M. Hashimoto et al.
Bioorganic & Medicinal Chemistry Letters 37 (2021) 127835
Fig. 1. (A) A schematic illustration of the construction of an NBD-modified phage library and selection of NBD-modified peptide biosensors against Gal-3. (B) A
helical wheel diagram of an NBD-modified -helical peptide library (GAEX LKCLEX LKAG) and a chemical structure of compound 1.
α
1
X
2
3 4
X
high affinity to a target molecule from the library. In particular, Winter’s
group constructed fluorophore-modified single chain Fv (scFv) libraries
and developed a labeled scFv that selectively responds to 2-phenyloxa-
zolone-conjugated BSA.1 Although folded proteins have the advan-
tage of properly defining random residues and modification sites in the
structure, peptide-based biosensors are preferred since small peptides
have high stability and can be organically synthesized, which reduces
the cost and facilitates chemical modification.20 Additionally, peptides
have the potential for intracellular applications.21 Taki’s group has
(Fig. S1). Next, to examine the NBD modification of expressed peptides
on pIII, model peptide phage was mixed with 1.0 µM compound 1 after
1.0 mM TCEP (tris(2-carboxyethyl)phosphine hydrochloride) reduction.
The NBD-labeled phages were subjected to SDS-PAGE and analyzed by
fluorescence gel imaging. A fluorescent band was observed at a molec-
ular weight corresponding to pIII of the model peptide phage (Fig. S2),
suggesting successful conjugation with NBD. In contrast, for the negative
control phage, in which Cys was substituted with Ala (C8A), no fluo-
rescent band was detected, indicating that the Cys residue employed in
the model helical peptide was modified with NBD. The reaction yield of
the NBD modification was almost saturated at 1.0 µM compound 1
(Fig. S3). The NBD modification did not significantly decrease the
infectivity of the phage at 1.0 µM, while a higher concentration of 1
attenuated the phage activity (Fig. S4). From these results, we deter-
8
3 3
constructed fluorophore-modified T7 phage libraries (X -C-X5-7-C-X ,
Cys are modified with a fluorescent dye, X = any amino acids) and
selected a peptide biosensor against glutathione S-transferase.19 How-
ever, the affinity-based selection of unstructured peptide libraries
carries the risk of selecting inappropriate peptides where the fluo-
rophore is not a core part of the interaction.
◦
mined that a one-hour incubation with 1.0 µM of 1 at 42 C was the
optimum condition.
In this research, for the rational development of peptide biosensors
integrating a dye into the binding interface, we devised a strategy to
stabilize the secondary structure that defines the positions of fluorescent
dye and randomized residues. Here, we report the construction of a
As a proof-of-concept, we performed a selection of fluorescent bio-
sensors from an NBD-modified peptide phage library against Gal-3. Gal-
3 is a galactose-specific lectin involved in tumor metastasis and angio-
genesis.27 In addition, secreted Gal-3 is known to induce cellular insulin
phage library displaying fluorophore-modified -helical peptides and
α
2
8
the selection of peptide biosensors from the library (Fig. 1A).
resistance, and the blood Gal-3 levels in subjects with obesity are high.
We chose a designed helical peptide as the biosensor backbone.
Helical structures play a vital role in protein–protein interactions (PPIs),
and approximately 62% of PPIs occur on one face of helical structures.22
For this reason, four residues located on one face of the helical peptide
were randomized. To stabilize the helical structure, a hydrophobic Leu
residue was introduced on one side to obtain an amphiphilic nature, and
hydrophilic Glu and Lys ion pairs were introduced on the other side for
salt bridges.23 Thus, we designed an amphipathic 16-mer helical peptide
Therefore, we selected Gal-3 as a physiologically significant target. In
the selection process, the carbohydrate recognition domain of Gal-3 was
chemically biotinylated and captured on streptavidin-magnetic beads.
The NBD-modified phage library was incubated with the beads for 30
◦
min at 4 C. Then, unbound phages were washed out, and the bound
phages were eluted by acidic glycine buffer (pH 2.2). The eluted phages
were infected with E. coli TG-1 and amplified for subsequent rounds of
selection. Selections were repeated 5 times in total. The recovery yields
in each round were determined by phage titration (Fig. S5). The yields
were dramatically increased in the 5th round. In addition, we conducted
enzyme-linked immunosorbent assays (ELISAs) to evaluate the binding
activity of phage pools to Gal-3. The results indicated that the amounts
of phage binding to Gal-3 dramatically increased in the 5th round
(Fig. S6), which is consistent with the recovery yields in Fig. S5.
We identified 33 kinds of peptides by DNA sequencing of individual
clones in the 5th round phage pool (Table S1). In general, Cys and Gln
residues were found in randomized positions at high frequency (4 times
higher than the theoretical frequency). Most of the peptides (17/33) had
an additional Cys at random residues in addition to the designed Cys. In
addition, peptides (10/33) without any Cys were also identified due to
mutation of the designed Cys to Arg, which resulted from PCR byprod-
ucts during the library construction. We attribute the enrichment of
clones lacking Cys or clones with two Cys residues to the low propaga-
tion efficiency of phages with unpaired Cys residues (odd numbers of
library (GAEX
1
X
2
LKCLEX
3
X
4
LKAG, Xn = randomized positions), as
1
5
reported previously. Furthermore, we have shown that a model pep-
tide in which all random residues are replaced with Ala (GAEAALK-
CLEAALKAG) forms
a
helical structure (86% helicity).15 The
microenvironments around the conjugated fluorochromes need to be
changed between unbound and bound states to read out the analyte
binding as a fluorescent response. Therefore, a Cys residue was placed at
the center of four randomized residues and modified with a dye by
nucleophilic substitution reaction (Fig. 1B). In this study, 4-nitrobenzox-
adiazole (NBD) was selected as a fluorescent dye since NBD is frequently
used for biomolecule imaging due to its high sensitivity to solvent
polarity.2
synthesized compound 1, which has an electrophilic bromoacetyl group
Fig. 1B).
First, to check the NBD modification of the synthetic model peptide,
4–26
For chemical modification of the Cys residue with NBD, we
(
the reaction was traced by HPLC. The model peptide disappeared, and
◦
29
the conjugate was detected after a one-hour incubation at 42
C
Cys residues). We excluded them from the candidates because they did
2