Toluidinesulfonamide hypoxia-induced factor 1 inhibitors: Alleviating drug-drug interactions through use of pubchem data and comparative molecular field analysis guided synthesis

Article abstract of DOI:10.1021/jm200272h

Inhibitors of hypoxia-inducible factor 1 (HIF-1) represent promising anticancer therapeutics. We have identified a series of potent toluidinesulfonamide HIF-1 inhibitors. However, the series was threatened by a potential liability to inhibit CYP2C9 which could cause dangerous drug-drug interactions when being coadministered with other drugs. We used structure-activity data from the PubChem database to develop a topomer CoMFAmodel that guided the design of novel sulfonamides with high selectivity for HIF-1 over CYP2C9 inhibition.

Full text of DOI:10.1021/jm200272h

BRIEF ARTICLE
pubs.acs.org/jmc
Toluidinesulfonamide Hypoxia-Induced Factor 1 Inhibitors:
Alleviating DrugꢀDrug Interactions through Use of PubChem Data
and Comparative Molecular Field Analysis Guided Synthesis
Bernd Wendt,* Marcel M€ulbaier, Sabine Wawro, Christoph Schultes,† Jorge Alonso, Bernd Janssen, and
Joe Lewis
Elara Pharmaceuticals GmbH, Boxbergring 107, 69127 Heidelberg, Germany
S Supporting Information
b
ABSTRACT: Inhibitors of hypoxia-inducible factor 1 (HIF-1) represent promising anticancer therapeutics. We have identified a
series of potent toluidinesulfonamide HIF-1 inhibitors. However, the series was threatened by a potential liability to inhibit CYP2C9
which could cause dangerous drugꢀdrug interactions when being coadministered with other drugs. We used structureꢀactivity data
from the PubChem database to develop a topomer CoMFA model that guided the design of novel sulfonamides with high selectivity
for HIF-1 over CYP2C9 inhibition.
’ INTRODUCTION
therapeutic index.10 Patients may face life-threatening situations
as a result of the diminished CYP2C9 enzyme activity and a
higher than expected exposure to their prescribed standard drug.
In attempts to eliminate potential CYP2C9 inhibitors early in
the drug discovery process, many computational models have
been proposed for the interaction of drugs with CYP2C9.11,12 In
addition, three crystal structures of CYP2C9 are available.13,14
Despite such valuable tools to generate structureꢀactivity
models, because of the inherent flexibility of the CYP2C9
protein’s active site and its ligand promiscuity, considerable
challenges remain for structure-based methods.15,16 However,
superior results could be achieved using ligand-based methods.17
In general, three-dimensional quantitative structureꢀactivity
relationship (3D-QSAR) studies have been very successful in
rationalizing manydatasets ofligandsandpredictingtheiraffinities
to CYP2C9.18ꢀ20 A prerequisite for good QSAR predictions is
activity data on structurally similar ligands. We were fortunate to
find a large number of ligands similar to our lead compound,
where CYP2C9 data are reported in one of the largest public
chemical biology databases.
Hypoxia-inducible factor-1 (HIF-1) is a transcription factor
that promotes in response to hypoxic conditions the production
and release of growth factors required for the development of
new blood vessels, e.g., vascular endothelial growth factor (VEGF),
proteins controlling glucose metabolism, and other mediators.1
Hypoxia is a common feature in solid tumors, as existing blood
vessels cannot supply the rapidly growing tumor with sufficient
oxygen and nutrients.2 The HIF signaling pathway therefore is
crucial for the growth of tumors, and overexpression of HIF is
often associated with an advanced disease stage and poor survival
prognosis of cancer patients.3
Because of the importance of HIF-1 in tumor development,
progression, and metastasis, a considerable amount of effort has
been devoted to identify HIF-1 inhibitors for cancer therapy.4ꢀ7
However, these compounds often have activities other than HIF-
1 inhibition, and most of them lack the desired pharmacokinetic
properties or toxicity profiles required for a useful pharmaceutical
agent. Furthermore, some of the compounds have the disadvan-
tage that they cannot be administered orally, such as the HIF-1
inhibitor EZN-2968, which is a locked nucleic acid antisense
oligonucleotide.8
’ RESULTS AND DISCUSSION
A substructure search in PubChem21 using N-phenylbenze-
nesulfonamide as a query yielded 978 virtual hits with asso-
ciated activity information from a CYP2C9 high-throughput
screen. Initially, the HQSAR22 (a 2D-QSAR method) was
attempted to derive a global QSAR model on this data set
using the percent inhibition values as dependent variable. This
yielded moderate relationships with an ensemble leave-one-out
cross-validated r2 (LOO-q2) of 0.3. Depending on the holo-
gram length, the predicted value for our lead compound varied
between 30% and 60% inhibition of CYP2C9 at 5 μM. Apart
from the inconclusive result from HQSAR, these models did
Our lead 1 is a potent inhibitor of hypoxia-induced HIF-1
activation as measured by a cell-based reporter assay (Invitrogen).
Compound 1 also has a strong inhibitory effect on cell prolifera-
tion in a panel of human cancer cell lines such as MCF7, HCT-
116, and MIAPaCa. According to literature precedence, arylsul-
fonamides have been characterized as CYP2C9 inhibitors;9
hence, we assumed that our lead compound also had a potential
liability as inhibitor for this enzyme. Therefore, we explored the
structural requirements for CYP2C9 inhibition, hoping to find
divergent SAR between CYP2C9 and HIF-1 inhibition.
CYP2C9 is a major cytochrome P450 enzyme involved in the
metabolic clearance of a wide variety of therapeutic agents.
Dangerous drugꢀdrug interactions can arise when a CYP2C9
inhibitor is administered together with drugs possessing a low
Received: March 8, 2011
Published: May 16, 2011
r
2011 American Chemical Society
3982
dx.doi.org/10.1021/jm200272h J. Med. Chem. 2011, 54, 3982–3986
|

Journal of Medicinal Chemistry
BRIEF ARTICLE
Table 1. Toluidinesulfonamides and Biological Activities for
HIF-1 and CYP2C9 and Predicted Values from Topomer
CoMFA Model
Figure 1. Actual pEC50 for CYP2C9 inhibition plotted against values
predicted from the topomer CoMFA model.
Figure 2. Topomer representation of 1: amino fragment on the left and
sulfonyl fragment on the right with contours from the topomer CoMFA
model of CYP2C9 inhibition. Green and yellow contours highlight
favored and disfavored steric interactions, whereas red and blue contours
highlight favored negative and positive electrostatic interactions
(disfavored positive and negative electrostatic interactions).
not allow for structural interpretation. Therefore, we explored
the use of the quantitative series enrichment analysis (QSEA)23
method for deriving local 3D-SAR from chemical biology
data.24 In QSEA, the ligands of the data set are organized
around its centroid, and topomer CoMFA25 models are being
created, starting with the centroid compound and its two nearest
neighbors and then iteratively expanding the training set one by
one with the next similar compound, each time creating a new
topomer CoMFA. The consecutive LOO-q2 values for our data
set reached a plateau between training set sizes of 115ꢀ127,
indicating a robust SAR. We chose the topomer CoMFA model
with a training set size of 122 and a q2 of 0.21 as 3D-QSAR model
for structural interpretation and prediction.26 Good and predic-
tive CoMFA models are suggested to have q2 >0.327 A low q2 may
simply reflect the lack of redundancy28 in the data set; however, a
high q2 does not sufficiently indicate good predictivity.29 Using
systematic exploration of topomer CoMFA models with QSEA,
where the aim is to navigate within the model space built from
available structures and activities, we suggest setting the thresh-
old of q2 to >0.2 whenever an enrichment of significant q2 is
observed. Surprisingly, irrespective of the low q2 derived from the
training set, the model’s predictions for seven newly synthesized
compounds showed a rather low SDEP of 0.54 (see Figure 1).
The overall correlation is good, indicating that the structural
modifications proposed by the model turned out to be predicted
in the right direction. However, the slope of the trend line is
rather flat. It seems that the quality of the biological information
at hand was sufficient to detect the trend but proved to be
inadequate to properly forecast its steepness.
The topomer CoMFA afforded a structural interpretation of
the altered patterns in the ligands that lead to differences in
CYP2C9 inhibition. The contour plots of the topomer CoMFA
model are shown for analogues of 1 in Figure 2, in its topomer
representation, as a pair of directly bonded amino and sulfonyl
fragments. The amino fragment (Figure 2, left) shows major
steric contributions to CYP2C9 activity in the 5- and 6-position
and major electrostatic contributions in the 1- and 2-position of
the aromatic ring. Whereas larger substitutions in position 5
(green contours) favor CYP2C9 inhibition, substituents in
position 6 (yellow contours) disfavor CYP2C9 activity. Blue
contours around the 1- and 2-position indicate favorable
interactions for hydrocarbon substituents.
The contour plots of the sulfonyl fragment (Figure 2, right)
revealed major steric and electrostatic interactions in the
4-position with minor steric interactions in the 3-position.
The model also suggests that extended substituents are in favor
of CYP2C9 inhibition (green contour) with heteroatoms
favored in the farther part of the side chain (red contours)
and electropositive hydrocarbons favored close to the aromatic
ring (blue contours). All these suggestions from the model were
further explored in the design of new analogues of our lead
3983
dx.doi.org/10.1021/jm200272h |J. Med. Chem. 2011, 54, 3982–3986

Journal of Medicinal Chemistry
BRIEF ARTICLE
Scheme 1. Synthetic Route to 1ꢀ7 in Table 1
Figure 3. Topomer representation of 5: amino fragment on the left and
sulfonyl fragment on the right with contours from the topomer CoMFA
model of CYP2C9 inhibition.
Scheme 2. Synthetic Route to 8 in Table 1
’ CONCLUSION
The 3D-QSAR model enabled the HIF-1 program to identify
structural patterns of the ligands that cause unwanted CYP2C9
inhibition. Careful interpretation of the contours allowed us to
design small changes within the toluidinesulfonamide series,
yielding potent HIF-1 inhibitors with high selectivity over
inhibition of CYP2C9. Combining the additive differences from
these studies resulted in the discovery of a new potent benzox-
azine series. Understanding potential CYP2C9 liabilities and
designing compounds that are less likely to act as inhibitors are
mandatory for a discovery project, especially at an early stage.
The utilization of publicly accessible SAR data through a
topomer CoMFA model is a suitable approach in this effort.
compound by introduction of new substituents at three posi-
tions (Table 1, Schemes 1 and 2).
Introduction of a methoxy group in the 4-position of the
sulfonylphenyl fragment, as seen with 2, shows a neutral effect
on HIF-1 activity. However, as the topomer CoMFA suggests,
this modification results in reduction of the CYP2C9 pot-
ency30 with the electronegative oxygen pointing toward the
blue contour, favoring electropositive interactions, and with
the electropositive methyl hydrogens pointing toward the red
contours, favoring electronegative interactions (Figure 3,
right).
’ EXPERIMENTAL SECTION
All reactions involving air or moisture sensitive reagents were
performed under nitrogen atmosphere using standard Schlenk techni-
ques. Unless otherwise stated, chemicals and solvents were of reagent
grade and used as obtained from commercial sources without further
purification. Water was deionized. The reactions were monitored by
thin-layer chromatography (TLC) analysis using silica gel (Merck 60-
F254) plates. Compounds were visualized by UV irradiation and/or
spraying with literature known stains followed by charring at 150 °C
(Seebach stain: 2.5 g of molybdatophosphoric acid, 1.0 g of Ce(IV)
sulfate, 6 mL of concentrated sulfuric acid, and 94 mL of water).
General Procedure for the Preparation of Sulfonamides.
In a typical procedure, a flask was charged with 8 mL of methylene
chloride. Then 165 mg (1.0 mmol) of methyl 3-amino-4-methylbenzo-
ate, 1.0 equiv of the corresponding aromatic sulfonic acid, and 0.4 mL of
pyridine were added. The flask was flushed with nitrogen, and the mixture
was stirred overnight at room temperature. After completion, the
reaction was quenched by addition of 10 mL of aqueous KHSO4
solution (5% w/w) followed by extraction with ethyl acetate (3 ꢁ
10 mL). The combined organic layers were dried with Na2SO4, and the
solvent was evaporated. The crude reaction product was further purified
by preparative LCMS to afford the final products. Compound 8 was
synthesized in an analogous procedure starting from commercially
available methyl 3,4-dihydro-2H-benzo[b][1,4]oxazine-6-carboxylate.
Compound 5 was synthesized in an analogous procedure starting from
commercially available methyl 3-amino-4-methoxybenzoate.
As seen with 3, adding another methoxy substituent in the
3-position of the sulfonylphenyl fragment does not signifi-
cantly reduce the off-target potency but interferes significantly
with HIF-1 activity. The model also suggests that extended
substitutions in the 4-position of the sulfonylphenyl fragment
would lead to higher CYP2C9 potency, which is confirmed by
4 having an acetamido side chain in the 4-position and
submicromolar activity reported for CYP2C9. With regard to
the toluidine fragment, any modification of the 5-position in
the toluidine ring had to be carefully judged because, based on
established SAR, the ester group was crucial for maintaining
HIF-1 activity. Here, the model suggests introducing a meth-
oxy group to the 2-position of the toluidine ring, thereby
exposing the electronegative oxygen next to a region where
electropositive interactions are favored and filling up space
where steric interactions are disfavored, as illustrated by 5
(Figure 3, left). This modification almost wipes out CYP2C9
potency; unfortunately this also goes along with a large drop in
HIF-1 activity. A further suggestion from the model is to
substitute the sulfonamide nitrogen with small hydrocarbons
like methyl. This results in a large reduction of CYP2C9
potency, as evidenced by 6. Combining the effects of a methoxy
substituent in the 4-position of the sulfonylphenyl fragment
with alkylation of the sulfonamide nitrogen, as shown in 7, is a
promising strategy, since the drop in CYP2C9 potency comes
with a lower impact on HIF-1 activity. Further combination of
the effects of a methoxy-substituted toluidine ring and the
methylated sulfonamide nitrogen eventually led to the discovery
of a new scaffold. The benzoxazine sulfonamide 8 shows a
promising selectivity window with a high preference for HIF-1
over CYP2C9 inhibition.
General Procedure for the Preparation of N-Methylated
Sulfonamides. In a typical procedure, 100 mg (1 equiv) of the
corresponding sulfonamide was dissolved in 5 mL of dry THF. Sodium
hydride (1.1 equiv) was added, and the mixture was stirred at room
temperature until evolution of hydrogen has stopped (5 min). Then
methyl iodide (1.1 equiv) was added and the mixture was further stirred
for 4 h at room temperature. After completion, the reaction was quenched
by addition of 20 mL of aqueous KHSO4 solution (5% w/w) followed by
extraction with ethyl acetate (3 ꢁ 10 mL). The combined organic layers
were dried with Na2SO4, and the solvent was evaporated. The crude
3984
dx.doi.org/10.1021/jm200272h |J. Med. Chem. 2011, 54, 3982–3986

Journal of Medicinal Chemistry
BRIEF ARTICLE
reaction product was further purified by preparative LCMS to afford the
final products.
(8) Greenberger, L. M.; Horak, I. D.; Filpula, D.; Sapra, P.;
Westergaard, M.; Frydenlund, H. F.; Albaek, C.; Schrøder, H;
Ørum, H. A RNA antagonist of hypoxia-inducible factor-1alpha,
EZN-2968, inhibits tumor cell growth. Mol. Cancer Ther. 2008,
7, 3598–3608.
’ ASSOCIATED CONTENT
(9) Miners, J. O.; Birkett, D. J. Cytochrome P4502C9: an enzyme of
major importance in human drug metabolism. Br. J. Clin. Pharmacol.
1998, 45, 525–538.
(10) Rettie, A. E.; Jones, J. P. Clinical and toxicological relevance of
CYP2C9: drugꢀdrug interactions and pharmacogenetics. Annu. Rev.
Pharmacol. Toxicol. 2005, 45, 477–494.
(11) Payne, V. A.; Chang, Y. T.; Loew, G. H. Homology modeling
and substrate binding study of human CYP2C9 enzyme. Proteins 1999,
37, 176–190.
S
Supporting Information. Spectroscopic data for 1ꢀ8,
b
assay protocols, dose-response curves, modeling parameters, and
SAR table from PubChem including statistical information from
QSEA. This material is available free of charge via the Internet at
http://pubs.acs.org.
’ AUTHOR INFORMATION
(12) De Groot, M. J.; Alex, A. A.; Jones, B. C. Development of a
combined protein and pharmacophore model for cytochrome P450
2C9. J. Med. Chem. 2002, 45, 1983–1993.
(13) Williams, P. A.; Cosme, J.; Ward, A.; Angove, H. C.; Matak-
Vinkovic, D.; Jhoti, H. Crystal structure of human cytochrome P450 2C9
with bound warfarin. Nature 2003, 424, 464–468.
(14) Wester, M. R.; Yano, J. K.; Schoch, G. A.; Griffin, K. J.; Stout,
C. D.; Johnson, E. F. Crystal structure of p4502c9 with flurbiprofen
bound. J. Biol. Chem. 2004, 279, 35630–35637.
Corresponding Author
*Phone: þ49 6221 387 8850. Fax: þ49 6221 387 8971. E-mail: b.
wendt@elarapharma.com.
Present Addresses
†Merck-Serono GmbH, Alsfelder Strasse 17, 64289 Darmstadt,
Germany.
(15) Afzelius, L.; Zamora, I.; Masimirembwa, C. M.; Karlꢀen, A.;
Andersson, T. B.; Mecucci, S.; Baroni, M.; Cruciani, G. Conformer-
and alignment-independent model for predicting structurally di-
verse competitive CYP2C9 inhibitors. J. Med. Chem. 2004,
47, 907–914.
(16) Sun, H.; Scott, D. O. Structure-based drug metabolism predic-
tions for drug design. Chem. Biol. Drug Des. 2010, 75, 3–17.
(17) Sykes, M. J.; McKinnon, R. A.; Miners, J. O. Prediction of
metabolism by cytochrome P450 2C9: alignment and docking studies
of a validated database of substrates. J. Med. Chem. 2008, 28, 780–791.
(18) Jones, B. C.; Hawksworth, G.; Horne, V. A.; Newlands, A.;
Morsman, J.; Tute, M. S.; Smith, D. A. Putative active site template
model for cytochrome P4502C9 (tolbutamide hydroxylase). Drug
Metab. Dispos. 1996, 24, 260–266.
(19) Ekins, S.; Bravi, G.; Binkley, S.; Gillespie, J. S.; Ring, B. J.; Wikel,
J. H.; Wrighton, S. A. Three- and four-dimensional-quantitative struc-
ture activity relationship (3D/4D-QSAR) analyses of CYP2C9 inhibi-
tors. Drug Metab. Dispos. 2000, 28, 994–1002.
(20) Rao, S.; Aoyama, R.; Schrag, M.; Trager, W. F.; Rettie, A.; Jones,
J. P. A refined 3-dimensional QSAR of cytochrome P450 2C9: computa-
tional predictions of drug interactions. J. Med. Chem. 2000,
43, 2789–2796.
’ ACKNOWLEDGMENT
This study is supported by the BMBF (Grant No. 0315418).
’ ABBREVIATIONS USED
3D, three-dimensional; CoMFA, comparative molecular field
analysis; CYP2C9, cytochrome P450 2C9; EC50, concentration
of drug that produces a half-maximal response; HIF-1, hypoxia-
inducible factor 1; HQSAR, hologram quantitative structureꢀ
activity relationship; IC50, concentration of drug that produces a
half-maximal inhibition; LOO, leave-one-out; QSAR, quantita-
tive structureꢀactivity relationship; QSEA, quantitative series en-
richment analysis; SDEP, standard deviation of prediction error
’ REFERENCES
(1) Semenza, G. L. HIF-1 and tumor progression: pathophysiology
and therapeutics. Trends Mol. Med. 2002, 8, S62–S67.
(2) Vaupel, P.; Mayer, A. Hypoxia in cancer: significance and
impact on clinical outcome. Cancer Metastasis Rev. 2007, 26,
225–339.
(21) PubChem Database. http://www.pubchem.org (accessed Oct
30, 2007).
(3) Semenza, G. L. Evaluation of HIF-1 inhibitors as anticancer
(22) Heritage, T. W.; Lowis, D. R. Molecular Hologram QSAR. In
Rational Drug Design: Novel Methodology and Practical Applications;
American Chemical Society: Washington, DC, 1999; pp 212ꢀ225.
(23) Wendt, B.; Cramer, R. D. Quantitative series enrichment
analysis (QSEA): a novel procedure for 3D-QSAR analysis. J. Com-
put.-Aided Mol. Des. 2008, 22, 541–551.
(24) Wendt, B.; Uhrig, U.; B€os, F. Capturing structureꢀactivity-
relationships from chemogenomic spaces. J. Chem. Inf. Mod 2011,
51, 843–851.
(25) Cramer, R. D. Topomer CoMFA: a design methodology for
rapid lead optimization. J. Med. Chem. 2003, 46, 374–388.
(26) There is another plateau of LOO-q2 values above 0.2 between
training set sizes of 85ꢀ95 compounds (see Supporting Information p
S10). No model was chosen from this plateau because of insufficient
structural variation within the training set.
agents. Drug Discovery Today 2007, 12, 853–859.
(4) Rapisarda, A.; Uranchimeg, B.; Scudiero, D. A.; Selby, M.;
Sausville, E. A.; Shoemaker, R. H.; Melillo, G. Identification of small
molecule inhibitors of hypoxia-inducible factor 1 transcriptional activa-
tion pathway. Cancer Res. 2002, 62, 4316–4324.
(5) Kung, A. L.; Zabludoff, S. D.; France, D. S.; Freedman, S. J.;
Tanner, E. A.; Vieira, A.; Cornell-Kennon, S.; Lee, J.; Wang, B.; Wang, J.;
Memmert, K.; Naegeli, H. U.; Petersen, F.; Eck, M. J.; Bair, K. W.; Wood,
A. W.; Livingston, D. M. Small molecule blockade of transcriptional
coactivation of the hypoxia-inducible factor pathway. Cancer Cell 2004,
6, 33–43.
(6) Welsh, S.; Williams, R.; Kirkpatrick, L.; Paine-Murrieta, G.;
Powis, G. Antitumor activity and pharmacodynamic properties of PX-
478, an inhibitor of hypoxia-inducible factor-1alpha. Mol. Cancer Ther.
2004, 3, 233–244.
(7) Chau, N. M.; Rogers, P.; Aherne, W.; Carroll, V.; Collins, I.;
McDonald, E.; Workman, P.; Ashcroft, M. Identification of novel small
molecule inhibitors of hypoxia-inducible factor-1 that differentially block
hypoxia-inducible factor-1 activity and hypoxia-inducible factor-1alpha
induction in response to hypoxic stress and growth factors. Cancer Res.
2005, 65, 4918–4928.
(27) Agarwal, A.; Pearson, P. P.; Taylor, E. W.; Li, H. B.; Dahlgren,
T.; Hersl€of, M.; Yang, Y.; Lambert, G.; Nelson, D. L.; Regan, J. W.;
Martin, A. R. Three-dimensional quantitative structureꢀactivity rela-
tionships of 5-HT receptor binding data for tetrahydropyridinylindole
derivatives: a comparison of the Hansch and CoMFA methods. J. Med.
Chem. 1993, 36, 4006–4014.
3985
dx.doi.org/10.1021/jm200272h |J. Med. Chem. 2011, 54, 3982–3986

Journal of Medicinal Chemistry
BRIEF ARTICLE
(28) Doweyko, A. M. 3D-QSAR illusions. J. Comput.-Aided Mol. Des.
2004, 18, 587–596.
(29) Golbraikh, A.; Tropsha, A. Beware of q2!. J. Mol. Graphics
Modell. 2002, 20, 269–276.
(30) Methoxy substituents represent potentially labile groups; how-
ever, in vivo studies later showed that introduction of the 4-methoxy
group did not result in a metabolic liability for this series.
3986
dx.doi.org/10.1021/jm200272h |J. Med. Chem. 2011, 54, 3982–3986