y W. Nilsen*
Center for RNA Molecular Biology
Department of Molecular Biology and Microbiology
Case Western Reserve University School of Medicine
to produce complex B. Upon tri-snRNP addition, an
10900 Euclid Avenue
intricate series of RNA-RNA rearrangements ensues, re-
Cleveland, Ohio 44106
sulting in the formation of the catalytically competent
complex C. Key steps in formation of the catalytically
active spliceosome include pairing of the 5 splice site
Summary
with U6 snRNA (an interaction that is mutually exclusive
with the U1/5 splice site pairing) and formation of U2/
A sensitive assay based on competition between cis-
U6 base-pairing interactions at the expense of U4/U6
pairing (Nilsen, 1998; Staley and Guthrie, 1998).
and trans-splicing suggested that factors in addition
While this canonical view of spliceosome assembly
to U1 snRNP were important for early 5 splice site
has proven to be reliable and predictive, there exist
recognition. Cross-linking and physical protection ex-
notable exceptions to the obligatory order of addition
periments revealed a functionally important interac-
of some components. For example, U2 snRNP can spe-
tion between U4/U6.U5 tri-snRNP and the 5 splice
cifically bind to the branch point region in certain mole-
site, which unexpectedly was not dependent upon
cules that lack a 5 splice site (e.g., Ruskin and Green,
prior binding of U2 snRNP to the branch point. The
1985; Chiara and Reed, 1995). In one well-established
early 5 splice site/tri-snRNP interaction requires ATP,
pathway, 5 splice site-independent recruitment of U2
occurs in both nematode and HeLa cell extracts, and
snRNP is mediated by exonic enhancer elements lo-
involves sequence-specific interactions between the
cated in the 3 exon (e.g., Lavigueur et al., 1993; Wang
highly conserved splicing factor Prp8 and the 5 splice
et al., 1995; Zuo and Maniatis, 1996). There are similar
site. We propose that U1 and U5 snRNPs functionally
exceptions to the order of assembly described above
collaborate to recognize and define the 5 splice site
that involve 5 splice site recognition. In a comprehen-
prior to establishment of communication with the 3
sive series of experiments, Konarska and colleagues
splice site.
have shown that oligoribonucleotides mimicking 5
splice sites can engage U6 snRNP directly, effectively
bypassing a role for U1 snRNP (Konforti et al., 1993;
Introduction
Konforti and Konarska, 1994, 1995). Indeed, productive
use of such oligonucleotides in splicing is observed only
Splicing of nuclear pre-mRNAs occurs in a large ribo-
when the 5 end of U1 snRNP is sequestered (Konforti
nucleoprotein complex known as the spliceosome,
and Konarska, 1995). A similar bypass of U1 snRNP
which is comprised of five snRNPs and a large number
function, thought to be mediated by direct engagement
( 50) of non-snRNP-associated proteins (reviewed in
of U6 snRNA, is observed when certain pre-mRNAs are
Burge et al., 1999; Staley and Guthrie, 1998). Unlike the
spliced in U1 snRNP inactivated or depleted extracts
ribosome, a ribonucleoprotein machine of comparable
(Tarn and Steitz, 1994, 1995; Crispino and Sharp, 1995;
size and complexity, the spliceosome must be assem-
Crispino et al., 1996). Processing of some of these mole-
bled de novo on each intron. Essential to initiation of
cules requires high concentrations of SR proteins. Nota-
assembly is recognition of intronic boundaries, the 5
bly, both of these examples of U1-independent splicing
and 3 splice sites. Despite intensive investigation, nei-
have been viewed as extraordinary reactions, not neces-
ther the precise determinants of splice site recognition
sarily reflective of normal spliceosome assembly.
nor the molecular details of splice site communication
In addition to the exceptions noted above, the consen-
are fully understood.
sus model of spliceosome assembly fails to adequately
Nevertheless, a vast amount of data has led to a con-
explain a number of phenomena associated with 5
sensus view of an ordered pathway of spliceosome as-
splice site choice. Specifically, a deterministic role for
sembly. In this model, U1 snRNP and associated factors
U1 snRNA base pairing (or U6 pairing) is unlikely to
recognize the 5 splice site, resulting in the formation of
account for activation of some cryptic donor sites
commitment (or E) complex; it is thought that functional
(Nilsen, 1994). Similarly, in certain unusual circum-
association of the 5 and 3 splice sites is established
stances, U5 snRNP unambiguously dictates the position
in this complex (Abovich and Rosbash, 1997; Reed and
of the 5 splice site (Newman and Norman, 1991, 1992;
Palandjian, 1997). The U1 snRNP-5 splice site interac-
Cortes et al., 1993). In this regard, Newman and Norman
tion proceeds at 0 C and is ATP independent. Subse-
(1992) demonstrated that, in the presence of a mutant
quently, U1 snRNP promotes U2 snRNP binding to the
5 splice site, base pairing between U5 snRNP and 5
branch point sequence; association of U2 snRNP is de-
exonic nucleotides activated a cryptic splice site several
pendent upon incubation at higher temperature and is
nucleotides upstream from the site of U1 interaction.
widely viewed as the first energy requiring step in
Here, we have used a sensitive assay to probe func-
tional recognition of the 5 splice site. This novel assay
is based upon competition between cis- and trans-splic-
* To whom correspondence should be addressed (e-mail: twn@po.
ing in an in vitro system derived from nematode em-
cwru.edu). These authors contributed equally to this work.
bryos. In nematodes, most pre-mRNAs are processed Molecular Cell
318
by both cis- and trans-splicing. In trans-splicing, the
helps to recruit U2 snRNP to the branchpoint (Figure
1A; Experimental Procedures).
5 most exon of the mature mRNA is acquired from
a specialized Sm snRNP, the spliced leader (SL) RNP
As an initial test of the competition assay, we inacti-
vated U1 snRNP via annealing of a 2 Ome oligonucleo-
(Blumenthal, 1995; Nilsen, 1997). Accurate processing
requires that SL addition be prevented at internal cis-3
tide complementary to its 5 end. Given the primary role
of U1 snRNP in 5 splice site recognition, we anticipated
splice sites. Several studies have demonstrated that
cis- and trans-3 splice sites appear to be functionally
that cis-splicing would be inhibited and trans-splicing
would be activated; trans-splicing does not depend the
equivalent, since cis-3 acceptors uncoupled from their
upstream 5 splice site are efficiently recognized and
5 end of U1 snRNP (Nilsen, 1997). While debilitation of
used by the trans-splicing apparatus (Conrad et al.,
U1 snRNP resulted in almost complete inhibition of cis-
1991; C. M. R., P. A. M., and T. W. N., unpublished data.).
splicing, activation of trans-splicing was surprisingly
Conversely, both in vivo and in vitro, a functional 5
modest under these conditions (Figure 1C, lane 2). By
splice site in cis-splicing is necessary and sufficient
contrast, occlusion of the 5 splice site with a comple-
to prevent inappropriate trans-splicing at downstream
mentary 2 Ome oligonucleotide (nucleotides
4 to
10)
splice sites and inactivation of a cis-5 splice site allows
led to a more significant increase in the level of trans-
trans-splicing (Conrad et al., 1993; Figure 1). Accord-
splicing even though cis-splicing was somewhat less
ingly, SL addition to an otherwise nonpermissive cis-3
inhibited (Figure 1C, lane 3). These results suggested
splice site provides a powerful reporter both for initial
either that oligonucleotide-blocked U1 snRNP could still
recognition of cis-5 splice sites and for functional com-
associate at some level with the 5 splice site or that
munication between 5 and 3 splice sites. Using an
factors in addition to U1 snRNP might be involved in
appropriate cis-splicing substrate, we generated a panel
early 5 splice site recognition.
of mutations in the 5 splice site region and monitored
activation of trans-splicing. The observed pattern of ac-
Activation of trans-Splicing in the Presence
tivation was not fully consistent with disruption of the
of Mutant 5 Splice Sites
U1 snRNP/5 splice site base-pairing interaction, and
As shown in Figure 1B, lane 3, and 1C, lane 4, mutation
an unexpected dependence on exonic nucleotides
1
of intronic nucleotides
1 to
6 led to full activation of
and
2 was revealed.
trans-splicing. To define more precisely the contribution
Site-specific cross-linking indicated that the impor-
of 5 splice site nucleotides to suppression of trans-
tance of these nucleotides is correlated with a specific
splicing, we introduced a series of dinucleotide muta-
interaction between the highly conserved U5 snRNP-
tions [( 1, 2)( 3, 4)( 5, 6)] in the intronic compo-
associated protein, Prp8, and the 5 splice site. This
nent of the 5 splice site; in addition, the 3 terminal
interaction requires ATP and incubation at 30 C and
two nucleotides of the 5 exon (-1,-2) were altered. As
unambiguously occurs independent of stable U2 snRNP
discussed below, the latter nucleotides are highly con-
association with the branch point. Several lines of evi-
served and have previously been implicated as determi-
dence indicate that the early Prp8/5 splice site inter-
nants of 5 splice site recognition. When tested in splic-
action occurs only if Prp8 is a constituent of U4/U6.U5
ing reactions, the series of mutant substrates yielded an
tri-snRNP. These results are likely to be of general signif-
intriguing range of phenotypes. Alteration of the nearly
icance because identical interactions were observed in
invariant GU ( 1, 2) resulted in full activation of trans-
HeLa cell extracts. The data define a previously unrec-
splicing, whereas alteration of nucleotides
3 and
4
ognized ATP-requiring step that proceeds in the ab-
yielded essentially no activation (compare Figure 2A,
sence of U2 snRNP