ABSTRACT
Introns of nuclear pre-mRNAs in dicotyledonous plants, unlike introns in vertebrates or yeast,
are distinctly rich in A+U nucleotides and this feature is essential for their
processing. In order to define more precisely sequence elements important for
intron recognition in plants, we investigated the effects of short insertions,
either U-rich or A-rich, on splicing of synthetic introns in transfected protoplasts of
Nicotiana plumbaginifolia
. It was found that insertions of U-rich (sequence UUUUUAU) but not A-rich (AUAAAAA) segments can activate splicing of a GC-rich synthetic intron, and that U-rich segments, or multimers thereof, can function
irrespective of the site of insertion within the intron. Insertions of multiple
U-rich segments, either at the same or different locations, generally had an
additive, stimulatory effect on splicing. Mutational analysis showed that
replacement of one or two U residues in the UUUUUAU sequence with A or C
residues had only a small effect on splicing, but replacement with G residues
was strongly inhibitory. Proteins that interact with fragments of natural and
synthetic pre-mRNAs
in vitro
were identified in nuclear extracts of
N.plumbaginifolia
by UV cross-linking. The profile of cross-linked plant proteins was considerably less complex than that
obtained with a HeLa cell nuclear extract. Two major cross-linkable plant proteins had apparent molecular mass of 50 and 54 kDa and
showed affinity for oligouridilates present in synGC introns or for poly(U).
Accurate splicing of nuclear pre-mRNAs requires that exon and intron sequences are effectively
distinguished from each other and that appropriate pairs of 5' and 3' splice sites at the intron/exon junctions are precisely
recognised and juxtaposed. Although some of the pre-mRNA
cis
-acting elements essential for accurate intron excision, such as the 5' splice site (5'ss) or the intron 3'-terminal AG, are similar in all eukaryotes,
others are either unique or their contribution differs significantly between
different organisms (for reviews see
1
-
6
). One of the characteristic features of introns in vertebrates and insects is
the presence of the polypyrimidine tract usually positioned immediately
upstream of the 3'ss (
1
,
3
). Early during spliceosome assembly the polypyrimidine tract is recognised by
the splicing factor U2AF (
7
,
8
; and references therein). U2AF is required for the binding of the U2 snRNP to
the branch point region the sequence of which is not highly conserved in
vertebrate introns (
1
,
3
,
9
,
10
). Pre-mRNAs in mammals and in
Drosophila
may also contain
cis
-acting elements located in exons. These elements, known as splicing
enhancers, play an essential role in splice site selection, during both
constitutive and alternative splicing, and are recognised by members of the SR
family of proteins or related splicing factors (
5
,
11
,
12
; and references therein).
The distinguishing feature of pre-mRNA introns in the yeast
Saccharomyces cerevisiae
is a highly conserved branch point sequence, UACUAAC, which base-pairs with the U2 snRNA (
2
,
6
,
13
). Moreover, most introns in yeast lack the extensive polypyrimidine tracts
between the branch site and the 3'ss. The extended U2 snRNA-branch point base-pairing in yeast appears to compensate for the absence of the
downstream auxiliary signal. Indeed, sequences positioned 3' of the branch site do not significantly contribute to the first step of
splicing in yeast (
2
,
14
; and references therein). Some yeast introns contain uridine-rich tracts in the 3'ss-proximal region. However, these tracts function during late
stages of splicing (
15
,
16
).
Requirements for intron recognition in higher plants differ from those in
vertebrates and yeast. The 5'ss and 3'ss in plant pre-mRNAs resemble their vertebrate counterparts but plant
introns contain neither distinct 3'ss-proximal polypyrimidine tracts nor conserved branch point sequences
similar to those found in vertebrate and yeast introns, respectively (reviewed
in
4
,
17
,
18
). Experiments with synthetic model introns have shown that these two signals
are indeed not essential for intron processing in transfected protoplasts of
Nicotiana plumbaginifolia
or maize (
19
,
20
). In two cases analysed, branching in plants was found to occur to A residues
positioned 31 and 32 nt upstream of the 3'ss (
21
). A characteristic feature of plant introns is their AU-richness and this property is essential for splicing. In dicot plants
studied more extensively, introns are on average ~15% more AU-rich than flanking exon sequences, with U residues usually
contributing much more to the AU-richness than the A residues. In only a very few cases does the A+U
content in introns fall below 60% and it is never <55% (
19
-
20
,
22
,
23
; reviewed in
4
). Experiments with synthetic and natural introns have shown that the high A+U
content is absolutely essential for efficient intron processing in dicot plant
cells (
19
,
20
). Consistent with this, when heterologous introns were tested for splicing,
only those which were AU-rich were processed at significant levels (
19
,
20
,
22
,
24
-
26
). Moreover, synthetic or natural non-intron sequences can function as introns as long as they are AU-rich (
19
,
20
,
27
,
28
). The requirement for AU-rich introns is not confined to dicot plants. Most introns in monocot
plants are AU-rich and this property facilitates splicing (
20
,
28
). Introns in
Drosophila
, nematodes, ciliates, slime moulds, and some other organisms are also AU-rich (
19
,
22
,
23
,
29
,
30
), and in nematodes AU-richness has been shown to be important for both
cis
- and
trans
-splicing (
31
,
32
).
It is not known how AU-rich sequences, usually distributed along the whole length of the intron,
contribute to intron recognition or splicing efficiency. Since hairpins
introduced into introns strongly inhibit splicing in dicot plant cells, the
role of AU-richness could be to minimise secondary structure in introns (
20
,
33
). A more plausible possibility is that AU-rich segments act as binding sites for heterogenous nuclear RNP (hnRNP)
proteins (for recent reviews on mammalian and yeast hnRNP proteins, see
34
,
35
) or other protein factors that help to delineate sequences to be excised as
introns. The latter possibility was originally suggested by the observation
that insertions of AU-rich segments of 30 nucleotides (nt) or longer restore efficient splicing
of an artificial GC-rich intron, suggesting a positive effect of AU-rich segments on splicing rather than its inhibition by GC-rich sequences (
19
). This notion is further supported by the finding that the 5'ss and 3'ss selected for splicing in tobacco cells are those present at
regions showing a transition from AU-rich to GC-rich sequences, while splice sites embedded within the AU-rich intron sequence are not efficiently utilised (
36
,
37
). Analysis of the 3'ss-proximal AU-rich region indicated that U residues may contribute more to
the definition of the 3'ss than A residues (
36
).
In this work, in order to define more precisely sequence elements important for
intron recognition in plants and the possible mode of their function, we have
investigated the effects of short insertions, either U-rich or A-rich, on splicing of synthetic GC-rich introns in protoplasts of
N.plumbaginifolia
. We find that short U-rich segments such as UUUUAU or multimers thereof, but not A-rich sequences, can activate splicing of the GC-rich intron. The U-rich elements can function irrespective of whether they
are inserted in the proximity of the 5' or 3'ss or in the middle of the intron, indicating that they function
differently from polypyrimidine tracts of metazoa. We have found that nuclear
extracts of
N.plumbaginifolia
contain a limited set of proteins which interact with fragments of natural and
synthetic pre-mRNAs
in vitro
. Two major cross-linkable plant proteins of apparent molecular mass of 50 and 54 kDa have
affinity for oligouridilates which makes them good candidates for factors
involved in intron recognition in plants.
Unless indicated otherwise, all DNA manipulations described in this and other
sections were carried out according to (
38
). Sequences of all inserts were verified by dideoxy sequencing.
The synGC and synGCm constructs are derivatives of a vector pDELbm, a modified
version of pDELb (
19
). To obtain pDELbm, the T7 promoter-polylinker-SP6 promoter region of pGEM1 was PCR-amplified, using T7- and SP6-promoter-specific primers (Promega), and cloned into
pDELb precut with
Sma
I and
Pst
I and treated with T4 polymerase to blunt the
Pst
I end. SynGC intron, a derivative of intron syn24, was first constructed in
plasmid pGS24 (
19
) by replacing its
Kpn
I-
Cla
I and
Nco
I-
Xho
I regions with new synthetic oligonucleotides, yielding pGSgc. The
Xba
I-
Pst
I fragment of pGSgc was then cloned into pDELbm cut with the same enzymes,
yielding pSynGC. pSynGCm was obtained by replacing the
Nco
I-
Xho
I fragment of pSynGC with the sequence CATGGCCGCCCCGCGG
The following templates were used for
in vitro
synthesis of either cold or
32
P-labelled RNAs: (i) for syn7 RNA, pGS7 linearised with
Pst
I (
19
); (ii) syn7/cDNA RNA, pGEM2/cDNA linearised with
Pst
I (
39
); (iii) syn7/IVS, pGS7c (derivative of pGS7 with removed
Xba
I-
Cla
I fragment) linearised with
Nco
I; (iv) human [beta]-globin (h[beta]) RNA, ph[beta]-2, linearised with
Bam
HI (
22
); (v) Leghemoglobin (Leg) RNA, pHbII linearised with
Hin
cII [pHbII is the pGEM1-based plasmid containing the 0.78 kb
Acc
I-
Hin
dIII fragment subcloned from pLb-1 which contains the soybean leghemoglobin c3 gene insert (
22
)]; (vi) phaseolin (Phas) RNA, pGphas3 linearised with
Hin
cII (pGphas3 is a pGEM2-based plasmid containing the 0.71 kb
Eco
RI-
Xba
I fragment of the French bean phaseolin gene subcloned from pDEphas (
20
); (vii) Waxy9 RNA, pGEM1-based plasmid containing the 0.75 kb
Bam
HI-
Pvu
II
waxy
gene fragment (
20
), linearised with
Pst
I; (viii) for synthesis of synGC-specific RNA and its variants containing U-island insertions, pSynGC and its derivatives were linearised with
Xho
I. For additional information about RNA transcripts, see the legend to Figure
4
. pSynGC- and pSynGCm-derived plasmids used for preparation of complementary probes were
linearised with
Eco
RV.
In vitro
transcriptions were performed as described (
40
). RNAs used for RNase A/T1 mapping and UV cross-linking were labelled with [[alpha]-
32
P]CTP (sp. act. 80 Ci/mmol) and [[alpha]-
32
P]UTP (100-200 Ci/mmol), respectively. All non-radioactive and radiolabelled RNAs were purifed by electrophoresis
in 8 M urea/PAGE.
Transfections of leaf protoplasts of
N.plumbaginifolia
, isolation of protoplast RNA and RNase A/T1 mapping were performed as described
(
20
,
40
). The efficiency of splicing was determined also as described (
33
). Values given represent the means of at least three independent experiments.
Isolation of plant nuclei.
A cell suspension of
N.plumbaginifolia
(obtained from Dr I. Negrutiu, University of Lyon, France) was cultured in NP
medium [MS medium (see
40
) with altered vitamins: 0.5 mg/ml Ca-pantothenate and 2.5 mg/ml thiamin-HCl] and diluted 1:5 once a week. Protoplasts (3-4 * 10
8
) were prepared from 400 ml of cell suspension collected 5 days after
subculturing. Pelleted cells (50 ml) were incubated overnight at 28oC in the dark with an equal volume of enzyme solution [5 mM MES, 5 mM CaCl
2
, 0.47 M sucrose, 1.5% Cellulase Onozuka R10, 0.5% Macerozyme R10 (both from
Yacult Honsha Co., Japan) (pH 5.6, 556 mOs)]. Protoplasts were filtered through
a 100 [mu]m sieve. Filtrate aliquots (40 ml) were overlayered with 5 ml of W5 solution
(
40
) and centrifuged at 600 r.p.m. for 10 min. Protoplasts, floating and present at
the interphase, were collected and washed twice with W5. Protoplasts were
homogenised in 30 ml of Buffer H (20 mM MES pH 6.0, 5 mM EDTA, 0.15 mM
spermine, 0.5 mM spermidine, 10 mM [beta]-mercaptoethanol, 1 mM PMSF, 1 mg/ml leupeptin, 1 mg/ml antipain)
with a few strokes in a Dounce homogeniser (pestle A). The slurry was filtered
through 60-, 30/40- and 10/15 [mu]m mesh nylon sieves. The nuclei were pelleted by centrifugation
at 1200
g
for 5 min, taken up in 50 ml of Buffer H and filtered through a 10/15 [mu]m mesh nylon sieve. The sedimentation/filtering step was repeated and the
pellet taken up in 20 ml of buffer L (20 mM HEPES pH 7.5, 50 mM KCl, 2 mM MgCl
2,
0.1 mM EDTA, 10% glycerol, 2 mM DTT, 1 mM PMSF, 1 mg/ml leupeptin, 1 mg/ml
antipain).
