Efforts towards an On-Target Version of the Groebke–Blackburn–Bienaymé (GBB) Reaction for Discovery of Druglike Urokinase (uPA) Inhibitors

Article abstract of DOI:10.1002/chem.201901917

Target-guided synthesis (TGS) has emerged as a promising strategy in drug discovery. Although reported examples of TGS generally involve two-component reactions, there is a strong case for developing target-guided versions of three-component reactions (3CRs) because of their potential to deliver highly diversified druglike molecules. To this end, the Groebke–Blackburn–Bienaymé reaction was selected as a model 3CR. We recently reported a series of druglike urokinase inhibitors, and these serve as reference compounds in the present study. Due to the limited number of literature reports on target-guided 3CRs, multiple experimental parameters were optimized here. Most challenging was the formation of imine intermediates under near-physiological conditions. This aspect was addressed by exploring chemical imine stabilization strategies. Notably, imines are also crucial intermediates of other 3CRs. Such systematic studies are strongly required for further development of the TGS domain but are largely absent in the literature. Hence, this work is intended as a reference for future multicomponent-based TGS studies.

Full text of DOI:10.1002/chem.201901917

A Journal of
Accepted Article
Title: Efforts towards an on-target version of the Groebke-Blackburn-
Bienaymé (GBB) reaction for discovery of druglike urokinase
(uPA) inhibitors
Authors: Rafaela Gladysz, Johannes Vrijdag, Dries Van Rompaey,
Anne-Marie Lambeir, Koen Augustyns, Hans De Winter, and
Pieter Van der Veken
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.201901917
Link to VoR: http://dx.doi.org/10.1002/chem.201901917
Supported by

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Efforts towards an on-target version of the Groebke-Blackburn-
Bienaymé (GBB) reaction for discovery of druglike urokinase
(uPA) inhibitors
Rafaela Gladysz,[a] Johannes Vrijdag,[a,b] Dries Van Rompaey,[a] Anne-Marie Lambeir,[b] Koen
Augustyns,[a] Hans De Winter,[a] Pieter Van der Veken*[a]
Abstract: Target-guided synthesis (TGS) has emerged as
a
that selects ligand building blocks with affinity for the target.
Subsequently, the selected building blocks are assembled into
finalized ligands based on complementary chemical reactivity
and close proximity in the ligand binding site. Building blocks
that lack target affinity are not brought to reaction, so that the
corresponding ligand constructs are not formed. This effect is
known as “selective ligand amplification”. As a consequence,
TGS allows combining several aspects of design, synthesis and
potency determination, in a single time- and cost-efficient step.
Two main target-assisted strategies have been reported in
literature: (1) dynamic combinatorial chemistry (DCC) and (2)
kinetic target guided-synthesis (KTGS) (Figure 1).[1-4] DCC uses
mixtures of building blocks, which react in a reversible manner.
promising strategy in drug discovery. While reported examples of
TGS generally involve two-component reactions, there is a strong
case for developing target-guided versions of three-component
reactions (3CRs) because of their potential to deliver highly
diversified, druglike molecules. To this end, the Groebke-Blackburn-
Bienaymé reaction was selected as model 3CR. We recently
reported
a series of druglike, urokinase inhibitors serving as
reference compounds in this study. Due to the limited number of
literature reports on target-guided 3CRs, multiple experimental
parameters were optimized here. Most challenging was formation of
imine intermediates under near-physiological conditions. The latter
was addressed in this study by exploring chemical imine stabilization
strategies. Noteworthy, imines are also crucial intermediates of other
3CRs. Such systematic studies are strongly required for further
development of the TGS domain, but absent in literature. Hence this
work intends to be a reference for future multicomponent-based TGS
studies.
Adding
a
biological target to this mixture, shifts the
thermodynamic equilibrium between the formed products, and in
consequence molecules demonstrating the highest target affinity
are being amplified “on-target”. Several types of reversible
reactions have been used in protein-templated DCC.[4] However,
these reversible reactions often produce metabolically unstable
moieties (e.g. imines, hydrazones, disulfides) that are
undesirable in drug development, and require isosteric
replacement.
Introduction
Small molecule drug discovery currently has a wide range of
technologies at its disposal for the generation of new “hit” and
“lead” compounds. Typically nonetheless, all contemporary
strategies revolve around iterative cycles of design, synthesis
and potency evaluation, each time producing further optimized
compounds. This is a very time-consuming and cost-intensive
process, that could be significantly shortened by implementing
target-guided synthesis (TGS).[1]
TGS has emerged as a promising novel strategy to identify
ligands for various biological targets.[1-3] It relies on direct
assistance of a drug target, which serves as a physical template
[a]
[b]
Dr. R. Gladysz, Dr. J. Vrijdag, Dr. D. Van Rompaey, Prof. Dr. K.
Augustyns, Prof. Dr. H. De Winter, Prof. Dr. P. Van der Veken
Laboratory of Medicinal Chemistry (UAMC), Department of
Pharmaceutical Sciences
University of Antwerp
Universiteitsplein 1, 2610 Wilrijk (Belgium)
E-mail: pieter.vanderveken@uantwerpen.be
Dr. J. Vrijdag, Prof. Dr. A.-M. Lambeir
Laboratory of Medical Biochemistry, Department of Pharmaceutical
Sciences
Figure 1. Simplified representation of “on-target” approaches in drug
discovery.
University of Antwerp
Universiteitsplein 1, 2610 Wilrijk (Belgium)
In the case of KTGS, an approach proposed during the last
decade by Sharpless and co-workers, on-target assembly of
ligand molecules proceeds via an irreversible step.[5-7] Here,
building blocks showing target affinity, are positioned by the
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.201901917
Chemistry - A European Journal
FULL PAPER
Figure 2. Schematic representation of three-component TGS for enzyme inhibitor discovery.
target in close proximity and correct mutual orientation in order
to react irreversibly on-target, affording a new ligand compound.
Contrary to DCC, this is a kinetically controlled process, as the
biological template brings the reaction partners together, thereby
dramatically increasing the reaction rate compared to a non-
templated reaction in solution. Hence, amplification and
selectivity of ligand formation arise from both target affinity of the
building blocks and the target’s ability to promote coupling.[8]
To date, limited but convincing proof for the KTGS
methodology has been published. Most examples of this
approach rely on a “click reaction”, namely the Huisgen 1,3-
dipolar cycloaddition reaction.[7,9,10] The latter posseses several
features which make it well-suited for on-target development,
including bioorthogonality, selectivity, and compatibility with
aqueous media. Other reports of KTGS involve the sulfo-click
reaction,[11,12] amidation,[13] alkylation of nucleophiles with alkyl
halides[14] or epoxides[15], as well as nucleophilic conjugate
addition[16,17]. In terms of biological targets, KTGS has been
applied to a number of medically relevant enzymes that mainly
belong to the class of hydrolases, e.g., HIV-protease[6], Factor
a review article by Weber,[23] and entails the application of an
Ugi-type reaction in the target-guided identification of a thrombin
inhibitor. It is remarkable however that to date no original report
on this work has appeared, rendering a detailed assessment of
this achievement impossible. The second, and most recent
example reported by Rademann and co-workers, describes a
target-assisted Mannich-3CR of STAT5 inhibitors involving an
aromatic amine, formaldehyde and several N-heterocycles.[24]
In response to the limited number of reports on target-
assisted MCRs, we decided to focus on development of a TGS
application of the Groebke-Blackburn-Bienaymé (GBB)
condensation, an isocyanide-based three-component reaction
with high interest to drug discovery. It features an aminopyridine,
aldehyde, and isocyanide building block that react to afford a
fused imidazo[1,2-a]pyridine (Scheme 1), equipped with
substituents that allow structural diversification. Imidazopyridines
and related heterobicyclics are frequently encountered scaffolds
in drug discovery. In this framework, the potential of the GBB
condensation to deliver compounds with a highly druglike,
“decorated scaffold” architecture is a particular characteristic
that sets it apart from several other isocyanide-based 3CRs.[25]
The GBB reaction proceeds via two key steps: (1) a reversible
reaction between the 2-aminopyridine component and aldehyde
forming a Schiff base (imine), and (2) an irreversible step
involving a non-concerted (4+1) cycloaddition between the
protonated Schiff base and isocyanide, finally followed by a 1,3-
H shift.[26-28]
Xa[13]
acetylcholinesterase (AchE)[7,9,17,18]. In addition, KTGS has been
used for inhibitor development for kinases[16]
carbonic
anhydrase-II (CA-II)[14,19] and biotin protein ligase (BPL)[20]
,
insulin-degrading
enzyme
(IDE)[10]
and
,
.
Interestingly, at least two molecules obtained by KTGS were
employed in in vivo research: (1) an inhibitor of CA-II served as
the basis for a PET-imaging probe, while (2) a catalytic-site
inhibitor of IDE (BDM44768) was evaluated for in vivo
pharmacokinetics and pharmacodynamics.[10,21]
In order to increase the practical value of target-guided
approaches for drug discovery, further methodological
innovation is strongly required. A main attention point is the
identification and elaboration of chemical reaction types that are
amenable to on-target approaches. Special attention in this
framework should be given to reaction types that are capable of
deliver “druglike” molecules and have a wide functional group
compatibility in order to allow diversity-oriented synthesis. With
this respect, multicomponent reactions (MCRs) that are widely
used in classical combinatorial drug discovery seem particularly
interesting (Figure 2).[22] So far, however, this area remains
largely unexplored: published examples mostly rely on two-
component reactions, and only two examples of a target-guided
MCR appeared in literature. The first example was mentioned in
Scheme 1. The classical Groebke-Blackburn-Bienaymé (GBB) reaction.
With this work, we wanted to address two main research
questions. First of all, we wanted to investigate whether the GBB
transformation is chemically and methodologically compatible
with typical TGS conditions (involving near-physiological
aqueous conditions and low concentrations of reagents and
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Scheme 2. Intermediate structures of the GBB reaction docked in the active site of uPA.
Figure 4. The proposed binding mode of reactants: isocyanide 7 and formimine 6, in the uPA active site (PDB code 4MNW)[33-35] in the (a) surface, (b) sticks
representation. Putative hydrogen bonds are shown as dashed yellow lines.
protein template). Second, we aimed to demonstrate that
“selective ligand amplification” involving the GBB-3CR can be
achieved under TGS conditions. We envisioned that this work
could be a reference for future KGTS-studies focusing on MCRs.
To obtain maximally reliable and verifiable results, we decided to
rely on a well-defined model system, involving urokinase
plasminogen activator (uPA) as the target protein. Urokinase is a
trypsin-like serine protease, and a valuable oncology target that
is overexpressed in metastasizing solid tumors.[29-31] A set of
potent imidazopyridine-based inhibitors of uPA 1-2, together with
a number of less potent analogues 3-5, served as reference
compounds in this study (Figure 3). The synthesis of these
molecules (via classical, solution phase GBB reaction), and their
uPA potencies were reported previously by us.[32]
To the best of our knowledge, the presence of such a well-
defined experimental framework is rather unique in TGS.
Nonetheless, at this early stage of research we deemed it
necessary to pursue reliable step-by-step exploration and
optimization of experimental parameters.
Results and Discussion
Before proceeding with TGS of uPA inhibitors via the GBB
reaction, we decided to investigate the feasibility of the approach
by means of molecular modeling. A docking study was pursued
of the different reaction intermediates leading to inhibitor 1,
starting from the irreversible step of the GBB reaction: the
addition of an isocyanide to an iminium function. Further steps of
the reaction comprise conformational rearrangement, ring
closure and tautomerization (Scheme 2).[25-28] With this respect,
the formimine 6 and isocyanide 7 were selected and the binding
mode of the two intermediates in uPA was assessed.[33-35]
Relative binding affinities (see Supporting Information, Figure
S3b, Table S1) were used as a guidance to estimate which of
Figure 3. Structures and inhibitory activities of imidazopyridine uPA inhibitors
1-5 used as references in this study.[32]
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Scheme 3. Synthesis of the isocyanide building block for the on-target GBB reaction.
the two intermediates is binding as the first one. The
phenylguanidine-moiety of 7 was found to bind uPA’s S1-pocket,
engaging in a salt bridge with Asp189, while the isonitrile group
of 7 extends into the solvent-exposed part of the active site.
experiment.[36] This N-terminal modification, however, is not
affecting the enzyme‘s active site. Moreover, aldehydes have
been widely used in target-assisted approaches.[4,24] Taking into
account the presence of aldehydes in living cells, and the
Formimine 6 was subsequently docked into the [uPA-7] complex, aforementioned arguments, in our opinion, the use of an
treating 7 as a part of the receptor. As shown in Figure 4,
reactant 6 is predicted to bind near reactant 7, with a distance
between the imine carbon and the isonitrile carbon of 3.6 Å. The
reactants can thus pre-coordinate in the protein cavity in a
conformation favourable for the GBB reaction to occur.
The subsequent intermediates and final compound 1 were
then docked (displayed in Figure 5). The latter revealed that all
intermediates could bind the target in a way that is largely
comparable to the binding mode of inhibitor 1, and that
conformational changes between the intermediates were
compatible with steric constraints in uPA’s active centre. In
conclusion, these results indicate that uPA may be amenable to
on-target synthesis using the GBB reaction.
aldehyde reagent should not interfere in our on-target
experiments. Regarding the isocyanide reaction component, the
reagents of this class are used in bioorthogonal click chemistry.
An example thereof was reported by Stöckmann and co-workers,
and it presents a bioorthogonal ligation of isocyanides and
tetrazines.[37]
Synthesis
Prior to the actual TGS experiments, appropriate building blocks
were synthesized. The most challenging was the preparation of
isocyanide 7 (Scheme 3). To the best of our knowledge, this is
the first reported compound that contains both an isocyanide
and a free guanidine moiety. We envisioned that the Boc-
protected analogue 11 that we reported earlier, would be a
suitable synthetic precursor. In line with earlier reports however,
we found that application of acidolytic protocols for Boc-removal,
under all evaluated conditions caused instability of the
isocyanide function. Surprisingly, we found that refluxing the
Boc-protected precursor 11 in water under neutral conditions,
allowed to obtain the desired isocyanide 7 in high yield and
without the need for further purification. Noteworthy, in aqueous
acidic media isocyanides tend to hydrolyse fast to their
corresponding formamides.[38] The other building blocks required
for the GBB reaction (aldehydes and aminopyridines) were
either obtained from commercial suppliers or prepared using
previously reported procedures.[32]
TGS protocol optimization
Figure 5. The proposed binding mode of the GBB reaction intermediate
structures: (a) intermediate I, (b) intermediate II, (c) intermediate III, and (d)
the final compound 1 in the uPA binding site (PDB code 4MNW).[33-35]
In subsequent steps, we aimed at developing a general on-
target protocol for the GBB reaction. Herein, specific open
questions were related to (1) the nature/composition of the
reaction medium, (2) the use of stoichiometric vs. non-
stochiometric concentrations of reactants and enzyme and (3)
the LC-MS methodology for analyzing the reaction mixtures.
Finally, (4) the lower limits for reagent concentrations were
investigated. Because attempts to express uPA in E. coli did not
produce sufficient amounts of the enzyme, medical uPA (trade
name: Actosolv) was used in this study.[39] The latter is a
registered thrombolytic used in hospitals. While protein purity of
medical uPA is > 99%, the enzyme is formulated with protein-
Next, we also considered the possibility of covalent
modification of the enzyme by the reaction partners (2-
aminopyridine, aldehyde, isocyanide) based on the literature.
Regarding the 2-aminopyridine reaction partner, we do not
expect any relevant modification of the enzyme. The aldehyde
reaction component is slightly more prone to modify the enzyme
by, for instance, reacting with lysine side chains, forming imines.
This is, however, a reversible reaction. Nonetheless, aldehydes
are known to also participate in N-terminal modifications of
proteins, that are not reversible under the conditions of the
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
stabilizing additives. Removal of the latter using a desalting
column proved to be crucial for the on-target experiments.
Overall, this is a potentially relevant attention point in all TGS
studies, because many protein stabilizers, even in trace
amounts, might interfere with the on-target experiments.
TGS experiments are commonly performed at relatively
high enzyme concentrations (i.e., in the micromolar range).[1]
Since most enzymes contain multiple ionizable amino acid side
chains, they can affect the pH of the medium. With an isoelectric
point (pI) of 8.69,[48] uPA has the potential to basify the reaction
medium if insufficiently buffered. In this study, a 0.2 M buffer
concentration was needed to exclude a direct influence on pH by
the enzyme.
Reaction medium
First, an appropriate reaction medium for the target-guided GBB
reaction was identified. In general, protein-templated reactions
are carried out in aqueous buffer solutions at physiological pH or
at the target’s pH optimum. Both aqueous conditions and
appropriate pH range ensure that the target protein is present in
a correctly folded conformation: the latter is a strict prerequisite
for templated synthesis. Incompatibility of the GBB reaction with
aqueous conditions was not expected because the
mechanistically related Ugi- and Passerini-type transformations
have been reported to run in water.[40-42] Finding appropriate pH
conditions however was anticipated to be particularly
challenging, mainly because the GBB condensation involves
In order to ensure sufficient solubility for a wide range of
organic building blocks employed in TGS experiments, the
influence of co-solvents was investigated. To this aim, the
enzymatic activity was evaluated in the presence of varying
concentrations of three typical organic co-solvents: methanol,
dimethyl sulfoxide (DMSO), and acetonitrile (see Supporting
Information, Figure S1). Based on the obtained results, DMSO
was selected as the most optimal co-solvent. The latter did not
cause appreciable loss of uPA’s activity when used at 5% v/v.
Additionally, DMSO’s generally known biocompatibility makes it
one of the most commonly used organic solvent in biochemical
assays.[49] The lifetime of uPA was monitored in the selected
buffer media, and was proven satisfactory (see Supporting
Information, Figure S2).
imine formation as its first and rate-determining step,
a
characteristic it shares with many other types of multicomponent
reactions. Imine formation is known to be an acid-catalyzed
process with highest rates at pH 4-6, while uPA’s pH optimum is
close to 8.5.[43] Therefore, a compromise had to be found that
allowed for the GBB reaction to proceed at acceptable rates at
relatively low reagent concentrations, while uPA remains
correctly folded. As a proxy for correct protein folding, we
measured enzymatic activity: this implies the presence of a
properly folded protein. Noteworthy, several other factors than
protein folding can affect enzyme activity as well, for example
the pH-dependent protonation state of catalytically active
residues. Absence or decrease of activity therefore not
necessarily means that the protein is denaturated.
Furthermore, many classical buffers used in enzymology
or in protein formulation contain functional groups that can
interfere with the desired GBB reaction and could lead to
formation of Ugi or Passerini-type side products. Two examples
thereof were encountered during this study. First, products of a
Passerini-type condensation between isocyanide 7, aldehyde,
and glycine were unexpectedly identified in on-target reaction
mixtures. These were traced back to glycine used as a stabilizer
in the ‘medical’ uPA formulation. Second, a Passerini-type
reaction was found to occur between isocyanide 7, aldehyde
and water, in specific reactions where phosphate ions were
present in the buffer mixture. Indeed, phosphates are known
catalysts of several reactions,[44-46] and this should be
considered while designing on-target experiments. More
information on the Passerini-type side reactions observed in this
study can be found in the Supporting Information (Schemes S6
and S21-S22, Figures S9-S10, Table S2).
Reactant Concentration
In general, reported examples of KTGS employ micromolar up to
low-millimolar reactant concentrations in order to allow robust
analysis of the formed ligation product. Both stoichiometric and
non-stoichiometric combinations of building blocks have been
reported.[1,13] In case of non-stoichiometric variants, the building
block which is expected to possess the highest affinity has
usually been the concentration-limiting reagent. Nonetheless, an
experimentally backed discussion on the rationale behind the
selection of a stoichiometric or non-stoichiometric experimental
set-up is virtually absent in literature.
In preliminary experiments, the rate of the GBB reaction
was first determined in the selected buffers in the absence of
enzyme. As a model, the GBB condensation of 2-aminopyridine
12a, glyoxylic acid 13a, and isocyanide 7 was taken, affording
one of the most potent uPA inhibitors within the reference set:
compound 1 (IC50 = 0.184 ± 0.007 μM) (Scheme 4). Glyoxylic
acid was used here as an efficient formaldehyde equivalent.[32]
Mechanistically, spontaneous decarboxylation is known to occur
as the last step of the GBB reaction using glyoxylic acid.
Based on these considerations, two structurally related
buffers were selected: (1) HEPES (pKa2(20 °C)= 7.55) for
experiments at pH 8.0 and pH 7.0 and (2) MES (pKa(20 °C)=
6.15) for experiments at pH 5.5.[47] Both of the selected buffers
are compatible with uPA’s catalytic activity and lack functional
groups that can participate in isocyanide-based 3CRs reaction.
Scheme 4. The GBB reaction for the formation of uPA inhibitor 1.
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Here, stochiometric reactant concentrations in the sub-
millimolar to low millimolar range were compared with non-
stochiometric variants (Scheme 4). In HEPES buffer at pH 8.0
(the ‘classical’ uPA assay buffer), the rate of the GBB reaction
was neglectable in all cases. Not unexpectedly, HEPES buffer at
pH 7.0, which is still close to the optimal pH of the enzyme,
increased the reaction rate when compared to the initial
conditions. In the latter case, the GBB reaction using all the
reagents at 2.5 mM concentration afforded 1.1 μM of product 1
after 24 h (0.044% yield). In MES buffer at pH 5.5, although
further from the uPA optimal pH, a further increase in the 3CR
rate was observed compared to the reaction at pH 7.0. In this
case, the same reaction afforded 5.2 μM of product 1 after 24 h
(0.21% yield). Although product formation rates at pH 7.0 and
5.5 were still low in absolute terms, this was not considered
problematic for the TGS-approach, taking into account that a
uPA-promoted process could be considerably faster.
Furthermore, we hypothesized imine-formation to be critically
implicated in the observed low product formation rates.
Chemical imine stabilization strategies (vide infra) therefore
were expected to further amend this situation. Based on all
these data, two standardized experimental conditions were
defined for use in the subsequent TGS experiments. The first
(referred to as conditions A) uses GBB-reactants at 1 mM
concentration in MES buffer / DMSO (95:5) at pH 5.5. The
second (referred to as conditions B) involves GBB reactants at
2.5 mM in HEPES buffer / DMSO (95:5) at pH 7.0. The higher
reactant concentration in the latter was chosen to compensate
for the intrinsically lower reactivity at pH 7.0.
Table 1. Optimized experimental parameters for TGS of uPA inhibitors.
Parameter
Optimized conditions for TGS
1. Reaction medium
a.
Buffer, pH
MES buffer of pH 5.5 (conditions A)
HEPES buffer of pH 7.0 (conditions B)
0.2 M
b.
c.
Buffer molarity
DMSO content
5% v/v
2. Reagent concentration
3. Enzyme[a] concentration
4. Reaction time
Stoichiometric, 1 mM (conditions A)
Stoichiometric, 2.5 mM (conditions B)
Catalytic, approx. 21 μM, corresponding to
approx.
1 mol% (cond. B) or 2 mol%
(cond. A)
24 h - 48 h
[a] Target enzyme, urokinase plasminogen activator,[39] was purified before
on-target experiments using desalting column, and the protein
a
concentration was determined spectrophotometrically according to the
equation: [uPA] = OD280 / ε(uPA); where OD280 - absorbance at λ = 280 nM, ε -
absorbance coefficient, for mature human urokinase it equals 1.527.[48]
knowledge, existing LC-MS-SIM-based TGS reports therefore
do not provide reliable quantitative data. Nonetheless, Manetsch
and co-workers[18] were able to determine a range in which the
Next, the concentration of uPA for use in the TGS
experiments, was studied using the same model reaction
towards inhibitor 1 (Scheme 4). A recent review claims that in
TGS experiments, the target should ideally be present in a
similar concentration range as the reactants.[1] Multiple literature
examples have nonetheless demonstrated ligand amplification
with a catalytic amount of the biological target.[10,13,20,50] In
response to this divergence, three possible experimental set-ups
were investigated: (a) catalytic enzyme concentrations (0.5 – 2
mol%), (b) enzyme concentrations comparable to the building
block concentration, and (c) enzyme concentrations higher than
the building block concentration. The two latter scenarios were
found not to be favorable. Experiments involving high
concentrations of uPA showed substantially lower product
formation rates than negative controls in buffer. As a result, the
final on-target experiments were performed in the presence of a
catalytic amount (approx. 1 – 2 mol%) of uPA. A summary of all
experimental conditions determined so far is given in Table 1.
Noteworthy, the preliminary experiments involving uPA were not
indicative of ligand amplification. Due to the early stage of the
study however, no conclusions were drawn at this point.
SIM-peak
areas
linearily
correlated
with
compound
concentrations. Alternatively, Tieu and co-workers performed
product quantitation separately by comparing compound peak
area in the UV trace to a standard curve obtained using the
reference compound.[20] It also deserves mentioning that
reference compounds are also used in many other TGS-studies,
but they rather serve to confirm the identity of the formed
products (in terms of m/z, tr), than for quantitative calibration of
10-11, 52]
the SIM-methodology.[6,
Moreover, Rademann and co-
workers have applied high-resolution HPLC-QTOF-MS (QTOF:
quadrupole-time-of-flight) in KTGS studies.[13, 24] In these reports,
the obtained ligation products were quantified based on
calibration curves determined with reference molecules.
Importantly, in one of these reports, also the rate of product
formation was determined.[13]
Also tandem mass spectrometry (MS/MS), by involving
several steps of ion fragmentation and selection, allows for
thorough quantitative analysis of reaction mixtures. Since we
considered quantitative analysis an absolute prerequisite for
correct interpretation of the TGS data, we relied on a UPLC-
TQD-MS instrument (TQD: tandem quadrupole detector) in the
Multiple Reaction Monitoring (MRM) mode. The latter allows to
monitor product ions based on the formed parent-daughter ion
couples, known also as transition couples.[53-54] This feature
reduces the risk of false positive results. Additionally, MRM-MS
Analytics
To date, most of the reported examples of TGS, use LC-MS in
the Selected Ion Monitoring (SIM) mode as analytical
methodology.[1, 51] Although highly sensitive, SIM is not a method
that by default allows quantitative analysis.[1] To the best of our
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
records the intensity of the detected transition couples, which is
proportional to the actual ligand concentration. Hence, MRM
enables both qualitative and quantitative analysis of the detected
hit compounds. The MRM method is created using reference
compounds. In this study, for each of the investigated
compounds 1-5, a separate MRM method was prepared, and
the product quantification was based on calibration curves (see
Supporting Information, Figures S4-S8).
in the absence of uPA. Benzotriazole adducts 15a-d were
prepared from benzotriazole and the corresponding
aminopyridine and aldehyde compounds (Scheme 6) based on
a modified reaction protocol from Katritzky et al.[55] The p-
toluenesulfinic acid adducts 17a-g, on the other hand, were
obtained by reacting p-toluenesulfinic acid with the
corresponding aminopyridine and aldehyde components
(Scheme 7). The structure of the imine adducts was confirmed
by 2D NMR experiments (HSQC, HMBC). Detailed experimental
procedures and characterization data for the preparation of
imine adducts 15a-d, 17a-g can be found in the Supporting
Information (Schemes S4-S5, and the General methods section).
Imine stabilization
Next, effort was put in addressing the problem of the sluggish
imine formation in aqueous media and at suboptimal pH. Hence,
we evaluated the possibility of chemically stabilizing preformed
imine intermediate in the form of a reversible adduct with (1)
benzotriazole or (2) p-toluenesulfinic acid. This strategy allows
for a quantitative pre-condensation of amine and aldehyde
building blocks. Furthermore, the reversible nature of adduct
formation, also ensures that the imines are released again over
time by the adducts, making them available for reaction with the
isocyanide via
(Scheme 5).
a pseudo two-component reaction (2CR)
Scheme 6. Synthesis of benzotriazole imine adducts 15a-d.
Scheme 5. Pseudo 2CR using stabilized imine intermediates towards
substituted imidazo[1,2-a]pyridines.
Benzotriazole adducts of imines and related functionalities
have been studied most extensively by Katritzky and co-
workers.[55-58] Such adducts have been used in, among others,
the Strecker reaction[55], and the alkylation of (hetero)aromatic
systems.[59] Interestingly, this strategy has also been reported for
the synthesis of functionalized imidazo[1,2-a]pyridines.[60-61]
Although involving the GBB reaction, the latter examples differ
from the strategy applied here by requiring an intermediate
Scheme 7. Synthesis of p-toluenesulfinic acid imine adducts 17a-g.
In a first set of experiments, the adducts were reacted with
isocyanide 7 in the absence of enzyme, under both experimental
conditions A and B (Scheme 8, results summarized in Tables 2-
4). Yields of a classical GBB protocol (involving condensation of
amine, aldehyde and isocyanide 7) under the same
experimental conditions were determined for comparison. The
reactivity of adducts that produce reference inhibitor 3 are
shown in Table 2. Compared to a ‘classical’ protocol, the use of
imine adducts was found to increase product yield under
experimental conditions A (Table 2, Entries 1-6). Noteworthy,
formaldehyde-based adducts perform considerably better than
glyoxylate-based analogues. The same trend was observed
amongst the non-carboxylated compounds, while performing the
reactions under conditions B (Table 2, Entries 7-12).
methylation step, and
a metal catalyst. Likewise, the p-
toluenesulfinic acid adducts of imines have been reported for the
Strecker reaction,[62-63] and for the solid phase synthesis of
dihydropyrimidinones.[64]
In total, four benzotriazole- and seven p-toluenesulfinic
acid adducts were synthesized in this study (Scheme 6 and
Scheme 7, respectively). Upon reaction with isocyanide 7, all of
these produce one of the reference inhibitors 1-5. Next to
phenylpropanal
and
3-pyridinecarboxaldehyde,
both
formaldehyde and its equivalent glyoxylate were used as the
aldehyde building blocks in the adducts. Because the -COOH
group in the latter is known to spontaneously decarboxylate after
the GBB condensation (vide supra) we wanted to evaluate the
reactivity of the corresponding adducts both in TGS settings and
In the second series of experiments, we investigated the
formation of inhibitor 1. First, we compared a 3CR involving
glyoxylic acid (Table 3, Entry 1) with the reaction of the
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Table 2. Comparison of 3CRs and pseudo 2CRs towards inhibitor
(Scheme 8).
3
Entry
Imine source
Experimental
conditions[a]
Yield (cpd. 3)[b] [%]
1
12b, 13a
15a
A
A
A
A
A
A
B
B
B
B
B
B
0.097
0.19
0.16
0.51
0.90
0.82
0.23
0.30
0.21
0.65
0.91
0.77
2
3
17a
4
12b, 13b
15b
5
6
17b
7
12b, 13a
15a
8
9
17a
10
11
12
12b, 13b
15b
17b
Scheme 8. “True” 3CRs and pseudo 2CRs in the absence of uPA towards
inhibitors 1, 3-4.
[a] Experimental conditions. Temperature: rt. A: Reaction medium: MES
buffer (pH 5.5)/ DMSO (95:5). Reagents concentration: 1 mM. B: Reaction
medium: HEPES buffer (pH 7.0)/ DMSO (95:5). Reagents concentration: 2.5
mM. [b] Determined by UPLC-TQD-MS measurement and calibration curves,
after 24 h reaction time. Average of two experiments.
p-toluenesulfinic acid adduct 17e (Entry 2) under experimental
conditions A. The benzotriazole analogue of adduct 17e was not
included in this set due its troublesome synthesis. Again, the
reaction involving an imine adduct gave higher conversion than
the classcial 3CR. The same tendency was observed while
comparing the 3CR involving formaldehyde (Entry 3) with the
reaction of adduct 17c (Entry 4). However, the reaction with
adduct 17c gave over 3-fold higher conversion than the reaction
using the corresponding carboxyl analogue 17e. The yield of the
reaction involving benzotriazole analogue 15c is not reported
here due to its insufficient solubility under the experimental
conditions. When these reactions were repeated under the
experimental conditions B (Table 3, Entries 5-8), the same
reactivity trend was observed.
On-target experiments
With these optimized conditions in hand, two types of TGS
experiments were investigated. A first setup involved a single
combination of reactants in the presence of uPA, giving rise to a
single inhibitor (Scheme 9). This experimental setup included
both (1) “true” 3CRs between separate sets of one
aminopyridine, one aldehyde and isocyanide 7 (Scheme 9a),
and (2) pseudo 2CRs between separate sets of one imine
adduct and isocyanide 7 (Scheme 9b). More specifically, we
compared here the outcome of reactions involving five
combinations of building blocks leading to uPA inhibitors 1-5.
Within this inhibitor series, relative binding affinities span three
orders of magnitude, and we expected the latter to be reflected
in the relative rate of the 5 individual enzyme-templated
reactions, based on theoretical considerations. A second setup
explored in this study, involved a pseudo 2CR featuring the
simultaneous introduction of two imine adducts to isocyanide 7,
while other experimental parameters remained constant
(Scheme 10). Although the latter experiment starts from the
same basic assumptions, it is different from the first one
because it includes competition between building blocks
demonstrating different binding affinities.
In the last series of experiments we compared a 3CR
forming inhibitor
4 with the corresponding pseudo two-
component reactions using imine adducts 15d and 17d (Table
4). Reactions performed under conditions A (Entries 1-3)
showed that both adducts (15d, 17d) gave comparable
conversion to the desired product over time, which was about 2-
fold higher than obtained from the 3CR. Under the experimental
conditions B (Entries 4-6), the use of imine adduct 17d again
resulted in increased formation of product 4 compared to the
classical 3CR.
As a result of their favorable solubility and reactivity, the
following on-target experiments were using p-toluenesulfinic acid
adducts without a carboxylic acid moiety (compounds 17b-d and
17f-g).
The first on-target experiment involved a “classical” 3CR
between 2-aminopyridine 12a, glyoxylic acid 13a and isocyanide
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Table 3. Comparison of 3CRs and pseudo 2CRs towards inhibitor
(Scheme 8).
1
Entry
Imine source
Experimental
conditions[a]
Yield (cpd. 1)[b] [%]
1
2
3
4
5
6
7
8
12a, 13a
17e
A
A
A
A
B
B
B
B
0.12
0.18
0.43
0.56
0.043
0.12
0.15
0.23
12a, 13b
17c
12a, 13a
17e
12a, 13b
17c
[a] Experimental conditions. Temperature: rt. A: Reaction medium: MES
buffer (pH 5.5)/ DMSO (95:5). Reagents concentration: 1 mM. B: Reaction
medium: HEPES buffer (pH 7.0)/ DMSO (95:5). Reagents concentration: 2.5
mM. [b] Determined by UPLC-TQD-MS measurement and calibration curves,
after 24 h reaction time. Average of two experiments.
Scheme 9. On-target version of the 3CRs and pseudo 2CRs towards uPA
inhibitors.
reaction medium MES buffer at pH 5.5 / DMSO (95:5), each of
the building blocks at 1 mM concentration, and uPA at 21 μM
concentration. Besides, a negative control reaction in buffer
(CTRL) was included here. Subsequently, the on-target reaction
as well as the control experiment were monitored over time. As
a result, in contrast to our expectations, this 3CR showed almost
identical conversion in both the experiment involving uPA and
the CTRL (Table 5, Entries 1-2). Moreover, the latter were
characterized by very low reaction rates. Here the reaction with
uPA resulted after 24 h in 0.13% yield, corresponding to 1.3 μM
of the product 1.
Therefore, we decided to investigate next the pseudo 2CR
reaction variant involving isocyanide 7 and adduct 17c (Scheme
9b). The later on-target experiment was performed first under
experimental conditions A. Additionally, two control experiments
were included this time: (1) a negative control reaction in buffer,
and (2) a reaction using human serum albumin (HSA) instead of
uPA. This experiment again showed that both the reaction
involving uPA as well as the CTRL demonstrated very
comparable product formation (Table 5, Entries 3-4). However,
not unexpectedly, this pseudo 2CR was characterized by an
over 5-fold higher conversion than the 3CR reaction variant. In
all performed experiments, the pseudo 2CR reaction mainly
proceeded during the first 24 h, with the biggest increase in
product formation within the first 5 h. After 24 h, the reaction rate
was neglectable. For the sake of completeness, experiments
were nonetheless monitored for 48 h, and after 1 week. Here,
the reaction involving uPA showed the same conversion as the
CTRL experiment in the whole timeframe. After 24 h, the
reaction resulted in 0.68% yield, what corresponeded to 6.8 μM
Table 4. Comparison of 3CRs and pseudo 2CRs towards inhibitor
(Scheme 8).
4
Entry
Imine source
Experimental
conditions[a]
Yield (cpd. 4)[b] [%]
1
2
3
4
5
6
12b, 13c
15d
A
A
A
B
B
B
0.33
0.58
0.61
1.3
17d
12b, 13c
15d
YND[c]
2.1
17d
[a] Experimental conditions. Temperature: rt. A: Reaction medium: MES
buffer (pH 5.5)/ DMSO (95:5). Reagents concentration: 1 mM. B: Reaction
medium: HEPES buffer (pH 7.0)/ DMSO (95:5). Reagents concentration: 2.5
mM. [b] Determined by UPLC-TQD-MS measurement and calibration curves,
after 24 h reaction time. Average of two experiments. [c] Yield of this reaction
was not determined (YND) due to insufficient solubility of compound 15d
under the experimental conditions.
7 to form imidazopyridine 1, a uPA inhibitor with nanomolar
potency (IC50 = 184 ± 7 nM) (Scheme 9a). The later experiment
was performed under experimental conditions A, using as
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Table 5. Results of on-target experiments towards uPA inhibitors 1-2 (Scheme 9).
Entry
1
Reagents
Product
Target or control
uPA
Experimental conditions[a]
Yield[b] [%]
0.13
0.12
0.68
0.68
0.61
0.20
0.21
0.18
0.19
0.22
0.78
0.76
0.73
0.23
0.23
0.21
Relative conc.[c]
12a, 13a, 7
1
A, [uPA] = 21 μM
1.1
1
2
buffer
uPA
A
3
17c, 7
17c, 7
1
1
A, [uPA] = 21 μM
1
4
buffer
HSA
A
1
5
A, [HSA] = 21 μM
B, [uPA] = 20.5 μM
B
0.89
0.98
1
6
uPA
7
buffer
HSA
8
B, [HSA] = 21 μM
B, [uPA] = 84 μM
B
0.88
0.84
1
9
17c, 7
17f, 7
1
2
uPA
10
11
12
13
14
15
16
buffer
uPA
A, [uPA] = 21.5 μM
A
1.02
1
buffer
HSA
A, [HSA] = 21 μM
B, [uPA] = 20 μM
B
0.95
0.97
1
17f, 7
2
uPA
buffer
HSA
B, [HSA] = 21 μM
0.89
[a] Experimental conditions. Temperature: rt. A: Reaction medium: MES buffer (pH 5.5)/ DMSO (95:5). Reagents concentration: 1 mM. B: Reaction
medium: HEPES buffer (pH 7.0)/ DMSO (95:5). Reagents concentration: 2.5 mM. [b] Determined by UPLC-TQD-MS measurement and calibration
curves, after 24 h reaction time. Average of two experiments. [c] Relative concentration refers to the product concentration in the sample calculated
relatively to the product concentration in the corresponding negative control in buffer, obtained by division. Within a single on-target experiment, product
concentration in the negative control in buffer refers to as 1.
of the product 1. The control experiment with HSA showed that
HSA does not promote product formation, but in fact interferes
with the reaction resulting in lower conversion (Table 5, Entry 5).
The pseudo two-component reaction on-target towards
inhibitor 1 was then repeated under experimental conditions B,
using HEPES buffer at pH 7.0 / DMSO (95:5) as the reaction
medium, and each of the building blocks (isocyanide 7 and
adduct 17c) at 2.5 mM concentration. As a result, again, the
reaction involving uPA and CTRL showed a very comparable
conversion over time, and the control with HSA gave lower
conversion (Table 5, Entries 6-8).
Due to the observed lack of difference between the
experiment involving uPA and CTRL, we decided to investigate
the effect of increased enzyme concentration. Hence, we
pursued the on-target experiment with 4-fold higher
concentration of uPA (84 μM) using experimental conditions B.
As a result, the on-target experiment showed lower reaction
outcome than the corresponding CTRL experiment (Table 5,
Entries 9-10). The possibility of incomplete analyte recovery
from solutions containing the highest protein concentrations was
also evaluated here, but the recovery was found to be identical
for solutions with and without protein (see Supporting
Information, Scheme S11, and the General protocol for pseudo
two-component (2CR) reactions on-target). We therefore
assume that the lower reaction outcome observed in case of
experiments involving the highest uPA concentration is due to
interference of the enzyme with the reaction. The latter may be a
result of e.g. (1) side reactions between enzyme amino acid side
chains and reagents or (2) faster decomposition of the imine
adduct 17c or isocyanide compound 7.
The second pseudo two-component reaction tested on-
target, involved isocyanide 7 and adduct 17f. This reaction
affords the most potent uPA-inhibitor within the reference set:
compound 2 (IC50 = 97 ± 10 nM). The same experimental
strategy as described in the previous experiment was applied.
First, we reacted isocyanide
7 and adduct 17f under
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Table 6. Results of on-target experiments towards uPA inhibitors 3-5 (Scheme 9).
Entry
1
Reagents
Product
Target or control
uPA
Experimental conditions[a]
A, [uPA] = 21.5 μM
A
Yield[b] [%]
0.68
Relative conc.[c]
17b, 7
3
1.07
1
2
buffer
HSA
0.63
3
A, [HSA] = 21 μM
B, [uPA] = 20 μM
B
0.64
1.02
0.93
1
4
17b, 7
3
uPA
0.81
5
buffer
HSA
0.87
6
B, [HSA] = 21 μM
A, [uPA] = 20 μM
A
0.77
0.88
1.01
1
7
17d, 7
4
5
uPA
0.46
8
buffer
uPA
0.45
9
12c, 13d, 7
A,[d] [uPA] = 22 μM
A[d]
0.0056
0.028
0.025
0.0086
0.045
0.042
0.013
0.045
0.041
0.21
1
10
11
12
13
14
15
16
17
buffer
HSA
A,[d] [HSA] = 22 μM
A,[d] [uPA] = 22 μM
A[d]
0.89
0.2
1
17g, 7
17g, 7
5
5
uPA
buffer
HSA
A,[d] [HSA] = 21 μM
B,[d] [uPA] = 22 μM
B[d]
0.93
0.29
1
uPA
buffer
HSA
B,[d] [HSA] = 22 μM
0.91
[a] Experimental conditions. Temperature: rt. A: Reaction medium: MES buffer (pH 5.5)/ DMSO (95:5). Reagents concentration: 1 mM. B: Reaction
medium: HEPES buffer (pH 7.0)/ DMSO (95:5). Reagents concentration: 2.5 mM. [b] Determined by UPLC-TQD-MS measurement and calibration
curves, after 24 h reaction time. Average of two experiments. [c] Relative concentration refers to the product concentration in the sample calculated
relatively to the product concentration in the corresponding negative control in buffer, obtained by division. Within a single on-target experiment, product
concentration in the negative control in buffer refers to as 1. [d] Due to a lower reactivity, this reaction required higher reagent concentration.
Experimental conditions A; reagents concentration: 2.5 mM. Experimental conditions B; reagents concentration: 5 mM.
experimental conditions A. Second, the same reaction was
repeated under experimental conditions B. However,
experiments involving a higher concentration of uPA were not
included because of the observed interference of high amounts
of enzyme on reaction yield. These experiments gave very
comparable results to the on-target reaction for the synthesis of
inhibitor 1 (Table 5, Entries 11-16). In particular, neither the
experiment in MES buffer nor HEPES buffer demonstrated uPA-
templated amplification of compound 2 in the whole timeframe.
Also here, control experiments with HSA showed lower reaction
outcome.
cause catalyst poisoning, thus inhibiting the enzyme to further
function as template during the on-target process. This situation
could be different in reactions affording less potent inhibitors.
To this end, isocyanide 7 was reacted with adduct 17b, to
afford the micromolar inhibitor 3 (IC50 = 9.04 ± 0.62 μM). The
same experimental strategy was applied as for compound 2.
Again, however, neither the on-target reaction performed under
conditions
A nor B showed uPA-templated formation of
compound 3 over 24 h (Table 6, Entries 1-6) as well as 48 h.
The experiment aiming at the on-target synthesis of high-
micromolar inhibitor 4 from the adduct 17d and isocyanide 7 had
a similar outcome (Table 6, Entries 7-8).
The next set of experiments was aiming at the uPA-
templated synthesis of uPA inhibitor 5 (IC50 = 6.89 ± 0.80 μM).
First, a 3CR between aminopyridine 12c, nicotinaldehyde 13d
Next, we compared the outcome of the above experiments
(involving highly potent inhibitors 1 and 2, Table 5), with the
outcome of reactions affording the less potent ligands 3-5 (Table
6). We hypothesized that ligands of nanomolar affinity could
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Scheme 10. Competition-based experiment using adducts 17b and 17f towards inhibitors 2 and 3.
Table 7. Results of the competition-based experiment towards inhibitors 2 and 3 (Scheme 10).
Entry
Target or control
uPA
Experimental conditions[a]
A, [uPA] = 21.5 μM
A
Yield (cpd. 2)[b] [%]
Rel. conc.[c] cpd. 2
Yield (cpd. 3)[b] [%]
Rel. conc.[c] cpd. 3
1
2
3
4
5
6
0.92
0.89
0.84
0.21
0.22
0.19
1.04
1
0.83
0.82
0.79
0.61
0.62
0.53
1.01
1
buffer
HSA
A, [HSA] = 21 μM
B, [uPA] = 20 μM
B
0.94
0.97
1
0.96
0.98
1
uPA
buffer
HSA
B, [HSA] = 21 μM
0.87
0.86
[a] Experimental conditions. Temperature: rt. A: Reaction medium: MES buffer (pH 5.5)/ DMSO (95:5). Reagents concentration: 1 mM. B: Reaction medium:
HEPES buffer (pH 7.0)/ DMSO (95:5). Reagents concentration: 2.5 mM. [b] Determined by UPLC-TQD-MS measurement and calibration curves, after 24 h
reaction time. Average of two experiments. [c] Relative concentration refers to the product concentration in the sample calculated relatively to the product
concentration in the corresponding negative control in buffer, obtained by division. Within a single on-target experiment, product concentration in the negative
control in buffer refers to as 1.
and isocyanide 7, in the presence of uPA was evaluated under
experimental conditions A. Due to the observed lower reactivity
of the reagents, this reaction required higher reagent
concentration. As a result, the reaction with uPA gave over 4
times lower conversion than the CTRL. However, the experiment
with HSA showed only slightly lower conversion than the CTRL
(Table 6, Entries 9-11).
The outcome of the 3CR towards inhibitor 5 was then
compared with the corresponding pseudo 2CR on-target. This
2CR involved imine adduct 17g and isocyanide 7. Although
notably faster, the pseudo 2CR on-target demonstrated the
same trend as its 3CR counterpart. More precisely, we observed
that both under adapted experimental conditions A and B, the
reaction involving uPA demonstrated 3-5 times lower outcome
than the corresponding CTRL reaction. Furthermore, similar as
before, the experiment with HSA showed only slightly lower
conversion than the CTRL (Table 6, Entries 12-17).
also interfere with the on-target reaction, what in some cases
could complicate the proper interpretation of TGS results. Here,
comparing negative control reactions in buffer with the reactions
involving uPA, can indeed hint towards interference of the
enzyme with the studied reaction. On the other hand,
discrepancy in the rates of reactions involving uPA and HSA can
be related to differences in amino acid sequence of uPA and
HSA reflected by their various isoelectric points (pI(uPA) = 8.69
vs. pI(HSA) = 5.8).[48, 65]
Next, we designed
a competition-based experiment
involving two uPA ligands displaying a 100-fold difference in
binding affinity: compounds 2 and 3. The rationale behind this
experiment was to investigate the possibility of TGS via an
alternative strategy than in the foregoing experiments, namely
by allowing direct competition between building blocks of
different affinity. This competition was expected to lead to a
product distribution reflecting the relative binding potencies
(represented by their IC50-values) of the inhibitors. In this way,
even in cases where the enzyme would slow down the reaction,
the relative amounts of the formed products could be an
indication of the enzyme-templated reactivity. Likewise, product
We do not have a clear explanation why both the classical
3CR and the pseudo 2CR with uPA towards inhibitor 5 result in
lower yield than the corresponding control reactions. Certainly,
this experiment shows that enzymes may not only promote, but
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
General methods. Commercially sourced reagents were used without
further purification. All solvents were reagent-grade, and in case of
purification of the final compounds HPLC grade were used. Where
necessary, flash purification was performed with a Biotage ISOLERA
One flash system equipped with an internal variable dual-wavelength
diode array detector (200-400 nm). SNAP cartridges were used for
normal phase purifications KP-Sil and for reversed phase purifications
KP-C18-HS. Dry sample loading was done by self-packing samplet
cartridges using silica and Celite 545, respectively, for normal and
reversed phase purifications. Gradients used varied for each purification.
Solvents used for the column chromatography were ethyl acetate,
heptane and methanol. 1H NMR and 13C NMR spectra were recorded
distribution should be notably different between the on-target
reaction and CTRL, in which products are formed only according
to the chemical reactivities of building blocks.
Herein, we reacted the corresponding imine adducts 17b
and 17f with isocyanide 7 in the presence of uPA under
conditions A and B (Scheme 10). As a result, the relative
amounts of products 2 and 3 formed in this experiment were
comparable under each of the applied experimental conditions
(A and B), whereas direct competition between building blocks
should be reflected by considerably higher amplification of
compound 2 over 3. In addition, the CTRL experiment showed
the same product distribution as the reaction involving uPA
(Table 7). Hence, it can be concluded that this experiment did
not show direct competition for uPA’s active site between the
selected building blocks.
1
with a Bruker Avance DRX 400 spectrometer (400 MHz for H, 101 MHz
for 13C) at 21 °C. Chemical shifts (δ) are reported in ppm relative to the
residual solvent peak. Splitting patterns are indicated as singlet (s),
doublet (d), triplet (t), quartet (q), multiplet (m), broad (br). The coupling
constants (J) are reported in hertz (Hz). Melting points were determined
with a melting point M-560 apparatus (BUCHI) and are uncorrected. A
Waters Acquity UPLC® system coupled to a Waters TUV detector and a
Waters TQD ESI mass spectrometer was used. The wavelength for UV
detection was 254 nm. A Waters Acquity UPLC® BEH C18 1.7 μm, 2.1
mm x 50 mm column was used. Column temperature was 30 °C.
Regarding conditions of the mobile phase, monitoring of the reaction
progress, as well as product purity determination was performed using
Method I. This method used as mobile phase solvent A: water with 0.1%
formic acid, and solvent B: acetonitrile with 0.1% formic acid. Conditions
of the mobile phase (Method I): flow 0.7 mL/min, gradient of the mobile
phase: 0-0.15 min: 95% solvent A, 5% solvent B (isocratic); 0.15-2.5 min:
linear gradient from 5% to 100% solvent B; 2.5-3.25 min: 100% solvent B
(isocratic); 3.25-4.0 min: further system equilibration (95% solvent A, 5%
solvent B), flow 0.35 mL/min.
Conclusions
In conclusion, we have described our efforts towards the
development of a target-guided version of the GBB reaction. Our
study relied on a well-defined model system involving urokinase
(uPA) as target protein. A set of imidazopyridine-based inhibitors
of uPA, which were previously reported by us, served as
reference compounds in this work.[32] A major hurdle to take
when developping an on-target protocol for the GBB reaction,
was found to be the imine formation step in aqueous media.
Most likely, the latter will represent a general challenge in the
development of TGS protocols for medicinally relevant MCRs
involving imine intermediates (e.g., the Ugi reaction, Hantzsch
synthesis, and A3-reaction). In order to address imine formation
problems, we applied a strategy based on trapping imines in the
form of a metastable adduct with either benzotriazole or p-
toluenesulfinic acid. Subsequently, a number of prepared imine
adducts were reacted with an isocyanide in the presence of uPA
following the developed on-target protocol. Unfortunately, our
on-target experiments for the formation of uPA inhibitors 1-5, did
not demonstrate a notable enzyme-templating activity. The latter
applied for both the reactions involving a single combination of
building blocks, and a competition-based experiment involving
two combinations of building blocks. These experimental results
were not in line with our initial molecular modeling study, which
indicated that the templated GBB reaction is sterically possible
in uPA’s active site.
Target-guided synthesis. Target-guided synthesis experiments were
monitored using a Waters Acquity UPLC® system and mobile phase
conditions described under Method II. The latter method used as mobile
phase solvent A: 5 mM ammonium acetate buffer (pH 4.0 adjusted with
formic acid); solvent B: acetonitrile. Conditions of the mobile phase
(Method II): 0.65 mL/min, gradient of the mobile phase: 0-0.15 min: 98%
solvent A, 2% solvent B (isocratic); 0.15-1.5 min: linear gradient from 2%
to 20% solvent B; 1.5-1.8 min: linear gradient from 20% to 50% solvent
B; 1.8-2.1 min: linear gradient from 50% to 100% solvent B; 2.1-2.2 min:
100% solvent B (isocratic); 2.2-2.6 min: 98% solvent A, 2% solvent B
(isocratic), 2.6-3.0 min: further system equilibration (98% solvent A, 2%
solvent B). Samples were analyzed using a full scan method and MRM
method. MRM methods were based on a tuning file and prepared for
each of the investigated products separately. Product quantification was
based on calibration curves prepared using reference compounds 1-5.
The calibration curves were prepared by plotting the response measured
with UPLC-TQD-MS in the MRM mode against the injected product
concentration. The experiments were performed in duplo, and the
reported reaction yields are the average of these two experiments. Based
on the magnitude of the calculated standard deviation (SD), reporting two
significant digits in the reaction yields was considered statistically justified.
Data collection and analysis were performed using Waters MassLynx
software and Microsoft Excel.
This study demonstrates that, in target-guided research,
moving from a 2CR to a 3CR is associated with increased
experimental complexity. Nonetheless, there might be niches of
reactivity within the MCR chemistry that are more amenable to
typical experimental conditions of TGS. Finally, the strategy of
stabilizing an imine intermediate in the form of a reversible
adduct, as described in this report, can be further implemented
while developing other MCRs on-target.
General protocol for TGS via the GBB reaction. In a 0.5 mL
Eppendorf tube were combined uPA solution in 0.2 M MES buffer at pH
5.5 (150 μL), isocyanide 7 solution (10 mM in MES buffer, 20 μL),
aminopyridine compound solution (20 mM in DMSO, 10 μL), and
aldehyde solution (10 mM in MES buffer, 20 μL). The final concentration
of uPA in the reaction mixture was approx. 21 μM. The resulting mixture
was shaken at rt for at least 48 h, and was monitored at time 0 h, 24 h,
48 h, and in case of a limited number of experiments after 1 week. When
Experimental Section
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
0.2 M HEPES buffer of pH 7.0 was used as reaction medium, higher
reagent concentrations were used. In this case, the reaction mixture
comprised uPA solution in 0.2 M HEPES buffer at pH 7.0 (150 μL),
isocyanide 7 solution (25 mM in HEPES buffer, 20 μL), aminopyridine
compound solution (50 mM in DMSO, 10 μL), and aldehyde solution (25
mM in HEPES buffer, 20 μL). Except reagent concentrations, all other
experimental conditions were as before.
Keywords: target-guided synthesis • multicomponent reactions •
imine adducts • urokinase • inhibitors
[1]
[2]
[3]
D. Bosc, J. Jakhlal, B. Deprez, R. Deprez-Poulain, Future Med. Chem.
2016, 8, 381-404.
M. Jaegle, E. L. Wong, C. Tauber, E. Nawrotzky, C. Arkona, J.
Rademann, Angew. Chem. Int. Ed. 2017, 56, 7358-7378.
E. Oueis, C. Sabot, P.-Y. Renard, Chem. Commun. 2015, 51, 12158-
12169.
General protocol for the pseudo two-component reaction on-target.
All imine adduct solutions were freshly prepared prior to use. In a 0.5 mL
Eppendorf tube, were combined uPA solution in 0.2 M MES buffer at pH
5.5 (170 μL), isocyanide 7 solution (10 mM in MES buffer, 20 μL), and
imine adduct solution (20 mM in DMSO, 10 μL). The final concentration
of uPA in the reaction mixture was approx. 21 μM. The resulting mixture
was shaken at rt for 48 h, and was monitored at time 0 h, 24 h, 48 h, and
in case of a limited number of experiments after 1 week. When 0.2 M
HEPES buffer of pH 7.0 was used as reaction medium, higher reagent
concentrations were used. In this case, the reaction mixture comprised
uPA solution in 0.2 M HEPES buffer at pH 7.0 (170 μL), isocyanide 7
solution (25 mM in HEPES buffer, 20 μL), and imine adduct solution (50
mM in DMSO, 10 μL). Except reagent concentrations, all other
experimental conditions were as before.
[4]
[5]
M. Mondal, A. K. Hirsch, Chem. Soc. Rev. 2015, 44, 2455-2488.
K. B. Sharpless, R. Manetsch, Expert Opin. Drug Discov. 2006, 1, 525-
538.
[6]
[7]
M. Whiting, J. Muldoon, Y.-C. Lin, S. M. Silverman, W. Lindstrom, A. J.
Olson, H. C. Kolb, M. G. Finn, K. B. Sharpless, J. H. Elder, V. V. Fokin,
Angew. Chem. Int. Ed. 2006, 45, 1435-1439.
W. G. Lewis, L. G. Green, F. Grynszpan, Z. Radić, P. R. Carlier, P.
Taylor, M. G. Finn, K. B. Sharpless, Angew. Chem. Int. Ed. 2002, 41,
1053-1057.
[8]
[9]
J. D. Cheeseman, A. D. Corbett, J. L. Gleason, R. J. Kazlauskas, Chem.
Eur. J. 2005, 11, 1708-1716.
A. Krasiński, Z. Radić, R. Manetsch, J. Raushel, P. Taylor, K. B.
Sharpless, H. C. Kolb, J. Am. Chem. Soc. 2005, 127, 6686-6692.
[10] R. Deprez-Poulain, N. Hennuyer, D. Bosc, W. G. Liang, E. Enée, X.
Marechal, J. Charton, J. Totobenazara, G. Berte, J. Jahklal, T. Verdelet,
J. Dumont, S. Dassonneville, E. Woitrain, M. Gauriot, C. Paquet, I.
Duplan, P. Hermant, F.-X. Cantrelle, E. Sevin, M. Culot, V. Landry, A.
Herledan, C. Piveteau, G. Lippens, F. Leroux, W.-J. Tang, P. van
Endert, B. Staels, B. Deprez, Nat. Commun. 2015, 6, 8250.
[11] X. Hu, J. Sun, H. G. Wang, R. Manetsch, J. Am. Chem. Soc. 2008, 130,
13820-13821.
General protocol for a competition experiment using a mixture of
two imine adducts. All imine adduct solutions were freshly prepared
prior to use. In a 0.5 mL Eppendorf tube, were combined uPA solution in
0.2 M MES buffer at pH 5.5 (170 μL), isocyanide 7 solution (10 mM in
MES buffer, 20 μL), and a solution containing two imine adducts (20 mM
of each adduct in DMSO, 10 μL). The final concentration of uPA in the
reaction mixture was approx. 21 μM. The resulting mixture was shaken at
rt for 48 h, and was monitored at time 0 h, 24 h, 48 h. When 0.2 M
HEPES buffer of pH 7.0 was used as reaction medium, higher reagent
concentrations were used. In this case, the reaction mixture comprised
uPA solution in 0.2 M HEPES buffer at pH 7.0 (170 μL), isocyanide 7
solution (25 mM in HEPES buffer, 20 μL), and a solution containing two
imine adducts (50 mM of each adduct in DMSO, 10 μL). Except reagent
concentrations, all other experimental conditions were as before.
[12] S. S. Kulkarni, X. Hu, D. Kenichiro, H.-G. Wang, R. Manetsch, ACS
Chem. Biol. 2011, 6, 724-732.
[13] M. Jaegle, T. Steinmetzer, J. Rademann, Angew. Chem. Int. Ed. Engl.
2017, 56, 3718-3722.
[14] R. Nguyen, I. Huc, Angew. Chem. Int. Ed. 2001, 40, 1774–1776.
[15] T. Maki, A. Kawamura, N. Kato, J. Ohkanda, Protein. Mol. BioSyst.
2013, 9, 940-943.
[16] F. E. Kwarcinski, M. E. Steffey, C. C. Fox, M. B. Soellner, ACS Med.
Chem. Lett. 2015, 6, 898-901.
Molecular Modeling. Molecular modeling was performed with help of
suitable software.[33] Molecular visualization was performed with
Molecular Operating Environment (MOE), 2018. Structure preparation of
the protein was performed using the system prepare module in MOE.
Docking was performed with FRED. Marvin was used for drawing and
displaying chemical structures. The crystal structure of uPA was taken
from the protein data bank; PDB code: 4MNW.[34] The enzyme was
complexed with the ligand bicyclic peptide UK749.[35] This ligand was
removed prior to docking. The enzyme structure was kept fixed.
[17] E. Oueis, F. Nachon, C. Sabot, P.-Y. Renard, Chem. Commun. 2014,
50, 2043-2045.
[18] R. Manetsch, A. Krasiński, Z. Radić, J. Raushel, P. Taylor, K. B.
Sharpless, H. C. Kolb, J. Am. Chem. Soc. 2004, 126, 12809-12818.
[19] V. P. Mocharla, B. Colasson, L. V. Lee, S. Röper, K. B. Sharpless, C. H.
Wong, H. C. Kolb, Angew. Chem. 2005, 117, 118-122.
[20] W. Tieu, T. P. Soares da Costa, M. Y. Yap, K. L. Keeling, M. C. J. Wilce,
J. C. Wallace, G. W. Booker, S. W. Polyak, A. D. Abell, Chem. Sci.
2013, 4, 3533-3537.
[21] V. P. Mocharla, J. C. Walsh, H. C. Padgett, H. Su, B. Fueger, W. A.
Weber, J. Czernin, H. C. Kolb, ChemMedChem 2013, 8, 43-48.
[22] A. Dömling, W. Wang, K. Wang, Chem. Rev. 2012, 112, 3083-3135.
[23] L. Weber, Drug Discov. Today Technol. 2004, 1, 261-267.
[24] E. L. Wong, E. Nawrotzky, C. Arkona, B. G. Kim, S. Beligny, X. Wang,
S. Wagner, M. Lisurek, D. Carstanjen, J. Rademann, Nat. Commun.
2019, 10, 66.
Supporting Information (see footnote on the first page of this article).
Detailed experimental procedures and characterization data for the
preparation of all reported compounds, as well as experimental data
concerning TGS experiments.
Acknowledgements
[25] N. Devi, R. K. Rawal, V. Singh, Tetrahedron 2015, 71, 183-232.
[26] K. Groebke, L. Weber, F. Mehlin, Synlett 1998, 661-663.
[27] C. Blackburn, B. Guan, P. Fleming, K. Shiosaki, S. Tsai, Tetrahedron
Lett. 1998, 39, 3635-3638.
R.G. is thankful to the Research Foundation – Flanders (FWO)
for the postdoctoral research grant. This work received support
from the University Research Fund (BOF-KP). The Laboratory of
Medicinal Chemistry is a partner of the Antwerp Drug Discovery
Network (www.addn.be).
[28] H. Bienaymꢀ, K. Bouzid, Angew. Chem. Int. Ed. 1998, 37, 2234-2237.
[29] K. Dass, A. Ahmad, A. S. Azmi, S. H. Sarkar, F. H. Sarkar, Cancer
Treat. Rev. 2008, 34, 122-136.
[30] A. H. Mekkawy, M. H. Pourgholami, D. L. Morris, Med. Res. Rev. 2014,
34, 918-956.
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
[31] S. Ulisse, E. Baldini, S. Sorrenti, M. D'Armiento, Curr. Cancer Drug
Targets 2009, 9, 32-71.
[48] ExPASy ProtParam tool; https://web.expasy.org/protparam/; provided
sequence: mature human urokinase (accessed january 30, 2019)
[49] L. Di, E. H. Kerns, Drug Discov. Today 2006, 11, 446-451.
[50] C. Peruzzotti, S. Borrelli, M. Ventura, R. Pantano, G. Fumagalli, M. S.
Christodoulou, D. Monticelli, M. Luzzani, A. L. Fallacara, C. Tintori, M.
Botta, D. Passarella, ACS Med. Chem. Lett. 2013, 4, 274-277.
[51] K. K. Murray, R. K. Boyd, M. N. Eberlin, G. J. Langley, L. Li, Y. Naito,
Pure Appl. Chem., 2013, 85, 1515-1609.
[32] R. Gladysz, Y. Adriaenssens, H. De Winter, J. Joossens, A.-M. Lambeir,
K. Augustyns, P. Van der Veken, J. Med. Chem. 2015, 58, 9238-9257.
[33] Software used: (a) Molecular Operating Environment (MOE), 2018;
Chemical Computing Group ULC, 1010 Sherbooke St. West, Suite
#910, Montreal, QC, Canada, H3A 2R7, 2018. (b) FRED; M. McGann, J.
Chem. Inf. Model. 2011, 51, 578-596. (c) Marvin; Marvin 15.10.26,
ChemAxon; http://www.chemaxon.com (accessed january 10, 2019).
[34] Crystal structure of uPA: PDB accession code 4MNW;
https://www.rcsb.org/structure/4mnw (accessed january 31, 2019).
[35] S. Chen, D. Bertoldo, A. Angelini, F. Pojer, C. Heinis, Angew. Chem. Int.
Ed. Engl. 2014, 53, 1602-1606.
[52] M. Mondal, M. Y. Unver, A. Pal, M. Bakker, S. P. Berrier, A. K. H.
Hirsch, Chem. Eur. J. 2016, 22, 14826-14830.
[53] A. K. Yocum, A. M. Chinnaiyan, Brief Funct. Genomic. Proteomic. 2009,
8, 145-157.
[54] J. J. Maury, D. Ng, X. Bi, M. Bardor, A. B. Choo, Anal Chem. 2014, 86,
395-402.
[36] A. R. Griswold, P. Cifani, S. D. Rao, A. J. Axelrod, M. M. Miele, R. C.
Hendrickson, A. Kentsis, D. A. Bachovchin, Cell Chem. Biol. 2019, 26.
1-7.
[55] A. R. Katritzky, L. Urogdi, J. Chem. Soc., Perkin Trans. 1, 1990, 0,
1853-1857.
[37] H. Stöckmann, A. A. Neves, S. Stairs, K. M. Brindle, F. J. Leeper, Org.
Biomol. Chem. 2011, 9, 7303-7305.
[56] A. R. Katritzky, M. Sutharchanadevi, L. Urogdi , J. Chem. Soc., Perkin
Trans. 1, 1990, 0, 1847-1851.
[38] H. M. Walborsky, G. E. Niznik, J. Org. Chem. 1972, 37, 187.
[39] Actosolv® (urokinase 600 000 IU, powder for solution for injection),
Eumedica Pharmaceuticals. Enzyme purification was performed using
a desalting column PD MiniTrap™ G-25, GE Healthcare.
[40] M. C. Pirrung, K. Das Sarma, J. Am. Chem. Soc. 2004, 126, 444-445.
[41] A. Dömling, Chem. Rev. 2006, 106, 17-89.
[57] A. R. Katritzky, B. Pilarski, L. Urogdi, Organic Preparations and
Procedures Int., 1989, 21, 135-139.
[58] A. R. Katritzky, S. Rachwal, B. Rachwal, J. Chem. Soc., Perkin Trans. 1,
1987, 0, 799-804.
[59] H.-J. Grumbach, B. Merla, N. Risch, Synthesis 1999, 1027-1033.
[60] J.-L. Moutou, M. Schmitt, V. Collot, J.-J. Bourguignon, Tetrahedron
Letters 1996, 37, 1787-1790.
[42] S. Marcaccini, T. Torroba, Nature Protocols, 2007, 2, 632-639.
[43] Y. Yabushita, Biotechnol. Appl. Biochem. 1988, 10, 294-300.
[44] J. B. Moffat, Catalysis Reviews: Science and Engineering, 1978, 18,
199-258.
[61] F. Yang, W. An, Z. Qian, T. Yu, Y. Du, L. Ma, X. Wang, T. Meng, J.
Shen, RSC Adv. 2015, 5, 32015-32019.
[62] R. Reingruber, T. Baumann, S. Dahmen, S. Bräse, Adv. Synth. Catal.
2009, 351, 1019-1024.
[45] G. A. Souza, M. M. M. Peixoto, A. P. F. Santos, W. J. Baader,
ChemistrySelect 2016, 1, 2307-2315.
[63] M. Petrini, Chem. Rev., 2005, 105, 3949-3977.
[64] W. Li, Y. Lam, J. Comb. Chem. 2005, 7, 721-725.
[65] M. Dockal, D. C. Carter, F. Rüker, J. Biol. Chem., 1999, 274, 29303-
29310.
[46] Y. Shimada, S. Kondo, Y. Ohara, K. Matsumoto, T. Katsuki, SYNLETT
2007, 15, 2445-2447.
[47] N. E. Good, G. D. Winget, W. Winter, T. N. Connolly, S. Izawa, R. M.
Singh, Biochemistry 1966, 5, 467-477.
This article is protected by copyright. All rights reserved.

10.1002/chem.201901917
Chemistry - A European Journal
FULL PAPER
Entry for the Table of Contents (Please choose one layout)
Layout 1:
FULL PAPER
While TGS examples generally
involve 2CRs, we selected the
Groebke-Blackburn-Bienaymé
reaction as model 3CR towards
urokinase inhibitors. Multiple
experimental parameters were
optimized. Additionally, chemical
imine stabilization strategies were
explored. Since imines are also crucial
intermediates of other MCRs, this
work intends to be a reference for
future multicomponent-based TGS
studies.
Rafaela Gladysz, Johannes Vrijdag,
Dries Van Rompaey, Anne-Marie
Lambeir, Koen Augustyns, Hans De
Winter, Pieter Van der Veken*
Page No. – Page No.
Efforts towards an on-target version
of the Groebke-Blackburn-Bienaymé
(GBB) reaction for discovery of
druglike urokinase (uPA) inhibitors
Layout 2:
FULL PAPER
Author(s), Corresponding Author(s)*
((Insert TOC Graphic here; max. width: 11.5 cm; max. height: 2.5 cm))
Page No. – Page No.
Title
Text for Table of Contents
This article is protected by copyright. All rights reserved.