other growth-related mRNAs which contain ARE-like motifs.
In this study, we demonstrate that like its human counterpart, Mel-N1 specifically selects RNAs which contain ARE core-like sequences such as AU
3
G, GU
3
A, AU
4
A and AU
5
A. In most of the Mel-N1-selected RNAs, the ARE core-like sequences are accompanied by a downstream tetranucleotide
GAAA motif, which has not been reported for the Hel-N1 and Hel-N2 selected RNAs. We further show that Mel-N1 binding is reduced by ~3-fold upon disruption of this GAAA sequence.
Examination of the Mel-N1 3'UTR revealed that interestingly, GAAA sequences occur in high
frequency near ARE core-like sequences. Such GAAA sequences are also found in the vicinity of the
mouse and human c-
fos
AREs (
43
,
44
) suggesting that this purine-rich sequence may act as a site to assist in the recognition of ARE by Mel-N1. In addition, we demonstrate for the first time that Mel-N1 can bind to the highly conserved region of the
c-fos
ARE which has been shown to function in mRNA destabilization. This finding
confirms previous observations from work involving the human Hel-N1, and expands it to show that multiple Mel-N1 molecules may associate with the ARE during regulation of mRNA
stability.
Two of the Mel-N1-selected RNAs contained an ARE core-like sequence, AU
3
G, but no GAAA. Instead, they have long runs of U initiated by an A residue, AU
7
and AU
11
. Our
in vitro
RNA binding analysis indicated that Mel-N1 bound these RNAs as efficiently as the RNAs containing the GAAA
sequence. This suggests that the U-rich stretches may be involved in further strengthening the binding of Mel-N1 to the ARE core. Alternatively, Mel-N1 may recognize AU
7
and AU
11
as a part of the ARE core, AUUUA. In any case, efficient RNA binding of Mel-N1 seems to require at least two stretches of three or more consecutive U
residues as suggested by the data of Hel-N1-selected sequences (
45
).
An important discovery is that Mel-N1 can bind specifically to its own 3'UTR RNAs. In addition, we were able to demonstrate that Mel-N1 is localized predominantly in the cytoplasmic region.
These results suggest that Mel-N1 may post-transcriptionally regulate the expression of its own mRNA in
addition to other ARE-containing mRNAs and contribute to various aspects of neuronal
differentiation and/or maintenance. It will be of great interest to examine
whether such an autonomous binding occurs also with the human Hel-N1 and other related proteins. We further found that two other proteins,
mHuC and mHuD, murine homologues of human HuC and HuD, were able to bind to
both the Mel-N1-selected RNAs and the Mel-N1 3'UTR RNAs (R.A. and H.S., unpublished data). Whether or
not these mHuC and mHuD also contain ARE-like sequences within their 3'UTRs remains yet unclear. Since the ARE has been shown to cause
destabilization of mRNA, Mel-N1 and other related proteins might function as
trans
-acting factors which facilitate mRNA decay
in vivo
. Alternatively, they might act as mRNA stabilizing factors which compete with
other potential destabilizing factors such as URBPs which have been shown to
bind the c-
fos
ARE (
44
). It remains to be shown whether Mel-N1 affects the expression of ARE-containing mRNA
in vivo
and, if does, at which post-transcriptional level Mel-N1 and its closely related proteins exert their effect.
ACKNOWLEDGEMENTS
We are grateful to Mutsuhito Ohno for pEF-BOS-T7 plasmid vector and Hiroshi Kawai for technical help and advice on
microscopy.
We thank Ruth Yu for critical reading of the manuscript. The nucleotide sequence
of Mel-N1 cDNA described in this paper will appear in GenBank with the accession
no. U29088. This work was supported in part by research grants from the
Ministry of Education, Science and Culture of Japan, the Asahi Glass
Foundation, the Inamori Foundation, and the Senri Life Science Foundation.
39 Maniatis, T., Fritsch, E.F. and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
40 Mizushima, S. and Nagata, S. (1990) Nucleic Acids Res., 18, 5322. MEDLINE Abstract
