A translational repression assay procedure (TRAP) for RNA-protein interactions in vivo.
ABSTRACT RNA-protein interactions are central to many aspects of cellular metabolism, cell differentiation, and development as well as the replication of infectious pathogens. We have devised a versatile, broadly applicable in vivo system for the analysis of RNA-protein interactions in yeast. TRAP (translational repression assay procedure) is based on the translational repression of a reporter mRNA encoding green fluorescent protein by an RNA-binding protein for which a cognate binding site has been introduced into the 5' untranslated region. Because protein binding to the 5' untranslated region can sterically inhibit ribosome association, expression of the cognate binding protein causes significant reduction in the levels of green fluorescent protein fluorescence. By using RNA-protein interactions with affinities in the micromolar to nanomolar range, we demonstrate the specificity of TRAP as well as its ability to recover the cDNA encoding a specific RNA-binding protein, which has been diluted 500,000-fold with unrelated cDNAs, by using fluorescence-activated cell sorting. We suggest that TRAP offers a strategy to clone RNA-binding proteins for which little else than the binding site is known, to delineate RNA sequence requirements for protein binding as well as the protein domains required for RNA binding, and to study effectors of RNA-protein interactions in vivo.
- [Show abstract] [Hide abstract]
ABSTRACT: RNA-protein interaction plays a significant role in regulating eukaryotic translation. This phenomenon raises questions about the ability of artificial biological systems to take the advantage of protein-RNA interaction. Here, we designed an oncogenic signal-processing system expressing both a Renilla luciferase reporter gene controlled by RNA-protein interaction in its 5'-untranslated region (5'-UTR) and a Firefly luciferase normalization gene. To test the ability of the designed system, we then constructed vectors targeting the nuclear factor-κB (NF-κB) or the β-catenin signal. We found that the inhibition (%) of luciferase expression was correlated to the targeted protein content, allowing quantitative measurement of oncogenic signal intensity in cancer cells. The systems inhibited the expression of oncogenic signal downstream genes and induced bladder cancer cell proliferation inhibition and apoptosis without affecting normal urothelial cells. Compared to traditional methods (ELISA and quantitative immunoblotting), the bio-systems provided highly accurate, consistent, and reproducible quantification of protein signals and were able to discriminate between cancerous and non-cancerous cells. In conclusion, the synthetic systems function as both "signal counters" and "signal blockers" in cancer cells. This approach provides a synthetic biology platform for oncogenic signal measurement and cancer treatment.Molecular BioSystems 04/2013; · 3.35 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Artificial genetic switches have been designed and tuned individually in living cells. A method to directly invert an existing OFF switch to an ON switch should be highly convenient to construct complex circuits from well-characterized modules, but developing such a technique has remained a challenge. Here we present a cis-acting RNA module to invert the function of a synthetic translational OFF switch to an ON switch in mammalian cells. This inversion maintains the property of the parental switch in response to a particular input signal. In addition, we demonstrate simultaneous and specific expression control of both the OFF and ON switches. The module fits the criteria of universality and expands the versatility of mRNA-based information processing systems developed for artificially controlling mammalian cellular behaviour.Nature Communications 09/2013; 4:2393. · 10.74 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Genetic devices that directly detect and respond to intracellular concentrations of proteins are important synthetic biology tools, supporting the design of biological systems that target, respond to or alter specific cellular states. Here, we develop ribozyme-based devices that respond to protein ligands in two eukaryotic hosts, yeast and mammalian cells, to regulate the expression of a gene of interest. Our devices allow for both gene-ON and gene-OFF response upon sensing the protein ligand. As part of our design process, we describe an in vitro characterization pipeline for prescreening device designs to identify promising candidates for in vivo testing. The in vivo gene-regulatory activities in the two types of eukaryotic cells correlate with in vitro cleavage activities determined at different physiologically relevant magnesium concentrations. Finally, localization studies with the ligand demonstrate that ribozyme switches respond to ligands present in the nucleus and/or cytoplasm, providing new insight into their mechanism of action. By extending the sensing capabilities of this important class of gene-regulatory device, our work supports the implementation of ribozyme-based devices in applications requiring the detection of protein biomarkers.Nucleic Acids Research 10/2014; · 8.81 Impact Factor
Proc. Natl. Acad. Sci. USA
Vol. 95, pp. 951–956, February 1998
A translational repression assay procedure (TRAP) for
RNA–protein interactions in vivo
EFROSYNI PARASKEVA*, ANN ATZBERGER, AND MATTHIAS W. HENTZE†
European Molecular Biology Laboratory, Meyerhofstrasse 1, D-69117 Heidelberg, Germany
Communicated by Kai Simons, European Molecular Biology Laboratory, Heidelberg, Germany, November 17, 1997 (received for review
September 4, 1997)
many aspects of cellular metabolism, cell differentiation, and
development as well as the replication of infectious pathogens.
We have devised a versatile, broadly applicable in vivo system
for the analysis of RNA–protein interactions in yeast. TRAP
(translational repression assay procedure) is based on the
translational repression of a reporter mRNA encoding green
fluorescent protein by an RNA-binding protein for which a
cognate binding site has been introduced into the 5? untrans-
lated region. Because protein binding to the 5? untranslated
region can sterically inhibit ribosome association, expression
of the cognate binding protein causes significant reduction in
the levels of green fluorescent protein fluorescence. By using
RNA–protein interactions with affinities in the micromolar to
nanomolar range, we demonstrate the specificity of TRAP as
well as its ability to recover the cDNA encoding a specific
RNA-binding protein, which has been diluted 500,000-fold
with unrelated cDNAs, by using fluorescence-activated cell
sorting. We suggest that TRAP offers a strategy to clone
RNA-binding proteins for which little else than the binding
site is known, to delineate RNA sequence requirements for
protein binding as well as the protein domains required for
RNA binding, and to study effectors of RNA–protein interac-
tions in vivo.
RNA–protein interactions are central to
RNA–protein interactions govern a plethora of biological
differentiation, human intelligence, and viral replication (1).
In many cases, RNA-binding proteins exert their functions in
cells or tissues that are not readily amenable to biochemical
analysis, such as specific areas of the brain or in the germ line.
The paucity of biochemical material imposes cumbersome
limitations to the identification and cloning of biologically
important RNA-binding proteins operating in such systems,
particularly when confounded by a lack of possible genetic
To circumvent such limitations, alternative strategies to
identify and study RNA–protein interactions have been de-
vised, based on phage display (2), transcription termination
(3), or translation (4) in Escherichia coli, or on transcription
initiation in yeast (5, 6). We have developed an approach to
study RNA–protein interactions in the cytoplasm of the uni-
cellular eukaryote Saccharomyces cerevisiae. This approach is
based on the realization that protein binding to specific sites
near the 5? end of an eukaryotic mRNA causes its translation
to be repressed both in mammalian cells and in yeast (7–9).
Because of its steric nature, translational repression appears
not to be restricted by the physiological function of the
RNA-binding proteins. We will refer to this approach as
TRAP (for translational repression assay procedure). By using
three different well-characterized RNA–protein interactions,
the binding of the spliceosomal protein U1A to loop 2 of U1
snRNA (10), the binding of iron regulatory protein (IRP)-1 to
iron-responsive elements (IREs) (11), and the the binding of
bacteriophage MS2 coat protein to the MS2 replicase mRNA
(12, 13), we demonstrate that TRAP allows the rapid and
specific identification of RNA–protein interactions in vivo and
cloning of cDNAs encoding a specific, cognate RNA-binding
MATERIALS AND METHODS
Plasmid Construction. Green fluorescent protein (GFP)
indicator plasmids YCp22F.bs-GFP are driven by the TEF1
(translation elongation factor) promoter and contain the
TRP1 selection marker. For the construction, the correspond-
ing YCp22FL plasmids (8, 14) were digested with NdeI, the
of E. coli DNA polymerase I, before an additional digestion
with XbaI and gel purification using QIAquick columns (Qia-
gen). The Ser-65-?Thr mutant GFP coding sequence was
amplified by PCR (Boehringer Mannheim PCR Master kit)
with primers 5?-TTTTTAATTATGAGCAAAGGAGAAG-
AAGAACTTTTC-3? and 5?-TTTTTTCTAGATTATTTG-
TATAGTTCATCCATG-3?, digested with XbaI, and then
introduced into the NdeI blunted-XbaI plasmids to create
plasmids YCp22F.IREwt-GFP, YCp22F.IREmt-GFP,
YCp22F.U1Awt-GFP, YCp22F.U1Amt-GFP, YCp22F.MSC-
GFP, YCp22F.MSA-GFP, and YCp22F.MSAdel-GFP.
Plasmids YCpIRP-1 [described as YCpIRF by Oliveira et al.
(14)], YCpU1A (8), and YCpCP (8) express IRP-1, U1A, and
MS2 coat protein, respectively, under the control of PGK?
GAL fusion promoter and contain URA3 as a selection
Yeast Culture Conditions and Preparation of Protein Ex-
tracts. The S. cerevisiae haploid strain RS453 (ade 2–1, trp 1–1,
leu 2–3, his 3–11, ura 3–52, lys?, can 1–100) was grown at 30°C
in YPD medium (1% peptone?1% yeast extract?2% glucose)
or YPD-agar plates. The yeast selective defined media con-
tained 0.67% (wt?vol) yeast nitrogen base without amino
acids, 2% glucose or 2% galactose, amino acids and nucleo-
tides as follows: adenine 20 mg?liter, arginine 20 mg?liter,
histidine 20 mg?liter, leucine 60 mg?liter, tryptophan 20
mg?liter, and uracil 20 mg?liter. Solid defined media prepa-
ration followed exactly the protocol above with the addition of
Yeast transformation was performed by the lithium acetate
method (15). For the mixed plasmid transformation,
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
© 1998 by The National Academy of Sciences 0027-8424?98?95951-6$2.00?0
PNAS is available online at http:??www.pnas.org.
Abbreviations: TRAP, translational repression assay procedure; GFP,
green fluorescent protein; UTR, untranslated region; FACS, fluores-
cence activated cell sorting; IRP, iron regulatory protein; IRE, iron
responsive element; CP, coat protein.
*Present address: Zentrum fu ¨r Moleculare Biologie Heidelberg, Im
Neuenheimer Feld 282, D-69120 Heidelberg, Germany.
†To whom reprint requests should be addressed. e-mail: hentze@
YCpIRP-1 and YCpU1A were mixed in 1:500,000 ratio (DNA
concentration was estimated by measuring the absorption at
260 nm and by ethidium bromide staining). The mixed plas-
mids were used to transform RS453 cells carrying plasmid
YCp22F.IREwt-GFP. Transformants were selected on plates
lacking tryptophan and uracil. Approximately 3 ? 106trans-
formants were pooled in liquid selective medium and used for
sorting the cells expressing IRP-1, as described below.
For induction of binding protein expression, cultures were
grown overnight in selective defined media containing 2%
galactose. Yeast protein extracts were prepared as follows:
yeast cultures of 10–20 ml were harvested by centrifugation.
The pellet was washed once with 10 ml of sterile distilled water
and resuspended in 1 ml of yeast lysis buffer (100 mM NaCl?50
mM Tris?HCl, pH 7.4). The cell suspension was briefly cen-
trifuged again and resuspended in 150 ?l of lysis buffer
containing 1 mM phenylmethylsulfonyl fluoride, 10 mg?ml
leupeptin, and 1 mM DTT. Cells were disrupted with glass
beads (0.45–0.50 mm) at 4°C by three cycles of 1-min vortexing
and 1-min incubation on ice, followed by centrifugation for 2
min at 6,000 g in a microfuge. The supernatant was centrifuged
again for 5 min at 1,500 g, transferred into a fresh tube and
stored at ?80°C. Protein concentration of the extracts was
determined by using the Bio-Rad protein assay.
Gel Retardation Assays.32P-labeled IRE and U1snRNA
synthesized in vitro from XbaI-linearized plasmids IREwt-
CAT (16) and U1Awt-CAT (7) with T7 RNA polymerase
(Stratagene). Aliquots (10–40 mg each) of yeast extracts were
incubated for 20 min at room temperature for IRP-1 or ice for
U1A with 1 ng of probe in 10 ml of lysis buffer. Subsequently,
samples were incubated for 10 min with heparin (0.5 mg?ml)
and analyzed by nondenaturing gel electrophoresis (17).
Fluorescence-Activated Cell Analysis and Sorting. A FAC-
Scan (Becton Dickinson) flow cytometer was used to analyze
the fluorescence levels of cells. The FACScan uses an argon
ion laser fixed at 488 nm for excitation and a 530-nm bandpass
were made of the cell size, internal complexity, and relative
fluorescence intensity of single cells. Debris were excluded
from the analysis by gating on the forward scatter versus side
For sorting, the cells were analyzed and isolated on a FACS
Vantage (Becton Dickinson) cell sorter. Excitation was at 488
nm, and the emitted fluorescence was collected with a 530-nm
bandpass filter. Sorting gates were set by using forward scatter
width versus side scatter height signals, to exclude debris and
clumps, and forward scatter height versus fluorescence height
signal, to sort cells that fell within a certain fluorescence
The Principle of TRAP. Fig. 1A outlines the functional
principle of TRAP: RS453 cells (genotype ade 2–1, trp 1–1, leu
2–3, his 3–11, ura 3–52, lys?, can 1–100) are transformed with
plasmids for the expression of the RNA-binding protein (or a
cDNA expression library) from a galactose-inducible pro-
moter (that is repressed in the presence of glucose), and the
GFP indicator (18) bearing the protein binding site in the 5?
untranslated region (UTR) of the encoded mRNA from a
constitutively active promoter. Both plasmids are centromeric
and maintained at 1–2 copies per cell. In the presence of
glucose, GFP is expressed and yields high fluorescence levels.
Replacement of glucose by galactose in the growth medium
activates the PGK?GAL promoter, and the expression of the
RNA-binding protein is induced. As a consequence, GFP
expression is translationally inhibited, and the fluorescence
level of the cells bearing the cDNA expression vector for the
cognate binding protein is diminished. This effect can be
quantitated by flow cytometry, cells displaying different GFP
levels can be separated by fluorescence activated cell sorting
(FACS), and viable cells recovered for additional analysis or
cycles of sorting.
Specificity of TRAP. The specificity of TRAP was chal-
lenged with three well-characterized RNA–protein interac-
tions: IRE?IRP-1, U1 snRNA loop 2?U1A, and the binding of
bacteriophage MS2 coat protein to the MS2 replicase mRNA.
The coding sequence for the S65T mutant GFP, which displays
improved excitation?emission characteristics for fluorometry
and FACS (19, 20), was subcloned into reporter gene plasmids
to generate YCp22F.IREwt-GFP, YCp22F.IREmt-GFP,
YCp22F.U1Awt-GFP, and YCp22F.U1Amt-GFP (Fig. 1 B
and C). Yeast cells were cotransformed with one of these GFP
reporter plasmids and either YCpIRP-1 or YCpU1A (Fig. 1B).
Double transformants were selected on plates lacking trypto-
phan and uracil, grown in selective liquid medium containing
TRAP. RS453 cells are transformed with the GFP indicator plasmid
that carries within the GFP 5?UTR the binding site of the protein
under study and the plasmid for expression of the RNA-binding
protein (or a cDNA library) under the control of a galactose-inducible
promoter. TRAP: In galactose medium, expression of the RNA-
binding protein is induced. The RNA–protein interaction at the 5? end
of the GFP mRNA represses its translation and the fluorescence of the
cell is diminished (cognate interaction). Controls: GFP mRNA trans-
lation yields high fluorescence levels when expression of the RNA-
binding protein is repressed in glucose medium (no binding protein
expression), or when the expressed protein cannot interact with the
GFP mRNA (noncognate binding protein or mutated binding site).
(B) Description of the plasmids used. The GFP reporter plasmids
YCp22F.BS-GFP contain the GFP S65T mutant ORF. The IRE and
U1 snRNA loop 2 wild type and mutated, as well as the MSC, MSA,
and MSAdel binding sites (BS) are cloned into the AflII site. The
binding sites are located 9 nucleotides downstream from the tran-
scription start site and 32 nucleotides upstream from the GFP trans-
lation initiation codon. The plasmids harbor a TRP1 selection marker.
YCpU1A, YCpIRP-1, and YCpCP are used for the galactose-
inducible expression of U1A, IRP-1, and MS2 coat protein (CP),
respectively. They contain a URA3 selection marker. (C) Sequences
and nomenclature of the GFP reporter plasmids. Binding site inser-
tions are represented by capital letters. Differences between wild-type
and mutated binding sites are indicated by arrows.
Schematic representation of TRAP. (A) The principle of
952Biochemistry: Paraskeva et al.Proc. Natl. Acad. Sci. USA 95 (1998)
either 2% glucose or 2% galactose for 14 hr to a density of
OD600 0.5–1, and GFP fluorescence was analyzed by flow
cytometry (Fig. 2). Cells grown in glucose and expressing GFP
from YCp22F.bs-GFP plasmids display high fluorescence
compared with control cells lacking a GFP reporter plasmid
(Fig. 2 a and d). Induction of the cognate RNA-binding
proteins in galactose-containing media significantly decreases
the mean fluorescence, manifested by a leftward shift of the
fluorescence distribution curve toward lower values (Fig. 2 b
and e) and the repression of the mean fluorescence of cells
(Table 1). This reduction specifically requires the presence of
the wild-type binding site and the cognate binding protein and
is not seen with either the mutant binding sites (Fig. 2 c and
f) or the noncognate binding proteins (Fig. 2 a and d).
Whereas the IRE?IRP-1 and the U1 snRNA loop 2?U1A
interactions display binding affinities in the low nanomolar
range (refs. 21–26, Table 1), many specific and physiologically
relevant RNA–protein interactions display lower binding af-
finities. To evaluate the utility of TRAP to study such lower
binding affinity interactions by using a well-characterized
example, defined binding sites for the MS2 coat protein (MS2
CP) in the subnanomolar [MSC, (12, 13)] and low micromolar
[MSA, 13)] range were introduced into the GFP reporter
plasmid and compared with a negative control (plasmid
YC22F.MSAdel-GFP, Fig. 1 B and C). Yeast cells were
cotransformed with one of the three GFP reporter plasmids
and the MS2 coat protein expression plasmid YCpCP (Fig.
1B). Induction of MS2 CP expression in galactose-containing
media resulted in specific repression of GFP expression (Table
1), whereas the MSAdel construct was unaffected (data not
shown). Importantly, the lower affinity MSA?MS2 CP inter-
action caused a 50% repression of GFP fluorescence, which
even exceeded the effect caused by the higher affinity U1 loop
2?U1A interaction (see Discussion). We conclude that TRAP
allows study of specific RNA–protein interactions with affin-
ities spanning a broad physiological range. Interestingly, for a
given RNA-binding protein the degree of repression appears
to reflect its binding affinity although different RNA–protein
interactions cannot be quantitatively compared, presumably
because additional parameters affect the degree of repression
Isolation of Cells Expressing Cognate RNA-Binding Pro-
teins by FACS. The specific reduction of GFP fluorescence
caused by the interaction of cognate RNA-binding proteins
with their wild-type binding sites introduced into the 5? UTR
of GFP reporter mRNAs (Fig. 2) suggests the possibility to
recover cells expressing the cognate RNA binding protein
from a background pool of cells expressing control proteins by
FACS. Cells cotransformed with plasmid YCp22F.IREwt-
GFP and either YCpIRP-1 or YCpU1A were mixed at a ratio
of 1:10,000. The mixed cell suspension was grown in galactose-
containing medium to induce the expression of the RNA-
binding proteins and sorted by FACS according to their GFP
fluorescence. Cells displaying low fluorescence levels were
regrown in galactose medium and sorted for a second time.
The enrichment for cells carrying plasmid YCpIRP-1 was
tested in a gel retardation assay for IRE-binding activity by
using extracts prepared from sorted cells (Fig. 3A). After two
rounds of sorting (F2), the extracts from galactose-induced
cultures of cells selected for low fluorescence levels (Fig. 3A,
F2?, lane 5) display IRE-binding activity comparable to that
indicator plasmid and either plasmid YCpIRP-1 or YCpU1A were grown in glucose or galactose containing medium. The fluorescence of cells
grown in glucose (green curves) or galactose (red curves) media was analyzed by flow cytometry. Control cells not carrying a GFP plasmid are
represented by the black curves in a and d. Fluorescence drops when expression of the cognate binding protein (b and e) is induced by the addition
of galactose. Expression of a noncognate binding protein (a and d) does not affect fluorescence levels. No change of fluorescence is observed with
a mutated binding site (c and f) present within the 5?UTR of GFP mRNA. The fluorescence intensity is plotted against counts (number of cells
analyzed per channel of fluorescence).
Fluorescence of GFP-expressing cells is specifically reduced by cognate RNA–protein interactions. Cells cotransformed with a GFP
fluorescence by RNA–protein interactions with different
Quantitative analysis of the repression of GFP
1.0–10 nM (21–23)
0.02–80 nM (24–26)
0.02–0.1 nM (12, 13)
0.1–1 mM (13)
U1 loop 2?U1A
MSC?MS2 coat protein
MSA?MS2 coat protein
*Binding affinities for the different RNA–protein interactions as
†GFP fluorescence of cells grown in glucose media was defined as
100%. The remaining mean fluorescence of cells after growth in
galactose media was expressed as a percentage of the above, and the
repression level calculated by substraction from 100%.
Biochemistry: Paraskeva et al.Proc. Natl. Acad. Sci. USA 95 (1998) 953
mixed at a ratio of 1:100 (not shown). No binding activity was
detectable in the initial 1:10,000 starting population (Fig. 3A,
M, lane 1), in extracts from cells selected for high fluorescence
levels (Fig. 3A, ?, lanes 2 and 4), or sorted only once (Fig. 3A,
F1?, lane 3). The F2? cells subsequently were regrown and
sorted again (F3) into four different pools of descending
fluorescence (R2-R5, Fig. 3B). R2-R5 extracts were compared
in a gel retardation assay to the F2 extracts (Fig. 3C). A
substantial further enrichment in IRE-binding activity is ap-
parent in cell populations exhibiting low fluorescence, R4 (Fig.
made from cultures of these clones tested for IRP-1 expression
(Fig. 3D). Ten of 10 clones tested from R4 expressed IRP-1
(Fig. 3D). Thus, starting from 1:10,000 (0.01%) cells carrying
plasmid YCpIRP-1, the IRP-1 expressing cells represented
100% of the cells analyzed from F3?R4.
We next performed an analogous experiment for the U1A-
encoding plasmid with YCp22F.U1Awt-GFP, by using a
10,000-fold excess of YCpIRP-1 as a background (Fig. 4 A–D).
After two rounds of initial sorting (F1, F2, Fig. 4A), the F2?
cells were regrown and sorted into regions (F3, R2-R5, Fig.
4B). As before, the strongest enrichment was observed in R4,
but there was substantially less U1A activity in R3 and R5 (Fig.
4C). Single colony analysis of F3?R4 revealed successful
cloning of YCpU1A in eight of 10 samples tested (Fig. 4D).
Taken together, the results obtained for IRP-1 and U1A
underscore the potential of TRAP to be used for the identi-
fication and recovery of cells expressing an RNA-binding
protein from a population of cells expressing unrelated pro-
Cloning by TRAP. To further challenge the utility of TRAP
for cloning, we attempted to reclone the IRP-1 cDNA. To
mimic the cloning process with a precisely preset required
enrichment factor, we mixed the plasmids YCpIRP-1 and
YCpU1A in a 1:500,000 ratio and transformed RS453 cells
carrying YCp22F.IREwt-GFP. Double transformants were
selected on plates lacking tryptophan and uracil, ?3 ? 106
transformants were pooled, expanded in selective medium
containing galactose to induce expression of the binding
proteins, and sorted by FACS. After two rounds of sorting,
cells from culture F2? showed IRE-binding activity (Fig. 5A,
lane 3). F2? cells were further sorted into regions of cells with
different fluorescence (Fig. 5B), and the greatest further
enrichment was again found in R4 (Fig. 5A, lane 6). When
single colonies from F3?R4 were analyzed, 1 of 20 expressed
IRP-1 (not shown). Thus, TRAP allowed successful recloning
of the cDNA encoding a cognate RNA-binding protein from
a 500,000-fold background.
The observation that RNA–binding proteins can sterically
repress translation from binding sites introduced close to the
5? end of a mRNA has paved the way for the development of
TRAP. The physiological principle (7, 8) appears to apply to
organisms as diverse as mammals and yeast, and not to be
limited by the origin or physiological functions of the RNA-
binding proteins under study. Our demonstration that a cDNA
can be recloned from a 0.5 million-fold background (Fig. 5)
indicates the potential of TRAP for the cloning of RNA-
binding proteins for which little other than the binding site
with U1A expressing cells by TRAP. (A) Cells cotransformed with
plasmid YCp22F-IREwt.GFP and either YCpIRP-1 or YCpU1A were
mixed in a ratio of 1:104and grown overnight in galactose-containing
medium (M). Cells displaying high (F1?) or low fluorescence levels
(F1?) were separated by FACS. Low fluorescent cells were regrown
and sorted for a second time (F2? and F2?). Total cell extracts were
incubated with 1 ng of IRE probe and analyzed by native PAGE. (B)
Dot plot profile of cells from F2?. Cells were analyzed by flow
cytometry, plotted according to forward scatter (size) vs. fluorescence
intensity and sorted for a third time (F3) to regions R2-R5 according
to their fluorescence. (C) Extracts from cells sorted twice (F2?, lane
1) and three times (F3, lanes 2–5) were incubated with 1 ng of IRE
probe and analyzed by native PAGE. (D) Extracts from single cell
clones of culture R4 were analyzed in a gel retardation assay (lanes
IRP-1 expressing cells can be recovered from a mixture
IRP-1 expressing cells by TRAP. (A) Cells cotransformed with
plasmid YCp22F-U1Awt.GFP and either YCpU1A or YCpIRP-1
were mixed in 1:104ratio and grown overnight in galactose-containing
medium (M). Cells displaying low fluorescence levels (F1?) were
isolated by FACS, regrown, and sorted for a second time (F2?). Total
cell extracts were incubated with 1 ng of U1A probe and analyzed by
native PAGE. (B) Dot plot profile of cells from F2?. Cells were
analyzed by flow cytometry, plotted according to forward scatter (size)
vs. fluorescence intensity and sorted for a third time (F3) into regions
R2-R4 according to their fluorescence. (C) Extracts from cells sorted
twice (F2?, lane 1) and three times (F3, lanes 2–5) were incubated
with 1 ng of U1A probe and analyzed by native PAGE. (D) Extracts
from single cell clones of culture R4 were analyzed in a gel retardation
assay (lanes 1–10).
U1A expressing cells can be recovered from a mixture with
954Biochemistry: Paraskeva et al.Proc. Natl. Acad. Sci. USA 95 (1998)
needs to be known. Because we have not attempted recloning
from a higher than 0.5 million-fold excess of unrelated plas-
mids, we have no reason to expect this figure to reflect the
The specific features of TRAP suggest that it comple-
ments the repertoire of recently reported strategies for
studying RNA-binding proteins (2–6) in several useful ways:
it represents a system that combines the use of a eukaryotic
host cell with the advantages implicit in assaying cytoplasmic
(rather than nuclear) expression and the possibilities af-
forded by successive rounds of enrichment. TRAP allows the
study of RNA-binding proteins in their native form without
having to fuse additional protein domains that might inter-
fere with their RNA-binding affinity or specificity. Although
TRAP detects the RNA-binding proteins in the cytoplasm,
it also was used successfully with the nuclear protein U1A
(Figs. 2 and 4), probably because sufficient amounts of newly
synthesized, overexpressed U1A remain cytoplasmic to af-
fect the translation of the GFP indicator mRNA. To clone or
study RNA binding proteins with more potent nuclear
import signals, such signals could be mutated (if known) or
the expression library modified by using a vector that directs
the expression of fusion proteins with the powerful leucine-
rich nuclear export signal of PKI (27). TRAP only requires
two components: the RNA-binding protein (or a suitable
cDNA expression library) and the indicator mRNA. This
simplicity minimizes the potential for inadvertent nonspe-
cific effects on the assay, a consideration that may be
particularly relevant when studying pharmacological effec-
tors of RNA-protein interactions. Furthermore, the utiliza-
tion of FACS permits the rapid processing of large numbers
of independent clones and the recovery of living cells after
sorting, thus allowing for multiple rounds of sorting while
monitoring enrichment. The highest level of specific enrich-
ment for both U1A and IRP-1 was found in the R4 fractions,
i.e., cells displaying reduced but not minimal fluorescence.
The minimally fluorescent population (R5) is possibly con-
taminated by cells that may harbor mutations in the GFP
cDNA or may have lost the GFP-expressing plasmid, even
during culture in selective medium. However, the design of
TRAP allows counterselection against nonspecific, consti-
tutive loss of GFP fluorescence, because cells displaying a
specific reduction in fluorescence specifically shift to and can
be recovered from the high fluorescent pool following
incubation in glucose-containing media (data not shown).
What are the current limitations of TRAP? The affinities
of the RNA–protein interactions tested have been reported
to be approximately 1.0–10 nM for IRE?IRP-1 (21–23),
0.02–80nM for U1 loop 2?U1A (24–26), 0.02–0.1 nM for
MSC?MS2 coat protein (12, 13), and 0.1–1.0 ?M for MSA?
MS2 coat protein (13) interactions. As shown in Table 1 and
Fig. 2, the affinities of the interaction of IRP-1 and U1A with
YCp22F.U1Amt-GFP, respectively, and of MS2 coat protein
for the MSAdel mutant (Fig. 1C and data not shown) are not
sufficient to cause diminished GFP fluorescence. The min-
imally required affinity is currently unknown. It certainly will
be subject to case-specific parameters such as the stability of
the corresponding RNA-binding protein and the maximal
level of its overexpression in yeast, its distribution between
different subcellular compartments, or possible complex-
ation with other proteins. Such effects probably account for
the lower degree of GFP repression by the high affinity U1
loop 2?U1A interaction compared with the lower affinity
MSA?MS2 CP interaction (Table 1). Although TRAP thus
appears suitable to study specific RNA–protein interactions
in a broad, physiological range, our findings suggest that
nonspecific and?or low affinity RNA–protein interactions do
not affect the performance of TRAP. This consideration is
supported by the lack of translational repression of IREmt,
U1Amt, and MSAdel binding site controls as well as the
ability to observe a specific interaction with an affinity in the
low micromolar range in a cytoplasmic compartment replete
with nonspecific RNA binding proteins. Because the three
proteins tested here bind to hairpin structures, we currently
cannot conclude that proteins binding to less structured
RNA motifs will perform equally well. Like the other
strategies to detect RNA–protein interactions (2–6), TRAP
requires that the protein under investigation be able to bind
RNA as a monomer or homopolymer. The specific principle
of TRAP also requires the introduction of the RNA-binding
region into the 5? UTR not to interfere with GFP expression
per se. Therefore, binding regions that are exceedingly highly
structured or harbor inhibitory ORFs may not represent
suitable target sequences. Ideally, the binding region should
be defined to 100 nucleotides or less, to minimize the
probability of introducing inhibitory features into the 5?
UTR, and also to increase the probability of introducing the
actual binding site close to the 5? end of the mRNA, a region
that allows maximal translational repression at least in
mammalian cells (7, 28, 29). However, a recent analysis of the
IRE?IRP-1 interaction in yeast has suggested that 5? end
proximity of the RNA binding site may not be a required
feature for translational repression in S. cerevisiae (30).
In summary, TRAP offers a versatile approach for studying
RNA–protein interaction pairs in yeast. The principle of
TRAP should be adaptable to other eukaryotes, including
mammalian cells. In addition to cloning novel RNA-binding
proteins, we envisage its utility to demonstrate specific inter-
actions between a protein and an RNA sequence in vivo, to
evaluate the impact of mutations of an RNA-binding protein
and?or the RNA-binding site on the interaction, and to
monitor the effect of pharmacological agents on known RNA–
We gratefully acknowledge Graham Smith for his help with some
initial cell sorting experiments and Kostas Pantopoulos for supplying
discussions and suggestions, and Anne Ephrussi, Juan Valcarcel, and
the members of the Hentze group for comments on the manuscript.
M.W.H. acknowledges support from the European Commission Bio-
technology Program (BIO-CT95-0045). E.P. was the recipient of a
expressing cells from a transformation with mixed plasmid DNA.
RS453 cells carrying plasmid YCpIREwt-GFP were transformed with
a DNA mixture containing plasmids YCpIRP-1 and YCpU1A in a
ratio of 1:500,000. Transformants were pooled and grown in galactose
medium. Cells with low fluorescence were sequentially sorted twice.
Total extracts were analyzed in a gel retardation assay with 1 ng of IRE
probe (lanes 1–3). The extract of cells sorted twice shows IRE binding
activity (lane 3). IRE binding activity is below detection limit in the
extract from pooled transformants (lane 1), or cells sorted once (lane
2). Cells from F2? were sorted for a third time (F3) into populations
R2-R5. The IRE binding activity of extracts from cultures R2-R5 was
analyzed (lanes 4–7). The highest activity is present in extract R4. (B)
Fluorescence analysis and sorting of cells from F2? into regions
R2-R4 according to their fluorescence.
Cloning of IRP-1 by TRAP. (A) Recovery of IRP-1
Biochemistry: Paraskeva et al.Proc. Natl. Acad. Sci. USA 95 (1998)955
Human Capital and Mobility fellowship from the European Commis-
Burd, C. G. & Dreyfuss, G. (1994) Science 265, 615–621.
Laird-Offringa, I. A. & Belasco, J. G. (1995) Proc. Natl. Acad.
Sci. USA 92, 11859–11863.
Harada, K., Martin, S. S. & Frankel, A. D. (1996) Nature
(London) 380, 175–179.
Jain, C. & Belasco, J. G. (1996) Cell 87, 115–125.
SenGupta, D. J., Zhang, B., Kraemer, B., Pochart, P., Fields, S.
& Wickens, M. (1996) Proc. Natl. Acad. Sci. USA 93, 8496–8501.
Putz, U., Skehel, P. & Kuhl, D. (1996) Nucleic Acids Res. 24,
Stripecke, R. & Hentze, M. W. (1992) Nucleic Acids Res. 20,
(1994) Mol. Cell. Biol. 14, 5898–5909.
Gray, N. K. & Hentze, M. W. (1994) EMBO J. 13, 3882–3891.
Scherly, D., Boelens, W., van Venrooij, W. J., Dathan, N. A.,
Hamm, J. & Mattaj, I. W. (1989) EMBO J. 8, 4163–4170.
Hentze, M. W. & Ku ¨hn, L. C. (1996) Proc. Natl. Acad. Sci. USA
Lowary, P. T. & Uhlenbeck, O. C. (1987) Nucleic Acids Res. 15,
Witherell, G. W., Gott, J. M. & Uhlenbeck, O. C. (1991) Progr.
Nucleic Acids Res. Mol. Biol. 40, 185–220.
Oliveira, C. C., Goossen, B., Zanchin, N. I. T., McCarthy, J. E. G.,
Hentze, M. W. & Stripecke, R. (1993) Nucleic Acids Res. 21,
Ito, H., Fukuda, Y., Murata, K. & Kimura, A. (1983) J. Bacteriol.
16.Gray, N. K., Quick, S., Goossen, B., Constable, A., Hirling, H.,
Ku ¨hn, L. C. & Hentze, M. W. (1993) Eur. J. Biochem. 218,
Leibold, E. A. & Munro, H. N. (1988) Proc. Natl. Acad. Sci. USA
Chalfie, M., Tu, Y., Euskirchen, G., Ward, W. W. & Prasher,
D. C. (1994) Science 263, 802–805.
Heim, R., Cubitt, A. B. & Tsien, R. Y. (1995) Nature (London)
Cubit, A. B., Heim, R., Adams, S. R., Boyd, A. E., Gross, L. A.
& Tsien, R. Y. (1995) Trends Biochem. Sci. 20, 448–455.
Leibold, E. A., Laudano, A. & Yu, Y. (1990) Nucleic Acids Res.
Bettany, A. J. E., Eisenstein, R. S. & Munro, H. N. (1992) J. Biol.
Chem. 267, 16531–16537.
Henderson, B. R., Menotti, E., Bonnard, C. & Ku ¨hn, L. C. (1994)
J. Biol. Chem. 269, 17481–17489.
Lutz-Freyermuth, C., Query, C. C. & Keene, J. D. (1990) Proc.
Natl. Acad. Sci. USA 87, 6393–6397.
K. (1991) EMBO J. 10, 3447–3456.
Stump, W. T. & Hall, B. K. (1995) RNA 1, 55–63.
Stade, K., Ford, C. S., Guthrie, C. & Weis, K. (1997) Cell 90,
Goossen, B., Caughman, S. W., Harford, J. B., Klausner, R. D.
& Hentze, M. W. (1990) EMBO J. 9, 4127–4133.
Goossen, B. & Hentze, M. W. (1992) Mol. Cell. Biol. 12,
Koloteva, N., Mueller, P. P. & McCarthy, J. E. G. (1997) J. Biol.
Chem. 272, 16531–16539.
956 Biochemistry: Paraskeva et al. Proc. Natl. Acad. Sci. USA 95 (1998)