ABSTRACT
Nuclear pre-mRNA splicing requires ATP at several steps from spliceosome assembly to
product release. Here, we demonstrate that an integral component of the 20S U5
snRNP is an RNA-dependent ATPase. The ATPase activity of 20S U5 and 25S [U4/U6.U5] snRNPs
purified by glycerol gradient centrifugation is strongly stimulated by
homopolymeric RNA but not ssDNA. Purified 12S U1 and U2 snRNPs do not exhibit
ATPase activity. Moreover, the U5-associated NTPase specifically hydrolyzes ATP and dATP. The additional
purification of 20S U5 snRNPs by Mono Q chromatography does not affect the
efficiency of ATP hydrolysis. Both U5 and tri-snRNPs bind ATP stoichiometrically in an RNA-independent manner. A candidate ATPase was identified by UV-irradiation of purified snRNPs with radiolabeled ATP. In the
presence of homopolymeric RNA, the 200 kDa U5-specific protein is the major crosslinked protein, even in Mono Q-purified U5 snRNPs. The correlation between RNA-dependent ATPase activity in the U5 snRNP and the RNA-dependent onset of this crosslink strongly suggests that the 200 kDa protein is an RNA-dependent ATPase. Furthermore, both the formation
of the crosslink and ATPase activity appear with a similar substrate
specificity for ATP.
Splicing of nuclear mRNA precursors (pre-mRNA) occurs via two consecutive transesterification reactions. In the
first, the 5" splice site is cleaved and a lariat/3" exon intermediate is formed. The second step involves cleavage of
the 3" splice site, exon ligation and the release of the intron in the form of a
lariat (for review, see
1
,
2
). Catalysis of nuclear pre-mRNA splicing requires a large trans-acting, ribonucleoprotein known as the spliceosome. This
multicomponent complex consists of the four major small nuclear
ribonucleoprotein particles (snRNPs) U1, U2, U4/U6 and U5, and an as yet
unidentified number of non-snRNP protein splicing factors (reviewed in
1
,
3
,
4
,
46
). The snRNPs associate with an intron in an ordered manner. Initially, the U1
snRNP binds to the 5" splice site and subsequently the U2 snRNP associates with the branch
site. Both recognition events involve base pairing interactions between snRNA
and the pre-mRNA. The U4/U6 and U5 snRNPs then associate with the pre-spliceosome as a 25S [U4/U6.U5] tri-snRNP complex to form the mature spliceosome, which is
catalytically active (reviewed in
1
,
2
,
5
,
6
).
Prior to or concomitant with the first step of the splicing reaction, the snRNAs
undergo a number of conformational transitions. For example, the two
intermolecular helices formed between U4 and U6 snRNAs within the U4/U6 snRNP
particle (
7
-
10
) dissociate (
11
-
14
). The dissociation of U4/U6 is accompanied by the formation of new U6/U2 snRNA
intermolecular helices (
15
-
19
). Subsequent to the interaction of U1 and U2 snRNA, additional snRNA/pre-mRNA base pairing interactions occur. In particular, the conserved loop I
of U5 snRNA interacts with exonic sequences at both splice sites (
20
-
23
,
26
) while the conserved ACAGAG sequence of U6 snRNA recognizes intron sequences at
the 5" splice site (
23
-
25
,
27
). It is believed that the resultant RNA network of snRNA/snRNA and snRNA/pre-mRNA interactions forms, at least in part, the catalytic center of the
spliceosome (reviewed in
28
-
30
).
Nuclear pre-mRNA splicing is an ATP consuming process (
31
,
32
). However, the role of ATP in splicing is poorly understood. Although the
chemical events of splicing should not require exogenous energy, several steps
in the pathway from spliceosome assembly to product release are ATP-dependent (reviewed in
1
,
2
). The recent demonstration that several essential splicing factors in yeast
contain putative ATP-binding RNA helicase domains has provided a link between the requirement
for ATP hydrolysis during splicing and conformational changes of the snRNAs in
the spliceosome (
1
,
30
,
33
). Specifically, two proteins, PRP 5 (
34
) and PRP 28 (
35
), belong to the DEAD-box family of putative ATP-dependent RNA helicases, while PRP 2 (
36
), PRP 16 (
37
) and PRP 22 (
38
) fall into a subfamily of putative RNA-helicases which is characterized by a DEAH-box (
33
,
39
).
While an
in vitro
helicase activity has as yet not been demonstrated for any of the above
proteins, PRP 2 and PRP 16 have been shown to exhibit RNA-dependent ATPase activity
in vitro
(
40
,
41
). Moreover, there is strong evidence that PRP 16-catalyzed ATP hydrolysis is accompanied by a conformational rearrangement
in the spliceosome (
42
).
Interestingly, the DEAD/DEAH-box PRP proteins are not stably bound to snRNPs and, as has been
demonstrated for PRP 2 and PRP 16, apparently interact only transiently with
the spliceosome (
43
,
41
). Given that the snRNAs undergo significant conformational transitions in the
course of splicing (see above), it is somewhat surprising that no snRNP protein
has been identified to date as a putative RNA helicase, either in yeast or in
metazoans.
Purified snRNPs from HeLa cells contain >40 distinct proteins which fall into two classes (
6
). One class comprises a set of eight proteins (designated B, B", D1, D2, D3, E, F and G) which are common to U1, U2, U4/U6 and U5 snRNPs.
The other proteins are snRNP-specific proteins which are involved in snRNP-specific functions. The 12S U1 snRNP, for example, contains three
specific proteins designated 70K, A and C. The 17S form of the U2 snRNP
contains at least 11 specific proteins, nine of which are bound in a highly
salt-sensitive manner. Thus, at salt concentrations >300 mM KCl, a 12S U2 snRNP
can be isolated which contains only two specific proteins, namely A" and B"" (
44
). Six of the 17S U2 snRNP-specific proteins are identical to the proteins contained in the splicing
factors SF3a and SF3b (
45
,
46
). The 20S U5 snRNP contains at least nine specific proteins with apparent
molecular weights of 15, 40, 52, 100, 102, 110, 116, 200 and 220 kDa (
47
), while two proteins with apparent molecular weights of 60 and 90 kDa are
associated with the 12S U4/U6 snRNP (
48
; J. Lauber and R. Lührmann, manuscript in preparation). The 25S [U4/U6.U5] tri-snRNP complex contains an additional five proteins with apparent
molecular weights of 15.5, 20, 27, 61 and 63 kDa (
49
; J. Lauber and R. Lührmann, manuscript in preparation). Only a small subset of the snRNP-specific proteins in metazoans has been cloned (primarily U1 and U2-specific proteins) and sequence comparisons have not
identified any of them as putative ATP-binding proteins (reviewed in
1
,
46
).
As a first step towards the identification of possible ATP-binding proteins in mammalian snRNPs, we have investigated whether purified snRNPs from HeLa cells exhibit ATPase activity. Interestingly, we
could demonstrate that the 20S U5 snRNP purified by Mono Q chromatography
contains an RNA-dependent ATPase activity. Moreover, it was possible to UV-crosslink the U5-specific 200 kDa protein with ATP. Similar to the ATPase
activity, the formation of this crosslink was greatly stimulated by the
presence of homopolymeric RNA. Our data thus suggest that an integral component
of the U5 snRNP, namely the 200 kDa protein, is an RNA-dependent ATPase.
Nuclear extracts were prepared from HeLa cells (Computer Cell Culture, Mons,
Belgium) according to Dignam
et al.
(
50
). Immunoaffinity purification of total snRNPs was performed at 0.25 M NaCl
using a monoclonal antibody raised against the m
3
G cap. Under these conditions, predominantly the 12S U1, 12S U2, 20S U5 and 25S
[U4/U6.U5] snRNP forms are isolated (
51
). For preparative fractionations, 3-4 mg of total snRNPs were layered on a 10-30% glycerol (w/w) gradient containing buffer G (150 mM KCl, 20 mM Hepes-KOH pH 7.9, 1.5 mM MgCl
2
) and centrifuged at 28 000 r.p.m. for 17 h at 4oC in a Beckman SW28 rotor. Fractions (0.5 ml) were collected and 100 and 20
[mu]l aliquots withdrawn from every second fraction for polyacrylamide gel electrophoresis (PAGE) and protein quantitation, respectively. To improve
the separation of high molecular weight proteins, separating gels were prepared with a 12% polyacrylamide (PAA)-10% glycerol (w/w) and 10% PAA step (unless otherwise stated).
Purification by FPLC on Mono Q columns of 20S U5 and 12S U1/U2 fractions
derived from glycerol gradients was performed essentially as described by Will
et al.
(
51
). The 10S core U5 and 12S U4/U6 snRNPs were obtained by Mono Q chromatography
as previously described (
52
).
Total snRNPs, 10 [mu]g, or 1 pmol of purified snRNPs (see above), were incubated in a volume of
50 [mu]l for 1 h in 50 mM triethanolamine pH 8.2, 75 mM KCl, 1.5 mM MgCl
2
, 1.25 mM DTT and 20 [mu]M [[alpha]-
32
P]ATP (Amersham; specific activity 0.4 Ci/mmol). Single-stranded homopolymeric RNA or DNA (Sigma) was added to a final
concentration of 0.6 [mu]g/[mu]l. At 0 and 60 min time points, aliquots were withdrawn from the
reactions, quenched by the addition of 1 vol of 0.5 M Na
2
EDTA pH 8.0, and 0.8 [mu]l analyzed on PEI cellulose thin-layer chromatography plates
(Machery and Nagel) in 0.75 M KH
2
PO
4
. Quantitation of hydrolysis products was performed with a PhosphorImager
(Molecular Dynamics). To identify hydrolysis products, a reaction of [[alpha]-
32
P]ATP incubated in the presence of 0.01 U calf intestinal phosphatase
(Boehringer Mannheim) for 30 min at 37oC was loaded. In comparison with thin-layer chromatography, similar results were obtained when the release
of
32
P from [[gamma]-
32
P]ATP was monitored using a charcoal assay (
53
). Activated charcoal, 0.5% (v/v; Norit) was washed with 50 mM HCl/5 mM H
3
PO
4
, centrifuged and the pellet suspended with 20 mM H
3
PO
4
. Of this, 0.5 ml was added to each ATPase reaction and incubated on ice for 10
min. After centrifugation of adsorbed ATP at 15 000 r.p.m., 400 [mu]l of the supernatant was quantitated by Cherenkov counting.
Purified snRNPs, 5-10 pmol, was preincubated under ATPase assay conditions (see above) for 15 min in a total volume of 100 [mu]l containing 50 mM triethanolamine pH 8.2, 150 mM KCl, 3 mM MgCl
2
, 1.25 mM DTT and 5 [mu]M [[alpha]-
32
P]ATP (specific activity 2 mCi/mmol). Nitrocellulose discs (Sartorius) were
degassed in buffer W (100 mM NH
4
Cl, 3 mM MgCl
2
, 50 mM Tris-HCl pH 7.5, 1 mM DTT) and subsequently saturated with 1 mM ATP for 45 min
at 4oC. Reactions were diluted to 250 [mu]l with buffer W to allow even coating of the nitrocellulose discs and
unbound ATP was removed with a Millipore filtration unit. After washing twice
with 3 ml of buffer W, the discs were dried and quantitated by Cherenkov
counting.
Purified snRNPs, 2 pmol, in a total volume of 50 [mu]l were preincubated as described above, except that each reaction contained
20 [mu]Ci [[alpha]-
32
P]ATP (specific activity 3000 Ci/mmol). After preincubation, the samples were
transferred to ice and crosslinked with a Sylvania G8T5 germicidal UV lamp for
5 min at a distance of 2 cm. Proteins were subsequently extracted with phenol-chloroform and separated by PAGE (see above). The gel was stained with
Coomassie Blue (Serva), and crosslinked proteins were detected by
autoradiography. Digestion of snRNPs with proteinase K (Sigma) was performed by
addition of 1 vol containing 10 [mu]g enzyme in 300 mM NaCl, 100 mM Tris pH 7.5, 1% SDS and subsequent
incubation at 37oC for 30 min. The hydrolysis of snRNA in snRNPs was accomplished with 10 U
micrococcal nuclease (Boehringer Mannheim) in the presence of 1.7 mM CaCl
2
after 1 h at 37oC. Subsequent to incubation, the reaction was stopped by the addition of
0.1 vol of 100 mM EGTA pH 8.0. In control experiments, no intact snRNA was
detected after micrococcal nuclease digestion (Fig.
3
C) and the nuclease was completely inhibited by EGTA (data not shown).
To investigate whether ATPase activity is associated with spliceosomal snRNPs
in vitro
, total snRNPs were isolated from HeLa nuclear extracts by [alpha]-m
3
G immunoaffinity chromatography (
51
) and incubated with [[alpha]-
32
P]ATP. As shown in Figure
1
, this mixture of spliceosomal snRNPs, which contains predominantly 12S U1, 12S
U2, 20S U5 and 25S [U4/U6.U5] snRNPs, catalyzes the hydrolysis of ATP (as
assayed by thin-layer chromatography). Comparison with hydrolysis products generated by
calf intestinal phosphatase (Fig.
1
) demonstrated that the main product of the ATPase activity is ADP. Thus, the
observed hydrolysis of ATP cannot be attributed to phosphatase activity. The
addition of poly(U) significantly stimulated the ATPase activity (Fig.
1
, compare lanes 1 and 2). Of the four polynucleotides tested, poly(U) and
poly(A) had the most pronounced effect on ATPase activity (see below).
We examined the substrate specificity of snRNP-catalyzed ATP hydrolysis by a competition experiment with non-labeled NTPs. In the presence of poly(U), only ATP competed for the
hydrolysis of [[alpha]-
32
P]ATP by 25S [U4/U6.U5] tri-snRNPs, while a 50-fold molar excess of GTP, CTP or UTP, showed no effect (Fig.
3
B, lanes 6-9, respectively). An identical substrate specificity was also observed
with 20S U5 snRNPs (data not shown). Inhibition of ATP hydrolysis was also
observed when non-labeled 2"-dATP or 3"-dATP was used as competitor (data not shown).
Consistent with this finding, [[alpha]-
32
P]2"-dATP was hydrolyzed efficiently (Fig.
3
A, lane 8). Thus, the substrate specificity is apparently restricted to the
recognition of the adenine base by the enzyme(s). Consistent with the
observation that the snRNP-associated ATPase hydrolyzes the [gamma]-phosphodiester bond (only ADP is formed), ATP hydrolysis
could be inhibited by the addition of the non-hydrolyzable ATP analogue [[gamma]-S]ATP (not shown). The effect of particle disruption on snRNP-catalyzed ATP hydrolysis was examined by hydrolyzing
the snRNA component of 20S U5 snRNPs with micrococcal nuclease (Fig.
3
A and C). As similar activities are observed prior to and after nuclease treatment (Fig.
3
A, compare lanes 6 and 2), we conclude that snRNP integrity is not required for
RNA-dependent ATP hydrolysis.
The stimulatory effect of RNA on ATPase activity was most pronounced with
poly(A) and poly(U) (Fig.
3
B, lanes 1-5). As compared to the level of ATP hydrolysis in the absence of
exogenous RNA, ATPase activity increased 50-75-fold upon the addition of poly(A) or poly(U) to 20S U5 snRNPs.
Poly(C) stimulated ATP hydrolysis <12-fold, while the addition of poly(G) had no detectable effect. A similar
effect of RNA on ATPase activity was observed with FPLC-purified 20S U5 snRNPs (Fig.
4
), which is consistent with the assumption that the ATPase activity is integral
to the U5 snRNP. While low levels of ATPase activity are detected in total
snRNP preparations even in the absence of exogenously added RNA (Fig.
1
), the fractionated snRNPs exhibit significant ATPase activity only in its
presence (Fig.
4
). Importantly, single-stranded poly (dA) and poly (dT) DNA did not significantly stimulate the
snRNP-associated ATPase activity (Fig.
4
).
We next tested whether purified snRNPs could bind ATP. Equimolar amounts of
purified snRNPs fractionated by glycerol gradient centrifugation (see Fig.
2
) were incubated with [[alpha]-
32
P]ATP, and the binding of ATP was measured by a nitrocellulose filter binding
assay. Significant ATP binding was observed with the 25S [U4/U6.U5] and 20S U5
snRNPs, while the 12S U1/U2 snRNP fractions produced background values (Fig.
5
A). At an ATP concentration of 5 [mu]M, ~0.3-0.4 pmol ATP per pmol 25S [U4/U6.U5] tri-snRNP were bound. A similar value was obtained with 20S U5
snRNPs (Fig.
5
A). In keeping with the substrate specificity of the ATPase activity described
above, the binding of ATP was competed by non-labeled ATP but not by any other rNTP tested (not shown). It is important
to note that poly(U) (as well as any other polynucleotide; data not shown) had
no significant effect on the binding efficiency of ATP to 25S [U4/U6.U5] and
20S U5 snRNPs (Fig.
5
A). Thus, while the ATPase activity is strongly stimulated by poly(U) (see
above), binding of ATP to the ATPase(s) of the U5 snRNP occurs independent of
exogenously added RNA. Figure
5
B shows that binding of ATP to 25S [U4/U6.U5] tri-snRNPs levels off at 15-20 [mu]M ATP with a ratio of ~0.4 pmol ATP bound per pmol tri-snRNP. Binding of ATP by 20S U5 snRNPs occurs with a
similar saturation behaviour (not shown). These data suggest that an integral
U5 snRNP protein rather than a contaminating non-snRNP protein is responsible for ATP binding.
Based on our finding that 20S U5 and 25S [U4/U6.U5] snRNPs stably bind ATP, we
were interested in determining whether ATP could be crosslinked to one or more
of the U5 snRNP proteins. Either 25S [U4/U6.U5] tri-snRNPs, purified by glycerol gradient centrifugation, or Mono Q-purified 20S U5 snRNPs were incubated with [[alpha]-
32
P]ATP in the presence or absence of poly(U) and irradiated with 260 nm UV light.
The snRNP proteins were then fractionated by SDS-PAGE and ATP-crosslinked proteins were detected by autoradiography. Upon UV
irradiation of gradient-purified 25S [U4/U6.U5] tri-snRNPs in the absence of poly(U), an ATP-crosslinked protein, migrating near the 90 kDa protein, was
detected (Fig.
6
A, lane 6). In addition, less pronounced crosslinked proteins with molecular
weights of ~100 and 200 kDa were reproducibly observed (Fig.
6
A, lane 6; compare with lanes 3 or 1 for Coomassie-stained proteins of the 25S [U4/U6.U5] tri-snRNP or molecular weight standards, respectively). In the presence
of poly(U), i.e. when ATPase activity is stimulated, the pattern of crosslinked
25S [U4/U6.U5] tri-snRNP proteins changes drastically (Fig.
6
A, lane 7): a major crosslinked protein with an apparent molecular weight of 200
kDa is now observed, while the other crosslinks are reduced significantly. When
20S U5 snRNPs were investigated, only minor crosslinks, if any, were detected
in the absence of poly(U) (Fig.
6
A, lane 8). Consistent with the crosslink pattern observed for 25S [U4/U6.U5]
tri-snRNPs, the addition of poly(U) to 20S U5 snRNPs strongly stimulated
crosslinking of ATP to a U5 snRNP protein in the 200 kDa range (Fig.
6
A, compare lanes 8 and 9). To confirm that the 200 kDa but not the 220 kDa
protein of the U5 snRNP is crosslinked, the proteins in the crosslink reactions
were Coomassie-stained, and the bands were superimposed on those of the autoradiograph
(Fig.
6
B). This approach allowed a clear distinction of these proteins and confirmed
that the crosslinked band migrated exactly with the 200 kDa U5-specific protein. The 200 kDa crosslink is sensitive to protease treatment
but resistant to nuclease (Fig.
6
A, lanes 13 and 14), indicating that ATP was crosslinked directly to the 200 kDa
protein and that the migration behavior of this crosslink is not due to the
presence of crosslinked RNA. The 12S U1 and U2 snRNP-containing gradient fractions, which were previously shown to lack
significant ATPase activity, do not yield any crosslink signal in the absence
or presence of poly(U) (Fig.
6
A, lanes 10 and 11).
We next investigated the crosslinking specificity of ATP to the 200 kDa protein
by competition studies with non-labeled NTPs. While crosslinking of
32
P-labeled ATP was abolished upon addition of an excess of non-labeled ATP, none of the other rNTPs tested significantly reduced
the labeling of the 200 kDa protein with ATP (Fig.
7
, lanes 6-9). This finding is consistent with the substrate specificities which we
have previously determined for both the ATPase activity and the binding of ATP
to 20S U5 and 25S [U4/U6.U5] snRNPs (see above). Crosslinking is also inhibited
by an excess of non-labeled ADP but not GDP (Fig.
7
, lanes 10 and 11). This finding is interesting given our observation that
stimulation of ATPase activity [i.e. the addition of poly(U)] greatly enhances
formation of the ATP-200 kDa crosslink. It is therefore possible that we observe an ADP-rather than an ATP-crosslink. In sum, our data suggest that the 200 kDa U5
snRNP protein is a candidate for an RNA-dependent ATPase.
In this manuscript, we demonstrate that 20S U5 snRNPs isolated from HeLa cells
possess an RNA-dependent ATPase activity. A number of observations indicate that this
activity is not due to a contaminating, non-snRNP NTPase, rather it resides with one or more of the U5-specific proteins.
First, when a mixture of spliceosomal snRNPs, obtained by [alpha]-m
3
G-immunoaffinity chromatography, is fractionated by glycerol gradient
centrifugation, the vast majority of ATPase activity cofractionated with 20S U5
and 25S [U4/U6.U5] tri-snRNPs (Fig.
2
). Secondly, and more significantly, even highly purified Mono Q-derived 20S U5 snRNPs exhibit ATPase activity with a similar specific
activity as gradient-purified U5 snRNPs, while 12S U1, U2 and U4/U6 snRNPs do not hydrolyze ATP
to a measurable extent (Fig.
3
A). The specific and stable association of the ATPase with U5 snRNPs even after
gradient centrifugation and ion-exchange chromatography would be unusual if it were a contaminating non-snRNP ATPase. Lastly, our findings that purified 20S U5 and 25S
[U4/U6.U5] tri-snRNPs bind ATP to a significant extent (~0.4 pmol ATP per pmol snRNP; Fig.
5
) and that 10S core U5 snRNPs do not hydrolyze ATP, strongly suggest that one or
more of the U5-specific protein(s) is an ATPase.
In an attempt to identify snRNP proteins which bind ATP, we UV-irradiated snRNPs in the presence of [[alpha]-
32
P]ATP. When purified U5 snRNPs were used as a source of snRNPs, the 200 kDa
protein was the major ATP-labeled protein. Most interestingly, crosslinking of ATP was greatly
enhanced in the presence of poly(U) despite the fact that purified U5 snRNPs
bound ATP equally well, both in the presence and absence of poly(U). The
simultaneous requirement of poly(U) for both the stimulation of ATPase activity
in 20S U5 snRNPs (Figs
3
and
4
) and for efficient affinity labeling of the 200 kDa protein with ATP (Fig.
6
) makes the 200 kDa protein a promising candidiate for a U5-specific ATPase. This idea is further supported by the fact that the
substrate specificity of the 20S U5 snRNP ATPase (Fig.
3
B) is identical to the binding substrate specificity of the 200 kDa protein as
assayed by photoaffinity labeling (i.e. in each case only an excess of non-labeled ATP competes for ATP hydrolysis or ATP crosslinking, respectively;
Fig.
7
).
It is currently not clear whether ATP or ADP [formed after poly(U)-stimulated ATP hydrolysis] is crosslinked to the 200 kDa protein. The
almost exclusive labeling of the 200 kDa protein with ATP in purified U5 snRNPs
does not necessarily exclude the possibility that an additional U5 protein may
also bind and hydrolyze ATP. We note that a comparatively weak, but
reproducible, ATP crosslink was also observed for a protein in the 100 kDa
region (Fig.
6
A). Moreover, with gradient-purified 25S [U4/U6.U5] tri-snRNPs, we observed crosslinking of ATP to a protein with an
apparent molecular weight of ~90 kDa (Fig.
6
A, lane 6). Interestingly, this protein was labeled only in the absence of
homopolymeric RNA. In the presence of poly(U), the dominant ATP-labeled protein was the U5-specific 200 kDa protein (Fig.
6
A, lanes 7 and 9). The nature of the ~90 kDa protein that is affinity-labeled by ATP in the tri-snRNP is not clear. As this protein is not associated with Mono
Q-purified 20S U5 snRNPs and does not comigrate with the tri-snRNP 90 kDa protein in one- (Fig.
6
A) or two-dimensional gels (unpublished data), it may be a minor, non-snRNP, ATP binding protein which cofractionates with gradient-purified 25S [U4/U6.U5] tri-snRNPs.
The specific affinity labeling of the 200 kDa but not the 220 kDa U5 snRNP
protein with ATP is interesting for another reason. In yeast, a 260 kDa protein
termed PRP 8 has previously been demonstrated to be a U5-specific protein. It has also been shown that PRP 8 is immunologically
related to one of the human high molecular weight U5 proteins (
54
-
56
). It has remained unclear, however, whether the human 200 and 220 kDa proteins
are distinct proteins or whether they are structurally related. The exclusive
labeling of the 200 kDa protein with ATP supports the idea that the two
proteins are functionally distinct. Moreover, the yeast PRP 8 protein does not
contain any of the motifs that are characteristic for an ATPase (
58
).
Although several putative ATP-dependent RNA helicases have been demonstrated to be essential for the pre-mRNA splicing process in yeast, none of these proteins appears to be
stably associated with either an snRNP or the spliceosome. For example, PRP 2
and PRP 16, which exhibit RNA-dependent ATPase activity
in vitro
, interact only transiently with the yeast spliceosome (
40
,
41
). Thus, our identification of RNA-dependent ATPase activity in the U5 snRNP is the first example of an snRNP-intrinsic ATPase activity.
Interestingly, PRP 2 and PRP 16 ATPase activity is stimulated by homopolymeric
RNA in a manner similar to that of the U5 snRNP. In each case, poly(A), poly(U)
and, to a lesser extent, poly(C) are efficient stimulators of ATPase activity,
while ssDNA has no effect (
40
,
41
). Despite these similarities, fundamental differences exist between these
ATPases which indicate that the U5 snRNP ATPase is not a homologue of PRP2 or
PRP16. First, PRP 2 and PRP 16 exhibit relaxed NTP substrate specificities (
40
,
42
). Secondly, as mentioned above, they are non-snRNP proteins which interact transiently with the yeast spliceosome. In
contrast, the probable U5 snRNP ATPase, namely the 200 kDa protein, is stably
associated with HeLa spliceosomal complexes B and C (
48
,
57
).
Analogous to the aforementioned PRP proteins, it is tempting to speculate based
on its RNA-dependent ATPase activity that the U5 snRNP may possess proteins with
putative ATP-dependent RNA helicase activity. The presence of RNA helicases within the
mammalian spliceosome has recently been suggested by the identification of HeLa
gene fragments which share sequence homology with the gene encoding the DEAH-box protein PRP 22 (
59
). Based on the presence of multiple RNA-RNA interactions within the spliceosome which are highly dynamic in
nature, it is easy to envisage several potential substrates for a U5 snRNP
helicase (see Introduction). Alternatively, the U5 snRNP ATPase could
exclusively bind to single-stranded RNA
in vivo
and directly affect RNA structure, in a manner more analogous to chaperones
than helicases (
60
,
61
).
The authors are highly indebted to C. L. Will and B. Kastner for helpful
discussions and comments on the manuscript. We also thank S. Becker, M. Wicke,
A. Badouin and I. Öchsner for excellent technical assistance. This work was supported by
grants from the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen
Industrie. B. L. was supported by a fellowship from the Verband der Chemischen
Industrie.
REFERENCES
Return