Can silicon make an excellent drug even better? An in vitro and in vivo head-to-head comparison between loperamide and its silicon analogue sila-loperamide

Article abstract of DOI:10.1002/cmdc.201500040

Loperamide (1a), an opioid receptor agonist, is in clinical use as an antidiarrheal agent. Carbon/silicon exchange (sila-substitution) at the 4-position of the piperidine ring of 1a (R₃COH→R₃SiOH) leads to sila-loperamide (1b). Sila-loperamide was synthesized in a multistep procedure, starting from triethoxyvinylsilane and taking advantage of the 4-methoxyphenyl (MOP) unit as a protecting group for silicon. The in vitro and in vivo pharmacokinetic (PK) and pharmacodynamic (PD) properties of the C/Si analogues 1a and 1b were determined and compared. Despite significant differences in the in vitro PK properties of loperamide and sila-loperamide regarding clearance, permeability, and efflux, both compounds exhibited nearly identical in vivo PK profiles. The increase in metabolic stability of the silicon compound 1b observed in vitro seems to be counterbalanced by an increase in efflux and diminished permeability compared to the parent carbon compound 1a. Overall, sila-loperamide exhibits high unbound clearance (CL_u), leading to a significant decrease in unbound concentration (C_u) and unbound area under the curve (AUC_u) after oral exposure, compared to loperamide. In vitro and in vivo metabolic studies showed an altered profile of biotransformation for the silicon compound 1b, leading to the formation of a more polar and quickly cleared metabolite and preventing the formation of the silicon analogue of the neurotoxic metabolite observed for the parent carbon compound 1a. These differences can be correlated with the different chemical properties of the C/Si analogues 1a and 1b. This study provides some of the most detailed insights into the effects of a carbon/silicon switch and how this carbon/silicon exchange affects overall drug properties.

Full text of DOI:10.1002/cmdc.201500040

DOI: 10.1002/cmdc.201500040
Full Papers
Can Silicon Make an Excellent Drug Even Better? An in
vitro and in vivo Head-to-Head Comparison between
Loperamide and Its Silicon Analogue Sila-Loperamide
[
a]
[b]
[b]
[a]
[c]
Marcel Geyer, Eric Wellner, Ulrik Jurva, Sebastian Saloman, Duncan Armstrong, and
[a]
Reinhold Tacke*
Loperamide (1a), an opioid receptor agonist, is in clinical use
compared to the parent carbon compound 1a. Overall, sila-lo-
as an antidiarrheal agent. Carbon/silicon exchange (sila-substi-
peramide exhibits high unbound clearance (CL ), leading to
u
tution) at the 4-position of the piperidine ring of 1a (R COH!
a significant decrease in unbound concentration (C ) and un-
3
u
R SiOH) leads to sila-loperamide (1b). Sila-loperamide was syn-
bound area under the curve (AUC ) after oral exposure, com-
3
u
thesized in a multistep procedure, starting from triethoxyvinyl-
silane and taking advantage of the 4-methoxyphenyl (MOP)
unit as a protecting group for silicon. The in vitro and in vivo
pharmacokinetic (PK) and pharmacodynamic (PD) properties of
the C/Si analogues 1a and 1b were determined and compared.
Despite significant differences in the in vitro PK properties of
loperamide and sila-loperamide regarding clearance, permea-
bility, and efflux, both compounds exhibited nearly identical in
vivo PK profiles. The increase in metabolic stability of the sili-
con compound 1b observed in vitro seems to be counterbal-
anced by an increase in efflux and diminished permeability
pared to loperamide. In vitro and in vivo metabolic studies
showed an altered profile of biotransformation for the silicon
compound 1b, leading to the formation of a more polar and
quickly cleared metabolite and preventing the formation of
the silicon analogue of the neurotoxic metabolite observed for
the parent carbon compound 1a. These differences can be cor-
related with the different chemical properties of the C/Si ana-
logues 1a and 1b. This study provides some of the most de-
tailed insights into the effects of a carbon/silicon switch and
how this carbon/silicon exchange affects overall drug proper-
ties.
Introduction
2
,2-Diphenyl-4-(4-aryl-4-hydroxypiperidino)butyramides were
into a drug can affect and, ideally, improve the pharmacody-
namics (PD) and pharmacokinetics (PK).
developed as antidiarrheal agents in the early 1970s by Jans-
sen and co-workers, and loperamide (1a) was selected for fur-
Herein we report the synthesis of sila-loperamide (1b) and
present a detailed comparison between the physicochemical
properties and the primary in vitro PD and PK properties of
the C/Si analogues 1a and 1b. We also report the profiling of
[
1]
ther clinical investigation. Today, loperamide is the most sold
over-the-counter antidiarrheal agent in the world, with a well-
known metabolic fate.
In context with our systematic studies on silicon-based
[
2,3]
drugs,
we sought to examine the biological effects of sila-
substitution (C/Si exchange) of the quaternary R COH carbon
3
atom of the 4-(4-chlorophenyl)-4-hydroxypiperidino group of
1a. The carbon/silicon switch strategy, i.e., the strategic re-
placement of a carbon atom with a silicon atom within a well-
known drug, with the rest of the molecule remaining unal-
tered, is a well-established method for the design and devel-
[
2,3]
opment of new silicon-based drugs.
Incorporation of silicon
primary safety aspects, such as CYP inhibition, time-dependent
CYP inhibition, CYP reaction phenotyping (CRP), and hERG in-
hibition. We also present an assessment of the impact of the
carbon/silicon switch on permeability, efflux, protein binding,
and reactive metabolite formation. Most importantly, we
report a head-to-head comparison of the in vivo PK and meta-
bolic profiles of the C/Si analogues 1a and 1b. This allows
unique insight into the behavior of a silicon-containing drug
and the translation of its in vitro properties into an in vivo sit-
uation.
[
a] M. Geyer, S. Saloman, Prof. Dr. R. Tacke
Universität Würzburg, Institut für Anorganische Chemie
Am Hubland, 97074 Würzburg (Germany)
E-mail: r.tacke@uni-wuerzburg.de
[b] Dr. E. Wellner, Dr. U. Jurva
AstraZeneca R&D Mçlndal, CVMD Innovative Medicines
Pepparedsleden 1, 43183 Mçlndal (Sweden)
[
c] Dr. D. Armstrong
AstraZeneca, Drug Safety & Metabolism
Alderley Park, Mereside, SK10 4TG (UK)
ChemMedChem 2015, 10, 911 – 924
911
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
thylamino-4-oxo-3,3-diphenylbutyl)-4-(4-methoxyphenyl)-4-sila-
piperidine (9). Treatment of 9 with trifluoromethanesulfonic
acid, followed by reaction with triethylammonium chloride
Results and Discussion
Synthesis
(
cleavage of the 4-methoxyphenyl protecting group to give
Sila-loperamide (1b) was synthesized in a multistep procedure,
starting from commercially available triethoxyvinylsilane (2)
and taking advantage of the 4-methoxyphenyl (MOP) unit as
a protecting group for silicon (Scheme 1). Thus, treatment of 2
with (4-chlorophenyl)magnesium bromide afforded (4-chloro-
phenyl)diethoxyvinylsilane (3), which upon sequential treat-
ment with (4-methoxyphenyl)magnesium bromide and vinyl-
magnesium chloride gave (4-chlorophenyl)(4-methoxyphenyl)-
divinylsilane (4). Reaction of 4 with 9-borabicyclo[3.3.1]nonane
the corresponding chlorosilane) and subsequent treatment
with methanol and triethylamine, gave 4-(4-chlorophenyl)-1-(4-
dimethylamino-4-oxo-3,3-diphenylbutyl)-4-methoxy-4-sila-
piperidine (10) as an intermediate (not further purified; identi-
fied by high-resolution mass spectrometry), which, upon hy-
drolysis, finally afforded the target compound 4-(4-chlorophen-
yl)-1-(4-dimethylamino-4-oxo-3,3-diphenylbutyl)-4-hydroxy-4-si-
lapiperidine (sila-loperamide, 1b).
Compounds 3, 4, and 6 were isolated as colorless liquids,
whereas 1b, 5, 7·HCl, and 9 were obtained as colorless solids.
The identities of all these compounds were established by ele-
mental analyses (3, 4, 5, 6, 7·HCl) or high-resolution mass spec-
[
4]
(9-BBN), followed by sequential treatment with aqueous solu-
tions of sodium hydroxide and hydrogen peroxide, yielded (4-
chlorophenyl)bis(2-hydroxyethyl)(4-methoxyphenyl)silane (5).
Reaction of 5 with methanesulfonyl chloride, in the presence
of triethylamine, and subsequent treatment with a large excess
of allylamine afforded 1-allyl-4-(4-chlorophenyl)-4-(4-methoxy-
phenyl)-4-silapiperidine (6). Reaction of 6 with 1-chloroethyl
1
13
trometry (1b, 9) and by NMR spectroscopic studies ( H, C,
29
and Si).
Physicochemical properties
[
5]
chloroformate, followed by treatment with methanol, yielded
-(4-chlorophenyl)-4-(4-methoxyphenyl)-4-silapiperidinium
chloride (7·HCl). Compound 7·HCl was then treated with dime-
4
Loperamide (1a) and sila-loperamide (1b) were studied for
their octanol/water (pH 7.4) distribution coefficient (logD
[6]
thyl(tetrahydro-3,3-diphenyl-2-furylidene)ammonium bromide
8) and sodium carbonate to give 4-(4-chlorophenyl)-1-(4-dime-
value), dissociation constant in water (pK value), and solubility
a
(
in HBSS buffer (pH 7.4). As can be seen from Table 1, the re-
Scheme 1. Synthesis of sila-loperamide (1b).
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
912
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Physicochemical properties of 1a and 1b.
[a]
[b]
[c]
Compd
logD
pK
a
Solubility [mm]
1
1
a
b
3.53Æ0.03
3.73Æ0.07
9.12Æ0.11
8.93Æ0.26
19Æ3
30Æ9
[
[
a] Determined at pH 7.4; values represent the meanÆSEM (n=3).
b] Values represent the meanÆSEM (1a: n=3; 1b: n=4). [c] Determined
in HBSS buffer (pH 7.4); values represent the meanÆSEM (1a: n=3; 1b:
n=5).
spective data for the C/Si analogues 1a and 1b are each within
the same range, indicating that the carbon/silicon switch has
no major impact on the physicochemical properties. Both com-
pounds are protonated at ~97% at pH 7.4, as shown by their
pK values.
a
In vitro pharmacology
The binding affinities and functional effects of loperamide (1a)
and sila-loperamide (1b) at the human m1 opioid receptor and
rat k1 opioid receptor were investigated. Binding affinities and
agonistic potencies are listed in Table 2. Loperamide is report-
Figure 1. A) Dose–response curves of 1a and 1b at the human m1 opioid re-
ceptor in a cell-based functional assay. B) Competition binding curves of
compounds 1a and 1b at the human m1 opioid receptor. The stronger left
shift of the curve of 1a relative to that of 1b in the agonistic efficacy assay
might result from a more effective engagement of the receptor reserve by
loperamide than by sila-loperamide.
Table 2. Binding affinity (pK
at the human m1 opioid receptor (m1OPR) and rat k1 opioid receptor
k1OPR).
i
) and agonistic potency (pEC50) of 1a and 1b
Affinity and functional potency of 1a and 1b at the rat k1
opioid receptor show a trend similar to that observed for the
human m1 opioid receptor system studied.
(
[
a,b]
[a,c]
Compd
m1OPR
k1OPR
pK
i
pEC50
pK
i
pEC50
1
1
a
9.91Æ0.17
10.19Æ0.15
(0.06 nm)
8.96Æ0.32
(1.1 nm)
7.19Æ0.09
(65 nm)
7.46Æ0.26
(35 nm)
hERG activity
(
0.12 nm)
9.27Æ0.09
0.53 nm)
b
6.44Æ0.06
6.29Æ0.23
The inhibitory effect of loperamide (1a) and sila-loperamide
(
(364 nm)
(517 nm)
(
1b) at the hERG potassium ion channel was tested to assess
[a] Molar values are given in parentheses. [b] Values represent the
[9]
the risk of QT interval prolongation. Loperamide showed no
inhibitory effect at the test concentration of 11 mm in the hERG
spot test. In the same spot test, sila-loperamide had an inhibi-
tory effect of 83%, and therefore the IC50 of 1b was deter-
mined. The silicon compound 1b blocked the potassium chan-
nel with a mean IC50 value of 3.64 mm. This is ~6700-fold
higher than the EC50 value of 1b at the human m1 opioid re-
ceptor.
meanÆSD (n=3). [c] Values represent the meanÆSD (n=4).
ed to have a K value of 3.3 nm for the human m opioid recep-
tor. In our m1 opioid receptor affinity binding assay, lopera-
i
[
7]
mide was much more potent (K =0.12 nm). Sila-loperamide
was found to be fourfold less potent (K =0.53 nm) than lopera-
i
i
mide. The competition binding and functional dose–response
curves of compounds 1a and 1b for the human m1 opioid re-
ceptor are shown in Figure 1. Both 1a and 1b were found to
be full agonists at the m1 opioid receptor, as can be seen from
Figure 1A. However, the potency shift in this functional assay
was more pronounced for sila-loperamide (EC =1.1 nm),
Apparent permeability in human Caco-2 and MDCK-MDR1
cells
[10]
Loperamide (1a) is a peripherally acting drug, and in vitro
and in vivo studies have shown that it is subject to P-glycopro-
tein (P-gp)-mediated efflux. The poor permeability and efflux
50
whereas the potency of the parent carbon compound lopera-
[11]
mide was in the low picomolar range (EC =60 pm). The po-
might limit the uptake of loperamide, resulting in low oral
bioavailability.
50
tency increase in the efficacy assay might result from overex-
pression of the receptor in CHO cells, resulting in a large re-
To obtain information about their apparent permeability
(Papp), loperamide (1a) and sila-loperamide (1b) were studied in
a human Caco-2 cell model (Table 3). For the apical-to-basolat-
eral (A!B) transport, the apparent permeability was moderate
[
8]
ceptor reserve. In context with our assay, the decreased po-
tency of 1b indicates that the silicon analogue is less effective
in activating the receptor than the parent carbon compound
À6
À1
1a.
for 1a (P =2.57”10 cms ) and low for 1b (P =0.63”
app app
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
913
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
À1
À1
and 46.9 mLmin mg for 1b) and medium to high in human
Table 3. Apparent permeability (Papp) and efflux ratio (ER) of 1a and 1b in
human Caco-2 and MDCK-MDR1 cells.
À1
À1
À1
À1
hepatocytes (13.1 mLmin mg for 1a and 8.60 mLmin mg
for 1b), indicating that the uptake into hepatocytes was not
limited by the medium-to-poor permeability and strong efflux
of the compounds. The intrinsic clearance in rat hepatocytes
was significantly higher, resulting in a substantial shortening
À6
À1
[c]
Compd Cells
P
app [10 cms
]
ER
Recovery [%]
A!B
B!A
[
a]
a]
1a
1b
1a
1b
Caco-2
Caco-2
MDCK-MDR1
MDCK-MDR1
2.57Æ0.08 13.8Æ0.38
0.63Æ0.02 7.72Æ0.44
5
50<x<66
30<x<47
40<x<66
20<x<38
[
12
13
34
(
Dt1/2 =t1/2huÀt1/2rat) of the half-life (Dt =46.7 min for 1a and
[
b]
b]
1/2
0.89
0.34
11.2
5.57
[
Dt1/2 =70.2 min for 1b).
[
(
a] Values represent the meanÆSD (n=3). [b] Values represent the mean
n=2). [c] Values represent the range for the A!B and B!A recovery.
CYP inhibition, time-dependent CYP inhibition, and CYP
reaction phenotyping
Loperamide (1a) and sila-loperamide (1b) are weak-to-medium
inhibitors of several CYP isoforms. The results of our studies of
CYP inhibition are summarized in Table 5. Both CYP2D6 (1a,
À6
À1
10
cms ). The efflux ratio ([P A!B]/[P B!A]) was high
app
app
for both compounds (ratio: 5 for 1a and 12 for 1b). Results
from the MDCK-MDR1 cell line confirmed that both com-
pounds are substrates for P-gp-mediated efflux (ratio: 13 for
1
a and 34 for 1b). The A!B permeability in this cell line was
Table 5. Cytochrome P450 inhibition (IC ) by 1a and 1b.
5
0
low for both compounds. Overall, these data suggest that the
oral bioavailability of loperamide and sila-loperamide might be
limited due to their 1) low permeability and 2) high efflux
[a]
Compd
IC50 [mm]
CYP3A4 CYP2C9 CYP1A2 CYP2C19 CYP2D6 CYP2C8
[
12]
1a
1.7Æ0.5 11 Æ 0.5
>20
11 Æ 0.2
4.8Æ0.7
0.9Æ0.1 >20
0.8Æ0.1 >20
ratio. The Caco-2 data indicate an even more pronounced
permeability limitation for the silicon compound 1b, which
could translate into improved safety margins for sila-lopera-
mide and decreased partition into the central nervous system
1
b
1.0Æ0.1 9.6Æ1.1 >20
[
a] Values represent the meanÆSD (n=3).
(
CNS). The recovery in Caco-2 and MDCK-MDR1 cells for both
A!B and B!A transport is generally low, indicating metabo-
lism of 1a and 1b within these cells or adsorption of the com-
pounds. Regarding the moderate intrinsic clearance of 1a and
IC =0.9 mm; 1b, IC =0.8 mm) and CYP3A4 (1a, IC =1.7 mm;
50 50 50
1b, IC =1.0 mm) exhibit the strongest inhibition by 1a and 1b.
50
No time-dependent inhibition of any isoform was observed for
either compound. This means that neither 1a nor 1b form me-
tabolites, which are capable of significantly inhibiting any of
the CYPs investigated.
1b in human hepatocytes (see below), adsorption of the com-
pounds at the surface is the more probable cause for the low
recovery rate.
CYP reaction phenotyping (CRP) is a tool to investigate the
contribution of various CYPs to total metabolic clearance. This
information can be used to evaluate the risk of drug–drug in-
Intrinsic clearance and half-life in human and rat hepato-
cytes and human liver microsomes
[13]
teractions. According to Kim et al., loperamide is metabo-
lized mainly by CYP3A4 and CYP2C8. Figure 2 summarizes our
results of CRP. It has been shown that while CYP3A4 is the pre-
dominant hepatic form of CYP3A, in many cases CYP3A5 con-
The clearance rates of loperamide (1a) and sila-loperamide
(1b) were measured in human liver microsomes (Table 4). Al-
though the microsome fractions contain many drug-metaboliz-
ing enzymes such as cytochrome P450s (CYPs), flavin monoox-
ygenases, carboxylesterases, and epoxide hydrolases, they do
not allow assessment of the possibility of phase II biotransfor-
mation. Therefore, to assess the potential impact of phase II
metabolism, we also measured the metabolic stability of the
C/Si analogues 1a and 1b in human and rat hepatocytes
[14]
tributes significantly to the total liver clearance. Therefore,
CYP3A5 was included in the CRP panel to provide a more dif-
ferentiated view on total CYP3A contribution relative to pub-
lished work thus far. Both loperamide and sila-loperamide are
metabolized by CYP3A4 to >95%. CYP3A5 contributes by
~1% to CYP oxidation, followed by CYP2C8. Sila-loperamide
displays a nearly identical CRP profile to that of the parent
carbon compound 1a. It offers no major alternative oxidation
pathway to escape the ex-
(
Table 4). The intrinsic clearance (CL ) was determined to be
int
À1
À1
medium in human liver microsomes (44.3 mLmin mg for 1a
tensive CYP3A4 metabo-
Table 4. Intrinsic clearance (CLint) and half-life (t1/2) of 1a and 1b in human/rat hepatocytes and human liver micro-
somes.
lism.
Compd
hu hep
rat hep
hu liv mic
CLint [mLmin mg ] t1/2 [min]
À1
À1
À1
À1
À1
À1
CLint [mLmin mg ] t1/2 [min]
CLint [mLmin mg ] t1/2 [min]
Rat in vivo PK and
[
a]
[a]
[b]
[c]
[c]
[c]
[c]
[c]
plasma protein binding
1
1
a
b
13.07Æ1.13
8.60Æ0.52
54.28Æ4.72
82.19Æ5.28
91.57Æ5.71
58.51Æ4.50
7.60Æ0.47
11.99Æ0.90
44.27Æ9.46
46.90Æ0.47
15.53Æ3.16
14.78Æ0.15
[
b]
[c]
[c]
[c]
Rat in vivo PK data for lo-
[
a] Values represent the meanÆSD (n=4). [b] Values represent the meanÆSD (n=6). [c] Values represent the meanÆ
peramide (1a) was report-
ed in 2012. Table 6 sum-
SD (n=3).
[11]
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
914
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
distribution into tissue. The half-life (t ) of 1a is 2 h (dose:
1/2
À1
À1
5
.1 mmolkg i.v.). For an oral dose of 10.1 mmolkg , the ob-
served half-life was much higher (t1/2 =11.5 h). Delayed dissolu-
tion, continuous absorption, decrease in gut motility, and the
[16,17]
enterohepatic recirculation of 1a
are possible explanations
for the prolongation of the half-life after oral dosing.
The C/Si analogues 1a and 1b have a similar solubility and
lipophilicity. Furthermore, both compounds are strong P-gp
substrates, having medium-to-poor permeability. The strong
efflux observed for 1a and 1b in Caco-2 and MDCK-MDR1 cells,
combined with their limited permeability and low solubility,
seems to substantially restrict their oral bioavailability and sys-
temic exposure in our models.
After i.v. and p.o. administration, sila-loperamide exhibits an
in vivo PK profile nearly identical to that of the parent carbon
compound loperamide. Parameters such as CL, V , and t are
Figure 2. CRP of compounds 1a and 1b. Both compounds are mainly metab-
olized by CYP3A4 and CYP3A5. While the silicon compound 1b is a weak
substrate for CYP2C19, it is more selective against CYP2E1 and CYP2D6 than
the parent carbon compound 1a.
SS
1/2
in the same range for both compounds. Differences can be ob-
served in oral bioavailability, which is 8% for sila-loperamide,
and total concentrations in plasma (Cmax =0.04 mm). Consider-
ing the lower unbound fraction of the silicon compound 1b in
[18]
Table 6. In vivo PK data of 1a in Sprague–Dawley rats and 1a and 1b in
Han Wistar rats.
the plasma, the unbound clearance (CLu) of 1b is ~3.5-fold
higher than that of its parent carbon compound 1a. The high
[
a]
[b]
[b]
CL value leads to a significant drop in the C
u
u,max
value of 1b
Parameter
1a
1a
1b
[19]
(
0.6 nm) relative to 1a (1.2 nm).
