TRAP display: A high-speed selection method for the generation of functional polypeptides

Article abstract of DOI:10.1021/ja312579u

Here, we describe a novel method that enables high-speed in vitro selection of functional peptides, peptidomimetics, and proteins via a simple procedure. We first developed a new cell-free translation system, the TRAP system (transcription-translation cou

Full text of DOI:10.1021/ja312579u

Article
pubs.acs.org/JACS
TRAP Display: A High-Speed Selection Method for the Generation of
Functional Polypeptides
Takahiro Ishizawa,† Takashi Kawakami,† Patrick C. Reid,‡ and Hiroshi Murakami*,†
†Department of Life Sciences, Graduate School of Arts and Sciences, The University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo
153-8902, Japan
‡PeptiDream Inc., 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan
S
* Supporting Information
ABSTRACT: Here, we describe a novel method that enables high-
speed in vitro selection of functional peptides, peptidomimetics, and
proteins via a simple procedure. We first developed a new cell-free
translation system, the TRAP system (transcription−translation coupled
with association of puromycin linker), which automatically produces a
polypeptide library through a series of sequential reactions: transcription,
association of puromycin-DNA linker, translation, and conjugation
between the nascent polypeptide and puromycin-DNA linker. We then
applied the TRAP system for the selection of macrocyclic peptides
against human serum albumin. Six rounds of selection using TRAP
display were performed in approximately 14 h, yielding macrocyclic peptides with nanomolar affinity to their target protein.
Because TRAP display enables high-speed selection of functional polypeptides, it will facilitate the generation of various
polypeptides that are useful for biological and therapeutic applications.
template DNAs in a coupled transcription−translation
system.6,7 Although these protein-based linkages are stable
under various conditions, this system cannot be easily applied
to the selection of peptidomimetic composed of non-
proteinogenic amino acids via genetic code reprogramming8−11
because these anchor proteins cannot be synthesized in a
translation system lacking proteinogenic amino acids.
INTRODUCTION
■
In vitro display methods have been used to generate various
functional peptides, peptidomimetics, and proteins. Success in
in vitro selection of polypeptides with desired properties require
one-to-one correspondence between the genotype (DNA or
mRNA) and phenotype (encoded polypeptide), maintained by
a physical linkage. Formation of such a physical linkage
determines the performance of the selection method, such as
(1) time and effort to complete the selection, (2) availability of
selective conditions that need to be precisely designed to select
polypeptides with desired properties, and (3) the variety of
amino acids that can be used as building blocks of the
polypeptide libraries.
For example, ribosome display can rapidly prepare the
peptidyl-tRNA/mRNA/ribosome complexes from the corre-
sponding DNAs because the linkages mediated by the ribosome
and peptidyl-tRNA can be automatically formed in a coupled
transcription−translation system.1,2 However, ribosome display
selection may suffer from unpredictable interactions between
the displayed polypeptide and the very large ribosome.3
Moreover, selection steps must be performed before reverse
transcription (RT) and under a high magnesium concentration
to avoid disassembly of the peptidyl-tRNA/mRNA/ribosome
complexes.1,2,4 Such conditions can potentially produce not
only desired polypeptide binders but also undesired RNA
aptamers5 because RNA folding is generally stabilized by
magnesium ions. Similarly, display technologies using template
DNA-binding anchor proteins (streptavidin or DNA replication
initiator protein, RepA) as the linkage also allow automatic
formation of the polypeptide/DNA complexes from the
The mRNA display (or in vitro virus) systems circumvent
these difficulties by using non-protein-based linkage mediated
by a small molecule, puromycin (Pu).12,13 Puromycin is an
analogue of the 3′-end of aminoacyl-tRNA and attacks a
nascent polypeptide in the ribosome. Therefore, only the
attachment of the puromycin linker to mRNA is required for
producing a robust polypeptide/mRNA complex. To date,
several attachment methods, such as ligation,12,13 photo-cross-
linking,14 and hybridization using 2′-O-methyl-RNA,15 have
been developed. However, these require multiple stepwise
manipulations, including transcription, attachment of the
puromycin linker, and translation, to be performed separately
because an independent step is required to modify the mRNA
with the puromycin linker. These multistep manipulations
make mRNA display complicated and time-consuming,
typically requiring 2−3 days per round.16,17 Therefore, there
is a need for more simple and convenient display methods that
can rapidly prepare polypeptide/mRNA complexes with small
linkages from the template DNAs.
Received: December 25, 2012
Published: March 18, 2013
© 2013 American Chemical Society
5433
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
Figure 1. Design of TRAP display. (a) Schematic view of TRAP display selection, which consists of four steps: the polypeptide/mRNA complex
formation in the TRAP system, reverse transcription (RT), selection against target-immobilized beads, and PCR. In the TRAP system, transcription,
association of a Pu-linker, translation, and conjugation between polypeptide and puromycin are coupled; therefore, the polypeptide/mRNA
complexes are automatically formed by simply adding template DNAs to the TRAP system. (b) A random pool of mRNA and the Pu-linker complex
used in TRAP display (An21). The Pu-linker is annealed to the 3′-UTR of the mRNA to form a 21-base-pair DNA/RNA duplex. The mRNA
sequence for a cysteine residue and the following spacer amino acid residues (Gly-Gly-Gly-Gly-Gly-Ser) and a blank UAG codon (colored in red)
are placed after the (NNK)12 random sequence. (c) Structure of mRNA/Pu-linker 13-base-pair duplex with ligation (Li13) or without ligation
(An13). Abbreviations: Pu-linker, puromycin-DNA linker; UTR, untranslated region.
Here, we describe TRAP (transcription−translation coupled
with association of puromycin linker) display, which facilitates
high-speed in vitro selection of polypeptides and peptidomi-
metics using a simple procedure (Figure 1a). In the TRAP
system, sequential reactions, including transcription of DNA
into mRNA, association of a puromycin-DNA linker (Pu-
linker), translation of mRNA into polypeptides, and con-
jugation between the polypeptide and Pu-linker, proceed
continuously. A stable polypeptide−Pu-linker/mRNA complex
is automatically formed from its template DNA so that the time
for preparation of polypeptide libraries displayed on the
corresponding mRNA can be markedly reduced. Thus, TRAP
display can greatly accelerate the speed of polypeptide
selections to typically less than 3 h per round.
In this study, we first compared the stability of peptide−Pu-
linker/mRNA complexes and the efficiency of formation of
random peptide−Pu-linker/mRNA complexes in a conven-
tional cell-free translation system and in the TRAP system. We
then demonstrated enrichment of a model T7-tag peptide (T7-
peptide) from a random peptide-encoding pool of DNAs
(random-DNAs) spiked with the T7-peptide-encoding DNA
(T7-DNA). Using TRAP display, we selected thioether-
macrocyclized peptides against human serum albumin (HSA)
from a peptide library containing as many as 1012−1013 unique
sequences. More importantly, six rounds of selection were
performed in only about 14 h, demonstrating the high-speed
advantage of TRAP display.
EXPERIMENTAL SECTION
■
Preparation of DNA, mRNA, and Aminoacyl-tRNAs. The
preparation of T7-DNA, random-DNA, T7-mRNA, and random-
mRNA is described in the Supporting Information. Phe-tRNAAsn‑E2
CUA
and biotin-Phe-tRNAfMet
were prepared according to the previous
CAU
reports10 (see also Supporting Information). Synthesis of N-[3-(2-
chloroacetamide)benzoyl]-L-Phe-CME (ClAB-L-Phe-CME, where
CME is cyanomethyl ester) and preparation of ClAB-L-Phe-
tRNAfMet
are also described in the Supporting Information.
PrepaCraAUtion of a Conventional Cell-Free Translation System
and the TRAP System. Creatine kinase, creatine phosphate, and
Escherichia coli tRNAs were purchased from Roche Diagnostics
(Tokyo, Japan). Myokinase was purchased from Sigma-Aldrich Japan
(Tokyo, Japan). Ribosome was purified from E. coli A19 strain by
using a similar procedure reported previously.18 Chemicals and
proteins for translation were prepared by a similar procedure of
previously published papers.9,19−21 The protein concentration in the
stock solution B (solB) and the final concentration in the reaction
mixture are listed in Table S2. The concentration of tRNAs and other
small-molecule factors in the stock solution A (solA) and in the
reaction mixture are listed in Table S3.
The reaction mixture of a conventional cell-free translation system
was prepared by mixing the solA (11%, v/v) and solB (10%, v/v) and
other solutions. The reaction mixture of the TRAP system was
prepared by mixing solA (11%, v/v), solC (11%, v/v), Pu-linker (1.1
μM in final concentration), and other solutions (Table S2).
Preparation of the Li13-, An13-, and An21-Type Pu-Linker/
mRNA Complex. Preparation of the Li13-type Pu-linker/mRNA
complex is described in the Supporting Information. The An13- and
An21-type Pu-linker/mRNA complexes were prepared by mixing
5434
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
mRNA and 1.1 equiv of Pu-linker. The Pu-linker/mRNA complex
concentrations described below were the concentrations of the mRNA.
Analysis of the T7-Peptide−Pu-Linker/mRNA/cDNA Com-
plex Stability. The translation solution was prepared as described
above. The translation mixture containing each 0.25 mM 20
proteinogenic amino acid, 0.1 μM Li13-, An13-, or An21-type Pu-
linker/T7-mRNA complex, and the corresponding 1 μM Pu-linker/
random-mRNA complex was used in this experiment. Alternatively, 25
of HSA) at room temperature for 20 min. The beads were washed with
60 μL of HBST three times and were suspended in 5 μL of 1× RT mix
[1 μM G5S-4.R20, 50 mM Tris-HCl pH 8.3, 75 mM KCl, 3 mM
MgCl2, 10 mM DTT, 0.5 mM each dNTP, 5 U M-MLV
(+)(Promega)]. After incubation at 42 °C for 30 min, the mixture
was added to 100 μL of 1× PCR mix (1× PCR buffer, 2 mM MgCl2,
0.25 mM each dNTP, 0.25 μM T7SDM2.F44, 0.25 μM G5S-
4an21.R41) and was heated at 95 °C for 5 min. A 1 μL aliquot of the
mixture was used for quantification of the recovered cDNA, and the
rest of the mixture was used for PCR (94 °C for 20 s, 55 °C for 20 s,
72 °C for 30 s) after addition of Taq DNA polymerase.
The reaction mixture (20 μL) of the TRAP system containing 0.25
mM each 19 proteinogenic amino acid (−Met), 10 μM ClAB-L-Phe-
tRNAfMetCAU, and 5% (v/v) PCR crude solution was incubated at 37
°C for 15 min. After addition of 5 μL of 100 mM EDTA (pH 7.5), the
mixture was added to 8.53 μL of 4× RT mix [200 mM Tris-HCl (pH
8.3), 300 mM KCl, 75 mM MgCl2, 4 mM DTT, 2 mM each dNTP, 10
μM G5S-4.R20, 6 U ReverTraAce], and incubated at 42 °C for 30 min.
After the reaction was quenched by adding 6 μL of 100 mM EDTA
(pH 7.5) and 4 μL of 500 mM Hepes, the solution was mixed with
HSA-immobilized SA beads (1.1 pmol of HSA) and incubated at 25
°C for 10 min. After the beads were washed with 60 μL of HBST three
times, the recovered beads were mixed with 40 μL of 1× PCR mix, and
the solution was heated at 95 °C for 5 min. Quantification of
recovered cDNA by real-time PCR and amplification of the cDNA by
PCR followed the same procedure as in the first round.
From the third round, the scale of the reaction mixture was reduced
four times. Additionally, negative selection using SA beads was
performed three times before the positive selection. In the fifth and
sixth rounds, stringent wash with 200 μL of HBST was performed at
37 °C for 30 min at the second wash. After the sixth round, the
amplified cDNA was cloned and sequenced.
Binding of Clonal Peptides and Their Derivatives to HSA in
Display Format. The mRNA of p1 and p2 was synthesized from
colony PCR product. Each mRNA of peptides with reverse or shuffled
sequence of p1 and p2 was prepared as follows. DNA template was
prepared using T7SD8M2.F44 as a forward primer and SD8No38r-
evG5S4.R69, SD8No38ranG5S4.R69, SD8No41revG5S4.R75, or
SD8No41ranG5S4.R75 as a reverse primer. Extension-PCR and
transcription were performed using a procedure similar to that used
for T7-mRNA preparation. The mRNA was purified by phenol/
chloroform extraction and 2-propanol precipitation. The pellet was
dissolved in ultrapure water.
μM Phe-tRNAAsn‑E2
was added to the mixture in order to reassign
CUA
the UAG codon from blank to Phe.
The reaction was performed as follows. The translation mixture (2
μL) was incubated at 37 °C for 15 min. After addition of 1 μL of 50
mM ethylenediaminetetraacetic acid (EDTA, pH 7.5), the mixture (3
μL) was added to 1 μL of 4× RT mix: 200 mM tris(hydroxymethyl)-
aminomethane (Tris)-HCl (pH 8.3), 300 mM KCl, 75 mM MgCl2, 4
mM dithiothreitol (DTT), 2 mM each dNTP (dATP, dTTP, dGTP,
dCTP), 10 μM G5S-4.R20, and 6 U ReverTraAce (TOYOBO). The
final concentrations of mRNA templates and primer were 0.05 μM T7-
mRNA, 0.5 μM random-mRNA, and 2.5 μM G5S-4.R20. The mixture
was incubated at 42 °C for 30 min.
T7-peptide pull-down was performed as follows. After 10-fold
dilution of the RT mixture with HBST [50 mM 2-ethanesufonic acid
(Hepes)-KOH pH 7.5, 300 mM NaCl, 0.05% Tween20], the diluted
mixture (10 μL) was mixed with anti-T7-peptide antibody (MBL)
immobilized on Dynabeads Protein G (VERITAS) and incubated for
15 min at 4 °C. The beads were washed with 10 μL of HBST three
times and were suspended in 25 μL of 0.5× PCR buffer [5 mM Tris-
HCl pH 8.4, 25 mM KCl, 0.05%(v/v) Triton X-100]. The solution
was heated at 95 °C for 5 min, and 1 μL of the supernatant was added
to 19 μL of 1× PCR mix (1× PCR buffer, 2 mM MgCl2, 0.25 mM
each dNTP, 0.25 μM T7SD8M2.F44, 0.25 μM G5S-4.R20, and Taq
DNA polymerase). After PCR was performed (94 °C for 20 s, 60 °C
for 20 s, 72 °C for 30 s), the amplified DNA was analyzed by native
PAGE.
Quantitation of the Random-Peptide−Pu-Linker/mRNA
Complex Formation in the TRAP System. The translation mixture
containing each 0.25 mM 19 proteinogenic amino acid (−Met), 25
μM biotin-Phe-tRNAfMet
and Li13- or An21-type 1 μM Pu-linker/
CAU
random-mRNA complex was used in this experiment. Alternatively, 25
μM Phe-tRNAAsn‑E2
was added to the mixture in order to reassign
CUA
the UAG codon from blank to Phe. The reaction mixture of the TRAP
system containing each 0.25 mM 19 proteinogenic amino acid
(−Met), 25 μM biotin-Phe-tRNAfMet
and 5% (v/v) random-DNA
RT-PCR solution was also used in this experiment.
CAU
Translation (2 μL), RT (4 μL), and the quenching reaction (5.25
μL) were performed as described above. A 2 μL aliquot of the resulting
solution was mixed with HSA-immobilized SA beads (0.3 pmol HSA)
or ligand-free SA beads. After the incubation at room temperature for
10 min, the beads were washed with 10 μL of HBST three times. At
the second wash, stringent wash with 100 μL of HBST was performed
at 37 °C for 30 min. Recovered cDNA was quantified by real-time
PCR.
Translation (2 μL) and RT (4 μL) were performed as described
above. The biotin-random-peptide−Pu-linker/mRNA/cDNA com-
plexes were recovered by using streptavidin (SA) magnetic beads,
and the recovered cDNA was quantified by real-time PCR. The
reaction mixture containing reverse transcribed random cDNA was
serial-diluted and used as standards.
Selective Enrichment of T7-Peptide Using a Conventional
mRNA Display and TRAP Display Format. The translation mixture
containing each 0.25 mM 20 proteinogenic amino acid, 0.3 nM Li13-
or An21-type Pu-linker/T7-mRNA complex, and the corresponding 1
μM Pu-linker/random-mRNA complex was used in this experiment.
For the conventional mRNA display format (Li13-type), 25 μM Phe-
tRNAAsn‑E2CUA was added to the mixture in order to reassign the UAG
codon from blank to Phe. The reaction mixture of the TRAP system
containing each 0.25 mM 20 proteinogenic amino acid and 5% (v/v)
RT-PCR solution of T7-DNA and random-DNA mixed in a 1:3,000
ratio was also used in this experiment.
RESULTS
■
Design of the TRAP System. mRNA display requires
multiple manipulation steps for preparation of the polypeptide/
mRNA complex, which makes this display selection compli-
cated and time-consuming. To prepare the polypeptide/mRNA
complexes more rapidly, we designed the novel TRAP coupling
system, which is a reconstituted cell-free transcription−
translation coupling system containing a Pu-linker and lacking
release factor 1 (RF1). In the TRAP system, multiple reactions
proceed sequentially in a single step, as shown in Figure 1a.
First, template DNA is transcribed to mRNA by T7 RNA
polymerase. Then, the transcribed mRNA is “trapped” by the
Pu-linker DNA through formation of a stable duplex between
the Pu-linker DNA, and the mRNA as a GC-rich sequence
complementary to the Pu-linker DNA is placed at the 3′-
untranslated region (3′-UTR) of the mRNA. The ribosome
Translation (2 μL), RT (4 μL), T7-peptide pull-down assay, and
PAGE analysis were similarly performed as described above.
In Vitro Selection of Macrocyclic Peptides by Using TRAP
Display. The translation solution was prepared as described above.
Translation mixture (50 μL) containing 0.25 mM each 19
proteinogenic amino acid (−Met), 10 μM ClAB-L-Phe-tRNAfMet
,
CAU
1 μM mRNA library and 1.1 μM Pu-linker.an21 was incubated at 37
°C for 15 min. After addition of 12.5 μL of 100 mM EDTA (pH 7.5),
the mixture was incubated with HSA-immobilized SA beads (1.6 pmol
5435
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
Figure 2. Examination of the peptide−Pu-linker/mRNA complex stability using the T7-peptide pull-down assay. (a) Scheme of the T7-peptide pull-
down assay. Mixtures of the T7-peptide and random pool (NNK12) templates were translated. After reverse transcription, the complexes were
recovered by anti-T7 antibody-immobilized beads. cDNAs of selected complexes were amplified and analyzed using PAGE. (b) Evaluation of the
peptide−Pu-linker/mRNA complex stability through the selection processes. Lanes 1−4 contain DNA markers synthesized from the T7-mRNA,
random-mRNA, and a 1:1 or 1:10 mixture of T7- and random-mRNAs. In lanes 5−10, T7- and random-mRNAs were mixed in a 1:10 ratio in the
translation system where a UAG codon is reassigned to Phe (lanes 5−7) or a blank (lanes 8−10). The following mRNA templates were used in
translation: lanes 5 and 8, Li13-type mRNAs; lanes 6 and 9, An13-type mRNAs; lanes 7 and 10, An21-type mRNAs. Fractions of the T7-DNA were
calculated based on the band intensity of T7- and random-DNA observed on the gel.
travels down the mRNA strand from the start codon to the end
of the open reading frame (ORF) and pauses immediately
before the Pu-linker/mRNA duplex region. Finally, the attached
Pu attacks the nascent polypeptide to form a peptide−Pu-
linker/mRNA complex. It should be noted that the UAG
codon, which is assigned as a blank due to the lack of RF1, is
placed to enhance the stalling of the ribosome immediately
before the Pu-linker/mRNA duplex region (vide infra).
Stability of the Peptide−Pu-Linker/mRNA Complex.
One of the initial concerns around TRAP display was the
stability of the peptide−Pu-linker/mRNA complex because the
Pu-linker is noncovalently attached to the mRNA. This
noncovalent linkage may cause an exchange of polypeptide−
Pu-linkers between different mRNAs, resulting in disarrange-
ment of the one-to-one correspondence between the mRNA
and encoded polypeptide. To evaluate the stability of the
peptide−Pu-linker/mRNA complex throughout the TRAP
display selection processes, we performed a T7-peptide pull-
down assay, as shown in Figure 2a. Pu-linker/mRNA
complexes of the T7-peptide-encoding mRNA (T7-mRNA)
and random peptide-encoding mRNAs (random-mRNAs),
both of which have the same Pu-linker annealing sequence at
their 3′-UTR, were mixed in a 1:10 ratio, and after translation
and RT using this template mRNA mixture, a T7-peptide−Pu-
linker/mRNA/cDNA complex was selectively recovered using
anti-T7 antibody-immobilized beads. Ideally, the T7-peptide−
Pu-linker is attached only to the T7-mRNA and only the
desired T7-mRNA/cDNA is recovered; but if the formed T7-
peptide−Pu-linker is dissociated from the T7-mRNA and is
reattached to random-mRNAs, a significant amount of the
undesired random-mRNA/cDNAs would also be recovered
(Figure S1).
We then examined the stability of the peptide−Pu-linker/
mRNA complexes of the conventional ligation-based mRNA
display (Figure 1c, Li13) and those of ligation-free 13-mer
annealing-based (Figure 1c, An13). In this experiment, we first
used an orthogonal Phe-tRNACUA prepared by a flexible tRNA
acylation ribozyme (flexizyme)10,22,23 to prepare a UAG codon
code for Phe in the translation reaction. As expected, when the
Li13-type Pu-linker/mRNA complexes were used, the T7-
mRNA/cDNA was recovered in a highly selective manner using
the anti-T7 antibody-immobilized beads (Figure 2b, lanes 4 vs
5). However, when the An13-type Pu-linker/mRNA complexes
were used, a significant amount of random-mRNA/cDNAs was
also recovered together with the T7-mRNA/cDNA (Figure 2b,
lane 6). We considered that this result was presumably because
of insufficient stability of the duplex formed between the Pu-
linker DNA and the An13-type mRNA. To further stabilize the
duplex, we increased the number of base pairs in the Pu-linker
DNA/mRNA duplex from 13 mers to 21 mers and prepared
the An21-type Pu-linker/mRNA, as shown in Figure 1b.
However, this improvement only increased the fraction of the
recovered T7-mRNA/cDNA from 34% to 40% (Figure 2b,
lanes 6 vs 7). This indicated that more than half of the T7-
peptide−Pu-linker dissociated from the T7-mRNA either
during translation, RT, or the T7-peptide pull-down process.
Given that the Pu-linker DNA/mRNA duplex was stable during
RT and the pull-down process, and under even more stringent
conditions (Figures S2 and S3), the dissociation must have
occurred during translation. We hypothesized that the helicase
activity of the translating ribosome caused the dissociation. To
reduce the effect of the ribosome helicase activity by stalling the
ribosome before the Pu-linker DNA/mRNA duplex region, we
placed a UAG blank codon immediately before the duplex
Before the pull-down assay, we first tested whether the T7-
DNA (Figure 2b, lane 1, 94 mers) and the random-DNA
(Figure 2b, lane 2, 103 mers) are synthesized from the
corresponding mRNAs by reverse transcription−polymerase
chain reaction (RT-PCR) with similar efficiency. Indeed, the
ratio of band intensity between the amplified T7-DNA and the
amplified random-DNAs was proportional to that of the
amount of their respective input DNA templates (Figure 2b,
lanes 3 and 4, lower bands vs upper bands).
region in the mRNA by omitting the Phe-tRNAAsn‑E2
from
CUA
the translation mixture (Figure 1b). As a result, the fraction of
the recovered T7-mRNA/cDNA was significantly increased
from 34% to 63% in the An13-type complex and from 40% to
73% in the An21-type complex (Figure 2b, lanes 6 vs 9 and
lanes 7 vs 10, respectively). This indicated that the major cause
of the T7-peptide−Pu-linker dissociation from the T7-mRNA
was indeed ribosome helicase activity, and extending the Pu-
linker DNA/mRNA duplex length and placing a blank UAG
5436
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
codon immediately before the duplex sufficiently stabilized the
T7-peptide−Pu-linker/mRNA complexes.
experiment. Quantification of recovered cDNA by real-time
PCR revealed that 7% of the Li13-type mRNA was recovered
when a UAG codon was assigned to Phe and when it was
assigned as a blank (Figure 3c, columns 1 and 2). However,
recovery of the An21-type mRNA markedly decreased from
11% to 4% by reassignment of the UAG codon from blank to
Phe (Figure 3c, columns 3 and 4). This suggested that the
ribosome could dissociate the Pu-linker DNA/mRNA duplex in
the An21-type Pu-linker/mRNA complex before Pu attacked
the nascent peptide, but placing the blank codon immediately
before the duplex region promoted the Pu attacking the nascent
peptide by stalling the ribosome at the blank codon. Lower
recovery of the Li13-type mRNA compared with the An21-type
mRNA (Figure 3c, columns 2 and 4) could result from the
moderate amount of appropriately ligated Li13-type Pu-linker/
mRNA (data not shown).
Efficiency of the Random-Peptide−Pu-Linker/mRNA
Complex Formation in the TRAP System. For selection of
peptides from a random library, the efficiency of the random-
peptide−Pu-linker/mRNA complex formation is highly im-
portant because it directly determines the diversity of the
peptide library. To evaluate the efficiency of the complex
formation, we added biotin-Phe (Figure 3a) charged onto
We then evaluated the efficiency of the An21-type peptide−
Pu-linker/mRNA complex formation by using the correspond-
ing DNA instead of mRNA as a template in the TRAP system.
To omit the time required to purify PCR-amplified template
DNAs in every selection round, we first optimized the TRAP
system for coupled transcription−translation. PCR mixture was
directly added to the translation mixture without purification,
and the transcribed mRNA was quantified by RT real-time
PCR. We confirmed that a sufficient amount of mRNA (<1
pmol/μL) was produced after incubation for 5−10 min in the
TRAP system containing 1 μM T7 RNA polymerase (Figure
S4). To this TRAP system, we added the An21-type random-
DNA as a template together with biotin-Phe-tRNAfMet
and
CAU
performed the biotin pull-down experiment. Recovery of the
An21-type peptide/mRNA complex generated from the DNA
template was reduced by half compared with that obtained
from the mRNA template because the ribosomes may be
consumed by the overproduced Pu-linker-free mRNA;
however, this was still comparable with that of the Li13-type
mRNA (Figure 3c, columns 2, 4, and 5). This level of efficiency
would particularly be acceptable for the second and later
rounds of selection, at least to ensure diversity in a polypeptide
library, because active sequences are enriched and amplified in
the previous rounds of selection. Therefore, we decided to use
the An21-type random-mRNA as a template for the first round
of selection to ensure library diversity and the amplified DNA
as a template for the later rounds to accelerate the speed of
selection.
Selective Enrichment for the T7-Peptide−Pu-Linker/
mRNA Complex. To evaluate the ability of TRAP display to
selectively enrich desired peptides, we compared the enrich-
ment in TRAP display (An21-type) and conventional mRNA
display (Li13-type) using a T7-peptide pull-down assay as
described above. The Pu-linker/mRNA complexes of the T7-
mRNA and random-mRNAs were mixed in a 1:3000 ratio and
were added to the TRAP system. After the pull-down assay,
RT-PCR products were analyzed using native polyacrylamide
gel electrophoresis (PAGE).
Figure 3. Analysis of the efficiency of the random-peptide−Pu-linker/
mRNA complex formation in the TRAP system. (a) Structure of the
biotin-L-Phe (BioF) used to synthesize biotin-label peptides. (b)
Scheme of the biotin pull-down assay. The N terminus of synthesized
peptides were labeled with biotin using BioF-tRNAfMet(CAU). After
translation and reverse transcription, the biotin-peptide−Pu-linker/
mRNA/cDNA complexes were captured by streptavidin magnetic
beads. The recovered complexes were quantified by real-time PCR. (c)
Efficiency of the random-peptide/mRNA complex formation. The
reaction was performed in the translation mixture where a UAG codon
is reassigned to Phe (columns 1 and 3) or a blank (columns 2, 4−6).
The following templates were used: columns 1 and 2, Li13-type
mRNAs; columns 3, 4, and 6, An21-type mRNAs; column 5, An21-
type DNA templates. Dagger (†) indicates no incubation. Recovery of
cDNA complexes was calculated by dividing the amount of recovered
cDNA by the theoretical amount of mRNA/Pu-linker (1 μM) in the
reaction mixture. Error bars represent standard deviation calculated
from experiments performed in triplicate.
tRNAfMet
by flexizyme to a translation mixture from which
CAU
Met had been removed to label synthesized peptides with
biotin at the N terminus (Figure 3b). Consequently, only the
mRNAs that display biotin-peptides can be selectively
recovered by pull-down using SA beads.
To compare the efficiency of random-peptide−Pu-linker/
mRNA complex formation between TRAP display and
conventional mRNA display, we first added the Li13- or
An21-type Pu-linker/mRNA complex as a template to the
translation mixture and performed the biotin pull-down
Quantitation of the ratio of T7-DNA and random-DNA in
the gel showed that the T7-mRNA was enriched by 3,700-fold
in the conventional mRNA display selection (Figure 4, lanes 4
and 5) and by 4,900-fold in the TRAP display selection (Figure
4, lanes 6 and 7). This result revealed that TRAP display using
the mRNA templates could selectively enrich the T7-DNA with
similar or even improved efficiency over that of conventional
mRNA display.
5437
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
ratio. The result of the T7-peptide pull-down assay revealed
that the T7-DNA was enriched by 2,000-fold (Figure 4, lanes 8
and 9), which indicated that TRAP display could generate the
polypeptide-Pu-linker/mRNA from not only mRNA templates
but also DNA templates and could enrich a target binding
polypeptide by pull-down experiments using target-immobi-
lized beads.
In Vitro Selection of Macrocyclic Peptides. To
demonstrate the ability of TRAP display to select functional
peptides, we selected macrocyclic peptides against HSA. We
used a macrocyclic peptide library24 because the constrained
macrocyclic structure often afforded high-affinity to tar-
gets.25−30 For the preparation of a macrocyclic peptide library,
we synthesized a novel amino acid for macrocyclization, N-[3-
(2-chloroacetamido)benzoyl]-L-phenylalanine (ClAB-L-Phe)
(Figure 5a), and we reassigned the AUG start codon to
Figure 4. Comparison of enrichment of the T7-peptide template
between conventional mRNA display (Li13) and TRAP display
(An21). Lanes 1−3 contain DNA markers synthesized from T7-
mRNA, random-mRNA, and a 1:3000 mixture of T7- and random-
mRNAs. The T7- and random-mRNAs (or DNA) were mixed in a
1:3000 ratio in the translation system, where a UAG codon is
reassigned to Phe (lanes 4, 5) or blank (lanes 6−9). The following
templates were used in translation: lanes 4 and 5, Li13-type mRNAs;
lanes 6 and 7, An21-type mRNAs; lanes 8 and 9, An21-type DNAs.
Enrichment of the T7-peptide template was calculated from the band
intensity observed on the gel before and after pull-down.
ClAB-L-Phe by adding ClAB-L-Phe-tRNAfMet
to a Met-
CAU
depleted TRAP system. We also prepared an mRNA library for
peptides consisting of 8−12 random amino acids flanked by
ClAB-L-Phe and Cys. The encoded peptides are spontaneously
macrocyclized via a thioether linkage formed between the
chloroacetyl and thiol groups31 (Figure 5b).
We also tested DNA as a template for selective enrichment of
the T7-peptide−Pu-linker/mRNA complex. The T7-DNA and
random-DNAs were added to the TRAP system in a 1:3,000
Figure 5. In vitro selection of macrocyclic peptides using TRAP display and mRNA display. (a) Structure of N-[3-(2-chloroacetamido)benzoyl]-L-
Phe (ClAB-L-Phe) used for peptide macrocyclization. (b) Macrocyclic peptide library used in this study. Random mRNA/Pu-linker complexes
displaying the encoded macrocyclic peptides consisting of 8−12 random amino acids flanked by ClAB-L-Phe and Cys. The chloroacetyl group of
ClAB-L-Phe and the thiol group of Cys spontaneously form an intramolecular thioether bond. (c) Progress of TRAP display selection and mRNA
display selection using HSA-immobilized SA beads. Recovery of the cDNA complexes was calculated in the same way as in Figure 3. (d) Sequences
of selected peptides. The X and C highlighted in blue indicate ClAB-L-Phe and Cys, respectively. (e) Evaluation of selected peptides. Selected
peptides (p1 and p2) and the derivative peptides with reversed and shuffled sequences of the original peptides were prepared as the An21-type
complex. The resulting complexes were examined by a pull-down assay using HSA-immobilized or HSA-free SA beads.
5438
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
The library of macrocyclic peptides displayed on the An21-
type Pu-linker/mRNA complexes was generated in the
translation mixture and used in the pull-down against HSA
immobilized on SA beads. The recovered DNA products,
amplified by RT-PCR, were directly added to the TRAP system
without purification, and the resulting macrocyclic peptide
library was used for the next round of selection. Since cDNA
recovery was significantly increased at the fourth round of
selection (Figure 5c), we performed a stringent wash in the
following two rounds to enrich peptide binders with higher
affinity. Notably, the six rounds of selection were performed in
approximately 14 h, which proved that TRAP display greatly
increased the speed of peptide selection.
To establish whether the conventional mRNA display
selection produces the same macrocyclic peptides obtained by
the TRAP display selection, we also performed mRNA display
selection of macrocyclic peptides (Figure 5c). The sequence
alignment of selected peptides obtained by mRNA display and
TRAP display showed that the two most abundant sequences in
each display method were the same (Figure 5d, p1
XTYNERLFWC and p2 XSQWDPWAIFWC, where X
indicates ClAB-L-Phe), which confirms the reliability of
TRAP display. The results of pull-down using HSA-
immobilized or HSA-free SA beads followed by quantification
of the recovered cDNA by real-time PCR showed that the p1
and p2 macrocyclic peptides indeed bound to HSA-
immobilized SA beads but not to the free SA beads (Figure
5e). Moreover, macrocyclic peptides consisting of reversed or
shuffled sequences of p1 or p2 did not bind to the HSA-
immobilized SA beads (Figure 5e). These results indicated that
not only the composition but also the sequence of the amino
acids in p1 and p2 were important for binding to HSA-
immobilized SA beads. We also chemically synthesized the
fluorescent-labeled p1 and p2 by Fmoc solid-phase synthesis
and determined their affinity to the target by using fluorescence
polarization assay. Both peptides bound to HSA−SA complex
with nanomolar affinity [Kd (App) = 41 and 43 nM for p1 and
p2, respectively] (Figure S5, see Supplementary Note). Most
importantly, these results show that TRAP display can produce
macrocyclic peptide leads with nanomolar affinity to their target
in as little as 14 h.
4), whereas no efficiency change was observed for the Li13-type
complex (Figure 3c, columns 1 and 2). The blank codon, which
stalls the ribosome, therefore, plays two key roles in TRAP
display selection: (1) stabilizing the Pu-linker/mRNA complex
to prevent dissociation of the peptide-Pu-linker from a
particular mRNA and subsequent reassociation with a different
mRNA, and (2) increasing the efficiency of the peptide/Pu-
linker/mRNA complex formation by providing a delay for Pu
attacking the nascent peptide (Figure S6).
The biotin pull-down experiment also provided insight into
the diversity of the polypeptide library. To our knowledge, no
investigations on in vitro display technologies have evaluated
the display efficiency of random peptides. Our attempt to
evaluate the inherent display efficiency of random peptides
revealed that 11% of the random pool of mRNA was converted
into a peptide−Pu-linker/mRNA complex (Figure 3c, column
4). This discovery suggests that highly diverse polypeptide
libraries (6 × 1013 molecules/mL) can be produced using
TRAP display selection.
To reveal the reliability of TRAP display, we compared the
results of selective enrichment and in vitro selection of TRAP
display with those of conventional mRNA display. We observed
better enrichment in mRNA-based TRAP display than in
conventional mRNA display (Figure 4, lanes 5 and 7), and the
enrichment reduced only 2-fold when a DNA template was
used (Figure 4, lanes 7 and 9). These results are in agreement
with the efficiency of the peptide−Pu-linker/mRNA complex
formation where 70% and 50% of relative efficiency were
observed in the Li13-type mRNA and the An21-type DNA,
respectively, as compared with the An21-type mRNA (Figure
3c, columns 1 and 5 vs 4). The relatively lower enrichment of
TRAP display using DNA templates was also observed in
macrocyclic peptide selection (Figure 5c). However, this
difference in enrichment between mRNA display and TRAP
display did not affect the results of the selection (Figure 5d).
Fluorescence polarization assay using chemically synthesized
peptides revealed that the selected peptides bound to the
HSA−SA complex with nanomolar affinity. In future studies,
the use of alternative types of target immobilization methods,
such as amine coupling or His-tag mediated immobilization,
should be investigated. Selection of more potent HSA binders
and the application of such peptides for therapeutic use32−36
would be of significant interest.
DISCUSSION
■
In this study, we developed a novel in vitro selection technology
that greatly accelerated the generation of functional polypep-
tides with desired properties. First, we constructed the TRAP
system in which the steps of transcription, association of a Pu-
linker, translation, and peptidyl-transfer reaction of polypeptide
to the Pu are sequentially performed in a single reaction
mixture (Figure 1a). Because the Pu-linker/mRNA complex is
maintained via a noncovalent linkage, the complex may not be
sufficiently stable to be used for selection, and thus we
evaluated the stability of the polypeptide-Pu-linker/mRNA
complexes. Although the complex was stable under various
conditions, such as RT and pull-down, it became dissociated in
the translation mixture. Based on the hypothesis that the
ribosome helicase activity causes dissociation of the Pu-linker
from the mRNA, we placed a blank UAG codon immediately
before the Pu-linker DNA/mRNA duplex region and
successfully prevented complex dissociation (Figure 2b, lanes
7 and 10). Moreover, the placement of this UAG blank codon
also increased the efficiency of the An21-type peptide/mRNA
complex formation from 4% to 11% (Figure 3c, columns 3 and
CONCLUSIONS
■
We have developed the TRAP system in which stable
polypeptide-mRNA complexes are automatically formed by
addition of the DNA template. In the TRAP system,
transcription−translation is coupled with association of a Pu-
linker and conjugation of the nascent polypeptide and
puromycin. A blank UAG codon placed immediately before
the Pu-linker DNA/mRNA duplex region to stall the ribosome
prevents dissociation of the Pu-linker DNA/mRNA complex
and increases the efficiency of the peptide/mRNA complex
formation. Using TRAP display, we completed six rounds of
selection in approximately 14 h and successfully obtained high
affinity macrocyclic peptides that bind to HSA−SA complex.
The ease of TRAP display should make it amenable to
automation, in which 10s to 100s of libraries could be run in
parallel, something that is not possible to do with conventional
mRNA display.
As TRAP display greatly accelerates the speed of functional
polypeptide selection, it will facilitate the discovery of the next
5439
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440

Journal of the American Chemical Society
Article
(20) Shimizu, Y.; Inoue, A.; Tomari, Y.; Suzuki, T.; Yokogawa, T.;
Nishikawa, K.; Ueda, T. Nat. Biotechnol. 2001, 19, 751.
(21) Kawakami, T.; Murakami, H.; Suga, H. Chem. Biol. 2008, 15, 32.
(22) Murakami, H.; Saito, H.; Suga, H. Chem. Biol. 2003, 10, 655.
(23) Ohuchi, M.; Murakami, H.; Suga, H. Curr. Opin. Chem. Biol.
2007, 11, 537.
(24) Kotz, J. SciBX 2012, DOI: 10.1038/scibx.2012.87.
(25) Heinis, C.; Rutherford, T.; Freund, S.; Winter, G. Nat. Chem.
Biol. 2009, 5, 502.
generation of peptides, peptidomimetics, and proteins that
could be useful for biological and therapeutic applications in the
future.
ASSOCIATED CONTENT
* Supporting Information
■
S
Detailed experimental procedures and additional tables and
figures. This material is available free of charge via the Internet
at http://pubs.acs.org.
(26) Yamagishi, Y.; Shoji, I.; Miyagawa, S.; Kawakami, T.; Katoh, T.;
Goto, Y.; Suga, H. Chem. Biol. 2011, 18, 1562.
(27) Hayashi, Y.; Morimoto, J.; Suga, H. ACS Chem. Biol. 2012, 7,
607.
(28) Morimoto, J.; Hayashi, Y.; Suga, H. Angew. Chem., Int. Ed. 2012,
AUTHOR INFORMATION
Corresponding Author
murah@bio.c.u-tokto.ac.jp
■
51, 3423.
(29) Schlippe, Y. V. G.; Hartman, M. C. T.; Josephson, K.; Szostak, J.
W. J. Am. Chem. Soc. 2012, 134, 10469.
(30) Hofmann, F. T.; Szostak, J. W.; Seebeck, F. P. J. Am. Chem. Soc.
2012, 134, 8038.
(31) Goto, Y.; Ohta, A.; Sako, Y.; Yamagishi, Y.; Murakami, H.; Suga,
H. ACS Chem. Biol. 2008, 3, 120.
(32) Nguyen, A.; Reyes, A. E., 2nd; Zhang, M.; McDonald, P.; Wong,
W. L.; Damico, L. A.; Dennis, M. S. Protein Eng. Des. Sel. 2006, 19, 291.
(33) Holt, L. J.; Basran, A.; Jones, K.; Chorlton, J.; Jespers, L. S.;
Brewis, N. D.; Tomlinson, I. M. Protein Eng. Des. Sel. 2008, 21, 283.
(34) Langenheim, J. F.; Chen, W. Y. J. Endocrinol. 2009, 203, 375.
(35) Neumann, E.; Frei, E.; Funk, D.; Becker, M. D.; Schrenk, H.-H.;
Mueller-Ladner, U.; Fiehn, C. Expert Opin. Drug Deliv. 2010, 7, 915.
(36) Angelini, A.; Morales-Sanfrutos, J.; Diderich, P.; Chen, S.;
Heinis, C. J. Med. Chem. 2012, 55, 10187.
Notes
The authors declare the following competing financial
interest(s): Authors Patrick C. Reid and Hiroshi Murakami
declare competing financial interests, as both are founders and
shareholders in PeptiDream Inc, a peptide drug discovery
company using similar technology.
ACKNOWLEDGMENTS
■
This work was supported by the Funding Program for Next
Generation World-Leading Researchers (LR011 to H.M.) from
Japan Society for the Promotion of Science (JSPS), and by the
Naito Foundation. T.K. was supported by the HFSP long-term
fellowship (LT000939/2010).
REFERENCES
■
(1) Mattheakis, L. C.; Bhatt, R. R.; Dower, W. J. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 9022.
(2) Hanes, J.; Pluckthun, A. Proc. Natl. Acad. Sci. U.S.A. 1997, 94,
̈
4937.
(3) Gold, L. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 4825.
(4) Zahnd, C.; Amstutz, P.; Hanes, J.; Pluckthun, A. Nat. Methods
̈
2007, 4, 269.
(5) Roberts, R. W. Curr. Opin. Chem. Biol. 1999, 3, 268.
(6) Odegrip, R.; Coomber, D.; Eldridge, B.; Hederer, R.; Kuhlman, P.
A.; Ullman, C.; FitzGerald, K.; McGregor, D. Proc. Natl. Acad. Sci.
U.S.A. 2004, 101, 2806.
(7) Sumida, T.; Doi, N.; Yanagawa, H. Nucleic Acids Res. 2009, 37,
e147.
(8) Forster, A. C.; Tan, Z. P.; Nalam, M. N. L.; Lin, H. N.; Qu, H.;
Cornish, V. W.; Blacklow, S. C. Proc. Natl. Acad. Sci. U.S.A. 2003, 100,
6353.
(9) Josephson, K.; Hartman, M. C. T.; Szostak, J. W. J. Am. Chem.
Soc. 2005, 127, 11727.
(10) Murakami, H.; Ohta, A.; Ashigai, H.; Suga, H. Nat. Methods
2006, 3, 357.
(11) Kawakami, T.; Murakami, H. J. Nucleic Acids 2012, 2012,
713510.
(12) Roberts, R. W.; Szostak, J. W. Proc. Natl. Acad. Sci. U.S.A. 1997,
94, 12297.
(13) Nemoto, N.; MiyamotoSato, E.; Husimi, Y.; Yanagawa, H. FEBS
Lett. 1997, 414, 405.
(14) Kurz, M.; Gu, K.; Lohse, P. A. Nucleic Acids Res. 2000, 28.
(15) Tabuchi, I.; Soramoto, S.; Nemoto, N.; Husimi, Y. FEBS Lett.
2001, 508, 309.
(16) Seelig, B. Nat. Protoc. 2011, 6, 540.
(17) Ma, Z.; Hartman, M. C. Methods Mol. Biol. 2012, 805, 367.
(18) Clemons, W. M., Jr.; Brodersen, D. E.; McCutcheon, J. P.; May,
J. L.; Carter, A. P.; Morgan-Warren, R. J.; Wimberly, B. T.;
Ramakrishnan, V. J. Mol. Biol. 2001, 310, 827.
(19) Baggott, J. E.; Johanning, G. L.; Branham, K. E.; Prince, C. W.;
Morgan, S. L.; Eto, I.; Vaughn, W. H. Biochem. J. 1995, 308 (Pt. 3),
1031.
5440
dx.doi.org/10.1021/ja312579u | J. Am. Chem. Soc. 2013, 135, 5433−5440