Development of a Synthetic Receptor for the Food Toxin Beauvericin: A Tale of Carbazole and Steroids

Article abstract of DOI:10.1021/acs.orglett.8b02630

The synthesis of the first synthetic receptor showing high affinity for the toxic ionophoric cyclodepsipeptide beauvericin is described. Binding results in a pronounced increase in fluorescence intensity of the receptor, while this increase is not observed for a very similar ionophore such as valinomycin. Experiments that shed light on the nature of this selectivity are discussed.

Full text of DOI:10.1021/acs.orglett.8b02630

Letter
pubs.acs.org/OrgLett
Cite This: Org. Lett. XXXX, XXX, XXX−XXX
Development of a Synthetic Receptor for the Food Toxin
Beauvericin: A Tale of Carbazole and Steroids
Vincent Ornelis,§ Andreja Rajkovic,† Benedikt Sas,‡ Sarah De Saeger,& and Annemieke Madder*,§
§Department of Organic Chemistry, Organic and Biomimetic Chemistry Research Group, Ghent University, Krijgslaan 281, Ghent,
Belgium
†Department of Food Safety and Food Quality, Laboratory of Food Microbiology and Food Preservation, Ghent University,
Coupure Links 653, Ghent, Belgium
‡Department of Food Safety and Food Quality, Food2Know, Ghent University, Coupure Links 653, Ghent, Belgium
&Department of Bioanalysis, Laboratory of Food Analysis, Ghent University,Ottergemsesteenweg 460, Ghent, Belgium
S
* Supporting Information
ABSTRACT: The synthesis of the first synthetic receptor showing
high affinity for the toxic ionophoric cyclodepsipeptide beauvericin is
described. Binding results in a pronounced increase in fluorescence
intensity of the receptor, while this increase is not observed for a very
similar ionophore such as valinomycin. Experiments that shed light on
the nature of this selectivity are discussed.
survival rate, illustrating the toxic effect of this ionophore.4
eauvericin (BEA) is a toxic cyclodepsipeptide produced
1
Bby the insect pathogenic fungus Beauveria bassiana or Moreover, when mice were treated intraperitoneally with 5 mg
BEA/kg body weight for three consecutive days, bioaccumu-
lation in the liver and lipophilic tissue was observed,5 probably
because of the hydrophobic nature of this toxin. It is this
combination of prevalence in food with toxic properties and
tendency to bioaccumulate that motivates the need to develop
efficient and rapid detection systems for beauvericin.
plant pathogenic Fussarium spp. (See Figure 1.) The latter
infects crops such as wheat and maize, which explains the high
concentrations (up to 520 mg/kg) of beauvericin that can be
found in infected kernels.2 This toxin is able to bind K+ and
NH4+ cations and can transport these ions across lipid bilayers.
By using membrane potential-sensitive fluorescent dyes,
Tonshin et al. showed that low micromolar concentrations of
beauvericin are sufficient to cause depolarization of human
neural (Paju) cells.3 Overnight incubation of human leukemia
cells with 3 μM beauvericin resulted in an 80% decrease in
Current methods rely on LC-MS(/MS),6 and even though
this technique provides low limits of detection/quantification,
important drawbacks remain that limit the straightforward
testing of large numbers of food samples for the presence of
the toxin.
The main limitation resides in the fact that LC-MS(/MS)
does not allow parallel measurements of many samples
simultaneously, nor can it be used on site (in the field or in
food processing facilities) where fungal growth and concom-
itant toxin production occurs. Such limitations have been
circumvented by the development of enzyme-linked immuno-
sorbent assays (ELISA) and lateral flow devices,7 which are
commercially available for the more well-known mycotoxins,
such as aflatoxins, deoxynivalenol, and many others. However,
both methods require antibodies to recognize the mycotoxin
under investigation. While mycotoxins such as aflatoxins,
zearalenone, and deoxynivalenol contain functional groups that
allow conjugation to carrier proteins (which is necessary to
Figure 1. Chemical structure of beauvericin (left) and valinomycin
(right), along with the design principles of the artificial receptor
(middle).
