Efficient one-cycle affinity selection of binding proteins or peptides specific for a small-molecule using a T7 phage display pool

Article abstract of DOI:10.1016/j.bmc.2008.09.061

Here, we report an efficient one-cycle affinity selection using a natural-protein or random-peptide T7 phage pool for identification of binding proteins or peptides specific for small-molecules. The screening procedure involved a cuvette type 27-MHz quartz-crystal microbalance (QCM) apparatus with introduction of self-assembled monolayer (SAM) for a specific small-molecule immobilization on the gold electrode surface of a sensor chip. Using this apparatus, we attempted an affinity selection of proteins or peptides against synthetic ligand for FK506-binding protein (SLF) or irinotecan (Iri, CPT-11). An affinity selection using SLF-SAM and a natural-protein T7 phage pool successfully detected FK506-binding protein 12 (FKBP12)-displaying T7 phage after an interaction time of only 10 min. Extensive exploration of time-consuming wash and/or elution conditions together with several rounds of selection was not required. Furthermore, in the selection using a 15-mer random-peptide T7 phage pool and subsequent analysis utilizing receptor ligand contact (RELIC) software, a subset of SLF-selected peptides clearly pinpointed several amino-acid residues within the binding site of FKBP12. Likewise, a subset of Iri-selected peptides pinpointed part of the positive amino-acid region of residues from the Iri-binding site of the well-known direct targets, acetylcholinesterase (AChE) and carboxylesterase (CE). Our findings demonstrate the effectiveness of this method and general applicability for a wide range of small-molecules.

Full text of DOI:10.1016/j.bmc.2008.09.061

Bioorganic & Medicinal Chemistry 16 (2008) 9837–9846
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Efficient one-cycle affinity selection of binding proteins or peptides specific
for a small-molecule using a T7 phage display pool
Yoichi Takakusagi a, , Kouji Kuramochi a, , Manami Takagi b, Tomoe Kusayanagi a, Daisuke Manita a,
Hiroko Ozawa b, Kanako Iwakiri b, Kaori Takakusagi a, Yuka Miyano a, Atsuo Nakazaki b,
a,
a,
*
Susumu Kobayashi b, , Fumio Sugawara , Kengo Sakaguchi
*
*
a Department of Applied Biological Science, Faculty of Science and Technology, Tokyo University of Science, 2641 Yamazaki, Noda, Chiba 278-8510, Japan
b Department of Medicinal and Life Science, Faculty of Pharmaceutical Sciences, Tokyo University of Science, 2641 Yamazaki, Noda, Chiba 278-8510, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Here, we report an efficient one-cycle affinity selection using a natural-protein or random-peptide T7
phage pool for identification of binding proteins or peptides specific for small-molecules. The screening
procedure involved a cuvette type 27-MHz quartz-crystal microbalance (QCM) apparatus with introduc-
tion of self-assembled monolayer (SAM) for a specific small-molecule immobilization on the gold elec-
trode surface of a sensor chip. Using this apparatus, we attempted an affinity selection of proteins or
peptides against synthetic ligand for FK506-binding protein (SLF) or irinotecan (Iri, CPT-11). An affinity
selection using SLF-SAM and a natural-protein T7 phage pool successfully detected FK506-binding pro-
tein 12 (FKBP12)-displaying T7 phage after an interaction time of only 10 min. Extensive exploration
of time-consuming wash and/or elution conditions together with several rounds of selection was not
required. Furthermore, in the selection using a 15-mer random-peptide T7 phage pool and subsequent
analysis utilizing receptor ligand contact (RELIC) software, a subset of SLF-selected peptides clearly pin-
pointed several amino-acid residues within the binding site of FKBP12. Likewise, a subset of Iri-selected
peptides pinpointed part of the positive amino-acid region of residues from the Iri-binding site of the
well-known direct targets, acetylcholinesterase (AChE) and carboxylesterase (CE). Our findings demon-
strate the effectiveness of this method and general applicability for a wide range of small-molecules.
Ó 2008 Elsevier Ltd. All rights reserved.
Received 20 August 2008
Revised 19 September 2008
Accepted 20 September 2008
Available online 30 September 2008
Keywords:
Small-molecule-oriented T7 phage display
QCM
Self-assembled monolayer
One-cycle screen
Drug-like small-molecule
Synthetic ligand for FK506-binding protein
(SLF)
Irinotecan (Iri, CPT-11)
1. Introduction
Recently, we reported the establishment of an effective T7 phage
display environment for small-molecules using a T7 phage display
In forward chemical genetics, the T7 phage display method is a
powerful tool that can be applied to the screening of small-mole-
cule binding proteins or peptides.1–3 Using this method, in theory,
drug-binding protein(s) or peptide(s) displayed on the T7 phage
capsid can be effectively panned from a diverse range of proteins
or peptides. However, the method relies on the chemically-defined
binding strength in vitro, which does not always result in
identification of the proteins responsible for bioactivity of the
small-molecule in vivo. Thus, a rapid and reliable classification of
small-molecule binding proteins from the ordered genome
database is highly desirable. Using this information it should be
possible to design a series of experiments to investigate the mode
of action of the corresponding small-molecules.
system and a cuvette type 27-MHz quartz-crystal microbalance
(QCM) apparatus (Affinix Q, Initium, Tokyo, Japan).4,5 This novel ap-
proach overcomes the technical difficulties of a conventional screen
using microplate or agarose beads and can be used for the rapid iso-
lation of binding phage specific for small-molecule.4,5 Here, we fur-
ther introduce a self-assembled monolayer (SAM) with multiple
functions as a means of small-molecule attachment for the sensor
chip.6 Two test molecules were used in this study: (i) a synthetic li-
gand for FK506-binding protein (SLF), an analog of the immunosup-
pressiveagent FK506,7 and (ii) irinotecan (Iri, CPT-11), ananti-tumor
camptothecin (CPT) derivative with clinical use.8 We verified selec-
tion from a natural-protein or random-peptide T7 phage pool. The
latter approach also involves the use of receptor ligand contact (RE-
LIC) software, a bioinformatics tool for phage display screening.9
This software system was developed using the concept that drug-
like small-molecules with a rigid structure immobilized on a carrier
appear to show weak interactions with peptide via surface–surface
contacts against a continuous or discontinuous three- to five-ami-
no-acid motif. Using a subset of small-molecule-selected random-
peptides, with the peptide sequences arbitrarily selected from the
* Corresponding authors. Tel.: +81
4 7124 1501x6543 (S.K.), +81 4 7124
1501x3400 (F.S.), +81 4 7124 1501x3409 (K.S.).
E-mail addresses: kobayash@rs.noda.tus.ac.jp (S. Kobayashi), sugawara@rs.noda.
tus.ac.jp (F. Sugawara), kengo@rs.noda.tus.ac.jp (K. Sakaguchi).
Both authors equally contributed to this study.
0968-0896/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2008.09.061

9838
Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
unscreened parent pool, a significant change of amino-acid popula-
tion, or the emergence of small-molecule-recognizing short
stretches discernable by eye, can be statistically detected.2,9,10 By
making use of this information and the subset of peptide sequences,
further exploration of binding proteins from the database, as well as
the validation of the peptide selection itself, is systematically feasi-
ble.9 The results of SLF and Iri in this paper demonstrate the effec-
tiveness and general applicability of our strategy for screening of
small-molecule binding proteins/peptides.
line). From kinetic analysis using AQUA ver1.5 software (Initium
Inc.), the KD value was 15–19 nM, which is consistent with previ-
ously reported data.13,14 The Bmax (100% binding ratio) of FKBP12
to immobilized SLF was 187–231 Hz (5.6–6.9 ng) in the case of
SAM immobilization, whereas 58–77 Hz (1.7–2.2 ng) was calcu-
lated from the data obtained by avidin–biotin immobilization.
These results indicate that the amount of active SLF molecule
immobilized by SAM is at least threefold greater than is immobi-
lized by avidin–biotin (Fig. 2C). Thus, the effectiveness of SAM
immobilization over the avidin–biotin method is clear, at least
using the QCM apparatus.
2. Results and discussion
2.2. One-cycle affinity selection of T7 phages exhibiting
preferential affinity for SLF from a natural-protein T7 phage
pool
2.1. Efficiency of SAM for SLF immobilization on the gold
electrode surface of a sensor chip
To date, SAM has been reused as an immobilization scaffold to
facilitate directed orientation and accumulation of the substrate
on a gold (Au) surface.6,11 Using SLF7,12–14 as a candidate small-
molecule, we initially synthesized the SLF derivative as shown in
Scheme 1 andFigure 2A. The molecule comprises four units: (i) sul-
fide bond for chemisorption on the gold electrode surface of the
sensor chip via thiol–Au interactions, (ii) a C11 alkyl chain15 for
accumulation of the SLF molecule on the gold, (iii) diethylene gly-
col (DEG) for reducing the non-specific binding (mainly via hydro-
phobic interaction) with the gold surface or linkers, and (iv) SLF to
act as bait during screening.
Using this synthetic molecule, we can attain one-step immobi-
lization of SLF on a gold electrode surface of QCM sensor chip. SLF
derivative was dissolved in 75% ethanol, dropped onto the gold
electrode and then incubated for 16 h under a humid and shaded
atmosphere at room temperature, which resulted in the formation
of SLF-SAM on the gold electrode surface. After washing the surface
of the electrode, the sensor chip was placed in the QCM apparatus
using a buffer filled cuvette. The QCM sensor was then fully stabi-
lized. To validate SAM immobilization, His-tagged FKBP12 was in-
jected into the cuvette and the frequency changes that occurred
upon binding to the immobilized SLF were monitored. Figure 2B
shows the resulting sensorgram that was obtained by monitoring
the interaction after injection of His-tagged FKBP12 at the indi-
cated concentration. His-tagged FKBP12 showed a clear frequency
decrease upon binding to SLF immobilized as SAM layers on the
gold electrode surface (Fig. 2B, red line). By contrast, a less dra-
matic frequency change was detected during avidin–biotin immo-
bilization using biotinylated SLF (SLF-bio, Fig. 2A) (Fig. 2B, black
We next applied this SLF-SAM to a T7 phage display screen.
When an aliquot of plaque forming units (pfu) of a freshly prepared
natural-protein T7 phage pool was injected into the cuvette, a fre-
quency change was detected (Fig. 3A). After monitoring the inter-
action for 10 min, the sensor chip was detached from the device.
The genomic DNA of SLF-binding phage on the gold electrode sur-
face was recovered using the host Escherichia coli (BLT5615) solu-
tion and amplified.
A short interaction time (10 min) was
sufficient to recover and amplify the bound phage, which showed
lysis of the host E. coli after 30 min incubation at 37 °C. Sixteen
phage particles were arbitrarily extracted from the resulting T7
phage solution and the enrichment was analyzed by sequencing
the DNA encoding the fusion protein of the phage capsid. The T7
phage display screen using SLF-SAM and QCM resulted in clear
detection of T7 phage whose capsid fusion protein corresponded
to FKBP12, which was included at a rate of 10 out of 16 phage iso-
lates (Fig. 3B, d). In addition, three shorter peptides (Fig. 3B, 4/16,
N: 1/16, j; 1/16, ꢀ), which included an amino-acid sequence sim-
ilar to the proposed SLF analog-binding site within FKBP12(Supple-
mentary Fig. 1) adventitiously generated by the random primer
strategy, were also isolated.4,16 Repeated cycles of screening
tended to bias phage distribution in terms of the growth character-
istics of each phage and lead to poor reproducibility. However, our
methodology does not require multiple rounds of screening.
A conventional phage display protocol involves repeated cycles
of screening in order to enrich for phage particles that specifically
bind to the bait. Technically, an input of 109–1010 pfu/ml of parent
phage pool is required because of the detection limit, which also
Figure 1. Schematic representation of a T7 phage display screen for drug-like small-molecule using a self-assembled probe and a cuvette type QCM apparatus. (1) An aliquot
of T7 phage pool is injected into a cuvette containing saline solution stirred at 1000 rpm. (2) The frequency change that occurs upon binding of the phages to a target of
interest immobilized by SAM onto the gold electrode is monitored in real time. (3) The bound phages are recovered using host cell (E. coli BLT5615, Novagen) solution. (4) The
resulting solution is suspended into another medium. (5) An aliquot (10–100 plaques) of the recovered phage is isolated. (6) The capsid gene-coding region of each phage is
cloned and analyzed by agarose gel electrophoresis followed by DNA sequencing. Using this platform, repeated rounds of selection is not required.

Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
9839
Scheme 1. Synthesis of SLF derivative.
increases the background of non-specifically binding phage parti-
cles. However, the high level of sensitivity of the QCM sensor com-
bined with the efficient SAM immobilization procedure allows the
parent phage pool to be reduced to 107 pfu/ml. Hence, there is a
stochastic reduction in the occurrence of non-specifically bound
phage in this case (Fig. 3B). Consequently, the total number of
phage recovered from the gold electrode surface after an interac-
tion period of 10 min is reduced to 102–103 pfu (data not shown)
compared with 105–106 pfu using a conventional platform. Thus,
our novel screening procedure enables the efficient identification
of phage particles that bind to small-molecules. In particular, our
method enhances the chances of a real-binding sequence emerging
from the screen (typically, 10–100 of plaques are arbitrarily picked
for sequencing) (see Fig. 1 (5 and 6)). For example, in the present
screen we successfully isolated an FKBP12-displaying phage by
sequencing only 16 phage particles. Thus, providing the SAM deriv-
ative can be synthesized, our phage screening procedure using a
QCM device is readily applicable to any small-molecule of interest.
cases where immobilization of a small-molecule may render it
inaccessible to the target protein binding site,4,5 or where the tar-
get protein itself cannot be displayed on a capsid because of prob-
lems of misfolding, low solubility or abundance. Taking these
factors into consideration, we constructed a 15-mer random-pep-
tide T7 phage pool and applied it to our screening platform. Use
of the phage pool enables identification not only of a lock-and-
key or induced-fit type interaction for a disordered flexible loop
but also the small-molecule binding site located on a protein.1,2
Moreover, the resulting small-molecule-selected peptides having
uniform length can conveniently be subjected to RELIC software,
a bioinformatics tool for elucidation of the phage display screen.9
In this experiment, using the same procedure as for the natural-
protein T7 phage pool, a subset of 15-mer peptides were obtained
after four sets of individual one-cycle screens with only 10 min of
monitoring (Fig. 4A and B). These peptides displayed a distribution
of amino-acids that had clearly changed by comparison with that
of the parent T7 phage pool (Supplementary Fig. 4B). In addition,
there are many continuous or discontinuous amino-acid motifs
in the 35 selected peptides calculated by MOTIF1 (Supplementary
Fig. 5) and MOTIF2 program (data not shown) in RELIC software,
which are not readily discernable by manual inspection alone.
Meanwhile, subsequent analysis of the selected peptides showed
a high degree of similarity with a part of SLF analog-binding site
within FKBP12(F37V) (Fig. 4C–F),12,17 as elucidated using HETERO-
align program in RELIC software.9 A similarity plot of 35 peptides
along with the entire sequence of FKBP12(F37V) is shown in
2.3. One-cycle affinity selection of T7 phages exhibiting a
preferential affinity for SLF from a random-peptide T7 phage
pool
The successful identification of FKBP12-displaying T7 phage in a
one-cycle procedure from a natural-protein T7 phage pool using
SLF-SAM was helped by the fact that the FK506 or SLF-binding site
is located at the surface of FKBP12. By contrast, there are many

9840
Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
Figure 2. Validation of immobilization efficiency by SAM. (A) Structure of SLF
derivative forming SAM and SLF-bio. (B) representative QCM sensorgram
A
obtained by monitoring the binding of His-tagged FKBP12 to an SLF-immobilized
gold electrode surface of a sensor chip. The SLF-immobilized ceramic sensor chip
generated by SAM (red line) or biotin–avidin interaction (black line) was individ-
ually attached to the QCM apparatus and the frequency decrease was monitored
after injecting 8
6:67 M) into a cuvette containing 8 ml of buffer. (C) Non-linear plot of frequency
change ( F) against various final concentrations (nM) of His-tagged FKBP12 in the
cuvette. Data are means SD of three independent experiments. SLF-SAM: red line,
SLF-bio: black line. AQUA ver1.5 software (Initium Inc.) was used to determine the
kinetic parameters. 1 Hz = 30 pg.
ll of each sample (1:2 lM, 2:4 lM, 3:8 lM, 4:17 lM, 5:33 lM,
l
Figure 3. Detection of phage binding to an SLF-SAM-immobilized gold electrode
surface of a sensor chip after injection of a natural-protein T7 phage pool. (A) The
SLF-immobilized ceramic sensor chip generated by SAM was first attached to the
QCM apparatus. After injecting a natural-protein T7 phage pool at the indicated
concentration, the frequency decrease was monitored for 10 min. (B) Enrichment of
T7 phage particles within the resulting solution. The binding phages on SLF-SAM-
immobilized gold electrode were recovered and amplified using E. coli (BLT5615)
solution. Sixteen T7 phage particles were arbitrarily extracted from the resulting
solution and then analyzed by PCR amplification of the DNA encoding the fusion
protein. Agarose gel electrophoresis of the PCR products followed by DNA sequence
analysis allowed the 16 clones to be classified into four groups (10/16, d; 4/16, N: 1/
16, j; 1/16, ꢀ). (C) (Poly)peptide sequence that was displayed on the T7 phage (10/
16, d). This sequence fully corresponds to that of FKBP12.
D
Figure 4B, which was calculated using a modified BLOSUM62 score.
A similarity plot of 103 peptides (arbitrarily selected from the
unscreened parent pool) along with the entire sequence of
FKBP12(F37V) was subtracted from the data before plotting. As a
result, this plot clearly pinpointed 38D, 92I, and 98L in
FKBP12(F37V) as residues of maximal similarity score (Fig. 4C).
These three residues had previously been demonstrated to be con-
tact sites for SLF analogs during the binding process. As shown in
Figure 4F, several other amino-acids, including 37V, 83Y, 88H,
91I, and 100F (highlighted in color), were also found to be rela-
tively high- or low-affinity sites of contact, consistent with previ-
ous binding experiments.12,17
protein T7 phage pool or protein extract, this approach is indepen-
dent of the physicochemical properties or abundance of the target
protein.
2.4. General applicability of one-cycle affinity selection of T7
phages exhibiting preferential affinity for small-molecules
By contrast, 27Y, 47F, 56V, 57I, and 60W were not assigned as
likely SLF-binding sites, even though contact of these amino-acids
had previously been demonstrated (Fig. 4F, blue). We reasoned
that these anomalous results could be due to a disturbance during
the selection process caused by the SAM linker, introduced at the
carboxylic group in SLF and aligned with high-density on the gold
electrode. Indeed, the highlighted amino-acids are all located at the
opposite side of the carboxylic group (Scheme 1 and Fig. 4E and F).
Alternatively, as a result of technical limitations, it is likely that
there are a number of sequences absent in the parent phage pool.
The variation of phages possible in this system is up to 108 pfu,
although the theoretical number of whole 15-mer peptides is
2015 (=3.28 ꢀ 1019). Thus, the peptide anticipated to be selected
by SLF (i.e., pinpoint 27Y, 47F, 56V, 57I, and 60W) may not have
been included in the phage pool. Rigorous inspection of the SLF-
binding site together with the identification of target protein is fea-
sible, even though the variation of random-peptide T7 phage pool
is ꢁ108. Furthermore, unlike target identification from a natural-
To further demonstrate the general applicability of this screen-
ing procedure, we attempted another experiment using Iri
(Scheme 2). Iri is a derivative of camptothecin (CPT), a natural
product from the Chinese bush Camptotheca acuminata, with
potent anti-tumor activity.8 This water-soluble compound is a
pro-drug with clinical use that is metabolized to the active SN-38
by liver carboxylesterase (CE) and acts as a topoisomerase I (top
I) inhibitor.18 Recently, acetylcholinesterase (AChE) has been
reported as an alternative direct target of Iri, and the mechanism
of docking between the two molecules has been solved by X-ray
crystallography.19 Using our screening procedure (Fig. 5A), we ob-
tained 29 peptides (Fig. 5B) and successfully pinpointed several
amino-acid residues (121Y, 225Q, 290F, 327E, 440H, and 442Y)
overlapping part of the Iri-binding site in AChE (Fig. 5C–E).19
Likewise, the identical subset of peptides also indicated several
amino-acids (99E, 100L, 252L, 305L, 387I, and 474V) (Fig. 6A) in

Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
9841
Figure 4. Detection of phage binding to SLF-SAM on the gold electrode after injection of a random-peptide T7 phage pool. (A) A representative sensorgram. Experiments were
carried out according to the same procedure with natural-protein T7 phage pool. (B) SLF-SAM-selected 15-mer peptide sequence after four sets of individual one-cycle
screens. T7 phage particles were arbitrarily extracted from the resulting solution and then analyzed by PCR amplification of the DNA encoding the fusion peptide. The PCR
products were then sequenced. (C) Similarity scores for the sequence of FKBP12(F37V) (1M-108E) against the sequences of 35 peptides selected for affinity to SLF-SAM
calculated using HETEROalign program in RELIC software (http://relic.bio.anl.gov/index.aspx). The similarity scores of random-peptides chosen without affinity selection
(Supplementary Fig. 2) have been subtracted from these scores to remove pool bias. The maximal similarity score was observed for 38D (35.9), 92I (22.2) and 98L (31.2) (each
peak is highlighted in blue). (D) Cluster diagram for the SLF-SAM-selected 15-mer peptide sequences in which each peptide is aligned to the input FKBP12(F37V) sequence at
the position of maximal similarity. Residues exhibiting identity or similarity to the protein sequence are highlighted in red or orange. X indicates an amino-acid residue
located within a distance of 5 Å from SLF analog-bound FKBP12(F37V). (E) Three structural models of SLF analog-FKBP12(F37V) complex (PDB ID: 1BL4). The peptide
backbone of FKBP12(F37V) is shown in cartoon; SLF analog is shown in CPK. Oxygen in SLF analog is shown in red. Similarity is coded by color; red indicates the highest
similarity and blue the lowest. (F) A close up view of the SLF analog-binding site rendered with RasMol with the amino-acid residues predicted to be involved in SLF-binding
represented as a stick. SLF analog is shown in stick.
the vicinity of the catalytic triad (221S, 353E, and 467H) in CE,
which potentially corresponds to the scaffold for Iri during the dee-
sterification (Fig. 6B and C).20 From our experimental data, to-
gether with the previous report, the likely location of each
amino-acid during the Iri docking can be determined (see
Fig. 6D). In our previous experiment using CPT-10-B, a portion of
beta strand (210G-220E) on the opposite side of 474V was also
identified,4 verifying the accuracy of this data. Unlike the case of
AChE, docking between Iri and native CE may not be directly de-
tected in either the X-ray or NMR structures because the enzymatic
reaction is likely to proceed smoothly under the experimental con-
ditions required for these methods. Nonetheless, our strategy,
which depends on the primary peptide sequence, enables direct
detection of the scaffold in the vicinity of the catalytic site.
positives or potential Iri-binding sites not involved in the bioactiv-
ity (generally categorized as ‘‘non-specific binding”). Furthermore,
there are amino-acid residues in AChE, and perhaps in CE, known
to be involved in Iri-binding that were undetected during our anal-
ysis. Our inability to select these residues could be due to problems
associated with the linker or the limited size of the parent phage
pool as discussed earlier for SLF. Thus, these amino-acids may be
rendered detectable by altering the position of the introduced lin-
ker in Iri, increasing the number of Iri-selected peptides, or increas-
ing the variation of the parent library to as close to the theoretical
number as possible.
3. Conclusion
However, part of the beta strand away from the well-defined
Iri-binding site was also highlighted for both AChE (Fig. 5D) and
CE (Fig. 6B). These highlighted residues can be regarded as false
In summary, we have successfully established an effective
small-molecule-oriented T7 phage display screen as a one-cycle

9842
Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
Scheme 2. Synthesis of Iri derivative.
method for identifying interacting species using both natural-pro-
tein and random-peptide T7 phage pools after an interaction time
of only 10 min. By combining a QCM apparatus and SAM immobi-
lization with T7 phage display, target discovery of various small-
molecules can be effectively and universally accomplished. As a
high-throughput screening platform, our method could dramati-
cally facilitate the detection of T7 phage(s) that can bind to an
immobilized small-molecule to validate the molecular interac-
tion(s), even using a new bioactive compound or a functionally un-
known target. We believe this methodology will contribute to the
development of forward chemical genetics together with a range of
other applications.
Novagen (Madison, WI). Klenow DNA polymerase I was purchased
from USB Corporation (Cleveland, OH). Neutravidin was obtained
from PIERCE (Rockford, IL). Ni–NTA resin was obtained from GE
healthcare (Amersham, UK).
4.2. General (chemistry)
All non-aqueous reactions were carried out using freshly dis-
tilled solvents under an atmosphere of argon. All reactions were
monitored by TLC, which was carried out on Silica Gel 60 F254
plates (E. Merck, Darmstadt, Germany). Flash chromatography sep-
arations were performed on PSQ 100B (Fuji Silysia Co., Ltd, Japan).
The NMR spectra (1H, 13C) were determined on a Bruker
600 MHz or 400 MHz spectrometer (Avance DRX-600, Avance
DRX-400) or a JEOL 400 MHz spectrometer (JNM-LD400), using
CDCl3 (with TMS for 1H NMR and chloroform-d for 13C NMR as
the internal reference) solution, unless otherwise noted. Chemical
shifts were expressed in parts per million (ppm) and coupling
constants are in Hertz. Optical rotations were recorded on a JASCO
P-1030 digital polarimeter using the sodium D line at room
4. Experimental
4.1. Materials
The QCM apparatus (AffinixQ) and ceramic sensor chip were
purchased from Initium Inc. (Tokyo, Japan). T7select10-3 OrientEx-
pressTM cDNA Cloning System and pET28a(+) vector were from

Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
9843
Figure 5. Detection of phage binding to Iri-SAM on the gold electrode after injection of a random-peptide T7 phage pool. Experiments were carried out according to the same
procedure with SLF-SAM. (A) A representative sensorgram. (B) Iri-SAM-selected 15-mer peptide sequence after three sets of individual one-cycle screens. (C) Similarity scores
for the sequence of TcAChE (4S-536A) against the sequences of peptides selected for affinity to Iri-SAM calculated using HETEROalign program in RELIC software. The
similarity scores of random-peptides chosen without affinity selection (Supplementary Fig. 2) have been subtracted from these scores to remove pool bias. The maximal
similarity score was observed for 121Y (38.8), 225Q (26.2), 290F (18.0), 327E (22.4), and 440H, 442Y (16.3 and 23.5) (highlighted in blue), which are involved in the Iri-
binding site. (D) Three structural models of Iri-TcAChE complex (PDB ID: 1U65). The peptide backbone of AChE is shown in cartoon; Iri is shown in CPK. Oxygen and nitrogen
in Iri are shown in red and cyan, respectively. Similarity is coded by color; red indicates the highest similarity and blue the lowest. (E) A close up view of the Iri-binding site
rendered with RasMol with the amino-acid residues predicted to be involved in Iri-binding represented as a stick. Iri is shown in stick.
temperature, using CHCl3 as a solvent unless otherwise noted.
Infrared spectra (IR) were recorded on a Jasco FT/IR-410 spectrom-
eter using NaCl (neat) or KBr pellets (solid), and were reported as
wavenumbers (cmꢂ1). Mass spectra (MS) were obtained on an Ap-
plied Biosystems mass spectrometer (API QSTAR pulsar i) under
conditions as high resolution, using polyethylene glycol as internal
standard.
166.6, 157.7, 154.6, 148.9, 147.4, 145.6, 141.8, 133.3, 129.9, 125.3
(2ꢀ), 122.2 (2ꢀ), 120.4, 120.1, 114.6, 113.1, 111.7, 111.3, 76.4,
65.3, 55.9, 55.8, 51.2, 46.7, 44.1, 38.0, 32.4, 31.2, 26.3, 24.9, 23.4,
23.1, 21.1, 8.7; IR (neat) 3021, 2937, 2861, 1786, 1738, 1701,
1640, 1592, 1521, 1444, 1348, 1210, 1143, 1082, 1029, 992, 921,
864, 756 cmꢂ1
;
HRMS (ESI) calcd for C38H44N2O11 [M+Na]+:
727.2837, found 727.2800.
4.3. Synthesis of the SLF derivative that forms SAM
4.3.2. Bis{10-[11-(N-tert-butoxycarbonylamino)-3,6,9-trioxa-
undecylcarbamoyl]undecanyl} disulfide (5)
4.3.1. (1R)-3-(3,4-dimethoxyphenyl)-1-[3-(4-nitrophenoxycarbonyl-
methoxy)phenyl]propyl (2S)-1-(3,3-dimethyl-2-oxovaleryl)-piper-
idine-2-carboxylate (SLF p-nitrophenyl ester, 2)
To a solution of 3 (221 mg, 0.75 mmol)21 in pyridine (3 ml) was
added
4
(204 mg, 0.30 mmol)22 followed by DMAP (3.7 mg,
0.03 mmol) at rt. The mixture was stirred for 17 h. Then the mixture
was quenched by the addition of H2O and diluted with EtOAc. The
layers were separated, and the organic layer was washed with brine
and dried with Na2SO4. After the organic layer was concentrated, the
residuewaspurifiedbysilicagelchromatographyto yield5 (249 mg,
84%) as a foam. 1H NMR (600 MHz, CDCl3) d = 6.18 (1H ꢀ 2, br s), 5.03
(1H ꢀ 2, br s), 3.64 (8H ꢀ 2, br m), 3.56 (4H ꢀ 2, m), 3.45 (2H ꢀ 2, m),
3.32 (2H ꢀ 2, m), 2.67 (2H ꢀ 2, t, J = 7.5 Hz), 2.17 (2H ꢀ 2, t,
J = 7.5 Hz), 1.66 (4H ꢀ 2, m), 1.45 (9H, s), 1.36 (2H ꢀ 2, m), 1.26
(10H ꢀ 2, br s); 13C NMR (100 MHz, CDCl3) d = 173.3 (2ꢀ), 156.0
(2ꢀ), 70.42 (2ꢀ), 70.39 (2ꢀ), 70.2 (2ꢀ), 70.14 (4ꢀ), 70.0 (2ꢀ), 40.3
(2ꢀ), 39.09 (2ꢀ), 39.06 (2ꢀ), 36.7 (2ꢀ), 29.41 (2ꢀ), 29.38 (2ꢀ),
29.32 (2ꢀ), 29.28 (2ꢀ), 29.17 (2ꢀ), 29.15 (2ꢀ), 28.5 (2ꢀ), 28.4
(6ꢀ), 25.7 (2ꢀ); IR (KBr) 3309, 3075, 2920, 2854, 1691, 1641, 1544,
1470, 1419, 1339, 1337, 1284, 1123, 1042, 868, 782, 758,
To a solution of 1 (172 mg, 0.29 mmol) and p-nitrophenol
(52 mg, 0.37 mmol) in CH2Cl2 (5 ml) was added a solution of DCC
(196 mg, 0.95 mmol) in CH2Cl2 (2 ml). Then, DMAP (43.2 mg,
0.35 mmol) was added to the mixture. The mixture was stirred at
rt for 80 min and then concentrated. The residue was purified by
silica gel chromatography (hexane/EtOAc = 4:1 to 2:1) to give 2
(194 mg, 94%) as a ca. 4:1 tautomeric mixture as yellow oil.
½
a 2D4
ꢃ
= ꢂ7.2 (c = 1.8 in CHCl3); 1H NMR (600 MHz, CDCl3, major tau-
tomer) d = 8.27 (2H, m), 7.33 (2H, m), 7.31 (1H, m), 7.03 (1H, d,
J = 7.7 Hz), 7.00 (1H, br s), 6.93 (1H, dd, J = 8.2 Hz, 2.5 Hz), 6.77
(1H, m), 6.68 (2H, m), 5.80 (1H, dd, J = 7.9 Hz, 5.7 Hz), 5.32 (1H,
br d, J = 5.3 Hz), 4.96 (2H, s), 3.85 (6H, s), 3.35 (1H, br d,
J = 12.7 Hz), 3.16 (1H, td, J = 13.1 Hz, 2.9 Hz), 2.57 (2H, m), 2.36
(1H, m), 2.24 (1H, m), 2.07 (1H, m), 1.72 (5H, m), 1.45 (1H, m),
1.35 (2H, m), 1.22 (3H, s), 1.21 (3H, s), 0.88 (3H, t, J = 7.4 Hz); 13C
NMR (600 MHz, CDCl3, major tautomer) d = 207.8, 169.7, 167.3,
713 cmꢂ1; HRMS (ESI) calcd for C48H94N4O12Na2S2 [M+2Na]2+
514.3135, found 514.3118.
:

9844
Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
centrated. Theresultingaminesaltwasuseddirectlyinthenextstep.
1H NMR (600 MHz, CD3OD) d = 3.70 (2H ꢀ 2, m), 3.68 (4H ꢀ 2, br s),
3.65 (2H ꢀ 2, m), 3.62 (2H ꢀ 2, m), 3.53 (2H ꢀ 2, t, J = 5.6 Hz), 3.35
(2H ꢀ 2, t, J = 5.6 Hz), 3.13 (2H ꢀ 2, brt, J = 4.9 Hz), 2.67 (2H ꢀ 2, t,
J = 7.3 Hz), 2.18 (2H ꢀ 2, t, J = 7.3 Hz), 1.66 (2H ꢀ 2, quintet,
J = 7.3 Hz), 1.59 (2H ꢀ 2, m), 1.39 (2H ꢀ 2, m), 1.31 (10H ꢀ 2, br s).
4.3.4. Bis(SLF-DEG-undecanyl) disulfide (7)
To a solution of 2 (58.4 mg, 0.076 mmol) in pyridine (3 ml) was
added a solution of 6 (136.4 mg, 0.194 mmol) in CH2Cl2 (3 ml) at
rt, and the mixture was stirred at rt for 7 h. Then the mixture was
concentrated, and the residue was purified by silica gel chromatog-
raphy to yield 7 (29.1 mg, 21%) as a ca. 4:1 tautomeric mixture in the
form of a colorless oil. ½a D24
ꢃ
= ꢂ3.6 (c = 0.67 in CHCl3); 1H NMR
(600 MHz, CDCl3, major tautomer) d = 7.31 (1H ꢀ 2, t, J = 7.9 Hz),
7.05 (1H ꢀ 2, brt, J = 5.6 Hz), 7.00 (1H ꢀ 2, br d, J = 7.7 Hz), 6.93
(1H ꢀ 2, br s), 6.86 (1H ꢀ 2, dd, J = 7.9 Hz, 2.0 Hz), 6.78 (1H ꢀ 2, m),
6.68 (1H ꢀ 2, br s), 6.67 (2H ꢀ 2, m), 6.09 (1H ꢀ 2, br m), 5.77 (1H,
dd, J = 7.9 Hz, 5.6 Hz), 5.31 (1H ꢀ 2, d, J = 5.4 Hz), 4.50 (2H ꢀ 2, s),
3.86 (3H ꢀ 2, s), 3.85 (3H ꢀ 2, s), 3.63 (8H ꢀ 2, br s), 3.60 (2H ꢀ 2,
m), 3.57 (2H ꢀ 2, m), 3.52 (2H ꢀ 2, m), 3.44 (2H ꢀ 2, m), 3.37
(1H ꢀ 2, br d, J = 12.3 Hz), 3.18 (1H ꢀ 2, td, J = 13.1 Hz, 2.9 Hz), 2.67
(2H ꢀ 2, t, J = 7.4 Hz), 2.61 (1H ꢀ 2, m), 2.53 (1H ꢀ 2, m), 2.38
(1H ꢀ 2, br d, J = 13.7 Hz), 2.25 (1H ꢀ 2, m), 2.15 (2H ꢀ 2, t,
J = 7.4 Hz), 2.05 (1H ꢀ 2, m), 1.80–1.60 (12H ꢀ 2, m), 1.50 (1H ꢀ 2,
m), 1.35 (4H ꢀ 2, m), 1.27 (10H ꢀ 2, br s), 1.23 (3H ꢀ 2, s), 1.21
(3H, s), 0.89 (3H ꢀ 2, t, J = 7.4 Hz); 13C NMR (100 MHz, CDCl3, major
tautomer) d = 207.8 (2ꢀ), 173.2 (2ꢀ), 169.7 (2ꢀ), 168.1 (2ꢀ), 167.3
(2ꢀ), 157.4 (2ꢀ), 148.9 (2ꢀ), 147.4 (2ꢀ), 141.9 (2ꢀ), 133.3 (2ꢀ),
130.0 (2ꢀ), 120.1 (2ꢀ), 120.1 (2ꢀ), 114.2 (2ꢀ), 113.3 (2ꢀ), 111.3
(2ꢀ), 76.5 (2ꢀ), 70.5 (4ꢀ), 70.3 (2ꢀ), 70.2 (2ꢀ), 70.0 (2ꢀ), 69.7 (2ꢀ),
67.4 (2ꢀ), 55.9 (2ꢀ), 55.8 (2ꢀ), 55.8 (2ꢀ), 51.3 (2ꢀ), 46.7 (2ꢀ), 44.1
(2ꢀ), 39.12 (2ꢀ), 39.10 (2ꢀ), 38.8 (2ꢀ), 36.7 (2ꢀ), 32.5 (2ꢀ), 31.2
(2ꢀ), 29.7 (2ꢀ), 29.5 (2ꢀ), 29.42 (2ꢀ), 29.36 (2ꢀ), 29.3 (2ꢀ), 29.21
(2ꢀ), 29.20 (2ꢀ), 28.5 (2ꢀ), 26.4 (2ꢀ), 25.7 (2ꢀ), 24.5 (2ꢀ), 23.5
(2ꢀ), 23.2 (2ꢀ), 21.2 (2ꢀ), 8.7 (2ꢀ); IR (neat) 3342, 2928, 2857, 1738,
1645, 1517, 1444, 1351, 1261, 1141, 1030, 992 cmꢂ1; HRMS (ESI) calcd
for C102H56N6O24Na2S2 [M+2Na]2+: 979.5198, found 979.5220. calcd
for C102H56N6O24NaS2 [M+Na]+: 1936.0504, found 1936.0549.
4.4. Syntheisis of biotinylated SLF derivative (SLF-bio)
The SLF-bio was synthesized according to previous reports.13,14
4.5. Synthesis of the Iri derivative that forms SAM
4.5.1. Bis{10-{[11-(4-nirtophenoxycarbonylamino)-3,6,9-
trioxaundecanyl]carbamoyl}undecanyl} disulfide (8)
To a solution of 5 (81 mg, 0.082 mmol) in CH2Cl2 (1 ml) was
added TFA (0.25 ml). The mixture was stirred at rt for 3.5 h and
then concentrated. To a solution of the resulting ammonium salt
in pyridine (1.2 ml) was added DMAP (1.6 mg, 0.013 mmol) and
bis(4-nirtophenyl)carbonate (100 mg 0.33 mmol). The mixture
was stirred at rt for 19 h and then concentrated, and the residue
was purified by silica gel chromatography (CHCl3/ MeOH = 50:1
to 25:1) to yield diester 8 (95 mg, quant) as a pale yellow crystal.
1H NMR (400 MHz, CDCl3) d= 8.24 (2H ꢀ 2, d, J = 9.3 Hz), 7.32
(2H ꢀ 2, d, J = 7.3 Hz), 6.25–6.00 (2H ꢀ 2, m), 3.90–3.32 (16H ꢀ 2,
m), 2.67 (2H ꢀ 2, t, J = 7.3 Hz), 2.16 (2H ꢀ 2, t, J = 7.7 Hz), 1.69–
1.51 (4H ꢀ 2, m), 1.49–1.20 (12H ꢀ 2, m); HRMS (ESI) calcd for
C52H84N6O16S2Na [M+Na]+: 1135.5277, found 1135.5254.
Figure 6. Analysis of interaction between Iri and rCE using a subset of Iri-SAM-
selected peptides. (A) Similarity scores for the sequence of rCE (23P-556R) against the
sequences of peptides selected for affinity to Iri-SAM (Fig. 5B) calculated using
HETEROalign program in RELIC software. The similarity scores of random-peptides
chosen without affinity selection (Supplementary Fig. 2) have been subtracted from
these scores to remove pool bias. The maximal similarity score was observed for 99E,
100L (63.9 and 63.6), 252L (53.4), 305L (14.0), 387I (18.0), and 474V (38.7)
(highlighted in blue), which are likely to be bound directly during the deesterification
of Iri. (B) Three structural models of rCE (PDB ID: 1K4Y). The peptide backbone of CE is
shown in cartoon. Similarity is coded by color; red indicates the highest similarity and
blue the lowest. Amino-acid residues of the catalytic triad (221S, 353L, and 467H) are
shown in yellow stick. (C) A close up view of the vicinity of the catalytic triad rendered
with RasMol with the amino-acid residues predicted to be involved in Iri-binding
represented as
a stick. (D) Possible location of Iri (white) in CE during the
deesterification. Gray arrow at the back shows part of beta strand (210G-220E),
which was identified during our previous T7 phage display screen using CPT-10-B.
4.3.3. Bis{10-[(11-amino-3,6,9-trioxaundecanyl)carbamoyl]-
undecanyl} disulfide, bistrifluoroacetic acid salt (6)
To a solution of 5 (20 mg, 0.02 mmol) in CH2Cl2 (2 ml) was added
TFA (0.5 ml). The mixture was stirred at rt for 45 min and then con-
4.5.2. 20-(N-Boc-b-alanyl) irinotecan (10)
To a solution of irinotecan (9) (51 mg, 0.075 mmol) in CH2Cl2
(9 ml) was added N-Boc-b-alanine (29 mg, 0.15 mmol). Then, 2-
chloro-1-methlpyridium iodide (116 mg, 0.45 mmol) and DMAP

Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
9845
(92 mg, 0.75 mmol) was added to the mixture successively at 0 °C.
The mixture was warmed and stirred at rt for 17 h and then concen-
trated. Then the mixture was quenched by the addition of 0.1 M HCl
and diluted with CH2Cl2. The layers were separated, and the organic
layer was washed with brine and dried with Na2SO4. After the or-
ganic layer was concentrated, the residue was purified by silica
gel chromatography (CHCl3/MeOH = 95:5 to 75:25) to yield 10
the screening procedure, the phage pool was amplified up to
1.0 ꢀ 1010 pfu/ml using E. coli (BLT5615) as the host strain.
4.7. Construction of the T7 phage pool using synthetic DNA
For the preparation of a duplex DNA library, oligonucleotide
GGGGATCCGAATTCT(NNK)15TGAAAGCTTCTCGAGGG (0.056 pM)
and CCCTCGAGAAGCTTTCA (0.56 pM) were mixed with Klenow
buffer, heated to 95 °C for 5 min and annealed by cooling the mix-
ture to 37 °C. The single-stranded regions were converted to du-
plex DNA by continuing the incubation at 37 °C for 2 h in the
(35 mg, 61%) as a yellow crystal. ½a D25
ꢃ
= ꢂ3.3 (c = 1.1 in CHCl3); 1H
NMR (400 MHz, CDCl3) d = 8.20 (1H, d, J = 9.3 Hz), 7.84 (1H, d,
J = 2.4 Hz), 7.60 (1H, dd, J = 2.6 Hz, 9.2 Hz), 7.17 (1H, s), 5.70 (1H,
d, J = 17.4 Hz), 5.41 (1H, d, J = 17.1 Hz), 5.26 (2H, s), 5.21–5.18 (1H,
m), 4.46–4.30 (2H, m), 3.65–3.35 (2H, m), 3.16 (2H, q, J = 7.7 Hz),
3.07 (1H, t, J = 12.7 Hz), 2.91 (1H, t, J = 12.2 Hz), 2.82–2.72 (1H, m),
2.80–2.50 (6H, m), 2.26 (1H, dq, J = 7.1 Hz, 14.7 Hz), 2.15 (1H, dq,
J = 7.1 Hz, 14.7 Hz), 2.03–1.26 (22H, m), 1.00 (3H, t, J = 7.6 Hz); 13C
NMR (100 MHz, CDCl3) d = 171.6, 167.7, 157.4, 155.9, 153.1, 151.4,
150.5, 147.1, 147.0, 146.0, 145.43, 145.40, 131.6, 131.0, 127.6,
127.1, 127.0, 125.9, 119.6, 114.6, 95.7, 79.4, 77.3, 76.3, 67.2, 62.5,
50.2, 49.3, 44.3, 44.0, 36.4, 34.69, 34.67, 31.8, 31.6, 29.7, 28.4,
28.0, 27.3, 25.6, 25.3, 24.3, 23.2, 22.7, 20.7, 14.1, 14.0, 7.6; IR (neat)
1716, 1662, 1610 cmꢂ1; HRMS (ESI) calcd for C41H52N5O9 [M+H]+:
758.3765, found 758.3753.
presence of dNTPs (2.5 mM) and Klenow enzyme (0.5 mU/ll). After
the reaction, double-stranded DNA was recovered by EtOH precip-
itation. The obtained DNA was then digested separately with EcoRI
and HindIII restriction enzyme and inserted into the T7select10-3b
vector according to the manufacture’s instructions.16,23 The pri-
mary titer of this T7 phage pool was 1.6 ꢀ 107 pfu/ml. For the
screening procedure, the phage pool was amplified up to
1.7 ꢀ 1010 pfu/ml using E. coli (BLT5615) as the host strain.
4.8. Expression of Six Histidine-tagged FK506-binding protein
12 (FKBP12)
4.5.3. 20-(b-alanyl) irinotecan ammumonium salt (11)
The cDNA encoding human FKBP12 was cloned into the expres-
sion vector pET28a(+) (Novagen), and the E. coli BL21(DE3) was
transformed with the vector. A single colony was inoculated into
To 10 (33 mg, 0.043 mmol) was added TFA (1.0 ml). The mixture
was stirred at rt for 15 min and then concentrated. The resulting
ammonium salt was used directly without further purification.
15 ml of LB medium containing 1% glucose and 50 lg/ml of kana-
½
a 2D5
ꢃ
= ꢂ39.1 (c = 1.0 in MeOH); 1H NMR (400 MHz, CD3OD)
mycin and cultured at 30 °C. The cells were grown overnight and
then used to inoculate another 1000 ml of the same medium at
37 °C. After 3 h of growth, heterologous expression of FKBP12
was induced by addition of 1 mM of IPTG. The culture was then
continued for a further 3 h at 37 °C. The cultured cells were har-
vested by centrifugation at 1500g for 10 min at 4 °C. The cells were
resuspended in PBS and centrifuged at 1100g for 10 min at 4 °C.
After removing the supernatant, the cell paste was frozen using li-
quid nitrogen and then stored at ꢂ80 °C until use.
d = 8.16 (1H, d, J = 9.0 Hz), 8.02 (1H, d, J = 2.4 Hz), 7.68 (1H, dd,
J = 2.4 Hz, 9.3 Hz), 7.37 (1H, s), 5.63 (1H, d, J = 16.8 Hz), 5.49 (1H, d,
J = 16.8 Hz), 5.35 (2H, s), 4.58 (1H, br s), 4.39 (1H, br s), 3.65–2.97
(13H, m), 2.40–1.73 (11H, m), 1.56 (1H, m), 1.40 (3H, t, J = 7.6 Hz),
1.05 (3H, t, J = 7.4 Hz); 13C NMR (100 MHz, CD3OD) d = 171.3,
169.3, 158.8, 154.5, 152.5, 151.6, 148.0, 147.8, 147.7, 147.6, 131.8,
130.0, 129.2, 128.7, 127.2, 120.3, 116.3, 97.6, 78.6, 67.8, 64.7, 58.3,
51.4, 50.7, 49.9, 44.3, 44.0, 36.1, 35.0, 33.0, 32.3, 31.9, 30.7, 30.5,
30.37, 30.34, 30.2, 27.6, 27.2, 26.8, 26.1, 25.3, 24.4, 23.9, 23.7, 22.9,
20.8, 18.4, 14.4, 14.3, 8.1; IR (neat) 1672, 1606 cmꢂ1; HRMS (ESI)
calcd for C36H44N5O7 [M+H]+: 658.3240, found 658.3221.
4.9. Purification of His-tagged FKBP12
The FKBP12-producing cells were sonicated in buffer A (50 mM
phosphate, pH 8.0, 300 mM NaCl, 10 mM imidazole (Im), 1 mM phen-
4.5.4. Bis[{20-(b-alanyl) irinotecan}-DEG-undecanyl] disulfide (12)
To a solution of 11 (6.4 mg, 0.0083 mmol) in pyridine (0.5 ml) and
CH2Cl2 (0.1 ml) was added 8 (4.0 mg, 0.0036 mmol) at rt, and the mix-
ture was stirred at rt to 40 °C for 3 days. Then the mixture was concen-
trated, and the residue was purified by silica gel chromatography
(CHCl3/MeOH = 95:5 to 2:1) to yield 12 (4.8 mg, 62%) as a yellow crys-
ylmethylsulfonylfluoride (PMSF), 1
fuged for 30 minat 20,400g at 4 °C. The supernatant was filtered using
a PVDF membrane (Millex, pore size 0.45 m, Millipore, Billerica, MA)
lg/mlleupeptin)andthen centri-
l
and then loaded onto a Ni–NTA resin (GE healthcare) pre-equilibrated
with buffer A. The column was washed in turn with buffer B (50 mM
phosphate, pH 8.0, 300 mM NaCl, 20 mM Im), buffer C (50 mM phos-
phate, pH 8.0, 300 mM NaCl, 30 mM Im) and buffer D (50 mM phos-
phate, pH 8.0, 300 mM NaCl, 40 mM Im). Bound proteins were then
eluted with buffer E (50 mM phosphate, pH 8.0, 300 mM NaCl,
250 mM Im). The fractions containing the His-tagged FKBP12 were
collected and dialyzed into TEMG buffer (50 mM Tris–HCl, pH 7.5,
1 mM EDTA, 50 mM NaCl, 10% (v/v) glycerol, 5 mM 2-mercap-
toethanol) for the following purification by FPLC. The protein was
loaded onto HiTrap DEAE column pre-equilibrated with TEMG buffer,
andtheflow-throughfractionwasthensubjectedtoHiTrapSPcolumn
chromatography. The flow-through fraction was collected and dia-
lyzed into FKBP buffer (25 mM Tris–HCl, pH 7.5, 50 mM NaCl, 1 mM
MgCl2, 1 mM CaCl2, 10%(v/v) glycerol, 5 mM 2-mercaptoethanol,
0.05%(v/v) Nonidet P-40) and then stored at ꢂ80 °C until use.
tal. ½a 2D6
ꢃ
= ꢂ8.6 (c = 0.45 in MeOH); 1H NMR (400 MHz, CD3OD)
d = 8.11 (1H ꢀ 2, d, J = 9.0 Hz), 7.89 (1H ꢀ 2, d, J = 0.96 Hz), 7.56
(1H ꢀ 2, dd, J = 9.3 Hz, 0.96 Hz), 7.31 (1H ꢀ 2, s), 5.61 (1H ꢀ 2, d,
J = 16.6 Hz), 5.46 (1H ꢀ 2, d, J = 16.6 Hz), 5.23 (2H ꢀ 2, s), 4.58
(1H ꢀ 2, br s), 4.39 (1H ꢀ 2, br s), 3.75–3.08 (26H ꢀ 2, m), 3.08–2.95
(1H ꢀ 2, m), 2.92–2.72 (2H ꢀ 2, m), 2.72–2.56 (2H ꢀ 2, m), 2.30–
1.15 (33H ꢀ 2, m), 1.03 (3H ꢀ 2, t, J = 7.5 Hz); 13C NMR (100 MHz,
CD3OD) d = 176.4, 172.7, 169.4, 160.8, 158.92, 158.87, 154.5, 151.7,
148.4, 148.0, 147.9, 133.5, 132.4, 132.3, 132.0, 130.1, 129.9, 127.0,
120.3, 116.5, 97.9, 77.8, 67.7, 67.3, 66.9, 64.8, 58.3, 51.4, 50.8, 44.4,
44.2, 44.0, 41.0, 40.3, 39.8, 37.1, 36.9, 35.7, 33.0, 32.0, 30.7, 30.5,
30.4, 30.4, 30.2, 30.1, 29.6, 29.4, 27.0, 26.9, 25.4, 24.5, 23.9, 23.7,
22.9, 18.4, 14.3, 8.2; IR (neat) 1682, 1435 cmꢂ1; HRMS (ESI) calcd for
C112H164N14O24S2 [M+4H]2+: 1076.5736, found 1076.5771.
4.6. Construction of a natural-protein T7 phage pool
4.10. Procedure for the T7 phage display screen using a cuvette
type QCM apparatus
A natural-protein T7 phage pool was constructed using Jurkat
cell-derived cDNA according to the manufacture’s instructions.16,23
The primary titer of this T7 phage pool was 5.8 ꢀ 106 pfu/ml. For
A 20
ll aliquot of SLF derivative (7) (1.3 mM in 75% EtOH) or Iri
derivative (12) (1 mM in 75% EtOH) was dropped onto the gold elec-

9846
Y. Takakusagi et al. / Bioorg. Med. Chem. 16 (2008) 9837–9846
trode of the ceramic sensor chip and left for 16 h under a humid and
shaded atmosphere at room temperature. The surface of the electrode
was washed for 10 min in buffer (10 mM Tris–HCl, pH 8.0, 200 mM
NaCl), which was stirred at 1000 rpm. The sensor chip was setup for
the QCM apparatus with the cuvette containing 8 ml of buffer. The
Supporting information available
Additional experimental details. This material is available free
of charge via the Internet at http://www.Xxx.xxxx. Supplementary
data associated with this article can be found, in the online version,
at doi:10.1016/j.bmc.2008.09.061.
QCM sensor was then allowed to fully stabilize. An aliquot of 8
l
l of
lofnatural-protein
ll of random-peptide T7
His-taggedFKBP12(2,4, 8, 17,33, and67 lM),16 l
T7 phage pool (1.0 ꢀ 1010 pfu/ml) and 8
References and notes
phage pool (1.7 ꢀ 1010 pfu/ml) were individually injected into the
cuvette. Frequencychanges,causedbybindingtothe SLFimmobilized
on the gold electrodesurface, were thenmonitored for 10 min. For the
1. Rodi, D. J.; Agoston, G. E.; Manon, R.; Lapcevich, R.; Green, S. J.; Makowski, L.
Comb. Chem. High Throughput Screen. 2001, 4, 553.
2. Rodi, D. J.; Janes, R. W.; Sanganee, H. J.; Holton, R. A.; Wallace, B. A.; Makowski,
L. J. Mol. Biol. 1999, 285, 197.
3. Smith, G. P.; Petrenko, V. A. Chem. Rev. 1997, 97, 391.
4. Takakusagi, Y.; Takakusagi, K.; Kuramochi, K.; Kobayashi, S.; Sugawara, F.;
Sakaguchi, K. Bioorg. Med. Chem. 2007, 15, 7590.
5. Takakusagi, Y.; Kuroiwa, Y.; Sugawara, F.; Sakaguchi, K. Bioorg. Med. Chem.
2008, 16, 7410.
recovery of bound phages, 10 ll of host E. coli (BLT5615) solution (cul-
tured for 30 min at 37 °C in the presence of 1 mM of IPTG beforehand)
was dropped onto the gold electrode and then incubated at 37 °C for
30 min. To the resulting solution was then added another 200 ll of
LB medium. An aliquot of phage was then extracted from this solution
and subjected to PCR analysis followed by agarose gel electrophoresis
and DNA sequencing.
For SLF-bio immobilization, the gold sensor surface was initially
coated by treatment with 100 ll of neutravidin solution (100 lg/ml
aq) for 30 min at room temperature. After washing the gold electrode
surface, SLF-bio was applied and left for 16 h under a humid and
shaded atmosphere at room temperature. The gold surface was thor-
oughly washed prior to conducting the screening experiments.
6. Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M. Chem. Rev.
2005, 105, 1103.
7. Holt, D. A.; Luengo, J. I.; Yamashita, D. S.; Oh, H.; Konialian, A. L.; Yen, H.; Rozamus,
L. W.; Brandt, M.; Bossard, M. J.; Levy, M. A.; Eggleston, D. S.; Liang, J.; Schultz, L.
W.; Stout, T. J.; Clardy, J. J. Am. Chem. Soc. 1993, 115, 9925.
8. Garcia-Carbonero, R.; Supko, J. G. Clin. Cancer Res. 2002, 8, 641.
9. Mandava, S.; Makowski, L.; Devarapalli, S.; Uzubell, J.; Rodi, D. J. Proteomics
2004, 4, 1439.
10. Rodi, D. J.; Makowski, L. Pac. Symp. Biocomput. 1999, 4, 532.
11. Furuya, M.; Haramura, M.; Tanaka, A. Bioorg. Med. Chem. 2006, 14, 537.
12. Clackson, T.; Yang, W.; Rozamus, L. W.; Hatada, M.; Amara, J. F.; Rollins, C. T.;
Stevenson, L. F.; Magari, S. R.; Wood, S. A.; Courage, N. L.; Lu, X.; Cerasoli, F. J.;
Gilman, M.; Holt, D. A. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 10437.
13. Keenan, T.; Yaeger, D. R.; Courage, N. L.; Rollins, C. T.; Pavone, M. E.; Rivera, V.
M.; Yang, W.; Guo, T.; Amara, J. F.; Clackson, T.; Gilman, M.; Holt, D. A. Bioorg.
Med. Chem. 1998, 6, 1309.
4.11. Agarose gel electrophoresis and DNA sequencing
Agarose gel electrophoresis and DNA sequencing were per-
formed as described previously.4
14. McKenzie, K. M.; Videlock, E. J.; Splittgerber, U.; Austin, D. J. Angew. Chem. Int.
Ed. Engl. 2004, 43, 4052.
15. Bain, C. D.; Troughton, E. B.; Tao, Y.; Evall, J.; Whitesides, G. M.; Nuzzo, R. G. J.
Am. Chem. Soc. 1989, 111, 321.
4.12. RELIC software
16. Novagen. OrientExpressTM cDNA Manual, 1999, TB247.
17. Yang, W.; Rozamus, L. W.; Narula, S.; Rollins, C. T.; Yuan, R.; Andrade, L. J.; Ram,
M. K.; Phillips, T. B.; van Schravendijk, M. R.; Dalgarno, D.; Clackson, T.; Holt, D.
A. J. Med. Chem. 2000, 43, 1135.
RELIC software (http://relic.bio.anl.gov/index.aspx) was used
for the screening with the T7 phage pool constructed from syn-
thetic DNA according to the supplier’s manuals.9
18. Yoon, K. J.; Hyatt, J. L.; Morton, C. L.; Lee, R. E.; Potter, P. M.; Danks, M. K. Mol.
Cancer Ther. 2004, 3, 903.
19. Harel, M.; Hyatt, J. L.; Brumshtein, B.; Morton, C. L.; Yoon, K. J.; Wadkins,
R. M.; Silman, I.; Sussman, J. L.; Potter, P. M. Mol. Pharmacol. 2005, 67,
1874.
20. Bencharit, S.; Morton, C. L.; Howard-Williams, E. L.; Danks, M. K.; Potter, P. M.;
Redinbo, M. R. Nat. Struct. Biol. 2002, 9, 337.
Acknowledgments
We thank Yakult Co. Ltd (Tokyo, Japan) and Dai-ichi Pharma-
ceutical Co. Ltd (Tokyo, Japan) for providing irinotecan. This work
was partially supported by a Grant-in-Aid for Scientific Research
(The Ministry of Education, Culture, Sports, Science and Technol-
ogy of Japan, Japan Society for the Promotion of Science)
21. Belser, T.; Stöhr, M.; Pfaltz, A. J. Am. Chem. Soc. 2005, 127, 8720.
22. Zhang, W.; Nowlan, D. T. r.; Thomson, L. M.; Lackowski, W. M.; Simanek, E. E. J.
Am. Chem. Soc. 2001, 123, 8914.
23. Novagen. T7SelectÒ System Manual, 2000, TB178.