Site-Selective Functionalization of Flagellin by Steric Self-Protection: A Strategy To Facilitate Flagellin as a Self-Adjuvanting Carrier in Conjugate Vaccine

Article abstract of DOI:10.1002/cbic.201700634

Flagellin (FliC) can act as a carrier protein in the preparation of conjugate vaccines to elicit a T-cell-dependent immune response and as an intrinsic adjuvant to activate the toll-like receptor 5 (TLR5) to enhance vaccine potency. To enable the use of FliC as a self-adjuvanting carrier, an effective method for site-selective modification (SSM) of pertinent amino-acid residues in the D2 and D3 domains of FliC is explored without excessive modification of the D0 and D1 domains, which are responsible for activating and binding with TLR5. In highly concentrated Na₂SO₄ solution, FliC monomers form flagellar filaments, in which the D0 and D1 domains are situated inside the tubular structure. Thus, the lysine residues (K219, K224, K324, and K331) in the D2 and D3 domains of flagellin are selectively modified by a diazo-transfer reaction with imidazole-1-sulfonyl azide. The sites with azido modification are confirmed by MALDI-TOF-MS, ESI-TOF-MS, and LC–MS/MS analyses along with label-free quantitation. The azido-modified filament dissolves to give FliC monomers, which can conjugate with alkyne-hinged saccharides by the click reaction. Transmission electron microscopy imaging, dynamic light scattering measurements, and the secreted embryonic alkaline phosphatase reporter assay indicate that the modified FliC monomers retain the ability either to bind with TLR5 or to reassemble into filaments. Overall, this study establishes a feasible method for the SSM of FliC by steric self-protection of the D0 and D1 domains.

Full text of DOI:10.1002/cbic.201700634

Accepted Article
Title: Site-Selective Functionalization of Flagellin by Steric Self-
Protection: A Strategy to Facilitate Flagellin as a Self-Adjuvanting
Carrier in Conjugate Vaccine
Authors: Chi-Jiun Peng, Hsiu-Ling Chen, Cheng-Hsun Chiu, and Jim-
Min Fang
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemBioChem 10.1002/cbic.201700634
Link to VoR: http://dx.doi.org/10.1002/cbic.201700634
A Journal of
www.chembiochem.org

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
Site-Selective Functionalization of Flagellin by Steric Self-Protection: A Strategy to Facilitate Flagellin as a Self-Adjuvanting Carrier in
Conjugate Vaccine
[
a]
[b]
[b,c,d]
[a,e]
Chi-Jiun Peng, Hsiu-Ling Chen, Cheng-Hsun Chiu,
and Jim-Min Fang*
Abstract: Flagellin (FliC) can act as a carrier protein in preparation
of conjugate vaccine to elicit the T-cell dependent immune response,
and as an intrinsic adjuvant to activate the toll-like receptor (TLR5)
for enhancement of the vaccine potency. To enable the use of FliC
as a self-adjuvanting carrier, an effective method for site-selective
modification (SSM) of pertinent amino acid residues in the D2 and
D3 domains of FliC is explored without excessive modification of the
D0 and D1 domains that are responsible for activation and binding
polysaccharides with proper protein carriers are required to
activate the T cell-dependent immune response for inducing the
maturation of plasma and memory B cells. This is an essential
event for long-term immunological memory and production of
high-affinity antibodies to prevent bacterial infection.
[
1]
Among a variety of carrier proteins, those derived from
bacteria, such as diphtheria toxoid (DT), cross-reacting material
197 (CRM197), tetanus toxoid (TT) and meningococcal outer
membrane protein complex (OMPC) are commonly used in
with TLR5. In
2 4
a high concentration of Na SO solution, FliC
[
2]
monomers form flagellar filaments, in which the D0 and D1 domains
are situated inside the tubular structure. Thus, the lysine residues
preparation of conjugate vaccines. However, the conjugate
vaccines incorporating these bacterial proteins are usually
produced for preventive treatment of the diseases other than the
infection by their source pathogens. The pathogen carriers
themselves are immunogenic and bring about carrier-specific
immune responses. If the vaccine containing a pathogen carrier
is not prepared for the purpose to fight against the source
pathogen, the preexisting immunity caused by such extensively
used pathogen carrier may diminish the efficacy in subsequent
(
K219, K224, K324 and K331) in the D2 and D3 domains of flagellin
are selectively modified by a diazo-transfer reaction with imidazole-
-sulfonyl azide. The sites with azido-modification are confirmed by
1
the MALDI−TOF−MS, ESI−TOF−MS and LC−MS/MS analyses
along with label-free quantitation. The azido modified filament
dissolves to give FliC monomers, which can conjugate with alkyne-
hinged saccharides by click reaction. The TEM imaging, DLS
measurement and secreted embryonic alkaline phosphatase (SEAP)
reporter assay indicate that the modified FliC monomers retain the
ability in either binding with TLR5 or reassembling to filaments.
Overall, this study establishes a feasible method for SSM of FliC by
steric self-protection of the D0 and D1 domains.
[
3]
administration of the conjugate vaccine. In a sharp contrast,
the challenge from the anti-carrier immune response may
enhance the effectiveness of the conjugate vaccine when a
specific pathogen-associated protein, such as flagellin (FliC), is
used to treat the disease caused by the same pathogen.
Flagellin is the structural protein of bacterial flagellar filament
that is responsible for the mobility, chemotaxis, adhesion and
invasion of bacteria. The immunization with flagellin or flagella
Introduction
[4]
can induce protective immunity against bacterial infection. FliC
can also trigger the T cell-dependent immune response via its
intrinsic adjuvant property mediated by toll-like receptor 5
Bacteria exhibit specific glycans such as capsular
polysaccharides (CPS) and lipopolysaccharides (LPS) on their
surface, and also secrete mucoid exopolysaccharide (MEP) to
form biofilm. These accessible polysaccharides with mild
immunogenicity are targets for vaccination against specific
pathogens. Like other haptens, conjugation of the
(
TLR5). Therefore, using flagellin as a carrier in conjugate
vaccine may have advantages due to its dual functions as the
antigen from the targeted bacteria and as an adjuvant for TLR5
[4a, 5]
activation.
Accordingly, flagellin has been utilized to
[6]
conjugate with low immunogenic antigens, such as O-antigens,
[
7]
[8]
mucoid exopolysaccharide (MEP) and cocaine analogs, for
producing the conjugate vaccines against bacterial infection and
drug abuse. By co-localization of antigen and adjuvant to
antigen presenting cells (APC), such antigen−adjuvant
conjugate [e.g., (analogous cocaine)−FliC conjugate] can elicit a
[
[
a]
b]
Chi-Jiun Peng, Jim-Min Fang
Department of Chemistry, National Taiwan University
1, Sec. 4, Roosevelt Rd., Taipei 10617, Taiwan
E-mail: jmfang@ntu.edu.tw
[
9]
Hsiu-Ling Chen, Cheng-Hsun Chiu
Molecular Infectious Disease Research Center, Chang Gung
Memorial Hospital
co-delivery effect to convey enhanced immune responses.
The lysine residues of FliC are generally applied in
conjugation chemistry by using the ε-amine group as a good
nucleophile. However, excessive modification of the abundant
lysine residues on the FliC surface may exert steric hindrance to
interfere with the binding with TLR5, so that the modified FliC
5, Fuxing St., Guishan Dist., Taoyuan 33302, Taiwan
[
c]
d]
Cheng-Hsun Chiu
Department of Pediatrics, Chang Gung Children’s Hospital
5, Fuxing St., Guishan Dist., Taoyuan 33302, Taiwan
[
Cheng-Hsun Chiu
Graduate Institute of Biomedical Sciences, College of Medicine,
Chang Gung University
[8b]
will lose its adjuvant effect. Though an appreciable progress in
[
10]
site-selective conjugation,
with defined lysines by kinetic control,
specific proximity-driven shape-selective protein modification,
including an exclusive conjugation
[11]
259 Wenhua 1st Road, Guishan Dist, Taoyuan 33302, Taiwan
and the sequence-
[
12]
[e]
Jim-Min Fang
The Genomics Research Center, Academia Sinica
has been achieved, most site-selective methods are only
applicable to some well-defined materials for limited purpose. To
facilitate the use of FliC as a self-adjuvant carrier, our objective
in this study is to explore a site-selective modification (SSM)
128, Sec. 2, Academia Rd, Taipei 11529, Taiwan
Supporting information for this article is given via a link at the end of
the document.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
method for precise control over the conjugation sites without
disrupting the TLR5 binding.
selective implantation of bio-orthogonal groups into the D2 and
D3 domains that are exposed to the surface of flagellar filament.
The steric hindrance provided by flagellar assembly will impede
the entry of chemical reagents to decorate the amino acid
residues in the D0 and D1 domains. After depolymerization of
the SSM flagellar assembly, the site-selective modified FliC
monomers will retain the ability for TLR5 activation.
The flagellar filament is a supercoiled assembly that is
constituted by polymerization of FliC monomers in a helical
manner. The D0 and D1 domains of FliC are conserved and
form a spoke region inside the assembled tubular structure of
flagellar filament (Scheme 1). The D1 domain of FliC provides
the most binding sites for recognition with TLR5 via hydrogen
In this study, we focus on developing a site-selective
modification method which will benefit the immunogenic flagellin
as a self-adjuvanting carrier. Taking Pseudomonas aeruginosa
FliC (paFliC) as an example, we demonstrate a feasible
approach by using the assembled flagellar filament to implant
bio-orthogonal azido groups selectively onto the lysine residues
[
13]
bonds and salt-bridge interactions.
The D0 domain is
unrelated to TLR5 binding but required to activate the flagellin-
[
13a]
mediated cellular activity.
In contrast, the hypervariable D2
[
13b,
and D3 domains are arranged outside the flagellar filament,
1
4]
[13]
away from the binding interface of the FliC−TLR5 complex.
[
17]
Adamo and coworkers have recently found that the homodimer
interface of CRM197 functions as a noncovalent self-protecting
in the D2 and D3 domains via a diazo-transfer reaction, while
the lysine residues in the D0 and D1 domains for TLR5
activation were effectively self-protected from the chemical
modification by the steric hindrance (Scheme 1). As a new
strategy for preparation of self-adjuvanting carrier, the human
TLR5 (hTLR5) reporter assay is designed to evaluate the
relative adjuvant efficacy of the conjugated paFliC samples.
[
15]
group.
The study by Kolarich and coworkers also illustrates
that the local environment significantly influences which lysine
[
16]
residues of CRM197 will be modified.
Such steric self-
protection is expected to exist as well in the flagellar structure.
Since the D0 and D1 domains are compactly sheltered by each
other inside the flagellar filament, there is a good opportunity for
Scheme 1. Site-selective azido modification using steric self-protection effect. (A) Monomeric FliC. (B) Flagellar filament comprising FliC monomers. (C) Azido
groups are site-selectively introduced to the protruding domains of flagellar filaments. (D) Azido-modified FliC monomer containing azido-modifications in the D2
and D3 domains.
Results
unstructured aggregation. The rod-like flagellar seeds had the
length around 100 nm on average (Figure 2A), and the flagellar
filament could extend to a length of 1.0 μm after a prolonged
incubation (Figure. 2B). The paFliC assembly was robust
enough to maintain its framework during the treatment with
Flagellin Assembling and Stability of the Flagellar Assembly
in Chemical Modification
The procedure for flagellin assembling was optimized by
sequential seeding and elongation processes as described in
the experimental section. The seeding was performed in a high
o
imidazole-1-sulfonyl azide at 4 C for 1 h to furnish the azido-
modification in
a site-selective manner (Figure 2C). In
concentration of Na
2
SO
4
solution (800−850 mM), and the
comparison, paFliC monomers did not form assembly or
aggregate on incubation in 1.25 M phosphate buffered saline
(PBS) without high concentration of Na SO (Figure 2D). The
2 4
assembly of paFliC monomers was monitored by DLS
microscopy (Figure 1). The Z-average size of the flagellar
assembly increased rapidly in the first 15 min incubation, and
then reached a plateau after 25 min along with the increment of
the sample dispersity. At this stage, the dispersity still slowly
increased even the size reached the plateau, implying the on-off
kinetics between paFliC monomers and the flagellar assembly
was ongoing.
conventional non-selective modification (NSM) of paFliC
monomers was carried out using imidazole-1-sulfonyl azide in
[
17]
PBS to give NSM-N
3
-paFliC,
no assembly or aggregate was
observed during this process (Figure 2E). The similar methods
were also applied to produce the azido-modified flagellins of
Salmonella typhimurium in either site-selective or non-selective
fashion.
The TEM imaging confirmed that the self-assembly of paFliC
monomers was in an ordered morphology rather than
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
Locations of Lysine Residues and Quantitative Analysis of
Site-Selective
Modification.
Modification
versus
Nonselective
The successful azido-modification of FliC was confirmed by the
IR spectral analysis, showing the characteristic peak of azido
-1
group at 2102 cm (Figure S1 in Supporting Information). To
investigate the efficiency of azido modification on Lys residues,
the MALDI−MS (Figure S2) and ESI−MS spectra (Figure S3) of
3 3
parental paFliC, NSM-N -paFliC and SSM-N -paFliC were taken.
The MALDI−MS spectrum of parental paFliC showed a signal at
+
m/z 49012, which was attributable to the [M + H] ion. After
Figure 1. Flagellin assembling process monitored by DLS measurements: The
trend of Z-average size (A) and dispersity (B). Black triangle: monomeric form
of paFliC in 50 mM PBS. Red dot: Both the size of flagellar assembly and the
heterogeneity of size distribution increase along the observation time in 30 mM
azido modification, the mass analyses indicated that at least 5
Lys-NH were transformed into Lys-N by either SSM or NSM
2 3
methods. Considering the high molecular weight of intact FliC,
the MALDI−MS analysis might not be an accurate measurement
of the number of azido decorations that only caused relatively
small mass changes. The deconvolution mass spectra of
ESI−MS analysis using standard software provided by the
instrument manufacturer showed the ionic mass of the native
2 4
PBS containing 800 mM Na SO and 150 mM NaCl.
paFliC as 49606 Da, while the mass signals of SSM-N
NSM-N -paFliC appeared at 50016 Da and 50024 Da,
respectively. Because the kon/koff dynamics over the preparation
of SSM-N -paFliC, some side products of NSM-N -paFliC might
3
- and
3
3
3
be generated. Thus, the ESI−MS and MALDI−MS spectra more
likely reflected the Gaussian distribution of azido modifications in
heterogeneity, rather than an accurate mass measurement of a
single species.
3
The subsequent modification of the SSM- and NSM-N -FliC
with propargyl maltoside was carried out by using the Cu(I)-
catalyzed (3+2) cycloaddition, giving the corresponding SSM-
and NSM-Mal-FliC. Presumably the carbohydrate modification
would cause heterogeneity of shape and mass-to-charge ratio in
polyacrylamide electrophoresis, the SSM- and NSM-Mal-FliC in
SDS-PAGE diagram appeared as diffused bands with mass
increment (Figure S4). The maltoside modified FliC also
displayed a typical MS spectral pattern of glycoprotein in
relatively low sensitivity with increasing molecular mass (Figure
S2). Accordingly, the MALDI−MS analysis in this study was
regarded as
a reference that provided evidence for the
maltoside modifications of FliC; no quantitative analysis was
attempted due to low resolution of the mass spectra.
To compare the SSM- and NSM-N -paFliC monomers, the
3
azido-modified Lys residues were searched by MS/MS analysis
and label-free quantitation. The relative abundance of each
identical Lys-N
3
containing peptide by SSM vs. NSM is
expressed as the fold-change, or preferably as the log2 fold-
change for a clearer comparison in Table 1. The negative value
of log2 fold-change indicates that using the SSM method gives
3
relatively less content of Lys-N at a specific site than that from
the NSM method. In comparison with the NSM of paFliC
monomer, the azido-modifications at K114, K427 and K448 from
SSM of filament are greatly reduced due to the steric self-
protection effect. On the other hand, the K219, K224, K324 and
K331 residues in the D2 and D3 domains still expose outside the
filament, and the azido-modifications of filament by the SSM
method is as effective as the NSM of paFliC monomer. To
understand the tendency of Lys residues in the chemical
Figure 2. TEM images of flagellar seeds and filaments: (A) Flagellar seeds
from self-assembly of paFliC monomers after 1 h incubation at 4 C in 1.25 M
PBS containing 850 mM Na SO and 17.5 M NaCl. (B) Elongated flagellar
2 4
o
filaments from the seeds in 167 mM PBS containing 2.3 M NaCl after 48 h
incubation. (C) The filamentous framework is robust in azido modification. (D)
No aggregation or polymerization of paFliC monomers (InvivoGen) in 1.25 M
o
PBS containing 17.5 M NaCl over 1 h period of incubation at 4 C. (E) No
aggregation or polymerization of NSM-N
modification.
3
-paFliC monomers during azido
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
modification, a study on the spatial positions of the Lys residues
in paFliC and flagellar filament was carried out. We prepared a
average size of the NSM group only slowly increased in the first
20 minutes. During the one-hour period of incubation, the size
3
D structure of paFliC (UniProt code: P72151) based on the
and dispersity of NSM-N
constantly increased without reaching a plateau.
The shapes of the SSM- and NSM-N -paFliC assemblies are
also different as shown by TEM imaging (Figure 4). Similar to
the parental paFliC, the SSM- and NSM-N -paFliC monomers do
not aggregate in 50 mM PBS solution (Figure 4C and 4D). By
incubation in Na SO solution (800 mM) for 1 h, the SSM-N
paFliC monomers reassembled in an orderly manner to form
filaments (Figure 4F), but the NSM-N -paFliC monomers only
3
-paFliC assembly gradually and
crystal structure of the FliC derived from Salmonella enterica
(
(
seFliC, PDB code: 3A5X), and found the identity is over 39%
Figure S5) to give a convincing simulation result (Figure S6).
3
The paFliC structure can be divided into four different domains,
including the conserved D0 and D1 domains as well as the
hypervariable D2 and D3 domains (Figure 3A). All the eight
modified Lys residues are tagged on the simulation structure in
different colors (Figure 3A). The K42, K114, K427, and K448
residues are located in the D0 and D1 domains, whereas the
K219, K224, K324 and K331 residues are located in the D2 and
D3 domains. The paFliC monomers assemble to form flagellar
filament, which exists as a hexamer (Figure 3B). The K42, K114,
K427, and K448 reside in a congested interior environment to
render the steric self-protection against chemical modification,
while the exterior K219, K224, K324 and K331 residues can
have access to chemical modification.
3
2
4
3
-
3
randomly aggregated to form mesh tangles (Figure 4E).
Table 1. Relative Abundance of Identical Azido-Modified Peptides
SSM/
log2
azido-modified peptides
NSM
p-value
[
a]
fold
fold
3
31
D2: VQGSDGK FEAAAK
0.74
0.67
0.48
0.47
0.41
0.19
0.09
0.02
–0.43
–0.57
–1.06
–1.09
–1.29
–2.43
–3.40
–5.55
< 0.05
0.064
324
D2: FGAQTGTATAGQVAVK VQGSDGK
224
D3: AIAEK MDGAIPNLSAR
< 0.05
< 0.05
< 0.05
< 0.05
< 0.05
< 0.05
219
D3: NIAIAAGDSAK AIAEK
[
b]
42
D0N : INSAK DDAAGLQISNR
[
b]
448
D1C : IK DTDFAAETAALSK
[
b]
114
D1N : AALQK EVAAQQAELTR
[
b]
427
D1C : FK NTIDNLTNISENATNAR
[
a] The MS1 intensity of each identical peptide was normalized by the
[18]
TOP3 method
(Figure S7–S9) before calculating the fold changes of
SSM/NSM. [b] D0N, D1C, and D1N denote the peptides at the N-terminal
in D0 domain, at the C-terminal in D1 domain, and at the N-terminal in D1
domain, respectively. The p values were determined by comparing the
MS1 intensity of each identical peptide (Table S2) with Student’s t test (n =
5).
Reassembling of Azido-modified Flagellin Monomers.
The reassembling function of SSM- and NSM-N
Figure 3. (A) The simulated three-dimensional structure of paFliC (UniProt:
P72151) using S. enterica FliC (PDB 3A5X) as the template. D1 domain (pink)
is the binding region of TLR5, while D0 domain (gray) is the activation region.
The K42, K114, K427, and K448 residues are located in the D0 and D1
domains, whereas the K219, K224, K324 and K331 residues are located in the
D2 and D3 domains. (B) The hexamer construct of paFliC filament. The K42,
K114, K427, and K448 reside in a congested interior environment to render
the steric self-protection against chemical modification, while the exterior K219,
K224, K324 and K331 residues can have access to chemical modification.
3
-paFliC
monomers was monitored by dynamic light scattering (DLS)
microscopy to investigate whether the azido-modification will
disrupt the protein-to-protein interactions. Similar to parental
paFliC monomers that will not undergo spontaneous
aggregation in 50 mM PBS (Figure 1), the Z-average size of
SSM- and NSM-N
solution over the first hour (Figure 4A). In the presence of high
concentration of Na SO the SSM- and NSM-N -paFliC
monomers started the polymerization process, nonetheless in
distinct modes. The SSM-N -paFliC monomers instantly
3
-paFliC was kept at a low level in PBS
2
4
,
3
hTLR5 Activation.
3
The capability of the azido- and maltoside-modified paFliC in
TLR5 activation was evaluated by a secreted embryonic alkaline
phosphatase (SEAP) reporter assay using the HEK-Blue hTLR5
cell line derived from the HEK293 cells with co-transfection of
hTLR5 and SEAP genes. Once paFliC activates TLR5, the
assembled and the average size reached a plateau within a few
minutes. In this polymerization process, the DLS measurement
showed a concurrent increase of dispersity that was maintained
at a high level for one hour (Figure 4B). In comparison, the
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
downstream event will induce the production of SEAP protein.
The level of SEAP in the culture medium can be determined
from the 620 nm absorbance derived from the enzymatically
hydrolyzed substrate, and thus the capability of paFliC in TLR5
activation can be readily evaluated.
Figure 5 shows the capability of flagellins at different
incubation concentrations for hTLR5 activation. In comparison
3
with the parental FliC, the capability of the SSM-N -paFliC and
The structure of FliC comprises four domains, in which the D0
and D1 domains are responsible for activation and binding with
TLR5. However, the conventional methods for chemical
modification of the lysine residues on the FliC surface are
nonselective to render excessive modifications in the D0 and D1
domains, and thus causing interference in binding with TLR5 to
lose the adjuvant effect. In contrast, we have explored a site-
selective method for modification of the lysine residues in the D2
and D3 domains that are arranged outside the flagellar filament,
while the D0 and D1 domains in the core of filament are under
steric self-protection. The MS/MS analysis reveals that the
azido-modification occurs at eight Lys residues, and the relative
abundance is highly correlated to their presenting domains
[
8]
SSM-Mal-paFliC in hTLR5 activation was not significantly
affected until the concentration decreased to 10 ng/mL (Figure
3
5A). On the other hand, the capability of NSM-N -paFliC in
hTLR5 activation was much lower than the parental FliC at 10
ng/mL concentration (Figure 5B). The hTLR activation by NSM-
Mal-paFliC also decreased at a concentration of 50 ng/mL, and
lost at 10 ng/mL.
(
Table 1). The K114, K427, and K448 residues in the D0 and D1
domains are dispositioned inside the filament core and obscured
by proximate proteins. The azido-modification for these Lys
residues is appreciably inhibited due to the formation of flagellar
filament. In comparison, the degree of azido-modification at the
K219, K224, K324 and K331 residues in the D2 and D3 domains
was less affected. The present 3D model of simulation
structures thus provides a rationale for the relative abundance in
Lys modifications as that deduced from the MS/MS experiments.
Figure 4. The trend of Z-average size (A) and dispersity (B) in reassembling of
SSM- and NSM-N
black blank triangle) and NSM-N
Spontaneous and quick reassembling of SSM-N
solution containing 800 mM Na SO (black filled triangle), compared with the
delayed aggregation of NSM-N -paFliC monomers (red filled circle). TEM
images of NSM-N -paFliC monomers (C), SSM-N -paFliC monomers (D),
NSM-N -paFliC aggregate in mesh tangles (E) and SSM-N -paFliC assembly
3
-paFliC monomers. Monomeric forms of SSM-N
-paFliC (red blank circle) in 50 mM PBS.
-paFliC monomers in PBS
3
-paFliC
Figure 5. hTLR5 activation by native and chemically modified flagellins. The
capability for activation of hTLR5 is retained in the SSM group of azido- and
maltoside-modified paFliC (A), but reduced appreciably in treatment with low
concentrations of NSM paFliC (B). The negative control (NC) is 1X PBS. *p <
(
3
3
2
4
3
##
0.05, **p < 0.01, and p < 0.01. The data were presented as mean ± standard
3
3
deviation (n = 5). The comparison of paired samples was performed by using
Student’s t test.
3
3
in filaments (F).
Based on the simulation structure, K42 is located in the D0
domain and expected to exhibit steric self-protection effect on
the SSM of flagellar filament. However, the MS/MS analysis
showed that the SSM/NSM ratio in azido-modification of K42 is
DISCUSSION
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
higher than other Lys residues in the D1 domain. To explain this
discrepancy, we speculate that the relatively high SSM/NSM
value for K42 is caused by underestimation of the azido-
modified K42 in the NSM of paFliC monomers, rather than the
enhancement in the SSM of filament. It is known that monomeric
flagellin is susceptible to random proteolytic degradation at the
reduced than that of SSM-N
3
-paFliC, even the maltoside
modifier has a larger size than azido modifier. This result
supports that the chemical modifications by the SSM method
mainly occur at the Lys residues in the region other than the
domains responsible for binding or activating hTLR5. In
comparison, the enhanced adverse effect of the NSM maltoside
modification is related to the surface hindrance that interferes
with the interactions between the NSM-Mal-paFliC and hTLR5
(Figure 5B). The experimental results demonstrated that the
[
19]
N- and C-termini of D0 and D1 domains.
As 10-kDa MWCO
membranes were applied to purify the azido-modified paFliC
samples by centrifugal filtration, the degraded fragments
containing the azido-modified K42 will be filtered off, and thus
underestimated in the MS/MS analysis.
3
SSM-N -paFliC and SSM-Mal-paFliC possessed higher activity
to stimulate the TLR5 signaling pathway compared with the
over-modified paFliC produced by the NSM method. If the NSM-
There are many bioorthogonal methods for site-selective
[10]
[20]
protein modifications,
including that targeting tryptophan,
3
N -FliC were used to conjugate with polysaccharide antigen for
[
21]
[22]
tyrosine, and N-terminal amine group. However, FliC lacks
tryptophan residue in the conserved sequence for the site-
selective modification. The tyrosine residues and N-terminal
amino group exist in the D0 and D1 domains, which would be
buried inside the FliC filament to hinder chemical modifications
vaccine preparation, one can reasonably expect that the
modified flagellin would not have adjuvant property because the
polysaccharide modified Lys residues in the D0 and D1 domains
would impose severe steric hindrance against binding with
hTLR5.
(
Figure S10). If the putative modification of tyrosine or N-
The click reaction in this study is readily carried out, and the
triazole-linked maltoside moiety does provide the desired steric
hindrance to disrupt protein−protein interactions, giving clear
and reliable results in the subsequent cellular assays. Despite
the click chemistry is successfully utilized in the preparation of
Mal-triazole-FliC, the triazole ring is known to elicit anti-linker
terminal amine did occur, the modified FliC would lose the
desired adjuvant activity. Thus, targeting the lysine residues is
more pertinent for the site-selective modification of FliC.
3
The reassembling function of SSM- and NSM-N -paFliC
monomers was investigated to know whether the excessive
azido-modification will disrupt the protein-to-protein interactions
[
23]
antibodies,
which may impair the effectiveness of conjugate
as proposed. Apparently, the SSM-N
possess reassembling ability with the efficiency and
polymerization kinetics similar to that of parental paFliC,
3
-paFliC monomers
vaccine. To avoid such undesired immunogenicity against
triazole linker, one may apply the traceless Staudinger ligation
[
24]
method
to connect the azido-FliC with amino-containing
whereas the NSM-N
3
-paFliC monomers having enriched azido-
antigen via a peptide bond to produce the conjugate vaccine.
Though this study only demonstrates a two-step method for
conjugation of carbohydrate molecule (maltose) with carrier
protein (FliC), the one-step crosslinking is possible by using a
dual functional linker, such as the advanced Staudinger ligation
reagent that contains an N-hydroxysuccinic (NHS) ester at the
other terminal to form another amide bond with amino-hinged
modifications in the D0 and D1 domains may cause a problem in
forming a compact core region of flagellar assembly (Figure 4A
and 4B). In addition to different kinetics in the formation of
flagellar assemblies, the shapes of the SSM- and NSM-N
paFliC assemblies are also different as shown by TEM imaging
Figure 4E and 4F). The random aggregation of NSM-N -paFliC
3
-
(
3
[
24a]
monomers is related to the constant increase in size and
dispersity as that observed in the DLS measurements. In
contrast, the orderly reassembling process of SSM-N -paFliC
3
monomers to flagellar filaments will eventually reach a kon/koff
equilibrium and show plateaus for the assembly size and
dispersity. The circular dichroic spectra showed that the azido-
modification did not significantly affect the secondary or tertiary
structure of flagellin (Figure S11). Thus, the excessive azido-
modification of the lysine residues in the D0 and D1 domains
possibly cause steric effect and change of electrostatic
molecule.
Collectively, the herein reported site-selective modification
method would have a better chance for using the chemically
modified flagellin as the self-adjuvanting protein carrier in
preparation of conjugate vaccines. The SSM of flagellar filament
seems to be a general method because most of the bacterial
flagella filaments have similar structure with remarkable
[
25]
conservation of amino acid sequences.
Furthermore, this
SSM method is practical for scale-up preparation at milligram
level across a variety of bacterial strains and species in less
than a few weeks, without genetic engineering, to produce
adjuvanting protein carrier.
interactions to interfere with the assembling of NSM-N
rather than conformational changes of the protein structure.
3
-FliC,
The D0 and D1 domains not only influence the formation of
flagellar filament, they are also responsible for TLR5 recognition
[
13a]
and activation.
As shown in Figure 3A, it is noted that K114
Conclusions
and K427 are located in the TLR5 recognition region, while K448
is close to the D0 domain which is responsible for TLR5
activation. The DLS and TEM experiments suggest that
excessive modifications in the D0 and D1 domains by the
conventional NSM method may also interfere with the capability
In this study, we demonstrate a new strategy for the site-
selective modification to keep the adjuvant activity of FliC by
using the steric self-protection effect from the flagellar filament
formation. The site-selective azido- or maltoside-modified
flagellin exhibits better capability in hTLR5 activation compared
with the over-modified FliC produced by the conventional
method. The feasibility and applicability of this site-selective
modification method would facilitate the use of flagellin as a self-
adjuvanting carrier in development of conjugate vaccines
against various bacterial pathogens. The present strategy of
site-selective modification is potentially useful for other research
purposes, for example, selective modifications at the desired
3
of NSM-N -paFliC in interactions with TLR5. In contrast, the
SSM-N -paFliC monomers derived from the site-selective
3
modification of flagellar filament based on the steric self-
protection effect of the Lys residues in the D0 and D1 domains
can effectively retain the activity for TLR5 stimulation. The
observations of DLS and TEM experiments were coordinated to
the results of hTLR5 activation assay (Figure 5A and 5B). As
3
shown in Figure 5A, the capability of the SSM-N -paFliC in
hTLR5 activation was not significantly affected until the
concentration decreased to 10 ng/mL. It is interesting to note the
capability of SSM-Mal-paFliC in hTLR5 activation is not further
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
exterior residues of a protein cage that can subsequently
undergo dissociation and reassemble to form various cage
complexes.
Orbitrap Mass Spectrometer (ThermoFisher, Waltham, MA, USA) with a
75 μm × 25 cm capillary C18 column (nanoACQUITY peptide BEH C18,
[
26]
Waters, MA, USA). Standard SDS-PAGE (9% gel) with Coomassie Blue
staining was applied to assay the purity and molecular mass of flagellin
samples.
Gram-negative bacteria possess a unique LPS membrane
outside the cell wall. The polysaccharide composition (O-antigen)
of LPS is considered a suitable antigenic component of vaccines
because LPS is located at the most superficial layer and the O-
[27]
Synthesis of Imidazole-1-sulfonyl Azide Hydrochloride.
To a
solution of sodium azide (0.65 g, 10 mmol) in acetonitrile (10 mL) cooled
in an ice-bath, sulfuryl chloride (SO Cl , 0.8 mL, 10 mmol) was added
[
1]
antigen has appropriate immunogenicity. We are currently
engaged in the preparation of the LPS conjugate vaccine using
the site-selective modified flagellin as an antigenic self-
adjuvanting carrier. The efficacy of as-synthesized LPS−SSM-
FliC vaccine will be compared with those vaccines using NSM-
FliC or classic carrier proteins.
2
2
dropwise. The mixture was stirred for 24 h at room temperature to give
chlorosulfonyl azide, which was cooled in an ice-bath and treated with
imidazole (1.3 g, 19 mmol) by addition in 5 portions. The mixture was
stirred for additional 3 h at room temperature, and then EtOAc (20 mL)
was added. The mixture was extracted with H
saturated aqueous NaHCO (20 mL × 2). The organic layer was dried
over MgSO and filtered. The filtrate was stirred in an ice-bath, and a
2
O (20 mL × 2) and
3
4
solution of acetyl chloride/EtOH (1:3.75, 4.75 mL) was added dropwise.
The white precipitate was collected by filtration, and rinsed with EtOAc
Experimental Section
(
20 mL × 3) to give imidazole-1-sulfonyl azide hydrochloride (1.19 g,
−
1 1
Materials. The paFliC was purchased from InvivoGen (San Diego, CA,
USA; cat. tlrl-pafla) or prepared in Dr. C.-H. Chiu’s lab [Chang Gung
Memorial Hospital (CGMH), Taoyuan, Taiwan; strain: PAO1]. The HEK-
Blue hTLR5 cells (cat. hkb-htlr5) and detection culture medium (cat. hb-
det2) were also purchased from InvivoGen. Bicinchoninic acid (BCA)
assay kit for protein quantification was obtained from ThermoFisher
3 5 5 2
56.6% yield). C H N O S, IR νmax (KBr) 2173 cm ; H NMR (400 MHz,
D
13
2
O) δ 9.40 (1 H, br s), 8.02 (1 H, br s), 7.61 (1 H, br s); C NMR (400
MHz, D O) δ 137.8, 122.7, 120.4.
2
[
28]
Synthesis of Propargyl Maltoside. A suspension of H SO -silica gel
2
4
(20 mg) in propargyl alcohol (0.5 mL) was added to a solution of maltose
monohydrate (360 mg, 1.0 mmol) in propargyl alcohol (0.5 mL). The
(Waltham, MA, USA; cat. 23227).
o
mixture was stirred for 6 h at 65 C, and then subjected to silica gel
All chemical reagents are commercial products and used as supplied
without further purification. The thin-layer chromatography (TLC) was
performed by using Silica Gel 60 F254 aluminum plate (Merck Millipore,
Billerica, MA, USA; cat. 105554). The silica gel (Geduran Si 60, cat.
column chromatography by elution with CH
MeOH/CH Cl (1:5, 100 mL). The fractions of desired product were
collected, and concentrated under reduced pressure to give propargyl
24 11
maltoside as an anomeric mixture (111.3 mg, 29.3% yield). C15H O ;
2
Cl
2
(50 mL) and
2
2
1
111567) used for the column chromatography was also purchased from
yellow viscous oil; TLC (MeOH/CH
MHz, D O) δ 5.10 (1 H, d, J = 3.8 Hz), 4.64 (1 H, d, J = 8.0 Hz), 4.48 (1 H,
dd, J = 4.8, 2.4 Hz), 4.34 (2 H, dd, J = 4.8, 2.6 Hz), 3.95–3.26 (11 H, m),
2 2 f
Cl (1:5)): R = 0.20; H NMR (400
Merck Millipore. The carbon-coated copper grid used for TEM was
purchased from Electron Microscopy Sciences (Montgomery, PA, USA;
cat. CF300-Cu).
2
13
2.94–2.88 (1 H, m); C NMR (400 MHz, D
7
5
2
O) δ 100.7, 97.3, 79.2, 79.0,
6.5, 76.2, 75.9, 73.13, 73.06, 72.3, 71.1, 69.7, 69.6, 60.9, 60.5, 56.7,
5.0.
Instrumentation. Fourier transform infrared (FT-IR) spectra were
recorded on
a Thermo Scientific Nicolet iS 5 FT-IR spectrometer
(
ThermoFisher, Waltham, MA, USA). Flagellin samples (lyophilized cake)
Assembling of Flagellin to Flagellar Filaments. Seeding: An aqueous
solution (8 μL) containing 850 mM Na SO and 150 mM NaCl was added
to paFliC monomers (InvivoGen, 100 μg; containing 5% saccharose and
1X PBS once reconstituted as adding 1.0 mL H O). The solution was
1
13
were pressed into KBr discs for recording FT-IR spectra. H and
C
2
4
NMR spectra were recorded on an Agilent Mercury 400 MHz NMR
spectrometer (Varian, Palo Alto, CA, USA). Chemical shifts (δ) were
recorded in parts per million (ppm). The splitting patterns are reported as
s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), br (broad) and
dd (doublet of doublets). Coupling constants (J) are given in hertz (Hz).
2
o
allowed to stand at 4 C for 1 h to grow flagellar proto-filament.
Elongation: To the above-prepared seeding solution was added sterilized
o
H
2
O (52 μL). The solution was gently shaken at 4 C for 48 h. For
Circular dichroic (CD) spectra were recorded on
a
J-815 CD
comparing the self-assembling process, 8 μL of sterilized H
2
O instead of
spectrometer (Jasco, Tokyo, Japan). Flagellin samples in 50 mM PBS
buffer containing 150 mM NaCl were used for CD measurements by
scanning 190 to 260 nm with a 10 nm/min speed in continuous mode.
Dynamic light scattering (DLS) analysis was carried out using Zetasizer
Nano ZS system (Malvern, Worcestershire, UK). Transmission electronic
the Na SO /NaCl solution was added to another 100 μg batch of paFliC
2
4
(InvivoGen) as a control group. The size and morphology of both
samples were characterized by TEM (Hitachi H-7100) at the end-point of
incubation.
2 4
Alternatively, an aqueous solution (16 μL) of 2 M Na SO and 150
microscopy (TEM) images were obtained on
a
Hitachi H-7100
mM NaCl was added to phosphate buffered saline (PBS, 24 μL of 50 mM
solution, pH = 7.0) containing 150 mM NaCl, and the paFliC (1.3 mg/mL)
obtained from CGMH. The solution was allowed to stand at room
temperature for 1 h to grow flagellar proto-filament. The assembling
process was monitored by using dynamic light scattering (DLS)
microscopy.
Transmission Electron Microscope (Tokyo, Japan). The High Speed
Mixer CM-1000 (EYELA, Tokyo, Japan) was used for the preparation of
flagella. The measurement of UV absorbance of BCA assay was
performed with Synergy H1 Hybrid Multi-Mode Reader (BioTek, Winooski,
VT, USA). The protein mass measurement was carried out using an
Autoflex Speed MALDI−TOF/TOF mass spectrometer (Bruker Daltonics,
Billerica, MA, USA). All full-range ESI−MS data were acquired with a
Premier ESI-quadrupole-time of flight mass spectrometer (Waters,
Preparation of Azido Modified Flagellar Filament and Flagellin
Monomer. To PBS (60 μL of 50−167 mM solution) containing 850 mM
Na SO and flagellar filaments (8−100 μg), the aqueous solutions of
Manchester, UK) with
a
nano-ESI source in positive mode.
2
4
Deconvolution of mass spectra was performed by using standard
software provided by the instrument manufacturer (MassLynx). LC–
MS/MS analysis was performed on an Orbitrap Elite Hybrid Ion Trap-
2 3
K CO (125 mM, 8 μL) and imidazole-1-sulfonyl azide hydrochloride (60
mM, 5 μL) were successively added. The mixture was gently shaken at
room temperature for 1 h to furnish the azido-modified flagellar filaments,
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
which were concentrated by spinning filtration against 50 mM PBS (pH =
assembling background; then Na
2
SO
4
(2 M solution) was added to a final
7
) through Amicon 100 kDa centrifugal filter (14,000× g, 15 min) for two
concentration of 800 mM to start the reassembling process. The progress
of polymerization was monitored in real time by DLS measurements for
another one hour.
times to purify the site-selective modified flagellar filaments (SSM-N
filament).
3
-
The filtrate was subsequently concentrated by using an Amicon 10
kDa centrifugal filter (7,000× g, 15 min) against 50 mM PBS (pH = 7) for
TEM Imaging. The paFliC sample (1−3 μL solution of 0.5−12.5 mg/mL)
two times to collected non-selective modified azido-paFliC (NSM-N
paFliC).
3
-
was placed on
absorbing the solution from the opposite side using a filter paper. The
residual PBS salt was washed off by using 20 μL of sterilized H O. An
a carbon-coated copper grid and dried by gently
3
The above-prepared SSM-N -filaments in PBS solution were
2
subjected to depolymerization by adding 1 M HCl until pH = 2 and gently
aqueous solution (10 μL) of 2% potassium phosphotungstate (PTA) was
sequentially added onto the copper grid and allowed to stand for 30
seconds to perform the negative staining. After absorbing the PTA stain
by a filter paper, the TEM imaging was taken on a Hitachi H-7100
transmission electron microscope.
[
29]
shaking at room temperature for 30 min
or simply by heating and
o
[30]
gently shaking at 65 C in PBS solution (pH = 7) for 30 min.
solution of SSM-N -paFliC monomers were concentrated by spinning
filtration, and stored at 4 C until use.
Alternatively, NSM-N -paFliC was prepared by adding an aqueous
The
3
o
3
solution of imidazole-1-sulfonyl azide hydrochloride (60 mM, 5 μL) to a
PBS solution (50−167 mM, 60 μL) containing monomeric paFliC (8−100
μg) and K CO (125 mM, 8 μL). The mixture was gently shaken at room
2 3
temperature for 1 h to give the azido-modified paFliC. The azido-modified
paFliC were concentrated by spinning filtration (Amicon 10 kDa
centrifugal filter, 7,000× g, 15 min) against 50 mM PBS (pH = 7) for two
3 3
Protein Digestion. The sample (SSM-N -paFliC, NSM-N -paFliC or
native paFliC) was treated with standard trypsin digestion procedures.
Briefly, the paFliC protein (10 μg) was dissolved in 25 mM ammonium
bicarbonate aqueous solution (30 μL) containing 6.5 M urea (pH = 8.0).
o
The sample was incubated with 100 mM dithiothreitol (DTT) at 95 C for
10 min to cleave the disulfide bonds, followed by alkylation with
times to gather the non-selective modified azido-paFliC (NSM-N
3
-paFliC).
iodoacetamide (35 mM final concentration, 2.4 μL) in the dark at room
o
3
The solution of NSM-N -paFliC monomers was stored at 4 C until use.
temperature for 30 min. Digestion was carried out by incubation with 0.1
The molecular mass of the azido modified flagellins was measured
by MALDI–TOF–MS and ESI−TOF−MS, and the site-selectivity was
estimated by label-free LC–MS/MS.
o
μg trypsin at 37 C for 16 h. After adding 10% trifluoroacetic acid (TFA) to
stop the digestion (pH
< 4), the peptides were lyophilized and
resuspended with 1% TFA for desalting (ZipTip C18, Merck Millipore, MA,
USA) before LC–MS/MS analysis.
Preparation of Maltoside-Modified paFliC by Click Reaction. The
stock solutions of CuSO
4 2
·5H O (4.1 mg, 16 mM), ascorbic acid (9.6 mg,
BCA Assay. The content of protein was measured by BCA assay.
According to the manufacturer’s instruction, the working reagent was
prepared by mixing BCA Reagent A and Reagent B (50:1 (v/v)) to yield a
27.5 mM) and propargyl maltoside (6.4 mg, 5.7 mM) were prepared,
respectively, by dissolving in 1.0 mL, 2.0 mL and 3.0 mL of degassed
PBS (50 mM, pH = 7) containing 150 mM NaCl. A mixture containing
clear green solution. The test samples in 25 μL of sterilized H
individually added into microplate wells (flat bottom 96-well microplate),
followed by addition of 200 μL working reagent. Sterilized H O was used
2
O were
CuSO
μL) stock solutions were added to a solution of NSM-N
-paFliC (containing 7−40 μg protein in 150 μL degassed 50 mM PBS).
4 2
·5H O (10 μL), ascorbic acid (40 μL) and propargyl maltoside (30
3
-paFliC or SSM-
2
N
3
o
as a negative control. The sample solution was incubated at 37 C for 30
min, cooled to room temperature, and the UV absorbance at 592 nm was
measured with a plate reader.
The mixture was gently shaken at room temperature for 72 h to complete
the click reaction, giving the corresponding maltoside-modified paFliC
(NSM-Mal-paFliC or SSM-Mal-paFliC). The product was concentrated by
spinning filtration through Amicon 10 kDa centrifugal filter (7,000× g, 15
min) against 50 mM PBS for three times. The content of protein was
measured by BCA assay, and the molecular mass of the maltoside
modified paFliC was measured by MALDI–TOF–MS.
MALDI–TOF–MS Measurement. The mixture containing 1 μL of sample
(SSM-paFliC, NSM-paFliC or native paFliC, 10 pmol/μL) and 1 μL of α-
cyano-4-hydroxycinnamic acid (CHCA) matrix (53.1 nmol/μL) was
prepared and pipetted onto a ground steel plate by the dried-droplet
method. The measurements were performed using an Autoflex Speed
MALDI–TOF/TOF mass spectrometer (Bruker Daltonics, Billerica, MA,
USA) in a linear-positive mode. The data analysis was carried out by
using flexAnalysis software (Bruker Daltonics, Billerica, MA, USA) with
centroid peak detection.
Reassembling of Azido Modified Flagellin Monomers to Azido-
Flagella. By the above-described procedure, the SSM-N -filament
3
prepared from the flagellin provided by CGMH was depolymerized by
treating with 1 M HCl (pH = 2) at room temperature for 30 min to give
3 3
SSM-N -paFliC monomers. The NMS-N -paFliC was similarly treated
with 1 M HCl for a side-by-side comparison of the reassembling efficacy.
The acidic solution was exchanged by spinning filtration through Amicon
LC–MS/MS Analysis. The LC–MS/MS analysis was performed on an
Orbitrap Elite mass spectrometer (ThermoFisher, Waltham, MA, USA).
The peptides were separated on a 75 μm × 25 cm capillary C18 column
1
0 kDa centrifugal filter (7,000× g, 15 min) against 50 mM PBS for three
times. By procedure similar to that for parental paFliC, the
reassembling of SSM-N -paFliC (or NSM-N -paFliC) monomers was
performed in 50 mM PBS solution containing 800 mM Na SO at room
a
(nanoACQUITY peptide BEH C18, Waters, MA, USA) using a gradient
3
3
elution program of 0.1% formic acid in H O and 0.1% formic acid in
2
2
4
acetonitrile, at a flow rate of 0.3 μL/min over 60 min. ESI cone voltage
was set at 2.8 kV with spray current as 0.13 μA. MS1 spectra were
acquired with a resolution of 60,000 at 400 m/z, covering the mass range
temperature. The reassembling process was monitored by DLS
microscopy. The size and morphology of assemblies were characterized
by TEM at the end-point of incubation.
350–2000 Da and accumulating data for 1.0 sec per cycle. The 10 most
intense ions were sequentially isolated for collision-induced dissociation–
MS/MS fragmentation and fragmented by normalized collision energy
setting at 35. Ions having unit charge or undefined charge was also
excluded. All samples were analyzed in pentaplicate.
All spectra were searched against the SwissProt database by using the
Mascot search engine (version 2.5, Matrix Science, London, UK). Trypsin
was specified as the digestion enzyme and up to 3 missing cleavages
Dynamic Light Scattering (DSL) Measurements. The flagellar
reassembling was performed in a low-volume quartz cuvette (ZEN2112,
Malvern, UK). The Z-average size and the polydispersity index (PDI)
were measured by the Zetasizer Nano ZS system (Malvern, UK). Briefly,
the monomeric native or azide-modified paFliC (0.8−1.3 mg/mL) was
incubated in 50 mM PBS (pH = 7) for 1 h to determine the self-
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
were allowed. The following modifications were set as variable
modifications: carbamidomethyl (C), carbamyl (K), carbamyl (N-term),
Acknowledgements
deamidated (NQ), oxidation (M), and ε-azido-modified lysine (N
with 26 Da mass increment. The mass tolerance of parent ion was set as
20 ppm, and ±0.6 Da was set for the fragment ion. The fragment ions of
3
-Lys)
This work was supported by research grants (MOST 103-2113-
M-002-017-MY3) from the Ministry of Science and Technology
(Taiwan) and Chang Gung Medical Research Project
±
target peptides containing ε-azido lysine residues were manually verified.
(
CMRPG3F114). We thank Ms. Ya-Yun Yang and Ms. Ching-
Label-Free Quantitation. The label-free quantitation was processed with
Skyline (version 3.6.0.10162, MacCoss Lab, University of Washington,
USA) for estimation of relative abundance of distinct azido-modified Lys
residues between SSM and NSM groups. The Skyline spectral libraries
were created from Mascot search results. The raw files acquired from the
Orbitrap Elite mass spectrometer were processed with the workflow of
data-independent acquisition (DIA) with data-dependent acquisition (DDA)
spectral library to produce the MS1 chromatograms. MS1 peak
integration of targeted peptides containing azido-modified Lys residues
was visually inspected with the matched MS2 libraries and manually
adjusted according to the ID annotations. Three isotopic precursors (M+0,
M+1 and M+2) were extracted and summed for quantification. The
abundance of each targeted peptide was normalized by the relative
protein concentration, which was inferred by the label-free TOP3
Yen Lin of Instrumentation Center, National Taiwan University,
for the assistance in TEM experiments. We also thank the
valuable technical supports of Ms. Meng-Ting Wu and Mr. Kuan-
Chieh Peng (Department of Chemistry, National Taiwan
University) in MS experiments. We thank Dr. Mei-Chun Tseng in
the Institute of Chemistry, Academia Sinica, Taiwan, for ESI-MS
experiment and consultation with the analysis of intact protein.
Conflict of Interest
The authors declare no competing financial interest.
[
18]
method.
Specifically, the integration of three most intense unique
Keywords: Flagellin • toll-like receptor • site-selective
modification • conjugate vaccine • adjuvant • antigen
peptides, without any modification, were averaged and assumed to be
proportional to the relative protein concentration of each sample. The
data were presented as the relative abundance of identical azido-
modified peptides. The p values were determined by comparing the MS1
intensity of each identical peptide with Student’s t test (n = 5).
[
1] a) R. Adamo, Acc. Chem. Res. 2017, 50, 1270-1279; b) J. B. Robbins, R.
Schneerson, J. Infect. Dis. 1990, 161, 821-832.
[2] M. E. Pichichero, Hum. Vaccines Immunother. 2013, 9, 2505-2523.
[
3] a) M. J. McCluskie, D. M. Evans, N. Zhang, M. Benoit, S. P. McElhiney, M.
Unnithan, S. C. DeMarco, B. Clay, C. Huber, A. Deora, J. M. Thorn, D. R.
Stead, J. R. Merson, H. L. Davis, Immunopharmcol. immunotoxicol. 2016,
38, 184-196; b) M. P. Schutze, C. Leclerc, M. Jolivet, F. Audibert, L.
Chedid, J. Immunol. 1985, 135, 2319-2322.
Molecular Modeling of paFliC and Assignment of Azido-modification
Sites. The homologous sequence searches and alignment of flagellin
(
(
UniProt code P72151) were performed with Chimera software
https://www.cgl.ucsf.edu/chimera/), version 1.11.2, using a BLAST web
[
4] a) V. L. Campodonico, N. J. Llosa, M. Grout, G. Doring, T. Maira-Litran, G.
B. Pier, Infect. Immun. 2010, 78, 746-755; b) W. J. Landsperger, K. D.
Kelly-Wintenberg, T. C. Montie, L. S. Knight, M. B. Hansen, C. C.
Huntenburg, M. J. Schneidkraut, Infect. Immun. 1994, 62, 4825-4830.
service provided by NCBI and the RCSB PDB as a searching database.
The structural model of native paFliC was generated using Modeller
version 9.17 (San Francisco, CA, USA). The full-length flagellin (PDB
code 3A5X, chain A) derived from Salmonella enterica was used as the
homology template (39.5% ID). The protofilament structure of flagella
was simulated by two sequential steps. First, the axially stacked dimer
structure of paFliC was generated by using ClusPro web service version
[5] S. B. Mizel, J. T. Bates, J. Immunol. 2010, 185, 5677-5682.
[
6] R. Simon, J. Y. Wang, M. A. Boyd, M. E. Tulapurkar, G. Ramachandran,
S. M. Tennant, M. Pasetti, J. E. Galen, M. M. Levine, Plos One 2013, 8,
e64680.
[7] V. L. Campodonico, N. J. Llosa, L. V. Bentancor, T. Maira-Litran, G. B.
Pier, Infect. Immun. 2011, 79, 3455-3464.
[
[
8] a) A. Kimishima, C. J. Wenthur, L. M. Eubanks, S. Sato, K. D. Janda, Mol.
Pharmaceutics 2016, 13, 3884-3890; b) J. W. Lockner, L. M. Eubanks, J.
L. Choi, J. M. Lively, J. E. Schlosburg, K. C. Collins, D. Globisch, R. J.
Rosenfeld-Gunn, I. A. Wilson, K. D. Janda, Mol. Pharmaceutics 2015, 12,
653-662.
2
.0 (Boston University and Stony Brook University, USA), based on the
[
14b]
interactive amino acid residues as previously reported.
The simulation
of tubular structure of flagella was performed with SymmDock program
http://bioinfo3d.cs.tau.ac.il/SymmDock/) using stacked dimer as a block
with D0 domain as the binding site, and the symmetry order was set as
1. The circle construct of two proto-filaments obtained from SymmDock
server was manually adjusted to produce the helical structure, based on
(
9] a) F. Qian, A. Guo, M. Li, W. Liu, Z. Pan, L. Jiang, X. Wu, H. Xu, Vaccine
2015, 33, 2038-2044; b) H. Liu, D. J. Irvine, Bioconjugate Chem. 2015, 26,
1
791-801; c) M. Kreutz, B. Giquel, Q. Hu, R. Abuknesha, S. Uematsu, S.
Akira, F. O. Nestle, S. S. Diebold, Plos One 2012, 7, e40208; d) J. M.
Blander, R. Medzhitov, Nature 2006, 440, 808-812; e) T. Maurer, A. Heit,
H. Hochrein, F. Ampenberger, M. O'Keeffe, S. Bauer, G. B. Lipford, R. M.
Vabulas, H. Wagner, Eur. J. Immunol. 2002, 32, 2356-2364.
[
14a]
the atomic model of the flagellar filament as previously reported.
Evaluation of hTLR5 Activity. The capability of native paFliC, azido-,
and maltoside-modified paFliC to stimulate TLR5 was evaluated by a
SEAP reporter cellular assay using an HEK-Blue hTLR5 cell line
[
10] O. Boutureira, G. J. Bernardes, Chem. Rev. 2015, 115, 2174-2195.
[11] X. Chen, K. Muthoosamy, A. Pfisterer, B. Neumann, T. Weil, Bioconjugate
Chem. 2012, 23, 500-508.
[
12] a) B. V. Popp, Z. T. Ball, Chem. Sci. 2011, 2, 690-695; b) Z. T. Ball, Acc.
Chem. Res. 2013, 46, 560-570.
13] a) W. S. Song, Y. J. Jeon, B. Namgung, M. Hong, S. I. Yoon, Sci. Rep.
2017, 7, 40878; b) W. S. Song, S. I. Yoon, Biochem. Biophys. Res.
Commun. 2014, 444, 109-115; c) S. I. Yoon, O. Kurnasov, V. Natarajan,
M. Hong, A. V. Gudkov, A. L. Osterman, I. A. Wilson, Science 2012, 335,
(
InvivoGen). According to the protocol, the suspension of HEK-Blue
4
hTLR5 cells was prepared at a concentration of 1.4 × 10 cells/mL in the
HEK-Blue Detection medium. Then the cell suspension was added to a
[
96-well plate (180 μL/well, ~25,000 cells per well). The paFliC purchased
from InvivoGen was used as a positive control, and 1X PBS buffer was
used as a negative control. In the presence of different concentrations of
the indicated paFliC samples (100, 50, and 10 ng/mL), the cells were
859-864; d) K. D. Smith, E. Andersen-Nissen, F. Hayashi, K. Strobe, M. A.
Bergman, S. L. R. Barrett, B. T. Cookson, A. Aderem, Nat. Immunol. 2003,
4, 1247-1253.
o
[
[
14] a) K. Yonekura, S. Maki-Yonekura, K. Namba, Nature 2003, 424, 643-
incubated at 37 C for 24 h; then the TLR5 activation was evaluated by
650; b) F. A. Samatey, K. Imada, S. Nagashima, F. Vonderviszt, T.
the absorbance at 620 nm due to the SEAP-catalyzed hydrolysis of
substrate. The data were presented as mean ± standard deviation (n = 5).
The comparison of paired samples was performed by using Student’s t
test.
Kumasaka, M. Yamamoto, K. Namba, Nature 2001, 410, 331-337.
15] S. Crotti, H. Zhai, J. Zhou, M. Allan, D. Proietti, W. Pansegrau, Q. Y. Hu,
F. Berti, R. Adamo, Chembiochem 2014, 15, 836-843.
[16] U. Moginger, A. Resemann, C. E. Martin, S. Parameswarappa, S.
Govindan, E. C. Wamhoff, F. Broecker, D. Suckau, C. L. Pereira, C. Anish,
P. H. Seeberger, D. Kolarich, Sci. Rep. 2016, 6, 20488.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
[
17] S. F. van Dongen, R. L. Teeuwen, M. Nallani, S. S. van Berkel, J. J.
[24] a) C. Grandjean, A. Boutonnier, C. Guerreiro, J. M. Fournier, L. A. Mulard,
J. Org. Chem. 2005, 70, 7123-7132; b) B. L. Nilsson, L. L. Kiessling, R. T.
Raines, Org. Lett. 2001, 3, 9-12.
Cornelissen, R. J. Nolte, J. C. van Hest, Bioconjugate Chem. 2009, 20,
20-23.
[
[
[
[
[
18]J. C. Silva, M. V. Gorenstein, G. Z. Li, J. P. Vissers, S. J. Geromanos, Mol.
Cell. Proteomics 2006, 5, 144-156.
19] a) Y. Lu, J. R. Swartz, Sci. Rep. 2016, 6, 18379; b) Y. Lu, J. P. Welsh, W.
Chan, J. R. Swartz, Biotechnol. Bioeng. 2013, 110, 2073-2085.
20] Y. Seki, T. Ishiyama, D. Sasaki, J. Abe, Y. Sohma, K. Oisaki, M. Kanai, J.
Am. Chem. Soc. 2016, 138, 10798-10801.
[25] S. A. Beatson, T. Minamino, M. J. Pallen, Trends Microbiol. 2006, 14,
151-155.
[26] N. P. King, J. B. Bale, W. Sheffler, D. E. McNamara, S. Gonen, T. Gonen,
T. O. Yeates, D. Baker, Nature, 2014, 510, 103-108.
[27] E. D. Goddard-Borger, R. V. Stick, Org. Lett. 2007, 9, 3797-3800.
[28] B. Roy, B. Mukhopadhyay, Tetrahedron Lett. 2007, 48, 3783-3787.
[29] G. F. Ibrahim, G. H. Fleet, M. J. Lyons, R. A. Walker, J. Clin. Microbiol.
1985, 22, 1040-1044.
21]J. M. McFarland, N. S. Joshi, M. B. Francis, J. Am. Chem. Soc. 2008, 130,
7639-7644.
22] A. O. Y. Chan, C. M. Ho, H. C. Chong, Y. C. Leung, J. S. Huang, M. K.
[30] S. Asakura, G. Eguchi, T. Iino, J. Mol. Biol. 1964, 10, 42-56.
Wong, C. M. Che, J. Am. Chem. Soc. 2012, 134, 2589-2598.
[
23]A. Nilo, I. Passalacqua, M. Fabbrini, M. Allan, A. Usera, F. Carboni, B.
Brogioni, A. Pezzicoli, J. Cobb, M. R. Romano, I. Margarit, Q. Y. Hu, F.
Berti, R. Adamo, Bioconjugate Chem. 2015, 26, 1839-1849.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201700634
FULL PAPER
Entry for the Table of Contents
FULL PAPER
Hidden gift: The D0 and D1
domains for TLR5 activation
were effectively protected from
the chemical modification by
the steric hindrance in the
assembled flagellar filament.
Herein, we demonstrated a
feasible site-selective
modification method which will
benefit the immunogenic
flagellin as a self-adjuvanting
carrier.
Chi-Jiun Peng, Hsiu-Ling Chen,
Cheng-Hsun Chiu, and Jim-Min
Fang*
Page No. – Page No.
Title: Site-Selective
Functionalization of Flagellin
by Steric Self-Protection: A
Strategy to Facilitate Flagellin
as a Self-Adjuvanting Carrier in
Conjugate Vaccine
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.