Received: August 17, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.8b02630
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
Scheme 1. Synthetic Scheme toward Artificial Receptor 6, with Visual Representation of BEA (Green) Binding to 6 (Orange),
Using VMD
elicit an immune response), beauvericin does not. Given the
difficulty to generate antibodies for beauvericin, and in view of
the inherent limitations of antibodies including limited long-
term stability, short shelf life, and cost, we became interested in
the development of a synthetic receptor capable of selectively
binding this toxin.
A wealth of artificial receptors has been developed for a
broad range of ligands such as anions,8 cations, saccharides9
and small linear dipeptides and tripeptides,10 to name just a
few. In the context of peptide ligand binding, association
constants in the range of 103−106 M−1 have been
reported.11−13 However, cyclic medium-sized depsipeptides
such as beauvericin have remained unexplored as ligands for
synthetic hosts. In terms of ligand scope, the work presented
herein is a first step toward filling this gap.
Our receptor design (see Figure 1) is based on the bridging
of two steroid arms via a linker to create a hydrophobic cavity
in an effort to accommodate the lipophilic surface of
beauvericin. The use of steroidal bile acid derivatives was
inspired by reports in the literature describing their use in
artificial receptors,14,15 as well as some of our previous
work.16,17 Incorporation of an easily protonated primary
amine functionality in the linking unit between the steroid
moieties should further promote binding, because of the
ionophoric nature of the toxin. Initially, a construct was
investigated in which the steroid arms were connected through
a flexible linker and further decorated with peptide chains
providing additional diversification points (see Figure S1 in the
Supporting Information (SI)). However, next to potential
scaleup being hampered by the required solid-phase peptide
synthesis, determining the binding affinity for beauvericin has
remained difficult without introducing adequate labels in either
of the two partners. Therefore, in the current optimized design,
shown in Figure 1, a carbazole moiety was introduced as
linking fragment between the steroid arms to ensure an
adequate level of rigidity, while, at the same time, equipping
the receptor with fluorescent properties. Although carbazoles
have been used in artificial receptors for caffeine18 and iodine19
among others, our design is the first example that combines
both bile acid derivatives and this heterocycle within one
receptor entity. In order to prevent the hydrophobic cavity
from being too small to accommodate the toxin, the envisaged
complex was visualized by taking the van der Waals volume of
both binding partners into account, using visual molecular
dynamics (VMD) (Scheme 1).20 This allowed to determine
the minimum length of the linkers connecting the carbazole
with the bile acid moieties.
The convergent synthesis of the receptor (Scheme 1)
involves two fragments: a steroid fragment 1 and a carbazole
fragment 2, which are joined together via a copper-catalyzed
alkyne azide cycloaddition (CuAAC) to yield the bipodal
receptor 6 after some manipulations. For the steroid fragment,
we started from commercially available lithocholic acid 3.
Reduction of the carboxylic acid and selective dimethoxytrityl
(DMT) protection of the primary alcohol in the presence of
the secondary alcohol yielded DMT-protected steroid
derivative 4, with a yield of 54% over three steps. Next,
acylation with bromoacetyl bromide, followed by a SN2
reaction with NaN3, resulted in the formation of the final
steroid fragment 1.
For the synthesis of the carbazole fragment, N-alkylation of
2,7-dibromocarbazole allowed the introduction of the required
primary amine, needed for complexation within the beauver-
icin interior. This was followed by a Sonogashira reaction and
alkyne-deprotection using NaH in toluene,21 which completed
the synthesis of compound 2.
With these two building blocks, we could proceed to the
construction of the bipodal receptor using the CuAAC
reaction. While typically this “click” reaction is very efficient,
no reaction occurred at room temperature using standard
Fokin−Sharpless conditions. By heating the mixture at 60 °C
overnight, the desired CuAAC product could be obtained. The
1,5-regioisomer that can potentially be formed as a result of a
Huisgen rather than copper-catalyzed pathway was not
observed. In principle, at this stage, deprotection of both
DMT- and Boc-group would yield a functional receptor.
However, the limited solubility of the resulting construct in
DMF alone, severely limits its future application. Therefore,
B
DOI: 10.1021/acs.orglett.8b02630
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
the free alcohols were acylated with bromoacetyl bromide to
render the receptor more hydrophobic. Subsequently,
following Boc-deprotection with 4 M HCl in dry dioxane, we
were glad to observe that the final receptor 6 was indeed
soluble in CHCl3. Note that the introduced α-bromoacetyl
moiety potentially allows for further derivatization (e.g.,
immobilization on solid support or labeling).
2:1 binding model confirms that the observed binding is best
explained based on a 1:1 stoichiometry, which is further
consistent with the proposed binding mode (recall the space-
filling model in Scheme 1).
As expected, when the amine is still Boc-protected no
change in fluorescence is observed (data not shown). More
importantly when only the carbazole fragment is tested without
the steroid arms, a much weaker association (factor 103
smaller) is observed (Figure S6 and S7 in the SI). This
decrease in binding strength clearly emphasizes the role of the
hydrophobic cavity generated by the steroid arms.
We then set out to determine the new receptor’s affinity for
beauvericin. It is known that fluorescence is highly dependent
on the molecular environment. Therefore, we reasoned that
binding of beauvericin in close proximity to the fluorescent
carbazole, would influence the local environment, thus
permitting the use of fluorescence titration for the determi-
nation of binding affinity. Gratifyingly, upon addition of
beauvericin to receptor 6 a clear concentration-dependent
increase in fluorescence is observed at ∼373 and 389 nm
(Figure 2), which corresponds to the typical blue carbazole
emission. When this increase in fluorescence intensity is
plotted versus the added ligand concentration, a binding curve
can be obtained. As shown in the inset of Figure 2, this curve
can be fitted to a 1:1 stoichiometry resulting in a mean
To assess the degree of selectivity, the capacity of receptor 6
to bind valinomycin (see Figure 1) was also tested. This
bacterial toxin is produced by Streptomyces spp.24 and, like
beauvericin, is an ionophoric cyclodepsipeptide with selectivity
for K+ and NH4 ions. The main difference between the two
+
toxins resides in their molecular weight and size (783 Da for
beauvericin and 1111 Da for valinomycin). Interestingly, even
though both ionophores are very similar, when valinomycin
was added to receptor 6, a negligible increase in fluorescence
occurred (see Figure 3, left).
association constant of 4.0 × 105 M−1
2.1 × 104 (Ka
standard deviation (SD)) (see also Figure S3 in the Supporting
Information). When performing the titration at a lower
receptor concentration (1 μM), a 2.6-fold higher association
constant was obtained, which seems to point toward a mass
balance limitation. As the limited fluorescence of carbazole
does not allow the reliable determination of Ka at lower
receptor concentrations, the value reported here for the
binding affinity should be considered as the minimal value,
which can be determined using fluorescence.
Figure 3. Fluorescence emission spectrum upon addition of 5 equiv
beauvericin or valinomycin to 2.5 μM receptor 6 using λex = 329 nm
(left) or to 10 μM of the carbazole fragment using λex = 290 nm
(right) in CHCl3.
Intrigued by this observation, we became interested in the
structural reason behind this selectivity. We originally
hypothesized that the cavity size was responsible for this
effect. In order to assess this hypothesis, we synthesized a
carbazole fragment that contains the amine linker but is devoid
of any hydrophobic cavity, thereby expecting that both toxins
would interact similarly. Surprisingly, even though beauvericin
can perfectly interact with this carbazole fragment, albeit with
lower affinity, valinomycin cannot (Figure 3, right). This seems
to indicate that the length of the chain connecting the primary
amine to the central carbazole moiety plays an important role
in the selectivity. This observation might be explained by
taking the size of the toxins into consideration. The short
amine linker forces the toxin in close proximity of the
carbazole fragment. We reason that valinomycin, being much
larger than BEA, is experiencing steric hindrance due to the
carbazole moiety, while this is not the case for the smaller
beauvericin.
Figure 2. Fluorescence titration of 0 to 8 equiv beauvericin (BEA) to
2.5 μM receptor 6 in CHCl3 using λex = 329 nm. The inset shows the
mean fluorescent intensity of 6 at 389 nm in function of added BEA
concentration. The mean and standard error at each titration point
were calculated based on three independent titrations. Data points
were fitted to a 1:1 stoichiometry (red line).
Even though the Job plot has been the golden standard in
terms of determining binding stoichiometry, multiple authors
have raised concerns and criticized its use in supramolecular
chemistry.22,23 Therefore, it has been recommended to fit the
data to different binding models and to assess which model is
most suitable for explaining the experimental data.22,23
When trying to fit the above titration points to a 1:2 or 2:1
host−guest stoichiometry (see Figures S4 and S5 in the SI), no
proper fit could be obtained for the individual titrations. The
fact that a good mathematical fit can be found for a 1:1 binding
model and no reasonable fit is achievable using either a 1:2 or
Finally, we investigated if the fluorescence sensing
capabilities of receptor 6 toward BEA would still stand in
more-complex solutions. To this end, 4 mL human urine was
extracted with 3 mL CHCl3. The organic phase was isolated
and the fluorescence emission was measured. This was
followed by the subsequent addition of 10 μM receptor 6
C
DOI: 10.1021/acs.orglett.8b02630
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
Author Contributions
and 2 equiv BEA. As can be seen in Figure 4, the receptor is
still able to sense the toxin, even in the chloroform extract of
urine. A similar experiment was performed using urine that was
spiked with BEA (see the SI).
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors would like to thank BOF-Ghent University for
GOA funding. V.O. is grateful to the Fonds voor
Wetenschappelijk Onderzoek for a VLAIO research grant.
REFERENCES
■
(1) Hamill, R. L.; Higgens, C. E.; Boaz, H. E.; Gorman, M.
Tetrahedron Lett. 1969, 10, 4255.
(2) Ritieni, A.; Moretti, A.; Logrieco, A.; Bottalico, A.; Randazzo, G.;
Monti, S. M.; Ferracane, R.; Fogliano, V. J. Agric. Food Chem. 1997,
45, 4011.
(3) Tonshin, A. A.; Teplova, V. V.; Andersson, M. A.; Salkinoja-
Salonen, M. S. Toxicology 2010, 276, 49.
(4) Jow, G.-M.; Chou, C.-J.; Chen, B.-F.; Tsai, J.-H. Cancer Lett.
2004, 216, 165.
̈
(5) Rodriguez-Carrasco, Y.; Heilos, D.; Richter, L.; Sussmuth, R. D.;
Heffeter, P.; Sulyok, M.; Kenner, L.; Berger, W.; Dornetshuber-Fleiss,
R. Toxicol. Lett. 2016, 247, 35.
Figure 4. Fluorescence emission spectrum of a CHCl3 extract from
urine (dotted line) and after addition of 10 μM receptor 6 (dashed
line) or both 10 μM receptor 6 and 2 equiv BEA (full line), using λex
= 329 nm.
(6) Santini, A.; Ferracane, R.; Meca, G.; Ritieni, A. Anal. Bioanal.
Chem. 2009, 395, 1253.
(7) Maragos, C. M. World Mycotoxin J. 2009, 2, 221.
(8) Beer, P. D.; Gale, P. A. Angew. Chem., Int. Ed. 2001, 40, 486.
(9) Mazik, M. Chem. Soc. Rev. 2009, 38, 935.
(10) Schmuck, C.; Wich, P. Angew. Chem., Int. Ed. 2006, 45, 4277.
(11) Bernard, J.; Wennemers, H. Org. Lett. 2007, 9, 4283.
(12) Chang, K.-H.; Liao, J.-H.; Chen, C.-T.; Mehta, B. K.; Chou, P.-
T.; Fang, J.-M. J. Org. Chem. 2005, 70, 2026.
(13) Henley, P. D.; Waymark, C. P.; Gillies, I.; Kilburn, J. D. J. Chem.
Soc., Perkin Trans. 2000, 1, 1021.
(14) Deng, L.-Q.; Lu, Y.-M.; Zhou, C.-Q.; Chen, J.-X.; Wang, B.;
Chen, W.-H. Bioorg. Med. Chem. Lett. 2014, 24, 2859.
(15) Erzunov, D. A.; Latyshev, G. V.; Averin, A. D.; Beletskaya, I. P.;
Lukashev, N. V. Eur. J. Org. Chem. 2015, 2015, 6289.
(16) Ruiz Garcia, Y.; Pabon-Martinez, V.; Smith, C. I. E.; Madder, A.
Chem. Commun. 2017, 53, 6653.
To conclude, we have described the first synthetic receptor
for the ionophore beauvericin with an association constant of
4.0 × 105 M−1 in CHCl3. This receptor shows a pronounced
increase in fluorescence intensity upon addition of the toxin,
thus allowing easy detection of binding in solution without
immobilization of either one of the binding partners.
Interestingly, when a very similar ionophore such as
valinomycin is added, no change in fluorescence is observed,
providing evidence for the selectivity of our receptor.
Preliminary experiments seem to indicate that not only the
cavity size, but also the length of the amine linker, is
responsible for this selective behavior. Artificial receptors have
been described for short linear peptides consisting of a
maximum of 3−4 natural amino acids (vide supra). The
receptor developed here binds the 6-mer cyclic depsipeptide
BEA with an affinity that is at least similar or slightly higher
than those previously reported for receptors binding small
peptides, making it the first synthetic receptor for beauvericin
reported to date.
́
(17) Bode, C. A.; Bechet, T.; Prodhomme, E.; Gheysen, K.; Gregoir,
P.; Martins, J. C.; Muller, C. P.; Madder, A. Org. Biomol. Chem. 2009,
7, 3391.
(18) Mahapatra, A. K.; Roy, J.; Sahoo, P.; Mukhopadhyay, S. K.;
Mukhopadhyay, A. R.; Mandal, D. Bioorg. Med. Chem. Lett. 2012, 22,
5379.
(19) Zhu, H.; Shi, B.; Chen, K.; Wei, P.; Xia, D.; Mondal, J. H.;
Huang, F. Org. Lett. 2016, 18, 5054.
(20) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graphics 1996,
14, 33.
(21) Havens, S. J.; Hergenrother, P. M. J. Org. Chem. 1985, 50, 1763.
(22) Ulatowski, F.; Dabrowa, K.; Balakier, T.; Jurczak, J. J. Org.
Chem. 2016, 81, 1746.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.or-
glett.8b02630.
Detailed synthetic procedures, spectroscopic character-
ization, fluorescence measurements, equations used for
Ka determination and experiments with urine samples
(PDF)
(23) Hibbert, D. B.; Thordarson, P. Chem. Commun. 2016, 52,
12792.
(24) Matter, A. M.; Hoot, S. B.; Anderson, P. D.; Neves, S. S.;
Cheng, Y. PLoS One 2009, 4, e7194.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: Annemieke.Madder@Ugent.be.
ORCID
Sarah De Saeger: 0000-0002-2160-7253
Annemieke Madder: 0000-0003-0179-7608
D
DOI: 10.1021/acs.orglett.8b02630
Org. Lett. XXXX, XXX, XXX−XXX