Molecular cloning of a RNA binding protein, S1-1
Akira
Inoue*
,
Kenichi Paulo
Takahashi
2
,
Masatsugu
Kimura
1
,
Takanori
Watanabe
and
Seiji
Morisawa
3
Department of Biochemistry and
1
Laboratory of Biophysics, Osaka City University Medical School, 1-4-54 Asahimachi,
Abeno-ku
, Osaka 545,
Japan
,
2
Department of Anatomy and Physiology, Osaka Prefectural College of Health
Sciences, 3-7-30 Habikino,
Habikino
, Osaka 583,
Japan
and
3
Seishin School of Nursing, 78-53 Kande-cho, Yoshinari,
Nishi-ku
, Kobe 651-23,
Japan
Received March 28, 1996;
Revised and Accepted June 17, 1996
DDBJ accession no. D83948
ABSTRACT
S1 proteins A-D constitute a nuclear protein family that are liberated rapidly in a set
from chromatin by mild digestion with a DNA or RNA hydrolyzing enzyme. With an
anti-S1-protein B antiserum that reacted with B2, C1 and D1, a cDNA clone,
pS1-1, was obtained, which encoded a protein of 852 amino acids. The S1-1 protein, encoded within the cells by a mRNA of 3480 nt, was a
novel protein and could be distinguished from the S1 proteins B, C and D by
their amino acid sequences. The S1-1 protein synthesized by
in vitro
translation bound to RNA homopolymers, with a preference for G and U
polyribonucleotides and little for poly(A). The protein contained two tandem
RNP motifs and several intriguing sequences, such as a novel repeat of five
octamers with a consensus sequence DP-S(Q/G)YYY and a potentially perfect amphipathic
[alpha]
-helix of five turns with basic and acidic amino acids positioned in an
ordered way. The two RNP motif sequences were similar, although homologies were
low, to the RNP motif sequences of yeast NSR1 protein, animal nucleolins,
Drosophila
hnRNP A1 and tobacco chloroplast RNP precursor protein, suggesting a functional
uniqueness of the S1-1 protein in RNA metabolism and also the evolution of its RNP motif
structure before plants and animals diverged. These results indicate that the
S1-1 protein encoded by the cDNA is a new class of RNA binding protein.
INTRODUCTION
A group of proteins that are functionally related can often be extracted in a
set under particular conditions and such a property helps further
characterization of the proteins. Examples of such nuclear proteins are
histones, HMG proteins and nuclear lamina pore complexes: they are isolated by
extraction from nuclei with dilute mineral acids (for a review see
1
), with 0.35 M NaCl and then 2% trichloroacetic acid (for a review see
2
), and with Triton X-100 and high salt solutions as an insoluble residue from the nuclei
digested with DNase I (
3
,
4
), respectively.
S1 proteins constitute another such group of nuclear proteins. They are
extracted selectively at pH 4.9 from the supernatant of nuclei treated mildly
with DNase or RNase (
5
-
7
). The S1 proteins are composed of proteins
A, B, C and D, each separable into doublets by SDS-PAGE: A1 (80.0 kDa) and A2 (76.1); B1 (49.5) and B2 (48.2); C1 (45.2) and
C2 (44.5); D1 (41.5) and D2 (39.5). They are liberated from nuclei with closely
similar kinetics on DNase I digestion, suggesting that the S1 proteins are
present in the same or very similar sites in the nucleus. They have been found
in all rat tissues examined so far (
5
) and in mammals, a bird (chicken), a fish (carp), an amphibian (frog;
unpublished data) and an echinoderm (starfish;
8
). Polyclonal antibodies raised in two rabbits with protein B as immunogen both
reacted with proteins B2, C1 and D1 (
9
,
10
). With these antisera, the S1 proteins were localized in the extranucleolar
nucleoplasm, in the euchromatin bordering the heterochromatic areas (
9
), where most RNA is synthesized (for a review see
11
). The S1 proteins constitute a family, as shown by their shared epitopes and
primary structures: in addition to the fact that the polyclonal antibodies
reacted with B2, C1 and D1, an anti-S1 protein monoclonal antibody produced by a cloned hybridoma reacted with
C2 and D2, and all of these proteins have identical or very similar amino acid
sequences, which have been partially analyzed (details to be published
elsewhere). In addition, it was recently found that they occur in association
with hnRNA in the cell nucleus; for example they sediment in association with
RNase-sensitive complexes of heterogeneous sizes with
S
values up to 200 or more (details to be published elsewhere).
To understand their structures and functions, we undertook molecular cloning
using the polyclonal antiserum and obtained a cDNA clone, pS1-1, from rat liver cDNA libraries. It was shown that while the S1-1 protein had similarities to the S1 proteins in immunoreactivity
and in having RNA binding activity, it was an RNA binding protein not
previously reported.
MATERIALS AND METHODS
Isolation of RNA
Total RNA was prepared from rat liver by the method of Chomczynski and Sacchi (
12
) and dissolved in DEPC-treated H
2
O. Poly(A)
+
RNA was isolated from total RNA with oligo(dT)-conjugated ceramic beads (Oligotex-dT30 Super; Roche, Japan).
Construction and screening of cDNA libraries
Double-stranded cDNA was synthesized from poly(A)
+
RNA, with random hexamers or oligo(dT) as a primer in the reverse transcription
reaction, and cDNAs were inserted into the
Eco
RI site of [lambda]gt 11 or between the
Eco
RI and
Xho
I sites of the [lambda]ZAP-II vector, with a kit from Amersham ([lambda]gt 11 cloning system RPN 1763) or a kit from Stratagene
(ZAP-cDNA synthesis kit 200400, Gigapack II packaging extract no. 200214). Packaged recombinant phages were screened by the standard method (
13
) with an anti-S1 protein antiserum raised in a rabbit (
10
) or
32
P-labeled cDNA fragments. Color was developed in the immunoscreening with a
horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody and 3-amino-9-ethylcarbazole/H
2
O
2
. The primary and secondary antibodies for immunoscreening were pretreated with
an extract from
Escherichia coli
. In DNA hybridization, blots on nylon membranes were hybridized at 65oC for 20 h in 5* SSPE (0.9 M NaCl, 50 mM sodium phosphate and 5 mM EDTA, pH 7.7) containing 5* Denhardt's solution, 1% SDS and 20 [mu]g/ml sonicated and heat-denatured salmon sperm DNA and then washed. Final
washes were in 0.1* SSPE containing 0.1% SDS at 65oC for 10 min and the membranes were analyzed by autoradiography.
DNA labeling
cDNA inserts were released from vector arms by digestion with a restriction
enzyme(s) and separated on low melting temperature agarose gels. DNAs in the
melted gel slices were labeled with random hexanucleotide primers, [[alpha]-
32
P]dCTP (3000-6000 Ci/mmol) and Klenow I fragment (Multiprime DNA labeling system RPN
1601Y; Amersham). Labeled DNAs were purified by reverse phase chromatography on
Nensorb 20 cartridges (NLP-022; NEN).
Oligonucleotide synthesis
Oligonucleotides were synthesized on a DNA synthesizer (model 381A; Applied
Biosystems), purified by extraction with phenol/chloroform or chromatography on
OPC columns (Applied Biosystems) and used as primers in PCR or in DNA
sequencing.
DNA sequencing
Sequencing of cDNAs subcloned into the pUC118 or pBluescript vectors was
performed after progressive unidirectional deletion of the cDNA insert with a
kit (Kilo-sequence deletion kit; Takara Shuzo Co., Kyoto) based on the methods of
Henikoff (
14
) and Yanisch-Perron
et al
. (
15
) or by sequence extension with synthetic oligonucleotide primers. Sequencing
reactions of double-stranded DNA were by the dideoxynucleotide chain termination method (
16
) with a Taq Dideoxy terminator cycle sequencing kit (Applied Biosystems) and the reaction products were analyzed on an Applied Biosystems model 370A DNA sequencer.
RT-PCR of S1-1 mRNA
RT-PCR was carried out with a kit (Gene Amp RNA PCR Kit N808-0017; Perkin Elmer) according to the manufacturer's protocol with the
following modifications. The rat liver total RNA (0.5-0.8 [mu]g) was heat denatured at 70oC for 3 min in 4 [mu]l H
2
O in the presence of the 3' primer and kept at 55oC for 1 min. To this total RNA was then immediately added 16 [mu]l of the reaction mixture, which contained all other ingredients
and had been kept at the same temperature for 5 min. Reverse transcription was
done, with the temperature being lowered gradually to 42oC over 15 min. The 5' primer in the following PCR was designed to break a putative
hairpin secondary structure of 39 nt (see Fig.
2
legend).
Sequence similarity search and structural analysis
Sequence similarity was searched for in the protein database from SWISS-PROT and the alignment was optimized as described by Dayhoff (
17
). Structural analysis was with the Genetyx system, version 7.3 (Software
Development Co. Ltd, Tokyo): protein secondary structures were predicted based
on the methods of Chou and Fasman (
18
,
19
) and of Robson (
20
), the wheel models on that of Schiffer and Edmundson (
2 1
) and hydropathy profiles on Hopp and Woods (
2 2
), averaging each hydropathy value for three successive amino acid residues.
In vitro
transcription and translation
The pBluescript plasmid containing S1-1 cDNA (pS1-1) was linearized with
Kpn
I, and S1-1 protein was synthesized by
in vitro
transcription with T3 RNA polymerase followed by translation in reticulocyte
lysate with kits from Stratagene (mCAP mRNA Capping Kit and
In Vitro
Express Translation Kit). For good translation, the RNA transcript (8 [mu]g) was incubated in H
2
O (15 [mu]l) at 68oC for 45 s and immediately mixed with 100 [mu]l nuclease-treated rabbit reticulocyte lysate (Wako Pure Chemicals,
Osaka) and l0 [mu]l [
35
S]methionine (100 [mu]Ci, 1000 Ci/mmol). The mixture was incubated at 37oC for 1 h.
RNA binding assays
The binding procedure was essentially the same as described by Siomi
et al
. (
2 3
). The [
35
S]S1-1 protein produced by
in vitro
translation was incubated at 4oC for 15 min by gentle rocking with RNA homopolymers bound to agarose
matrix [poly(A), poly(U) and poly(C); Sigma] or polyacrylhydrazido-agarose matrix [poly(U) and poly(G); Sigma] in a total of 0.5 ml binding
buffer (10 mM Tris-HCl, pH 7.4, 2.5 mM MgCl
2
and 0.5% Triton X-100). The beads were pelleted by a brief spin at 4oC and washed five times with cold binding buffer before resuspension
in 20 [mu]l SDS-PAGE loading buffer at 100oC for 10 min. SDS-PAGE (7.0%) of the centrifuged supernatant was as
described by Laemmli (
2 4
) and proteins were visualized by fluorography.
RESULTS
Cloning of S1-1 cDNA
A [lambda]gt11 library of rat liver cDNAs, synthesized with random hexanucleotides
as primers, was screened with a polyclonal anti-S1 protein B antibody. A clone with an insert of 317 bp (clone 12) was
isolated (Fig.
1
). By rescreening with the
32
P-labeled cDNA, clone 32 was obtained. The clone 32 cDNA was 2012 bp, with
an open reading frame (ORF) of 1668 nt and a start Met codon at nt 345. It did
not, however, have a stop codon and the cDNAs of most other clones were shorter
than sequence 12. A library was reconstructed in the ZAPII-[lambda] phage vector, this time using oligo(dT) as primer. Its five
clones, isolated with sequence 12 as probe, contained the cDNA inserts with a
polyadenylation signal, AAUAAA, a poly(A) tail and identical upstream sequences. The 5'-ends of these clones started, however, at 11-33 nt downstream of sequence 32, except for clone 3Z,
indicating premature termination in reverse transcription. Sequence 3Z (1423 bp) overlapped the 32 sequence, perfectly over 313 nt. The composite complete
coding sequence was named S1-1 (Fig.
1
).
Construction of complete S1-1 cDNA
Occurrence of S1-1 mRNA in rat liver cells
Northern analysis of total rat liver RNA demonstrated the presence of S1-1 mRNA. It was 3480 nt long (Fig.
2
a).
Northern analysis also verified the genuineness of the composite S1-1 sequence: the 5' and 3' probes, which interposed the overlapping region of the 32
and 3Z sequences (Fig.
1
), gave images of the same molecular size, and the band intensity obtained with
the 5' probe was ~2-fold, as expected (this probe was 2-fold larger in size than and had almost the same
specific radioactivity as the 3' probe).
RT-PCR of total rat liver RNA and sequencing of its products also confirmed the
S1-1 sequence (not shown). RT-PCR of the interposing overlapping region
amplified an expected 503 bp fragment (Fig.
1
). To obtain successful amplification, however, the 5' primer was specifically designed to break a putative hairpin structure of 39 nt located immediately upstream of the overlapping region (Fig.
2
b).
Thus, it was concluded that the S1-1 mRNA represented a true sequence present in the mRNA population in the
liver cells. S1-1 cDNA encodes a novel protein of 852 amino acids, with a molecular mass
of 94.3 kDa. It is composed of 14.4 and 13.3 mol% basic and acidic amino acids
respectively.
Relationship between S1-1 protein and S1 proteins B, C and D
The amino acid sequences were compared. The S1 proteins B, C and D (all N-termini blocked) were digested with
Staphylococcus aureus
V8 protease and microsequencing of the peptides was performed by Edman
degradation. The S1 proteins B2, C1 and D1 and proteins C2 and D2 constitute
two groups respectively, as judged by very similar peptide maps on SDS gels and
their similar amino acid sequences. In addition, the two groups are related:
for example both groups had similar pentadecamer sequences, which differed by
only three amino acids in the middle and one at the C-terminus. The S1 proteins were shown to associate with hnRNAs in the cell
nucleus (details to be published elsewhere).
In vitro
transcription and translation
The S1-1 protein had RNA binding activity. This was shown using [
35
S]Met-labeled S1-1 protein synthesized from the pS1-1 cDNA by
in vitro
transcription and translation. The estimated molecular mass of the translation
product was 102 kDa (Fig.
3
a), 8 kDa larger than the theoretical value. Since S1-1 protein contains a number of potential phosphorylation sites (18 sites
for casein kinase II, five for C-kinase, two for tyrosine kinase and one for A-kinase), the larger molecular mass probably resulted from
phosphorylation, which causes slower migration on SDS-PAGE. Truncated products at 77, 72 and 63 kDa are thought from the
molecular weights to have been synthesized from the internal methionines at
127, 236 and 312 respectively. The 47 kDa protein was a product of an
endogenous RNA in a rabbit reticulocyte lysate (Fig.
3
a).
RNA binding assays
The [
35
S]S1-1 protein bound various RNA homopolymers attached to solid matrix beads,
strongly to poly(G) and poly(U), less to poly(C) and little to poly(A) (Fig.
3
b). Binding was affected by increasing ionic strength and abolished at NaCl
concentrations >0.4 M, suggesting that it involved an ionic interaction.
In the presence of competing free RNA homopolymers, binding of S1-1 protein to poly(U) beads was competed more strongly by free poly(G),
while that to poly(G) beads by free poly(U) (Fig.
3
c). The results not only suggest that S1-1 protein interacts with both poly(G) and poly(U), but also that binding
of the S1-1 protein to RNA occurs by some unique mechanism.
DNA and total RNA also had competing activities, but their activities were low
even at concentrations at which free poly(U) and poly(G) caused almost complete
competition. Single-stranded DNA had a stronger activity than double-stranded DNA or total RNA (Fig.
3
d).
Structural features
The S1-1 protein contained two RNP motif sequences, I and II (Fig.
4
), at amino acids 61-131 and 217-307. Besides this, it had quite a few characteristic amino acid
sequences. Some of them are as follows.
The central ROH region of 100 amino acids (Y430-Y529) is rich in hydroxyl amino acids, 42% of the residues being Ser, Thr
and Tyr (Fig.
4
). Besides this, it has a novel tandem repeat of five octamers in its C-terminal half, with a consensus sequence of DP-S(Q/G)YYY [Fig.
5
a(i)]. Corresponding to the repeated structure, its
hydrophilicity/hydrophobicity profile shows a characteristic path of five up-and-downs, as shown in Figure
5
a(ii).
DISCUSSION
Cloned sequence S1-1
Upon screening of an expression library using a polyclonal anti-S1 protein antiserum, a positive clone was isolated. With this cDNA as
probe, libraries were re-screened and a composite clone (pS1-1) containing a complete coding sequence was obtained. The
genuineness of the sequence was verified by Northern blotting, RT-PCR and
sequencing of the amplified product. The S1-1 cDNA coded a new protein of 852 amino acid residues with RNA binding
activity.
The S1-1 protein was ~2-fold larger than the S1 proteins B, C and D. The possibility
that the S1-1 protein is a precursor of these proteins is unlikely, because it did not
match the amino acid sequences of the S1 proteins. On the other hand, the S1-1 protein had similarities to the S1 proteins in that it had RNA binding
activity and immunoreactivity to the anti-S1 protein antiserum. The latter indicates that the S1-1 protein has a common or closely similar epitope structure to those
of the S1 proteins B2, C1 and D1. For these reasons we named the protein coded
by the cDNA S1-1.
Premature termination sites in cDNA synthesis
A cDNA clone containing a complete coding sequence could not be isolated because
of premature terminations occurring in the region of nt 1700-2100. When cDNA synthesis was primed with random primers, MuLV reverse
transcriptase stopped around nt 1780 in most cases. Similarly, with an
oligo(dT) primer, the enzyme stopped mostly at nt 2024-2046. Computer-assisted examination for intramolecular base pairings indicated
that there were a number of possible long base pairings, particularly densely
concentrated in the regions nt 1460-1825 and 1935-2078. In fact,
in vitro
translation of S1-1 mRNA and RT-PCR to amplify this region (nt 1700-2100) were difficult. To obtain efficient
in vitro
translation, the S1-1 mRNA had to be used immediately after heat treatment (if it was left
longer than 5 min on ice it resulted in poor translation). Similarly, the
RT-PCR conditions were the best ones that could be achieved among many attempts
using various primers (Fig.
2
). Whether or not there is a functional significance in these stable structures
is intriguing.
RNP motif structures
In accord with the
in vitro
RNA binding activity, S1-1 protein had two RNP motifs (I and II, Fig.
4
) at amino acids 61-131 and 217-306 respectively (Fig.
6
). These RNP motif regions seemed to be responsible for the RNA binding activity
of S1-1 protein, as truncated S1-1 proteins synthesized from internal methionines and lacking the RNP
motifs did not usually show RNA binding activity (Fig.
3
).
RNP sequences are generally composed of ~90 amino acids and contain two short submotif sequences, RNP1 and RNP2
(reviewed in
2 5
,
2 6
). The S1-1 RNP I had a typical RNP1 submotif (RGFAFVEF), but lacked a RNP2 submotif
on the N-terminal side. In fact, RNP2 is generally less strictly conserved (
2 7
) and many RNP proteins lack this submotif (
2 5
,
2 8
). On the other hand, the second RNP II was more similar to the canonical
structure (
2 6
,
2 9
) and had the submotifs RNP1 (RGFAFIQL) and RNP2 (IILRNL).
The RNP motif I was similar to those of chicken nucleolin (
30
,
3 1
), yeast nuclear localization sequence binding protein NSR 1 (
3 2
), and
Drosophila
hnRNP A1 (
3 3
,
3 4
) and the RNP II to those of the yeast NSR 1 (
3 2
) and tobacco chloroplast RNP precursor proteins (
3 5
) (Table
1
and Fig.
6
). Some of these proteins possess two tandem RNP motifs and others four (Table
1
). Functionally similar RNP-containing proteins have similar domain organizations with respect to copy
number of the motif and auxiliary domains (
2 5
,
3 6
). This suggests that the S1-1 protein represents a new class of RNA binding protein. The
characteristic of the S1-1 protein RNP motifs is that they resemble both animal and plant RNPs. It
is likely that this S1-1 RNP motif structure evolved in the early stage of evolution, before
plants and animals appeared.
The range of sequence similarity between two unrelated RNP motifs has been
estimated at 10-20% identity by Birney and others (
24
, figs 1 and 4 therein); in the same regions as those analyzed by them, the S1-1 RNP motif I (amino acids 51-135) showed only 29% identity to that of chicken nucleolin (amino
acids 550-631), and RNP motif II (amino acids 222-310) 26% identity to that of yeast NSR 1 (amino acids 266-348) (Table
1
). Interestingly, the identity between RNP I and II of the S1-1 protein was also low, at 25%. These rather low homologies suggest that
the S1-1 protein should have a special function in RNA metabolism.
RNA binding properties
The S1-1 protein had RNA binding activity. It bound to poly(G) and poly(U) more
strongly than to other RNA homopolymers. Accordingly, it is suggested that the
S1-1 protein is a RNA binding protein that preferentially binds to regions
rich in G and U ribonucleotides.
In the competition experiments, binding of S1-1 protein to poly(U) beads was competed strongly by free poly(G), and that
to poly(G) beads by free poly(U). Possible duplex formation of poly(U) beads
with free poly(G) or of poly(G) beads with free poly(U) could not explain these
results, since free poly(A) or poly(C), which can form more stable duplexes
with poly(U) or poly(G) beads, exerted only small inhibitory effects under the
assay conditions used. These results imply that some unique mechanism operates
in the binding of S1-1 protein to RNA. The binding could occur in a random manner, a concerted
manner or a stepwise manner. The present results seem to favor a mechanism in
which multiple RNA binding sites participate in an ordered manner.
.
Similarity of known RNP sequences to the RNPs I and II of S1-1 protein
RNP of
Protein with similar
No. of
Similar RNP
Optimum
Identity
Reference
S1-1
RNP sequence
RNPs
in the repeat
score
(%)
RNP I
Chicken nucleolin
4
4th RNP
102
29
30
Yeast NSR1 protein
2
2nd RNP
102
25
32
Drosophila
hnRNP A1
2
2nd RNP
95
26
33, 34
Mammalian nucleolins
4
4th RNP
92
21
44-46
RNP II
Yeast NSR1 protein
2
2nd RNP
93
26
32
Tabacco 33 kDa RNP
2
1st RNP
87
25
35
Xenopus
nucleolin
4
4th RNP
86
26
47
Human HuD protein
3
3rd RNP
85
24
48
Sequences similar to the RNP I (amino acids 51-135) or II (amino acids 222-306) of S1-1 protein were searched in the SWISS-PROT database. Analyzed RNP regions correspond to
those specified in the alignment of 70 RNP proteins (25). Optimum score was
obtained according to Dayhoff (17). The top four proteins with the highest scores,
together with their RNP copy numbers and the position of the similar RNP in the
repeats, are cited. Identical amino acids (%) are also shown. Mammalian
nucleolins are those of mouse (44), rat (45) and golden hamster (46). The (possible) functions of these proteins are: nucleolins, in pre-rRNA transcription and ribosome assembly; NSR1 protein, in pre-rRNA processing; hnRNP A1, as a component of heterogeneous nuclear RNPs; tobacco 33 kDa RNP
precursor protein, in chloroplast RNA processing; HuD protein, in neuron-specific RNA processing. HuD protein has been characterized as a paraneoplastic
encephalomyelitis antigen (48). The score for tobacco 33 kDa RNP protein was given for the amino acid 141-191 region (amino acids 252-306 of the S1-1 protein), where the score was highest.
The S1-1 protein had binding affinity for single-stranded DNA. RNA binding proteins often show a similar activity
in vitro
; the significance of the S1-1 affinity for single-stranded DNA is unknown.
Other structural features
Another characteristic feature of the S1-1 protein was a novel tandem repeat of five octamers. This repeat sequence
may be functionally important by playing a role as a structural element in the
molecular architecture.
In addition, a positively charged region, K735-K759, had a bipartite structure, which satisfies a motif sequence for a
nuclear localization signal (
3 7
,
3 8
). Also, the Lys/Glu-rich sequence at K556-K568 satisfies the so-called KEKE motif, proposed by Realini and others (
3 9
) as a motif for promotion of association between proteins (Fig.
4
).
We conclude that the cloned cDNA encoded a protein, S1-1, which was a novel RNA binding protein with characteristic and unique
structural features.
ACKNOWLEDGEMENTS
AI thanks the fourth year students of Osaka City University Medical School for
their participation and enthusiasm, who chose the S1-1 study in their laboratory course in November and December. They are
(year): Shunnsuke Fujimoto (1990), Yuko Asechi, Kinuko Kono and Yamazaki Shioko
(1991), Sinnichiro Ishino, Shuji Iwai and Rie Shibata (1992), Kazumitsu Ueda
and Hidetada Fukushima (1993), Hideo Kuniba and Tomoaki Morioka (1994) and Takanori Watanabe (1995).
REFERENCES
1 Johns,E.W. (1971) In Phillips,D.M.P. (ed.), Histones and Nucleohistones. Plenum, New York, NY, pp. 2-45.
