ABSTRACT
We have established an
in vitro
reconstitution/splicing complementation system which has allowed the
investigation of the role of mammalian U1 snRNP components both in splicing and
at the early stages of spliceosome formation. U1 snRNPs reconstituted from
purified, native snRNP proteins and either authentic or
in vitro
transcribed U1 snRNA restored both early (E) splicing complex formation and
splicing activity to U1-depleted extracts.
In vitro
reconstituted U1 snRNPs possessing an m
3
G or ApppG cap were equally active in splicing, demonstrating that a
physiological cap structure is not absolutely required for U1 function.
However, the presence of an m
7
GpppG or GpppG cap was deleterious to splicing, most likely due to competition
for the m
7
G cap binding proteins. No significant reduction in splicing or E complex
formation was detected with U1 snRNPs reconstituted from U1 snRNA lacking the
RNA binding sites of the U1-70K or U1-A protein (i.e., stem-loop I and II, respectively). Complementation
studies with purified HeLa U1 snRNPs lacking subsets of the U1-specific proteins demonstrated a role for the U1-C, but not U1-A, protein in the formation and/or stabilization of early
splicing complexes. Studies with recombinant U1-C protein mutants indicated that the N-terminal domain of U1-C is necessary and sufficient for the stimulation of E complex
formation.
Nuclear pre-mRNA splicing requires the formation of a large ribonucleoprotein complex,
the spliceosome, wherein the catalysis of the two sequential transesterification reactions responsible for intron
removal and exon ligation occurs. Spliceosomes are formed by the ordered
interaction of numerous splicing factors and the four small nuclear
ribonucleoproteins (snRNPs), U1, U2, U5 and U4/U6, with conserved regions of
the pre-mRNA (reviewed in
1
). One of the initial contacts with the pre-mRNA in the spliceosome assembly pathway is the binding of U1 snRNP to the
5' splice site. This interaction involves base pairing between the 5' end of the U1 snRNA and conserved sequences spanning the 5' splice site (
2
,
3
). A stable U1 snRNP/pre-mRNA complex, referred to as the commitment complex, was first identified
in yeast (
4
,
5
). A similar complex, designated the early or E complex, was subsequently
identified in mammalian splicing extracts by gel filtration (
6
,
7
). Subsequent to E complex assembly, the U2 snRNP interacts with the branch site
of the intron, thereby forming the so-called pre-spliceosome. Mature spliceosomes are ultimately formed by the
interaction of the U4/U6 and U5 snRNPs, in the form of a pre-assembled [U4/U6.U5] tri-snRNP complex (reviewed in
1
).
The spliceosomal snRNPs consist of one snRNA molecule (or two in the case of
U4/U6) complexed with eight so-called Sm or core proteins (B, B', D1, D2, D3, E, F, G), which are present in all snRNP species, and
a number of particle-specific proteins (reviewed in
8
). The snRNP core proteins interact with an evolutionarily conserved region of
the U1, U2, U4 and U5 snRNAs, the Sm site, which consists of a single-stranded, uridylic acid-rich region that is flanked by two stem-loop structures (
9
). The association of the snRNP Sm proteins results in the hypermethylation of
the snRNA's monomethylguanosine (m
7
G) 5' cap structure to a 2,2,7-trimethylguanosine (m
3
G) form (
10
). The m
3
G cap, together with the snRNP Sm proteins, forms the karyophilic signal
required for the nuclear import of the spliceosomal snRNPs (
11
,
12
).
The mammalian U1 snRNP contains, in addition to U1 snRNA and the common snRNP
proteins, three U1-specific proteins denoted 70K, A and C. The U1-70K and U1-A proteins possess highly conserved RNA binding domains (RBDs)
characteristic of a number of RNA binding proteins (
13
,
14
). These proteins are thus able to interact directly with the U1 snRNA;
specifically, U1-70K binds stem-loop I of the U1 snRNA and U1-A interacts with stem-loop II (
15
-
17
). Protein-protein contacts also appear to contribute to the association of U1-70K and U1-A with the U1 particle. For example, a stable interaction
between a U1-70K deletion mutant containing the N-terminal 97 amino acids, which do not bind U1 snRNA, and U1 snRNPs
containing only the core snRNP proteins has been demonstrated (
18
). Recent studies with SNP1, the
Saccharomyces cerevisiae
U1-70K homolog, indicate that contacts between 70K and other U1 proteins are
sufficient for the formation of functional U1 snRNPs (
19
). The 70K protein also exhibits structural features, including regions rich in
serine and arginine residues, characteristic of the SR family of essential
splicing factors (reviewed in
20
).
