A scalable route to the SMO receptor antagonist SEN826: Benzimidazole synthesis via enhanced in situ formation of the bisulfite-aldehyde complex

Article abstract of DOI:10.1021/op4002092

A practical and scalable route to the SMO antagonist SEN826 1 is described herein, including the discussion of an alternative approach to the synthesis of the target molecule. The optimized route consists of five chemical steps. A new and efficient access to the key intermediate 6 via the bisulfite-aldehyde complex was developed, significantly enhancing the yields and reducing costs. As a result, a synthetic procedure for preparation of multihundred gram quantities of the final product has been developed.

Full text of DOI:10.1021/op4002092

Article
pubs.acs.org/OPRD
A Scalable Route to the SMO Receptor Antagonist SEN826:
Benzimidazole Synthesis via Enhanced in Situ Formation of the
Bisulfite−Aldehyde Complex
Matteo Betti,*,† Eva Genesio,‡ Guido Marconi,¶ Salvatore Sanna Coccone,§ and Paul Wiedenau⊥
†Process Chemistry Unit, Siena Biotech SpA, 53100 Siena, Italy
‡Compound Management & Analysis Unit, Siena Biotech SpA, 53100 Siena, Italy
S
* Supporting Information
ABSTRACT: A practical and scalable route to the SMO antagonist SEN826 1 is described herein, including the discussion of an
alternative approach to the synthesis of the target molecule. The optimized route consists of five chemical steps. A new and
efficient access to the key intermediate 6 via the bisulfite−aldehyde complex was developed, significantly enhancing the yields
and reducing costs. As a result, a synthetic procedure for preparation of multihundred gram quantities of the final product has
been developed.
INTRODUCTION
■
The SMO receptor mediates Hedgehog (Hh) signaling critical
to development, differentiation, growth, and cell migration.1 In
normal conditions, activation of the pathway is induced by
binding of specific endogenous ligands (i.e., Sonic Hh) to its
receptor Patched (Ptch), which in turns reverts the Ptch
inhibitory effect on SMO. SMO activation ultimately
determines specific target genes activation through a family of
three transcription factors, Gli1, Gli2 and Gli3. Although Hh
signaling is significantly curtailed in adults, it retains functional
roles in stem cell maintenance, and aberrant Hh signaling has
been described in a range of tumours.2 Mutational inactivation
of the inhibitory pathway components results in a constitutive
ligand-independent activation seen in tumours such as basal cell
carcinoma (BCC) and medulloblastoma. Ligand-dependent
activation is seen in tumours such as prostate cancer, pancreatic
cancer, gastrointestinal malignancies, melanoma, gliomas, breast
cancer, ovarian cancer, leukemia, and B-cell lymphomas. A
significant body of evidence supports the conclusion that SMO
receptor antagonism will block the downstream signaling
events.3,4
Figure 1. SMO receptor antagonist SEN826.
path in Scheme 1). Sodium hydrogen sulfite is used to promote
the condensation of the corresponding o-phenylenediamine
with the Br-aromatic aldehyde 3.6b The next step is the
coupling of the aryl bromide with isonipecotic ethyl ester in
Buchwald conditions. After acidic hydrolysis with HCl under
microwave irradiation, the final amide 1 was synthesized with
CDI as coupling agent. The major issues of this pathway for a
multihundred gram scale are:
As part of a program to address unmet medical need with
regard to tumours in the CNS, Siena Biotech has designed and
investigated selective antagonists of the SMO receptor. The
newly designed API development candidate SEN826 1 (Figure
1) is part of a group of potent antagonists of the Hedgehog
pathway.5
(a) poor yield in the synthesis of the bromobenzene
intermediate 6;
(b) poor reproducibility (in terms of conversion, impurity
profile, reaction time) of the Buchwald coupling when
increasing the scale;
(c) silica gel purification of each intermediate;
(d) final product 1 as free base is an amorphous solid, which
is difficult to handle.
RESULTS AND DISCUSSIONS
■
MedChem Synthesis. The first synthesis of SEN826,
carried out in our Medicinal Chemistry laboratories, delivered a
few grams of product to support in vitro testing (Scheme 1).
The synthesis starts with the formation of the 2-
arylbenzimidazole derivative 6 which can be carried out starting
from N-methylphenylenediamine 2 (Method A; blue path in
Scheme 1) or employing o-phenylenediamine 4 in the ring
closure reaction followed by N-methylation (Method B; orange
Evaluation of Synthetic Alternatives. At first we decided
to investigate an alternative scalable synthetic pathway. This
new route to SEN826 1 is shown in Scheme 2. Preliminary
attempts were carried out on gram scale (1−2 g) in order to
Received: August 7, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. MedChem Route to compound 1
Scheme 2. Alternative synthetic pathway to compound 1
evaluate HPLC profile (time of conversion, major impurities)
and feasibility of each reaction.
suitable functional group for the ring-closure step to SEN826.7
Moreover, Buchwald couplings on this compound with primary
amines have already been described.8 The reactivity between
the 3-Br-phenyl derivatives and isonipecotic ethyl ester or
amine 9 (Scheme 2, orange arrow) was tested using the same
conditions of the original Medchem procedure (Scheme 1).9
All the experiments gave clean HPLC profile but lower yields
3-Br-benzoic-tert-butyl ester and 3-Br-phenyl-1,3-dioxolane
were considered to be suitable starting materials, as they
contain a masked group (respectively a carboxylic acid or an
aldehyde) that would allow selective basic hydrolysis of the
ethyl ester moiety and, later in the synthesis, would represent a
B
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(60%) with respect to the original route. Buchwald coupling
between Br-phenyl derivatives and isonipecotic acid (Scheme 2,
blue arrow) gave only poor conversion and was not pursued
any further.
Next, the direct amidation of the ester (12 or 13, Scheme 2)
with N-methyl-piperazine was examined. By carrying out the
reaction in neat conditions, only low conversions (25% a/a by
HPLC) were observed. NaH and LDA as strong bases to
deprotonate the amine did not improve conversion in this case.
Running the reaction with DABCO−AlMe3 complex,12 an air-
stable adduct of trimethyl aluminum, also resulted in low
conversion (35% a/a by HPLC), although with a very clean
HPLC reaction profile. The ester hydrolysis with aqueous
NaOH in dioxane followed by CDI-mediated amidation in
acetonitrile worked with moderate yields (yield on two steps:
40−45%) and turned out to be, in our opinion, the best way to
obtain the desired amide 15 or 16 (Scheme 2).
At this point, under time and resource limitations the final
ring-closure step of the alternative sequence was not tried, and
the Medchem pathway was selected for the optimization. The
original synthesis appeared much more attractive because of its
higher yields in Buchwald coupling, hydrolysis, and amidation
stages. However, a useful indication for the design of a new
synthetic approach was found: amine 9 underwent the desired
cross-coupling and would be an interesting intermediate for an
alternative approach. Because of its high cost, a two-step lab
synthesis of 9 was developed with an overall yield of 74% giving
50 g of 9 as free base in excellent NMR purity after distillation
under reduced pressure (Scheme 3).
Figure 2. Workflow of 250 g and 500 g process campaigns.
Scheme 3. Two-step synthesis of amine 9
profile determined by HPLC−MS analysis is reported below
(Figure 3).
Starting amine 2 (<1% a/a) and 2% a/a of 3-Br-benzoic acid
(peak 3 was the major impurity of the commercial aldehyde 3)
were detected together with the desired product 6 (peak 1,
Figure 3) and two unidentified impurities (peaks 2 and 4,
Figure 3). We hypothesized their structures solely by mass
spectra interpretation (Figure 4).
In the next 20-g scale experiment a quite complex sequence
of washes was necessary in order to eliminate these impurities
(Figure 5).
The convergent route via amine 9 could be worth
considering, provided that the Cbz-protected amine 10 can
be sourced at a reasonable price.11
Fortunately, during the scale up experiment of the first
campaign (250 g scale) a solid precipitated while cooling the
reaction mixture to room temperature. After filtration, a
treatment with active charcoal followed by a trituration with
cyclohexane (3 vol) was necessary to obtain the desired
intermediate 6 as a brown solid in 94% HPLC purity but only
in 49% yield. Method B. Starting from 1,2-phenylidenediamine
4, the benzimidazole synthesis on gram scale gave a relatively
low yield range (20−37%) and a messy HPLC profile under the
same conditions employed in method A. Longer reaction time
(18 h) or further addition of sodium hydrogen sulfite (up to 0.9
equiv) did not affect the yield or the reaction.13 We observed a
slight improvement in the HPLC profile of the condensation,
maintaining the reaction temperature at 60 °C over 6 h (43%
a/a of 5) instead of 75 °C (35% a/a of 5). The HPLC−MS
check at the end of reaction is reported below (Figure 6).
As in method A, starting amine 4 (<1% a/a) and 3-Br-
benzoic acid were detected. The latter (peak 2, m/z = 199 [M
− 1]−) had the same retention time of an unidentified
oligomeric impurity with m/z = 547 [M + 1]+. This explained
Process Optimization. Two campaigns were carried out to
synthesize SEN826 batches for preclinical studies (Figure 2).
The first one furnished 250 g of the final compound 1 as HBr
salt and included some reaction optimization work; the aim of
the second campaign, which furnished 500 g of final product as
HBr salt, was to improve the yield of the initial stages of the
synthesis. The achieved results are described in detail in the
corresponding section.
Stages 1 and 2 - Comparison between Methods A
and B. 250 g Campaign. Method A. Studies directed toward
the synthesis of the benzimidazole ring began by exploring on
gram scale the feasibility of employing 3-Br-benzaldehyde 3
(1.1 equiv), N-methylphenylenediamine 2 (1.0 equiv), and
sodium hydrogen sulfite (0.5 equiv) in ethanol (10 vol) as
solvent. After 3 h at 75 °C (Tint) we observed partial conversion
(80% a/a by HPLC). On further addition of 0.2 or 0.3 equiv of
compound 3, the reaction was deemed to be complete by
HPLC analysis (respectively in 3 or 2 h). The relative impurity
C
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 3. HPLC−MS check at the end of reaction (Method A).
Scaling the method B procedure to 250 g of starting material
4, we recovered after filtration the desired product 5 in 95%
HPLC purity and 37% yield. The N-methylation of the
benzimidazole ring was afforded with NaH (1.3 equiv) in THF
(3 vol) at 25 °C in order to isolate 6 in 61% yield and excellent
purity (99.1% by HPLC).
We concluded that the two-step procedure (Method B),
despite the lower yield, had important advantages with respect
to direct formation of 6 (Method A): the use of the cheaper
and more stable starting phenylenediamine 4; the easier
workup procedure which avoided the active charcoal treatment;
the higher purity of the isolated product which improved the
rate of conversion and the general profile of the subsequent
Buchwald coupling (see Stage 3 - Buchwald Coupling).
500 g Campaign. At the beginning of the second campaign,
we focused our attention on the first two steps of Method B
which had 23% overall yield and needed a further optimization.
Adjusting the order of addition of the reactants, we confirmed
that the water insoluble bisulfite−aldehyde complex 14 could
be quantitatively formed by addition of the ionic nucleophile
Figure 4. Hypothesized structures a, a′ and b for the major HPLC
impurities peaks of Method A.
the increased % a/a of the corresponding signal with respect to
Figure 3. We found in literature that in the absence or with a
substoichiometric amount of the oxidizing agent (we employed
0.5 equiv), the reaction may lead either to 2-substituted
benzimidazole or to “aldehydine” d by the o-phenylenediimine
c rearrangement (Scheme 4).16 Peaks 3 and 4 in Figure 6
actually confirmed this side reaction; since they had the same
m/z value, we carried out a further reaction between o-
phenylenediamine 4 and aldehyde 3 (without sodium bisulfite)
in EtOH in order to distinguish o-diimine c from aldehydine d.
Stirring 4 and 3 in ethanol at room temperature we only
observed (by HPLC−MS) the condensation reaction to obtain
the diimine intermediate c (retention time and [M + 1]+
identical with peak 4 in Figure 6), which rearranged to give d
(retention time and [M + 1]+ identical with peak 3 in Figure 6)
under heating at 60 °C.
6a,14
‑
HSO3 as reported in Scheme 5.
After the addition of ethanol as cosolvent (5.5 vol) to the
aqueous suspension, the adduct 14 seemed to react (on 5 g
scale) with 1,2-phenylenediamine (1.1 equiv) in a cleaner way
compared to the free aldehyde 3. After heating at 70 °C for 2 h,
the cyclization reaction was complete by HPLC analysis with
11% a/a of the aldehydine d.
Repeating the experiment on 50 g scale, we adjusted the
reaction conditions in sight of a 500 g scale up. The volumes of
Figure 5. Sequence of washes used to isolate 6 on 20-g scale experiment.
D
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 6. HPLC−MS check at the end of reaction (6 h at 60 °C; Method B).
Scheme 4. Aldehydine formation in the absence of NaHSO3
a large amount (25% a/a by HPLC) of the bis-methylated side
product f (Figure 7) lowered the yield to 63%.
Scheme 5. In situ formation of bisulfite−aldehyde complex
14
Figure 7. Chemical structure of the benzimidazolium iodide f.
solvent (H2O/EtOH = 1/1) were reduced from 11 to 7; a
stoichiometric amount of 4 was used and its rate of addition
was regulated in order to maintain the internal temperature of
the system below 30 °C. The HPLC−MS check at the end of
reaction confirmed a clean reaction profile with major
impurities (c and d) lower than 2% a/a. On 500 g scale, we
finally obtained the desired benzimidazole 5 in 82% yield and
98% HPLC purity.
Direct formation of the benzimidazole 6 by cyclization of N-
methylphenylenediamine 2 with the aldehyde/bisulfite complex
14 was attempted. In the same conditions reported above, we
observed complete HPLC conversion after 18 h at 70 °C. A
tarry reaction mixture was obtained; after the workup, we
recovered 6 as a pale-brown solid in 75% yield and 90% HPLC
purity.
We tried to contain the side reaction by adding MeI (1
equiv) to the suspension of benzimidazole 5 (1 equiv) and
potassium carbonate (2 equiv) in acetone (7 vol) while heating
at 45 °C (Tint). In this case the conversion was very low (40%
a/a by HPLC after 1 h) and did not change after further
addition of MeI (0.5 equiv).15 Next, we decided to direct our
attention to the employment of a stronger base for the
benzimidazole deprotonation. The system KOH/DMSO (2.5
equiv in 2 vol) turned out to be optimal for our purpose: on 10
g scale the reaction worked at room temperature, and complete
conversion was achieved within 20 min. A simple workup
(water addition and filtration) gave the product 6 in 97% yield
and 99% HPLC purity. Carrying out the experiment on 40 g
scale the concentration was decreased using 3 volumes of
solvent in order to minimize the formation of thick slurry
throughout the reaction. We observed an exotherm during the
deprotonation stage (ΔT = +7 °C, 2.5 equiv of KOH added all
at once), the methyl iodide addition (ΔT = +8 °C, 1.0 equiv
added over 20 min) and the quench with water (ΔT = +15 °C,
Moving to the methylation step, we initially employed MeI
(1.2 equiv), potassium carbonate (2 equiv) and acetone (7 vol),
but no reaction was observed while stirring them together with
5 at room temperature. Even if complete conversion was
achieved upon heating at 35 °C (Tint) for 18 h, the formation of
E
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 8. NMR spectra of contaminated 6 and the impurity e.
2 vol added over 45 min at 25 °C). Even if the product 6 was
recovered in 86% yield and 97% HPLC purity, NMR analysis of
the filtered compound showed free aniline peaks, confirming
the presence of a partially ring-opened structure (3% mol by
NMR spectrum, Figure 8) with m/z = 319 (by HPLC−MS).
By the comparison between NMR spectra of contaminated
compound 6 and the impurity 319 m/z (see Supporting
Information for isolation of m/z 319), we clarified its structure
which conforms to the “pseudo base” e originated from
benzimidazolium iodide f by reaction with potassium hydroxide
(Figure 8).16 This was probably due to the excess of KOH
employed that generated a strongly basic aqueous environment
during the reaction quench.
This issue was solved by adding MeI over 40 min (in order
to minimize the formation of f), lowering the amount of base
employed (2.0 equiv instead of 2.5 equiv) and quenching the
reaction at 10−15 °C (Tint). In the second campaign
experiment (600 g scale), 1.1 equiv of MeI (instead of 1.0
equiv) was necessary to complete the conversion.17 The
efficient heat exchange of the 10 L jacketed reactor allowed an
improved control of the exotherms, while the preliminary
calorimetric study of the reaction mixture by DSC analysis
confirmed a low thermal risk. After crystallization from iPrOAc
(2.5 vol), N-methylbenzimidazole 6 was isolated in 92% yield,
98% HPLC purity and clean NMR profile.
carried out by varying one at time the following parameters:
solvent, batch of benzimidazole 6, temperature, catalyst/ligand
loading, cesium carbonate and ethyl isonipecotate equivalents.
Solvent. Toluene and 1,4-dioxane gave the same HPLC
reaction profile in preliminary trials. Toluene was preferred
since it was a suitable solvent for the crystallization of the
desired product 7.
Batch of Benzimidazole 6. In the experimental trials we
observed that rate of conversion was lower when employing the
aryl bromide 6 obtained from Method A (94% HPLC purity)
with respect to Method B (99% HPLC purity).
Temperature. Employing 5% mol of catalyst/ligand, 5 equiv
of Cs2CO3 (Alfa Aesar batch, vide infra) and 2 equiv of ethyl
isonipecotate at 85 °C (Tint) in toluene (10 vol) the conversion
was partial (92% a/a by HPLC) in 18 h; in the same lapse of
time at 100 °C (Tint), a complete conversion by HPLC analysis
was observed employing the same batch of 6.
Ethyl isonipecotate. In order to reduce costs and facilitate
workup, 1.1 equiv (instead of 2) of ethyl isonipecotate were
used without negative effect on the rate of conversion.
Cesium Carbonate. We employed 5 or 3 equiv of base and
an identical rate of conversion was observed at 100 °C (Tint) in
both experiments. Literature precedent indicates a strong
influence of the Cs2CO3 particle size on the reaction rate of Pd-
catalyzed N-alkylations.18 In fact, we observed that when using
a batch of Cs2CO3 with a smaller particle size, the coupling
reaction was fast and complete conversion was achieved in less
Stage 3 - Buchwald Coupling: Reproducibility at
Larger Scales. Optimization of the coupling reaction was
F
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a
than 18 h. We qualitatively confirmed by microscope (Figure 9)
that the Cabot high-purity grade Cs2CO3 (Aldrich product no.
Table 1. Preliminary screening of different catalyst/
phosphine loadings (1:1 molar ratio)
b
Pd(OAc)2/BINAP loading
(% mol)
conversion percentage /time at
entry
1
100 °C (Tint)
2%
5%
55% - 4 h
62% - 5 h
74% - 7 h
100% - 18 h
61% - 2 h
79% - 4 h
95% - 7 h
100% - 18 h
85% - 5 h
100% - 8 h
100% - 3.5 h
2
Figure 9. Comparison between the particle sizes of different batches of
Cs2CO3.
3
4
8%
562572) had a smaller particle size than the initially used
Cs2CO3 (Alfa Aesar product no. 12887). We milled some of
the poorly performing Cs2CO3 with pestle and mortar, and
tested this material in a Buchwald reaction with a similar
substrate where it performed equally well compared to the
Aldrich Cs2CO3. For convenience, however we purchased and
used only the Aldrich material from thereon.
10%
a
Poorly performing Cs2CO3 was employed during this preliminary
b
screening. % a/a by HPLC analysis of a sample of the reaction
mixture.
filtrate to an optimal volume (3 vol) produced a precipitated
product with the desired impurity profile. So the desired ester 7
was isolated after crystallization from toluene in 77% yield
(instead of 70% as in the MedChem procedure, Scheme 1) and
98% HPLC purity.
By replacing cesium carbonate with NaOtBu, the tert-butyl
ester g (Figure 10) was observed as side product (5% a/a by
HPLC−MS).
Stage 4 - Ester Hydrolysis. The synthesis of intermediate
8 was achieved, without further optimization, employing basic
hydrolysis conditions (NaOH 15 wt %; 1.8 equiv) in dioxane
(10 vol) while heating at 70 °C (Tint) for 2 h in both
campaigns. After the workup, the desired product 8 was
isolated by filtration at pH = 5 in more than 90% yield and
more than 98% HPLC purity.
Stage 5 - Amidation and Salt Formation. Initially, for
the synthesis of the amide 1 the direct aminolysis of the starting
ester was considered. A literature survey showed that a catalytic
amount (0.3 equiv) of triazabicyclodecane (TBD) promotes
aminolysis.20 On gram scale the reaction was performed in a
pressure tube, heating at 95 °C overnight. Purification by silica
column gave the product in 92% yield. However, the scalability
of this promising approach was not investigated further because
of time limitations.
The synthesis of the final intermediate 1 was achieved by
employing CDI as coupling agent (1.2 equiv) in acetonitrile
(8.5 vol) at 55 °C, without using TEA as in the MedChem
procedure (Scheme 1). The presence of residual water in acid 8
(KF = 1−1.5 wt %) required an overcharge of the coupling
agent to achieve full conversion to intermediate imidazolide. In
these conditions the reaction profile was clean (by HPLC) and
the conversion complete in 2 h. The main issue of this step was
the final purification of the amorphous free base 1. The
formation of a filterable salt was then examined in order to
enable the purification of the API by crystallization and ensure
a suitable and stable physical form. A preliminary screen of four
acids (fumaric, sulfuric, hydrochloric and hydrobromic acid)
showed that the fumarate and the hydrobromide salt
precipitated from an ethanol solution of the free base. While
the fumarate appeared to be hygroscopic, the hydrobromide
formation was carried out in the 250 g campaign to investigate
its purity and recovery. The crude free base isolated after the
campaign run contained imidazole (10% mol by NMR) as
impurity. It was dissolved in EtOH (4 vol) and treated with 48
wt % aqHBr at 45 °C, followed by seeding the solution and
cooling first to 25 °C then to 5 °C. This gave the salt in 90%
Figure 10. Side-product g observed by replacing Cs2CO3 with
NaOtBu.
Mechanical Stirring. We also observed that on a smaller
scale (10−50 g) mechanical stirring improved the reaction rate,
as opposed to magnetic stirring, underlining the importance of
efficient agitation when using cesium carbonate in heteroge-
neous reaction systems.
Catalyst/Ligand. Even if during the preliminary trials the
reaction worked well both with Pd2(dba)3/Xphos in dioxane or
Pd(OAc)2/BINAP in toluene, we chose the latter reagents due
to their lower cost. Maintaining the temperature at 100 °C, we
carried out preliminary experiments with different catalyst
loadings (Table 1).
Using 2% mol of catalyst a complete conversion by HPLC
was achieved after 18 h; in this case we observed the formation
of the dehalogenated starting benzimidazole (5% a/a by
HPLC), however removable by trituration in cyclohexane.19
We decided to employ these conditions for the scale up during
the 250 g campaign, with a considerable cost saving and making
the purification easier because of the employment of less
BINAP. We isolated the desired product 7 in 53% yield and
98% HPLC purity. During the next 500 g scale campaign the
catalyst loading was further optimized by premixing in toluene
3.5% mol of Pd(OAc)2 and 5% mol of phosphine in order to
favor the Pd(II) precatalyst reduction and improve the reaction
rate. Using cesium carbonate with a smaller particle size, we
achieved complete conversion in 3 h. Although the trituration
was determined to be potentially useful in the purging of
impurities, it was found that concentration of the toluene
G
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 6. 500 g campaign final synthetic route
yield with 99% HPLC purity, but still containing 6% mol of
imidazole by NMR analysis. Another crystallization from EtOH
(3.5 vol) completely removed the coproduct. Then, in spite of
drying in the vacuum oven (Toven = 65 °C, p = 10 mbar, 4 d),
the EtOH content was reduced only to 2.4 wt %. During drying
the colour changed from off-white to yellow without affecting
the HPLC purity and giving the desired compound in 80%
yield. It is important to note that by using HBr in ethanol for
the salt formation and drying the salt at 65 °C for prolonged
periods, the formation of EtBr (a genotoxic impurity) was a
potential issue. In the 500 g campaign, with the view to the
difficulty to remove residual ethanol from the HBr salt in the
first campaign, acetonitrile instead of EtOH was employed as
solvent for the salt formation. The salt 9 started to precipitate
already without adding a seed and was recovered in 71% yield,
99% HPLC purity and less than 1 wt % of residual solvent by
NMR (Scheme 6). However, replacing EtOH with acetonitrile
did not obviate the GTI issue, since acetamide can form
hydrolytically in the presence of HBr and trace of moisture.
These aspects have to be considered as forward-looking
statements for manufacture of clinical material in the later
stages of development.
(3.5% mol of precatalyst; 5% mol of ligand), reasons of
poor reproducibility at larger scales elucidated;
(3) No chromatographic purification;
(4) Five-step synthesis with good overall yield (37%); the
intermediates are all solids apart from the free base
intermediate 1;
(5) Generally low-cost (except Pd-catalyst) and nonhazar-
dous/toxic (except for MeI and phenylenediamine)
reagents.
EXPERIMENTAL SECTION
■
The reported yields are corrected for purity and water/solvent
content of the products. Generally, the reactions were
monitored by HPLC and purities/conversions quoted refer to
HPLC area % at 215 nm. HPLC method: Waters Symmetry
C18 3.5 μm 4.6 mm × 75 mm column; flow rate 1.0 mL/min;
mobile phase A: 10 mM aq K2HPO4 buffer or 0.1% aq formic
acid; mobile phase B: acetonitrile or acetonitrile 0.1% formic
acid; gradient (13 min): 95:5 A/B to 10:90 A/B over 10 min, 2
min at 10:90 A/B then 10:90 A/B to 95:5 A/B over 1 min.
UPLC−MS analyses were run using an Acquity Waters
UPLC equipped with a Waters SQD (ES ionization) and
Waters Acquity PDA detector. UPLC method: BEH C18 1.7
μm, 2.1 mm × 50 mm column; flow rate 0.6 mL/min; mobile
phase A: 0.1% NH4HCO3 in H2O 95%/MeCN 5%; mobile
phase B: 100% MeCN; gradient (4 min): 95:5 A/B to 25:75 A/
B over 2.1 min, 25:75 A/B to 15:85 A/B over 0.2 min then 1.0
min at 15:85 A/B, 15:85 A/B to 95:5 A/B over 0.2 min then
0.5 min at 95:5 A/B. Retention times were expressed in
minutes. Temperature: 40 °C. UV Detection at 215 and 254
nm. ESI+ detection in the 80−1000 m/z range.
CONCLUSIONS
■
Two synthesis campaigns of SEN826 as HBr salt were carried
out to give 250 g in the first campaign and 500 g in the second
campaign. In order to provide API material for preclinical work,
a suitable synthetic route was developed. An alternative route
was investigated initially, but was found to be inferior to the
original route; however, some useful indications for the design
of a new convergent approach were found. Thus, the original
route was further optimized. Its main features are:
Synthesis of 2-(3-Bromo-phenyl)-1H-benzimidazole
(5). Sodium bisulfite (281 g; 2.7 mol; 1.0 equiv) was charged
to the 10 L jacketed reactor under nitrogen flow and water (1.8
L; 3.5 vol) was added. A clear solution formed after 5 min
under stirring. Neat 3-Br-benzaldehyde 3 (500 g; 2.7 mol; 1.0
equiv) was added dropwise over 5 min. The heterogeneous
(1) Robust and high-yielding formation of the key
intermediate 6 via the in situ bisulfite−aldehyde complex
formation;
(2) Buchwald−Hartwig coupling with improved yield (77%),
short reaction time (3 h), reasonably low catalyst loading
H
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
mixture was stirred for 20 min while observing an exotherm
(ΔT = +5 °C; from 20 to 25 °C) and a white precipitate
formation. Ethanol (1.8 L; 3.5 vol) and 1,2-phenylenediamine 4
(307 g; 2.7 mol; 1.0 equiv) were respectively added. The
resulting mixture was heated to 70 °C (Tint), and after 3 h
HPLC analysis showed complete conversion. The reaction
suspension was cooled to room temperature (Tint = 20 °C) and
2.0 mol; 1.0 equiv) was added followed by Cs2CO3 (1.95 kg;
6.0 mol; 3.0 equiv) and toluene (0.2 L). The resulting
suspension was heated to 100 °C (Tint) for 3 h and then
checked by HPLC (conversion complete). The mixture was
cooled to room temperature and filtered through a cellulose
pad, washing with toluene (2 × 0.5 L). The mother liquors
were concentrated under reduced pressure (Tint = 70 °C; p =
400 mbar; 5.0 L of solvent distilled) and cooled to −20 °C
filtered on a Buchner funnel, washing with EtOH/H O = 1/1
̈
2
(2 × 1.0 L) and water (1 × 1.0 L). The product was dried in an
oven (Toven = 40 °C; p = 10 mbar; 18 h) until constant weight.
The intermediate 5 (605 g) was isolated in 82% yield as a white
solid.
(T ) for 1 h. The yellow suspension was filtered on a Buchner
̈
int
funnel and the cake was washed with toluene (2 × 0.15 L
cooled at −20 °C). The solid was dried in an oven until
constant weight (Toven = 40 °C; p = 10 mbar, 4 h). The desired
intermediate 7 (560 g) was isolated as a pale-yellow solid in
77% yield.
UPLC−MS: tR = 1.44 min; m/z = 273 [M + 1]+.
HRMS calcd for C13H10BrN2 [M + 1]+ 273.00274, found
273.00278.
UPLC−MS: tR = 1.76 min; m/z = 364 [M + 1]+.
HRMS calcd for C22H26N3O2 [M + 1]+ 364.20251, found
364.20248.
HPLC: tR = 5.04 min; 97.5% purity.
Water content (KF): 0.06 wt %.
1H NMR (400 MHz DMSO-d6): δ 8.36 (t, J = 1.5 Hz, 1H),
8.18 (dt, J = 7.8, 1.5 Hz, 1H), 7.65 (dm, J = 7.9 Hz, 1H), 7.61
(m, 2H), 7.48 (t, J = 7.9 Hz, 1H), 7.20 (m, 2H).
13C NMR (100 MHz DMSO-d6): δ 150.3, 150.2, 133.08,
131.83, 129.57, 126.06, 123.10, 122.94, 119.8 (broad), 111.3
(broad).
HPLC: tR = 4.41 min; 98.1% purity.
1H NMR (400 MHz DMSO-d6): δ 7.66 (d, J = 7.8 Hz, 1H),
7.56 (d, J = 7.8 Hz, 1H), 7.36 (t, J = 7.8 Hz, 1H), 7.32 (m, 1H),
7.28−7.16 (m, 3H), 7.09 (m, 1H), 4.06 (q, J = 7.1 Hz, 2H),
3.83 (s, 3H), 3.71 (m, 2H), 2.82 (m, 2H), 2.48 (m, 1H), 1.90
(m, 2H), 1.66 (m, 2H), 1.17 (t, J = 7.1 Hz, 3H).
13C NMR (100 MHz DMSO-d6): δ 174.9, 154.3, 151.7,
143.13, 137.2, 131.5, 129.9, 122.9, 122.5, 120.0, 119.6, 117.7,
117.1, 111.1, 60.6, 48.5, 40.9, 32.3, 28.1, 14.8.
Synthesis of 2-(3-Bromo-phenyl)-1-methyl-benzimi-
dazole (6). Potassium hydroxide (241 g in pellets; 4.3 mol;
2.0 equiv) was charged to the 10 L jacketed reactor under
nitrogen flux and DMSO (0.5 L; 0.9 vol) was added. The
resulting suspension was stirred for 15 min while observing an
exotherm (ΔT = +5 °C; from 20 to 25 °C). Solid
benzimidazole 5 (588 g; 2.2 mol; 1.0 equiv) was then added
portion-wise (6 × ∼100 g) over 25 min; an exotherm was
observed during the addition (ΔT = +10 °C; from 20 to 30
°C). A solution of MeI (335 g; 2.4 mol; 1.1 equiv) in DMSO
(1.0 L) was added over 40 min (ΔT = +3 °C; from 20 to 23
°C), and a solid started to precipitate. After 2 h, HPLC analysis
showed complete conversion. Ice (2.0 kg) was added portion-
wise followed by water (1.0 L), maintaining the temperature at
15 °C (Tint). Next, the white suspension was stirred at 20 °C
Synthesis of 1-[3-(1-Methyl-benzoimidazol-2-yl)-phe-
nyl]-piperidine-4-carboxylic acid (8). The starting ester 7
(550 g; 1.5 mol; 1.0 equiv) was charged to the 10 L jacketed
reactor under nitrogen flux and dioxane (4.4 L) was added. The
suspension was stirred at 20 °C for 10 min in order to obtain a
clear homogeneous solution. A solution of NaOH (108 g; 2.7
mol; 1.8 equiv) in water (0.6 L) was added all at once; the
biphasic mixture was heated to 70 °C (Tint) under mechanical
stirring. After 2 h, HPLC analysis showed complete conversion.
The solvent was concentrated under reduced pressure (Tint
=
70 °C; p = 400 mbar; 4.0 L of solvent distilled). The residue
was diluted with water (2.0 L) and acidified with HCl 1 N to
pH = 4.7 (measured by digital pH-meter) while cooling at 20
°C (Tint). The white suspension was stirred at 20 °C for 3 h
for 2 h and then filtered on a Buchner funnel, washing with
̈
water (2 × 0.5 L). The filtered product was crystallized from
iPrOAc (2.5 vol) and dried in an oven (Toven = 40 °C; p = 10
mbar; 12 h) until constant weight. The desired intermediate 6
was recovered as pale-yellow solid in 92% yield.
and then filtered on a Buchner funnel, washing with water (2 ×
̈
0.5 L). The solid was dried in an oven (Toven = 40 °C; p = 10
mbar; 18 h) until constant weight (458 g). The desired product
8 was recovered as an off-white solid in 91% yield.
UPLC−MS: tR = 0.82 min; m/z = 334 [M + 1]−.
HRMS calcd for C20H22N3O2 [M + 1]+ 336.17121, found
336.17124.
UPLC−MS: tR = 1.60 min; m/z = 289 [M + 1]+.
HRMS calcd for C14H12BrN2 [M + 1]+ 287.01839, found
287.01842.
HPLC: tR = 4.10 min; 98.3% purity.
1H NMR (400 MHz DMSO-d6): δ 8.02 (t, J = 1.7 Hz, 1H),
7.85 (dt, J = 7.8, 1.7 Hz, 1H), 7.73 (dm, J = 7.8 Hz, 1H), 7.67
(d, J = 7.7 Hz, 1H), 7.60 (d, J = 7.7 Hz, 1H), 7.52 (t, J = 7.7 Hz,
1H), 7.30 (dt, J = 7.2, 1.2 Hz, 1H), 7.27 (dt, J = 7.2, 1.2 Hz,
1H), 3.87 (s, 3H).
HPLC: tR = 3.35 min; 98.7% purity.
Water content (KF): 0.9 wt %.
1H NMR (400 MHz DMSO-d6): δ 12.27 (broad, 1H), 7.65
(d, J = 7.3 Hz, 1H), 7.57 (d, J = 7.3 Hz, 1H), 7.35 (t, J = 7.9 Hz,
1H), 7.31 (t, J = 2.0 Hz, 1H), 7.27 (td, J = 7.3, 1.2 Hz, 1H),
7.22 (td, J = 7.3, 1.2 Hz, 1H), 7.16 (d, J = 7.9 Hz, 1H), 7.10
(dd, J = 7.9, 2.0 Hz, H), 3.84 (s, 3H), 3.71 (m, 2H), 2.82 (m,
2H), 2.41 (m, 1H), 1.90 (m, 2H), 1.65 (m, 2H).
13C NMR (100 MHz DMSO-d6): δ 176.6, 154.3, 151.7,
143.0, 137.2, 131.4, 129.9, 122.9, 122.6, 119.9, 119.6, 117.7,
117.1, 111.2, 48.6, 40.1, 32.3, 28.1.
13C NMR (100 MHz DMSO-d6): δ 152.0, 151.9, 143.0,
137.3, 133.1, 132.4, 131.5, 128.9, 123.4, 122.8, 122.6, 119.8,
119.7, 111.4, 32.3.
Synthesis of Ethyl 1-[3-(1-Methyl-benzoimidazol-2-
yl)-phenyl]-piperidine-4-carboxylate (7). Pd(OAc)2 (15.7
g; 0.07 mol; 3.5% mol), BINAP (63.9 g; 0.1 mol; 5% mol) and
toluene (5.0 L; 8.8 vol) were charged to the 10 L jacketed
reactor under nitrogen atmosphere. The resulting suspension
was stirred at 20 °C (Tint) for 30 min. A solution of ethyl
isonipecotate (346 g; 2.2 mol; 1.1 equiv) in toluene (0.5 L) was
added dropwise in 5 min. Next, 3-Br-benzimidazole 6 (570 g;
Synthesis of {1-[3-(1-Methyl-benzoimidazol-2-yl)-phe-
nyl]-piperidin-4-yl}-(4-methyl-piperazin-1-yl)-metha-
none Hydrobromide (9). The starting carboxylic acid 8 (450
g; 1.3 mol; 1.0 equiv) was charged to the 10 L jacketed reactor
under nitrogen atmosphere and suspended in MeCN (2.7 L;
I
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
6.0 vol) at 20 °C (Tint). CDI (253 g; 1.6 mol; 1.2 equiv) was
added all at once and the resulting mixture was heated to 50 °C
(Tint). After 1 h, the HPLC analysis of a sample (quenched with
butyl amine) showed complete activation. A solution of N-
methyl-piperazine (182 g; 1.8 mol; 1.4 equiv) in MeCN (1.1 L;
2.5 vol) was added dropwise in 15 min; next, the reaction
mixture was heated to 55 °C (Tint) for 2 h (conversion
complete by HPLC). The mixture was concentrated under
reduced pressure (Tint = 40 °C; p = 350 mbar; 3.5 L of solvent
distilled), the residue was diluted with DCM (2.5 L) and
washed with water (2 × 1.5 L). The organic layer was
concentrated under reduced pressure (Tint = 25 °C; p = 400
mbar; 2.0 L of solvent distilled) and the residue was diluted in
MeCN (1.7 L; 4 vol). Aqueous HBr 48 wt % (285 g; 1.7 mol;
1.3 equiv) was added dropwise over 30 min at 20 °C, observing
a slight exotherm (ΔT = +1 °C; from 20 to 21 °C). After the
addition was complete, the suspension was stirred at 20 °C
(Tint) for 3 h and then cooled to 5 °C (Tint) for 2 h. After
REFERENCES
■
(1) Jiang, J.; Hui, C. Developmental Cell 2008, 15, 801−812.
(2) Yang, L.; Xie, G.; Fan, Q.; Xie, J. Oncogene 2010, 29, 469−481.
(3) Mas, C.; Ruiz i Altaba, A. Biochem. Pharmacol. 2010, 80, 712−
723.
(4) LoRusso, P. M.; Rudin, C. M.; Reddy, J. C.; Tibes, R.; Weiss, G.
J.; Borad, M. J.; Hann, C. L.; Brahmer, J. R.; Chang, I.; Darbonne, W.
C.; Graham, R. A.; Zerivitz, K. L.; Low, J. A.; Von Hoff, D. D. Clin.
Cancer Res. 2011, 17, 2502−2511.
(5) Pericot Mohr G., Thomas R. J., Minetto G., Bellini M., Caramelli
C. PCT Int. Appl. WO/2010/142426.
(6) (a) Ridley, H. F.; Spickett, R. G. W.; Timmis, G. M. J. Heterocycl.
Chem. 1965, 2, 453. (b) Xiangming Han.; et al. Russ. J. Org. Chem.
2008, 44, 863−865.
(7) For the standard synthesis to benzimidazoles via cyclo-
condensation of o-phenylenediamine (or substituted o-phenylenedi-
amines) with carboxylic acids (or their derivatives) see: (a) Sluiter, J.;
Christoffers, J. Synlett 2009, 63. (b) Czarny, A.; Wilson, W. D.; Boykin,
D. W. J. Heterocycl. Chem. 1996, 33, 1393. (c) Tidwell, R. R.; Geratz, J.
D.; Dann, O.; Volz, G.; Zeh, D.; Loewe, H. J. Med. Chem. 1978, 21,
613. (d) Fairley, T. A.; Tidwell, R. R.; Donkor, I.; Naiman, N. A.;
Ohemeng, K. A.; Lombardy, R. J.; Bentley, J. A.; Cory, M. J. Med.
Chem. 1993, 36, 1746. (e) Treu, M.; Karner, T.; Kousek, R.; Berger,
H.; Mayer, M.; McConnell, D. B.; Stadler, A. J. Comb. Chem. 2008, 10,
863.
(8) Wolfe, J. P.; Buchwald, S. L. J. Org. Chem. 2000, 65, 1144−1157.
(9) While 3-Br-phenyl-1,3-dioxolane is commercially available,
several attempts were carried out in order to perform a laboratory
synthesis of 3-Br-benzoic-tert-butyl ester starting from 3-Br-benzoic
acid. Employing SOCl2 and t-BuOH no conversion was observed. We
turned our attention to tert-butylacetoacetate as the in situ source of
isobutylene10a since it could offer several advantages.10b Moderate to
good yields (55−65%) were achieved using this reagent, but the
recovered product still contained the starting tert-butylacetoacetate as
main impurity (25 wt % by NMR analysis). Lower conversion was
observed on heating the reaction to 50 °C instead of 25 °C.
(10) (a) Taber, D. F.; Gestenhaber, D. A.; Zhao, X. Tetrahedron Lett.
2006, 47, 3065. (b) This procedure does not call for the use of
expensive reagents, the handling of gaseous isobutylene or harsh
conditions. Further, the only byproducts would be acetone and CO2,
so it would not be necessary to remove water in order to drive the
reaction to acceptable conversions.
filtration on a Buchner funnel and washing with MeCN (2 ×
̈
0.25 L), the recovered solid was dried in oven until constant
weight (Toven = 50 °C; p = 10 atm; 18 h) in order to obtain 460
g of 1 as hydrobromide salt. Yield: 71%.
UPLC−MS: tR = 1.24 min; m/z = 418 [M + 1]+.
HRMS calcd for C25H33N5O [M + 1]+ 418.26069, found
418.26075.
HPLC: tR = 5.99 min; purity 99.1%.
1H NMR (400 MHz DMSO-d6): δ 9.80 (broad, 1H), 7.89
(m, 1H), 7.77 (m, 1H), 7.55−7.45 (m, 3H), 7.38 (s, 1H), 7.24
(m, 2H), 4.48−4.15 (m, 2H), 3.96 (s, 3H), 3.86 (m, 2H),
3.55−3.15 (m, 3H), 3.10−2.82 (m, 6H), 2.81 (s, 3H), 1.76−
1.57 (m, 4H).
13C NMR (100 MHz DMSO-d6): δ 173.5, 152.3, 151.5,
135.1, 135.0, 130.5, 126.2, 125.6, 125.3, 119.9, 119.1, 117.1,
116.5, 113.0, 53.2, 48.2, 42.7, 38.8, 37.4, 33.1, 28.2.
Water content (KF): 3.5 wt %.
Pd content (ICP-MS): 128 ppm.
Bromine content (ionic exchange LC): 20 wt % (1.2 equiv).
ASSOCIATED CONTENT
■
(11) At the time of this process research the price was $5.71/g
(Acros). The scalability of this promising approach was not
investigated further.
S
* Supporting Information
1
Copies of H NMR and 13C NMR spectra. Isolation and
(12) Novak, A.; Humphrey, L. D.; Walker, M. D.; Woodward, S.
identification of impurity e. This material is available free of
charge via the Internet at http://pubs.acs.org.
Tetrahedron Lett. 2006, 47, 5767−5769.
(13) The publication cited does not make clear the role of NaHSO3.
The authors wrote that addition of 30% mol of NaHSO3 ensures the
best yield of the product.6b
AUTHOR INFORMATION
■
(14) Taylor, H. M.; Hauser, C. R. Organic Syntheses; Wiley and Sons,
New York, 1963; Collect. Vol. 5, p 437.
(15) The partial evaporation of methyl iodide at 45 °C may be a
reason for the slowness of reaction.
(16) Wright, J. B. Chem. Rev. 1951, 48, 397.
(17) We needed this excess because of the partial evaporation under
reduced pressure of methyl iodide during the charging operations to
the jacketed reactor glass line.
(18) Meyers, C.; Maes, B. U. W.; Loones, K. T. J.; Bal, G.; Lemiere,
G. L. F.; Dommisse, R. A. J. Org. Chem. 2004, 69, 6010. Betti, M.;
Castagnoli, G.; Panico, A.; Sanna Coccone, S.; Wiedenau, P. Org.
Process Res. Dev. 2012, 16 (11), 1739.
(19) Lower loadings of catalyst/phosphine were not screened in
order to avoid much longer reaction time and relevant formation of
the debrominated side product.
Corresponding Author
*E-mail: mbetti@sienabiotech.it
Present Addresses
⊥Aptuit (Verona), 37135 Verona, Italy.
§Novartis Vaccines, 53100 Siena, Italy.
¶Autifony Srl, 37135 Verona, Italy.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We acknowledge Claudia Lombardi for her analytical support.
We especially recognize Sandro Pertini for managing multiple
aspects of the Process Chemistry workflow. We also gratefully
acknowledge Chiara Caramelli, Giacomo Minetto and Gal.la
Pericot Mohr for the early Medicinal Chemistry work in Siena
Biotech.
(20) Sabot, C.; Kumar, K. A.; Meunier, S.; Mioskowski, C.
Tetrahedron Lett. 2007, 48, 3863.
J
dx.doi.org/10.1021/op4002092 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX