AMD070, a CXCR4 chemokine receptor antagonist: Practical large-scale laboratory synthesis

Article abstract of DOI:10.1021/op8000993

An efficient and convergent four-step synthetic route to the CXCR4 chemokine receptor antagonist AMD070 (1) has been developed which employs only a single chromatographic step in the entire sequence. Novel reductive amination methods have been developed for the coupling of 2 and 3 in which a dehydrative imine formation is followed by reduction with an attenuated borohydride reagent (zinc chloride and sodium borohydride). Selective extraction methods were employed to purify synthetic intermediates and remove reagents and impurities. A procedure has also been developed to isolate 1 in a pure crystalline form.

Full text of DOI:10.1021/op8000993

Organic Process Research & Development 2008, 12, 823–830
Full Papers
AMD070, a CXCR4 Chemokine Receptor Antagonist: Practical Large-Scale
Laboratory Synthesis
,†
Jason B. Crawford,* Gang Chen, David Gauthier, Trevor Wilson, Bryon Carpenter, Ian R. Baird, Ernie McEachern,
‡
Alan Kaller, Curtis Harwig, Bem Atsma, Renato T. Skerlj, and Gary J. Bridger
AnorMED Inc. (a wholly owned subsidiary of Genzyme Corp.), 200-20353 64th AVenue, Langley, British Columbia, Canada
4
Abstract:
AMD070 (1) is a small-molecule CXCR4 antagonist. Phase
5
1
a data showed that AMD070 is generally safe, well tolerated,
An efficient and convergent four-step synthetic route to the
CXCR4 chemokine receptor antagonist AMD070 (1) has been
developed which employs only a single chromatographic step in
the entire sequence. Novel reductive amination methods have been
developed for the coupling of 2 and 3 in which a dehydrative imine
formation is followed by reduction with an attenuated borohydride
reagent (zinc chloride and sodium borohydride). Selective extrac-
tion methods were employed to purify synthetic intermediates and
remove reagents and impurities. A procedure has also been
developed to isolate 1 in a pure crystalline form.
and orally bioavailable. Preliminary clinical data also showed
that AMD070 is active in HIV patients: 4/8 patients at a dose
of 200 mg BID had significant reductions in CXCR4 viral load
with an average reduction of 1.3 log. These results are the first
demonstration of clinical efficacy against HIV for an orally
bioavailable CXCR4 antagonist.
A retrosynthetic analysis of 1 provides a convergent strategic
route based on two disconnections and three starting materials
(Scheme 1). The synthesis of 8-amino-5,6,7,8-tetrahydroquino-
line, and the isolation of the enantiopure (S)-isomer 2 from
the corresponding racemate Via enzymatic resolution have been
6
7,8
9
described previously. Installation of the two remaining frag-
Introduction
The CXCR4 chemokine receptor is a 7-transmembrane
G-protein coupled receptor (GPCR) expressed on the surface
of a variety of cell types including epithelial, endothelial,
neuronal, and hematopoietic cells including CD4+ T cells, the
target for HIV infection. The natural ligand for the receptor is
the chemokine CXCL12 (SDF-1). The CXCR4/CXCL12 axis
plays major roles in hematopoiesis, hematopoietic stem cell and
(
3) (a) Schols, D.; Claes, S.; De Clerq, E.; Hendrix, C.; Bridger, G.;
Calandra, G.; Henson, G. W.; Fransen, S.; Huang, W.; Whitcomb,
J. M.; Petropoulos, C. J. AMD3100, a CXCR4 Antagonist, Reduced
HIV Viral Load and X4 LeVels in Humans, presented at the 9th
Conference on Retroviruses and Opportunistic Infections (CROI),Feb-
ruary 24-28, 2002,Seattle, WA. (b) Hendrix, C. W.; Collier, A. C.;
Lederman, M. M.; Schols, D.; Pollard, M.; Brown, S.; Jackson, J. B.;
Coombs, R. W.; Glesby, M. J.; Flexner, C. W.; Bridger, G. J.; Badel,
K.; MacFarland, R. T.; Henson, G. W.; Calandra, G. J. Acquired
Immune Defic. Syndr. Hum. RetroVirol. 2004, 37 (2), 1253–1262.
4) Bridger, G.; Skerlj, R.; Kaller, A.; Harwig, C.; Bogucki, D.; Wilson,
T. R.; Crawford, J.; McEachern, E. J.; Atsma, B.; Nan, S.; Zhou, Y.;
Schols, D.; Smith, C. D.; Di Fluri, M. Preparation of Tertiary
N-(5,6,7,8-Tetrahydro-8-quinolinyl)-N-(1H-benzimidazol-2-ylmethy-
l)amines and Analogs as Chemokine Receptor Modulators for Treat-
ment of HIV or FIV. Int. Appl. WO2002034745, PCT, 2002.
1
lymphocyte trafficking and homing, and neonatal development.
(
T-tropic strains of HIV utilize CXCR4 as a coreceptor during
2
infection, enabling the entry of viral RNA into the human cell.
Through the use of a specific CXCR4 chemokine receptor
antagonist, the interaction between HIV and the CXCR4
coreceptor can be interrupted, and infection can therefore be
(5) (a) Stone, N.; Dunaway, S.; Flexner, C.; Calandra, G.; Wiggins, I.;
Conley, J.; Snyder, S.; Tierney, C.; Hendrix, C. W.; and the ACTG
A5191 Study Team., Biologic activity of an orally bioavailable CXCR4
antagonist in human subjects, abstr 7225. XV International AIDS
Conference, Bangkok, Thailand, 2004. (b) Stone, N.; Dunaway, S. B.;
Flexner, C.; Tierney, C.; Calandra, G. B.; Becker, S.; Cao, Y.-J.;
Wiggins, I. P.; Conley, J.; MacFarland, R. T.; Park, J.-G.; Lalama,
C.; Snyder, S.; Kallungal, B.; Klingman, K. L.; Hendrix, C. W.
Antimicrob. Agents Chemother. 2007, 51 (7), 2351–2358.
3
blocked. Thus, a specific CXCR4 chemokine receptor could
be of therapeutic benefit as an HIV entry inhibitor.
*
To whom correspondence should be addressed. E-mail: JCrawford@cdrd.ca.
Telephone: (604)-221-7750 (x119).
†
Current address: Centre for Drug Research and Development, Suite 364 -
2
1
259 Lower Mall, University of British Columbia, Vancouver, BC, Canada V6T
Z4. Telephone: 604-221-7750.
Current address: Genzyme Corp. 153 2nd Ave., Waltham, MA 02451, U.S.A.
(6) In open-label dose-finding phase 2a clinical study conducted by Adult
Aids Clinical Trial Group (ACTG Study A5210), patients were dosed
twice daily for 10 consecutive days: Saag, M.; Rosenkranz, S.; Becker,
S.; Klingman, K.; Kallungal, B.; Zadzikla, A.; Coakley, E.; Acosta,
E.; Calandra, G.; Johnson, V.; and NIAID AIDS Clinical Trial Group.
Proof of concept of ARV actiVity and AMD11070 (an orally admin-
istered CXCR4 entry inhibitor): results of the first dosing cohort A
studied in ACTG Protocol A5210. Abstr. 512, presented at the 14th
Conference on Retroviruses and Opportunistic Infections (CROI),Feb-
ruary 25-28, 2007,Los Angeles, CA.
‡
(
1) (a) Zou, Y. R.; Kottmann, A. H.; Kuroda, M.; Taniuchi, I.; Littman,
D. R. Nature 1998, 393, 595–599. (b) Nagasawa; Tachibana; K;
Kishimoto, T. Semin. Immunol. 1998, 10 (3), 179–85. (c) Murdoch,
C. Immunol. ReV. 2000, 177, 175–184. (d) Lavi, E.; Strizki, J. M.;
Ulrich, A. M.; Zhang, W.; Fu, L.; Wang, Q.; O’Conner, M.; Hoxie,
J. A.; Gonzalez-Scarano, F. Am. J. Pathol. 1997, 151 (4), 1035–1042.
(
e) Watt, S. M.; Forde, S. P. Vox Sang. 2008, 94, 18–32.
(
2) (a) Caroll, R. G.; Riley, J. L.; Levine, B. L.; Feng, Y.; Kaushal, S.;
Ritchey, D. W.; Bernstein, W.; Weislow, O. S.; Brown, C. R.; Berger,
E. A.; June, C. J. Science 1997, 276, 273–276. (b) Schramm, B.; Penn,
M. L.; Speck, R. F.; Chan, S. Y.; De Clerq, E.; Schols, D.; Connor,
R. I.; Goldsmith, M. A. J. Virol. 2000, 74, 184–192.
(7) Skupinska, K. A.; McEachern, E. J.; Skerlj, R. T.; Bridger, G. J. J.
Org. Chem. 2002, 67, 7890–7893.
(8) McEachern, E. J.; Yang, W.; Chen, G.; Skerlj, R. T.; Bridger, G. J.
Synth. Commun. 2003, 33, 3497.
1
0.1021/op8000993 CCC: $40.75

2008 American Chemical Society
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
823
Published on Web 07/18/2008

Scheme 1
of the racemic amine 2 on scales up to 45 kg. Specifically,
n-hexyllithium was selected as a preferred base over n-
butyllithium for safety reasons (less pressurization of reaction
vessel) and isoamyl nitrate was selected over tert-butyl nitrite
due to its lower volatility and commercial availability in high-
purity.
Subsequent resolution of the racemic amine rac-2 with
Candida antarctica lipase B (CaLB) to give the (S)-isomer 2,
and the corresponding (R)-isomer as the acetamide 10, were
9
ments was planned to proceed Via reductive amination to install
based on our established procedures (Scheme 4). Larger-scale
3
, and N-alkylation to install 4.
The initial medicinal chemistry route to 1 is described in
resolutions were carried out in toluene instead of neat ethyl
acetate or isopropyl ether, with ethyl acetate as the acylating
agent, and afforded 41% yields (82% of theoretical maximum)
of (S)-amine 2 in high enantiopurity (97% ee). Typically, the
amine was isolated as the solid bis-HCl salt, due to its enhanced
stability and handling convenience over the liquid freebase. It
should be noted that it was possible, in small-scale proof of
concept experiments, to isolate the acetamide 10, racemize the
stereocenter at elevated temperature (150 °C, neat), and
hydrolyze the acetamide to provide rac-2, thus providing a
Scheme 2, and was used to provide material for preclinical
10
evaluation. The aminobutyraldehyde 3 was generated from
-aminobutanol, after protection of the primary amine as a
4
11
phthalimide by oxidation with tetrapropylammonium perru-
thenate (TPAP) to provide the aldehyde 3. Installation of the
tetrahydroquinoline fragment 2 was accomplished through
12
13
established reductive amination methods. The secondary amine
was purified by column chromatography before being
subjected to an N-alkylation with N-BOC-2-chloromethylben-
5
9
potential pathway for recycling the reagent.
14
zimidazole 4, which, in turn, was synthesized from com-
mercially available 2-chloromethylbenzimidazole. Removal of
the phthaloyl protecting group was achieved with hydrazine,
and the product 1 was initially isolated as the amorphous
hydrochloride salt. Overall, the yield from commercial reagents
was approximately 8% (six linear steps), and four of the six
steps required column chromatography. Synthetic refinement
would clearly be necessary to furnish the kilogram amounts of
Development of the aminoaldehyde (3) synthesis was
initiated from 4-aminobutyric acid and made use of the
11
phthaloyl protecting group (Scheme 5) to generate 11.
Reduction of the acid was accomplished Via hydrogenolysis of
16
the acid chloride 12. A point of note is that significant
variability in the efficiency of the hydrogenolysis during early
development was traced to carryover of p-TsOH which was
initially used in the formation of 11. Reduction of the residual
p-toluenesulfonic acid, or sulfonyl chloride after treatment with
oxalyl chloride, provided 4-methylbenzenethiol as a potential
catalyst poison. Its presence was confirmed by the isolation of
1
necessary for preclinical and clinical evaluation.
Development of Practical Routes to Intermediates 2, 3,
and 4. In the initial route, it was possible to generate racemic
-amino-5,6,7,8-tetrahydroquinoline 2 from the corresponding
unsubstituted 5,6,7,8-tetrahydroquinoline 7, Via formation of an
17
8
the thioester side product 13 as a contaminant in 3. Once the
p-TsOH was removed from the process, however, the hydro-
genolysis became predictable and scaleable to provide up to
30-kg amounts of 3. Interestingly, the reaction rate for the
protection of the amine was only decreased by approximately
2-fold in the absence of p-TsOH. By using 10% Pd/C as a
15
N-oxide followed by a Boekelheide rearrangement to install
the alcohol functionality at the benzylic position in 8, then
functional group conversion to generate 2 (Scheme 3). However,
the route was six steps long, and the yields were low (33%).
Accordingly, we developed a preferred chromatography-free
route using a regioselective oxime formation as a key trans-
2
catalyst with 15 psi H (g) at ambient temperature, the hydro-
genolysis reaction was complete in approximately 6 h. Both
11 and 3 were isolated as solids in high purity (>95% by
HPLC) through precipitative workup procedures.
8
formation. With the regioselective nitrosation of 7 to generate
the oxime 9, followed by a zinc-mediated reduction, it was
possible to efficiently access the racemic amine 2 in two steps.
Further refinement of reagent stoichiometries and reaction
conditions enabled the safe and reliable large-scale preparation
The benzimidazole fragment 4 was generated by protecting
commercially available 2-chloromethylbenzimidazole with
14
BOC. Changing the reaction solvent from THF to DMF
enabled the reaction to be performed at concentrations up to 2
M, and at an increased reaction rate. In THF, the reaction
typically required 4-6 days to consume the reagent, whereas
in DMF the reaction was typically complete in 12-16 h. Again,
a precipitative workup was applied to isolate 4 in 96+% area
purity by HPLC.
(
9) (a) Skupinska, K. A.; McEachern, E. J.; Baird, I. R.; Skerlj, R. T.;
Bridger, G. J. J. Org. Chem. 2003, 68, 3546–3551. (b) Bridger, G. J.;
Skerlj, R. T.; Kaller, A.; Harwig, C.; Bogucki, D.; Wilson, T. R.;
Crawford, J.; McEachern, E. J.; Atsma, B.; Nan, S.; Zhou, Y.; Schols,
D.; Smith, C. D.; DiFluri, R. Preparation of azolulmethylaminotet-
rahydroquinolines and related compounds as chemokine receptor
binding agents. PCT Int. Appl. WO 0222600, 2002.
(
(
(
(
(
(
10) Bridger, G. J.; Pelus, L. M. Chemokine combinations to mobilize
progenitor/stem cells. U.S. Pat. 20060035829, 2006.
11) Hamilton, R.; Walker, B. J.; Walker, B. Tetrahedron Lett. 1993, 34,
Assembly of 1. For the coupling steps, efforts were focused
on optimizing reaction conditions to minimize the formation
of side products so that chromatographic purifications could
be avoided. Initial evaluation of the reductive amination-based
2
847–2850.
12) Ley, S. V.; Marsden, S. P.; Norman, J.; Griffith, W. P. Synthesis 1994,
6
39–666.
13) Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Marynaoff, C. A.;
Shah, R. D. J. Org. Chem. 1996, 61, 3849–3862.
(16) Balenovi c´ , K.; Jambre sˇ i c´ , I.; Furi c´ , I. J. Org. Chem. 1952, 17, 1459–
1462.
14) An, H.; Wang, T.; Mohan, V.; Griffey, R. H.; Cook, P. D. Tetrahedron
1
998, 54, 3999–4012.
(17) McElhinney, R. S.; McCormick, J. E.; Bibby, M. C.; Double, J. A.;
Radacic, M.; Dumont, P. J. Med. Chem. 1996, 39, 1403–1412.
15) Beschke, H. Aldrichimica Acta 1978, 11, 13–16.
8
24
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

Scheme 2
Scheme 3
Scheme 5
Scheme 4
gel column chromatography. With these methods, it was
possible to isolate 5 in 45-70% yields from 3. For large-scale
synthesis, however, it was not feasible to use such an excess
of the comparatively costly substrate 2 nor was it practical to
employ silica gel column chromatography.
coupling of 2 and 3 using sodium borohydride or sodium
triacetoxyborohydride revealed that a complicated mixture of
side products was formed (Scheme 6). In general, acidic reaction
The first goal of the synthetic development was to attempt
to carry out the reductive amination with an equimolar stoi-
chiometry of 2 and 3. Early attempts were based on a
preformation of the imine 14 Via a dehydrative coupling
conditions favored the aldol-type self-condensation of 3 to form
1
1
5 (identified by H NMR and LC/MS as a presumed mixture
1
of E/Z isomers). The di-addition product 16 was formed if the
aldehyde 3 was present in excess, or if the relative rate of imine
(monitored by H NMR). An examination of various dehydrat-
ing agents and solvents led to the selection of potassium
carbonate in THF as a preferred combination to form 14. The
basic conditions during dehydration suppressed aldol-type
condensations, and the carbonate was easily removed by
filtration. In comparison, when magnesium sulfate was used, it
led to the formation of higher levels of impurities (>5 mol %
14 reduction was fast relative to the condensation of chiral amine
2
and N-phthaloyl aldehyde 3 (i.e. 5 was available as a reagent
during the course of reaction). Overreduction (to generate 17)
occurred if sodium borohydride or other strong reducing agents
were used.
In the initial small-scale syntheses using sodium triacetoxy-
borohydride, the practical solution to balance side-reactivity
issues was to use 1.5 equiv of chiral amine reagent 2, relative
to the aldehyde 3, to suppress di-addition. Following reaction
workup, the desired secondary amine 5 was purified by silica
1
by H NMR).
Having established that the imine 14 could be prepared as a
1
discrete reagent (formed in >95% conversions by H NMR),
various reducing agents were examined (Table 1). Unfortu-
nately, the commercial hydride-based reagents proved to be
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
825

Scheme 6
Table 1. Reagent selection for reduction of imine 14
a
side productsb
entry
reagent/solvent/temperature
conversion to 5 (%)
1
2
3
4
NaBH(OAc)
3
(1.5 equiv)/CH Cl (or THF)/ambient
2
2
70-75
60
∼10% di-addition (16)
∼20% over-reduced (17)
<5% over-reduced (17)
∼2% aldol (15)
NaBH
NaBH
NaBH
4
4
4
(1.0 equiv)/THF/-20 °C to ambient
(1.0 equiv) + AcOH (1.5 equiv)/THF/-20 °C
80
(1.0 equiv) + ZnCl
2
(1.1 equiv)/THF/-10 °C
80-90
a
Yields determined by 1H NMR of reaction mixtures prior: product integral/imine + product integral. b Measured by 1H NMR integration as fraction of product integral.
Scheme 7
unsuitable: sodium triacetoxyborohydride gave the di-addition
side product (16) as an impurity, presumably Via hydrolysis of
the imine in situ. In comparison, use of sodium borohydride
led to a degree of overreduction (17). Sodium cyanoborohydride
was not considered for toxicological reasons. Conversely,
attempts at catalytic hydrogenation (using Pd/C) were low-
yielding.
diisopropyl ether as an antisolvent to induce precipitation. By
controlling the solvent ratios and temperatures in the precipita-
tive conditions, the product 5 could be isolated in sufficient
purity (90+% area by HPLC) to eliminate the need for column
chromatography. The process was successfully carried out on
up 20-kg scales, with yields of approximately 65% from the
amine 2 (bis-HCl salt).
We determined that the ideal chemical reducing agent should
have a reducing activity between that of sodium triacetoxy-
borohydride and sodium borohydride. Indeed, evaluation of such
attenuated borohydride agents described in literature reports,
Installation of the benzimidazole 4 was accomplished with
only minor modifications of the early developed work (Scheme
7). The reaction was conducted at ∼0.5 M in acetonitrile at 50
°C, using N,N-diisopropylethylamine (DIPEA) as a base in the
presence of catalytic potassium iodide. Careful control of
stoichiometry, and an HPLC test for reaction completion was
implemented to ensure complete reaction (typically 6-12 h).
At the completion of the reaction, the carbamate protecting
group was hydrolyzed with aqueous HCl, which also served to
hydrolyze any residual benzyl chloride 4. The tertiary amine
then became soluble in the aqueous layer, and nonpolar
impurities could be removed Via a wash with tert-butyl methyl
ether (MTBE). After adjusting the pH to approximately 13, the
18
19
such as sodium acetoxyborohydride and zinc borohydride
reagents generated in situ) gave a cleaner, higher-yielding
(
reaction profile. Continued experimentation led to the selection
of a 1.1:1 ratio of zinc chloride:sodium borohydride as the
optimal reducing agent (Table 1, entry 4), since it not only gave
a clean, high-yielding reaction profile, but also enabled the
reaction to be run at concentrations up to 1 M in substrate (vs
the 0.1 M maximum using sodium acetoxyborohydride reagent).
To avoid chromatography, a procedure was developed to
isolate the secondary amine 5 as a solid in 90+% purity using
(
19) Bhattacharyya, S.; Chatterjee, A.; Duttachowdhury, S. K. J. Chem.
Soc., Perkin. Trans. 1 1994, 1–2.
(
18) Reetz, T. J. Am. Chem. Soc. 1960, 82, 5039–5042.
8
26
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

Scheme 8
product 18 was selectively extracted into toluene. It should be
noted that excessively basic conditions (pH g 14) could lead
to the partial hydrolysis of the phthalimide protecting group.
The tertiary amine 18 was typically stored as a toluene stock
solution at -10 °C.
based removal of the cyclic diamide side product 21, as shown
in Scheme 8: it was possible to selectively extract 21 into
dichloromethane at pH 6, leaving 1 in the aqueous layer.
The workup consisted of repeated washes of the dichlo-
romethane solution of 1 with dilute aqueous base (0.5 N NaOH)
in order to remove residual hydrazine. Treatment with activated
charcoal served to effectively decolorize the solution of 1, and
The final goal of the synthetic development was to isolate a
pure, stable, crystalline form of 1. Although hydrazine is often
20
used for cleaving the phthaloyl protecting group, the high
toxicity of the reagent provided a compelling incentive to search
for alternatives. A number of amines were evaluated for the
4
a silica gel pad (elution with 1% NH OH, 79% dichlo-
romethane, and 20% methanol) was applied to remove residual
low-polarity and high-polarity (baseline) impurities. At 50-70
g laboratory scale, approximately seven fractions of 500 mL
each were collected, the product would be present in fractions
3-7, and the high-polarity impurity would remain on the
column. It should be noted that the silica pad was successfully
applied in the pilot plant (at 7-kg scale) by filling a Nutsche
filter with silica and eluting 50-L fractions. A final wash with
0.5 M NaOH was applied to the pooled, concentrated fractions
to ensure that 1 was not partially protonated, due to exposure
to silica gel, and was later used in conjunction with an
in-process, specifically developed HPLC test to ensure that the
residual hydrazine level was <8 ppm. At this stage in the
sequence, 1 was typically 98+% area pure by HPLC analysis
of the solution.
Isolation of 1 as a Crystalline Solid. After extensive
laboratory experimentation, ethyl acetate was selected as the
recrystallization solvent, as it enabled the removal of residual
dichloromethane from the solution of 1 Via a vacuum distillation
solvent exchange. The recrystallization was then conducted by
heating to 65-67 °C to solubilize the substrate and then cooling
to 40 °C and seeding to induce crystallization.
2
1
22
deprotection, including methylamine, butylamine, hydroxy-
23
lamine hydrochloride with sodium methoxide), ammonia, and
ethylene diamine. Generally, the reactions were conducted in
alcohol solvents (ethanol or butanol), at both room and elevated
24
(80 °C) temperatures. The introduction of the first equivalent
of nucleophilesto form a mixed diamidesusually proceeded
efficiently and could be tracked by NMR or LC/MS. The
problem step was the displacement of 1 from the phthalic
diamide with a second equivalent of nucleophile. In the case
of methyl- and butylamine, the reaction proceeded to only
30-40% completion following overnight exposure to an excess
(>15 equiv) of the reagent at 80 °C. The conversions with
hydroxylamine and ammonia were even lower. Ethylenediamine
was more reactive but was ultimately abandoned as a reagent
since the pendant primary amine functionalities on the phthalic
diamide side product would make extractive removal of the
“impurity” from an aqueous solution of the polyamine 1
difficult. Attempts at using a reductive removal of the phthal-
25
imide with reagents such as sodium borohydride/acetic acid
were also unsuccessful due to racemization of the chiral center
presumably occurring upon exposure to hot acetic acid).
(
The crystallization of 1 was capable of rejecting some
chemical impurities, though more importantly, it enhanced the
enantiomeric excess (ee) of 1. In small-scale laboratory trials
at 50-g scale, it was possible to upgrade the ee from 85% to
98%. This property was effectively exploited in the large-scale
(10 kg) conversion of a 90% ee batch of 1, which yielded
crystalline material with an enantiopurity of >99% ee. An X-ray
crystal structure determination confirmed the crystalline nature
of 1, whose asymmetric centre had previously been determined
Consequently, hydrazine remained the only practical reagent,
in spite of the hazards, for the transformation at large scale.
Eventually, suitable conditions were developed for the
reaction: eight equivalents of hydrazine in methanol at room
temperature. Typically, the hydrolysis was complete in 16-24
h. The isolation of 1 was based on the selective extraction-
(
(
20) Sasaki, T.; Minamoto, K.; Itoh, H. J. Org. Chem. 1978, 43, 2320–
2
325.
9
to be (S) Via analysis of 2 (as a Re complex).
21) (a) Sen, S. E.; Roach, S. L. Synthesis 1995, 756–758. (b) Ojima, I.;
Lin, S.; Chakravarty, S.; Fenoglio, I.; Park, Y. H.; Sun, C.-M.;
Appendino, G.; Pera, P.; Veith, J. M.; Bernacki, R. J. J. Org. Chem.
1
998, 63, 1637–1645.
Conclusions
(
(
22) Kanie, O.; Crawley, S. C.; Palcic, M. M.; Hindsgaul, O. Carbohydr.
Res. 1993, 243, 139–164.
An efficient four-step synthetic route to crystalline 1 has been
developed which employs only a single chromatographic step:
a silica gel pad in the purification of 1 prior to crystallization.
The key step in the sequence is a novel reductive amination,
employing an attenuated borohydride reducing agent, between
23) Carroll, F. I.; Berrang, B.; Linn, C. P. J. Med. Chem. 1985, 28, 1564–
1
567.
(
(
24) Wentworth, P.; Janda, K. D. Synlett 1997, 537–543.
25) Osby, J. O.; Martin, M. G.; Ganem, B. Tetrahedron Lett. 1984, 25,
2
093–2096.
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
827

the chiral (S)-amine 2 and the aminobutyraldehyde 3 to afford
the secondary amine 5. N-alkylation with 4 provided the
requisite protected tertiary amine 18, which was purified Via a
pH-selective extraction. Hydrolysis of the phthalimide protecting
group with hydrazine was followed by further pH-selective
extraction methods to afford 1 in solution. In the final step, a
crystallization procedure for the isolation of 1 was developed,
which allows for the enhancement of both chemical and chiral
purity. The synthetic methods described herein have been
successfully applied on large scale (with minor modification)
to supply 7-kg amounts of 1.
h and is evaluated by comparison of R-proton resonances,
expecting >95% conversion). The reaction was cooled to a
maximum temperature of -8 °C and agitated at this temperature
for ∼1 h. The reaction mixture was filtered, and the filter cake
was rinsed with toluene (-8 °C, 75 mL). The filter cake was
dried under vacuum at a maximum temperature of 50 °C. The
protected acid 11 was isolated in a yield of 84% (64 g, 99.5%
HPLC purity). The acid 11 (66.4 g) was then dissolved in
dichloromethane (280 mL), and oxalyl chloride (40 g) was
added over ∼8 h (to control the gas evolution). The completion
of the reaction was confirmed by in-process NMR (R-proton
resonances). The solution was then concentrated. Tetrahydro-
furan (300 mL) and 2,6-lutidine (31.7 g) were added. The
solution was transferred to a hydrogenator, rinsing forward with
THF (2 × 30 mL). To this solution was added a slurry of 10%
Pd/C (3.3 g) in dichloromethane (10 mL). The mixture was
It should be noted that an alternate procedure was subse-
quently developed and has been carried out on approximately
20-kg scale to provide a crystalline salt form of AMD070
without intermediate isolation of crystalline 1; however, the
synthetic development was suspended before the process for a
high-purity conversion to the crystalline freebase could be
developed. The procedure uses an N,N-di-BOC to protect the
aminoaldehyde fragment, which enables the removal of hydra-
zine and the silica gel column from the synthetic process. We
intend to report on these procedures in due course.
2
hydrogenated under 15 psi H gas at 22 °C. After 6 h, in-process
NMR analysis was performed to monitor conversion to the
aldehyde (>95% conversion). The reaction mixture was filtered
through a Celite pad, and the filter cake was rinsed with THF
(30 mL). The filtrate was combined and concentrated to about
120 mL; 1 N HCl solution (100 mL) was added, and the
resulting solution was agitated at 22 °C for 16 h. THF was
removed under vacuum, and then dichloromethane (150 mL)
was added. The organic phase was separated and the aqueous
phase discarded. The organic phase was returned to the flask.