Ultimately, this results in
i.v. dosing
CL [mLmin kg
À1
À1
[c]
[d]
[d]
a
lower area under the curve unbound (AUCu; 1b,
]
128
2667
[c]
30
27
À1
À1
À1
À1
[c]
[d]
[d]
CL
u
[mLmin kg
]
517
[d]
1800
[d]
0.01 mmolh ; 1a, 0.02 mmolh ) for 1b despite the slight in-
crease in oral bioavailability. It is well understood that P-gp-
mediated secretion of a compound into the intestine and low
permeability can have a profound impact by increasing the in
vivo blood clearance (CL). In such cases, CL is often underpre-
À1
VSS [Lkg
]
9
1
4
2
4
3
[
c]
[d]
[d]
t1/2 [h]
furat [%]
fuhu [%]
4.8
8.1
5.8
8.1
1.5
2.8
[20]
p.o. dosing
F [%]
dicted by the in vitro intrinsic clearance (CLint). For lack of
Mdr1a knockout rats, we tried to assess the qualitative impact
of transporters and permeability on the in vivo clearance of 1a
and 1b by comparing it with the predicted clearance of the
compounds. Applying the well-stirred model,
[
c]
[d]
[d]
43
13
4
8
15
[
c]
[d]
[d]
t1/2 [h]
11.5
0.02
1.2
[
c]
[d]
[d]
Cmax [mm]
0.19
c]
0.04
0.6
[
[d]
[d]
Cu,max [nm]
9
À1
[c]
[d]
[d]
[21]
AUC [mmolh
AUC [mmolh ]
u
]
À1
1.69
0.08
0.39
0.02
0.72
0.01
the in vitro
CLint of 1b in hepatocytes results in a predicted blood clear-
[
22]
À1
À1
[
a] Sprague–Dawley; results taken from reference [11]. [b] Han Wistar;
values represent the mean (n=2). [c] Dose: 5.5 mmolkg
1 mmolkg p.o. [d] Dose: 5.1 mmolkg i.v. or 10.1 mmolkg p.o.
ance CLH of 50 mLmin kg and reflects the observed in
À1
i.v. or
À1
À1
vivo blood clearance (CL=30 mLmin kg ) surprisingly well,
À1
À1
À1
2
[20]
despite the high efflux ratio and low permeability of 1b. The
intrinsic clearance of the parent carbon compound 1a ob-
served in rat hepatocytes leads to an overprediction of the in
À1
À1
vivo clearance (CL=30 mLmin kg ) versus predicted blood
À1
À1
marizes the published PK data for 1a in male Sprague–Dawley
rats and our own data for 1a and its silicon analogue sila-loper-
amide (1b) in male Han Wistar rats. Differences observed be-
tween published data and our in-house data might result from
differences in dosing or more likely from the use of different
rat strains.
clearance (CL =52 mLmin kg ). Overall, the blood clearance
H
of the C/Si analogues 1a and 1b is predicted well by their in-
trinsic in vitro hepatocyte clearance, indicating that efflux and
low permeability do not particularly impact the blood clear-
ance, or are counterbalanced by other mechanisms such as re-
distribution from other compartments.
In our hands, loperamide has an oral bioavailability (F) of
Considering that the motility effect of loperamide is mainly
achieved via the opioid receptor in the gut, the lower Cu,max
À1
À1
4
5
%
and
a
À1
clearance (CL) of 30 mLmin kg
[15]
(dose:
.1 mmolkg i.v.). The rat plasma binding (fu ) of 1a was
and AUC values of sila-loperamide could improve safety mar-
rat
u
determined in Han Wistar rat plasma to be 5.8%, with a recov-
ery rate of 95%, indicating the high stability of loperamide in
rat plasma. The total plasma concentration (Cmax) after oral ad-
gins, particularly with regard to side effects mediated by the
[23]
opioid receptors in the brain. Figure 3 shows the ratio of Cu
after oral dosing and the EC50 values reported in Table 2, and it
represents the free exposure above the EC50 value. While the
free exposure of 1a at the human m1 opioid receptor is ~27-
fold above the EC50 value, the low potency at the rat k1 opioid
À1
ministration (dose: 10.1 mmolkg p.o.) is 0.02 mm, resulting in
total plasma concentration unbound (Cu,max) of 1.2 nm. The
À1
volume of distribution (V ) is high (4 Lkg ), indicating a broad
ss
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
915
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
receptor leads to no free exposure of the compound above its
EC50 value. The free concentrations of the silicon analogue 1b
are just about to reach the EC50 value at the m1 opioid recep-
tor, but are almost 1000-fold below the EC50 value at the k1
opioid receptor. Therefore, these data suggest that sila-lopera-
mide provides a decrease in the free drug exposure after oral
administration in rats, possibly leading to improved safety mar-
gins. In combination with the high efflux ratio and low perme-
ability observed in Caco-2 and MDCK-MDR1 cells, the exposure
of 1b into the CNS might be even further decreased.
Identification of major in vitro metabolites and reactive me-
tabolites
Two major metabolites of loperamide (1a), M1a and M2a,
were identified in human hepatocytes (Table 7, Scheme 2). Sim-
[24]
ilar observations were made in human liver microsomes. In-
terestingly, we did not observe any formation of conjugates in
the hepatocytes, which means that the aforementioned
CYP3A4 metabolism seems to be the predominant route of
biotransformation for both 1a and 1b. Considering the meta-
[
3c,25,26]
bolic fate of haloperidol,
which is metabolized into the
+
potentially neurotoxic pyridinium cation HPP , we were inter-
+
ested in the formation of the possible neurotoxic LPP metab-
[
24]
olite M4a (Scheme 3).
human hepatocytes, M4a was observed to a minor extent
0.6%). In addition, the minor loperamide metabolites M3a
After 40 min incubation of 1a in
Figure 3. Ratio between measured free plasma concentration C
EC50, plotted against time, for compounds 1a and 1b. A) -Fold exposure of
a and 1b at the human m1 opioid receptor. B) -Fold exposure of 1a and 1b
u
and agonist
(
1
and M5a were detected. The main metabolites of sila-lopera-
mide (1b) in human hepatocytes are M1b and M2b, the silicon
analogues of the main loperamide metabolites M1a and M2a
at the rat k1 opioid receptor. Both compounds were orally administered at
À1
1
2.2 mmolkg . The horizontal dotted lines represent C
u
/EC50 =1 in each case
and indicate the limit for free exposure above or below the measured ago-
nist EC50 value. The two concentration maxima observed in the concentra-
tion versus time curves of 1a might indicate enterohepatic recirculation.
(
Table 8, Scheme 2). Further, a-oxidation (relative to the nitro-
gen atom) at the silapiperidine ring, followed by ring-opening
Table 7. Loperamide (1a) metabolites after an incubation period of 40 min in human hepatocytes.
Metabolite
t
R
[min]
m/z Values (HRMS)
Estimated
Protonated parent species
Major MS–MS ions
formation [%]
1
a
3.25
2.87
3.02
2.68
3.61
3.07
477.2303
493.2256
463.2142
479.2107
455.1847
491.2101
477.2319, 266.1568, 238.1238, 210.1290
493.2265, 463.2140, 282.1490, 252.1384, 196.1123
463.2149, 252.1391, 224.1088, 196.1124
479.2117, 449.1976, 268.1331, 238.1226, 182.0975
455.1829, 430.4557, 297.0898, 266.0975
450.1807, 280.1351, 252.1382, 224.0900, 196.1107
64
23
13
0.6
0.6
<0.1
M1a
M2a
M3a
M4a
M5a
Scheme 2. Proposed metabolic pathways of loperamide (1a) in human hepatocytes. The structures of the proposed metabolites M1a, M2a, M3a, M4a, and
M5a are tentative assignments based on accurate mass measurements and interpretation of MS–MS spectra.
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
916
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 8. Sila-loperamide (1b) metabolites after an incubation period of 40 min in human hepatocytes.
Metabolite
t
R
[min]
m/z Values (HRMS)
Estimated
Protonated parent species
Major MS–MS ions
formation [%]
1
b
3.42
3.03
3.19
2.86
3.11
3.25
493.2076
509.2021
479.1907
495.1862
483.1855
507.1861
493.2069, 266.1555, 238.1233, 210.1284
509.2010, 479.1899, 282.1499, 252.1390, 196.1126
479.1925, 252.1388, 196.1126
495.1855, 465.1765, 420.1179, 268.1347, 238.1233
483.1845, 420.1191, 266.1543, 238.1238, 210.1282
507.1884, 466.1593, 420.1189, 252.1394, 196.1118
78
12
3
0.3
5
M1b
M2b
M3b
M4b
M5b
1
Human microsome metabolite trapping reactions
to identify glutathione conjugates and nitrile or me-
thoxyamine adducts of 1a and 1b were all negative
and did not reveal the formation of any reactive me-
tabolites.
Identification of major metabolites in vivo in
plasma from male Han Wistar rats
The plasma samples from the in vivo PK experiments
(
see Rat in vivo PK and plasma protein binding section
above) were analyzed with respect to metabolites as
well as parent drug to allow the generation of meta-
bolic profiles. The relation between loperamide (1a)
and sila-loperamide (1b) and their metabolites identi-
fied in rat plasma following i.v. administration of 1a
Scheme 3. Metabolism of loperamide (1a). Left: formation of the neurotoxic pyridinium
metabolite LPP (M4a). Right: formation of sila-LPP from sila-loperamide (1b) is not
possible owing to chemical reasons (unstable Si=C double bond).
À1
À1
(
dose: 5.5 mmolkg ) and 1b (dose: 5.1 mmolkg ) is
+
[24]
+
shown in Figure 4. For both compounds, the two
major metabolites M1a/M1b and M2a/M2b from the
in vitro metabolite identification experiments
reaction, leads to the formation of a third major metabolite,
the silanediol M4b. This result suggests that sila-loperamide
has access to an alternative metabolic pathway leading to
a more polar metabolite than loperamide itself. In addition,
the two minor sila-loperamide metabolites M3b and M5b (the
silicon analogues of the loperamide metabolites M3a and
M5a) were detected. Because it is well understood that silapi-
peridine analogues cannot undergo a bioactivation to a silapyri-
(Tables 7 and 8, Schemes 2 and 4) were also identified as major
circulating metabolites in vivo. The origin of these metabolites
is hydroxylation on the dimethylamino moiety, and clearly, this
is a favored metabolic pathway for both 1a and 1b in vitro as
well as in vivo. Metabolites M2a/M2b are formed chemically
from M1a/M1b, and it should be noted that the difference in
levels between M1a and M1b could originate from small differ-
ences in sample handling. The metabolites M1a/M1b and
M2a/M2b were relatively stable in the circulation. Inspection
[
3e]
+
dinium cation, the generation of sila-LPP is not expected.
Scheme 4. Proposed metabolic pathways of sila-loperamide (1b) in human hepatocytes. The structures of the proposed metabolites M1b, M2b, M3b, M4b,
and M5b are tentative assignments based on accurate mass measurements and interpretation of MS–MS spectra.
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
917
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
type metabolite M4a was not observed in vivo, whereas rela-
tively low levels of the minor in vitro metabolites M5a and
M5b were detected.
Under the assumption that the AUC of the PK profiles in
Figure 4 reflect, to some extent, the concentration over time
and number of the main metabolites in the body, it appears
that the burden of the metabolites of sila-loperamide in the
blood is lower than that of the main metabolites of lopera-
mide.
