Convergent syntheses of carbon-13 labeled midazolam and 1′-hydroxymidazolam

Article abstract of DOI:10.1016/j.tetlet.2005.01.128

For the purpose of drug interaction studies, the stable-isotope labeled [¹³C₃]midazolam and its metabolite, 1′-hydroxy- [¹³C₃]midazolam were synthesized in four and five steps in overall yields of 25.5% and 14.2%, from 7-chloro-5-(2-fluorophenyl)-2-(N- nitrosomethylamino)-3H-1,4-benzodiazepine, respectively, by a convergent synthesis, in which a key imidazoline ring formation was achieved by the facile reaction of [¹³C]2-aminomethyl-7-chloro-2,3-dihydro-5-(2- fluorophenyl)-1H-1,4-benzodiazepine with varying ethyl imidate hydrochlorides. The scrambling of C-3 and C-4 labeling in intermediate diamine, and consequently in the final products, as well as the formation of a Δ^4,5 isomer of 1′-hydroxy-[¹³C₃]midazolam was observed and explained.

Full text of DOI:10.1016/j.tetlet.2005.01.128

Tetrahedron
Letters
Tetrahedron Letters 46 (2005) 2087–2091
Convergent syntheses of carbon-13 labeled midazolam
and 10-hydroxymidazolamI
Yinsheng Zhang,* Peter W. K. Woo, Jon Hartman, Norman Colbry,
Yun Huang and Che C. Huang
Chemical R&D, Michigan Pharmaceutical Sciences, Pfizer Global R&D, Pfizer Inc., 301 Henrietta Street Kalamazoo, MI 49007, USA
Received 10 December 2004; revised 22 December 2004; accepted 25 January 2005
Abstract—For the purpose of drug interaction studies, the stable-isotope labeled [13C3]midazolam and its metabolite, 10-hydroxy-
[13C3]midazolam were synthesized in four and five steps in overall yields of 25.5% and 14.2%, from 7-chloro-5-(2-fluorophenyl)-2-
(N-nitrosomethylamino)-3H-1,4-benzodiazepine, respectively, by a convergent synthesis, in which a key imidazoline ring formation
was achieved by the facile reaction of [13C]2-aminomethyl-7-chloro-2,3-dihydro-5-(2-fluorophenyl)-1H-1,4-benzodiazepine with
varying ethyl imidate hydrochlorides. The scrambling of C-3 and C-4 labeling in intermediate diamine, and consequently in the final
products, as well as the formation of a D4,5 isomer of 10-hydroxy-[13C3]midazolam was observed and explained.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
methane, catalytic reduction of the resulting unsatu-
rated-nitro compound 2, cyclization of diamine 3 with
triethyl orthoacetate, and oxidation of imidazoline 4
with MnO2. Such small molecules as nitromethane and
ethyl orthoacetate involved in the synthesis could be
the source of stable-isotope labeled reagents. However,
the synthesis of unlabeled 10-hydroxymidazolam 8, re-
ported by the same authors, was achieved by a highly
laborious linear synthesis that involved the N-oxide forma-
tion of midazolam, Polonovsky rearrangement of the
2-N-oxide or di-N-oxide 6 to 1-acetoxymidazolam or
its 5-N-oxide analog 7, and removal of oxygen from
any 5-N-oxide, followed by saponification of the acetoxy
group (Scheme 1). The overall synthesis from diamine 3
thus required six steps with a calculated yield of 1.8%.
Many important drugs are subject to oxidation by cyto-
chrome P450. Midazolam, being a substrate and also the
inducer of cytochrome P450, is often used for drug inter-
action study to see how the oxidation of a particular
drug may be influenced by the presence of other drugs,
which are also subject to oxidation, by the enzyme.2
Ordinarily the drug interaction study by IV route and
by oral route needs to be studied separately. By using la-
beled midazolam and 1-hydroxymidazolam, the method
can be modified so that the effects by the two routes of
administration can be determined simultaneously.
Several syntheses of unlabeled midazolam have been re-
ported in literature.3 Considering the availability and
cost of labeled reagents, we aimed to put the label into
the molecule as late in the synthetic scheme as possible
and to use the least amount of labeled reagents possible.
Conceivably, stable-isotope labeled midazolam 5 may be
synthesized by an adaptation of WalserÕs procedure3a
(Scheme 1), involving the sequential condensation of a
2-N-nitrosomethylamino benzodiazepine 1 with nitro-
We now wish to report the first isotope-synthesis, which
is convergent and allows the preparation of both car-
bon-13 labeled midazolam 13a and 10-hydroxymidazo-
lam 14 from the common labeled intermediate diamine
10 (Scheme 2).
2. Synthesis of the common labeled intermediate
diamine 10
Keywords: Drug interaction; Isotope labeling; Midazolam; 10-Hydroxy-
midazolam; Imidazoline ring formation; Imidate hydrochloride.
q See Ref. 1.
2-N-Nitrosomethylamino benzodiazepine 1 was pre-
pared (Scheme 2) according to a published procedure.3
Compound 1 was treated with nitro[13C]-methane to
*
Corresponding author. Tel.: +1269 833 6454; fax: +1269 833
7989; e-mail: yinsheng.zhang@pfizer.com
0040-4039/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2005.01.128

2088
Y. Zhang et al. / Tetrahedron Letters 46 (2005) 2087–2091
MeNO2
O2N
H2N
N
N
O
(10 equiv.)
Me
DMF
H
H
H
Ra-Ni
H2,155psi
N
N
N
tBuOK
N
N
Cl
83%
Cl
N
53%
Cl
F
F
F
2
N
N
1
3
N
N
CH3C(OEt)3
(3 equiv.)
xylene
MnO2
Toluene
m-CPA
11%
N
Cl
N
Cl
45%
68%
F
F
O
N
4
5
HOH2C
N
AcOH2C
N
1) PCl3
2) MeONa
MeOH
Ac2O
reflux
N
N
N
N
N
Cl
Cl
N
80%
Cl
O
F
71%
O
F
F
8
6
7
Scheme 1. Original synthesis of midazolam and 10-hydroxy-midazolam.
*MeNO2
O2N
H2N
N
N
O
(1.2equiv.)
DMF
H
H
#
H
*
Ra-Ni
Me
N
N
N
H2, 50psi
tBuOK
#
N
N
Cl
91%
Cl
N
50%
Cl
F
F
F
9
10
1
*
N
*
N
*
N
X
*
N
X
H
*
Cl
#
#
#
N
*
MnO2
X
#
OEt
Toluene
N
Cl
N
11a/11b
Cl
THF, EtOH
F
F
50-60%
78-93%
11a) X=H
12a) X=H
12b) X=AcO
13a) X=H
13b) X=AcO
11b) X=AcO
*
N
*
N
HO
#
MeONa, MeOH
80%
#
*
= C
13C, # = 1/2 13C, 1/2 12
N
Cl
F
14
Scheme 2. Convergent synthesis of [13C3]midazolam and 10-hydroxy[13C3]midazolam.
give the nitro derivative 9, which was then hydrogenated
in MeOH–THF in the presence of Raney nickel to pro-
vide the crucial common labeled intermediate 10.
affecting the yield. In the synthesis of 10 through hydro-
genation of 9 we initially obtained highly inconsistent
results, often encountering poor yields with large num-
ber of impurities if the progress of the reduction reaction
went slowly. The data showed that achievement of fast
initial reduction was necessary to avoid the impurities
found in slow reactions. We reasoned that fast initial
reaction should be favored by the use of active catalyst,
large catalyst/substrate ratio. Indeed, using a high cata-
lyst/substrate ratio (4/1) and fresh catalyst, we were able
In the literature procedure for the synthesis of unlabeled
diamine 3, a large amount of nitromethane (10 equiv)
was used for the reaction. However, by increasing the
concentration of reactants in the reaction mixture, we
were able to decrease the molar ratio of the labeled
nitromethane from 10-fold to 1.25-fold without largely

Y. Zhang et al. / Tetrahedron Letters 46 (2005) 2087–2091
2089
H2N
H
*
N
NH2
+ H2O
-H2O
O
Cl
Cl
H2N
H2N
F
H
*
H
N
N
*
A
N
Cl
N
Cl
H
F
F
N
HN
*
NH
F
10a
10b
B
Scheme 3. Possible mechanisms for the scrambling of C-3 and C-4 labeling.
to achieve high yield, fast reaction rate, and pure prod-
uct. We also found that during hydrogenation, a scram-
bling of the 13C label occurred. Two possible
mechanisms for the scrambling are shown in Scheme
3: one involving ring-opening by water (from
Raney nickel), the other resulting from direct nucleo-
philic attack by the primary amine on the imino car-
bon in the diazepine ring. Partial de-fluorination (5%)
also occurred during the hydrogenation as a side
reaction.
We reasoned, however, that the protonated imino group
in an iminoester hydrochloride (such as 11a and 11b),4,5
should be much more reactive (than the corresponding
orthoesters 16a and 16b) to the nucleophilic diamine
function in 10. An initial reaction of the primary amine
in 10 would lead to an intermediate 23a and 23b still
containing a protonated imino group, in a favorable po-
sition to react with the c-aromatic amino group to form
an imidazoline ring by eliminating NH4Cl. Further-
more, the availability of varying substituted iminoester
hydrochloride would provide a facile convergent synthe-
sis to various midazolam derivatives. Thus, when ethyl
2-(p-methoxyphenoxy)acetamidate hydrochloride 11d
was reacted with diamine 3 (Scheme 5), the desired cyclic
compound 21 was produced in 37% yield at rt instead
of 150 ꢁC. However, compound 21 was not stable and
the ring-opened compound 22 was isolated in the next
oxidation reaction. Nevertheless, the results confirmed
our reasoning that the reagent, imidate hydrochloride
11, is much more reactive than triethyl orthoacetate,
and prompted us to apply various imidate hydro-
chlorides 11a and 11b to form the imidazoline ring
system. The results are partially shown in Table 1. It
was found that the treatment of diamine 10 with ethyl
imidate hydrochloride 11a in ethanol/THF at a 1:1.5
molar ratio at 0 ꢁC provided the product 12a in 93.5%
yield.6 Also similar reaction of diamine with acetoxy
imidate hydrochloride 11b in ethanol/THF gave 12b in
78% yield.
3. Imidazoline ring formation
The imidazoline ring formation was initially carried out
by the condensation of diamine 3 and triethyl orthoace-
tate (Scheme 1). The yields were not as good as Walser
reported,3a and were further decreased when the amount
of orthoacetate was reduced from 3 to 2 equiv. We
thought that the treatment of diamine 3 with triethyl
2-acetoxyorthoacetate 16b might produce a precursor
17 of 10-hydroxymidazolam 8. However, the condensa-
tion provided undesired 18 (71%) rather than the desired
compound 17 (Scheme 5), apparently due to the higher
reactivity of the acetoxy group in the reagent 16b
(Scheme 4). By changing acetoxy group to methoxy
group, we were able to prepare the desired cyclic com-
pound 19, but could not remove the methyl group with-
out decomposition. An attempt to convert ethyl 2-(p-
methoxyphenoxy)acetamidate hydrochloride 11d to the
corresponding orthoacetate was unsuccessful (Scheme
4).
Because of the significantly higher reactivity of the imi-
date hydrochlorides 11a/11b as compared with the
EtOH/ether
dry HCl gas
NH2Cl
OEt
2 EtOH
OEt
X
X
X
CN
*
15
*
OEt
*
*
OEt
80-84%
*
100%
*
11
16
a) X = H
a) X = H
a) X = H
b) X = AcO
c) X = MeO
b) X = AcO
c) X = MeO
b) X = AcO
c) X = MeO
d) X = 4-methoxyphenoxy
d) X = 4-methoxphenoxy
Scheme 4. Preparation of the labeled acetimidate 11a–d and orthoacetate 16a–c.

2090
Y. Zhang et al. / Tetrahedron Letters 46 (2005) 2087–2091
Ac
HN
N
H2N
N
AcO
H
N
H
16b
xylene, reflux
N
N
Cl
+
N
Cl
N
Cl
F
F
F
18 (71%)
3
17 (5%)
N
N
H3CO
H3CO
Cl
N
N
16c
MnO2
toluene
xylene, reflux
46%
N
3
Cl
N
65%
F
F
20
19
O
N
N
X
NH
N
X
H
N
N
MnO2
Toluene
11d
Cl
Cl
3
F
39%
THF, EtOH
F
x= 4-methoxyphenoxy
21
22
Scheme 5. Attempted new approaches to 10-hydroxymidazolam.
Table 1. A novel approach to labeled midazolam and 10-hydroxymidazolam
HClN
*
NH2
N
NH
#
*
N
X
*
*
X
#
#
#
H
N
NH
11a or 11b
-NH4Cl
#
#
N
N
Cl
see table 1
N
F
F
F
a) X=H
12a/b
10
23a/b
b) X=AcO
Entry
10/11 (mol)
Conditions
Yields (%)
1
2
3
4
5
6
7
8
10/11a = 1/2
10/11a = 1/2
10/11a = 1/1
10/11a = 1/1
MeOH, 0 ꢁC–rt, 10 h
37
71
32
41
93.5
35
40
78
MeOH, THF, 0 ꢁC–rt, 10 h
MeOH, THF, 0 ꢁC–rt, 10 h
EtOH, THF, 0 ꢁC–rt, 10 h
EtOH, THF, 0 ꢁC–rt, 3 h
EtOH, THF, 0 ꢁC–rt, 3 h
EtOH, THF, 0–10 ꢁC, 4 h
EtOH, THF, ꢀ78–5 ꢁC, 2 h
10/11a = 1/1.5
10/11b = 1/1.5
10/11b = 1/1.5
10/11b = 1/1.5
7
corresponding orthoacetates, and the fact that orthoace-
tates are generally prepared from the imidates as pre-
cursors (Scheme 4), the modified procedure was much
more convenient and led to significantly higher yields,
especially in the synthesis of 10-hydroxymidazolam.
Thus the latter was prepared in three steps from diamine
9 in 28.4% overall yield, while the reported synthesis via
the modification of midazolam required six steps with an
overall yield of 1.8% from diamine 9.
4. Oxidation of midazoline with MnO2
The conversion of imidazoline to imidazole was studied
using several kinds of reagents, such as Pd/C, BaMnO4,
DDQ, and MnO2. Our study showed that only the reac-
tion with MnO2 produced the desired product. It was
found that the source of MnO2 dramatically affected
the yields and material recovery rate. The reaction with
the reagent from Aldrich without any treatment gave

Y. Zhang et al. / Tetrahedron Letters 46 (2005) 2087–2091
2091
Supplementary data
*
*
N
N
*
N
*
N
X
HO
#
#
#
#
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.tetlet.
2005.01.128.
N
N
Cl
Cl
F
References and notes
24a) X=H, 24b) X=OH
Figure 1. Side products.
25
1. Zhang, Y. et al. 15th Annual Central US International
Isotope Society Chapter Meeting, Mason, OH, 2001.
2. Gorski, J. C.; Jones, D. R.; Haehner-Daniels, B. D.;
Hamman, M. A.; OÕMara, E. M., Jr.; Hall, S. D. Clin.
Pharmacol. Ther. 1998, 64(2), 133–135.
3. (a) Walser, A. et al. J. Org. Chem. 1978, 43(5), 936; (b)
Fryer, R. I. et al. J. Heterocycl. Chem. 1991, 28(7), 1661;
(c) Fryer, R. I. et al. J. Org. Chem. 1978, 43(23), 1661; (d)
Fryer, R. I. et al. J Org. Chem. 1978, 43(23), 4480; (e)
Field, G. F. et al. U.S. Patent 4,194,049, 1980; (f) Khann, J.
M. et al. European Patent 0835874, 1998; (g) Huber, J. E.
U.S. Patent 5,831,089, 1998; (h) Bender, K. et al. European
Patent 768310, 1997; (i) Kowalczyk, E. et al. Pol Patent
162851, 1994; (j) Walser, A. U.S. Patent 4,226,771, 1980.
4. Miyano, M.; Smith, J. N. J. Heterocycl. Chem. 1982, 19, 659.
5. Reagent 11a is stable for several months when stored at
<0 ꢁC. Reagent 11b was similarly prepared from 2-acetoxy-
[13C2]acetonitrile, which was obtained by treatment of
bromo[13C2]acetonitrile with potassium acetate. Reagent
11b was unstable at room temperature and completely
decomposed after about 6 months. Thus it is best to use a
freshly prepared reagent or one kept at <0 ꢁC and used
within 2 weeks.
poor recovery rate (38% by weight); however, the oxida-
tion with the new reagent prepared by boiling AldrichÕs
MnO2 in toluene for 5 h afforded high recovery rate even
with a high loading of the new MnO2. The results prob-
ably indicate that the MnO2 from Aldrich absorbed the
product irreversibly, while the new reagent was satu-
rated with toluene, and did not absorb the product.
Using the new reagent we were able to convert an unsta-
ble compound 12b to the desired compound 13b in
50% yield (Scheme 2). In the oxidation of acetoxy deriv-
ative 12b, it was especially essential to use mang-
anese dioxide that had been pretreated with toluene at
130 ꢁC and prepared in neutral and not basic
conditions.
5. Side products formation
6. Typical procedures for imidazoline ring formation: To a
solution of the crude compound 10 (12.5 g, 41 mmol) in
EtOH (100 mL) and THF (60 mL) was added compound
11a (13.0 g) in three portions at ꢀ10 ꢁC. The resulting
mixture was stirred at 0 ꢁC for 1h, and warmed up slowly
to rt during 2 h period. Then the solvent was evaporated
under vacuum at 25 ꢁC. To this residue was added 300 mL
of CH2Cl2 and NaHCO3 saturated solution. The aqueous
layer (pH about 7) was extracted with CH2Cl2 three times
(100 mL · 3). The combined organic layers were washed
with brine (50 mL · 3), and then dried over Na2SO4. The
solvent was evaporated to give the crude product 12a,
which was dried under vacuum and used for next reaction.
Mp: 143–145 ꢁC [lit.3a 142–145 ꢁC (unlabeled)]. 1H NMR
(CDCl3) d 7.6–6.9 (m, 7H), 4.6 (m, 1H), 4.1–3.2 (m, 4H), 1.3
(dd, 6.81, 28.7 Hz, 3H).
7. To a suspension of the crude compound 12a (ca. 41mmol)
in dry toluene (180 mL) was added MnO2 (80.0 g, Aldrich
MnO2 was heated in toluene for 5 h, dried under vacuum
for 24 h at 60 ꢁC). The mixture was stirred at 110–115 ꢁC
for 1.5 h. The solid was filtered off, washed with toluene and
CH2Cl2. The solvent was evaporated under reduced
pressure to give the crude product 13a (12.4 g), which was
applied to flash chromatography (Biotage, 250 mL of
CH2Cl2/MeOH/Et3N = 500/10/1 treated the column first,
then eluted with 2% MeOH/CH2Cl2). The product was
further recrystalized from CH2Cl2/ether/hexane = 1/2/4.
Total weight: 8.0 g (60%). Mp: 159–161 ꢁC. [lit.3a 158–
160 ꢁC (unlabeled)]. 1H NMR (CDCl3) d 7.61(dt, .17,
7.5 Hz, 1H), 7.54 (dd, 2.4, 8.5 Hz, 1H), 7.43 (m, 1H), 7.37
(d, 8.7Hz, 1H), 7.24–7.20 (m, 1H), 7.01 (m, 1H), 6.92 (d,
10.7 Hz, 1H), 6.93 (dd, 190.2, 10.7 Hz, 1H), 4.93 (ddd, 88.2,
76.2, 14.1 Hz, 1H), 4.02 (ddd, 88.1, 77.8, 13.5 Hz, 1H),
2.54 (dd, 129.2, 7.2 Hz, 3H). 13C NMR (CDCl3) d 15.0
(d, 13CH3), 46.3 (13CH2N@), 124.2 (d, 13CH@N); 144.3 (d,
N13CH@N); 19F NMR ꢀ112.58; MS 329 [(M+3)+H]+.
The defluorinated contaminant (2–5%) in diamine 10,
which was produced during reduction over Raney Ni,
was carried to the final products 24a and 24b (Fig. 1).
Suppression of this side reaction would eliminate subse-
quent purification problems.
The preparation of 14 was accompanied by the forma-
tion of a very insoluble side product 25 (5%). It might
have formed by isomerization initiated by methoxide
abstraction of the proton at C4.
In summary, stable-isotope labeled midazolam (99.56%
purity) and 10-hydroxymidazolam (98.3% purity) were
synthesized using a novel efficient imidate approach. It
was found that acetimidate is a fairly stable and more
efficient labeling reagent than orthoacetate. Our study
also showed that the pretreatment of commercially
available MnO2 was necessary for high recovering rate
and yields. An interesting scrambling of label position
was observed and rationalized. Several by-products were
isolated and identified by means of spectroscopy and
X-ray crystal structure. These modified syntheses
provided much higher yields than those from the
literatures.
Acknowledgements
We thank Dr. Om Goel and Dr. Brian Tobias for help-
ful discussions and Dr. John R. Rubin for determining
single crystal X-ray structures.