An aqueous solution of 8% sodium bicarbonate (200 mL) was
added, and the mixture was agitated for 60 min. The layers
were separated, and the organic layer was concentrated to about
Experimental Section
1
13
H NMR (300 MHz) and C NMR (75 MHz) spectra were
recorded using CDCl solvent with TMS as an internal standard
unless otherwise noted). HPLC was performed using reversed-
phase conditions.
3
(
Isolation of (S)-8-Amino-5,6,7,8-tetrahydroquinoline Free
Base (2). (S)-8-amino-5,6,7,8-tetrahydroquinoline hydrochloride
2
50 mL; then heptane (200 mL) was added. After agitating the
mixture for 15 min the solution was concentrated to ∼250 mL
to remove dichloromethane), and additional heptane (200 mL)
(
80 g, 0.361 mol) was dissolved in deionized water (200 mL)
and neutralized to pH 7 with a 10 N sodium hydroxide solution
∼32 mL), then diluted with additional deionized water (100
mL). The mixture was extracted with dichloromethane (3 ×
00 mL) until 8-acetamido-5,6,7,8-tetrahydroquinoline was not
(
(
was added. The mixture was adjusted to <5 °C and agitated at
that temperature for 1 h. The solid was filtered, and the filter
cake was rinsed with heptane (150 mL). The cake was dried
under vacuum at ambient temperature. The yield of 3 was 77%
3
present by thin layer chromatography (TLC conditions: dichlo-
romethane/methanol/ammonium hydroxide (88:10:2). 8-Aceta-
mido-5,6,7,8-tetrahydroquinoline R
1
0
(
96.6% area HPLC purity). Spectral data matched literature
f
) 0.6. (S)-8-amino-5,6,7,8-
1
values: H NMR (CDCl
3
) δ 2.00 (m, 2H), 2.50 (t, 2H, J ) 6.3
tetrahydroquinoline R ) 0.5). The aqueous layer was adjusted
f
Hz), 3.73 (t, 2H, J ) 6.3 Hz), 7.70 (m, 2H), 7.82 (m, 2H), 9.75
s, 1H).
Preparation of tert-BOC-2-chloromethylbenzimidazole
4). 2-Chloromethylbenzimidazole (285 g, 1.71 mol) and di-
tert-butyl dicarbonate (410 g, 1.88 mol) were suspended in DMF
850 mL). N,N-diisopropylethylamine (30 mL, 0.17 mol) was
to pH 13-14 with 10 N sodium hydroxide (30 mL) and was
extracted with dichloromethane (3 × 300 mL). The combined
organic layers were dried (sodium sulfate) and concentrated in
Vacuo to afford (S)-8-amino-5,6,7,8-tetrahydroquinoline (2,
(
(
51.3 g, 96%) as a dark-brown oil. Note: the amine freebase
(
should be stored under an inert atmosphere at low temperatures
1
added to the agitated reaction mixture. The reaction mixture
(
-10 °C or below) if not used immediately. H NMR (CDCl
3
)
1
was agitated at 25 °C for 5 h (an in-process H NMR confirmed
δ 1.64-1.84 (m, 2H), 1.94-2.01 (m, 1H), 2.14-2.23 (m, 1H),
>5:1 ratio of 4:2-chloromethylbenzimidazole by comparison
2
1
.69-2.87 (m, 2H), 3.99 (dd, 1H, J ) 7.7, 5.3 Hz), 7.06 (dd,
H, J ) 7.7, 4.4 Hz), 7.36 (d, 1H, J ) 7.5 Hz), 8.41 (d, 1H, J
of benzylic resonances). Then additional di-tert-butyl dicar-
bonate (82 g, 0.36 mol) was added, and the reaction mixture
was warmed to 40 °C and agitated for a further 6 h (until >97%
)
4.4 Hz). Chiral purity was determined by gas chromatography
to be 97.5% ee (separated by chiral GC, J&W CycloSil B
column, isothermally run at 130 °C for 40 min, (S)-(+)-
enantiomerrt ) 26.3 min, (R)-(-)-enantiomerrt ) 28.7 min).
Preparation of 4-(1,3-Dioxo-1,3-dihydroisoindol-2-yl)bu-
tyraldehyde (3). 4-Aminobutyric acid (35 g) and phthalic
anhydride (52.5 g) were suspended in toluene (300 mL) and
the reaction mixture was adjusted to reflux. Water was removed
by a Dean-Stark trap. In-process NMR was performed to
confirm the completion of the reaction (typically requires 6-12
1
conversion by H NMR). Water (250 mL) was added, and the
reaction mixture was agitated for 2 h at 40 °C. The reaction
mixture was then concentrated under vacuum at 40 °C to
approximately half-volume. The temperature was reduced to
22 °C, water (100 mL) and DMF (200 mL) were added, and
the mixture agitated for 2 h, during which time a precipitate
formed. The contents were filtered through a glass frit. The filter
cake was washed with water (250 mL). The filter cake was
8
28
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

then dried under vacuum. The yield of (4) was 365 g (80%).
By HPLC: 96% area pure. Spectral data matches literature
desired product was obtained in a 68.6 g yield (84%) as a light-
brown crystalline solid. H NMR (CDCl ) δ 1.59-2.17 (series
3
14
1
1
values: H NMR (CDCl
m, 2H), 7.73 (dd, 1H, J ) 4.1, 0.9 Hz), 7.98 (dd, 1H, J ) 4.1,
.9 Hz).
Preparation of Secondary S)-2-[4-(5,6,7,8-Tetrahydro-
3
) δ 1.73 (s, 9H), 5.06 (s, 2H), 7.37
of m, 8H), 2.74 (m, 4H), 3.72 (t, 2H, J ) 7.2 Hz), 3.72 (m,
1H), 7.04 (dd, 1H, J ) 7.8, 4.8 Hz), 7.35 (dd, 1H, J ) 7.8, 0.6
Hz), 7.70 (m, 2H), 7.82 (m, 2H), 8.36 (dd, 1H, J ) 4.8, 0.6
(
0
13
Hz). C NMR (CDCl ) δ 20.0, 27.0, 28.1, 29.1, 29.3, 47.7,
3
quinolin-8-ylamino)butyl]isoindole-1,3-dione (5).
58.5, 122.2, 123.5 (2C), 132.5, 134.2 (2C), 137.3, 147.2, 158.3,
168.8 (2C). ES-MS m/z 350 (M + H); Purity by HPLC 94%
area. Chiral purity 97% ee (by chiral HPLC).
Part 1. Imine Formation [(S)-2-[4-(5,6,7,8-Tetrahydroquino-
lin-8-ylamino)but-4-enyl]isoindole-1,3-dione (14)]. To a solution
of 8-amino-5,6,7,8-tetrahydroquinoline (2, 50.0 g, 338 mmol,
Preparation of Tertiary Amine (S)-2-{4-[1-Methyl-1H-
benzimidazol-2-ylmethyl)-(5,6,7,8-tetrahydroquinolin-8-
yl)amino]butyl}isoindole-1,3-dione (18). A 5-L three-necked
flask was charged with secondary amine 5 (65 g, 0.186 mol,
1.0 equiv), acetonitrile (0.37 L), N,N-diisopropylethylamine (48
mL, 0.28 mol, 1.5 equiv), N-BOC-2-chloromethylbenzimidazole
(4, 52.0 g, 196 mmol, 1.05 equiv), and potassium iodide (3.08
g, 18.6 mmol, 0.1 equiv) in that order. The mixture was stirred
mechanically at a 50 °C internal temperature under a nitrogen
1.0 equiv) in THF (1700 mL, [0.2 M]) in a 3-L flask was added
4-(N-phthalimidylamino)butanal (3, 73.3 g, 338 mmol, 1.0
equiv) and 325 mesh potassium carbonate (325 mesh, 46.6 g,
38 mmol, 1.0 equiv). The mixture was then stirred for 1 h
and filtered. An NMR aliquot is used to monitor complete imine
14) formation, using the triplet at 4.31 ppm as a diagnostic
resonance for product formation (reaction is deemed complete
3
(
1
when <3 mol % 2 and 3 exist relative to 14). H NMR (CDCl
3
)
1
δ 1.76-2.19 (series of m, 6H), 2.35 (m, 2H), 2.78 (m, 2H),
atmosphere for 4 h. A H NMR of an aliquot indicated the
3
4
7
.73 (m, 2H), 4.31 (t, 1H, J ) 5.1 Hz), 7.05 (dd, 1H, J ) 7.8,
.8 Hz), 7.38 (d, 1H, J ) 7.8 Hz), 7.69 (m, 2H), 7.80 (m, 2H),
.82 (t, 1H, J ) 4.1 Hz), 8.38 (d, 1H, J ) 4.8 Hz).
reaction was complete (the appearance of a multiplet at 4.18
ppm (CDCl ) is indicative of product formation, >95% conver-
3
sion required), and the mixture was then allowed to cool to
RT. Water (300 mL) and tert-butyl methyl ether (500 mL) were
added. The pH was adjusted to pH 3 by addition of 6 N HCl
(∼35 mL). The aqueous and organic layers were separated, the
ether layer discarded, and the aqueous layer was stirred at RT
for 8-16 h. Toluene (600 mL) was added, and the pH was
adjusted to 13 with 10 M sodium hydroxide. The mixture was
then filtered through a pad of diatomaceous earth (∼8 cm
diameter × 1 cm depth), washing forward with approximately
50 mL of toluene. After separation of the aqueous and organic
layers, the aqueous layer was extracted with toluene (500 mL).
The combined organic layers were washed with 5% w/w
aqueous sodium hydroxide (20 mL), and the tertiary amine 18
was held at -10 °C (or colder) as a stock solution. Yield 74%
Part 2. Reducing Agent Formation. To a flask containing
THF (160 mL) was added zinc(II) chloride (35.1 g, 257 mmol).
A mild exotherm occurred upon dissolution. Sodium borohy-
dride (8.85 g, 234 mmol) was then added slowly. The mixture
was stirred for 1 h, during which time a homogeneous solution
formed. The solution was cooled to -20 °C.
Part 3. Reduction. A solution of imine 14 (234 mmol) in
THF (80 mL) was cooled to -20 °C, and was then added slowly
to the cooled solution of zinc chloride and sodium borohydride
Via cannula, maintaining the internal temperature of the reaction
flask between -10 and -20 °C. The reaction was then stirred
at -15 °C for 2 h. The reaction completion was confirmed by
1
3
H NMR (CDCl ) by the disappearance of the imine-specific
1
triplet resonance in the proton NMR at 4.3 ppm, or the triplet
at 7.82 ppm (<2 mol % residual reagent). A solution of 6 N
aqueous HCl was added dropwise, maintaining the temperature
below -5 °C, until the pH of the aqueous layer measured 2-3
(by H NMR estimate). A sample of the solution was concen-
1
trated: H NMR (CDCl
3
) δ 1.42-2.17 (series of m, 8H), 2.74
(m, 4H), 3.52 (t, 2H, J ) 7.2 Hz), 3.98 (m, 1H), 4.02 (d, 1H,
J ) 16.8 Hz), 4.11 (d, 1H, J ) 16.8 Hz), 7.04 (dd, 1H, J )
7.8, 4.8 Hz), 7.15 (m, 4H), 7.35 (d, 1H, J ) 7.8 Hz), 7.40 (br
s, 1H (NH)), 7.64 (m, 2H), 7.76 (m, 2H), 8.81 (d 1H, J ) 4.8
Hz). ES-MS m/z 480.3 (M + H (LC/MS)); Purity by HPLC
93% area. Chiral Purity 97.5% ee by chiral HPLC.
(by pH paper). The reaction was allowed to warm to room
temperature, and a solution of 13% aqueous sodium carbonate
was added until the pH reached 4, as measured with pH paper.
The reaction flask was placed under vacuum, and the THF
solvent was removed by distillation. Water (345 mL) and
dichloromethane (260 mL) were then added. The mixture was
agitated, and then the aqueous and organic layers were
separated. The organic layer was washed with concentrated
aqueous ammonium hydroxide (110 mL) and then water (150
mL). The dichloromethane solution was concentrated to ∼75
mL under vacuum, then diisopropyl ether (520 mL) was added.
Caution: It should be noted that diisopropyl ether is very
flammable and can form explosive peroxides upon storage. As
a matter of safety, the solvent should be tested for peroxide
content with a commercial peroxide test strip prior to use. The
solution was concentrated under vacuum to ∼160 mL and was
then cooled slowly to -10 °C, with agitation, during which
time a precipitate formed. The precipitate (5) was filtered, and
washed with diisopropyl ether. After drying under vacuum, the
Preparation of (S)-N′-(1H-Benzimidazol-2-ylmethyl)-N′-
(5,6,7,8-tetrahydroquinolin-8-yl)butane-1,4-diamine (1, solu-
tion). Tertiary amine 18 (0.137 mol, in toluene) was concen-
1
trated under vacuum to a foam (0.25 equiv PhCH
3
by H NMR).
The residue was taken up in methanol (550 mL), transferred to
a 2 L round bottomed flask, which was purged with nitrogen,
and treated with hydrazine hydrate (63 mL, 1.1 mol, 8.0 equiv).
The mixture was stirred mechanically in a capped vessel for
20 h at 25 °C (monitored by HPLC, >98% conversion
required). The thick beige slurry was filtered through a glass
frit (washing forward with 500 mL of dichloromethane). The
filtrate was washed with 200 mL of 0.5 M NaOH. The aqueous
layer was discarded. To the organic layer was added water (300
mL), and 6 M HCl was added until the pH was adjusted to
5-6 (with stirring). The layers were separated, and the organic
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
829

layer was discarded. To the aqueous layer was added activated
charcoal (Norit G-60, 5 g), and the mixture was stirred for 1 h.
The suspension was filtered, and to the filtrate (pale yellow)
was added dichloromethane (200 mL). The pH of the aqueous
layer was adjusted to 13 with the addition of 4 M NaOH. The
layers were then separated, and the organics were concentrated.
The residue was taken up in 50 mL of dichloromethane and
was passed through a pad of silica gel (60 g, preconditioned
The filter cake was placed under a cone of nitrogen and was
allowed to dry for 5 min while still under vacuum pressure.
The crystals were transferred to a flask and were placed under
vacuum for 1 h. The crystals were placed in a 40 °C vacuum
oven (<1.0 mmHg) and were dried for 24 h. The yield of (off-
1
white) crystalline 1 was 32.45 g (67%). H NMR (CDCl
3
) δ
1.23-1.49 (m, 4H), 1.62-1.77 (m, 1H), 1.85-1.97 (m, 1H),
2.00-2.10 (m, 1H), 2.16-2.26 (m, 1H), 2.51 (t, 2H, J ) 6.8
Hz), 2.54-2.62 (m, 1H), 2.67-2.78 (m, 1H), 2.81-2.92 (m,
1H), 7.15 (d, 1H, J ) 7.6 Hz), 7.18-7.23 (m, 2H), 7.59 (br s,
4
with eluent: 15% methanol, 2% conc. NH OH in dichlo-
romethane). The product was eluted with the preconditioning
13
solvent mixture in fractions of ∼500 mL volume. Typically
1H), 8.60 (d, 1H, J ) 4.4 Hz). C NMR (CDCl ) δ 21.8, 23.9,
3
7-8 fractions were collected. The combined fractions containing
26.4, 29.6, 31.4, 42.1, 49.8, 50.9, 62.2, 115.2, 121.8 (2C), 122.5,
134.9, 137.7 (2C), 147.0 (2C), 157.0, 157.9. ES-MS m/z 350
product were concentrated, then taken up in 500 mL of
dichloromethane, washed with 100 mL of 0.5 N NaOH and
then water (100 mL). The organic layer was dried over sodium
sulfate. The solution of 1 in dichloromethane was stored at low
temperature (-10 °C or lower) in preparation for crystallization.
Crystallization of 1. The solution of 1 in dichloromethane
(
>
M + H). Chemical purity: 99.4% (area HPLC). Chiral purity:
99% ee (chiral HPLC). Melting point: 108-110 °C. Anal.
Calcd for C11 : C, 72.17; H, 7.79; N, 20.04. Found: C,
2.07; H, 7.71; N, 19.95.
27 5
H N
7
Acknowledgment
(47.8 g, 0.137 mol theoretical yield) was concentrated to a
We thank the following for their contributions: Raylo
Chemicals Inc. (Edmonton, AB, Canada) and Aptuit (formerly
Evotec, Abingdon, Oxon, UK) for collaborative development
and for large-scale manufactures of AMD070 under GMP
conditions; the analytical staff at AnorMED for their method
development and analyses; Brian Patrick (University of British
Columbia) for the crystal structure determination.
volume of approximately 120 mL under vacuum (at 25 °C).
Ethyl acetate (600 mL) was then added, and the mixture was
concentrated under vacuum to a final volume of 350 mL. The
1
solvent ratio was checked by H NMR (if the dichloromethane
content is >1 mol% with respect to ethyl acetate, the concentra-
tion cycle was repeated). The suspension was then heated, with
stirring, to an internal temperature of 65-67 °C (over a period
of 30-60 min). The mixture was then stirred at 65 °C for five
minutes, until all solid material dissolved, and the solution was
transparent and homogeneous. The solution was allowed to cool
to 45 °C, at which point 30 mg of a sample of crystalline 1
was added to seed crystallization. The solution was allowed to
slowly cool to room temperature (over 3 h), with stirring. The
flask was cooled to 5 °C for 1 h. The off-white crystals were
isolated by suction filtration (through filter paper), and the
crystals were washed with 50 mL of cold (0 °C) ethyl acetate.
Supporting Information Available
Characterization and purity data for new compounds (1, 5,
6, and 18), representative spectra for known compounds; CIF
file and crystal structure report for 1. This material is available
free of charge Via the Internet at http://pubs.acs.org.
Received for review April 22, 2008.
OP8000993
8
30
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development