Conserved tertiary base pairing ensures proper RNA folding and efficient assembly of the signal recognition particle Alu domain.

Abstract

Proper folding of the RNA is an essential step in the assembly of functional ribonucleoprotein complexes. We examined the role of conserved base pairs formed between two distant loops in the Alu portion of the mammalian signal recognition particle RNA (SRP RNA) in SRP assembly and functions. Mutations disrupting base pairing interfere with folding of the Alu portion of the SRP RNA as monitored by probing the RNA structure and the binding of the protein SRP9/14. Complementary mutations rescue the defect establishing a role of the tertiary loop-loop interaction in RNA folding. The same mutations in the Alu domain have no major effect on binding of proteins to the S domain suggesting that the S domain can fold independently. Once assembled into a complete SRP, even particles that contain mutant RNA are active in arresting nascent chain elongation and translocation into microsomes, and, therefore, tertiary base pairing does not appear to be essential for these activities. Our results suggest a model in which the loop-loop interaction and binding of the protein SRP9/14 play an important role in the early steps of SRP RNA folding and assembly.

Full text

Page 1

Conserved tertiary base pairing ensures proper RNA
folding and efficient assembly of the signal
recognition particle Alu domain
Laurent Huck, Anne Scherrer, Lionel Terzi, Arthur E. Johnson1, Harris D. Bernstein2,
Stephen Cusack3, Oliver Weichenrieder3and Katharina Strub*
De ´partement de Biologie Cellulaire, Universite ´ de Gene `ve, CH-1211 Gene `ve 4, Switzerland,1Department of Medical
Biochemistry and Genetics, Texas A&M University, System Health Science Center, College Station, TX 77843-1114,
USA,2GeneticsandBiochemistry Branch, National Institute of Diabetes andDigestive andKidney Diseases,National
Institute of Health 10, Center Drive, Bethesda, MD 20892-1810, USA and3European Molecular Biology Laboratory,
Grenoble Outstation, 6, Rue Jules Horowitz, F-38042 Grenoble, CEDEX 9, France
Received July 29, 2004; Revised and Accepted September 2, 2004
ABSTRACT
Proper folding of the RNA is an essential step in the
assemblyoffunctionalribonucleoproteincomplexes.
We examined the role of conserved base pairs
formed between two distant loops in the Alu portion
ofthemammaliansignal
RNA (SRP RNA) in SRP assembly and functions.
Mutations disrupting base pairing interfere with fold-
ingoftheAluportionoftheSRPRNAasmonitoredby
probing the RNA structure and the binding of the
protein SRP9/14. Complementary mutations rescue
the defect establishing a role of the tertiary loop–
loop interaction in RNA folding. The same mutations
in the Alu domain have no major effect on binding of
proteinstotheSdomainsuggestingthattheSdomain
can fold independently. Once assembled into a
complete SRP, even particles that contain mu-
tant RNA are active in arresting nascent chain
elongation and translocation into microsomes, and,
therefore, tertiary base pairing does not appear
to be essential for these activities. Our results
suggest a model in which the loop–loop inter-
action and binding of the protein SRP9/14 play an
important role in the early steps of SRP RNA folding
and assembly.
recognition particle
INTRODUCTION
The assembly of ribonucleoprotein particles (RNPs) is a com-
plex process thatincludesmultiplestepssuchasproperfolding
of the RNA moiety and the ordered concomitant or sequential
association of the proteins at specific subcellular locations.
Folding of RNAs with sizes >300 nt is rather slow and
may be assisted by proteins with RNA chaperone activities
and by specific RNA binding proteins, which may remain part
of the RNP. Reasons for the slow folding of RNA molecules
include their highly complex tertiary structure as well as their
intrinsic propensity to become kinetically trapped in inactive
structures [for review see (1,2) and references therein].
The goal of this study was to examine the specific role of an
RNA tertiary structure in the Alu domain of the mammalian
signal recognition particle (SRP). SRP is probably one of the
best-characterized ribonucleoprotein complexes, since it can
be assembled and studied in vitro. SRP plays an essential role
in targeting proteins to the endoplasmic reticulum (ER) [for
review see (3)], which is the first step in routing proteins into
the secretory pathway. SRP specifically recognizes signal
sequences, a common hallmark of ER-targeted proteins, in
nascent polypeptide chains as they exit from the ribosome
and subsequently delays elongation of the nascent chain.
SRP then targets the ribosome-nascent chain complex to the
translocon, a specific structure in the ER membrane that pro-
motes the transfer of the nascent chain across or into the ER
membrane [for review see (4)].
SRP has a two-domain structure in most but not all organ-
isms [for review see (5)]. The S domain of mammalian SRP
comprises the central part of SRP RNA (or 7SL RNA) and the
proteins SRP68/72, SRP54 and SRP19 (Figure 1). The
conserved RNA helix VIII and SRP54 harbour the signal
recognition and the targeting functions of SRP (6,7). The
Alu domain, which comprises the 50and 30ends of SRP
RNA and the protein SRP9/14, retardsor arrests the elongation
of the nascent chain and thereby increases the translocation
efficiency in vitro (6,8,9). Over the last few years, a vast body
of atomic structure information was obtained [for review see
(10–12)] and the positioning of SRP on the ribosome was
reconstructed from cryo-electron microscopy (cryo-EM)
images (13). This information has yielded new insights into
the mechanism of SRP functions and has facilitated a more
specific biochemical analysis.
SRP is thought to assemble in nucleoli, because SRP RNA
and SRP proteins, except SRP54, can be observed in this
*To whom correspondence should be addressed. Tel: +41 22 379 6724; Fax: +41 22 379 6442; Email: Strub@cellbio.unige.ch
Present address:
Oliver Weichenrieder, Dept. of Molecular Carcinogenesis—H2, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, Netherlands
Nucleic Acids Research, Vol. 32 No. 16 ª Oxford University Press 2004; all rights reserved
Nucleic Acids Research, 2004, Vol. 32, No. 16 4915–4924
doi:10.1093/nar/gkh837

Published online September 21, 2004

Page 2

structure (14–16). The localization of SRP RNA to the nucle-
olus is dependent on the Alu sequences and on the RNA helix
VIII (17). In addition, import of SRP proteins into nucleoli and
export of SRP RNA are mediated by specific receptors
(18–20). Till date, nothing is known about the specific events
that coupleRNA synthesisandfolding totheorderedassembly
of SRP proteins.
An Alu domain-like structure is present in all eukaryotic and
archaeal and in few eubacterial SRP RNAs. The common
signature of this domain is a three-way junction of stems
and a highly conserved single-stranded region (21,22). Of
the SRP RNAs including an Alu domain, the yeast and
protozoan RNAs are missing one or both hairpins. In trypano-
somes, the absence of one hairpin in the Alu domain may have
been compensated for by the acquisition of an additional
transfer RNA (tRNA)-like RNA in the SRP particle (23,24).
As revealed by crystallography, the Alu domain contains a
compactly folded three-way junction of stems. A stack of two
stems (H1.1 and H1.2) and a third stem (H2) are connected by
a single-stranded U-turn at this junction [Figure 1; (25)].
Note that we use the topological nomenclature in which
helix H2 and loop L2 represent stem–loop III and helix
H1.2 and loop L1.2 stem–loop IV in the helix diagram of
Figure 1A [according to the nomenclature in (26)]. The
single-stranded connecting region is highly conserved in
primary sequence and represents the major protein-binding
site [Figure 1B, U-turn motif; (22,25)]. The U-turn determines
the relative orientation of the two helical stems. The RNA fold
is further stabilized by a tertiary structure consisting of three
G-C base pairs, which are formed between the two distant
loops L1.2 and L2 (Figure 1D, G13-C37, G14-C34 and
C15-G33). The three base pairs form an extended stack
together with two purines from the loops L2 and L1.2 (G16
and A36). The existence of loop–loop interactions had pre-
viously been suggested based on sequence complementarities
observed between the two loops (27).
Theprotein SRP9/14 also makescontacts tothe centralstem
connecting the Alu and the S domains (Figure 1B). Further
biochemical and structural data support a model in which the
central stem would fold back by up to 180?to align alongside
the 50domain allowing contacts to the SRP9 moiety of the
protein (25,28). The model is consistent with the requirement
that the 30and the 50domains have to be flexibly linked for
fully efficient binding of SRP9/14 (29).
After the existence of loop–loop interactions had been
confirmed and their identity revealed by crystal structure,
we decided to examine their role in SRP assembly and
function. Base pairs were disrupted by specific individual
mutations in both loops and restored with a different primary
sequence by simultaneous complementary mutations. The
mutated RNAs were analysed by native gel electrophoresis
and by limited V1 ribonuclease digestion experiments and
assayed for SRP9/14 binding. The mutated SRP RNAs
were also reconstituted into SRP and their activities compared
to the activities of SRP comprising wild-type (WT) RNA. Our
results demonstrate that base pairing between the loops is
Figure 1. Structure of SRP and the SRP Alu domain. (A) Schematic representation of SRP, (B) secondary structure of the minimal Alu RNA that still binds 9/14
efficiently. Stems and loops are named according to the topological nomenclature. Bases in the loops that form tertiary base pairs are highlighted in red. Protein
footprintsareshowninboldface.U-turnsaremarkedwithasterisksandstretchesof10ntaremarkedinblue.TheU-turnmotifmarksahighlyconservedsequencethat
represents the major protein-binding site. (C) Structure of the SRP Alu 50domain. SRP9 and SRP14 are displayed as dark and light grey ribbons, respectively. The
RNA is shown as a yellow ribbon with the loop sequences and the U-turn motif shown in orange. Nucleotides from loops L2 and L1.2 that are involved in tertiary
interactionsbetweentheloopsareshownaswireframe.(D)Detailedviewofthetertiarybasepairingbetweenloops.BasesG13,G14andC15formhydrogenbonds
with C37, C34 and G33, respectively (dotted greenlines). One basefrom each loop, G16 and A36, are positionedto extend the stack formed by the three base pairs.
4916Nucleic Acids Research, 2004, Vol. 32, No. 16

Page 3

important for proper folding of the SRP RNA Alu domain. The
mutations in the Alu domain had no significant impact on the
binding of the S domain proteins, suggesting that it can fold
independently. Despite the folding defects, all but one mutant
RNA could be assembled into SRPs with intact elongation
arrest and translocation activities, indicating that base pairing
is not essential for these activities.
MATERIALS AND METHODS
Materials
Sources of materials were as follows: urea, AppliChem;
4-(2-hydroxyethyl)-1-piperazine-1-ethane
(HEPES), AppliChem; potassium and magnesium acetate
[KOAc and Mg(OAc)2], Fluka; Nikkol–BL-85Y, Nikko
Chemical Co.; Triton X-100, Interchim; [35S]methionine
(1500 mCi/mmol), Amersham Biosciences; Aquaphenol,
QbioGene; DTT, Eurobio; spermidine, Tris[hydroxymethyl]
aminomethane (Tris), BSA and ribonucleotides, Sigma. The
T7 RNA polymerase expression plasmid was a kind gift of
Dr Mallet (30). The T7 RNA polymerase was kindly prepared
by Viviane Simonet, Julien Hasler and L. Terzi.
sulfonicacid
Mutation and expression of the SRP RNA gene
Mutations at specific sites in the SRP RNA gene were intro-
duced byPCR using the QuickChangeTMmethod (Stratagene).
The primer oligonucleotides contained the desired mutations
flanked by 15 nt (Microsynth). The template for PCR was the
plasmid p7Sswt (22). Two sequential rounds of mutations
yielded the clones with complementary mutations. All clones
were verified by automatic sequencing. Plasmids were linear-
ized with XbaI (Gibco BRL) before synthesis of the RNA with
T7 RNA polymerase in 100 ml reactions containing 40 mM
Tris-HCl,pH8.1,7mMMgCl2,1mMspermidine,5mMDTT,
1 mM of each rNTPs, 1 mg/ml BSA, 100 ng/ml DNA template
and 0.1mg /ml T7 RNA polymerase. After transcription, the
DNA was digested with RQ1 RNAse-free DNAse (Promega)
and the samples extracted with phenol and purified on 1 ml
Sephadex G50 (fine, Pharmacia) or on Nucleobond (Machery
& Nagel) columns. The RNAs were precipitated and dissolved
in sterile water. To optimize protein binding with biotinylated
SRP RNAs, we tested different ratios of UTP/bio-11-UTP
(ENZO Life Sciences) in the transcription reactions. Protein
binding was optimal when using equal ratios of both nucleo-
tides. RNA concentrations were determined by OD260and the
quality of the RNA samples analysed by denaturing 6%
PAGE. The concentrations of biotinylated RNAs were
determined by comparing three dilutions of each sample to
a dilution series of an RNA standard of unmodified WT RNA
with ethidium-bromide-stained denaturing PAGE. Canine
SRP RNA was extracted from the DEAE resin used to dis-
assemble SRP into RNA and protein subunits (6).
RNA analysis
Native gels (acrylamide–bisacryamide ratio of 40:1) were run
in 50 mM Tris, pH 7.5 and 10 mM Mg(OAc)2in the cold room
at 0.22 W/cm3for 4 h. The samples were loaded in 20 mM
KOAc and 3 mM Mg(OAc)2neutralized to pH 7.5 and 40%
glycerol. Native RNA was stained with GelStar1(Cambrex).
One microgram of RNA was digested with V1 RNAse (Pierce,
0.05 U per reaction) in 20 ml of 300 mM KOAc, 3 mM
Mg(OAc)2and 10 mM HEPES, pH 7.5 for 20 min on ice.
The reactions were stopped and the RNAs purified with
phenol/dichloromethane extractions. The precipitated RNA
was dissolved in formamide loading buffer and displayed
by 10% PAGE. The RNA was revealed by ethidium bromide
staining of the gel and visualized on an ultraviolet light box.
RNA binding assays
Synthetic mRNAs for human SRP14, human SRP19 and
canine SRP54 were prepared as described previously in
(31–33), respectively. The [35S]methionine-labelled h14 was
synthesized inwheatgerm translation reactionsinthe presence
of 3 pmol of recombinant h9 (34) and 6 pmol of recombinant
h9/14 in 10 ml translation reaction. For the binding reactions,
an aliquot of 3.3 ml was combined with the desired amounts of
biotinylated RNA in a final volume of 50 ml of binding buffer
(50 mM HEPES–KOH, pH 7.5, 500 mM KOAc, pH 7.5, 5 mM
Mg(OAc)2and 0.01% Nikkol). The complex was allowed to
form for 10 min on ice and then for 10 min at 37?C. The
magnetic streptavidin beads (Dynabeads M-280, Dynal
Biotech ASA) were prepared as indicated by the manufacturer
and equilibrated with binding buffer. The beads (60 ml) were
combined with the protein–RNA samples and incubated for
another hour at 4?C. The bound samples were washed three
times for 10 min with binding buffer including 0.1% Triton
X-100. Bound proteins were analysed by 15% SDS–PAGE,
visualized by autoradiography and quantified with the
PhosphoImager (Bio-Rad). We confirmed by denaturing
PAGE and ethidium bromide staining that no detectable
amounts of SRP RNA were present in the unbound fractions
atthehighestRNAconcentrationsused(detectionlimit25ng).
In addition, the protein contents of the supernatant fractions
were also quantified in the initial experiments. The bound and
free proteins together accounted for the total protein input
(100 – 10%) confirming that dissociation of the complex
was negligible during the wash. A negative control RNA,
rat 4.5S RNA (34), showed no detectable protein-binding
activity (data not shown). For the binding reactions with
h19 and c54, the translation reactions contained 6 pmol of
recombinant h19 and 6 pmol of recombinant h19 and c54,
respectively.
SRP proteins
Theproteinsh9,h14andh19wereexpressedinEscherichiacoli
from the plasmid pEh9, pEh14 and pE19 (8,34) using the T7
RNA polymerase expression system (35). Cells were lysed in a
FrenchPressTMandtheextractsofh9andh14werecombinedfor
the purification of the complex. Hi-Trap Heparin (Amersham
Biosciences), hydroxyl apatite (Bio-Rad) and Superdex-200
(Amersham Biosciences) chromatography was used to purify
h9/14. Hi-Trap Heparin, CM and Superdex-200 (Amersham
Biosciences)chromatographywasusedtopurifyhf19.c54was
expressedininsectcellsusingthebaculovirusexpressionsystem
(7). Following lyses of the cells in the homogenizer, the
protein was purified by CM and Hi-Trap Heparin chromato-
graphy. The purified proteins were quantified by spectrophoto-
metry at 280 nm. Molar extinction coefficient eh9/14,
15 130 M?1cm?1; eh19, 12 570 M?1cm?1; and e54,
Nucleic Acids Research, 2004, Vol. 32, No. 164917

Page 4

22 220 M?1cm?1. Canine SRP68/72 was purified from canine
SRP as described previously (6). SRP68/72 was quantified by
comparing the SRP68 signal to known concentrations of BSA
in a Coomassie-stained SDS–PAGE.
Particle reconstitutions and activity assays
Particleswerereconstitutedasdescribedpreviously(36)in50ml
reactions at 2 mM protein concentrations and 4 mM RNA con-
centrationsforWTand2CompRNAs.2L2and3L2RNAswere
usedatfinalconcentrationsof16and24mM,respectively.Fifty
microliters DEAE Sephacel (Pharmacia) columns were eluted
twice with 50 ml of 20 mM HEPES–KOH, pH 7.5, 500 mM
KOAc, pH 7.5, 5 mM Mg(OAc)2, 0.01% Nikkol and 1 mM
DTT. Active SRP is eluted in the first fraction. In the reactions
with 2L2 and 3L2 RNAs, we first eluted with 30 ml followed
bytwofractionsof50mltooptimizetheyieldofactiveSRP.On-
tenth of the total fraction was analysed by 5–20% SDS–PAGE
and visualized by silver staining. c54 samples containing 50,
100,150and200ngofproteinwererunonthesamegelandused
to estimate the particle concentrations. For the direct assays,
reconstitutions were done in 8 ml at 0.5 and 1 mM final concen-
trations of canine SRP68/72, recombinant h9/14, h19 and c54
and in vitro synthesized SRP RNAs, respectively. Activity
assays were done as described previously (8) at final particle
andmembraneconcentrationsof100nMand0.15eq/ml,respec-
tively. Preprolactin, prolactin and cyclin D were quantified by
the useofaphosphorescence imagingsystem(Bio-Rad). Elon-
gation arrest and translocation efficiencies were calculated as
follows:
EA =
1 ?
Ps=Cs
Po=Co
??
· 100,
where EA is the percentage elongation arrest activity, Ps and
Cs are the amounts of preprolactin and cyclin quantified in the
sample and Po and Co the amounts of preprolactin and cyclin
present in the negative control (SRP buffer or SRP reconstit-
uted without h9/14). T = 100 · [P/(pP + P)], where T is the
percentage translocation, P is the amount of prolactin and pP
the amountofpreprolactin quantifiedineach sample. AllSRPs
were tested in at least two independent experiments.
Computer programs and data bases
For the sequence comparison of SRP RNAs, we used the SRP
database (21). Molecular graphics images were produced
using the UCSF Chimera package from the Computer
Graphics Laboratory, University of California, San Francisco
(supported by NIH P41 RR-01081) and the Swiss PDB viewer
using Pov-RayTM(http://www.expasy.org/spdbv/).
RESULTS
Tertiary base pairing and structural determinants
in loop L2 are conserved in evolution
To examine the conservation of structural determinants in the
loops L1.2 and L2, we created a new structure-based align-
ment of the loop sequences. In mammalian SRP RNA, loop L2
is well defined by the sheared G-G base pair and has a rather
stiff character determined by another U-turn (U12, Figure 1B,
asterisk). In contrast, loop L1.2 lacks internal stabilizing
elements and is, therefore, presumed to be flexible. Its most
important structural feature is its capacity to form base pairs
with loop L2 (25). For the analysis, we included SRP RNAs
[SRPDB; (21)] containing the three-way junction of stems
forming the specific fold of the mammalian SRP Alu domain
(Figure 2). As illustrated by the helix diagrams, the lengths of
the stems and the sizes of the loops are remarkably different.
Figure 2. Structure-based alignment of loop L1.2 and L2 sequences in SRP RNAs of animal metazoans, of plants and of eubacterial and archaeal species.
(A) Nucleotides proposed to base pair are shown in red, sheared G-G pairs in blue and U-turns in green. First nucleotides in stems are shown in italic. Black bars
delineate the loop sequences. For each species, only one representative is shown. Sequences were obtained from SRPDB. (B) Helix diagrams of mammalian, plant
and archaeal Alu RNA 50domains. Asterisks highlight a conserved nucleotide in the archaeal RNAs. The helix diagram of a eubacterial SRP RNA resembles most
closely the one of archaeal SRP RNAs.
4918 Nucleic Acids Research, 2004, Vol. 32, No. 16

Page 5

The G-G base pair and the adjacent U-turn are well
conserved in loop L2 of all but archaeal SRP RNAs
(Figure 2, blue and green). In these organisms, the boundaries
of the loop L2 are well defined by these structural elements.
Since the G-G base pair is absent in archaeal RNAs,
the boundaries of loop L2 remain uncertain and the putative
U-turn motif may, therefore, precede the one shown. In a few
species, the uridine is replaced by a cytidine, which may also
introduce U-turn-like bends as revealed by the pseudoknot
structure of beet western yellow virus RNA (37). In four
Arabidopsis species, the G-G base pair is replaced by a
G-A base pair.
The potential to form base pairs is conserved in all SRP
RNAs with a remarkable bias for G-C pairs (Figure 2, red
nucleotides). More strikingly, in animal metazoans even the
primary sequences of the nucleotides are conserved. In plants,
only 2 bp may form between the loops because of the smaller
size of loop L1.2. In contrast, in Archaea and Eubacteria the
increased size of loop L2 expands the predicted base comple-
mentarities between the loops. In these organisms, helix H1.2
is not well defined, since the loop-adjacent strands are purine-
rich on both sides and it remains uncertain to which degree
they may form non-canonical base pairs to extend the four
Watson–Crick base pairs.
In summary, the high conservation of loop–loop base
pairing and of structural determinants in loop L2 points
towards a role of these structures in SRP assembly and
function. However, apart from base pairing, the shape and
size of the tertiary structure is expected to vary considerably,
since the numberof base pairs and the sizes ofthe loops are not
conserved between different RNAs.
Mutations in loop L2 induce conformational changes
in the RNA Alu domain
ToexaminethefunctionsofthetertiarybasepairsinhumanSRP
RNA, we changed bases in single loops to disrupt base pairing
andinbothloopstorestorebasepairingwithadifferentsequence
(see Table 1). Guanidine was replaced by cytidine and vice
versa. The RNAs were synthesized in vitro. They were never
denatured and were purified under native conditions to avoid
refoldingartefacts.Themutatedandwild-typesyntheticRNAs
migrated as a single band at the position expected of their size
in denaturing gel electrophoresis (Figure 3A, lower panel).
To examine the conformation of the synthetic RNA var-
iants, they were displayed in parallel with native canine SRP
RNA by native PAGE (Figure 3A). Native canine SRP RNA
migrated as a single band. In vitro synthesized RNAs are more
heterogeneous, containing RNA aggregates and RNAs with
Table 1. Mutations in SRP RNA
Wild type
WTLoop L2
Loop L1.2
50-G13
30-C37-(A36-U35)-C34-G33-50
— G14-C15-30
Mutations in a single loop
2L2 Loop L2
Loop L1.2
Loop L2
Loop L1.2
Loop L2
Loop L1.2
Loop L2
Loop L1.2
50-C13
30-C37-(A36-U35)-C34-G33-50
50-C13—
30-C37-(A36-U35)-C34-G33-50
50-G13—
30-G37-(A36-U35)-G34-G33-50
50-G13—
30-G37-(A36-U35)-G34-C33-50
—C14-C15-30
3L2C14-G15-30
2L1.2G14-C15-30
3L1.2G14-C15-30
Compensatory mutations in both loops
2CompLoop L2
Loop L1.2
Loop L2
Loop L1.2
50-C13
30-G37-(A36-U35)-G34-G33-50
50-C13—
30-G37-(A36-U35)-G34-C33-50
—C14-C15-30
3Comp C14-G15-30
Figure 3. Analysis of wild-type and mutated synthetic SRP RNAs. (A) Native
(upper panel) and denaturing (lower panel) 6% PAGE. Equal amounts of RNA
wereloadedonbothgels.TheRNAswerevisualizedbystainingwithGelstar1.
SRP RNA was extracted from purified canine SRP. The synthetic RNAs are
labelled as shown in Table 1. (B) Limited V1 ribonuclease digestion
experiments. The digestion products were displayed by 10% denaturing
PAGE and the RNA fragments visualized with ethidium bromide staining.
The bracket highlights the region that contains RNAs obtained by single
cleavages in the Alu portion of SRP RNA. In, 50% of the RNA used in the
experiments.
Nucleic Acids Research, 2004, Vol. 32, No. 16 4919