The development of practical synthetic routes to a CB2 agonist: Efficient construction of a densely substituted purine core

Article abstract of DOI:10.1021/op300278c

An efficient and scalable process for the preparation of a purine-based CB₂ agonist was developed. The production route to the requisite purine core relies on N-acylation and sequential substitution of a 5-amino-4,6-dichloropyrimidine with two amine building blocks followed by a cyclocondensation reaction. The chemistry was successfully employed to rapidly prepare over 5 kg of the active pharmaceutical ingredient. To further improve efficiencies, postproduction development resulted in a rearranged synthesis which reduced the need for pressure reactors and introduced the most costly reagent in the final step.

Full text of DOI:10.1021/op300278c

Article
pubs.acs.org/OPRD
The Development of Practical Synthetic Routes to a CB2 Agonist:
Efficient Construction of a Densely Substituted Purine Core
Amy C. DeBaillie, Chauncey D. Jones, Michael E. Laurila, Nicholas A. Magnus, and Michael A. Staszak*
Chemical Product Research and Development Division, Eli Lilly and Company, Indianapolis, Indiana 46285, United States
ABSTRACT: An efficient and scalable process for the preparation of a purine-based CB2 agonist was developed. The production
route to the requisite purine core relies on N-acylation and sequential substitution of a 5-amino-4,6-dichloropyrimidine with two
amine building blocks followed by a cyclocondensation reaction. The chemistry was successfully employed to rapidly prepare
over 5 kg of the active pharmaceutical ingredient. To further improve efficiencies, postproduction development resulted in a
rearranged synthesis which reduced the need for pressure reactors and introduced the most costly reagent in the final step.
Scheme 1. Initial route to CB2 agonist 1
1. INTRODUCTION
The search for orally active compounds for treating pain has
been a longstanding goal of the pharmaceutical industry.1 In
particular, osteoarthritic pain presents a large and challenging
field, with much effort expended to discover pharmacological
agents which exhibit limited side effects during potentially long-
term treatments.
The cannabinoid receptors CB1 and CB2 have been studied
extensively;2 in broad terms, the CB1 site is expressed mainly in
the brain,3 whereas the CB2 receptor is expressed in the
immune system and hematopoietic cells.4 A 2004 Pfizer patent
described the use of purine compounds as CB1 receptor
antagonists and their use as anti-obesity agents.5 The
therapeutic potential of the CB2 receptor has been reviewed,
and that receptor site is considered a prime target for treating
pain.6 More recent Lilly patents7 describe the use of certain
purine compounds as selective CB2 agonists, which could
provide a means of treating pain while limiting side effects.
Herein we describe the process development leading to a
scalable synthesis of LY2828360 (1), a CB2 agonist which was
being developed by Eli Lilly and Company as an oral treatment
for osteoarthritis pain.8
philes) for screening and subsequently to synthesize gram
amounts of select compounds for early-stage testing. Finally,
the chemistry outlined was used to produce several hundred
grams of the lead molecule 1 (as the phosphate salt 1a) to fund
toxicological evaluations.
The discovery route began with a SNAr reaction of the
appropriate amine nucleophile and methylpyrimidine 2. This
first step was typically high yielding (∼90%) but required
extended reaction times (>5 days) and multiple charges of the
nucleophile.11 The second step included imine formation, an
oxidative cyclization, and a second SNAr reaction, all in one
reaction vessel. The process required upwards of 7 days at
elevated temperature to afford the desired purine. Nitrobenzene
was used as the oxidant in the system, with anisole as the high-
boiling solvent (bp 154 °C). Although the crude weight yields
were acceptable (∼75%), the product potency at the free base
stage was unacceptably low. Salt formation was used both to
2. DISCUSSION AND RESULTS
2.1. Discovery Chemistry Synthesis. The route shown
below (Scheme 1) was developed by discovery chemists at
Lilly’s Erl Wood Laboratories.9 This route is illustrated using
the synthesis of 1, although it was employed to synthesize a
number of SAR10 candidates (using various amine nucleo-
Received: October 7, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 2. Generic route to purines14
render a crystalline product and as an attempted aid in
purification. Overall, the assay-corrected yields for this one-pot
reaction were <50%.
h with isopropylamine and 1.4 equiv of isopropylethyl amine
(DIPEA or, Huenig’s base) at 100 °C in sealed pressure vessels.
Yields for this step were typically >90%.16
The discovery route was certainly functional for delivering
material to support SAR and toxicology studies. However, the
long reaction times and ineffective purification strategies made
its use for a multikilogram campaign undesirable. An additional
issue, and one that could be a long-term problem, arose from
the availability of the key 4-aminotetrahydropyran (4-amino-
THP) reagent for the synthesis of 1. This noncomplex, low-
molecular weight amine was both surprisingly unavailable and
expensive (∼$10,000/kg). Domestic suppliers were few, and
those only had limited amounts of the compound available as
the HCl salt.12
2.2. Process Development Synthesis. Considering the
aforementioned issues, we began both to seek alternative
synthetic strategies and to investigate sources of the 4-amino-
THP. For the synthetic route, a literature search led to a
number of possibilities,13 but the most compelling one was
found within a Pfizer process patent describing CB1
antagonists.14 The generic route from that reference is shown
in Scheme 2.