A model pre-mRNA used to investigate the effect of U- or A-rich sequence elements on splicing is a derivative of the syn7
pre-mRNA (
19
; see Fig.
1
). This pre-mRNA, called synGC, contains a synthetic intron (IVS1) which is 75% GC-rich and is devoid of AU-rich stretches of >= 4 nt. In addition, synthetic exons that flank the intron
contain T7 and SP6 promoters to facilitate the
in vitro
synthesis of sense and antisense RNAs. The 5'ss of the synGC intron conforms to the consensus derived for dicot plant
introns (AG/GTAAGT; ref.
4
), while the 3'ss (CGCAG/GT) deviates from the consensus (TGCAG/GT) at position -5 (Fig.
1
B). Plasmids expressing the synGC pre-mRNA or derivatives thereof were transfected into protoplasts of
N.plumbaginifolia
. Splicing efficiency was analysed by RNase A/T1 mapping using RNA isolated from
transfected protoplasts and
32
P-labelled RNA probes complementary to the unspliced form of RNA. Consistent
with previous findings, demonstrating that pre-mRNAs containing GC-rich introns are not spliced in dicot plant cells, the efficiency of
synGC intron processing was <5% (Fig.
2
). The control syn7 intron which is 75% AU-rich was spliced with an efficiency of 85% (Fig.
2
A).
We investigated the effect of short, U-rich (sequence UUUUUAU) or A-rich (sequence AUAAAAA) insertions on splicing of the synGC pre-mRNA. These sequences (referred to as U- or A-rich islands) or multimers thereof, flanked by a
few G or C nucleotides in order to facilitate cloning, were inserted either
close to the 5'ss or 3'ss or in the middle of the intron (insertions
Cla
I,
Sac
II and
Mlu
I, respectively; see Figs
1
B and
2
B). Insertions of one, two or three U-rich islands near the 5'ss resulted in splicing of 15, 29 and 41% of the pre-mRNA, respectively. The insertion of a single U-rich island in the middle of the intron resulted in 18%
of splicing, while pre-mRNAs with two U-rich islands separated by either two (intron MluU2) or six (intron MluU2*) G+C nucleotides were spliced 55 and 42%, respectively. The presence of one U-rich island at the
Sac
II site, 11 nt upstream of the acceptor AG, did not activate splicing, but pre-mRNAs with two and three islands at this location were efficiently spliced
(41 and 56%, respectively; Fig.
2
). Insertions of U-rich islands at two different locations had an additive stimulatory
effect. Introns containing two U islands at the
Sac
II site and either one (intron ClaU/SacU2) or two (intron ClaU2/SacU2) islands inserted at the
Cla
I site were spliced at levels of 57% and 66%, respectively (Fig.
2
, and data not shown), i.e. more efficiently then introns bearing insertions at
only one site.
The accuracy of splicing of introns containing U-rich island insertions at different locations was verified by reverse
transcription-polymerase chain reaction (RT-PCR) analysis, using exon 1- and exon 2-specific oligonucleotides as primers. In all instances, products
with the expected sizes of unspliced and spliced RNAs were identified; no
products suggestive of alternatively spliced RNAs were apparent. The product
diagnostic of spliced ClaU2/SacU2 RNA was isolated from a gel and cloned. Of 14
clones sequenced, all represented accurately spliced RNA (data not shown).
Although most dicot plant introns contain more U residues than A residues, ~10% have more As than Us (
17
,
19
). Therefore, we investigated whether A-rich sequences also have the potential to activate splicing when inserted
into the synGC intron. Insertions of one or two A-rich islands at the
Cla
I,
Mlu
I or
Sac
II site had no effect on processing of the synthetic intron (Fig.
2
). Taken together, these results demonstrate that insertion of U-rich but not A-rich elements can activate splicing of the GC-rich intron, and that U-rich elements can function irrespective of the position
of insertion within the intron.
We and others (
18
,
19
,
36
) have proposed that intronic AU- or U-rich sequences may be recognised by specific proteins that help to
delineate intron borders or to attract components of the splicing machinery. To
investigate whether plant nuclear extracts contain proteins which can interact
with pre-mRNA transcripts, the technique of UV-cross-linking was employed. In the experiment shown in Figure
4
A, uniformly [[alpha]-
32
P]UTP-labelled RNAs containing synthetic intron syn7, intron 2 of the soybean
leghemoglobin (Leg) gene and intron 1 of the bean phaseolin (Phas) gene were
used (in each RNA the intron is flanked by exon sequences; see legend to Fig.
4
A). All these introns are AU-rich (71-75% AU) and are efficiently processed in transfected protoplasts of
N.plumbaginifolia
(
20
). The RNAs were incubated with nuclear extracts prepared either from
N.plumbaginifolia
or HeLa cells. The major cross-linkable proteins identified in plant extracts had apparent molecular mass
of 50 and 54 kDa; other less intensely labelled proteins of 68 and 80 kDa were
also found to be reproducibly cross-linked to the synthetic and natural plant RNAs (lanes 1-3). The intensity of the ~45 kDa band varied with different extract preparations; as did
intensities of other minor bands of ~38 and ~30 kDa (Fig.
4
A, see also below). The profile of cross-linked proteins seen with the plant extract was significantly less complex
than that seen with the HeLa cell nuclear extract (lanes 4-6). In the latter case, patterns of cross-linked proteins differed considerably between individual RNA
substrates (Fig.
4
A). This is consistent with previous findings that HeLa cell nuclei contain a
large population of hnRNP proteins which interact with pre-mRNAs with evident sequence preferences (reviewed in
34
,
35
,
44
).
The experiment presented in Figure
4
B is similar to that shown in Figure
4
A but compares two types of extracts prepared from nuclei of
N.plumbaginifolia
(see legend to Figure
4
B, and Materials and Methods), and four additional RNA substrates were tested.
Syn7/cDNA corresponds to a cDNA version of syn7, devoid of the intron, while
syn7/IVS is an internal fragment of the syn7 intron. Waxy contains intron 9 of
the maize
waxy
gene and h[beta] contains intron 1 of the human [beta]-globin gene. The latter two introns, being GC-rich (60 and 55% GC, respectively), are not processed in
N.plumbaginifolia
protoplasts (
20
). The results show that both methods of extract preparation yield comparable
patterns of cross-linked proteins. As in the experiment shown in Figure
4
A, Leg RNA acted as an efficient substrate but Waxy and h[beta] transcripts, containing non-functional introns, did not become cross-linked to
N.plumbaginifolia
proteins. Comparison of syn7/IVS and syn7/cDNA RNAs revealed that it is intron
and not exon sequences of syn7 which become cross-linked to the proteins (Fig.
4
B, compare lanes 1 and 2, and lanes 6 and 7). Additional experiments showed no
significant differences in cross-linking of the syn7 intron internal fragments and intron fragments which
extend into exons and encompass either the 5'ss or the 3'ss (data not shown). Thus, neither splice site seems to be
essential for binding of proteins identified by the cross-linking assay.
The finding that U- but not A-rich islands can trigger splicing of the inactive synGC intron is
consistent with the observation that in natural plant introns U residues
generally contribute more to the A+U nucleotide bias than A residues. On
average, dicot plant introns are ~41% U and 30% A, and runs of U residues, either uninterrupted or
interspersed with few A residues, are a common feature (reviewed in
4
,
17
,
18
,
45
). These U-rich stretches are usually found randomly distributed throughout the
intron sequence (but see below). Our results extend the findings of Lou
et al.
(
36
) who studied properties of the 3'ss-proximal AU-rich elements in the maize
Adh
1 gene intron expressed in tobacco cells. Replacement of uridines in these
elements by A residues decreased utilisation of the adjacent 3'ss and activated a cryptic 3'ss located upstream in the intron. Inactivation of two AU-rich elements, by either U to A or U to C mutations, had
more effect than did mutation of a single element, suggesting that individual
elements function cooperatively, which also appears to be the case for the U-rich islands studied in this work (Figs
2
and
3
). In addition, Carle-Urioste
et al
. (
46
) have recently identified an intra-intronic U-rich element, the integrity of which may be important for efficient
splicing of the
bronze
gene intron in maize cells.
Using another set of synthetic introns we have shown previously that replacement
of a central region in the AU-rich (50% U and 25% A) syn7 intron with an A-rich sequence still allows efficient splicing. This led to the
conclusion that A-rich sequences may be as efficient as U-rich sequences in activating splicing in
N.plumbaginifolia
protoplasts (
19
). In view of the results presented here, indicating that U-rich stretches interspersed with a few A residues (e.g
.
, sequences AAUUUAU or UUAUUAU; see Fig.
3
) are as effective in promoting splicing as the prototype U-rich island (UUUUUAU), it is likely that segments composed of Us and As,
still present in the synthetic A-rich intron, were responsible for its efficient splicing.
One of the most important findings of this work is that U-rich sequences can activate splicing irrespective of whether they are
inserted near the 5'ss or 3'ss or in the middle of the synGC intron (Fig.
2
). This indicates that U-rich islands in plant introns function differently from the metazoan
polypyrimidine tracts which are usually located proximal to the 3'ss and which are always downstream of the branch point region (
1
-
3
). At an early step of spliceosome assembly, the polypyrimidine tract is
recognised by the splicing factor U2AF, which facilitates binding of the U2
snRNP to the branch point region located upstream (
3
,
5
,
7
,
9
). Other proteins able to interact with the polypyrimidine tract have also been
characterised. They may participate in later stages of splicing or play a
regulatory role (
5
,
8
,
47
, and references therein). All available evidence indicates that U-islands in introns of the ClaU and MluU series activate splicing from
positions upstream of the branch point. The U-islands in the ClaU and MluU introns are separated from the 3'ss by a minimum of 76 and 52 nt, respectively. We have recently
demonstrated that A residues selected for branching during splicing
in vivo
of intron 3 of the
Arabidopsis
rubisco activase gene and of the intron syn7 are positioned 32 and 31 nt
upstream of the 3'ss, respectively (
21
). It is most probable that A
-31
is also selected for branching in the U-island-containing derivatives of synGC since these introns share many
structural features with syn7 including a branch point consensus of identical
sequence and location (CTAAC, positions -34/-30; see Fig.
1
B, and ref.
19
). Requirements for a minimal functional distance between the 5'ss and the branch point also argue in favour of branching in ClaU and
MluU introns taking place downstream of U-islands. The 5'ss and the branch point have to be separated by 45-50 nt in order to allow productive interaction of U1 and U2
snRNPs with vertebrate pre-mRNAs, and a similar distance appears to be essential for splicing in
plants (
33
, and refs therein). Branching to sequences upstream of U-islands would require that regions as short as 12 nt (introns of the ClaU
series) or 39 nt (introns of the MluU series) are competent to simultaneously
bind U1 and U2 snRNPs. Even if the U-islands themselves act as branch acceptors, a possibility we consider
unlikely, the distance between the 5'ss and the branch point in the ClaU introns would be too short to allow
splicing.
While single U-islands inserted at the
Cla
I and
Mlu
I sites can activate processing, a similar insertion at the
Sac
II site, close to the 3'ss, had no stimulatory effect. On the other hand, insertions of two or
three U-islands at the
Sac
II site stimulated splicing very strongly (Figs
2
and
3
). Inactivity of a single U-island at the
Sac
II site is rather unexpected since 3'ss-proximal regions (residues -21/-6) of plant introns are on average 6-8% more U-rich than regions further upstream (
4
). It is unlikely that the inactivity of a single U-island is due to its being too close to the acceptor AG since an identical
sequence inserted at positions -27/-21, 9 nt further upstream than in the intron SacU, also did not
activate splicing (unpublished result). It is possible that sequence
requirements for protein binding in the vicinity of the 3'ss are different from requirements at the 5'-proximal and central parts of the intron. Contrary to the
results obtained with nuclear extracts (Fig.
6
), different proteins may bind to U-rich sequences present at different intron positions
in vivo
.
Accessibility of U-rich sequences within the intron may also be important for determining
their activity. We have found (unpublished results) that the stimulatory effect
of U-rich islands on processing of synthetic introns is diminished by
insertions of complementary A-rich islands likely to result in the formation of hairpins which sequester
U-rich sequences. Formation of secondary structure might explain why some
introns do not undergo splicing in protoplasts despite the fact that they
contain one or two short stretches of Us [e.g., intron syn19 (
19
) or intron 1 of the human [beta]-globin gene (
20
)] which would be expected to act as functional U-rich islands.
UV cross-linking experiments have revealed that extracts prepared from nuclei of
N.plumbaginifolia
contain a relatively limited set of proteins which interact with natural and
synthetic pre-mRNAs
in vitro
. The two major cross-linkable plant proteins have apparent molecular mass of 50 and 54 kDa.
These two proteins, as well as a few others which are labelled less intensely
(e.g., 68 and 80 kDa proteins), appear to interact with pre-mRNAs which undergo splicing in protoplasts of
N.plumbaginifolia
(syn7, Leg and Phas RNAs) but not with RNAs which are not spliced in this
system (Waxy and h[beta] RNAs) or with the intron-less syn7/cDNA RNA (Figs
4
and
5
A). These observations, together with the finding that the 50 and 54 kDa
proteins have affinity for poly(U) (Fig.
5
B) and also interact with oligouridilates present in the U-island-containing derivatives of synGC (Fig.
6
), make these proteins good candidates for factors involved in intron
recognition in plants. We have recently cloned a cDNA encoding the 50 kDa
protein which cross-links to plant introns
in vitro
. The deduced protein sequence contains three RNP-type RNA binding domains (also known as RRM or RBD-CS domains;
34
,
44
) but it does not appear to be an equivalent of the characterised mammalian or
yeast hnRNP proteins (
34
,
35
). Neither does it represent the plant counterpart of the splicing factor U2AF,
for which a cDNA clone has also been isolated from
N.plumbaginifolia
(unpublished results, together with G. Simpson and C. Domon). Since the 50 kDa
protein is enriched in the nucleus (our unpublished results), it probably also
differs from the 50 kDa cytoplasmic poly(U) binding protein implicated in the
elicitor-induced destabilization of mRNA encoding a proline-rich protein in bean (
48
).
The profile of plant proteins identified by UV cross-linking in this work is considerably less complex than that seen with a
HeLa cell nuclear extract (Fig.
4
). More than 20 hnRNP proteins which associate with nascent pre-mRNAs in mammalian cell nuclei and which range in size from 34 to 120 kDa,
have been characterised (
34
,
35
,
44
,
49
). The exact functions of most of the hnRNP proteins are not known. Some of them
are implicated in constitutive or alternative splicing, while others may
participate in other processing reactions in the nucleus or in RNA transport
(reviewed in
34
,
44
). Several hnRNP-like proteins have also been recently characterised in the yeast
S.cerevisiae
(reviewed in
35
). Although our results show that only a small number of proteins present in
plant nuclear extracts are efficiently cross-linked to pre-mRNA fragments
in vitro
, they do not eliminate the possibility that plant cell nuclei contain
additional hnRNP-like proteins which are either less abundant or less readily cross-linkable to the substrates used in this work.
Evidence that the 50 and 54 kDa proteins interact with AU-rich or U-rich intron sequences
in vivo
is still missing, but it is interesting to speculate how such sequences and
proteins interacting with them could contribute to the processing of plant
introns. U-islands are usually randomly distributed along the intron and proteins
binding to them could delineate sequences to be excised as introns. The bound
proteins could assist U-snRNPs or other splicing factors in identifying the splice sites at the
transition regions between the AU-rich and GC-rich sequences. The 50 and 54 kDa proteins may represent specialised
sequence-specific hnRNP proteins which, as already proposed for some mammalian
hnRNP proteins (
34
,
35
), maintain discrete regions within pre-mRNAs in a conformation suitable for interactions with RNA splicing
factors.
We thank J. Petruska for excellent technical assistance and Drs T. Hohn, P. King
and H. Rothnie for critical reading of the manuscript.
REFERENCES
Return