Dynamic combinatorial chemistry employing boronic acids/boronate esters leads to potent oxygenase inhibitors

Article abstract of DOI:10.1002/anie.201202000

Dynamic duo: The reversible reaction of boronic acids with alcohols to form boronate esters, coupled to protein mass spectrometry analyses, was used to discover potent oxygenase inhibitors. This dynamic combinatorial mass spectrometry technique could potentially be applied to the identification of other protein inhibitors. Copyright

Full text of DOI:10.1002/anie.201202000

.
Angewandte
Communications
DOI: 10.1002/anie.201202000
Boron Chemistry
Dynamic Combinatorial Chemistry Employing Boronic Acids/
Boronate Esters Leads to Potent Oxygenase Inhibitors**
Marina Demetriades, Ivanhoe K. H. Leung, Rasheduzzaman Chowdhury, Mun Chiang Chan,
Michael A. McDonough, Kar Kheng Yeoh, Ya-Min Tian, Timothy D. W. Claridge,
Peter J. Ratcliffe, Esther C. Y. Woon,* and Christopher J. Schofield*
The application of dynamic reactions is a promising approach
for the discovery of small-molecule ligands for proteins. To
date, however, this method is limited by the few appropriate
reactions and the techniques used for the analysis of protein–
ligand complexes.[1] “Dynamic” functional group intercon-
vertions that have been employed include the conversion of
thiols to disulfides, the aldol reaction, and the addition of
nucleophiles to ketones and aldehydes.[2] The reaction of
boronic acids with diols to form boronate esters is attractive
for dynamic-library formation, because it is reversible in
aqueous solution in a pH-dependent manner.[3] The dynamic
boronic acid/boronate ester system has been used to form
supramolecular switches, some of which have been used for
sugar detection.[4,5] However, this system has not been used
for the identification of protein ligands. Proof of principle
work with proteases, which react reversibly with boronic
acids, suggests that boronic acid/boronate ester systems might
be useful for the identification of enzyme inhibitors.[6]
One issue with the application of reversible reactions for
ligand identification is the need to analyze labile complexes
that are derived from mixtures. High-resolution techniques,
such as NMR spectroscopy and X-ray crystallography, are
applicable, but these are time-consuming.[7] Our research
group and that of Poulsen, have used non-denaturing protein
mass spectrometry to identify protein–ligand complexes
formed from equilibrating mixtures of thiols/disulfides[8–11]
and aldehydes/hydrazones.[7] The dynamic-combinatorial
mass spectrometry (DCMS) technique has the advantages
of being efficient and providing information on mass shifts,
which can be used for assigning structures to the ligands that
bind preferentially.
Herein we demonstrate that boronic acid/boronate ester
dynamic systems coupled with protein mass spectrometry
analysis are useful for the identification of protein inhibitors
(Scheme 1). Our target model enzyme was prolyl hydroxylase
domain isoform 2 (PHD2), which is a FeII and 2-oxoglutarate
(2OG) oxygenase that regulates the human hypoxic response.
PHD2 inhibition is of therapeutic interest for the treatment of
anemia and ischemia-related diseases.[12]
DCMS experiments were carried out using “support
ligands” 2 and 3 (Scheme 2), which were designed to
participate in FeII chelation in the active site and, through
the incorporation of a boronic acid moiety, participate in
boronate ester exchange. We selected the 2-(picolinamido)-
acetic acid scaffold because, based on crystal structures of
PHD2,[13] it is predicted to fit into the active site through its
chelation with FeII. The low potency of 2-(picolinamido)acetic
acid (IC50 > 1 mm) enabled the effect of boronate ester
substitution to be monitored.
Modeling studies suggested that whereas the boronic acid
group in support ligand 2 would fit into the active-site
subpocket, that of 3 would clash with the active-site wall.[14]
Hence, it was envisaged that the reactivity of 3 might serve as
a control to investigate possible non-specific binding. The
analysis of mixtures of 2 or 3 with PHD2·FeII through the use
of non-denaturing ESI-MS led to the observation of a new
peak at 27887 Da (187 Æ 2 Da shift), corresponding to a small
molecule/protein adduct, in which the OH groups of the
boronic acids moiety are cleaved. We have previously
observed, through the use of non-denaturing ESI-MS, anal-
ogous apparent fragmentation of boronic acids complexed
with other enzymes.[15] Notably, the mixture of boronate ester
4 and PHD2·FeII gave the same mass shift (187 Æ 2 Da) as that
observed with 2 and 3 at a cone voltage of 80 V.[14] However,
when a lower cone voltage was used (30 V), the mass shift
corresponding to an adduct of 4 with the protein, without
fragmentation, was apparent (358 Æ 2 Da), demonstrating
that boronate ester formation can be observed when suffi-
ciently mild ionization is used.
[*] M. Demetriades, I. K. H. Leung,[+] Dr. R. Chowdhury,[+] M. C. Chan,
Dr. M. A. McDonough, Dr. K. K. Yeoh, Dr. T. D. W. Claridge,
Prof. C. J. Schofield
Chemistry Research Laboratory, University of Oxford
12 Mansfield Road, Oxford, OX1 3TA (UK)
E-mail: christopher.schofield@chem.ox.ac.uk
Dr. Y. M. Tian, Prof. P. J. Ratcliffe
Nuffield Department of Clinical Medicine, University of Oxford
Roosevelt Drive, OX3 7BN (UK)
Dr. E. C. Y. Woon
Department of Pharmacy, National University of Singapore
18 Science Drive 4, 117543 Singapore (Singapore)
E-mail: phaewcy@nus.edu.sg
[+] These authors contributed equally to this work.
[**] We thank Dr. T. Brown Jr. for the initial work and Dr. R. J. Hopkinson
for discussions, Dr. M. K. Lee for providing a sample of HyAsn803
antibody, and the Universiti Sains Malaysia (fellowship to K.K.Y).
Both
2 and 3 compete with the 2OG analogue
Supporting information, including syntheses and characterizations,
MS methods, inhibition assays, protein purification, crystallization,
and structure solution methods for this article is available on the
WWW under http://dx.doi.org/10.1002/anie.201202000.
N-oxalylglycine (NOG) for the 2OG binding site of
PHD2.[14] To ensure that boronate ester formation involving
2 and 3 was favorable under the conditions used (NH4OAc
6672
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 6672 –6675

Angewandte
Chemie
Scheme 1. Schematic representation of the dynamic combinatorial mass spectrometry (DCMS) method that uses the conversion between boronic
acids and boronate esters as the reversible reaction.
We then investigated whether boronate esters could be
used in DCMS. Forty diols were separated into four sets (i–iv)
of ten diols[14] according to molecular weight (see the
Supporting Information for details). Each set was treated
with either support ligands, 2 or 3, in the presence of
PHD2·FeII. ESI-MS was then used to determine which of
the in situ formed boronate esters bind preferentially to
PHD2·FeII (Figure 1).
The reaction of support ligand 2 with the diol sets i–iv
resulted in the observation of peaks with mass shifts corre-
sponding to complexes of PHD2 with boronate esters of 2
(5, Scheme 2), which were derived from 7 diols (7–13,
Scheme 2). ESI-MS analysis of mixtures of 2 with the
individual diols validated the above DCMS results.[14] In
some cases, diols were observed to bind to PHD2·FeII in the
absence of 2, presumably through Fe chelation (Figure 1).[14]
In contrast to the results with 2, when 3 was employed, no
binding of boronate esters was observed in the corresponding
assay. Direct binding of some diols to PHD2·FeII was again
observed (Figure 1). The results imply that the mass shifts
observed with 2 represent protein adducts involving boronate
esters derived from support ligand 2 rather than adducts
derived from simultaneous binding of 2 and diols. In the case
of diol 8, which was observed to bind to PHD2·FeII both in the
absence of a support ligand (2 and 3) and in the form of the
boronate ester derived from 2, addition of the 2OG com-
petitor NOG to the PHD2/FeII/8 mixture resulted in separate
peaks representing the adducts derived from the binding of
PHD2·FeII with both 8 and NOG; no peak derived from
simultaneous binding of 8 and NOG was observed. These
observations support the hypothesis that boronate esters
derived from 2 are formed and bind to the protein and that, in
the absence of 2, diol 8 and also the diols 11, 13, and 34–38
(see the Supporting Information for details) bind to PHD2
through chelation with FeII.[14]
Scheme 2. Structures of selected compounds used in this work.
buffer, pH 7.5), the pKas of the boronic acids were measured
by NMR spectroscopy. The pKa values of 2 (approximately
The results of ESI-MS analyses are not always represen-
tative of what exists in solution,[16] therefore the DCMS
results were validated by solution studies. NMR-based water
[14]
7.4) and 3 (approximately 6.9)
are appropriate for
boronate ester formation under the incubation conditions.[14]
Angew. Chem. Int. Ed. 2012, 51, 6672 –6675
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6673

.
Angewandte
Communications
13.2 mm, 2/39: KD,app = 18.2 mm).[14] Hence there is qualitative
agreement between the gas-phase and solution studies.
To verify the potential of the boronic acid/boronate ester-
based DCMS for identification of PHD2 inhibitors, the
synthesis of stable analogues that mimic the structure of the
identified hits (boronate esters derived from 2 and diols 7–13)
from DCMS experiments was attempted. To prepare stable
analogues of the boronate esters, we made use of the boronic
acid functional group, which was also employed in the DCMS
chemistry, in Suzuki cross-coupling reactions.[14] Benzofuran-
and naphthalene-based analogues, 16 and 18, respectively,
were prepared; the benzofuran and naphthalene moieties
function as 6/5- and 6/6-fused bicyclic mimics of the catechol
boronate ester moiety in the boronate ester derived from 2
and catechol (6). The binding constants of 16 and 18 were
similar (16: KD = 9.5 mm, 18: KD = 7.0 mm) suggesting that
both ring systems are accommodated in the protein binding
pocket. The 5-methoxynaphthalene derivative 20 was used as
a mimic of the boronate esters derived from 4-substituted
catechols (7–11) and, notably, 20 (IC50 = 107 mm, KD = 1.6 mm)
showed greater inhibition and binding affinity than 16 and 18;
these results are consistent with the DCMS results. NMR
water relaxation measurements showed that all the stable
analogues bind significantly stronger to PHD2 than the
parent compound 2 (Table 1); however they were generally
poorer binders than the boronate esters identified by
DCMS.[14]
Analysis of PHD2 structures[14] suggests that hydrogen-
bonding interactions between the inhibitor and the side chain
Table 1: Binding constants (KD) as measured by NMR spectroscopy and
IC50 values for selected inhibitors.[14]
Figure 1. DCMS analyses of PHD2 with boronic acids 2/3 and diol sets
i–iv. Spectra of deconvoluted non-denaturing ESI-MS (cone voltage
30 V) showing: a) PHD2·FeII; b) PHD2·FeII with 2 (1:1); c–f) PHD2·FeII
with 2 and i–iv (1:1:1), respectively; g) PHD2·FeII with 3 (1:1),
h–k) PHD2·FeII with 3 and diols i–iv (1:1:1), respectively. Assignment
of labeled peaks: PHD2·Fe·2 (peak B); PHD2·Fe·3 (peak O); PHD2·Fe·-
diol (peaks C, E, H, I, L, M, P, and Q); PHD2·Fe·5 (peaks D, F, G, J, K,
and N).
Compound
R
X
KD [mm]
N/A
IC50 [mm]
>1000
>500
14
16
H
H
H
9.5
[17,18]
18
2
H
H
H
7.0
24.8
1.6
>500
126
relaxation experiments
were used for the measurement
of the apparent binding constants (KD,app) of 2 with PHD2, in
the presence of different diols. In this method, paramagnetic
MnII was used as a substitute for the diamagnetic FeII; KD,app
values were determined by monitoring the bulk water
relaxation rate, which decreases when water access to the
paramagnetic metal center is hindered through binding of
compounds to the metal. The support ligand 3 displayed no
significant PHD2 binding, as determined by NMR spectros-
copy.[14] When the NMR experiments were carried out using 2
in the absence or in the presence of diols (6–13), it was found
that its apparent affinity for PHD2 (KD,app = 24.8 mm) was
enhanced in the presence of the diols that had been identified
by DCMS as being able to form boronate esters (e.g. 2/12:
B(OH)2
20
107
22
H
8.7
>100
23
30
H
OH
OH
3.5
0.5
409
0.017
28
32
OH
OH
0.8
0.9
0.013
0.004
K
D,app = 3.0 mm, 2/10: KD,app = 1.3 mm, 2/11: KD,app = 0.6 mm).
The presence of diols that were not observed to form
boronate esters in the DCMS assays caused a much weaker
increase in the apparent binding affinity of 2 (2/6: KD,app
=
6674 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 6672 –6675

Angewandte
Chemie
hydroxy group of Tyr303 might promote inhibitor binding.[13]
Thus, a derivative of inhibitor 20 containing a hydroxy group
at the C3 position, 28, was prepared.[14] Compound 28 showed
significantly higher potency and affinity (IC50 = 13 nm, KD =
0.8 mm) than 20. As indicated by DCMS, the aryl substituent
at the 5-position is important: 28 showed higher potency than
23 (Table 1).[11] In the presence of the C3 hydroxy group, the
methoxy group on the naphthalene ring, which improved the
activity of the initial inhibitors, is less important, as shown by
the similar potency of 28 and 30 (Table 1). However,
substitution at the C5 position of the naphthalene moiety
remains important as indicated by the higher activity of 32
relative to 30.
To investigate the mode of binding of 28, it was crystalized
in complex with the PHD2 catalytic domain incorporating
MnII, which substitutes for FeII.
The crystal structure reveals that 28 complexes to the
metal in a bidentate manner, with the C3 hydroxy group
interacting with Tyr303 through a hydrogen bond. The
naphthalene group of 28 is twisted relative to the pyridinyl
ring (torsion angle 52.88), thus allowing the naphthalene
moiety to occupy the substrate-binding site (Figure 2).
other proteins. Suitably protected boronic acids were also
used for the synthesis of stable analogues through organo-
metallic chemistry. We have found that there are potential
limitations of the DCMS approach, for example, boronic acids
can sometimes fragment under the conditions of MS. How-
ever, we foresee that improved MS screening methods that
give results that are more representative of the composition of
the analyte in solution will become available as the technol-
ogy develops. We have also shown that NMR-based screening
methods, and other types of analysis, can also be employed in
the evaluation of boronate ester binding.
Received: March 13, 2012
Published online: May 25, 2012
Keywords: boron · combinatorial chemistry ·
.
mass spectrometry · oxygenase · prolyl hydroxylase
[1] O. Ramstrom, J. M. Lehn, Nat. Rev. Drug Discovery 2002, 1, 26.
[2] P. T. Corbett, J. Leclaire, L. Vial, K. R. West, J. L. Wietor,
J. K. M. Sanders, S. Otto, Chem. Rev. 2006, 106, 3652.
[3] G. Springsteen, B. H. Wang, Tetrahedron 2002, 58, 5291.
[4] H. G. Kuivila, A. H. Keough, E. J. Soboczenski, J. Org. Chem.
1954, 19, 780.
[5] J. Z. Zhao, T. M. Fyles, T. D. James, Angew. Chem. 2004, 116,
3543; Angew. Chem. Int. Ed. 2004, 43, 3461.
[6] I. K. H. Leung, T. Brown, C. J. Schofield, T. D. W. Claridge, Med.
Chem. Commun. 2011, 2, 390.
[7] S. A. Poulsen, J. Am. Soc. Mass Spectrom. 2006, 17, 1074.
[8] B. M. R. Liꢀnard, N. Selevsek, N. J. Oldham, C. J. Schofield,
ChemMedChem 2007, 2, 175.
[9] N. R. Rose, E. C. Y. Woon, G. L. Kingham, O. N. F. King, J.
Mecinovic, I. J. Clifton, S. S. Ng, J. Talib-Hardy, U. Oppermann,
M. A. McDonough, C. J. Schofield, J. Med. Chem. 2010, 53, 1810.
[10] B. M. R. Liꢀnard, R. Hueting, P. Lassaux, M. Galleni, J. M.
Frere, C. J. Schofield, J. Med. Chem. 2008, 51, 684.
[11] E. C. Y. Woon, M. Demetriades, E. A. L. Bagg, W. S. Aik, S. M.
Krylova, J. H. Y. Ma, M. C. Chan, L. J. Walport, D. Wegman,
K. N. Dack, M. A. McDonough, S. N. Krylov, C. J. Schofield, J.
Med. Chem. 2012, accepted, DOI: 10.1021/jm201417e.
[12] N. R. Rose, M. A. McDonough, O. N. F. King, A. Kawamura,
C. J. Schofield, Chem. Soc. Rev. 2011, 40, 4364.
Figure 2. View from a crystal structure of PHD2·MnII (light blue)
complexed with 28 (yellow) superimposed with a structure of
PHD2·MnII in complex with HIF-1a residues 558-574 (“CODD”: blue
surface; PDB ID 3HQR). The image shows the proximity of the
C5 substituent of 28 (5-methoxynaphthalene) to the substrate bind-
ing site. The active-site metal is in purple and Pro564 of HIF-1a is
labeled.
[13] R. Chowdhury, M. A. McDonough, J. Mecinovic, C. Loenarz, E.
Flashman, K. S. Hewitson, C. Domene, C. J. Schofield, Structure
2009, 17, 981.
[14] For more details see the Supporting Information.
[15] T. Brown, Jr., D.Phil Thesis, University of Oxford (UK), 2009.
[16] J. M. Daniel, S. D. Friess, S. Rajagopalan, S. Wendt, R. Zenobi,
Int. J. Mass Spectrom. 2002, 216, 1.
[17] I. K. H. Leung, E. Flashman, K. K. Yeoh, C. J. Schofield,
T. D. W. Claridge, J. Med. Chem. 2010, 53, 867.
Finally, to investigate the efficacy of the C5 pyridyl
compounds in cells, we tested the methyl ester derivative of 30
(33, Scheme 2) owing to its better cell penetration. 33 was
shown to upregulate HIF1a in cells by selectively inhibiting
PHDs but not Factor Inhibiting Hypoxia (FIH).[14]
Overall, we have demonstrated the potential of boronic
acids/boronate esters for inhibitor discovery using a biologi-
cally relevant human hydroxylase. We envision that similar
approaches that use this boron chemistry will be applicable to
[18] I. Bertini, M. Fragai, C. Luchinat, E. Talluri, Angew. Chem. 2008,
120, 4609; Angew. Chem. Int. Ed. 2008, 47, 4533.
Angew. Chem. Int. Ed. 2012, 51, 6672 –6675
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6675