Several features of Pfizer’s purine chemistry interested us.
First, employing an amide similar to structure IV, a crystalline
intermediate, would afford a potential purification point prior
to final product formation, and this was a feature missing from
the discovery route. Second, under the Pfizer conditions, the
cyclization to the purine VI was mediated by strong acid, and
this could lead either to direct formation of a desired API15 salt
from the reaction mixture and/or afford an additional
purification point.
Once supplies of 4-amino-THP allowed, the knowledge
gained from the use of isopropylamine was applied to the
preparation of 3 with only minor modifications. Laboratory-
scale trials showed this first SNAr reaction required approx-
imately 36−40 h at 100 °C (again in sealed vessels) with 1.6
equiv of the amine to reach completion. The reaction could be
performed in as few as 5 volumes17 (vol) of IPA with 90−94%
isolated yields after an ethyl acetate extractive workup.
However, the workup was cumbersome due to the need for
multiple large-volume extractions. Thus, a simple precipitation
method was developed using water, which afforded the product
in 80−84% recovery on 50-g scale. The lower recovery was
viewed as a fair trade-off for the simplicity achieved for large-
scale synthesis.
For the amide formation step (III to IV in Scheme 2), the
Pfizer patent procedure employed a slight excess of an acid
chloride at 0−5 °C with dimethylacetamide (DMAc) as
solvent. Adding water to the ending reaction mixture had
afforded their amides in good yields (80−82%) and high purity.
Although we initially tested a number of logical replacements
for DMAc,18 that solvent was found to be superior in terms of
both yield and product purity. For our derivatives (see middle
step of Scheme 4), we found the reactions worked best in 7 vol
of DMAc with 1.2 equiv of 2-chlorobenzoyl chloride 4.
Reversing the quench order resulted in a more easily filterable,
crystalline product with an increase of purity. Yields of 5 were
typically 80−85%. Significantly, to avoid overacylation, no
amine base was used to scavenge the HCl byproduct.
Due to the limited supplies of 4-amino-THP, our develop-
ment research began with isopropylamine as a surrogate
nucleophile, since the material was both inexpensive and readily
available. We felt that information garnered from such a model
study would likely be applicable to the synthesis of 3 once
supplies of 4-amino-THP were available. We successfully
demonstrated that this reaction could be accomplished in 24
The Pfizer patent procedures (Et3N, IPA, 80 °C) afforded
only very low conversions to the desired material from our
compound (5) reacting with 1-methylpiperazine. Switching to
N-methylpyrrolidinone (NMP) as solvent, using DIPEA as
base, and heating the reaction mixture to 120 °C in a sealed
vessel, we were able to isolate the desired (penultimate)
intermediate 6 in 54% yield (Scheme 3). Thus, with this
B
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
portion of the synthesis demonstrated, we were ready to test
the final cyclization to the purine.
While investigating the SNAr reaction of 5 with 1-
methylpiperazine, we detected a small peak (in the HPLC
data) corresponding to the desired final product 1. To this
point, we had believed that strong acid was required to cyclize
to the purine, a belief supported by the literature. Yet, the data
showed that small amounts of cyclized material were forming at
100 °C under basic reaction conditions. Our next goal was to
determine if we could use this result to our advantage by
forcing cyclization to occur under thermal stress, thus providing
an opportunity for a one-pot process to effect the final two
steps of the synthesis.
Scheme 3. Initial two-step sequence to 1b
Using pressure-rated equipment, we combined amide 5, 1-
methyl piperazine, and DIPEA in IPA and heated the mixture
to 160 °C in a sealed glass reactor. After 24 h, HPLC analysis
showed that all of the starting amide had been consumed with
only 3% of the uncyclized intermediate 6 remaining. The major
product (97 area %) was 1. After simply adding water to the
reaction mixture, the free base 1 crystallized and was isolated in
89% yield and >99% HPLC purity.
Although our original goal was to reliably produce 1 which
could then be converted to a pharmaceutically acceptable salt
(i.e., 1a or 1b), the crystalline free base had never been available
for testing. After performance of comparative microscopy,
XRPD, DSC/TGA, and hygroscopicity studies on the free base
and the phosphate salt, the data indicated that the free base
exhibited preferable physicochemical characteristics. Addition-
ally, a polymorph screen for the crystalline free base indicated
that the polymorph obtained via our new process appeared to
be the most stable form.19 To date, no other crystal forms,
hydrates nor solvates have been identified for this material. On
the basis of these data, the free base 1 became our target for
scale-up.
The Pfizer patent conditions for the cyclization of V to VI
employed 3 equiv of conc H2SO4 in refluxing IPA. For our
THP derivative, we found it necessary to use 9 equiv of H2SO4
to achieve conversion. Yields of 1b (the sulfate of 1) were, at
best, 68% when the reaction was performed in ethanol at 100
°C in a sealed vessel (Scheme 3). Again, this chemistry was
demonstrated to be feasible with our derivative (6), and we
were poised to further develop the synthetic route.
The finalized synthetic route to 1 is shown in Scheme 4. This
chemistry was laboratory tested using 50 g of the starting
pyrimidine 2. The two steps requiring high temperature/
pressure (the conversions of 2 to 3 and 5 to 1) were performed
Scheme 4. Finalized laboratory route to 1
C
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 5. Proposed formation of impurity 7
in Hastelloy C Parr reactors. The overall yield to crystalline 1
from 2 was 63%, with 99% purity by HPLC.
isolation. Ultimately, LC/MS data confirmed that the impurity
was the dimethylamino derivative 7 (Scheme 5).
Considering these results, especially in the context of our
development work on this step, where DMAc was found to be
an excellent solvent for this reaction, we proposed the following
cascade of events. First, DMAc is likely not just a solvent for
this reaction but is also a participant in a Vilsmeier−
Haack-22type activation. Outlined in Scheme 5, the acid
chloride could react with DMAc to form an iminium ion,
which then reacts with the poorly basic and poorly nucleophilic
amine in our system. During the water quench/crystallization
portion of the process, the byproducts (including any iminium
chloride from excess acid chloride reagent) are ultimately
hydrolyzed to dimethylamine hydrochloride and acetic acid.23
The higher concentrations, more occlusive crystallization,
and less efficient water washes of the large-scale run likely all
contributed to significant amounts of dimethylamine hydro-
chloride remaining in/on the amide product. Still, this impurity
might not have been problematic (due to its low boiling point)
in downstream chemistry if not for the fact that the second
SNAr and cyclization steps are performed in sealed vessels. In
that process, the DIPEA employed would neutralize the
dimethylamine hydrochloride to its free base form which
then acts as just another nucleophile for the SNAr reaction,
hence, the formation of the dimethylamino analogue (7) of the
API when the production lot of amide 5 was laboratory tested.
To purify the crude bulk amide, the material was refluxed in
IPA and then cooled, filtered, and rinsed with IPA. The material
was recovered in an overall 81% yield and >99% HPLC purity
for this step. Significantly, the 1H NMR of the material showed
2.3. Production Results. A production plan to prepare 5
kg of 1 was devised in conjunction with our outsourcing
partner.20 The following paragraphs summarize the successes
and additional learning points achieved while performing the
first scale-up of this chemistry.
For the first step, SNAr, several minor process modifications
were initiated, which led to increased yields over those of
laboratory experiments. The reaction was performed using 1.4
equiv of the 4-amino-THP and 2 equiv of DIPEA in only 3.3
vol of IPA at 100 °C. The desired product 3 was isolated by
simple concentration followed by precipitation with water.
Although the reaction did take slightly longer on scale,
complete conversion of 2 was reached in 2.5 days, and the
yields improved. Two combined 20-L autoclave batches
afforded 5.9 kg of 3 in 96% yield and >99% purity by HPLC.
To prepare for the large-scale amide formation, the reaction
was demonstrated in the laboratory to work well in 3.5 vol of
DMAc (versus the original 7 vol). This reduced volume would
allow all the required amide to be formed in one production
run, saving processing time. Unfortunately, this volume change
for production purposes led to a slower filtration (on scale) and
formation of a new impurity in isolated 5, which exhibited a
1
sharp singlet (2.7 ppm) in the H NMR (CDCl3) spectrum of
the amide product.21 When a small portion of the amide 5 was
forward processed through the final steps, a new impurity was
also observed in 1. This new impurity was rather similar in
behavior to 1 and was difficult to reject during workup and
D
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 6. Rearranged synthesis to amide 5
no dimethylamine hydrochloride signal, and a laboratory test
using purified 5 afforded 1 free of dimethylamino impurity 7.
The final SNAr and cyclization steps were completed in an
autoclave in two successful production runs, each requiring
approximately 48 h at 160 °C to reach completion. The yield
for these last steps was 92%, and a total of 5.16 kg of 1 was
produced. The purity (HPLC) of the final material was >97%,
which was within our specifications. Thus, the overall
production-scale yield for this first campaign was 72% from
the starting pyrimidine 2.
2.4. Post Campaign Development Efforts. The first
campaign to synthesize 1 was deemed successful on the basis of
the goals of the project. Anticipating a larger second campaign,
we critically reviewed the new route for additional development
opportunities, and several were identified.
(0.5 equiv) prior to telescoping this step. HPLC analysis
indicated an 89% solution yield of the amide, and the small
amount of residual water did not have a deleterious effect on
the desired substitution reaction. As in the two-step process,
the conversion was complete in less than 4 h at 80 °C in NMP.
The yield of 5 for the two-step process was 82% from the
pyrimidine 2, and product purity was 99% (HPLC). This
material behaved identically to that from the earlier route when
forward processed (using the pressure protocol) to 1.
