Improved, selective, human intestinal carboxylesterase inhibitors designed to modulate 7-ethyl-10-[4-(1-piperidino)-1-piperidino]carbonyloxycamptothecin (irinotecan; CPT-11) toxicity

Article abstract of DOI:10.1021/jm9001296

CPT-11 is an antitumor prodrug that is hydrolyzed by carboxylesterases (CE) to yield SN-38, a potent topoisomerase I poison. However, the dose limiting toxicity delays diarrhea that is thought to arise, in part, from activation of the prodrug by a human i

Full text of DOI:10.1021/jm9001296

3742
J. Med. Chem. 2009, 52, 3742–3752
Improved, Selective, Human Intestinal Carboxylesterase Inhibitors Designed to Modulate
7-Ethyl-10-[4-(1-piperidino)-1-piperidino]carbonyloxycamptothecin (Irinotecan; CPT-11)
Toxicity
Latorya D. Hicks,† Janice L. Hyatt, Shana Stoddard,‡ Lyudmila Tsurkan,† Carol C. Edwards,† Randy M. Wadkins,‡ and
Philip M. Potter*,†
Department of Molecular Pharmacology, St. Jude Children’s Research Hospital, 262 Danny Thomas Place, Memphis, Tennessee 38105-2794,
Department of Chemistry and Biochemistry, UniVersity of Mississippi, UniVersity, Mississippi 38677
ReceiVed February 2, 2009
CPT-11 is an antitumor prodrug that is hydrolyzed by carboxylesterases (CE) to yield SN-38, a potent
topoisomerase I poison. However, the dose limiting toxicity delays diarrhea that is thought to arise, in part,
from activation of the prodrug by a human intestinal CE (hiCE). Therefore, we have sought to identify
selective inhibitors of hiCE that may have utility in modulating drug toxicity. We have evaluated one such
class of molecules (benzene sulfonamides) and developed QSAR models for inhibition of this protein. Using
these predictive models, we have synthesized a panel of fluorene analogues that are selective for hiCE,
demonstrating no cross reactivity to the human liver CE, hCE1, or toward human cholinesterases, and have
Ki values as low as 14 nM. These compounds prevented hiCE-mediated hydrolysis of the drug and the
potency of enzyme inhibition correlated with the clogP of the molecules. These studies will allow the
development and application of hiCE-specific inhibitors designed to selectively modulate drug hydrolysis
in vivo.
Introduction
We have sought therefore to identify compounds that can
inhibit CEs without impacting human acetyl- or butyrylcho-
linesterase (hAChE and hBChE, respectively). Initially, we
screened a library of compounds from Telik19 and identified
several compounds that were selective inhibitors of CEs.20,21
Of these, one class demonstrated selectivity toward hiCE versus
the human liver CE, hCE1 (CES1).21 The majority of these
compounds were benzene sulfonamides, and preliminary studies
indicated that halogen substitution tended to increase the potency
of the inhibitors. However, these studies were based on a series
of 9 compounds (4-12 in Table 1) with a disparate set of
different chemotypes.21 Here we have considerably expanded
these analyses and have now assayed and analyzed 57 benzene
sulfonamides for their ability to inhibit hiCE, hCE1, hAChE,
or hBChE. Using detailed QSAR models, we have designed a
series of novel fluorene analogues that are highly potent hiCE
inhibitors and can modulate 1 metabolism. Potentially, these
molecules would be lead compounds for subsequent drug design.
Carboxylesterases (CEa) are ubiquitously expressed enzymes
that are thought to be responsible for the hydrolysis of
xenobiotics.1 They catalyze the conversion of esters to their
corresponding alcohols and carboxylic acids. Because numerous
clinically used compounds are esterified, an approach used by
the pharmaceutical industry to improve the water solubility of
molecules, they are substrates for these enzymes. Hence, drugs
such as heroin, cocaine, 1 (irinotecan; CPT-11;2 Figure 1),
capecitabine, oseltamivir, lidocaine, and meperidine are all
hydrolyzed by CEs.3-16 Therefore, identifying compounds that
modulate the hydrolysis of these agents may be useful in either
altering the half-life and/or toxicities associated with these drugs.
For example, flestolol, a ꢀ-blocker is rapidly hydrolyzed by CEs
to an inactive metabolite and hence its biological activity is
rapidly lost.17 Inhibition of the enzyme responsible for this
hydrolysis would increase the in vivo stability of the molecule
and likely improve its therapeutic utility. In contrast, the delayed
diarrhea that is associated with 1 treatment is thought to arise,
in part, from hydrolysis of the drug in the intestine by the human
intestinal CE (hiCE, CES2)12,13,18 to yield 2 (7-ethyl-10-
hydroxycamptothecin; SN-38; Figure 1). Because this is the dose
limiting toxicity for this highly effective anticancer agent,
approaches that ameliorate this side effect would improve patient
quality of care and potentially allow drug dose intensification.
This could potentially be achieved by an inhibitor that targets
hiCE within the gut.
Results
Selective Inhibition of hiCE by Benzene Sulfonamides. On
the basis of our previous work,21 we had identified benzene
sulfonamides as selective inhibitors of hiCE. The 3D-QSAR
analysis presented in this article indicated that (i) halogenation
of the phenyl rings resulted in more potent compounds, and
(ii) that the central region of the inhibitor-enzyme complex
was hydrophobic and could accommodate a large aromatic
structure. Therefore, we synthesized or acquired a total of 57
sulfonamide analogues, mostly containing halogen atoms ap-
pended to the benzene rings, and assessed their inhibitory
potency toward the CEs hiCE and hCE1, as well as hAChE
and hBChE. These assays used 3 (o-nitrophenyl acetate) as a
substrate for the former enzymes and acetylthiocholine and
butyrylthiocholine for the respective cholinesterases.
* To whom correspondence should be addressed. Phone: 901-595-3440.
Fax: 901-595-4293. E-mail: phil.potter@stjude.org.
†
Department of Molecular Pharmacology, St. Jude Children’s Research
Hospital.
‡
Department of Chemistry and Biochemistry, University of Mississippi.
a Abbreviations used: AChE, acetylcholinesterase; BChE, butyrylcho-
linesterase; CE, carboxylesterase; hAChE, human AChE; hBChE, human
BChE; hCE1, CES1, human carboxylesterase 1; hiCE, CES2, human
intestinal carboxylesterase; HPLC, high performance liquid chromatography;
Ki, inhibition constant.
As indicated in Table 1, the vast majority of the compounds
were excellent inhibitors with Ki values ranging from 41 to 3240
10.1021/jm9001296 CCC: $40.75
2009 American Chemical Society
Published on Web 05/27/2009

Human Intestinal Carboxylesterase Inhibitors
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12 3743
Figure 1. The chemical structure and hydrolysis of 1 resulting in the formation of 2.
nM. The sulfonamides were selective for hiCE, with only one
molecule demonstrating weak activity toward hCE1 (compound
11). However, 11 was still over 250-fold more potent against
hiCE (Ki values were 53 vs 13700 nM for hiCE and hCE1,
respectively). None of the sulfonamides inhibited either hAChE
or hBChE (data not shown), consistent with our previously
reports of these types of compounds.21 Because both CEs and
cholinesterases demonstrate very similar crystal structures,4,22
we presume that the specificity of the sulfonamides for hiCE is
due to unique interactions with amino acids within the active
site of this protein.
Six compounds were inactive toward all enzymes (28, 39,
41, and 44-46). The majority of these compounds contained
large bulky atoms or moieties present within either the terminal
benzene rings (28, 39, 41) or the central domain of the molecule
(44-46). These groups would likely impede access of the
inhibitor to that active site gorge, thereby mitigating their
biological activity.
The mode of enzyme inhibition by the sulfonamides inhibitors
was partially competitive, indicating that while their structure
resembled the substrate molecule, they were unable to com-
pletely inhibit substrate hydrolysis.23 For this series of com-
pounds, therefore, the inhibitory potency will be partially
dependent upon the structure of the substrate. To confirm that
these molecules could indeed inhibit the hydrolysis of other
substrates, their effect on the metabolism of 1 was assessed.
Inhibition of hiCE-Mediated 1 Hydrolysis. Having dem-
onstrated potent inhibition of the hydrolysis of 3 by the
sulfonamides, we determined their ability to prevent conversion
of 1 to 2. As indicated above, this was necessary because the
inhibitors acted in a partially competitive fashion. Table 2
displays the Ki values for the inhibition of 1 hydrolysis by these
compounds. As can be seen, very potent inhibitors were
obtained, with inhibition constants as low as 23.4 nM being
observed. A comparison of the inhibition constants for 1 and 3
demonstrated a good linear correlation (r2 ) 0.72; Figure 2) as
well as an excellent Spearman r value (r ) 0.80; p ) 0.0003).
Again, only compounds that inhibited the hydrolysis of both
substrates were included in these analyses. These results indicate
that, in general, there is a commonality in the potency of enzyme
inhibition and ability of selected compounds to inhibit the
hydrolysis of specific substrates. Furthermore, we have also
demonstrated that compounds 18 and 54 can also inhibit the
metabolism of heroin and cocaine (data not shown; Hatfield et
al., manuscript in preparation). In summary, these results
demonstrate that while determination of the actual Ki values
for specific substrates will always be necessary, it is likely that
compounds that can inhibit o-NPA hydrolysis will also be active
toward other esters.
30 Å into the interior of the enzyme.4,22,25 These hydrophobic
regions are also present in the 3D-QSAR models and are thought
to reflect the nature of the amino acids that line the gorge.
Therefore, we graphed the inhibition constants for the sulfona-
mides with hiCE versus the clogP values for the respective
compounds. For analyses with 3, we excluded the data sets for
compounds 28, 39, 41, and 44-46 because they did not inhibit
the enzyme. Similarly with 1 as a substrate, we excluded data
obtained from compounds 4, 6, 7, 10, 12, 20, 46, 50, and 52.
As indicated in Figure 3, reasonable correlations were
observed between the Ki values and clogP for both substrates,
with linear correlation coefficients (r2) for the curve fit of 0.49
and 0.43 for 1 and 3, respectively. Analysis of these data
however, using a Spearman rank order calculation, demonstrated
highly significant results (Table 3). Because this statistical test
uses a ranking system for determining significance, these
analyses cannot be used to predict the Ki value for novel
sulfonamides. Nevertheless, it is clear that the relative hydro-
phobicity of the molecule is an important factor in determining
the biological potency of these compounds.
Inhibition of hiCE by Fluorene Analogues. On the basis
of the information from our earlier 3D-QSAR studies and the
clogP correlations, we hypothesized that a planar hydrophobic
ring structure, larger than benzene, in the central core of the
molecule might improve inhibitor potency. We excluded
compounds that might increase the bulkiness of the molecule
because the active site for hiCE is thought to exist as a long
deep gorge that projects into the interior of the protein.
Therefore, we synthesized a series of fluorene derivatives
(compounds 56-60) and assessed their ability to selectively
inhibit hiCE. As can be seen from Table 4, all of these analogues
were potent inhibitors of hiCE with Ki values ranging from
14-91 nM. In addition, all of these molecules inhibited hiCE
in a partially competitive fashion, similar to that seen for the
other sulfonamides. Direct comparison of the inhibition constants
for compounds 8 and 56, which differ only by the fluorene
moiety in the central domain of the molecule, indicate that the
latter is ∼11-fold more potent at inhibiting hiCE. Furthermore,
addition of halogen atoms to the terminal benzene rings also
increased the efficacy of the molecules, with compounds 57,
58, and 60 being among the most potent at inhibiting both the
hydrolysis of 1 and 3 (Table 2). Indeed, these sulfonamides
yielded Ki values of 58, 40, and 38 nM, with 1, respectively.
These results indicate that the inclusion of a planar, aromatic,
hydrophobic domain within the center of the molecule is
beneficial for hiCE inhibition.
3D-QSAR Analyses. Having developed relatively simple
relationships for enzyme inhibition based upon the clogP value
of the inhibitors, we undertook detailed 3D-QSAR analyses with
the goal of identifying specific domains within the molecules
that might improve (or ablate) inhibitory potency. We have
previously performed similar studies, and the pseudoreceptor
site models that have been generated have significantly enhanced
subsequent inhibitor design. Therefore, compounds 5, 8, 8-11,
Correlation between clogP and Inhibitor Potency. We have
previously demonstrated that for a series of isatins, the inhibitory
potency of the compounds toward CEs was related to their clogP
values.24 This is likely due to the fact that the active sites of
these proteins are highly hydrophobic gorges that project up to

3744 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12
Hicks et al.
Table 1. Ki Values for the Inhibition of Human CEs and Cholinesterases by the Benzene Sulfonamidesd

Human Intestinal Carboxylesterase Inhibitors
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12 3745
Table 1. Continued

3746 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12
Table 1. Continued
Hicks et al.
a Bz represents benzene ring. b Data taken from Wadkins et al.21 c Data taken from Hatfield et al. (manuscript submitted). d For the CEs, 3 was used as
a substrate and the respective thiocholines were used for hAChE and hBChE. The general structure of the sulfonamides is indicated. The X in the subfragment
represents the point of attachment to the sulfonamide moiety.
14-31, 35-37, 47-56, and 60 were used as a training set for
the development of the QSAR model. This was then validated
using the other molecules (the test set). As can be seen from
Table 5 and Figure 4, the linear correlation coefficients (r2) for
the observed versus predicted Ki values for the training set were
∼0.9, with similar values for the cross correlation coefficients
(q2). Because q2 values greater than 0.4 are considered statisti-
cally significant for biological systems,26 these models are likely

Human Intestinal Carboxylesterase Inhibitors
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12 3747
Table 2. Ki Values for the Inhibition of hiCE-Mediated Hydrolysis of 1
by Selected Benzene Sulfonamidesa
compd
Ki 1 (nM)
Ki 3 (nM)
4
5
6
7
8
9
10
11
12
13
15
16
17
18
19
20
23
46
50
52
54
>100000b
3220 ( 950b
>100000b
>100000b
892 ( 67b
238 ( 29b
>100000
141 ( 64b
>100000
110 ( 23b
152 ( 29
79 ( 15
218 ( 45b
1310 ( 176b
165 ( 33b
767 ( 285b
1060 ( 133b
365 ( 87b
194 ( 23b
53.2 ( 5.5b
451 ( 39b
41.5 ( 6.5b
248 ( 26
355 ( 70
62 ( 15
74.0 ( 5.5
67.0 ( 9.6c
85.2 ( 10.1
5260 ( 620
210 ( 33
121 ( 15c
44 ( 10
>100000
192 ( 44
>100000
>100000
>100000
28 ( 4c
Figure 3. Graph demonstrating the correlation between Ki values and
clogP for the sulfonamide inhibitors. Assays were run using either 1
(red) or 3 (blue) as substrates. Linear correlation coefficients (r2) for
the curve fits are indicated.
>100000
3240 ( 1780
2600 ( 1010
23.4 ( 2.7c
Table 3. Linear Correlation Coefficients and P Values for the Statistical
Analysis of the Relationship between the clogP and the Ki Values for
hiCE Inhibition by the Sulfonamidesa
a Inhibition constants using 3 are included for comparison. b Data taken
from Wadkins et al.21 c Data taken from Hatfield et al. (manuscript in
preparation).
parameter
substrate
r2
Spearman r
P
1
3
0.486
0.432
-0.832
-0.615
<0.0001
<0.0001
a Results are presented using either 1 or 3 as substrates.
ered good inhibitors of hiCE (Ki values in the low µM range),
only 58 was predicted with any great accuracy (Table 4 and
Figure 4). This is likely due to the fact that only 12 compounds
were used as the training set for this model, and this did not
include any of the fluorene analogues. However, these QSAR
models will allow for rapid screening of analogues for inhibitor
potency prior to the initiation of chemical synthesis.
Pseudoreceptor Site QSAR Models. To provide a graphical
representation of the QSAR results, we developed a pseu-
doreceptor model for hiCE with data sets derived from the
inhibition of hydrolysis of 3 (Figure 5A). This figure outlines
the interactions that describe receptor-ligand binding in the
enzyme. The model has primarily anionic (red areas) regions
of charge located in a cluster at the base of the model and a
weakly cationic (blue spheres) domain located at the top. This
is consistent with all our previous analysis of hiCE inhibitors,
where charge asymmetry was observed both in the QSAR
models20,21,24,27-29 and in the enzyme structure determined from
homology modeling.21 We also present an electrostatic potential
map of the model (Figure 5B) oriented to emphasize the charge
asymmetry. While the origin of this charge distribution is not
completely understood, at least for homology structures of hiCE,
it maps well onto the position of charged amino acids present
within the active site.21 It should be remembered that these
figures represent a description of the interior surface of the active
site and not the combined surfaces of the inhibitor molecules.
However, in general, the electrostatic potential inside the active
site was negative30 and hence it is not clear whether the positive
areas required to fit the inhibition data reflect interactions within
the active site alone, or represent potential interactions with
positively charged residues near the active site opening. The
interpretation of these models will be greatly enhanced if a
crystal structure of hiCE becomes available.
Figure 2. Graph demonstrating the correlation between the Ki values
for inhibition of 1 hydrolysis vs those for 3 hydrolysis when using
hiCE. The linear correlation coefficient (r2) for the curve fit is indicated.
to have excellent predictive value in the design of novel
sulfonamide-based hiCE inhibitors. In addition, the q2/r2 values
were close to unity (Table 5), confirming the validity of the
models.
As indicated in Table 4, the QSAR models were reasonable
at predicting the Ki values for the fluorene analogues 56-60,
with all compounds except 58 being considered excellent
inhibitors of hiCE (<100 nM) when using 3 as a substrate.
Interestingly, the latter molecule had the highest clogP value
(5.69), suggesting that for these models, this parameter is not a
major determinant of biological potency. Significantly, while
the fluorene moiety increased the length of the molecule by 4-5
Å as compared to the benzene derivative, this did not signifi-
cantly impact enzyme inhibition. Also it should be noted (see
Figure 3) that the model was very poor at predicting the efficacy
of 7. This is potentially due to the fact that this compound
contains a disulfide chemotype (unlike all of the other molecules
that were assayed) and can potentially adopt alternate conforma-
tions that would not be accurately predicted by the model.
The QSAR predicted Ki values for the inhibition of hiCE-
mediated 1 metabolism were not as good as though seen for 3
(Table 4). Indeed, while the fluorene compounds were consid-
Discussion
In this article, we have demonstrated that potent, selective
inhibitors of hiCE based upon the benzene sulfonamide scaffold

3748 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12
Hicks et al.
Table 4. Predicted and Observed Ki Values for hiCE with Five Fluorene Analogues (Compounds 56-60) that Were Postulated from the QSAR
Analyses to be Excellent CE Inhibitors
Table 5. QSAR Validation Parameters Pbtained from Quasar Software
when Using the Ki Values for hiCE Inhibition with Either 1 or 3 as a
Substrate
would be more likely to localize within the hydrophobic active
site gorge of the protein.
In contrast, substitution with multiple larger atoms (e.g.,
compound 28) resulted in loss of biological activity and is likely
due to the fact that this molecule is too large to fit within the
active site of hiCE. Furthermore, molecules containing a
carboxylic acid or amide group at the 4-position (39 and 41)
were inactive. Whether the loss of enzyme inhibition this is
due to electronic effects on the other atoms within the sulfon-
amide, or a steric interaction that forces the inhibitor into a
conformation such that it can no longer interact with amino acids
that line the active site, or a combination of both, is unclear.
However, it is apparent that introducing substitutions that
increase the clogP without dramatically increasing the size of
the molecules can improve the potency of these analogues.
substrate
r2
q2
q2/r2
1
3
0.89
0.91
0.83
0.88
0.93
0.96
can be developed. This has resulted in the development of a
series of fluorene analogues that have Ki values in the low nM
range for both the inhibition of hydrolysis of 1 and 3. These
compounds (56-60) were designed and synthesized based upon
prior 3D-QSAR pseudoreceptor site models that indicated that
a bulky, hydrophobic central domain within the inhibitors
improved their potency.
The benzene sulfonamide analogues that we assayed fell into
four broad classes. Compounds 4-13 were originally identified
in a small scale library screen,21 and they essentially contained
three domains. This included terminal and central phenyl rings
bonded via sulfonamide chemotypes and substitutions within
the rings that altered the chemical properties of the compounds.
Because compound 13 demonstrated the lowest Ki for hiCE
inhibition, this molecule was used as a scaffold for the design
and synthesis of analogues 14-55. Due to the fact that the most
potent compounds tended to be halogen substituted,21 we
concentrated our efforts on the generation of inhibitors contain-
ing these atoms, principally in the terminal benzene rings
(molecules 14-41). Among this latter series of compounds, we
noted that substitution in either the 3- (meta-) or 4- (para-)
position with chlorine or bromine (compounds 17-19; Ki values
ranging from 67-85 nM) resulted in significantly lower
inhibition constants as compared to that of the unsubstituted
molecule (8; Ki ) 1060 nM). Indeed, 17-19 were the most
potent of the molecules containing substitutions within the
terminal benzene rings. These halogens increase the hydropho-
bicity of the molecules (clog P for 17-19 range from 4.3-4.7,
∼1 log greater than compound 8), and therefore these analogues
Compounds 42-54 contained substitutions within the central
benzene ring coupled with halogens appended in the terminal
phenyl groups. In general, these modifications resulted in
reduced potency toward hiCE inhibition. This is exemplified
by compounds 44-46, which are inactive in this assay.
However, 54, which contains the 2,3,5,6-tetrafluoro substitution,
resulted in increased activity as compared to the unsubstituted
molecule (19). Because these compounds demonstrate very
similar clogP values, the increase in biological activity is likely
due to a change in the electron distribution afforded by the very
electronegative fluorine atoms. As we have not yet obtained
the X-ray structure of hiCE, it has not been possible to identify
the specific amino acids with which these sulfonamides interact.
Hence, it is unclear how these small molecules demonstrate
specificity for hiCE. However, based on previous observations
that sulfonamides can inhibit thrombin, we presume that a
similar mechanism of enzyme inhibition occurs.31-34 These
studies demonstrated that the oxygen atoms within the sulfon-
amide group can hydrogen bond with amino acids present within
the active site of this protein. Therefore, we believe that the

Human Intestinal Carboxylesterase Inhibitors
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12 3749
Figure 5. 3D-QSAR pseudoreceptor site models generated from the
enzyme inhibition data for hiCE using the benzene sulfonamides with
3 as a substrate. These models were generated using results obtained
from all 57 sulfonamides and include compound 8 for reference. Both
figures were generated using Raster3D43 and Molscript.44 (A) The
model is depicted as colored spheres on a hydrophobic gray grid. Areas
that are hydrophobic are indicated in gray, where dark-blue spheres
represent regions that are positively charged (+0.25e) and light-blue
spheres correspond to reduced charge (+0.1e). Dark-red spheres indicate
domains that are negatively charged (-0.25e), and light-red spheres
represent areas of reduced negative charge (-0.1e). In all cases, e is
the charge of the proton. (B) Stereo view of a 3D-QSAR pseudoreceptor
site model that describes sulfonamide binding as a molecular surface
upon which the electrostatic potential is mapped. The electrostatic
potential is calculated from the Quasar software partial charges, which
were defined as -0.25e, -0.1e, +0.1e, and +0.25e for negative salt
bridge, hydrophobic negative, hydrophobic positive, and positive salt
bridge characteristics, respectively. As above, e is the charge of the
proton. The figure is oriented to emphasize the charge asymmetry that
appears in all QSAR models that we have observed in previous
analyses.20,21,24,27-29
Figure 4. Graphs demonstrating the correlation between the predicted
and the observed Ki values obtained from the QSAR models. (A) Graph
when using 3 as a substrate, and (B) when using 1. Linear correlation
coefficients for the line fits (r2) are indicated on the graphs.
unique arrangement of residues present within the hiCE catalytic
gorge, and their ability to form hydrogen bonds with the
sulfonamides, represents the key interactions responsible for
selective CE inhibition.
Consistent with our previous reports describing CE inhibi-
tors,20,24,27 molecules that contained substitutions that increased
the bulkiness or width of the compound were generally poor
inhibitors. This was exemplified in this series of analogues by
28, 39, 41, and 44-46. All of these sulfonamides contained
groups or atoms that significantly increased either the width or
length of the molecule that would preclude facile access of the
inhibitor to the CE active site. Because the hydrolysis of
compounds by CEs is dramatically influenced by their ability
to interact with the catalytic amino acids,14,30 it is highly likely
that the same holds true for inhibitor molecules. Therefore,
compounds that exceed the dimensions of the active site gorge
in these proteins would not be inhibitors of these proteins.
Molecules 28, 39, 41, and 44-46 are much larger and bulkier
than the other compounds described here and therefore do not
inhibit hiCE.
On the basis of the results of the QSAR analyses, we
hypothesized that introducing a larger, planar, aromatic core
domain within the center of the molecule should increase its
potency. This was due to the fact that this moiety would increase
the clogP (hydrophobicity) of the compound, without impeding
the ability of the sulfonyl groups to interact with the amino
acids present within the active site. This is consistent with the
3D-QSAR model depicted in Figure 4. This data set was
generated using both compounds 56 and 60 in the training set.
Excluding these compounds resulted in QSAR relationships that
predicted 56-60 to be good inhibitors, but the experimental
data revealed them to be excellent inhibitors: much better than
predicted and some of the most potent compounds we have
examined to date. One difficulty with any QSAR analysis is
that the fewer members of a class of molecules in the training
set, the less likelihood of correct prediction of that series of
compounds. Hence, we rebuilt the models, including 56 and
60, to determine whether this would result in improved
prediction of the Ki values for 57-59. As expected, the
predictive properties of the data were improved by at least 1
order of magnitude by including 56 and 60, suggesting that as
we synthesize and test more fluorene-containing molecules, we
should be able to produce a more accurate QSAR relationship
for other classes of molecules beyond those used in this study.

3750 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12
Hicks et al.
in the presence of increasing concentrations of inhibitor. The
fractional inhibition was then graphed versus inhibitor concentration
and data fitted to the following equation;23
In general, halogen substitution of the distal phenyl rings
slightly decreased the Ki value as compared to that of the un-
substituted analogue 11, whereas methyl groups added to the
central phenyl ring did not much affect the binding of the
analogues. This was somewhat surprising, as the 3D-QSAR
pseudoreceptor model (Figure 5) is highly hydrophobic in the
central ring area, as indicated by gray lines. However, replace-
ment of the central phenyl ring with a naphthyl or fluorene
chemotype greatly reduced the inhibition constants of the
compounds, i.e., making them more potent hiCE inhibitors. We
attribute this to π-π interactions of the small molecules with
the tryptophan and phenylalanine rings that are known to line
the active site gorge of hiCE.25 Similarly, this would explain
the potent inhibitory power of the indole-containing isatins that
we have previously characterized.24
[I]{[S](1 - ꢀ) + Ks(R - ꢀ)}
[I]{[S] + RKs} + KI{R[S] + RKs}
i )
(1)
where [I] ) inhibitor concentration, [S] ) substrate concentration,
i ) fractional inhibition, Ks is the dissociation constant for the
enzyme-substrate complex (assuming negligible commitment to
catalysis), R ) change in affinity of the substrate for the enzyme
in the presence of the inhibitor (where R > 0), ꢀ ) change in the
rate of enzyme-substrate complex decomposition in the presence
of the inhibitor (where 1 > ꢀ > 0), and Ki is the inhibitor constant.
Routinely, results were evaluated using Perl Data Language and
graphed with GraphPad Prism software with the mode of enzyme
inhibition being determined using Akaike’s information criteria.20
Ki values were then calculated using the best fit model described
from these analyses.
Conclusions
Inhibition of Acetyl- and Butyrylcholinesterase. Inhibition of
hAChEandhBChEweredeterminedaspreviouslyreported20,21,24,28,29
using either acetylthiocholine or butyrylthiocholine as substrates,
respectively.
In summary, we have synthesized and analyzed the inhibitory
properties of a panel of bisbenzene sulfonamides, and demon-
strated that these compounds have specificity for hiCE. This
has allowed the design of a series of fluorene analogues that
are highly potent inhibitors, with Ki values in the low nM range
when using either 1 or 3 as substrates. We are currently
determining the ability of these compounds to inhibit hiCE
intracellularly and evaluating whether any of these molecules
represent valid lead compounds for in vivo use. If so, they may
allow the design of clinical candidates, suitable for the modula-
tion of 1-induced toxicity, in patients treated with this drug.
Inhibition of hiCE-Mediated 1 Hydrolysis. The inhibition of
the conversion of 1 to 2 by hiCE was determined using the
following conditions. Inhibitor concentrations, ranging from 1 nM
to 100 µM, were added to reactions (100 µL) containing 25 U of
hiCE (1U represents the amount of enzyme that catalyzes the
conversion of 1 nmole of o-nitrophenyl acetate per min), 20 µM 1,
and 50 mM Hepes pH7.4. Reactions were incubated at 37 °C for
5 min and terminated by the addition of 100 µL of cold acidified
methanol. Samples were clarified by centrifugation at 14000g for
10 min, and concentrations of 2 in the supernatant were then
determined by reverse phase HPLC, as previously described.21,37
Ki values were calculated in an identical fashion to that indicated
above.
Calculation of clogP Values. clogP values were predicted using
ChemSilico Predict software (ChemSilico, Tewksbury, MA).
Data Graphing and Statistical Analyses. Results were graphed
and analyzed using GraphPad Prism software. Statistical correlations
were determined using the Spearman method (nonparametric) to
generate two-tailed P values. Spearman correlation coefficients (r)
equal to -1 or +1 represent perfect negative or positive correlations,
respectively. P values <0.05 were considered statistically significant.
3D-QSAR Modeling of Inhibitors. Multidimensional-QSAR
modeling of carboxylesterase inhibitors was performed using Quasar
5 software38-41 running on a Macintosh G5. The structure for each
compound was built using Chem 3D, and each structure was
minimized with MM2 and PM3 formalisms. The solvation energies
and charges were calculated using the AMSOL 7.1 program42 using
the SM5.42R solvation method. The gas phase charges obtained
were used for all QSAR analyses. A resulting receptor surface was
then generated using Quasar 5.0 and analyzed to yield over 200
independent models and subsequently refined to generate ∼7000
pseudoreceptor site models. Repeated analyses of the data sets were
then performed until the cross correlation coefficients (q2) were
greater than 0.7 for the experimentally determined versus the
predicted Ki values. Routinely, this yielded correlation coefficients
(r2) of >0.8.
Experimental Section
Reagents and Chemicals. General reagents and chemicals were
obtained from Sigma Aldrich (St. Louis, MO). Sulfonamides were
obtained from the following commercial sources: 4-12 from Telik
(Palo Alto, CA), 25 and 32 from Akos (Steinen, Germany), 28,
36, 42, 44, and 45 from Princeton (Monmouth Junction, NJ), 29,
30, 37-41, and 55 from Enamine (Monmouth Junction, NJ), 31
and 33-35 from Scientific Exchange (Center Ossipee, NH), and
43 from AsinEx (Winston-Salem, NC). The synthesis of compound
13 has been previously described.21 Compounds were dissolved in
DMSO (typically at a concentration of 10 mM) immediately prior
to biochemical analysis.
Sulfonamide Synthesis. Sulfonamides 14-24, 26, 27, 46-54,
and 56-60 were synthesized by condensation of the respective
diamines and sulfonyl chlorides. Briefly, diamine (5 mmol) was
dissolved in CH2Cl2 (∼30 mL) with stirring and 2 mL of pyridine
and 0.2 g of dimethylaminopyridine were added. The sulfonyl
chloride (12 mmol), dissolved in 15 mL of CH2Cl2, was added in
one portion and the reaction was allowed to stir for 2-16 h. During
this time, the reaction became dark-pink/orange-red and a fine
precipitate appeared. The N,N′-bis-arene sulfonamide was recovered
by filtration and washed with 100 mL of CH2Cl2 and 100 mL of
ethyl acetate. After drying under vacuum, the material was
resuspended in 250 mL of 0.1 M HCl and stirred at room
temperature for 2-3 h. The mixture was filtered, washed with water,
and dried to yield the pure product. All compounds were validated
1
by TLC, H and 13C NMR, HRMS, melting point and elemental
Two independent QSAR models were constructed using two
different training sets. The first was constructed using 37 sulfona-
mides as inhibitors of hiCE using 3 as a substrate. The second was
obtained using the Ki values of 12 inhibitors of 1 hydrolysis. In all
cases, a range of Ki values spanning 3 orders of magnitude were
used to ensure a reasonable model. We then included the structures
of a number of potential inhibitors of hiCE in the test set to identify
novel chemical scaffolds that might be potential selective inhibitors
of hiCE. These included the fluorene-containing compounds 56-60.
analysis, and were greater than 95% pure. Full details are provided
as Supporting Information.
HumanEnzymes.hCE1waspurifiedaspreviouslydescribed4,22,35
and hiCE was generated in a similar manner, except that a slightly
different purification procedure was adopted (Hatfield et al.,
manuscript in preparation). hAChE and hBChE were purchased
from Sigma Aldrich.
Carboxylesterase Inhibition Assays and Determination of
Ki Values. Parameters for CE inhibition were determined as
previously described.20,21,24,28,29,36 Briefly, the inhibition of hy-
drolysis of 3 to o-nitrophenol was measured spectrophotometrically
Acknowledgment. We thank Dr. J. P. McGovren for the
gift of compound 1. This work was supported in part by NIH

Human Intestinal Carboxylesterase Inhibitors
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12 3751
(20) Wadkins, R. M.; Hyatt, J. L.; Wei, X.; Yoon, K. J.; Wierdl, M.;
Edwards, C. C.; Morton, C. L.; Obenauer, J. C.; Damodaran, K.;
Beroza, P.; Danks, M. K.; Potter, P. M. Identification and
characterization of novel benzil (diphenylethane-1,2-dione) ana-
logues as inhibitors of mammalian carboxylesterases. J. Med. Chem.
2005, 48, 2905–2915.
(21) Wadkins, R. M.; Hyatt, J. L.; Yoon, K. J.; Morton, C. L.; Lee, R. E.;
Damodaran, K.; Beroza, P.; Danks, M. K.; Potter, P. M. Identification
of novel selective human intestinal carboxylesterase inhibitors for the
amelioration of irinotecan-induced diarrhea: Synthesis, quantitative
structure-activity relationship analysis, and biological activity. Mol.
Pharmacol. 2004, 65, 1336–1343.
(22) Bencharit, S.; Morton, C. L.; Hyatt, J. L.; Kuhn, P.; Danks, M. K.;
Potter, P. M.; Redinbo, M. R. Crystal structure of human carboxy-
lesterase 1 complexed with the Alzheimer’s drug tacrine. From
binding promiscuity to selective inhibition. Chem. Biol. 2003, 10,
341–349.
(23) Webb, J. L. Enzyme and Metabolic Inhibitors, Volume 1. General
Principles of Inhibition; Academic Press Inc.: New York, 1963.
(24) Hyatt, J. L.; Moak, T.; Hatfield, J. M.; Tsurkan, L.; Edwards, C. C.;
Wierdl, M.; Danks, M. K.; Wadkins, R. M.; Potter, P. M. Selective
inhibition of carboxylesterases by isatins, indole-2,3-diones. J. Med.
Chem. 2007, 50, 1876–1885.
(25) Potter, P. M.; Wadkins, R. M. Carboxylesterasessdetoxifying enzymes
and targets for drug therapy. Curr. Med. Chem. 2006, 13, 1045–1054.
(26) Lundstedt, T.; Seifert, E.; Abramo, L.; Thelin, B.; Nystrom, A.;
Pettersen, J.; Bergman, B. Experimental design and optimization.
Chemom. Intell. Lab. Syst. 1998, 42, 3–40.
(27) Hicks, L. D.; Hyatt, J. L.; Moak, T.; Edwards, C. C.; Tsurkan, L.;
Wierdl, M.; Ferreira, A. M.; Wadkins, R. M.; Potter, P. M. Analysis
of the inhibition of mammalian carboxylesterases by novel fluo-
robenzoins and fluorobenzils. Bioorg. Med. Chem. 2007, 15, 3801–
3817.
(28) Hyatt, J. L.; Stacy, V.; Wadkins, R. M.; Yoon, K. J.; Wierdl, M.;
Edwards, C. C.; Zeller, M.; Hunter, A. D.; Danks, M. K.; Crundwell,
G.; Potter, P. M. Inhibition of carboxylesterases by benzil (diphe-
nylethane-1,2-dione) and heterocyclic analogues is dependent upon
the aromaticity of the ring and the flexibility of the dione moiety.
J. Med. Chem. 2005, 48, 5543–5550.
(29) Wadkins, R. M.; Hyatt, J. L.; Edwards, C. C.; Tsurkan, L.; Redinbo,
M. R.; Wheelock, C. E.; Jones, P. D.; Hammock, B. D.; Potter,
P. M. Analysis of mammalian carboxylesterase inhibition by
trifluoromethylketone-containing compounds. Mol. Pharmacol. 2007,
71, 713–723.
(30) Wadkins, R. M.; Morton, C. L.; Weeks, J. K.; Oliver, L.; Wierdl,
M.; Danks, M. K.; Potter, P. M. Structural constraints affect the
metabolism of 7-ethyl-10-[4-(1-piperidino)-1-piperidino]carbony-
loxycamptothecin (CPT-11) by carboxylesterases. Mol. Pharmacol.
2001, 60, 355–362.
(31) Brandstetter, H.; Turk, D.; Hoeffken, H. W.; Grosse, D.; Sturzebecher,
J.; Martin, P. D.; Edwards, B. F.; Bode, W. Refined 2.3 A X-ray crystal
structure of bovine thrombin complexes formed with the benzamidine
and arginine-based thrombin inhibitors NAPAP, 4-TAPAP and MQPA.
A starting point for improving antithrombotics. J. Mol. Biol. 1992,
226, 1085–1099.
(32) Sturzebecher, J.; Hauptmann, J.; Steinmetzer, T., Thrombin. In
Proteinase and Peptidase Inhibition: Recent Potential Targets for Drug
DeVelopment; Smith, H. J.; Simmons, C., Eds.; Taylor & Francis:
London, 2002; pp 185-201.
(33) Okamoto, S.; Kinjo, K.; Hijikata, A.; Kikumoto, R.; Tamao, Y.;
Ohkubo, K.; Tonomura, S. Thrombin inhibitors. 1. Ester derivatives
of N-alpha-(arylsulfonyl)-L-arginine. J. Med. Chem. 1980, 23, 827–
830.
(34) Sturzebecher, J.; Markwardt, F.; Voigt, B.; Wagner, G.; Walsmann,
P. Cyclic amides of N-alpha-arylsulfonylaminoacylated 4-amidinophe-
nylalaninestight binding inhibitors of thrombin. Thromb. Res. 1983,
29, 635–642.
(35) Morton, C. L.; Potter, P. M. Comparison of Escherichia coli,
Saccharomyces cereVisiae, Pichia pastoris, Spodoptera frugiperda and
COS7 cells for recombinant gene expression: application to a rabbit
liver carboxylesterase. Mol. Biotechnol. 2000, 16, 193–202.
(36) Hyatt, J. L.; Wadkins, R. M.; Tsurkan, L.; Hicks, L. D.; Hatfield,
M. J.; Edwards, C. C.; Ii, C. R.; Cantalupo, S. A.; Crundwell, G.;
Danks, M. K.; Guy, R. K.; Potter, P. M. Planarity and constraint
of the carbonyl groups in 1,2-diones are determinants for selective
inhibition of human carboxylesterase 1. J. Med. Chem. 2007, 50,
5727–5734.
grant CA108775, a Cancer Center core grant CA21765, a NSF
EPSCoR grant EPS-0556308, and by the American Lebanese
Syrian Associated Charities.
Supporting Information Available: Physical and spectral
parameters for the sulfonamide compounds synthesized. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Cashman, J.; Perroti, B.; Berkman, C.; Lin, J. Pharmacokinetics and
molecular detoxification. EnViron. Health Perspect. 1996, 104, 23–
40.
(2) Kunimoto, T.; Nitta, K.; Tanaka, T.; Uehara, N.; Baba, H.; Takeuchi,
M.; Yokokura, T.; Sawada, S.; Miyasaka, T.; Mutai, M. Antitumor
activity of 7-ethyl-10-[4-(1-piperidino)-1-piperidino]carbonyloxy-
camptothecin, a novel water-soluble derivative of camptothecin, against
murine tumors. Cancer Res. 1987, 47, 5944–5947.
(3) Pindel, E. V.; Kedishvili, N. Y.; Abraham, T. L.; Brzezinski, M. R.;
Zhang, J.; Dean, R. A.; Bosron, W. F. Purification and cloning of a
broad substrate specificity human liver carboxylesterase that catalyzes
the hydrolysis of cocaine and heroin. J. Biol. Chem. 1997, 272, 14769–
14775.
(4) Bencharit, S.; Morton, C. L.; Xue, Y.; Potter, P. M.; Redinbo, M. R.
Structural basis of heroin and cocaine metabolism by a promiscuous
human drug-processing enzyme. Nat. Struct. Biol. 2003, 10, 349–356.
(5) Kamendulis, L. M.; Brzezinski, M. R.; Pindel, E. V.; Bosron, W. F.;
Dean, R. A. Metabolism of cocaine and heroin is catalyzed by the
same human liver carboxylesterases. J. Pharmacol. Exp. Ther. 1996,
279, 713–717.
(6) Brzezinski, M. R.; Spink, B. J.; Dean, R. A.; Berkman, C. E.; Cashman,
J. R.; Bosron, W. F. Human liver carboxylesterase hCE-1: binding
specificity for cocaine, heroin, and their metabolites and analogs. Drug
Metab. Dispos. 1997, 25, 1089–96.
(7) Shi, D.; Yang, J.; Yang, D.; LeCluyse, E. L.; Black, C.; You, L.;
Akhlaghi, F.; Yan, B. Anti-influenza prodrug oseltamivir is activated
by carboxylesterase human carboxylesterase 1, and the activation is
inhibited by antiplatelet agent clopidogrel. J. Pharmacol. Exp. Ther.
2006, 319, 1477–84.
(8) Zhang, J.; Burnell, J. C.; Dumaual, N.; Bosron, W. F. Binding and
hydrolysis of meperidine by human liver carboxylesterase hCE-1.
J. Pharmacol. Exp. Ther. 1999, 290, 314–318.
(9) Alexson, S. E.; Diczfalusy, M.; Halldin, M.; Swedmark, S. Involvement
of liver carboxylesterases in the in vitro metabolism of lidocaine. Drug
Metab. Dispos. 2002, 30, 643–647.
(10) Tsuji, T.; Kaneda, N.; Kado, K.; Yokokura, T.; Yoshimoto, T.; Tsuru,
D. CPT-11 converting enzyme from rat serum: purification and some
properties. J. Pharmacobiodynamics 1991, 14, 341–349.
(11) Potter, P. M.; Pawlik, C. A.; Morton, C. L.; Naeve, C. W.; Danks,
M. K. Isolation and partial characterization of a cDNA encoding a
rabbit liver carboxylesterase that activates the prodrug Irinotecan (CPT-
11). Cancer Res. 1998, 52, 2646–2651.
(12) Humerickhouse, R.; Lohrbach, K.; Li, L.; Bosron, W.; Dolan, M.
Characterization of CPT-11 hydrolysis by human liver carboxylesterase
isoforms hCE-1 and hCE-2. Cancer Res. 2000, 60, 1189–1192.
(13) Khanna, R.; Morton, C. L.; Danks, M. K.; Potter, P. M. Proficient
metabolism of CPT-11 by a human intestinal carboxylesterase. Cancer
Res. 2000, 60, 4725–4728.
(14) Wierdl, M.; Tsurkan, L.; Hyatt, J. L.; Hatfield, M. J.; Edwards, C. C.;
Danks, M. K.; Redinbo, M. R.; Potter, P. M. An improved human
carboxylesterase for use in enzyme/prodrug therapy with CPT-11.
Cancer Gene Ther. 2008, 15, 183–192.
(15) Tabata, T.; Katoh, M.; Tokudome, S.; Nakajima, M.; Yokoi, T.
Identification of the cytosolic carboxylesterase catalyzing the 5′-deoxy-
5-fluorocytidine formation from capecitabine in human liver. Drug
Metab. Dispos. 2004, 32, 1103–1110.
(16) Quinney, S. K.; Sanghani, S. P.; Davis, W. I.; Hurley, T. D.; Sun, Z.;
Murry, D. J.; Bosron, W. F. Hydrolysis of capecitabine to 5′-deoxy-
5-fluorocytidine by human carboxylesterases and inhibition by lop-
eramide. J. Pharmacol. Exp. Ther. 2005, 313, 1011–1016.
(17) Stampfli, H. F.; Quon, C. Y. Polymorphic metabolism of flestolol and
other ester containing compounds by a carboxylesterase in New
Zealand white rabbit blood and cornea. Res. Commun. Mol. Pathol.
Pharmacol. 1995, 88, 87–97.
(18) Morton, C. L.; Iacono, L.; Hyatt, J. L.; Taylor, K. R.; Cheshire, P. J.;
Houghton, P. J.; Danks, M. K.; Stewart, C. F.; Potter, P. M.
Metabolism and antitumor activity of CPT-11 in plasma esterase-
deficient mice. Cancer Chemother. Pharmacol. 2005, 56, 629–636.
(19) Beroza, P.; Damodaran, K.; Lum, R. T. Target-related affinity profiling:
Telik’s lead discovery technology. Curr. Top. Med. Chem. 2005, 5,
371–381.
(37) Guichard, S.; Morton, C. L.; Krull, E. J.; Stewart, C. F.; Danks,
M. K.; Potter, P. M. Conversion of the CPT-11 metabolite APC to
SN-38 by rabbit liver carboxylesterase. Clin. Cancer Res. 1998, 4,
3089–3094.
(38) Stewart, J. J. MOPAC: a semiempirical molecular orbital program.
J. Comput.-Aided Mol. Des. 1990, 4, 1–105.

3752 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 12
Hicks et al.
(39) Vedani, A.; Dobler, M. Multidimensional QSAR: moving from three-
to five-dimensional concepts. Quant. Struct.-Act. Relat. 2002, 21,
382–390.
(40) Vedani, A.; Dobler, M. 5D-QSAR: the key for simulating induced
fit. J. Med. Chem. 2002, 45, 2139–2149.
(41) Vedani, A.; Dobler, M.; Lill, M. A. Combining protein modeling and
6D-QSAR simulating the binding of structurally diverse ligands to
the estrogen receptor. J. Med. Chem. 2005, 48, 3700–3703.
(42) Hawkins, G. D.; Giesen, D.; Lynch, G.; Chambers, C.; Rossi, I.; Storer,
J.; Li, J.; Zhu, T.; Thompson, J.; Winget, P.; Lynch, B.; Rinaldi, D.;
Liotard, D.; Cramer, C. J.; Truhlar, D. G., V. Amsol, version 7.1;
University of Minnesota: Minneapolis, 2004.
(43) Merritt, E. A.; Bacon, D. J. Raster 3D: Photorealistic molecular
graphics. Methods Enzymol. 1997, 277, 505–524.
(44) Kraulis, P. J. MOLSCRIPT: A program to produce both detailed
and schematic plots of protein structures. J. Appl. Cryst. 1991, 24,
946–950.
JM9001296