In contrast to U1-70K and U1-A, the U1-specific C protein does not contain an RBD and its association
with the U1 snRNP appears to be mediated primarily by protein-protein
interactions (
21
,
22
). Binding studies carried out with U1-C deletion mutants have shown that the N-terminal 45 amino acids suffice for its association with the U1
snRNP (
22
). Interestingly, this region of the U1-C protein contains a zinc finger-like motif which appears to be essential for binding, since point
mutations in the cysteine and histidine residues of this putative zinc finger
abolish U1-C binding (
22
). The association of U1-C with the U1 snRNP requires the presence of the N-terminal region of the 70K protein, as well as one or more Sm
protein (
18
). More recent
in vitro
studies have demonstrated that the U1-C protein can form homodimers; U1-C dimerization also requires amino acids located in its N-terminal domain (
23
).
The U1 snRNP appears to function primarily during the early steps of splicing
complex formation. Its main role is the recognition of the 5' splice site which appears to be a nucleation event for spliceosome
assembly. While U1 snRNP function relies heavily upon base pairing interactions
of the U1 snRNA, protein components have also been shown to contribute to its
activity. A general role for U1-specific proteins was initially suggested by splicing complementation
studies in
Xenopus
oocytes using mutant U1 snRNAs; several mutants which did not support the
stable association of U1-specific proteins were unable to restore splicing activity to oocytes that
had been depleted of their endogeneous U1 snRNA (
24
). The first indication that a stable U1/5' splice site interaction is mediated by U1 snRNP proteins was provided by
studies demonstrating that mild proteolysis of the mammalian U1 particle
inhibits its association with the 5' splice site (
25
). Consistent with this observation, filter binding and gel mobility shift
assays have provided evidence that the U1-C protein can augment the binding of the U1 snRNP to the 5' splice site (
26
,
27
). The splicing factor ASF/SF2 has also been shown to enhance the interaction of
U1 with the 5' splice site; this enhancement appears to involve an interaction between
the SR domain of ASF/SF2 and that of the U1-70K protein (
28
).
The bulk of functional information regarding the mammalian U1-specific proteins has been obtained by studies carried out with highly
purified components in the absence of splicing extract. Here we have
investigated the function of the mammalian U1-specific proteins in both splicing and early splicing complex formation,
using U1-depleted splicing extracts and either
in vitro
reconstituted or biochemically purified U1 snRNPs. Complementation studies
performed with purified U1 snRNPs lacking one or more of the U1-specific proteins demonstrated that the U1-C protein, but not U1-A, enhances the formation of early spliceosomal complexes.
Mutagenesis experiments indicated that the N-terminal 60 amino acids of the U1-C protein are sufficient for this enhancement.
Nuclear extracts were prepared from HeLa cells (Computer Cell Culture Center,
Mons) as described (
29
). Native U1 snRNPs or U1 snRNPs specifically lacking either the U1-A ([Delta]A), U1-A and -C ([Delta]A,C) or U1-A -C, and -70K proteins (core U1 snRNPs)
were isolated from HeLa nuclear extract by anti-m
3
G immunoaffinity chromatography followed by Mono Q chromatography (
30
). Core U1, [Delta]A, and [Delta]A,C particles contained maximally 5% of each of the depleted U1-specific proteins. Native, RNA-free snRNP proteins (TPs) were isolated from a mixture of
immunoaffinity purified U1, U2, U5 and U4/U6 snRNPs by dissociation in the
presence of EDTA and the anion exchange resin DE53 (
31
). HeLa U1-A and U1-C proteins were isolated from native snRNP proteins by Mono S
chromatography as previously described (
18
). Recombinant his-tagged U1-C proteins were constructed and purified as described previously (
23
). The substitution mutant (s28/29) was constructed in essentially the same
manner.
HeLa U1 and U2 snRNAs were isolated from purified snRNPs as described previously
(
31
).
In vitro
transcribed human U1 snRNA was prepared from
Pst
I linearized pHU1a (
32
).
Wild-type and mutant
Xenopus
U1 snRNAs were prepared by
in vitro
transcription of
Bam
HI linearized plasmids with T7 polymerase (
15
). U1 snRNAs were transcribed in the presence of 1 mM chemically synthesized GpppG, ApppG, m
7
GpppG or m
3
GpppG (as indicated) and 0.1 mM GTP and were purified by denaturing polyacrylamide gel electrophoresis. Radiolabeled MINX (
33
) and pSP62[Delta]i (
34
) pre-mRNA, with a specific activity of 2.5 * 10
6
c.p.m./pmol, were transcribed
in vitro
in the presence of GpppG essentially as described (
35
). 5' and 3' MINX pre-mRNAs, with a specific activity of 6.25 * 10
6
c.p.m./pmol, were transcribed from
Hin
dIII linearized pMINX and
Bam
HI linearized p3'MINX. p3'MINX was generated by isolating a
Hin
dIII-
Bam
HI fragment from pMINX and subcloning into pGEM4.
U1-depleted nuclear extract was prepared by affinity selection with a biotinylated 2'-
O
-methyl RNA oligonucleotide and streptavidin-agarose beads (
36
). An oligonucleotide complementary to U1 snRNA nucleotides 1-13 with the following sequence was used: 5'-GCCAGGUAAGUAUdC*dC*dC*dC*dT-3' (where dC* denotes a biotinylated 2'-deoxycytidine and A, U, G and C
represent 2'-
O
-methyl-ribonucleotides). Titration experiments and Western blotting indicated that >97.5% of the U1 snRNA, and 95% of each of the U1-specific proteins A, C and 70K, had been removed from the U1-depleted extract. Splicing reactions (12.5 [mu]l) were incubated for 60 min at 30oC and typically contained 35% extract, 45
mM KCl, 2.5 mM MgCl
2
, 2.0 mM ATP, 10 mM creatine phosphate and 12 fmol
32
P-labeled pSP62[Delta]i pre-mRNA (3 * 10
4
c.p.m.). For complementation of U1-depleted extract, 0.6 pmol (200 ng) of Mono Q purified HeLa U1 snRNP were
added directly to the reaction mixture. Complementation with
in vitro
reconstituted particles was accomplished by combining 1.8 pmol (100 ng) of authentic or
in vitro
transcribed U1 snRNA and 3.8 pmol (750 ng) of purified native snRNP proteins. This mixture was incubated for 30
min at 30oC in the presence of splicing reactions lacking pre-mRNA, and splicing was initiated by the addition of the pre-mRNA. No differences in complementation efficiency were
observed when reconstitution was carried out either directly in splicing
extract or by additionally pre-incubating in the absence of extract. Splicing intermediates and products
were isolated by phenol/chloroform extraction followed by ethanol precipitation
and analysed on 14.0% polyacrylamide-7 M urea gels.
E complex formation was assayed by gel filtration on Sephacryl S-500 (
6
). Splicing extracts were initially depleted of ATP and magnesium by dialysis
and subsequent incubation at room temperature for 30 min. Under these
conditions no A complex formation was observed by native gel electrophoresis after a 30 min incubation at 30oC. Standard E complex reactions (75 [mu]l) contained 30% splicing extract, 45 mM KCl and 0.18 pmol (4.5 * 10
5
c.p.m.) of
32
P-labeled MINX pre-mRNA and were incubated at 30oC for 25 min. Complexes were fractionated on a 60 * 0.9 cm column at a flow rate of 10 ml/h, and the amount
of radioactivity present in 350 [mu]l fractions was determined by Cherenkov counting. Assays performed with 5' or 3' MINX pre-mRNA contained 72 fmoles (4.5 * 10
5
c.p.m.) of radiolabeled transcript. For complementation of U1-depleted extracts, 4 pmol of the indicated Mono Q purified U1 snRNPs were
added either directly to the reaction mixture or after a 30 min incubation on
ice with 20 pmol of Mono S purified or recombinant HeLa U1-A or U1-C protein, as indicated. For
in vitro
reconstitution studies, 600 ng of U1 snRNA and 3.75 [mu]g native snRNP proteins (in a total of 3 [mu]l) were incubated for 30 min at 30oC prior to incubation with extract.
As a potential means to investigate the function of the mammalian U1-specific proteins in pre-mRNA splicing, we established an
in vitro
reconstitution/splicing complementation system for HeLa U1 snRNPs (depicted
schematically in Fig.
1
). HeLa nuclear extracts specifically depleted of U1 snRNPs were prepared by
affinity selection with a biotinylated 2'-
O
-methyl RNA oligonucleotide complementary to nucleotides 1-13 of the U1 snRNA. Mock-depleted extract was handled in an identical manner, except
that oligonucleotide was omitted. Reconstitution of U1 snRNPs was carried out
by incubating purified U1 snRNA and native snRNP proteins (TPs) in the presence
of splicing extract. TPs, which are essentially free of any snRNA, consist of
the snRNP Sm proteins, B, B', D1, D2, D3, E, F and G, and the U1-specific proteins A and C, but only trace amounts of U1-70K (
31
,
37
). Since the association of U1-C is strictly dependent on the presence of the U1-70K protein (
18
),
in vitro
reconstituted U1 snRNPs thus consist predominantly of the snRNP Sm proteins and
the U1-A protein. Due to low levels of the U1-70K protein in TPs and U1-depleted extract, a small amount of wild-type U1 snRNP is also reconstituted (data not shown).
Splicing of an adenovirus major late II pre-mRNA (pSP62[Delta]i) was significantly reduced in U1-depleted extract (Fig.
2
, lanes 1-2). However, the addition of a physiological amount (200 ng) of highly
purified U1 snRNP restored splicing to mock extract levels (Fig.
2
, lane 3), demonstrating that the observed reduction in activity is due
specifically to the absence of U1 snRNPs. Splicing was also restored if 2 pmol
of purified HeLa U1 snRNA (a 2-fold excess as compared to the mock extract level) was added (Fig.
2
, lane 5). U2 snRNA, on the other hand, did not complement splicing (lanes 6-7), indicating that this effect is specific for U1 snRNA. Splicing
complementation was, however, significantly enhanced, especially at the lower
U1 snRNA concentration, if native snRNP proteins were added to the
reconstitution mixture (compare lanes 4-5 with 8-9). TPs alone had little or no effect on the splicing activity of
U1-depleted extract (Fig.
3
, lane 23). Since the extent of U1 snRNA degradation was significantly reduced when TPs were present during reconstitution and
in vitro
splicing, the TP-induced enhancement of splicing complementation appeared to be primarily
the result of increased U1 snRNA stability (data not shown). In sum, splicing
could be restored by the addition of purified U1 snRNA alone, but even more
efficiently by the combination of U1 snRNA and purified snRNP proteins.
We next tested whether the splicing activity of U1-depleted extract could be restored by
in vitro
transcribed U1 snRNA. Concurrently, the role in splicing of the U1 snRNA 5' cap was investigated by comparing
in vitro
transcribed RNAs possessing various cap structures. Synthetic HeLa U1 snRNA
containing an m
3
G or non-physiological ApppG cap (either alone or with TPs) restored splicing to U1-depleted extract as efficiently as U1 snRNA isolated from HeLa U1
snRNPs (compare lanes 13-14, 17-18 and 19-20). In contrast, U1 snRNAs containing either an m
7
G or GpppG cap not only were unable to complement splicing (lanes 15-16 and 21-22), but also, in the absence of TPs, inhibited the activity of
the mock extract as well as the residual splicing activity of the U1-depleted extract (compare lane 1 with lanes 3 and 9, and lane 12 with
lanes 15 and 21). Since the stability of the m
7
G- and GpppG-capped U1 snRNAs was similar to that of the m
3
G- and ApppG-capped ones (data not shown), the decreased complementation
efficiency of these RNAs cannot be attributed to an increase in their turnover.
Rather, the latter result is consistent with previous data demonstrating that
short RNAs possessing an m
7
G or GpppG cap can inhibit splicing by competing for proteins (i.e., CBP20 and
CBP80) which normally bind the m
7
G cap of the pre-mRNA (
38
). Thus, although there is no absolute requirement for an m
3
G cap, not all cap structures are compatible with U1 snRNP splicing activity
in vitro
.
The ability to complement splicing with
in vitro
transcribed U1 snRNA allowed us to investigate the effect of U1 snRNA mutation on the activity of
in vitro
reconstituted U1 snRNPs. Since we were particularly interested in investigating the function of the U1-specific proteins, we tested the activity of
Xenopus
U1 snRNA mutants which lacked the RNA binding sites of the U1-70K and U1-A protein (mutants A and B, respectively). As shown in Figure
4
,
in vitro
transcribed wild-type
Xenopus
U1 snRNA restored splicing activity to U1-depleted extract (lanes 6-7), albeit slightly less efficiently than U1 snRNA isolated from
HeLa U1 snRNPs (lanes 4-5). As compared to wild-type, deletion of stem-loop I (mutant A) or stem-loop II (mutant B)
had no significant effect on the complementation efficiency of
in vitro
reconstituted U1 snRNPs (Fig.
4
, lanes 8-11). Thus, reconstitution of U1 snRNPs active in splicing does not
require the presence of stem-loop I or II. This in turn suggests that either U1-70K and U1-A are dispensible for U1 snRNP function
in vitro
or, alternatively, that these proteins can functionally associate with the U1
snRNP by other means (e.g. by protein-protein interactions).
Since substoichiometric amounts of the U1 snRNP may suffice for the complete
restoration of splicing activity, we reasoned that alterations in the U1 snRNP
that affect its function might be more readily detected by an assay which
directly measures the amount of functional snRNPs present. We thus analyzed the
formation of the earliest detectable functional splicing complex, the E
complex, by gel filtration (
6
). Due to inefficient splicing complex formation with SP62[Delta]i transcripts (data not shown), E complex assays were performed with MINX
pre-mRNA, which is also a derivative of the adenovirus major late II
transcript. As compared to the mock-depleted extract, E complex formation was severely reduced in U1-depleted extract; predominantly pre-mRNA/ hnRNP protein complexes (i.e., H complexes) were formed
(Fig.
5
A). The addition of physiological amounts of highly purified U1 snRNPs shifted
the gel filtration profile of the U1-depleted extract to that of the mock extract (Fig.
5
A), indicating that the reduction in early splicing complex formation is
specifically due to the absence of U1 snRNPs. E complex formation could also be
partially restored by the addition of U1 snRNPs reconstituted from TPs and
either authentic HeLa or
in vitro
transcribed
Xenopus
U1 snRNA (Fig.
5
B). The addition of
in vitro
transcribed
Xenopus
U1 snRNA or TPs alone had little or no effect on complex formation (data not
shown).
The ratio of E to H complex did not change significantly, as compared to wild-type
Xenopus
U1 snRNA, when complementation was performed with reconstituted U1 snRNPs
lacking stem-loop I or II (mutants A and B; Fig.
5
C). These results thus provide additional evidence that functional U1 snRNPs can
be reconstituted even if the U1 snRNA binding sites for the U1-70K and U1-A protein are deleted. Deletion of the 5' end of the U1 snRNA, on the other hand, led to a marked
decrease in E complex formation (data not shown). The latter is consistent with
the previous observation that deletion of the 5' splice site leads to a dramatic reduction in E complex formation (
7
) and supports the idea that the early splicing complexes which we detect are
indeed functional ones.
Previous studies have demonstrated that specific ATP-independent complexes, denoted E5' and E3', assemble on RNAs containing only the 5' or 3' portion of a pre-mRNA (
7
). E5' complexes contain predominantly U1 snRNPs, whereas E3' complexes are enriched in the splicing factor U2AF (
7
). To test whether the U1-C protein acts at the 5' or 3' splice site, RNAs consisting of the 5' or 3' half of the MINX pre-mRNA were prepared and their ability to
form E5' and E3' complexes, respectively, was tested in mock and U1-depleted extract. Consistent with the known composition of
E3' complexes (i.e., they are for the most part devoid of U1 snRNPs), no
significant difference in E3' assembly was detected in mock versus U1-depleted extract (data not shown). In contrast, with the 5' MINX transcript, an equal amount of E5' and H complex was observed with mock-depleted extract, but essentially only H
complex with U1-depleted extract (Fig.
7
A). As observed with full-length MINX, the addition of physiological amounts of either wild-type or [Delta]A U1 snRNPs shifted the gel filtration profile of U1-depleted extract to that of the mock-depleted extract (Fig.
7
A and B). Addition of [Delta]A,C U1 snRNPs, on the other hand, resulted in only a partial shift (Fig.
7
B). However, pre-incubation of [Delta]A,C U1 snRNPs with the U1-C protein enhanced E5' assembly such that the levels of E5' and H complex were indistinguishable from
those of the mock-depleted extract (Fig.
7
C). In contrast, addition of the U1-A protein had no effect on the complementation activity of [Delta]A,C U1 snRNPs (Fig.
7
C). Moreover, addition of the U1-C protein alone also had no significant effect on E5' complex formation (data not shown). Thus, the U1-C protein stimulates interactions occurring on the 5' half of the MINX pre-mRNA, namely those between the U1 snRNP and the
5' splice site, even in the absence of a 3' splice site.
To determine whether distinct regions of the U1-C protein are necessary and/or sufficient for the stimulation of E complex
formation, we tested the activity of several HeLa U1-C mutants. To this end, histidine-tagged U1-C deletion and substitution mutants were constructed and over-expressed in
E.coli
(Fig.
8
A). While the addition of [Delta]A,C U1 snRNPs alone only partially enhanced E complex formation in U1-depleted extract (Fig.
8
B), preincubation with wild-type recombinant U1-C protein led to a significant increase in the ratio of E to H
complex formed (Fig.
8
B). Surprisingly, deletion of the C-terminal 99 amino acids of the U1-C protein did not significantly reduce complementation efficiency,
indicating that the N-terminal 60 amino acids are sufficient for activity. In contrast, deletion
of amino acids 1-29 abolished the ability of the U1-C protein to enhance E complex formation (Fig.
8
B). These results thus demonstrate that residues within the first 29 amino acids
of the U1-C protein are essential for its activity. We next tested the activity of
two point mutants, (s25) and (s28/29), which exhibit both reduced binding and
dimerization activity (
23
, data not shown). As shown in Figure
8
C, substitution of the cysteine at position 25 with a serine led to only a
slight decrease in the ratio of E to H complex, while substitution of arginine
and lysine at positions 28 and 29 with glycine and serine, respectively,
abolished the ability of U1-C to enhance E complex formation. These results thus indicate that the
latter amino acids play an important role in the U1-C protein-mediated augmentation of E complex assembly.
We have established an
in vitro
reconstitution/splicing complementation system for HeLa U1 snRNPs which should
facilitate future investigation of both structural and functional aspects of
the U1 snRNP. Reconstitution/complementation systems have thus now been
described for each of the mammalian spliceosomal snRNPs and a number of
comparisons can be made among them (
37
,
39
,
40
). Restoration of the splicing activity of U1-depleted extracts could be achieved by the addition of an excess of
purified U1 snRNA alone (Fig.
2
), suggesting that sufficient amounts of those proteins required for the
assembly of functional U1 snRNPs are present in U1-depleted extracts. While similar results were previously reported for U4-depleted extracts (
39
), U2- or U5- depleted extracts could not be complemented by the addition of U2
or U5 snRNA alone. Rather, reconstitution of functional U2 or U5 snRNPs
required the addition of purified snRNP Sm proteins (
37
). In this respect, it is interesting to note that U1 and U4/U6 snRNPs, which
contain only three and two particle-specific proteins, respectively, are relatively simple RNP complexes as
compared to U2 and U5 snRNPs which contain eleven and nine particle-specific proteins, respectively. Functional reconstitution of the former
may thus be more readily achieved in the presence of relatively low levels of
Sm proteins or, alternatively, may not be strictly dependent on their presence
(
39
).
U1 snRNPs reconstituted from
in vitro
transcribed U1 snRNA were as active in E complex formation and splicing as
those reconstituted from U1 snRNA that had been isolated from U1 snRNPs (Figs
3
and
5
). Since
in vitro
transcribed U1 is devoid of modified internal nucleotides, and it is unlikely
that modification occurs during reconstitution and
in vitro
splicing, the three pseudouridines and two 2'-
O
-methylated nucleotides normally present at the 5' end of U1 snRNA appear to be dispensible for U1 snRNP function in
splicing. The ability to restore splicing to U1-depleted extract with synthetic U1 snRNA indicates that it may be possible
to reconstitute functional U1 snRNPs containing photoactivateable nucleosides
(e.g. 4-thiouridine). The
in vitro
reconstitution/splicing complementation system described here could thus
potentially be used to study the interactions of the U1 snRNP with other
components of the splicing reaction and thereby further our understanding of
the three-dimensional architecture of the spliceosome.
As previously reported for the U5 snRNP (
37
), a modified 5' cap structure (i.e., m
3
G) was also dispensible for the activity of U1 snRNPs in
in vitro
splicing (Fig.
3
). However, apparently due to competition for the m
7
G cap-binding proteins CBP20 and CBP80, the presence of an m
7
G or GpppG cap led to inhibition of the splicing reaction. Since we had
previously reported that m
7
G-capped U5 snRNPs could restore splicing activity to a U5-depleted extract (
37
), and the amount of m
7
G-capped U5 snRNA used was only 2-fold less than the amount of m
7
G-capped U1 snRNA used here, we initially considered whether this inhibitory
effect might be specific for the U1 snRNP. A direct comparison of the
inhibitory effects of m
7
G-capped U1 and U5 snRNPs on
in vitro
splicing indicated that both compete at a comparable level for the binding of
the cap-binding proteins (data not shown). The detrimental effect of m
7
G-capped snRNPs on splicing thus appears to be more pronounced in U1-depleted extracts. Since the cap-binding proteins have also been shown to stimulate E complex
formation (
41
,
42
), it is perhaps not surprising that the effects of competition for these
proteins are more readily detected in a system where E complex assembly is
severely compromised due to the decreased level of U1 snRNPs. Alternatively,
since CBP20 and CBP80 have been proposed to interact either directly or
indirectly with components of the U1 snRNP (
41
), it is conceivable that they have been partially co-depleted with the U1 snRNP. The deleterious effect of m
7
G-capped U1 or U5 snRNPs on splicing suggests that the requirement of an m
3
G cap for the import of the spliceosomal snRNPs into the nucleus may have
evolved, at least in part, to prevent the accumulation of m
7
G-capped snRNPs in the nucleus.
Whereas
in vitro
splicing activity could be restored by U1 snRNA alone, complementation of E
complex assembly required both U1 snRNA and TPs. This result could be explained
by differences in the nature of these two assays. In contrast to
in vitro
splicing, the E complex assay is a binding assay and, therefore, should
directly reflect the amount of functional U1 snRNP that is reconstituted. Since
E complex formation may not be rate limiting for
in vitro
splicing, the reconstitution of small amounts of functional U1 snRNP, while
having little impact on the overall level of E complex formation, may suffice
for the complete restoration of splicing. In addition, the amount of E complex
detected by gel filtration appears to be very dependent upon the stability of
the complexes which are formed (
7
). Structural alterations in the U1 snRNP which reduce the amount of E complex
detected by gel filtration (e.g., the removal of the U1-C protein), could thus primarily influence the stability, rather than the
assembly, of these complexes. Thus, not only quantitative, but also qualitative
differences in the U1 snRNP should be more readily apparent in the E complex
assay. This could also explain why, in contrast to
in vitro
splicing, E complex formation was only partially restored by
in vitro
reconstituted U1 snRNPs (Fig.
5
B). It is at present not clear, however, whether this partial complementation is
due to the reconstitution of insufficient amounts of U1 snRNPs or,
alternatively, to the formation of particles that are unable to support stable
E complex formation, for example due to limiting amounts of the U1-70K protein which would limit binding of the U1-C protein.
Complementation studies with U1 snRNPs reconstituted from U1 snRNA deletion
mutants indicated that functional U1 snRNPs can be formed even after extensive
mutagenesis of the U1 snRNA, including deletion of stem-loop I or II (Figs
4
and
5
). These results are somewhat surprising given the fact that these mutations
significantly inhibited the ability of U1 snRNPs to complement splicing in
Xenopus
oocytes whose endogeneous U1 snRNA had been inactivated by oligonucleotide-directed RNase H cleavage (
24
). The basis for this difference is not clear, but may simply reflect
differences in the assay systems employed. The ability of U1 snRNPs lacking
stem-loop I or II to complement splicing suggests, at first glance, that U1-A and U1-70K are dispensible for U1 snRNP function. However, there is a
significant amount of evidence suggesting that these proteins can stably
associate with the U1 snRNP solely via protein-protein interactions (
16
,
18
,
24
). In addition, recent studies carried out in
S.cerevisiae
indicated that contacts between the yeast 70K homologue and other U1 snRNP
proteins are sufficient for the assembly of a functional U1 snRNP particle (
19
). Our results are thus consistent with the idea that, despite the removal of
their primary RNA binding site, the U1-70K and U1-A proteins can stably and functionally associate with the U1 snRNP.
Evidence that the U1-C, but not U1-A, protein can enhance E complex formation and/or stability was
provided by complementation studies with biochemically purified U1 snRNPs
lacking one or more of the U1-specific proteins (Figs
6
and
7
). These results are consistent with previous studies, performed in the absence
of nuclear extract,which indicated that the U1-C protein enhances the interaction of the U1 snRNP with the 5' splice site (
26
,
27
). The U1-C protein could enhance E complex formation by interacting directly with
the 5' splice site. Indeed, it has recently been shown that U1-C can be crosslinked to a short oligonucleotide containing a 5' splice site (
43
). Alternatively, the binding and dimerization of the U1-C protein could indirectly enhance interactions with the pre-mRNA by affecting the overall structure of the U1 snRNP. The reduced
activity of [Delta]A,C,70K (core) U1 snRNPs as compared to [Delta]A,C U1 snRNPs further suggested that the U1-70K protein also contributes to the formation and/or
stability of E complexes. Consistent with this hypothesis, a stable, ASF/SF2-mediated interaction between U1 and the 5' splice site appeared to involve an interaction between the U1-70K protein and ASF/SF2 (
27
,
28
). However, since U1-C protein binding is dependent on the U1-70K protein, the reduced activity of core U1 snRNPs as compared to [Delta]A,C U1 snRNPs could also potentially result from a decrease
in the amount of U1-C association. Since neither U1 snRNPs specifically depleted of the U1-70K protein nor sufficient amounts of purified U1-70K protein were available, it was not possible to directly
test the effect of this U1-specific protein on E complex formation.
Mutagenesis of the U1-C protein provided strong evidence that its N-terminal domain is necessary and sufficient for its function during
the early stages of spliceosome assembly (Fig.
8
). A functional role for the N-terminal but not C-terminal domain in E complex formation is consistent with recent
studies demonstrating that solely amino acids in the N-terminal domain of the protein have been evolutionarily conserved from
yeast to man (
44
). However, since we have only measured effects on E complex formation, we
cannot rule out that the methionine/proline-rich C-terminal domain is involved in subsequent steps of the splicing
process. The U1-C N-terminus contains residues essential for the interaction of U1-C with the U1 particle, as well as those required for
dimerization (
22
,
23
). Due to the inability to overexpress sufficient amounts of point mutants which
are solely deficient in U1 snRNP association, we cannot conclude at present
whether a stable U1-C/U1 snRNP interaction is essential for the promotion of E complex
assembly. However, given that the U1-C protein enhances the interaction of U1 with the 5' splice site, it is reasonable to assume that the stabilization of
this interaction requires the stable association of U1-C with the U1 snRNP. Complementation studies performed with point mutants
demonstrated that the arginine and lysine present at positions 28 and 29,
respectively, are essential for U1-C activity; substitution of the cysteine at position 25 with serine, on
the other hand, only slightly reduced its ability to stimulate E complex
formation. Since both mutations have been reported to diminish, but not abolish
U1-C dimerization (
23
, data not shown), these results suggest that dimerization, while not absolutely
essential, may still contribute to the activity of the U1-C protein during E complex formation. However, since these mutants also
exhibit reduced binding (see Fig.
8
), we cannot exclude that their phenotypes reflect, at least in part, their
decreased ability to interact with the U1 particle. A more detailed mutational
analysis of the U1-C protein should, in the future, allow us to more precisely define those
amino acids essential to U1-C function. This system should thus allow a finer examination of those
factors influencing the formation of the first functionally important complex
in the spliceosome assembly pathway.
We thank Iain Mattaj for kindly providing the wild-type and mutant
Xenopus
U1 snRNA plasmids, as well as Edward Darzynkiewicz for providing m
3
G cap. We also thank Silke Börner for excellent technical assistance, Christopher Marshallsay and Veronica Raker for constructive comments on the manuscript, and Véronique Ségault for helpful advice and discussions. We are grateful
to Michael Krause for preparing biotinylated 2'-
O
-methyl oligonucleotides and Gabi Plessel for help in constructing p3'MINX. This work was supported by the Deutsche
Forschungsgemeinschaft (SFB 272/A3), the Netherlands Foundation for Chemical
Reasearch (SON) with financial aid from the Netherlands Organization for
Scientific Research (NWO) and European Community grants ERBCHRXCT 930191 and
930176.
REFERENCES
Return