Towards the goal of conserving 4-amino-THP through its
later use in the synthesis, the isolated amide 8 was reacted with
1-methylpiperazine in NMP (Scheme 7). This reaction did not
Scheme 7. Rearranged synthesis from amide 8 to 1
First, using DMAc as solvent has both health and
environmental issues. Furthermore, as we discovered during
production, the solvent also has the potential to lead to
impurity formation. Second, although the new processes were
shorter than the original reactions, two steps still required >48
h at elevated temperatures to reach completion. Third, the
most expensive reagent (4-amino-THP) was used early in the
synthesis, and we wondered if rearranging the order of the steps
would be possible, thus potentially leading to lower cost of
goods for the final product. Finally, with high temperature and
pressure reactions prominent in the synthesis, we wanted to
investigate how to make the entire route more amenable to
standard plant reactors.
While developing the amide formation step, a number of
alternatives to DMAc had been screened, including CH3CN,
MTBE, DMSO, CH2Cl2, THF, and N-methylpyrrolidinone
(NMP). Of these solvents, NMP was the only one to allow
product formation of reasonable quality and yield. We
wondered if NMP would work better when the amidation
was performed at an earlier stage.24 Significantly, we showed
that the methylpyrimidine 2 (Scheme 6) could react with a
slight excess of the acid chloride 4 in NMP at 70 °C. The
reaction was complete in 5 h, and 8 could be isolated as a solid
in 91% yield.
Subsequently, we demonstrated that compound 8 could be
converted to intermediate 5 by reaction with 4-amino-THP in
NMP with DIPEA. This SNAr reaction was significantly more
facile with 8 than when forming our original intermediate 3
(Scheme 4). The conversion was affected at 80 °C in ∼3 h
without the need for pressure equipment. After precipitation
using water, intermediate 5 was recovered in 90% yield from 8.
To further streamline the synthesis, we showed that amide 8
could be carried directly into the first SNAr reaction (using 4-
amino-THP). The slight excess of acid chloride present at the
end of the amide formation needed to be quenched using water
require elevated temperature nor pressure, as intermediate 9
was formed in 2.5 h at room temperature and was isolated in
92% yield. Subsequently, compound 9 was treated with 4-
amino-THP (1.75 equiv) in IPA (5 vol) using 3 equiv of
DIPEA. After 26 h at 160 °C in a pressure vessel, the desired
product 1 was obtained in 87% yield, providing proof of
concept for the reordered synthesis from pyrimidine 2 (Scheme
7).
3. CONCLUSIONS
In summary, we have presented an efficient, safe, and scalable
synthesis of 1. The synthesis was successfully demonstrated on
∼5 kg scale, producing 1 of 97% purity and in overall yield of
72% from pyrimidine 2. Additionally, the route has been
developed postproduction to afford further yield and processing
gains. These gains were realized on laboratory scale but due to
loss of project funding were not tested on pilot-plant scale.
E
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
4. EXPERIMENTAL SECTION
Article
at reflux for 48 h. After cooling to 25 °C and filtration, the
product cake was washed with IPA (12.5 L) and dried under
vacuum at 80 °C. Compound 5 was isolated as a white solid
(7520 g, 81.6%, 99.6% purity by HPLC). 1H NMR (DMSO-d6,
400 MHz): δ 1.50 (m, 2H), 1.78 (dd, 2H), 2.36 (s, 3H), 3.38
(t, 2H), 3.83 (dd, 2H), 4.14 (m, 1H), 6.45 (d, 1H), 7.50 (m,
3H), 7.75 (d, 1H), 9.74 (bs, 1H). LC/MS data showed an M +
1 of 382.
HPLC analyses were performed using a reverse-phase
technique as follows. Instrument: Agilent 1200 series. Column:
YMC Pack Pro C18 100 mm × 4.6 mm, 3 μm. Flow: 1.0 mL/
min. Solvent A = 0.05% TFA in H2O; Solvent B = 0.05% in
CH3CN. Gradient: 90% A to 40% A over 10 min; change to 5%
A over 1 min and hold at 5% A for 2 min. Change back to 90%
A over 2.5 min and hold for 2.5 min. Column temperature: 40
°C. Wavelength: 230 nm. LC/MS analyses were performed as
follows. Instrument: Agilent 1200 MSD. Column: SeQuant
ZIC-HILIC 150 mm × 4.6 mm, 5 um. Mobile phase consisting
of solvent A: 2 mM aqueous NH4Cl; Solvent B: 2 mM NH4Cl
in 95/5 CH3CN/H2O; Gradient: 90% B for 5 min, then change
to 50% B over 5 min and hold for 6 min. Change to 90% B over
2 min and hold for 2 min. Flow rate: 1.0 mL/min; Column
temperature 40 °C. Wavelength: 230 nm; Ionization mode: ESI
Positive.
4.1. Production Scale Procedures (cf. Scheme 4).
4.1.1. 6-Chloro-2-methyl-N4-(tetrahydro-2H-pyran-4-yl)-
pyrimidine-4,5-diamine (3). A 20-L autoclave was charged
with 2 (2243 g, 12.6 mol), 4-amino-THP (1782 g, 17.6 mol, 1.4
equiv), DIPEA (3263 g, 25.2 mol, 2 equiv), and IPA (3.7 L).
The reaction mixture was heated to 100 °C. After 48 h, HPLC
analysis of a reaction aliquot indicated 100% conversion of 2.
The reaction mixture was cooled to 20 °C and transferred to a
100-L vessel; the autoclave was rinsed out with IPA (5 L).
Concurrently, a second run (2240 g of starting material 2) was
performed in a 20-L autoclave using the same procedure. The
combined runs were concentrated in vacuo to remove ∼20 L of
the solvent. The internal temperature was adjusted to 75 °C,
and H2O (33 L) was added over 45 min. A suspension formed,
and concentration in vacuo was continued to collect another 10
L of solvent. The resulting suspension was filtered and the cake
washed with H2O (23 L) and dried on the filter using vacuum
and a nitrogen flow for 22 h. The cake was further dried under
vacuum at 80 °C to a constant weight. Compound 3 was
recovered as an off-white solid (5870 g, 95.9%, 99.3% purity by
HPLC). 1H NMR (DMSO-d6, 400 MHz): δ 1.42 (m, 2H), 1.83
(dd, 2H), 2.19 (s, 3H), 3.96 (t, 2H), 3.85 (dd, 2H), 4.08 (m,
1H), 4.78 (bs, 2H), 6.45 (bs, 1H). LC/MS data showed an M +
1 of 243.
4.1.2. 2-Chloro-N-(4-chloro-2-methyl-6-((tetrahydro-2H-
pyran-4-yl)amino)pyrimidin-5-yl)benzamide (5). To DMAc
(17.6 L) was added 3 (5854 g, 24.1 mol), and the mixture was
stirred at 25 °C to attain a solution (∼10 min) and then cooled
to 0 °C. 2-Chlorobenzoyl chloride 4 (5180 g, 29.6 mol, 1.23
equiv) was added over 30 min, and the transfer line was rinsed
with DMAc (2.9 L). After 1 h at 0−5 °C, the mixture was
warmed to 20 °C over 5 h, stirred for 8 h, and checked by
HPLC, which indicated >99% conversion of starting material.
The reaction mixture was transferred to a separate addition
tank, and the transfer line was rinsed with DMAc (1.3 L). The
original reaction vessel was charged with water (60 L), which
was cooled to 15 °C. The reaction mixture was added to the
water over 1.5 h. The resulting suspension was warmed to 25
°C and stirred for 20 h. The suspension was filtered, washed
with water (28 L), and dried on the funnel under a nitrogen
purge for 48 h.
4.1.3. 8-(2-Chlorophenyl)-2-methyl-6-(4-methylpiperazin-
1-yl)-9-(tetrahydro-2H-pyran-4-yl)-9H-purine, LY2828360
(1). A 20-L autoclave was charged with 5 (2021 g, 5.3 mol),
1-methylpiperazine (897 g, 8.95 mol, 1.75 equiv), DIPEA (758
g, 5.9 mol, 1.1 equiv), and IPA (10 L). The reaction mixture
was heated to 160 °C. After 58 h, HPLC analysis of a reaction
aliquot indicated >99% conversion of starting material 5. The
reaction mixture was cooled to 20 °C and transferred to a 200-
L vessel, and the autoclave was rinsed out with IPA (10 L).
Concurrently, a second run (2033 g of the amide starting
material) was performed in a 20-L autoclave in like fashion. The
combined runs were diluted with IPA (62 L) and heated to 80
°C to obtain a clear solution. After 1 h the mixture was cooled
to 70 °C and filtered through an in-line filter.25 The tank and
transfer line were rinsed with warm (∼45 °C) IPA (5 L), and
the clear solution was concentrated by distillation (∼76 L of
distillate was collected). The mixture was cooled to 20 °C over
4 h. Water (70 L) was added to the suspension over 1 h, and
stirring was continued for 2 h. The contents of the tank were
filtered, and the cake was washed with water (2 × 20 L). The
cake was dried overnight on the funnel under a nitrogen flow
and then further dried under vacuum at 80 °C. 1 was obtained
as a crystalline white solid (4.13 kg, 92.2%, 97.4% purity by
HPLC). Mp (DSC) (10 °C/min) onset 185.6 °C; peak 187.8
°C. 1H NMR (DMSO-d6, 400 MHz, VT = 75 °C): δ 1.70 (bd,
2H), 2.21 (s, 3H), 2.41 (bs, 4H), 2.47 (bs, 3H), 2.7 (m, 2H),
3.24 (t, 2H), 3.90 (d, 2H), 4.02 (m, 1H), 4.17 (bs, 4H), 7.51
(m, 1H), 7.63 (m, 3H). 13C NMR (100 MHz, DMSO-d6): δ
26.6, 30.8, 40.6, 46.2, 54.2, 55.1, 66.8, 118.0, 128.0, 130.0,
130.3, 132.5, 133.1, 134.0, 145.8, 152.5, 153.3, 160.5.
4.2. Laboratory-Scale Procedures for the Re-ordered
Synthesis (Post API-1). 4.2.1. 2-Chloro-N-(4,6-dichloro-2-
methylpyrimidin-5-yl)benzamide (8) (cf. Scheme 6). Under a
nitrogen atmosphere, 2 (10.0 g, 56.2 mmol) was dissolved in
NMP (50 mL). 2-Chlorobenzoyl chloride 4 (10.8 g, 61.8
mmol, 1.1 equiv) was added and the mixture heated to 70−75
°C. After 5 h, the mixture was cooled to ambient temperature
and water (100 mL) was added over 10 min. The resulting
slurry was stirred for 1 h, filtered, and the solids were washed
with water (2 × 100 mL). After vacuum drying at 50 °C,
1
compound 8 was recovered as white solid (16.1 g, 90.7%). H
NMR (DMSO-d6, 400 MHz): δ 2.63 (s, 3H), 7.45−7.61 (m,
4H), 10.80 (bs, 1H). 13C NMR (100 MHz, DMSO-d6): δ 25.3,
125.9, 127.7, 129.4, 130.4, 130.6, 132.2, 135.6, 160.1, 165.6,
167.1. HRMS (ESI+): calcd for C12H9Cl3N3O (M+): 315.9806,
found 315.9803.
4.2.2. 2-Chloro-N-(4-chloro-2-methyl-6-((tetrahydro-2H-
pyran-4-yl)amino)pyrimidin-5-yl)benzamide 5 (using a step-
wise process; cf. Scheme 6). 8 (13.5 g, 42.6 mmol, prepared as
described in the previous procedure) was combined with 4-
amino-THP (6.0 g, 59.7 mmol, 1.4 equiv), DIPEA (14.9 mL,
85.4 mmol, 2.0 equiv), and NMP (72 mL). The mixture was
heated to 80 °C, stirred for 3.25 h, and cooled to ambient
temperature. Water (140 mL) was added over 15 min, and the
resulting slurry was stirred for 1 h. The mixture was filtered, and
The overweight product cake was divided into three portions
and further dried under vacuum at 80 °C; the resulting solids
were then manually crushed into smaller-sized particles. The
combined materials were suspended in IPA (78 L) and heated
F
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
the solids were washed with water (3 × 140 mL). After vacuum
drying at 50 °C, compound 5 was recovered as a white solid
(14.6 g, 89.9%). 1H NMR (DMSO-d6, 400 MHz): δ 1.47−1.56
(m, 2H), 1.80 (d, 2H), 2.36 (s, 3H), 3.38 (t, 2H), 3.83 (d, 2H),
4.09−4.19 (m, 1H), 6.45 (d, 1H), 7.43−7.56 (m, 3H), 7.74
(dd, 1H), 9.74 (s, 1H). 13C NMR (100 MHz, DMSO-d6): δ
25.8, 32.7, 47.2, 66.3, 110.1, 127.5, 130.0, 130.3, 130.8, 131.9,
136.0, 155.8, 159.5, 165.7, 165.8. HRMS (ESI+): calcd for
C17H19Cl2N4O2 (M+): 381.0880, found 381.0888.
1.70 (bd, 2H), 2.21 (s, 3H), 2.41 (bs, 4H), 2.47 (bs, 3H), 2.7
(m, 2H), 3.24 (t, 2H), 3.90 (d, 2H), 4.02 (m, 1H), 4.17 (bs,
4H), 7.51 (m, 1H), 7.63 (m, 3H). 13C NMR (100 MHz,
DMSO-d6): δ 26.6, 30.8, 40.6, 46.2, 54.2, 55.1, 66.8, 118.0,
128.0, 130.0, 130.3, 132.5, 133.1, 134.0, 145.8, 152.5, 153.3,
160.5.
AUTHOR INFORMATION
Corresponding Author
*E-mail: staszak_michael_a@lilly.com
■
4.2.3. 2-Chloro-N-(4-chloro-2-methyl-6-((tetrahydro-2H-
pyran-4-yl)amino)pyrimidin-5-yl)benzamide (5) (using a
telescoped process; cf. Scheme 6). Under a nitrogen
atmosphere, 2 (9.9 g, 55.5 mmol) was dissolved in NMP (45
mL). 4 (10.3 g, 57.2 mmol, 1.03 equiv) was added and the
mixture heated to 80 °C. After 4.5 h, water (0.5 mL, 27.7 mmol,
0.5 equiv) was added, and stirring was continued at 80 °C for
0.5 h. DIPEA (29.0 mL, 166.3 mmol, 3.0 equiv) was added
followed by 4-amino-THP (7.9 g, 78.6 mmol, 1.4 equiv) and
NMP (10 mL rinse). After 3.25 h at 80 °C, the mixture was
cooled to ambient temperature, and water (110 mL) was added
over 15 min. The resulting slurry was stirred for 0.5 h and
filtered, and the solids were washed with water (3 × 100 mL).
After vacuum drying at 50 °C, compound 5 was recovered as an
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Special thanks to the process chemistry group at CarboGen
AG, Aarau, Switzerland, and in particular to Dr. Marc Lanz and
Dr. Jurgen Beyer for their development efforts and scale
demonstrations of the chemistry shown in Scheme 3.
REFERENCES
■
(1) Dronsfield, A.; Brown, T.; Ellis, P. Pain Relief: From Coal Tar to
Paracetamol. Education in Chemistry (Royal Society of Chemistry) 2005,
42 (4), 102−105.
1
off-white solid (17.4 g, 82.3%). H NMR (DMSO-d6, 400
(2) For example, see (a) Mackie, K. Cannabanoid Receptors: Where
They Are and What They Do. J. Neuroendrocrinol. 2008, 20
(Supplement 1), 10−14. and (b) Graham, E. S.; Ashton, J. C.
Cannabanoid Receptors: A Brief History and ‘What’s Hot’. Front.
Biosci. 2009, 14, 944−957.
(3) Matsuda, L. A.; Lolait, S. J.; Brownstein, M. J.; Young, A. C.;
Bonner, T. I. Nature 1990, 346, 561−564.
(4) Pacher, P.; Mechoulam, R. Prog. Lipid Res. 2011, 50, 193−211.
(5) Griffith, D. A. (Pfizer Corp.) Preparation of Purines as
Cannabinoid Receptor Ligands (CB1 Receptor Antagonists). WO/
2004/037823, 2004; Chem. Abstr. 2004, 140, 391293.
(6) Guindon, J.; Hohmann, A. G. Cannabinoid CB2 Receptors: A
Therapeutic Target for the Treatment of Inflammatory and Neuro-
pathic Pain. Br. J. Pharmacol. 2008, 153, 319−334.
(7) (a) Astles, P. C.; Guidetti, R.; Tidwell, M. W.; Hollinshead, S. P.
(Eli Lilly and Co.). Preparation of Piperazinyl Purine Compounds for
the Treatment of Pain. U.S. Patent Appl. 0160288, 2010; Chem. Abstr.
2010, 153, 115955. (b) Hollinshead, S. P. (Eli Lilly and Co.).
Preparation of Purine Compounds for Use as CB2 Agonists. WO/
2011/123482, 2011; Chem. Abstr. 2011, 155, 483848.
MHz): δ 1.47−1.56 (m, 2H), 1.80 (d, 2H), 2.36 (s, 3H), 3.38
(t, 2H), 3.83 (d, 2H), 4.09−4.19 (m, 1H), 6.45 (d, 1H), 7.43−
7.56 (m, 3H), 7.74 (dd, 1H), 9.74 (s, 1H). 13C NMR (100
MHz, DMSO-d6): δ 25.8, 32.7, 47.2, 66.3, 110.1, 127.5, 130.0,
130.3, 130.8, 131.9, 136.0, 155.8, 159.5, 165.7, 165.8. HRMS
(ESI+): calcd for C17H19Cl2N4O2 (M+): 381.0880, found
381.0888.
4.2.4. 2-Chloro-N-(4-chloro-2-methyl-6-(4-methylpipera-
zin-1-yl)pyrimidin-5-yl)benzamide (9) (Scheme 7). 8 (15.0 g,
47.4 mmol) was dissolved in NMP (75.0 mL) under a nitrogen
atmosphere. 1-Methylpiperazine (5.5 mL, 49.8 mmol, 1.05
equiv) was added over 5 min, and this material was rinsed into
the flask with NMP (1.5 mL). After 10 min, DIPEA (9.9 mL,
56.8 mmol, 1.2 equiv) was added followed by an NMP rinse
(1.5 mL), and the resulting mixture was stirred at ambient
temperature for 2.5 h. Water (150 mL) was added over 10 min
followed by ethyl acetate (75 mL), and the resulting slurry was
stirred for 1 h. After filtration, water wash (3 × 150 mL), and
ethyl acetate wash (30 mL), the isolated solids were vacuum-
dried at 50 °C. Compound 9 was recovered as a white solid
(8) Ibid. 7a.
(9) The discovery chemistry route was developed by Dr. Jeffrey
Richardson at Eli Lilly’s Erl Wood Laboratories in Surrey, England. We
are indebted to Dr. Richardson for the groundwork he laid for the
chemistry of these CB2 agonists, the insight he afforded us from a
synthetic standpoint, and the samples of intermediates and reagents he
supplied.
(10) SAR is shorthand for structure−activity relationship (the
relationship between the chemical structure of a molecule and its
biological activity).
1
(16.5 g, 91.5%). H NMR (DMSO-d6, 400 MHz): δ 2.15 (s,
3H), 2.35 (dd, 4H), 2.39 (s, 3H), 3.72 (bm, 4H), 7.44−7.59
(m, 4H), 10.15 (s, 1H). 13C NMR (100 MHz, DMSO-d6): δ
25.6, 46.0, 46.7, 55.0, 111.2, 127.7, 129.3, 130.5, 131.9, 136.1,
160.0, 161.2, 164.8, 165.8. HRMS (ESI+): calcd for
C17H20Cl2N5O (M+): 380.1039, found 380.1045.
4.2.5. 8-(2-Chlorophenyl)-2-methyl-6-(4-methylpiperazin-
1-yl)-9-(tetrahydro-2H-pyran-4-yl)-9H-purine, LY2828360
(Scheme 7). A Hastelloy C Parr reaction vessel was charged
with 9 (3.5 g, 9.2 mmol), 4-amino-THP (1.7 g, 16.6 mmol, 1.8
equiv), DIPEA (1.8 mL, 10.2 mmol, 1.1 equiv), and IPA (17.5
mL). The vessel was purged with nitrogen and sealed, and the
internal temperature was raised to 160 °C. After 26 h, the
contents were cooled to 45−50 °C, transferred to a separate
reaction flask, and treated with cold water (60 mL). The
resulting slurry was stirred for 1.75 h and filtered. The solids
were washed with cold water (17.5 mL) and dried under
vacuum at 50 °C. LY2828360 was obtained as a white solid (3.4
(11) In the case of the 4-aminotetrahydropyran reagent, the reagent
itself was unstable to extended periods of heating, and thus extra
equivalents of the material were necessary for this process.
(12) Domestically, there were two companies which had supplies of
this material. Combi-Blocks Corporation had two 1 g samples of the
HCl salt available at $134/g. Austin Chemicals had prepared the HCl
salt of the 4-amino-THP for the discovery chemistry efforts, but had
only 5 g of the material remaining.
(13) For a general treatise on purines and pyrimidines, see
Cumulative Index of Heterocyclic Systems; Brown, D. J., Evans, R. F.,
Cowden, W. B., Fenn, M. D., Eds.; Chemistry of Heterocyclic
Compounds: The Pyrimidines, 2008; Vol. 65, p 52 (DOI: 10.1002/
9780470187395). For more specific syntheses, see the following:
(a) Young, R. C.; Jones, M.; Milliner, K. J.; Rana, K. K.; Ward, J. G. J.
1
g, 86.6%). H NMR (DMSO-d6, 400 MHz, VT = 75 °C): δ
G
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Med. Chem. 1990, 33, 2073−2080. (b) Dang, Q.; Brown, B. S.; Erion,
M. D. Tetrahedron Lett. 2000, 41, 6559−6562. (c) Harada, H.; Asano,
O.; Kawata, T.; Inoue, T.; Horizoe, T.; Yasuda, N.; Nagata, K.;
Murakimi, M.; Nagaoka, J.; Kobayashi, S.; Tanaka, I.; Abe, S. Bioorg.
Med. Chem. 2001, 9, 2709−2726. (d) Tandel, S.; Bliznets, I.; Ebinger,
K.; Ma, Y.; Bhumralkar, D.; Thiruvazhi, M. Tetrahedron Lett. 2004, 45,
2321−2322.
(14) Ragan, J. A. (Pfizer Corp.). Intramolecular Cyclocondensation
Process for Preparing Purine Compounds. WO/2006/043175; Chem.
Abstr. 2006, 144: 412537.
(15) API = active pharmaceutical ingredient.
(16) Using isopropylamine as the surrogate nucleophile, the isolated
yields of the corresponding product 6-chloro-N4-isopropyl-2-methyl-
pyrimidine-4,5-diamine were typically 90−92% after concentration in
vacuo.
(17) A “volume” (vol) as used here and in all later occurrences, refers
to milliliters of the solvent per gram of the limiting reagent (in this
case, compound 4).
(18) For a useful guide to solvents, see the American Chemical
Society Green Chemistry Institute’s (ACS-GCI) Pharmaceutical
Roundtable Solvent Selection Guide, version 2.0; 2011 (available online
at www.acs.org/gcipharmaroundtable). For specific information on
DMAc, see Kennedy, G. L.; Sherman, H. Drug Chem. Toxicol. 1986, 9
(2), 147−l70.
(19) We are grateful to Dr. Susan Reutzel-Edens of Lilly’s
preformulations Solid State Group for performing the polymorph
screening.
(20) The outsourcing partner chosen for these efforts was
CARBOGEN AMCIS AG in Switzerland.
(21) The HPLC assay (at 230 nm) of isolated 5 showed no signal
other than that of the desired material. The impurity was evident in the
1H NMR data as only the sharp singlet mentioned at 2.7 ppm. We
were immediately suspicious that the impurity was dimethylamine
hydrochloride. When a sample of 5 was spiked with a bonafide sample
1
of dimethylamine hydrochloride, the H NMR data supported that
suspicion. When 5 was forward processed in the chemistry of the final
step, the identification of 7 confirmed that dimethylamine hydro-
chloride was indeed the impurity in the material and there is no other
possibility for it to arise other than from the DMAc used as solvent.
(22) For a general review of the Vilsmeier−Haack reaction, see
Meth-Cohn, O.; Stanforth, S. Comp. Org. Synth. 1991, 2, 777−794.
(23) This series of events effectively results in the acid-catalyzed
hydrolysis of the solvent DMAc, which is the only plausible source of
dimethylamine hydrochloride in the reaction sequence.
(24) In theory, NMP is susceptible to the same type of acid
hydrolysis mentioned earlier. However, this solvent is regarded as
quite resistant to such hydrolysis, and our analyses of products from
reactions using NMP showed no signs of hydrolysis-related impurities.
(25) For GMP processing, once the final, in-line filtration is
completed, all subsequent rinses with solvents and/or deionized water
were performed with filtered materials.
H
dx.doi.org/10.1021/op300278c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX