KinITC—One Method Supports both Thermodynamic and Kinetic SARs as Exemplified on FimH Antagonists

Article abstract of DOI:10.1002/chem.201802599

Affinity data, such as dissociation constants (K_D) or inhibitory concentrations (IC₅₀), are widely used in drug discovery. However, these parameters describe an equilibrium state, which is often not established in vivo due to pharmacokinetic effects and they are therefore not necessarily sufficient for evaluating drug efficacy. More accurate indicators for pharmacological activity are the kinetics of binding processes, as they shed light on the rate of formation of protein–ligand complexes and their half-life. Nonetheless, although highly desirable for medicinal chemistry programs, studies on structure–kinetic relationships (SKR) are still rare. With the recently introduced analytical tool kinITC this situation may change, since not only thermodynamic but also kinetic information of the binding process can be deduced from isothermal titration calorimetry (ITC) experiments. Using kinITC, ITC data of 29 mannosides binding to the bacterial adhesin FimH were re-analyzed to make their binding kinetics accessible. To validate these kinetic data, surface plasmon resonance (SPR) experiments were conducted. The kinetic analysis by kinITC revealed that the nanomolar affinities of the FimH antagonists arise from both (i) an optimized interaction between protein and ligand in the bound state (reduced off-rate constant k_off) and (ii) a stabilization of the transition state or a destabilization of the unbound state (increased on-rate constant k_on). Based on congeneric ligand modifications and structural input from co-crystal structures, a strong relationship between the formed hydrogen-bond network and k_off could be concluded, whereas electrostatic interactions and conformational restrictions upon binding were found to have mainly an impact on k_on.

Full text of DOI:10.1002/chem.201802599

A Journal of
Accepted Article
Title: KinITC - One method supports both thermodynamic and kinetic
SARs as exemplified on FimH antagonists
Authors: Beat Ernst, Pascal Zihlmann, Marleen Slbermann, Timothy
Sharpe, Xiaohua Jiang, Tobias Mühlethaler, Roman P. Jakob,
Said Rabbani, Christoph P. Sager, Priska Frei, and Timm
Maier
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: Chem. Eur. J. 10.1002/chem.201802599
Link to VoR: http://dx.doi.org/10.1002/chem.201802599
Supported by

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
KinITC – One method supports both thermodynamic and kinetic
SARs as exemplified on FimH antagonists
Pascal Zihlmann,[a]‡ Marleen Silbermann,[a]‡ Timothy Sharpe,[b] Xiaohua Jiang,[a] Tobias Mühlethaler,[a]
Roman P. Jakob,[c] Said Rabbani,[a] Christoph P. Sager,[a] Priska Frei,[a] Timm Maier,[c] Beat Ernst* [a]
‡ These authors contributed equally to this work
Abstract: Affinity data, such as dissociation constants (KD) or
inhibitory concentrations (IC50), are widely used in drug discovery.
However, these parameters describe an equilibrium state, which is
often not established in vivo due to pharmacokinetic effects and they
are therefore not necessarily sufficient for evaluating drug efficacy.
More accurate indicators for pharmacological activity are the kinetics
of binding processes, as they shed light on the rate of formation of
protein–ligand complexes and their half-life. Nonetheless, although
highly desirable for medicinal chemistry programs, studies on
structure-kinetic relationships (SKR) are still rare. With the recently
introduced analytical tool kinITC this situation may change, since not
only thermodynamic but also kinetic information of the binding
process can be deduced from isothermal titration calorimetry (ITC)
experiments. Using kinITC, ITC data of 29 mannosides binding to
the bacterial adhesin FimH were re-analyzed to make their binding
kinetics accessible. To validate these kinetic data, surface plasmon
resonance (SPR) experiments were conducted. The kinetic analysis
by kinITC revealed that the nanomolar affinities of the FimH
antagonists arise from both (i) an optimized interaction between
protein and ligand in the bound state (reduced off-rate constant koff),
and (ii) a stabilization of the transition state or a destabilization of the
unbound state (increased on-rate constant kon). Based on
congeneric ligand modifications and structural input from co-crystal
structures, a strong relationship between the formed hydrogen bond
network and koff could be concluded, whereas electrostatic
interactions and conformational restrictions upon binding were found
to have mainly an impact on kon.
century ago (“corpora non agunt nisi fixata”),[1] little attention has
been paid to kinetic aspects until recently. Only after Robert A.
Copeland introduced the drug-target residence time concept in
2006, kinetic considerations gained increasing interest.[2] The
key message of this model is that residence time (τ = 1/koff)[3] or
half-life (t1/2 = ln2/koff)[2,4] of a binary drug–target complex, and
not the binding affinity (e.g. expressed by the dissociation
constant KD), controls the in vivo pharmacological activity.
The launch of a new drug is associated not only with high
financial costs but also with a substantial risk of failure, mainly
due to insufficient efficacy and unwanted side effects.[5] Several
studies have shown that kinetic information is more reliable than
affinity data to predict in vivo potency as well as the resistance
potential of a drug candidate.[6] One of many examples is the
HIV reverse transcriptase inhibitor efavirenz that exhibits only a
low affinity for its target (5 µM) but a long dissociation half-life
(t1/2 = 2.8 h).[7] Furthermore, kinetic studies of Maschera et al. on
HIV-1 protease mutants revealed a correlation between drug
resistance to the HIV protease inhibitor saquinavir and an
increased dissociation rate of the drug–target complex.[8]
Nevertheless, early-phase drug discovery still focuses mainly on
the optimization of KDs, although it has been recognized that
kinetic rate constants for association (kon) and dissociation (koff)
are a necessary precondition for a comprehensive description of
the binding process.[9]
Long dissociation half-lives are a crucial feature of many
small molecule drugs on the market, e.g. for the neuraminidase
inhibitor oseltamivir (47 min),[10] the selective COX-2 inhibitor
rofecoxib (9 h)[11] or the HIV-1 protease inhibitor darunavir
(> 240 h).[12] Generally, a drug–target complex with a long
dissociation half-life can compensate for unfavorable pharmaco-
kinetics, i.e. a bimolecular complex can still exist while the
unbound drug molecule is already cleared from the body.[13] This
is of particular importance for substances with short plasma half-
lives.[4] In contrast, in the case of drug toxicity fast off-rates are
favored as it is the case for antipsychotics that occupy D2
receptors (t1/2 < 1 min).[14] In vivo, fast on-rates play a central
role in drug rebinding, which can be influenced for example by
the local accumulation of the target and spatial characteristics of
the binding site.[15] In contrast to off-rates, which are indepen-
dent of free ligand concentration, on-rates can be increased by
the administration of higher doses. However, this requires
sufficient oral bioavailability and increases the risk of undesired
side effects based on the elevated amount of drug in
circulation. [16]
Introduction
Although the theoretical foundation of binding kinetics in
drug-target interactions was set by Paul Ehrlich more than a
[a]
Dr. P. Zihlmann, M.Silbermann, Dr. X.Jiang, T.Mühlethaler,
Dr. S. Rabbani, Dr. C. P. Sager, P. Frei, Prof. Dr. B. Ernst
Institute of Molecular Pharmacy
Pharmazentrum, University of Basel
Klingelbergstrasse 50, 4056 Basel (Switzerland)
E-mail: beat.ernst@unibas.ch
[b]
[c]
Dr. T. Sharpe
Biophysics Facility
Biozentrum University of BaselI
Klingelbergstrasse 70, 4056 Basel (Switzerland)
Dr. R. P. Jakob, Prof. Dr. Timm Maier
Focal Area Structural Biology
It is surprising that studies on the correlation of molecular
structures and their binding kinetics, so-called structure-kinetic
relationships (SKRs), are rare.[17] In a rough approximation,
more rotatable bonds and higher molecular weights correlate
with long complex half-lives.[18] Moreover, water-shielded
Biozentrum University of BaselI
Klingelbergstrasse 70, 4056 Basel (Switzerland)
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
hydrogen bonds also tend to improve the lifetime of protein–
ligand complexes.[19] On the other hand, the on-rate is limited by
diffusion and can be influenced by steric and electrostatic factors
as well as conformational dynamics.[9]
of heat release or uptake that is measured in the calori-
meter. As the concentration of species changes throughout the
titration, the relaxation kinetics changes, and so the shape
(width) of the peak changes, with narrower peaks at the
beginning of the titration and wider peaks nearer the point of
stoichiometric equivalence. The observed shape of the peak
actually depends on both the rate of relaxation to equilibrium,
and the intrinsic rate constant of the power compensation
electronics of the calorimeter, so it is necessary to deconvolute
these contributions.
Surface plasmon resonance (SPR) spectroscopy has
evolved into the method of choice for measuring binding kinetics
in drug discovery.[20] It monitors non-covalent interactions in real-
time by detecting a mass-dependent change in the refractive
index close to the sensor surface.[21] SPR is also applied to
obtain thermodynamic information from affinity data as
a
function of temperature by van’t Hoff analysis. However, since
the heat capacity change (ΔCp) of macromolecular interactions
with small molecules is mostly different from zero, enthalpy and
entropy changes for binding are usually temperature dependent.
This introduces curvatures in van't Hoff plots and can limit the
accuracy of linear approximations compared to direct measure-
ments.[22]
KinITC extracts the underlying kinetic information from the
ITC thermogram by analysing the shape of each injection peak,
and has been described by Dumas.[25] Here we have employed a
simplified version of this analysis, kinITC-ETC (Equilibration
Time Curve), implemented in the commercially available ITC
analysis software AFFINI-meter.[25a] This method calculates the
time necessary for the differential power curve to return to the
baseline after an injection, using the baseline-corrected
thermogram, yielding a bell-shaped ETC curve. This curve is
fitted as a function of koff and the intrinsic rate constant for the
instrument electronics. The enthalpy change for association and
the association constant, derived from fitting of the integrated
binding isotherm (or determined in other experiments), are used
to constrain this fit, so the kinetic and equilibrium parameters
determined by this analysis are necessarily correlated and a
reliable fitted value of the equilibrium constant is essential
(Figure 1).
Therefore, the method of choice to directly determine the
thermodynamics of a molecular binding event is isothermal
titration calorimetry (ITC). In contrast to SPR, where one
interaction partner has to be immobilized on the sensor chip
surface, ITC measures the heat change of a binding interaction
in solution. This heat change can be converted into a binding
isotherm that allows the direct measurement of KD and the
change in enthalpy (ΔH°), whereas the change in free energy
(ΔG° = RTlnKD, with R being the universal gas constant and T
the absolute temperature) and the change in entropy (ΔS°,
ΔG° = ΔH° – TΔS°) are calculated.[23] Both techniques are the
gold-standard in their field of application, SPR for the determi-
nation of kinetic and ITC for thermodynamic parameters.[24]
Recent publications from Dumas and coworkers have the
potential to transform the role of ITC in drug discovery, since it
describes an approach to derive binding kinetics from ITC data
(kinITC).[25]
Obviously, kinITC-ETC greatly increases the value of ITC
data, allowing the direct determination of all relevant biophysical
constants describing a binding event within one experiment.
Although bearing a great potential to expand the positive impact
of kinetic investigations in drug discovery, kinITC has only
hesitantly found its way into SKR studies yet.[26] To reduce the
initial reservation for the application of kinITC, kinetic
parameters derived from ITC experiments are independently
validated for the first time in the present study. In addition, we
demonstrate that kinITC is a powerful tool for “data mining”.
In an ITC experiment, after each injection and mixing of
titrant, an association reaction in the calorimeter cell relaxes to
the equilibrium position dictated by the new total concentrations
of reactants. This relaxation to equilibrium occurs at a rate that is
determined by the association and dissociation rate constants
for the reaction, together with the concentrations of free and
complexed species. This relaxation process gives rise to a peak
Figure 1. Illustrative representation of the kinITC-ETC obtained from the thermogram of a 1:1 interaction. The peak broadening observed at mid-titration (injection
8) as compared to the initial peaks (injection 2) is an indicative that the thermogram contains kinetic information that can be determined with kinITC. The
equilibration time for each peak is plotted against titrant/titrate molar ratio to yield the ETC. Curve fitting of ETC according to the simplified kinITC method yields
the values of the off-rate constant (koff) and the response time.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
Therefore, the binding kinetics of a series of congeneric
mannosides binding to the lectin domain of FimH (FimHLD) were
determined. FimH is a virulence factor located at the tip of type 1
pili of uropathogenic E. coli (UPEC) strains.[27] It interacts with
the highly mannosylated glycoprotein uroplakin Ia, which is part
of the urothelial mucosa and thereby mediates bacterial
adhesion to the bladder epithelium as the initial step of urinary
tract infections (UTI).[28] An anti-adhesion therapy with FimH
antagonists, which block the adhesion and thereby prevent the
infection, could therefore be beneficial for patients suffering from
recurrent UTI.[29] Re-analyzing data from numerous ITC[30] and
SPR[31] studies we published in the last 5 years, enabled the
validation of the kinITC-ETC approach and the correlation of
kinetic parameters with structural properties of a large ligand
dataset. The kinetic fingerprints of FimH antagonists offer the
opportunity to further improve the binding characteristics
essential for a clinical application.
kinetic and thermodynamic parameters from ITC raw data
avoiding the risk of user bias as can arise through manual
evaluation of thermograms. We applied kinITC-ETC to deduce
binding kinetics from ITC data of a large set of structurally
diverse FimH antagonists (compounds 1-29, see Table 1, 2 and
S1). This kinetic information proved to be extremely valuable as
it provided a more detailed insight into the binding process of
carbohydrate-based antagonists to the adhesin FimHLD
.
Validation of kinITC-ETC by SPR. Although Dumas et al.
proved the potential of kinITC with the evaluation of multistep
kinetic RNA folding,[25b] there is, to the best of our knowledge, no
independent comparison of data obtained by kinITC-ETC and
those derived from alternative biophysical methods available to
date. For this purpose, a subset of four mannose-based FimH
antagonists (compounds 1-4) was investigated by SPR. For the
on-rate constants kon, the comparison revealed an excellent
correlation with absolute values differing by less than a factor of
two (Table 1-A), whereas up to 6-fold deviation was observed for
koff values (Table 1-B). These differences might be a conse-
quence of ligand rebinding on the SPR chip surface.[32] However,
when we normalized the koff values obtained by SPR and
kinITC-ETC to that of n-heptyl α-D-mannopyranoside (1) (giving
relative koff, rkoff) differences of less than a factor of 2 arose
(Table 1-B). In their comparison of SPR and kinITC, Burnouf et
al. reported a similar difference for thiamine pyrophosphate
interacting with mRNA.[25b]
Results and Discussion
The publication on kinITC by Burnouf et al. in 2012 gained
wide interest, but the complexity of the analysis hampered a
broader application.[25b] However, with kinITC-ETC integrated
into the commercially available ITC analysis software
AFFINImeter (Version 1.0415 – 1.1510) this hurdle could be
overcome. It uses fully automated data processing to derive
Table 1. Comparison of kinetic data obtained by kinITC-ETC and SPR for the interaction of FimHLD with the mannose-derived antagonists 1-4.[a]
OH
OH
O
OH
OH
O
OH
OH
O
OH
OH
O
HO
HO
HO
HO
HO
HO
HO
HO
O
O
OMe
1
2
3
4
Table 1-A. Association rate constants kon of FimHLD with compound 1-4 determined by ITC and SPR.
kinITC-ETC
SPR
SPR
kon
kon
konkinITC-ETC/kon
Compound
kon [M-1s-1]
kon [M-1s-1]
rkon
rkon
absolute
relative
1
2
3
4
3.32*104
9.71*103
2.31*103
2.00*103
1.00
0.29
0.07
0.06
2.45*104
5.57*103
2.45*103
2.67*103
1.00
0.23
0.10
0.11
1.35
1.74
0.94
0.75
1.00
1.29
0.70
0.55
Table 1-B. Dissociation rate constants koff of FimHLD with compound 1-4 determined by ITC and SPR.
kinITC-ETC
SPR
SPR
koff
koff
rkoff
koffkinITC-ETC/koff
Compound
koff [s-1]
t1/2 [min]
koff [s-1]
t1/2 [min]
rkoff
absolute
relative
1
2
3
4
7.27*10-4
1.15*10-3
2.86*10-3
1.82*10-3
1.00
1.58
3.93
2.50
15.9
10.1
4.1
1.54*10-4
1.89*10-4
1.11*10-3
6.51*10-4
1.00
1.23
7.21
4.23
75.0
61.1
10.4
17.7
4.72
6.07
2.57
2.80
1.00
1.29
0.54
0.59
6.3
[a]
Absolute (kon and koff) and relative (rkon and rkoff) kinetic parameters are reported for both methods (Table 1-A for on-rates and Table 1-B for off-rates). Relative
values are normalized to n-heptyl α-D-mannopyranoside (1), which is set to 1. The discrepancies between kinITC-ETC and SPR for kon (Table 1-A) and koff (Table
1-B) values are displayed in the rightmost columns. KD values (Table S2) as well as confidence intervals for kon and koff (ITC, Table S4 and SPR, Table S5) are part
of the Supplementary Information.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
Kinetics of FimHLD binding. Only recently, when correla-
tions between prolonged drug-target half-lives and clinical
efficacy of drugs were reported, binding kinetics attracted the
interest of the drug discovery community.[14a,33] The initial
assumption that association rate constants are only diffusion
controlled and rather constant within a congeneric set of ligands
for a specific target had to be revised.[17,18,34] Indeed, our test
subset clearly indicates that the variations of the on-rate
constants within the series of congeneric FimH antagonists 1-4
are more pronounced (for kinITC ≈ 1 : 16, Table 1-A) than of the
off-rate constants (for kinITC ≈ 1 : 4, Table 1-B).
To exclude that the length of the aliphatic aglycone affects
the extended hydrogen bond network formed by the mannose
moieties of compounds 1-4, co-crystal structures of compound 3
(PDB-code: 5JCR) and 4 (PDB-code: 5MUC) were solved and
compared with published structures of compound 1 (PDB-code:
4XO8)[35] and 2 (PDB-code: 1UWF)[36] (Table S3). For all four
ligands, both protein structure and binding mode were found to
be identical (Figure 2). Hence, the higher on-rate constants of
antagonist 1 and 2 must be a result of their elongated
aglycones. This finding is unexpected since the favorable
interactions of the elongated aglycones of compound 1 and 2
with Tyr48 and Tyr137 (called tyrosine gate) were presumed to
primarily lower the off-rate resulting in a prolonged complex half-
life.
Figure 2. Binding mode of compounds 1-4 to FimHLD
. The co-crystal
structures of 1 (red, PDB-code: 4XO8), 2 (yellow, PDB-code: 1UWF), 3
(green, PDB-code: 5JCR) and 4 (blue, PDB-code: 5MUC) show coinciding
binding modes. Their mannose moieties form identical hydrogen bond
networks with FimHLD and a structural water molecule (W1), whereas only the
aglycones of 1 and 2 interact with the two tyrosines 48 and 137.
or worsened interaction with the target protein, affect the stability
of the various states. Thus, for example, an additional interaction
predominantly formed in B and to a lesser extent in TS‡ will
increase ΔGTS-B (decrease koff) and to a lesser extent decrease
ΔGTS-U (increase kon). Concomitantly, the change of ΔG° leads to
tighter binding. Furthermore, structural modifications that affect
long-range electrostatics of the ligand and thus its proper
orientation can affect the rate constants by altered diffusion,
significantly impeding an interpretation based only on changes in
the relative thermodynamic stability of states.
The kinetic data of antagonists 1-4 obtained by both ITC and
SPR demonstrate that the improvements of on-rates go hand in
hand with the elongation of the aglycone, whereas off-rates are
decreased, however to a lesser extent, leading to an overall
improved KD value (Table S2). Hence, increased on-rates are
related to favorable interactions realized in TS‡ or a destabili-
zation of U relative to TS‡, whereas decreased off-rates imply an
additional stabilization of B relative to TS‡. Although structural
information regarding TS‡ is not available, it seems plausible to
speculate that a hydrophobic aglycone first establishes a contact
A one-step model with ligand and protein in the unbound
state (U), in the transition state (TS‡), and in the bound state (B)
characterized by the kinetic rate constants kon and koff is the
simplest way to describe
a
protein-ligand interaction
(Figure 3-A). However, more common is a multistep binding
mechanism with apparent rate constants composed of multiple
elementary rate constants.[9] Rate constants depend upon the
free energy difference ΔGTS-U (between TS‡ and U) for kon and
the free energy difference ΔGTS-B (between TS‡ and B) for koff,
respectively (Figure 3-A). The energy difference between B and
U represents the equilibrium free energy of binding (ΔGB-U, or
ΔG°). Changes in the relative stability of these states due to
structural differences of the ligand (or the protein as
a
consequence of mutations) provoke changes in one or both rate
constants. A stabilization of U and B relative to TS‡ (or destabili-
zation of TS‡ relative to U and B) decreases the on-rate and the
off-rate, respectively. Ligand modifications leading to improved
Figure 3. A) The energy diagram illustrates a simple one-step binding process with the transition state (TS‡) separating the unbound state (U) from the bound
state (B). ΔGTS-U is the activation free energy of association, ΔGTS-B the activation free energy of dissociation, ΔG° the free energy of binding and kon and koff are
the kinetic rate constants. B) The correlation of the kinetic parameters kon and C) koff with changes of ΔG° depends on the site of a ligand’s modification. Ligands
with modified aglycones (2-4; 10-23; for structures and binding parameters see Supporting Information, full symbols, solid regression lines) are separated from
ligands with modified mannose moieties (5-9; 24-29; for structure and binding parameters see Supporting Information, hollow symbols, dotted regression lines): A,
aglycone; M, mannose moiety.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
with the surface-exposed tyrosine gate of FimHLD. Only then the
protein–ligand complex relaxes from the TS‡ to the bound state
B by a successful completion of the hydrogen bond network
within the binding site. A similar effect was described for the
human carbonic anhydrase II (hCAII), where the on-rate
increased in parallel with the chain length of interacting alkyl
benzenesulfonamides,[37] i.e. in a pre-binding state, an inter-
action with a hydrophobic patch of the enzymatic cavity is
formed.[38]
FimHLD, affect the koff values (slope = 0.268, R2 = 0.906) to a
greater extent than those of the aglycone (slope = 0.133,
R2 = 0.801). In contrast, modifications of the aglycone, forming
beneficial contacts with the tyrosine gate in the bound state,[30c]
influence kon values (slope = -0.270, R2 = 0.943) to a larger
degree than those of the mannose moiety (slope = -0.135,
R2 = 0.710).
Thus, the mannose moiety establishes stronger interactions in
the bound state B than in the transition state TS‡, resulting in the
observed stronger effect on the off-rate constant koff
(slope = 0.268) than on the on-rate constant kon (slope = -0.135).
For structural modifications of the aglycone, however, the impact
on kon values (slope = -0.270) is more pronounced than on koff
values (slope = 0.133). This suggests that either the TS‡ is
stabilized relative to the bound state B (e.g. when the aglycone
establishes stronger interactions with the receptor in the TS‡
than in the bound state), or that the unbound state U is
destabilized relative to TS‡.
To investigate the kinetic behavior of FimHLD antagonists in
more detail, its binding interaction with 29 antagonists (for
structural details see Supporting information, Table S1) covering
a broad range of affinities (1 nM to 100 µM) were analyzed by
kinITC-ETC. Thermodynamic and kinetic fits (Table S1) and
binding parameters (Table S3 and S4) are listed in the Support-
ing Information.
The results observed for compounds 1-4 seamlessly fit into
the trend observed for all studied FimH antagonists. However,
these correlations are based on a rough ligand classification
(modifications of the glycan or aglycone), neglecting various
factors, e.g. electrostatic interactions or flexibility of the ligand.
Contributions from these factors can only be revealed by a
closer look at individual structural details of the various ligands
and at their interactions with the protein.
Hydrogen bonds, electrostatic interactions, and confor-
mational changes. To rationalize the trend that mannose
modifications have a larger impact on off-rates than on on-rates
(Figure 3-B and 3-C), interactions of ligands 5-9 with FimHLD
were analyzed in more detail (Table 2). The 2-hydroxyl of
n-heptyl α-D-mannopyranoside (1) forms two hydrogen bonds
with FimHLD, once as H-bond donor to a structural water
molecule (W1) and once as H-bond acceptor from the positively
charged N-terminal Phe (Figure 2). Substitution of the 2-hydroxy
group with a fluorine (1 → 5) leads to a loss of hydrogen bonds,
Structure-kinetic relationship (SKR)
Influence of the site of ligand modification. With the set of
29 carbohydrate-based antagonists, we studied how the site of
the structural modification affects the correlation between
binding energies and kinetic binding rates. The ligands were
divided into two subsets; one subset (n = 17) containing all
representatives with varied aglycones (compounds 2-4 with alkyl
chains of different length; 10-21 with biphenyl aglycones, 22 with
a
squaric acid aglycone and 23 with an indolinylphenyl
aglycone) and second subset (n = 11) representing
a
compounds with modified mannose moieties (deoxy- and deoxy-
halogeno derivatives 5-8 and 24-27, a septulose derivative 9
and two C-2 branched mannose derivatives 28 and 29 (see
Supporting Information, Table S1). For the two subsets,
significantly different correlations of ΔG° with kon- (Figure 3-B,
p < 0.001) and koff values (Figure 3-C, p < 0.001) were obtained.
Modifications of the mannose moiety, known to form an
extended hydrogen bond network in the deep binding pocket of
Table 2. Kinetic binding parameters for the interaction of FimHLD with mannose derivatives of n-heptyl α-D-mannopyranoside (1).[a]
OH
F
O
OH
Cl
O
OH
Br
O
HO
HO
HO
HO
HO
HO
O
O
O
5
6
7
OH
OH
OH
O
O
HO
HO
HO
HO
O
O
8
9
koff [s-1]
Compound
1
kon [M-1s-1]
1/rkon
rkoff
t1/2 [min]
3.32*104
1.0
7.27*10-4
1.0
15.9
5 (2-F)
6 (2-Cl)
7 (2-Br)
8 (2-H)
2.04*104
8.90*103
7.90*103
5.35*103
1.6
3.7
4.2
6.2
1.02*10-2
1.19*10-2
1.45*10-2
4.13*10-2
14.0
16.4
19.9
56.8
1.1
1.0
0.8
0.3
9 (7-ring)
5.10*103
6.5
1.35*10-3
1.9
8.6
[a]
Confidence intervals of the fitted parameters kon, koff, KD and the response time of the calorimeter feedback circuit are part of the Supplementary Information (Table S4).
Relative changes (rkon, rkoff) are compared to n-heptyl α-D-mannopyranoside (1).
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
while the electrostatic interactions are largely preserved as
oxygen and fluorine share a comparable polarity and a close
isosteric relationship. Upon removal of the 2-hydroxy group
(1 → 8) both the H-bonds and the electrostatic interactions are
lost. In terms of kinetics, the loss of H-bonds (1 → 5) mainly
affects the off-rate. In contrast, decreasing electronegativity
reduces the electrostatic interaction with the N-terminus from
fluorine to chlorine to bromine (5 → 6 → 7) and affects the kon
value to a larger extent than the koff value. This complex
behavior results formation of the TS‡ which enhances on- and
off-rates and (ii) the formation of more specific interactions in the
bound state which decreases off-rates. However, mannose
modification may also affect the on-rate constant, as observed
for the ring extended compound 9. As shown in the study of
Sager et al.,[30a] septulose 9 (PDB-code: 5CGB) and pyranoside
1 (PDB-code: 4XO8) establish an identical H-bond network with
FimHLD. Upon binding to the lectin, the markedly increased
flexibility of septulose 9 in solution leads to an entropy penalty
due to a loss of conformational freedom upon binding. The
present work displays that the increased flexibility in solution of
ligand 9 stabilizes U relative to TS‡ and to a lesser extent to B
where the flexibility of all ligands is largely restricted by binding
interactions. This results in a considerably lowered on-rate and a
relatively small increase in the off-rate when compared to 1
(Table 2).
destabilization of the unbound state. On a structural level, the
kinetic contribution of the carbohydrate moiety could be
separated from that of the aglycone. The structural variations of
the sugar moiety, which forms an extended hydrogen bond
network in the deep binding pocket of FimHLD, mainly influence
the off-rate constant koff, while modifications on the aglycone
establishing hydrophobic interactions with the two tyrosines at
the entrance to the mannose-binding pocket, predominantly
affect kon. The latter observation could be an indication that the
aglycone initially establishes a contact with the tyrosine gate to
facilitate the mannose moiety to enter the carbohydrate
recognition domain. This finding is in excellent agreement with
previous observations from Gaspari et al. for the interaction of
hCAII with hydrophobic ligands.[38] In general, initial hydrophobic
interactions to stabilize protein-ligand interactions established
before the final binding mode is reached might facilitate ligand
recruitment from solution. Furthermore, electrostatic interactions
may beneficially influence the on-rate as exemplified by the
replacement of the 2-hydroxy group of the mannose moiety by
fluorine, chlorine, bromine or hydrogen. Finally, due to its
flexibility in solution, the unbound state of septulose derivative 9
is stabilized relative to its transition state, resulting in a reduced
on-rate.
These kinetic considerations are essential for the success of
a therapeutic treatment of UTI, because the half-life of the
complex formed by a carbohydrate-based antagonist is a
cardinal parameter. Only an extended half-life of the antagonist–
FimH complex prevents bacteria from interacting with the
urothelial cells of the host for long enough to allow bacterial
elimination by urination. In contrast to our in vitro results,
rebinding of monovalent antagonists to the CRD of neighboring
pili may result in a longer FimH occupancy in vivo.
Our kinetic study with FimHLD is the first systematic analysis
of protein–ligand interactions using kinITC-ETC. Additional
studies will improve the insight into the effects originating from
ligand modifications upon binding kinetics and may lead to
generally applicable rules. For numerous projects over the past
20 years ITC data had been acquired and the hidden kinetic
treasures can now be raised by kinITC-ETC. We therefore
expect kinITC-ETC to become a popular instrument for “data
mining.”
Conclusions
Based on ITC measurements, the new analytical tool kinITC-
ETC allowed the determination of kinetic data for 29 FimH
antagonists. To test the reliability of kinITC-ETC, the kinetic rate
constants for a subset of 4 ligands were compared with data
measured by SPR. The systematically lower dissociation rate
constant (koff) obtained by SPR may originate from
a
fundamental difference between the two approaches. ITC
measures molecular interactions in free solution, whereas SPR
requires the immobilization of one binding partner on a chip
surface. The reduction of the apparent koff in SPR
measurements may be due to the analyte’s rebinding to the
immobilized partner. However, our attempts to mitigate this
effect by using a low immobilization density had only a limited
effect. Another rationale for the observed difference in koff may
arise from errors associated with the kinITC-ETC method, i.e. Experimental Section
the observed kinetics of heat evolution in ITC experiments
includes a contribution from the intrinsic instrumental response
(τITC), which can vary between instruments and for the same
instrument depending on cleanliness. Since we re-analyzed
existing data for this study, τITC was included as a fitting
parameter in the kinITC-ETC analysis and not measured for the
instrument at the time of the experiment. Nevertheless, it should
be noted that the systematic disagreement in off-rates between
SPR and ITC measurements is at most in the order of a factor of
6 and therefore in an acceptable range and that the relative rate
constants obtained with kinITC-ETC and SPR are in good
Protein cloning, expression, purification and biotinylation. For
all ITC experiments FimHLD of E.coli K-12 strain was expressed with a
C-terminal thrombin cleavage site and a 6His-tag (FimHLD-Th-6His,
173 residues) following a previously published protocol.[39] For SPR
experiments FimHLD-AVITag-6His gene construct was cloned by overlap
extension PCR using the FimHLD-Th-6His[39] gene construct and an
AVITag-6His gene sequence as templates and was subsequently ligated
into the plasmid pNT.[40] The histidine-tagged recombinant protein was
expressed in the protease-deficient E. coli strain HM125 and purified with
Ni-NTA affinity chromatography.[39] AVITag is a specific 15-amino acid
peptide sequence (GLNDIFEAQKIEWHE) that can be biotinylated by the
E. coli biotin ligase BirA. To keep the BirA enzyme active, the protein was
dialyzed with Slide-A-Lyzer MINI Dialysis Units (3’500 MWCO, Thermo
Scientific) in 50 mM bicine, pH 8.3. Biotinylation was done according to
the manufacturer's protocol (Avidity). In order to remove excessive biotin
the protein sample was dialyzed in HBS-EP, 7.4, overnight.
agreement. Therefore, kinITC-ETC can be regarded as
a
reliable method to derive kinetic information from ITC
experiments.
The insight into the binding kinetics of the 29 mannose-
based antagonists strongly improved our understanding of their
binding characteristics. Their enhanced KD values not only stand
for prolonged complex half-lives, but also for increased on-rates
indicating
a
stabilization of the transition state and
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
Surface plasmon resonance (SPR). Surface plasmon resonance-
based experiments were performed using a Biacore T200 instrument (GE
Healthcare). Biotinylated FimHLD (100 nM) was immobilized on the
surface of a streptavidin chip (sensor chip SA) using the “aim for
immobilized level wizard“ (1’300 RU, 5 µL/min). A reference surface with
Amino PEG biotin (50 µM, Polypure) was prepared (time and flow rate:
60 s and 10 µL/min, respectively) to correct for unspecific binding events
of the glycomimetics with streptavidin on the sample surface. Kinetic
experiments were run at 25°C using HBS-EP (0.01 M HEPES pH 7.4,
0.15 M NaCl, 3 mM EDTA, 0.005% surfactant P20, from GE Healthcare)
as running buffer at a flow rate of 30 µL/min. Since a convenient
regeneration condition, which keeps the protein active, was not found,
single cycle kinetics (SCKs) without regeneration steps was used.
Instead of surface regeneration with chemical agents after each injection,
a complete dissociation of the respective compound from the protein was
allowed, before a new run was started. Blank injections (HBS-EP running
buffer) were conducted under equal conditions in order to apply double
referencing. Binding data was evaluated using Biacore T200 Evaluation
software version 1 (GE Healthcare).
check if the slopes of the two subgroups (modified mannose moiety vs.
modified aglycone moiety) are identical or different, were calculated with
Prism version 5.0c from GraphPad, Inc. (La Jolla, CA, USA).
Crystallization and structure. FimHLD–3 and FimHLD–4 (compound
3 and 4 in complex with FimHLD.) were crystallized by sitting-drop vapor
diffusion at 4°C. FimHLD (residues 1-158) was used at
a final
concentration of 10 mg/mL (ca. 0.5 mM) with a 5-fold molar ligand
excess in HEPES buffer (pH 7.4, 20 mM). After several months, plate like
crystals appeared in 1.6 M (NH4)2SO4, 0.1 M Hepes pH 7.5, 1% PEG
3350 (w/v) and were flash-cooled to 100 K after a quick soak in 2.5 M
Li2SO4.[44] Data were collected at the PX beamline (X06SA) of the Swiss
Light Source (Paul Scherrer Institute, Switzerland), indexed, integrated,
and scaled with XDS.[45] The structures were solved by molecular
replacement with PHASER[46] with 4X50.pdb[30c] as a search model. The
models were built in COOT[47] and refined with the PHENIX software[48]
.
Geometric ligand restraints were generated with PRODRG.[49] The
atomic coordinates and structure factors are deposited in the Protein
Data Bank with PDB code 5JCR (3) and 5MUC (4).
Isothermal titration calorimetry. Standard ITC experiments were
performed at 25°C using a VP-ITC (Malvern Instruments, Worcestershire,
UK) with an injection volume between 3 µl and 15 µl, a reference power
of 10 µcal/sec, a stirring speed of 307 rpm, in high feedback mode and
with a filter period of 2 sec. Preceding the measurements, FimHLD-Th-
His6 was dialyzed against a 10 mM HEPES buffer adjusted to pH 7.4,
containing 150 mM NaCl. Ligand and protein were dissolved in the same
buffer. Protein concentration was determined by NanoDrop ND-1000
Spectrophotometer (Thermo Scientific, MA, USA) using an extinction
coefficient of 24’180 M-1 cm-1. The active protein concentration was
determined by an ITC experiment with FimHLD binding to compound 1.[41]
The thermodynamic parameters KA (association constant) and ΔH°
(change in enthalpy) and the kinetic parameter koff (dissociation rate
constant) were measured by ITC. All parameters were evaluated using
the fully automated analysis software package from AFFINImeter
(Version 1.0415 – 1.1510, Software for Science Developments, Santiago
de Compostela, Spain).[25b] The parameters ΔG° (free energy of binding)
and ΔS° (change in entropy) were calculated from Experimental Details.
equation 1, and kon (association rate constant) was calculated according
to equation 2:
Molecular Modeling. Protein-Ligand complexes of the crystal
structures (PDB accession codes: 1UWF, 4XO8, 5JCR, 5MUC)
were processed with the Protein Preparation Wizard.[50] Figure 2
was produced using PyMOL.[51]
Acknowledgements
We gratefully thank Eva Muñoz from AFFINImeter-S4SD,
Edificio Emprendia, Campus Vida, Santiago de Compostela, for
providing Figure 1. This work was supported by Swiss National
Science Foundation Grants 200020_146202 and R’EQUIP
145023.
Keywords: kinITC • method validation • kinetics • thermo-
dynamics • FimH
[1]
[2]
a) F. Bosch, L. Rosich, Pharmacology 2008, 82, 171-179; b) P. Ehrlich,
Lancet 1913, 182, 445-451.
R. A. Copeland, D. L. Pompliano, T. D. Meek, Nat. Rev. Drug Discov.
2006, 5, 730-739.
[3]
[4]
[5]
R. A. Copeland, Nat. Rev. Drug Discov. 2016, 15, 87-95.
G. Dahl, T. Akerud, Drug Discov. Today 2013, 18, 697-707.
M. Hay, D. W. Thomas, J. L. Craighead, C. Economides, J. Rosenthal,
J. Nat. Biotechnol. 2014, 32, 40-51.
ΔG° = ΔH° − TΔS° = −RT lnKA
KA = ꢀ!"/ꢀ!"" = 1/ꢀ!
(eq. 1)
(eq. 2)
with T being the absolute temperature and R the universal gas
constant (8.314 Jmol-1K-1). The thermodynamic and kinetic parameters of
compounds 1-10 and 24-29 were calculated from the average of two
independent experiments. The c values of compounds 15-21 and 23
were significantly above 1,000 for the direct titrations, therefore additional
competitive ITC experiments were performed to achieve valid KD
values.[42] These compounds were titrated into protein, which was
preincubated with excess of the weak binding compound 8, resulting in
sigmoidal titration curves. The experimental conditions and the analysis
method with Origin 7.0 (OriginLab, Northampton, MA, USA) and
SEDPHAT version 10.4 (National Institute of Health)[43] for the
competitive experiments of compounds 19 and 23 are described in Fiege
et al. 2015[30c] and for compounds 15-18, 20, and 21 in Kleeb et al.
2015.[30d] These independently derived KD values from competitive
experiments of compounds 15-21 and 23 were manually fixed for the
kinetic analysis with AFFINImeter. For the low-affinity compounds 26 and
27, the c value was below 1. To extract reliable thermodynamic and
kinetic data from low c value experiments the stoichiometry was manually
fixed to 1.[23] Correlations of the kinetic parameters kon (B) and koff (C)
with the change of free energy of binding (ΔG°), as well as the F-test to
[6]
a) F. B. E. Garah, J. L. Stigliani, F. Coslédan, B. Meunier, A. Robert,
ChemMedChem 2009, 4, 1469-1479; b) J. F. Vanhauwe, M. Ercken, D.
van de Wiel, M. Jurzak, J. E. Leysen, Psychopharmacology (Berlin,
Ger.) 2000, 150, 383-390; c) S. Nishino, J. Arrigoni, J. Shelton, T.
Kanbayashi, W. C. Dement, E. Mignot, J. Neurosci. 1997, 17, 6401-
6408.
[7]
[8]
[9]
V. A. Braz, L. A. Holladay, M. D. Barkley, Biochemistry 2009, 49, 601-
610.
B. Maschera, G. Darby, G. Palú, L. L. Wright, M. Tisdale, R. Myers,
E. D. Blair, E. S. Furfine, J. Biol. Chem. 1996, 271, 33231-33235.
A. C. Pan, D. W. Borhani, R. O. Dror, D. E. Shaw, Drug Discov. Today
2013, 18, 667-673.
[10] W. M. Kati, D. Montgomery, R. Carrick, L. Gubareva, C. Maring, K.
McDaniel, K. Steffy, A. Molla, F. Hayden, D. Kempf, Antimicrob. Agents
Chemother. 2002, 46, 1014-1021.
[11] C. Chan, S. Boyce, C. Brideau, S. Charleson, W. Cromlish, D. Ethier,
J. Evans, A. Ford-Hutchinson, M. Forrest, J. Gauthier, J. Pharmacol.
Exp. Ther. 1999, 290, 551-560.
[12] I. Dierynck, M. De Wit, E. Gustin, I. Keuleers, J. Vandersmissen, S.
Hallenberger, K. Hertogs, J. Virol. 2007, 81, 13845-13851.
[13] P. J. Tummino, R. A. Copeland, Biochemistry 2008, 47, 5481-5492.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
[14] a) H. Lu, P. J. Tonge, Curr. Opin. Chem. Biol. 2010, 14, 467-474;
b) S. Kapur, P. Seeman, J. Psychiatry Neurosci. 2000, 25, 161.
[15] G. Vauquelin, Expert Opin. Drug Discov. 2010, 5, 927-941.
[16] a) R. A. Copeland, Drug-Target Residence Time. In Thermodynamics
and Kinetics of Drug Binding, Mannhold, R.; Kubinyi, H.; Folkers, G.;
Keserü, G.; Swinney, D. C., Eds. John Wiley & Sons: Weinheim, 2015;
Vol. 65, 157-167; b) P. Y. Muller, M. N. Milton, Nat. Rev. Drug Discov.
2012, 11, 751-761.
Zalewski, F. Schadler, O. Schwardt, B. Ernst, BChemMedChem 2012,
7, 1404-1422.
[31] M. Scharenberg, X. Jiang, L. Pang, G. Navarra, S. Rabbani, F. Binder,
O. Schwardt, B. Ernst, ChemMedChem 2014, 9, 78-83.
[32] M. Gopalakrishnan, K. Forsten-Williams, T. R. Cassino, L. Padro, T. E.
Ryan, U. C. Täuber, Eur. Biophys. J. 2005, 34, 943-958.
[33] a) R. A. Copeland, Expert Opin. Drug Discov. 2010, 5, 305-310.; b) D.
C. Swinney, Curr. Opin. Drug Discov. Dev. 2009, 12, 31-39.
[34] E. V. Schneider, J. Böttcher, R. Huber, K. Maskos, L. Neumann, Proc.
Natl. Acad. Sci. U. S. A. 2013, 110, 8081-8086.
[17] G. Klebe, ChemMedChem 2015, 10, 229-231.
[18] D. C. Miller, G. Lunn, P. Jones, Y. Sabnis, N. L. Davies, P. Driscoll,
MedChemComm 2012, 3, 449-452.
[35] M. M. Sauer, R. P. Jakob, J. Eras, S Baday, D. Eriş, G. Navarra, S.
Bernèche, B. Ernst, T. Maier, R. Glockshuber, Nat. Commun., 2016, 7,
10738-10738.
[19] P. Schmidtke, F. J. Luque, J. B. Murray, X. Barril, J. Am. Chem. Soc.
2011, 133, 18903-18910.
[20] U. Jönsson, L. Fägerstam, B. Ivarsson, B. Johnsson, R. Karlsson, K.
Lundh, S. Löfås, B. Persson, H. Roos, I Rönnberg, BioTechniques
1991, 11, 620-627.
[36] J. Bouckaert, J. Berglund, M. Schembri, E. De Genst, L. Cools, M.
Wuhrer, C.-S. Hung, J. Pinkner, R. Slättegård, A. Zavialov, D.
Choudhury, S. Langermann, S. J. Hultgren, L. Wyns, P. Klemm, S.
Oscarson, S. D. Knight, H. De Greve, Mol. Microbiol. 2005, 55, 441-
455.
[21] R. Karlsson, J. Mol. Recognit. 2004, 17, 151-161.
[22] G. A. Holdgate, W. H. Ward, Drug Discov. Today 2005, 10, 1543-1550.
[23] W. B. Turnbull, A. H. Daranas, J. Am. Chem. Soc. 2003, 125, 14859-
14866.
[37] a) R. King, A. Burgen, Proc. Royal Soc. B 1976, 193, 107-125; b) P. W.
Taylor, R. W. King, A. S. Burgen, Biochemistry 1970, 9, 2638-2645.
[38] R. Gaspari, C. Rechlin, A. Heine, G. Bottegoni, W. Rocchia, D.
Schwarz, J. Bomke, H.-D. Gerber, G. Klebe, A. Cavalli, J. Med. Chem.
2016, 59, 4245-4256.
[24] G. Holdgate, S. Geschwindner, A. Breeze, G. Davies, N. Colclough, D.
Temesi, L. Ward in Protein-Ligand Interactions, Vol. 1008 (Eds. : M. A.
Williams, T. Daviter) Humana Press, 2013, pp. 327-355.
[25] a) P. Dumas, E. Ennifar, C. Da Veiga, G. Bec, W. Palau, C. Di Primo,
A. Piñeiro, J. Sabin, E. Muñoz, J. Rial in Methods in Enzymology,
Vol. 567 (Eds.: L. F. Andrew) Academic Press, 2016, pp. 157-180; b) D.
Burnouf, E. Ennifar, S. Guedich, B. Puffer, G. Hoffmann, G. Bec, F.
Disdier, M. Baltzinger, P. Dumas, J. Am. Chem. Soc. 2012, 134, 559-
565.
[39] S. Rabbani, X. Jiang, O. Schwardt, B. Ernst, Anal. Biochem. 2010, 407,
188-195.
[40] T. Klein, D. Abgottspon, M. Wittwer, S. Rabbani, J. Herold, X. Jiang, S.
Kleeb, C. Luthi, M. Scharenberg, J. Bezencon, E. Gubler, L. Pang,
M. Smiesko, B. Cutting, O. Schwardt, B. Ernst, J. Med. Chem. 2010,
53, 8627-8641.
[26] S. Igde, S. Röblitz, A. Müller, K. Kolbe, S. Boden, C. Fessele, T. K.
Lindhorst, M. Weber, L. Hartmann, Macromol. Biosci. 2017, 17,
1700198.
[41] H. Edelhoch, Biochemistry 1967, 6, 1948-1954.
[42] B. W. Sigurskjold, Anal. Biochem. 2000, 277, 260-266.
[43] J. C. Houtman, P. H. Brown, B. Bowden, H. Yamaguchi, E. Appella, L.
E. Samelson, P. Schuck, Protein Sci. 2007, 16, 30-42.
[27] a) E. V. Sokurenko, V. Chesnokova, D. E. Dykhuizen, I. Ofek, X.-R.
Wu, K. A. Krogfelt, C. Struve, M. A. Schembri, D. L. Hasty, Proc. Natl.
Acad. Sci. U. S. A. 1998, 95, 8922-8926; b) N. Sharon, FEBS Lett.
1987, 217, 145-157.
[44] K. A. Rubinson, J. E. Ladner, M. Tordova, G. L. Gilliland, Acta
Crystallogr. 2000, D56, 996-1001.
[45] W. Kabsch, Acta Crystallogr. 2010, D66, 133-144.
[46] A. J. McCoy, R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, M. D.;
Storoni, L. C.; Read, R. J., J. Appl. Crystallogr. 2007, 40, 658-674.
[47] a) P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Acta Crystallogr.
2010, D66, 486-501; b) P. Emsley, K. Cowtan, Acta Crystallogr. 2004,
D60, 2126-2132.
[28] X. R. Wu, T. T. Sun, J. J. Medina, Proc Natl. Acad. Sci. U S A 1996, 93,
9630-9635.
[29] a) D. Abgottspon, B. Ernst, Chimia 2012, 66, 166-169; b) M, Hartmann,
T. K. Lindhorst, Eur. J. Org. Chem. 2011, 2011, 3583-3609.
[30] a) C. P. Sager, B. Fiege, P. Zihlmann, R. Vannam, S. Rabbani, R. P.
Jakob, R. C. Preston, A. Zalewski, T. Maier, M. W. Peczuh, B. Ernst,
Chem. Sci. 2018, 9, 646-654; b) W. Schönemann, M. Lindegger, S.
Rabbani, S.; Zihlmann, P.; Schwardt, O.; Ernst, B., 2-C-Branched
Mannosides, A Novel Family of FimH Antagonists - Synthesis and
Biological Evaluation. Perspecti. Sci. 2017, 11, 53-61; c) B. Fiege, S.
Rabbani, R. C. Preston, R. P. Jakob, P. Zihlmann, O. Schwardt, X.
Jiang, T. Maier, B. Ernst, ChemBioChem 2015, 16, 1235-1246; d) S.
Kleeb, L. Pang, K. Mayer, D. Eriş, A. Sigl, R. C. Preston, P. Zihlmann,
T. Sharpe, R. P. Jakob, D. Abgottspon, A. S. Hutter, M. Scharenberg,
X. Jiang, G. Navarra, S. Rabbani, M. Smieško, N. Lüdin, J. Bezençon,
O. Schwardt, T. Maier, B. Ernst, J. Med. Chem. 2015, 58, 2221-2239;
e) L. Pang, S. Kleeb, K. Lemme, S. Rabbani, M. Scharenberg, A.
[48] a) P. D. Adams, P. V. Afonine, G. Bunkoczi, V. B. Chen, I. W. Davis, N.
Echols, J. J. Headd, L. W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve,
A. J. McCoy, N. W. Moriarty, R. Oeffner, R. J. Read, D. C. Richardson,
J. S. Richardson, T. C Terwilliger, P. H. Zwart, Acta Crystallogr. 2010,
D66 (Pt 2), 213-221; b) P. D. Adams, R. W. Grosse-Kunstleve, L.-W.
Hung, T. R. Ioerger, A. J. McCoy, N. W. Moriarty, R. J. Read, J. C.
Sacchettini, N. K. Sauter, T. C. Terwilliger, Acta Crystallogr. 2002, D58,
1948-1954.
[49] A. W. Schuttelkopf, D. M. Van Aalten, Acta Crystallogr. 2004, D60,
1355-1363.
[50] G. M. Sastry, M. Adzhigirey, T. Day, R. Annabhimoju, W. Sherman, J.
Comput. Aided Mol. Des. 2013, 27, 221-234.
[51] F. Bosch, L. Rosich, Pharmacology, 2008, 82, 171-17
This article is protected by copyright. All rights reserved.

10.1002/chem.201802599
Chemistry - A European Journal
FULL PAPER
FULL PAPER
In addition to thermodynamic profiles,
ITC gives access to kinetic information
of protein–ligand interactions. In this
study, kinITC-ETC, a new method to
extract kinetic parameters from ITC
data, was independently validated for
the first time by comparison with SPR.
Re-analysis of ITC data of a large set
of mannosides binding to the virulence
factor FimH revealed new insights into
their structure-kinetic relationship,
supporting the development of thera-
peutically relevant FimH antagonists.
KinITC proved to be a powerful tool to
raise kinetic information from existing
ITC data.
Dr. Pascal Zihlmann,‡ Marleen
Silbermann,‡ Dr. Timothy Sharpe,
Dr. Xiaohua Jiang, Tobias Mühlethaler,
Dr. Roman P. Jakob, Dr. Said Rabbani,
Dr. Christoph P. Sager, Priska Frei,
Prof. Dr. Timm Maier, Prof. Dr. Beat
Ernst*
Page No. – Page No.
((Insert TOC Graphic here: max.
width: 5.5 cm; max. height: 5.0 cm))
KinITC – One method supports both
thermodynamic and kinetic SARs as
exemplified on FimH antagonists
This article is protected by copyright. All rights reserved.