Conclusions
Sila-loperamide (1b), a silicon analogue of the opioid receptor
agonist loperamide (1a), was prepared in a multistep synthesis,
starting from triethoxyvinylsilane. In this synthesis, the use of
the 4-methoxyphenyl (MOP) moiety as a protecting group for
silicon played a key role. The physicochemical properties of
the C/Si analogues 1a and 1b were found to be nearly identi-
cal. The PD profiles of 1a and 1b were studied at the human
m1 opioid and rat k1 opioid receptors. Both compounds exhib-
ited sub-nanomolar binding affinities and agonist potency at
the m1 receptor, which is the primary target of loperamide. The
compounds are full agonists at both receptors studied. Signifi-
cant differences were observed in their ability to permeate
Caco-2 cell membranes, where sila-loperamide showed lower
permeability and a higher efflux ratio than loperamide. Both
compounds are strong substrates for the P-gp transporter
(
MDR1/ABCB1). Clearance in human and rat hepatocytes is not
limited by the low permeability of 1a and 1b. Loperamide and
sila-loperamide are both readily oxidized in human liver micro-
somes. The CRP profile of both compounds confirms that the
CYP3A4 isoform is the main enzyme responsible for metabolic
clearance in vitro. No time-dependent CYP inhibition was ob-
served, and the CYP inhibitory potency was found to be
Figure 4. A) PK profiles of compound 1a and its metabolites identified in rat
plasma following i.v. administration of 1a at 5.5 mmolkg . B) PK profiles of
À1
compound 1b and its metabolites identified in rat plasma following i.v. ad-
À1
ministration of 1b at 5.1 mmol kg
.
<
1 mm only for CYP2D6. The in vivo PK profile of the C/Si ana-
of the later time points in Figure 4 allows a rough estimation
of their clearance, which, for these metabolites, appears to be
similar to or slightly lower than that of the parent drugs (1a,
logues 1a and 1b was studied in male Han Wistar rats and
compared with available published data for 1a. Basic PK pa-
rameters such as clearance (CL), volume of distribution (Vss),
p.o. and i.v. half-life (t1/2), and oral bioavailability (F) were simi-
lar for both drugs. However, the lower unbound fraction of
sila-loperamide in the plasma protects 1b from elimination,
À1
À1
À1
À1
CL=30 mLmin kg ; 1b, CL=27 mLmin kg ). Hydroxyl-
ation on the N-methyl moiety of M2a/M2b is the origin of the
minor in vitro metabolite M3a/M3b. Whereas M3a was a major
metabolite in the circulation, the corresponding sila-analogue
M3b was not detected in the plasma. One possible reason for
this difference is that M3b is not formed in vivo. On a more
speculative note, M3b, in contrast to M3a, could be a good
substrate for active transport into urine and/or bile. The third
major metabolite of sila-loperamide in the in vitro experiments,
silanediol M4b, formed presumably via hydroxylation on the si-
lapiperidine moiety at the a-position to the nitrogen atom fol-
lowed by hydrolytic ring opening, was also detected in vivo as
a major metabolite at the early time points. In contrast to the
relatively stable metabolites M1b and M2b, metabolite M4b
displayed a completely different PK profile and was rapidly
cleared from the circulation. There could be several reasons for
such a PK profile: rapid elimination of M4b into urine and/or
bile or metabolism to secondary metabolites that could not be
detected with the applied LC–MS conditions. The pyridinium-
whereas the CL value is much higher for 1b than for 1a. Ulti-
u
mately, this leads to lower Cu,max and AUC values for 1b than
u
for 1a. The combination of increased efflux ratio and decreased
permeability and Cu observed for 1b might result in a de-
creased exposure of 1b into the CNS and increased safety mar-
gins. Furthermore, there are marked differences in the in vitro
and in vivo metabolite pattern between loperamide and sila-
loperamide. Whereas two main in vivo metabolites (M1a/M1b,
M2a/M2b) and one minor in vivo metabolite (M5a/M5b) for
1a and 1b represent C/Si analogues, the formation of the sila-
analogue of the metabolite M3a was not observed. Instead,
sila-loperamide forms a third main metabolite, silanediol M4b.
This metabolite is readily cleared in vivo, whereas the three
other less polar metabolites have much longer half-lives. This
demonstrates that sila-loperamide has access to different met-
abolic biotransformations, leading to an altered elimination
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
918
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
compared to the parent carbon compound loperamide, there-
by decreasing the overall burden of formed metabolites over
time. Interestingly, no conjugate formation in hepatocytes or
in vivo was observed for either 1a or 1b.
ratus (Elementar Analysensysteme GmbH) or a EURO EA Elemental
Analyzer (Euro Vector). High-resolution mass spectra (HRMS, ESI+)
were recorded on a Waters LCTP mass spectrometer using solu-
tions in DMSO.
In summary, it is increasingly unlikely in today’s market envi-
ronment that a carbon/silicon switch would replace a safe and
inexpensive drug. However, with improved synthetic access to
organosilicon compounds and a better understanding of their
physicochemical and PK and PD properties in vitro and in vivo,
organosilicon compounds can be successfully used to modify
several properties using smart drug design. For example, it is
possible to introduce metabolic soft spots, avoid formation of
reactive glucuronides, and alter the in vivo pharmacology (in
this context, see also reference [31]). Although carbon/oxygen
or carbon/nitrogen switches are well-established approaches in
medicinal chemistry, the carbon/silicon switch offers an addi-
tional option with a much higher probability of retaining PD
activity than the other switches, due to more subtle changes
in structural and electronic properties. On the other hand, dif-
ferences in metabolic pathways and changes in the PK proper-
ties can be significant, given the different chemical properties
of C/Si analogues. Our data provide clear evidence that the
carbon/silicon switch strategy is more than a “patent busting”
activity; it is a powerful tool to improve the overall properties
of drugs. In answer to the title question: Silicon can make an
excellent drug even better.
4
-(4-Chlorophenyl)-1-(4-dimethylamino-4-oxo-3,3-diphenylbu-
tyl)-4-hydroxy-4-silapiperidine (sila-loperamide 1b): Trifluoro-
methanesulfonic acid (153 mg, 1.02 mmol) was added dropwise at
0
8C within 1 min to a stirred solution of 9 (300 mg, 514 mmol) in
CH Cl (15 mL), and the mixture was stirred at 08C for 30 min and
2
2
then at 208C for a further 30 min. Subsequently, the mixture was
cooled to 08C, triethylammonium chloride (200 mg, 1.45 mmol)
was added in a single portion, and the mixture was stirred at 08C
for 30 min. The solvent was removed under reduced pressure, THF
(15 mL) was added to the stirred residue at 208C, and the mixture
was then kept undisturbed at À188C for 2 h. The upper layer of
the resulting two-phase system was separated from the lower
layer (triethylammonium trifluoromethanesulfonate) by means of
a syringe and cooled to 08C. To this solution, MeOH (200 mg,
6
.24 mmol) and Et N (500 mg, 4.94 mmol) were added sequentially
3
in single portions, and the mixture was stirred at 08C for 30 min
and then kept undisturbed at À188C for 16 h. The resulting precip-
itate was filtered off and discarded, and the solvent, excess MeOH,
excess Et N, and anisole (formed in the SiÀC cleavage reaction)
3
were removed from the filtrate under reduced pressure in a bulb-
to-bulb distillation apparatus to yield 4-(4-chlorophenyl)-1-(4-dime-
thylamino-4-oxo-3,3-diphenylbutyl)-4-methoxy-4-silapiperidine (10)
as an oily crude product. This product was dissolved in CH CN
3
(5 mL), and a 1m ethereal HCl solution (1.0 mL, 1.0 mmol of HCl)
was added at 208C, immediately followed by the addition of H O
2
(1.0 mL). The solvents were removed under reduced pressure, the
residue was dissolved in CH Cl (10 mL), and the resulting solution
2
2
Experimental Section
was washed sequentially with H O (10 mL) and a 1m aqueous solu-
2
tion of NaOH (10 mL). The aqueous extracts were discarded, the or-
ganic extract was dried over anhydrous Na SO , and the solvent
Chemistry
2
4
was removed under reduced pressure to furnish 1b as a crude
product, which was dissolved in DMSO (1.0 mL) and then purified
by preparative HPLC on silica gel (XBridge C18 column; 10 mm,
General procedures: All syntheses involving chemicals sensitive to
water and/or oxygen were carried out under a dry argon atmos-
phere. The organic solvents used were dried and purified accord-
ing to standard procedures and stored under dry nitrogen. High-
pressure liquid chromatography (HPLC) was performed by using
a Gilson Preparative HPLC with a gradient pump system 333/334,
GX-281 injector, and UV/Vis detector 155. Bulb-to-bulb distillations
were carried out with a Büchi GKR-50 Glass Oven apparatus. Lyo-
2
50”19 mm) using a gradient of 45–85% CH CN in H O/CH CN/
3 2 3
À1
NH (95:5:0.2 v/v/v) buffer over 20 min (flow rate: 19 mLmin , de-
3
tector l: 220 nm). The relevant fractions were combined and
freeze-dried to give 1b as a colorless powder (79 mg, 160 mmol;
1
3
1%). H NMR ([D ]DMSO): d=0.67–0.75 and 0.77–0.87 (m, 4H,
6
1
SiCH
2.25–2.34 (m, 2H, NCH
SiCH CH N), 2.85 (brs, 3H, NCH
7.48–7.50 ppm (m, 14H, C , C
14.0 (SiCH CH N), 36.6 (NCH ), 38.5 (NCH
(SiCH CH N), 54.2 (NCH CH C), 58.9 (NCH
127.69 (C3/C5, C Cl), 127.73 (C3/C5, C H
6 5
2
CH
2
N), 1.99–2.08 (m, 2H, NCH
CH C), 2.50–2.56 and 2.57–2.64 (m, 4H,
), 6.07 (s, 1H, SiOH), 7.24–7.42 and
2
CH C), 2.24 (brs, 3H, NCH ),
2
3
philization was performed with a Christ alpha 1-2 apparatus. H,
1
3
29
2
2
C, and Si NMR spectra were recorded at 238C on Bruker DRX-
1
13
29
2
2
3
3
00 [compound 5 ( H, 300.1 MHz; C, 75.5 MHz; Si, 59.6 MHz)],
13
1
6
H
5
6
H
4
Cl); C NMR ([D
), 41.4 (NCH
CH C), 126.5 (C4, C
), 128.3 (C2/C6, C
6
]DMSO): d=
Bruker Avance 500 [compounds 3, 4, 6, 7·HCl, and
5
9
( H,
1
3
29
2
2
3
3
2
CH C), 51.6
2
00.1 MHz; C, 125.8 MHz; Si, 99.4 MHz)], or Bruker Avance III
1
13
2
2
2
2
2
2
6
H
H
5
),
),
NMR spectrometers [compound 1b ( H, 500.0 MHz; C, 125.7 MHz;
2
9
H
6
4
6
5
Si, 99.3 MHz)] using CD Cl , CDCl , C D , or [D ]DMSO as solvent.
2
2
3
6
6
6
1
34.4 (C1, C H Cl), 135.2 (C2/C6, C H Cl), 137.1 (C4, C H Cl), 141.1
6 4 6 4 6 4
Chemical shifts (d, ppm) were determined relative to internal
29
1
13
(C1, C
HRMS (ESI):
493.2079.
H
6
5
), 172.2 ppm (C=O); Si NMR ([D
6
]DMSO): d=À6.8 ppm;
C H33ClN O Si [M+H] , calcd: 493.2100; found:
28 2 2
CHDCl2 ( H, d=5.32 ppm; CD Cl ), internal CD Cl2 ( C, d=
2
2
2
+
1
5
3.8 ppm; CD Cl ), internal CHCl ( H, d=7.24 ppm; CDCl ), internal
2 2 3 3
CDCl ( C, d=77.0 ppm; CDCl ), internal C HD ( H, d=7.28 ppm;
C D ), internal C D ( C, d=128.0 ppm; C D ), internal [D ]DMSO
H, d=2.49 ppm; [D ]DMSO), internal [D ]DMSO ( C, d=
6 6
9.5 ppm; [D ]DMSO), and external TMS ( Si, d=0.0 ppm; CD Cl ,
6 2 2
CDCl , C D , [D ]DMSO). Analysis and assignment of the H NMR
1
3
1
3
3
6
5
1
3
6
6
6
6
6
6
5
Triethoxyvinylsilane (2): This compound was commercially avail-
able.
1
13
(
29
3
1
(4-Chlorophenyl)diethoxyvinylsilane (3): A 1m solution of (4-
3
6
6
6
1
1
spectroscopic data were supported by H, H gradient-selected
chlorophenyl)magnesium bromide in Et O (800 mL, 800 mmol of 4-
2
1
3
1
13
1
COSY, C, H gradient-selected HMQC, and C, H gradient-selected
ClC H MgBr) was added dropwise at 08C within 4 h to a stirred so-
6
4
13
HMBC experiments. Assignment of the C NMR spectroscopic data
lution of 2 (457 g, 2.40 mol) in Et O (800 mL), and the mixture was
2
1
3
1
was supported by DEPT-135 and the aforementioned C, H corre-
lation experiments. Coupling constants are given as their absolute
values. Elemental analyses were performed with a VarioMicro appa-
then stirred at 208C for 16 h. Subsequently, the mixture was con-
centrated under reduced pressure to a volume of ~500 mL, fol-
lowed by the addition of n-pentane (800 mL). The resulting sus-
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
919
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
pension was stirred at 208C for 10 min, the precipitate was re-
moved by filtration and washed with n-pentane (3”100 mL), and
the solvent of the filtrate (including the wash solutions) was re-
moved under reduced pressure. The excess of 2 was separated
from the residue by distillation in vacuo (34 mbar, 698C), and the
combined, and the solvents were removed under reduced pressure
to give 5, after crystallization from Et O at 208C, as a colorless crys-
2
1
talline solid (71.2 g, 211 mmol, 91%). H NMR (C D ): d=1.30–1.35
6
6
(m, 4H, CH CH OH), 1.53 (s, 2H, CH CH OH), 3.32 (s, 3H, C H OCH ),
2
2
2
2
6
4
3
3.57–3.62 (m, 4H, CH CH OH), 6.82–6.85 (m, 2H, H-2/H-6,
2
2
residue was then further distilled in vacuo (1 mbar, 788C) to give 3
C H OCH ), 7.16–7.18 (m, 2H, H-2/H-6, C H Cl), 7.21–7.25 (m, 2H, H-
6 4 3 6 4
3/H-5, C H Cl), 7.31–7.35 ppm (m, 2H, H-3/H-5, C H OCH ); C NMR
6 4 6 4 3
(C D ): d=18.4 (CH CH OH), 54.6 (CH CH OH), 59.2 (C H OCH ),
6 6 2 2 2 2 6 4 3
114.2 (C2/C6, C H OCH ), 125.8 (C1, C H OCH ), 128.4 (C1, C H Cl),
6 4 3 6 4 3 6 4
1
13
as a colorless liquid (169 g, 658 mmol, 82%). H NMR (CDCl ): d=
3
2
3
1
6
6
1
7
.28 (d ), 3.86 (d ), and 3.88 (d ) (ABX system, J =10.3 Hz, J
=
X
3
A
B
3
AB
AX
.8 Hz, JBX =7.2 Hz, 10H, OCH H C(H ) ), 5.94 (d ), 6.12 (d ), and
A
B
X 3
A
M
3
2
3
.22 (dX) (AMX system, JAM =4.6 Hz, JA(E)X =14.7 Hz, JM(Z)X
=
134.9 (C4, C H Cl), 136.0 (C2/C6, C H Cl), 136.58 (C3/C5, C H Cl),
6 4 6 4 6 4
2
9
9.8 Hz, 3H, CH =CH H ), 7.38–7.41 (m, 2H, H-2/H-6, C H Cl), 7.60–
136.59 (C3/C5, C H OCH ), 161.4 ppm (C4, C H OCH ); Si NMR
6 4 3 6 4 3
X
A
M
6
4
13
.64 ppm (m, 2H, H-3/H-5, C H Cl); C NMR (CDCl ): d=18.3
(C D ): d=À9.3 ppm; anal. calcd (%) for C H ClO Si (336.89): C
6
4
3
6
6
17 21
3
(
OCH CH ), 58.8 (OCH CH ), 128.1 (C2/C6, C H Cl), 131.5 (C1,
60.61, H 6.28; found: C 60.98, H 6.25.
2
3
2
3
6
4
C H Cl), 131.7 (CH=CH ), 136.0 (C3/C5, C H Cl), 136.6 (C4, C H Cl),
6
4
2
6
4
6
4
2
9
1
(
37.3 ppm (CH=CH2); Si NMR (CDCl ): d=À33.2 ppm; anal. calcd
1-Allyl-4-(4-chlorophenyl)-4-(4-methoxyphenyl)-4-silapiperidine
3
%) for C H ClO Si (256.80): C 56.13, H 6.67; found: C 56.33, H
(6): Methanesulfonyl chloride (17.1 g, 149 mmol) and Et N (18.0 g,
1
2
17
2
3
6
.75.
178 mmol) were added sequentially in single portions at À208C to
a stirred solution of 5 (20.0 g, 59.4 mmol) in CH Cl (500 mL), and
2
2
(
4-Chlorophenyl)(4-methoxyphenyl)divinylsilane (4): A 0.5m solu-
the mixture was then stirred at À208C for 3 h. Subsequently, allyl-
amine (100 g, 1.75 mol) was added in a single portion at À208C to
the mixture, which was then warmed to 208C and stirred at 208C
for 16 h. The solvent and excess allylamine were removed from the
reaction mixture under reduced pressure, followed by sequential
tion of (4-methoxyphenyl)magnesium bromide in THF (624 mL,
3
3
and the mixture was then stirred at 208C for 16 h. Subsequently,
a 0.7m solution of vinylmagnesium chloride in THF (491 mL,
12 mmol of 4-MeOC H MgBr) was added dropwise at 208C within
6 4
h to a stirred solution of 3 (80.0 g, 312 mmol) in Et O (800 mL),
2
addition of EtOAc (200 mL) and a saturated aqueous NaHCO solu-
3
3
2
44 mmol of CH =CHMgCl) was added dropwise at 208C within
h, and the mixture was heated under reflux for 16 h. The mixture
tion (100 mL) at 208C. The organic layer was separated, the aque-
ous layer was extracted with EtOAc (3”50 mL) and discarded, the
combined organic extracts were dried over anhydrous Na SO , and
2
was then concentrated under reduced pressure to a volume of
200 mL, followed by sequential addition of Et O (500 mL) and
2
4
~
the solvent was removed under reduced pressure to give an oily
2
a saturated aqueous Na CO solution (200 mL) at 08C. The organic
residue. This crude product was dissolved in Et O (100 mL), and
2
3
2
layer was separated, the aqueous layer was extracted with Et O
a 2m ethereal HCl solution (33.0 mL, 66.0 mmol of HCl) was added
dropwise at 208C within 5 min. The resulting precipitate was isolat-
ed by filtration and recrystallized from 2-propanol (50 mL) by slow
cooling of the boiling solution to 208C. The product was isolated
2
(
3”100 mL) and discarded, the combined organic extracts were
dried over anhydrous Na SO , the solvent was removed under re-
2
4
duced pressure, and the residue was purified by bulb-to-bulb distil-
lation in vacuo (1 mbar, 1808C) to give 4 as a colorless liquid
by filtration and washed with Et O (2”10 mL) to give 6·HCl as a col-
orless crystalline solid, which was subsequently added in a single
portion at 208C to a stirred two-phase system consisting of a 2m
2
1
(
80.0 g, 266 mmol, 86%). H NMR (CD Cl ): d=3.81 (s, 3H,
2
2
2
C H OCH ), 5.79 (d ), 6.27 (d ), and 6.48 (d ) (AMX system, J =
AM
6
4
3
A
M
X
3
3
3
.7 Hz, J =14.6 Hz, J
=20.2 Hz, 6H, CH =CH H ), 6.92–6.95
aqueous NaOH solution (33.0 mL, 66.0 mmol of NaOH) and Et O
A(E)X
M(Z)X
X
A
M
2
(
m, 2H, H-2/H-6, C H OCH ), 7.35–7.38 (m, 2H, H-2/H-6, C H Cl),
(200 mL), and the mixture was then stirred at 208C for 30 min, fol-
6
4
3
6
4
7
.43–7.46 (m, 2H, H-3/H-5, C H Cl), 7.46–7.48 ppm (m, 2H, H-3/H-5,
lowed by the addition of H O (200 mL). The organic layer was sep-
6
4
2
1
3
C H OCH ); C NMR (CD Cl ): d=55.4 (C H OCH ), 114.1 (C2/C6,
arated, the aqueous layer was extracted with Et O (3”50 mL) and
6
4
3
2
2
6
4
3
2
C H OCH ), 124.6 (C1, C H OCH ), 128.4 (C2/C6, C H Cl), 133.8 (C1,
discarded, the combined organic extracts were dried over anhy-
drous Na SO , and the solvent was removed under reduced pres-
6
4
3
6
4
3
6
4
C H Cl), 134.1 (CH=CH ), 136.1 (C4, C H Cl), 136.8 (CH=CH ), 137.2
6
4
2
6
4
2
2
4
1
(
C3/C5, C H Cl), 137.3 (C3/C5, C H OCH ), 161.5 ppm (C4,
sure to give 6 as a colorless oil (8.52 g, 23.8 mmol, 40%). H NMR
(CD Cl ): d=1.26–1.35 (m, 4H, SiCH CH N), 2.71–2.80 (m, 4H,
6
4
9
6
4
3
2
C H OCH ); Si NMR (CD Cl ): d=À20.9 ppm; anal. calcd (%) for
6
4
3
2
2
2
2
2
2
C H ClOSi (300.86): C 67.87, H 5.70; found: C 67.80, H 5.80.
SiCH CH N), 3.03–3.06 (m, 2H, CH CH=CH ), 3.80 (s, 3H, C H OCH ),
2 2 2 2 6 4 3
1
7
17
5
.07–5.11 and 5.12–5.17 (m, 2H, CH CH=CH ), 5.82–5.90 (m, 1H,
2
2
(
4-Chlorophenyl)bis(2-hydroxyethyl)(4-methoxyphenyl)silane (5):
CH
2
CH=CH
2
), 6.90–6.94 (m, 2H, H-2/H-6, C
6
H
4
OCH
3
), 7.33–7.36 (m,
Cl), 7.46–
): d=11.8
), 61.9 (NCH CH=
CH=CH ), 126.1 (C1,
Cl), 135.2 (C1, C Cl), 135.8 (C4,
Cl), 136.43 (C3/C5, C OCH ), 136.6
); Si NMR (CD Cl ): d=
24ClNOSi (357.95): C 67.11, H
6.76, N 3.91; found: C 66.80, H 6.80, N 3.91.
A solution of 9-borabicyclo[3.3.1]nonane [63.1 g, 259 mmol (based
on the 9-BBN dimer)] and 4 (70.0 g, 233 mmol) in THF (1 L) was
stirred at 208C for 16 h, followed by sequential addition of H O
2H, H-2/H-6, C
7.49 ppm (m, 2H, H-3/H-5, C
(SiCH CH N), 52.4 (SiCH CH N), 55.4 (C
CH ), 114.1 (C2/C6, C OCH ), 117.0 (NCH
), 128.4 (C2/C6, C
Cl), 136.42 (C3/C5, C
CH=CH ), 161.3 ppm (C4, C
À15.6 ppm; anal. calcd (%) for C20
H Cl), 7.42–7.45 (m, 2H, H-3/H-5, C H
6 4 6 4
1
3
H
6
OCH
4
); C NMR (CD
OCH
Cl
2 2
3
H
6
2
2
2
2
2
4
3
2
(
200 mL) and a 4m aqueous NaOH solution (300 mL, 1.20 mol of
2
H
6
4
3
2
2
NaOH) at 08C. Subsequently, an aqueous H O solution (30 wt%,
4
tion mixture, which was then heated under reflux for 3 h, followed
by sequential addition of a 0.1m aqueous solution of K CO3
C
C
6
H
H
OCH
4
3
H
6
4
H
6 4
2
2
20 mL) was added dropwise at 08C within 4 h to the stirred reac-
6
4
6
H
4
6
H
4
3
2
9
(NCH
H
6
OCH
4
2
2
3
2
2
H
2
(
500 mL) and CH Cl (500 mL) at 208C. The organic layer was sepa-
2 2
rated, the aqueous layer was extracted with CH Cl2 (3”200 mL)
2
and discarded, the combined organic extracts were dried over an-
hydrous Na SO , and the solvent was removed under reduced pres-
sure to give a colorless oil. The byproduct cyclooctane-1,5-diol was
separated from this crude product by bulb-to-bulb distillation
4-(4-Chlorophenyl)-4-(4-methoxyphenyl)-4-silapiperidinium chlo-
ride (7·HCl): 1-Chloroethyl chloroformate (2.32 g, 16.2 mmol) was
added dropwise at 08C within 5 min to stirred solution of 6
2
4
(5.25 g, 14.7 mmol) in CHCl (250 mL), and the mixture was stirred
3
(
0.07 mbar, 1608C), and the residue was then purified by column
at 08C for 10 min and then heated under reflux for 2 h. Subse-
quently, the mixture was cooled to 208C, the solvent was removed
under reduced pressure, the residue was dissolved in MeOH
chromatography on aluminum oxide [Al O , Brockmann III; eluent,
n-hexane/CH Cl /EtOH (20:50:4 v/v/v)]. The relevant fractions were
2
3
2
2
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
920
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(
2
100 mL), and the resulting solution was heated under reflux for
h. Approximately 80 mL of MeOH were removed by distillation
put on an Edmund Bühler shaker for 18 h at 208C. Aliquots (5 mL)
of octanol were transferred and diluted with 495 mL of CH CN/H O
3
2
under atmospheric pressure, and the remaining mixture was then
kept undisturbed at À208C over 16 h. The resulting precipitate
was isolated by filtration and recrystallized from 2-propanol
(1:1 v/v) and, to avoid contamination of the buffer, the rest of the
octanol was removed before 150 mL of buffer samples were trans-
ferred. Octanol and buffer samples were diluted with CH CN/H O
3
2
(
10 mL) by slow cooling of the boiling solution to 208C. The pre-
(1:1 v/v) in four 10-fold steps to yield octanol samples diluted 100-
to 1000000-fold and buffer samples diluted 1- to 10000-fold. LC–
MS/MS was used for analysis, and logD was calculated from the in-
tegrated peak areas of the samples in the linear MS response
range.
cipitate was isolated by filtration and washed with Et O (2”10 mL)
to give 7·HCl as a colorless crystalline solid (3.28 g, 9.26 mmol,
6
(
2
1
3%). H NMR (CD Cl ): d=1.62–1.80 (m, 4H, SiCH CH N), 3.27–3.51
2
2
2
2
m, 4H, SiCH CH N), 3.81 (s, 3H, C H OCH ), 6.95–6.98 (m, 2H, H-2/
2 2 6 4 3
H-6, C H OCH ), 7.38–7.41 (m, 2H, H-2/H-6, C H Cl), 7.47–7.49 (m,
6
4
3
6
4
Solubility in HBSS buffer (pH 7.4): Solutions of test compounds in
DMSO (30 mL, 10 mm) in glass vials were dried using a Genevac
vacuum evaporator. When samples were dry, 300 mL of 0.1m phos-
phate buffer, pH 7.4, was added to the glass vials. The vials were
put on an Edmund Bühler shaker for 18 h at 208C. Samples were
filtered through a Whatman GF/B 96-well filter, and 20 mL of the fil-
trated samples were transferred to separate wells in a plate con-
taining 180 mL of CH CN/H O (1:1 v/v). Standards were prepared by
2
9
4
H, H-3/H-5, C H Cl), 7.50–7.53 (m, 2H, H-3/H-5, C H OCH ),
6 4 6 4 3
13
.76 ppm (brs, 2H, NH2); C NMR (CD Cl ): d=9.5 (SiCH CH N),
2
2
2
2
4.4 (SiCH CH N), 55.5 (C H OCH ), 114.6 (C2/C6, C H OCH ), 122.5
2
2
6
4
3
6
4
3
(
C1, C H Cl), 128.8 (C2/C6, C H Cl), 131.9 (C1, C H OCH ), 136.45
6 4 6 4 6 4 3
(
1
C3/C5, C H Cl), 136.55 (C3/C5, C H OCH ), 136.9 (C4, C H Cl),
6 4 6 4 3 6 4
29
62.0 ppm (C4, C H OCH ); Si NMR (CD Cl ): d=À16.9 ppm; anal.
6
4
3
2
2
calcd (%) for C H Cl NOSi (354.35): C 57.62, H 5.97, N 3.95; found:
1
7
21
2
3
2
C 57.56, H 6.05, N 4.02.
diluting the 10 mm solutions of the test compounds with CH CN/
3
4
-(4-Chlorophenyl)-1-(4-dimethylamino-4-oxo-3,3-diphenylbu-
H O (1:1 v/v) to 200 mm. Three further 10-fold dilution steps were
2
tyl)-4-(4-methoxyphenyl)-4-silapiperidine (9): A mixture of 7·HCl
applied to both the samples and the standards, and they were all
analyzed by LC–MS/MS. The solubility was determined by using
the integrated peak areas of the samples in the linear MS response
range.
(
(
2.43 g, 6.86 mmol), 8 (2.61 g, 7.54 mmol), anhydrous Na CO3
2
3.20 g, 30.2 mmol), and CH CN (200 mL) was heated under reflux
3
for 16 h. The mixture was then cooled to 208C, the solvent was re-
moved under reduced pressure, and H O (100 mL) and CH Cl
2
2
2
pKa values: The pKa values were obtained according to refer-
(
100 mL) were added sequentially to the residue. The organic layer
ence [28]. The method uses pressure-assisted capillary electropho-
was separated, the aqueous layer was extracted with CH Cl (3”
2
2
3D
resis (HPCE , Agilent Technologies) coupled online with an ion
5
0 mL) and discarded, the combined organic extracts were dried
trap mass spectrometer (1100 series LC/MSD trap).
over anhydrous Na SO , and the solvent was removed under re-
2
4
duced pressure. The solid residue was purified by column chroma-
tography on silica gel [40–63 mm, 300 g (Merck), treated with con-
centrated aqueous ammonia solution (7% by weight relative to
the silica gel); eluent, CH Cl /MeOH (97:3 v/v)]. The relevant frac-
In vitro pharmacology
Radioligand binding assays: Radioligand binding assays were per-
formed at CEREP (Poitiers, France). Briefly, displacement of the radi-
oligands by the test compounds was measured at each receptor
using membrane preparations from HEK cells expressing human
m1 opioid receptors or CHO cells expressing rat k1 opioid recep-
tors. Membrane homogenates (16–60 mg of protein) were incubat-
2
2
tions were combined, and the solvents were removed under re-
duced pressure to give 9 as a colorless crystalline solid (2.52 g,
1
4
.32 mmol, 63%). H NMR (CD Cl ): d=1.06–1.33 (m, 4H,
2
2
SiCH CH N), 2.06–2.22 (m, 2H, NCH CH C), 2.30 (brs, 3H, NCH ),
2
2
2
2
3
2
.36–2.52 (m, 2H, NCH CH C), 2.60–2.76 (m, 4H, SiCH CH N), 2.92
2 2 2 2
3
(
7
brs, 3H, NCH ), 3.78 (s, 3H, C H OCH ), 6.87–6.90 and 7.25–
ed with 1) 0.5 nmol [ H]DAMGO for m opioid receptors in the ab-
3
6
4
3
13
.43 ppm (m, 18H, C H , C H OCH , C H Cl); C NMR (CD Cl ): d=
sence or presence of the test compound in a buffer containing
6
5
6
4
3
6
4
2
2
1
1.5 (SiCH CH N), 37.2 (NCH ), 39.3 (NCH ), 42.2 (NCH CH C), 52.3
50 mmol of Tris·HCl (pH 7.4) and 5 mmol of MgCl for 120 min at
2
228C or 2) 1 nmol [ H]U69593 for k opioid receptors in the ab-
2
2
3
3
2
2
3
(
SiCH CH N), 55.3 (NCH CH C), 59.9 (NCH CH C), 114.1 (C3/C5,
2
2
2
2
2
2
C H OCH ), 126.2 (C1, C H OCH ), 126.9 (C4, C H ), 128.3 (C2/C6,
sence or presence of the test compound in a buffer containing
6
4
3
6
4
3
6
5
C H Cl), 128.5 (C3/C5, C H ), 128.6 (C2/C6, C H ), 135.3 (C1, C H Cl),
50 mmol of Tris·HCl (pH 7.4), 10 mmol of MgCl , and 1 mmol of
2
6
4
6
5
6
5
6
4
1
35.7 (C4, C H Cl), 136.39 (C3/C5, C H Cl), 136.41 (C2/C6,
EDTA for 60 min at 228C. Nonspecific binding was determined in
the presence of 10 mmol naloxone. Following incubation at 228C
for 60 min (k) or 120 min (m), samples were filtered rapidly under
vacuum through glass fiber filters (GF/B, Packard), pre-soaked with
6
4
6
4
C H OCH ), 141.5 (C1, C H ), 161.2 (C4, C H OCH ), 173.3 ppm (C=
6
4
2
3
6
5
6
4
3
9
O); Si NMR (CD Cl ): d=À15.6 ppm; HRMS (ESI): C H ClN O S
2
2
35 39
2
2
+
[M+H] , calcd: 583.2548; found: 583.2558.
0
.3% PEI, and rinsed several times with an ice-cold buffer contain-
ing 50 mmol of Tris·HCl (pH 7.4) using a 96-sample cell harvester
Unifilter, Packard). The filters were dried and then counted for ra-
4
-(4-Chlorophenyl)-1-(4-dimethylamino-4-oxo-3,3-diphenylbu-
tyl)-4-methoxy-4-silapiperidine (10): This compound was isolated
in the synthesis of 1b as a crude product and was not further puri-
fied. HRMS (ESI): C H ClN O Si [M+H] , calcd: 507.2235; found:
(
dioactivity in a scintillation counter (Topcount, Packard) using
a scintillation cocktail (Microscint 0, Packard).
+
2
9
35
2
2
5
07.2212.
Cell-based functional assays (agonist/antagonist determination):
m and k opioid receptor function was determined by measurement
of receptor-driven inhibition of NKH 477-stimulated cAMP accumu-
lation in HEK cells expressing human m opioid receptors or in CHO
cells expressing rat k opioid receptors. To determine whether the
test compounds were agonistic, cells were suspended in HBSS
buffer (Invitrogen) complemented with 20 mmol of HEPES (pH 7.4)
and 500 mmol IBMX and then distributed in microplates at a density
Physicochemical properties
logD values (pH 7.4): Partitioning of the test compounds between
1
-octanol and 0.1m phosphate buffer, pH 7.4, at 208C was deter-
mined by using a modified version of the shake-flask method de-
[27]
scribed by Leo et al. The compounds were dissolved, in a 96-
well plate, in 400 mL of octanol, and 400 mL of buffer were added
to each well. The plate was vigorously stirred for 5 min and then
3
4
of 7”10 cells per well (HEK/m) or 10 cells per well (CHO/k) and
pre-incubated for 5 min at 208C in the presence of HBSS or the
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
921
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
test compounds, following which NKH 477 was added to a concen-
tration of 10 mmol. After a further 10 min incubation at 378C, the
cells were lysed, and the fluorescence acceptor (D2-labeled cAMP)
and fluorescence donor (anti-cAMP antibody labeled with europi-
um cryptate) were added. After 60 min at 208C, the fluorescence
transfer was measured at the excitation wavelength 337 nm and
emission wavelengths 620 and 665 nm using a microplate reader
Caco-2 cells, with the modification that HBSS buffer, pH 7.4, was
used on both sides of the monolayer. The permeability was studied
in the apical-to-basolateral direction (Papp A!B) as well as in the
basolateral-to-apical direction (Papp B!A). Efflux ratios were derived
from the following equation: (Papp B!A)/(P A!B).
app
Permeability and efflux in MDCK-MDR1 cells
(
Rubystar, BMG). The cAMP concentration was determined by di-
viding the signal measured at 665 nm by that measured at 620 nm
ratio) relative to a standard curve. To determine whether the test
MDCK-MDR1 cells seeded onto a Millipore 96-well plate at a density
of 60000 cells per well were used to study P-gp efflux. The process
was automated by a robotic Tecan EVO. HBSS buffer, pH 7.4, was
dispensed to either the apical or basal side of the monolayer. The
(
compounds had antagonistic effects, a reference agonist was
added along with NKH 477: 20 nmol DAMGO for m opioid receptors
or 5 nmol U50488 for k opioid receptors.
À1
assay was initiated by adding the test substrate (1 mmolmL in
Data analysis: Concentration–effect data were fitted to a four-pa-
rameter logistic equation using nonlinear regression. The concen-
tration of the test compound causing 50% maximal effect (A ) for
HBSS buffer, pH 7.4) to the apical or basal side of the monolayer
(with HBSS buffer, pH 7.4, on the other side of the monolayer).
Samples were withdrawn before the addition of the test com-
pound and at 150 min post-addition of the test compound. During
incubation, the plates were placed in a shaking incubator at 378C.
All samples were analyzed by LC–MS/MS, and the areas were used
to determine the Papp A!B and Papp B!A values. Efflux ratios were
derived from the following equation: (P B!A)/(P A!B).
50
each individual curve was determined, and mean pA50 (ÀlogA )
50
values were calculated with standard deviation. For radioligand
binding assays, the affinity of the test compounds was estimated
[29]
by the method of Cheng and Prusoff.
app
app
hERG activity
Intrinsic clearance and half-life in human liver microsomes
and human and rat hepatocytes
Assessment of hERG blockade was made as described by Bridg-
land-Taylor et al.
scribed by Persson et al. were grown to semi-confluence at 378C
in a humidified environment (5% CO ) in Ham’s F-12 nutrient mix-
ture and l-glutamine (Sigma) supplemented with 10% fetal calf
serum (Invitrogen) and 600 mgmL hygromycin (Invitrogen). Cells
were incubated at 378C for 24 h and then incubated at 288C for
[
30]
CHO k1 cells expressing hERG channels de-
[31]
Metabolic stability in human liver microsomes: Human liver mi-
À1
crosomes were defrosted on ice and diluted to 1 mgmL of mi-
2
crosomal protein in 0.1m phosphate buffer, pH 7.4. The test com-
pounds at 1 mm were incubated with the liver microsome suspen-
sions and 1 mm NADPH at 378C in a 96-well plate. At 0.5, 5, 10, 15,
À1
2
0, and 30 min, aliquots (30 mL) were transferred to a 96-well plate
4
8–72 h, following which a single voltage pulse was applied to
containing 120 mL of CH CN. This plate was centrifuged for 20 min,
3
evoke the pre- and post-compound hIERG currents using an Ion-
Works HT (Molecular Devices) platform. A holding potential of
À70 mV was applied for 20 s, followed by a 160 ms step to
À60 mV and a 100 ms step back to À70 mV. The voltage was then
stepped to 40 mV for 1 s, and a steady-state current was observed.
A 2 s step down to À30 mV, inducing the tail current, was then fol-
lowed by a 0.5 s step to À70 mV. The current signal was sampled
and the supernatant was removed and diluted (1:1 v/v) with H O
2
before analysis by LC–MS/MS. Peak areas were determined from
extracted ion chromatograms, and the in vitro intrinsic clearance
À1
À1
(
in vitro CL , mLmin mg microsomal protein) of the parent
int
compound was calculated from the slope in the regression analysis
of the natural logarithm of parent concentration versus time curve.
at 2.5 kHz. The hI current magnitude was measured automatical-
ERG
Metabolic stability in human and male Han Wistar rat hepato-
ly from the leak-subtracted traces by the IonWork software by
taking a 40 ms average of the baseline current measured at
À70 mV and subtracting this from the peak of the tail-current re-
sponse measured during the voltage step at À30 mV. The degree
cytes: Hepatocyte metabolic stability was determined in accord-
[32]
ance with the method described by Jacobson et al.
served hepatocytes at a concentration of 10 viable cells per mL
were used. After thawing, hepatocytes were incubated for 10 min
Cryopre-
6
of inhibition of the hI current was assessed by dividing the post-
ERG
to warm to 378C, and the test compound, dissolved in CH CN, was
3
scan hIERG current by the respective pre-scan hIERG current for each
well.
added to give a final concentration of 1 mm. At 0.5, 5, 15, 30, 45,
6
0, 80, 100, and 120 min, the incubation system was mixed, and
aliquots (20 mL) were transferred at each time point to wells in
a separate plate filled with 80 mL of CH CN to stop the reaction.
3
Permeability and efflux in human Caco-2 cells
The quenching plate was then vortexed, followed by centrifuga-
tion, and the supernatants were analyzed by LC–MS/MS. Peak
areas were determined from extracted ion chromatograms, and
A monolayer of Caco-2 cells, cultured on semi-permeable poly-
carbonate surfaces, was used to study the permeability in the
apical-to-basolateral direction. The process was automated by a ro-
botic Tecan EVO platform and 24-transwell plates from Costar.
HBSS buffer, pH 7.4, was dispensed to the basal side of the mono-
À1
6
the in vitro intrinsic clearance (in vitro CL , mLmin 10 cells) of
int
the parent compound was calculated from the slope in the regres-
sion analysis of the natural logarithm of parent concentration
versus time curve.
layer. The assay was initiated by adding the test compound
À1
(
10 mmolmL in HBSS buffer, pH 6.5) to the apical side of the
monolayer. Samples were withdrawn before the addition of the
test compound and at 45 and 120 min post-addition of the test
compound. During incubation, the transwell plates were placed in
a shaking incubator at 378C between sampling. All samples were
analyzed by LC–MS/MS, and the apparent permeability (Papp) was
calculated from the peak areas. Efflux was studied in the same
CYP inhibition, time-dependent CYP inhibition, and CYP reac-
tion phenotyping
[33]
CYP inhibition: A fluorescence-based method in 96-well format
was used to determine the inhibition of five different CYPs (1A2,
2C9, 2D6, 3A4, and 2C19). The recombinant human enzymes used
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
922
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
À1
were prepared in house, except for CYP2D6 (Cypex Ltd.). Various
coumarin substrates, biotransformed into fluorescent metabolites,
were used as probes for each individual CYP. A fluorescence plate
reader (SpectraMax GeminiXS, Molecular Devices) was used to
measure the levels of metabolites formed. A dilution series of the
test compounds was prepared at eight different concentrations.
For each CYP, a mixture of the enzyme, the corresponding coumar-
tration. The catheters were filled with heparin (100 IUmL ), exteri-
orized at the nape of the neck, and sealed. The surgery was per-
formed under isoflurane (Forene, Abbott) anesthesia. After surgery,
the rats were housed individually and had free access to food and
water. Approximately 16 h prior to dosing, the animals were de-
prived of food and fasted until 4 h after dosing. The rats had free
access to drinking water during the experiment. On the experi-
ment day, the test item formulation was administered orally by
gavage or intravenously in the jugular vein. At pre-defined time
points, blood samples of ~150 mL were withdrawn from the carotid
artery up to 24 h after dosing. A total of 10 samples were with-
drawn. The blood samples were collected in heparinized plastic
tubes and centrifuged, within 30 min, for five minutes at 10000 g
and 48C. The plasma was transferred to a 96-well plate and stored
at À208C until analysis by LC–MS/MS. The studies were approved
by the Gçteborg Animal Research Ethical Board.
in substrate, potassium phosphate buffer (pH 7.4), and H O (con-
2
centrations and volumes were CYP-dependent) were added to
each well in a black 96-well plate. The test compounds at different
concentrations were added. After 10 min pre-incubation, the cofac-
tor NADPH was added to initiate the reaction. After 20–50 min
(
CYP and substrate dependent), the reaction was terminated by
the addition of Tris base/CH CN (20:80 v/v). The plates were trans-
3
ferred to the fluorescence plate reader, where the wavelengths
were set individually for the different coumarin substrates and
their respective fluorescent metabolite. The responses were export-
ed to Microsoft Excel, in which the IC50 curves were plotted (per-
cent inhibition versus concentration) and IC50 values calculated for
each test substrate and enzyme using XL-fit.
Plasma protein binding
The test compounds were added to a 96-well plate containing
blood plasma to give a final concentration of 5 mm. After mixing,
the samples were added to an equilibrium dialysis device (RED,
Thermo Fischer Scientific Inc.) and dialyzed against phosphate
buffer (0.1m, pH 7.4) for 18 h at 378C. Standard curves used for
quantification were prepared in plasma in a concentration range of
7 mm to 1.4 nm. Standard curve samples and dialysis samples were
Time-dependent CYP inhibition: Time-dependent inhibition of
CYP was investigated for six CYPs (1A2, 2C8, 2C9, 2C19, 2D6, and
À1
3
A4) in a pool of human liver microsomes (1 mgmL , BD Gentest
batch 38289). The test compounds were pre-incubated at two sep-
arate concentrations (10 and 50 mm) in the presence and absence
of NADPH (1 mm) for 30 min. In addition, an incubation without
the test compound was performed with and without addition of
NADPH. A fraction of each incubation (25 mL) was then diluted ten-
fold with NADPH (1 mm) and phosphate buffer (pH 7.4) containing
a cocktail of probe substrates for each CYP and incubated for an-
other 15 min. All incubations were performed in a 96-well format
at 378C in a total volume of 250 mL using a Tecan Freedom Evo
Plus 200 robot. The proteins (95 mL aliquot) were then precipitated
by acidic MeOH (190 mL) and put in a freezer for 30 min, followed
by centrifugation. The supernatant was transferred to a separate
plate, and the response of the metabolite formation from each
probe substrate was estimated by LC–MS/MS (Agilent 1100 LC and
TSQ Quantum Ultra triple quadrupole mass spectrometer equipped
with an APCI ion source). For each probe substrate, an LC–MS/MS
response of the formed metabolite was generated in the following
incubations: A) (À) test compound and (À) NADPH, B) (À) test
compound and (+) NADPH, C) (À) test compound and (+) NADPH,
and D) (À) test compound and (À) NADPH. The percent inhibition
of compounds at both 10 and 50 mm was calculated for each CYP
with the following formula: [1À(C/D)/(B/A)]”100.
precipitated with CH CN, and after centrifugation, supernatants
3
were analyzed by LC–MS/MS. The fraction unbound in the incuba-
tion was calculated as the concentration in the media buffer
sample divided by the concentration in the plasma sample.
Identification of major in vitro metabolites and reactive me-
tabolites
The experimental setup for metabolite identification in hepato-
cytes was the same as for the determination of the metabolic sta-
bility, with the following modifications: The test compound con-
centration was 4 mm, and the reaction was stopped by the addition
of cold CH CN/H O (3:1 v/v). Samples for metabolite identification
3
2
were taken after 40 min incubation. In addition, a blank sample
without test compound was analyzed. The samples were analyzed
by ultra-performance liquid chromatography (Waters ACQUITY
UPLC) coupled to a Xevo G2S TOF instrument equipped with an
electrospray interface. The software used to process the data was
MetaboLynx (Waters). Product ion spectra of the major metabolites
were acquired to allow interpretation and structural assignments.
CYP reaction phenotyping (CRP): The assay consisted of a panel
of six recombinant cytochrome P450 (rCYP) isoforms (1A2, 2C8,
2
C9, 2C19, 2D6, and 3A4; Cypex Ltd.). The incubation was per-
formed in 96-well plates at 1 mm test compound concentration
and with 50 nm rCYP at 378C. Aliquots for analysis by LC–MS/MS
were taken at 0, 7, 15, and 30 min. Values of intrinsic clearance
Identification of major in vivo metabolites in plasma from
male Han Wistar rats
À1
À1
(
CL , mLmin pmol ) were calculated for each individual rCYP.
Plasma samples from the in vivo rat PK experiments were analyzed
with respect to metabolites as well as parent drug to allow the
generation of metabolic profiles. To 50 mL of plasma was added
int
With the aid of relative activity factors derived from human liver
microsomes, the rCYP CLint values were then used to calculate the
relative contribution by a certain CYP isoform to the total metabo-
lism observed in the experiment.
150 mL of CH
vortexed for 10 s and then centrifuged for 10 min at 48C at
2750 g. The supernatants were diluted 1:1 with 0.1% formic acid
prior to analysis by ultra-performance liquid chromatography
Waters ACQUITY UPLC) coupled to a Waters Xevo G2S TOF instru-
CN to precipitate plasma proteins. The samples were
3
~
(
In vivo rat PK
ment equipped with an electrospray interface. The software used
to process the data was MetaboLynx (Waters). Product ion spectra
of the major metabolites were acquired to allow interpretation and
structural assignments.
Two days prior to dosing, male Han Wistar rats were prepared by
cannulation of the left carotid artery for blood sampling and by
cannulation of the right jugular vein for intravenous (i.v.) adminis-
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
923
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
[
7] D. L. DeHaven-Hudkins, L. C. Burgos, J. A. Cassel, J. D. Daubert, R. N. De-
haven, E. Mansson, H. Nagasaka, G. Yu, T. Yaksh, J. Pharmacol. Exp. Ther.
Acknowledgements
1
999, 289, 494–502.
The following people at AstraZeneca R&D are gratefully acknowl-
edged for their work on the various biological studies and DMPK
assays: Lyn Rosenbrier-Ribeiro, Malin Forsgard, Johan Wernevik,
Siavash Tavakoli, Linda Fredlund, Alan Clark, Neil Shearer, Sara
Amberntsson, Martina Furasen, and Charlotta Vedin.
8] S. Grimwood, P. R. Hartig, Pharmacol. Ther. 2009, 122, 281–301.
[9] C. E. Pollard, N. Abi Gerges, M. H. Bridgland-Taylor, A. Easter, T. G. Ham-
mond, J.-P. Valentin, Br. J. Pharmacol. 2010, 159, 12–21.
10] J. Vandenbossche, M. Huisman, Y. Xu, D. Sanderson-Bongiovanni, P.
Soons, J. Pharm. Pharmacol. 2010, 62, 401–412.
11] M. J. Zamek-Gliszczynski, D. W. Bedwell, J. Q. Bao, J. W. Higgins, Drug
Metab. Dispos. 2012, 40, 1825–1833.
12] G. Mukwaya, T. MacGregor, D. Hoelscher, T. Heming, D. Legg, K. Kava-
naugh, P. Johnson, J. P. Sabo, S. McCallister, Antimicrob. Agents Chemo-
ther. 2005, 49, 4903–4910.
[
[
[
Keywords: C/Si exchange · metabolic fate · opioid receptors ·
sila-drugs · silicon
[
[
13] K.-A. Kim, J. Chung, D.-H. Jung, J.-Y. Park, Eur. J. Clin. Pharmacol. 2004,
6
0, 575–581.
14] J. A. Williams, B. J. Ring, V. E. Cantrell, D. R. Jones, J. Eckstein, K. Ruterbo-
[
1] R. A. Stokbroekx, J. Vandenberk, A. H. M. T. Van Heertum, G. M. L. W. van
Laar, M. J. M. C. Van der Aa, W. F. M. Van Bever, P. A. J. Janssen, J. Med.
Chem. 1973, 16, 782–786.
ries, M. A. Hamman, S. D. Hall, S. A. Wrighton, Drug Metab. Dispos. 2002,
3
0, 883–891.
15] P. L. Toutain, A. Bousquet-MØlou, J. Vet. Pharmacol. Ther. 2004, 27, 415–
25.
[
[
[
[
2] Reviews dealing with silicon-based drugs: a) R. Tacke, H. Linoh in The
Chemistry of Organic Silicon Compounds, Part 2 (Eds.: S. Patai, Z. Rappo-
port), Wiley, Chichester, 1989, pp. 1143–1206; b) W. Bains, R. Tacke,
Curr. Opin. Drug Discovery Dev. 2003, 6, 526–543; c) G. A. Showell, J. S.
Mills, Drug Discovery Today 2003, 8, 551–556; d) J. S. Mills, G. A. Show-
ell, Expert Opin. Invest. Drugs 2004, 13, 1149–1157; e) P. K. Pooni, G. A.
Showell, Mini-Rev. Med. Chem. 2006, 6, 1169–1177; f) S. McN. Sieburth,
C.-A. Chen, Eur. J. Org. Chem. 2006, 311 –322; g) S. Gately, R. West, Drug
Dev. Res. 2007, 68, 156–163; h) A. K. Franz, Curr. Opin. Drug Discovery
Dev. 2007, 10, 654–671; i) N. A. Meanwell, J. Med. Chem. 2011, 54,
4
16] J. D. O’Brien, D. G. Thompson, A. McIntyre, W. R. Burnham, E. Walker,
Gut 1988, 29, 312–318.
17] J. Heykants, M. Michiels, A. Knaeps, J. Brugmans, Arzneim.-Forsch. 1974,
2
4, 1649–1653.
[
[
[
18] Z. Zhivkova, I. Doytchinova, Mol. Pharm. 2013, 10, 3758–3768.
19] D. A. Smith, L. Di, E. H. Kerns, Nat. Rev. Drug Discovery 2010, 9, 929–939.
20] L. Huang, L. Berry, S. Ganga, B. Janosky, A. Chen, J. Roberts, A. E. Colletti,
M.-H. J. Lin, Drug Metab. Dispos. 2010, 38, 223–231.
[
[
21] K. Ito, J. B. Houston, Pharm. Res. 2004, 21, 785–792.
2529–2591; j) A. K. Franz, S. O. Wilson, J. Med. Chem. 2013, 56, 388–
405; k) S. McN. Sieburth, Top. Med. Chem. 2014, DOI: 10.1007/7355_
2014_80; l) R. Tacke, S. Dçrrich, Top. Med. Chem. 2014, DOI: 10.1007/
7355_2014_55.
22] A.-K. Sohlenius-Sternbeck, C. Jones, D. Ferguson, B. J. Middleton, D. Pro-
jean, E. Floby, J. Bylund, L. Afzelius, Xenobiotica 2012, 42, 841–853.
23] R. N. Upton, Clin. Exp. Pharmacol. Physiol. 2007, 34, 695–701.
24] A. S. Kalgutkar, H. T. Nguyen, Drug Metab. Dispos. 2004, 32, 943–952.
25] K. Igarashi, F. Kasuya, E. Usuki, N. Castagnoli, Jr., Life Sci. 1995, 57,
[
[
[
[
3] Recent original publications dealing with silicon-based drugs: a) R.
Tacke, F. Popp, B. Müller, B. Theis, C. Burschka, A. Hamacher, M. U. Kas-
sack, D. Schepmann, B. Wünsch, U. Jurva, E. Wellner, ChemMedChem
2
439–2446.
26] E. Usuki, R. Pearce, A. Parkinson, N. Castagnoli, Jr., Chem. Res. Toxicol.
996, 9, 800–806.
[
2
008, 3, 152–164; b) R. Tacke, B. Nguyen, C. Burschka, W. P. Lippert, A.
Hamacher, C. Urban, M. U. Kassack, Organometallics 2010, 29, 1652–
660; c) T. Johansson, L. Weidolf, F. Popp, R. Tacke, U. Jurva, Drug
1
[
[
27] A. Leo, C. Hansch, D. Elkins, Chem. Rev. 1971, 71, 525–615.
1
28] H. Wan, A. G. HolmØn, Y. Wang, W. Lindberg, M. Englund, M. B. Någård,
R. A. Thompson, Rapid Commun. Mass Spectrom. 2003, 17, 2639–2648.
29] Y.-C. Cheng, H. W. Prusoff, Biochem. Pharmacol. 1973, 22, 3099–3108.
30] M. H. Bridgland-Taylor, A. C. Hargreaves, A. Easter, A. Orme, D. C. Hen-
thorn, M. Ding, A. M. Davis, B. G. Small, C. G. Heapy, N. Abi-Gerges, F.
Persson, I. Jacobson, M. Sullivan, N. Albertson, T. G. Hammond, E. Sulli-
van, J.-P. Valentin, C. E. Pollard, J. Pharmacol. Toxicol. Methods 2006, 54,
Metab. Dispos. 2010, 38, 73–83; d) M. J. Barnes, C. Burschka, M. W. Bütt-
ner, R. Conroy, J. O. Daiss, I. C. Gray, A. G. Hendrick, L. H. Tam, D. Kuehn,
D. J. Miller, J. S. Mills, P. Mitchell, J. G. Montana, P. A. Muniandy, H.
Rapley, G. A. Showell, D. Tebbe, R. Tacke, J. B. H. Warneck, B. Zhu, Chem-
MedChem 2011, 6, 2070–2080; e) R. Tacke, R. Bertermann, C. Burschka,
S. Dçrrich, M. Fischer, B. Müller, G. Meyerhans, D. Schepmann, B.
Wünsch, I. Arnason, R. Bjornsson, ChemMedChem 2012, 7, 523–532;
f) J. B. G. Gluyas, C. Burschka, S. Dçrrich, J. Vallet, H. Gronemeyer, R.
Tacke, Org. Biomol. Chem. 2012, 10, 6914–6929; g) P. Luger, M. Weber,
C. Hübschle, R. Tacke, Org. Biomol. Chem. 2013, 11, 2348–2354.
[
[
189–199.
[
[
[
31] F. Persson, L. Carlsson, G. Duker, I. Jocobson, J. Cardiovasc. Electrophy-
siol. 2005, 16, 329–341.
32] L. Jacobson, B. Middleton, J. Holmgren, S. Eirefelt, M. Frçjd, A. Blom-
gren, L. Gustavsson, Assay Drug Dev. Technol. 2007, 5, 403–415.
33] C. L. Crespi, V. P. Miller, B. W. Penman, Anal. Biochem. 1997, 248, 188–
[
[
4] Synthesis of 9-BBN: J. A. Soderquist, H. C. Brown, J. Org. Chem. 1981, 46,
4599–4600.
5] 1-Chloroethyl chloroformate as an N-dealkylation agent: R. A. Olofson,
J. T. Martz, J.-P. Senet, M. Piteau, T. Malfroot, J. Org. Chem. 1984, 49,
190.
2081–2082.
[6] Synthesis of 8: Z. Chen, E. Davies, W. S. Miller, S. Shan, K. J. Valenzano,
Received: January 28, 2015
D. J. Kyle, Bioorg. Med. Chem. Lett. 2004, 14, 5275–5279.
Published online on March 20, 2015
ChemMedChem 2015, 10, 911 – 924
www.chemmedchem.org
924
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim