Development of a multigram asymmetric synthesis of 2-(R)-2-(4,7,10-tris tert-butylcarboxymethyl-1,4,7,10-tetraazacyclododec-1-yl)-pentanedioic acid, 1-tert-butyl ester, (R)-tert-Bu4-DOTAGA1

Article abstract of DOI:10.1021/op8002932

A process for the multigram asymmetric synthesis of the chiral tetraazamacrocycle 2-(R)-2-(4,7,10-tris tert-butylcarboxymethyl- 1,4,7,10-tetraazacyclododec-1-yl)-pentanedioic acid, 1-tert-butyl ester ((R)-tert-Bu₄-DOTAGA, 4) has been devised and demonstrated. The nine-step synthesis features an improved synthesis of 2-(S)-5- oxotetrahydrofuran- 2-carboxylic acid, tert-butyl ester 8, the precursor to the novel alkylating agent (S)-5-benzyl 1-tert-butyl 2-(methylsulfonyloxy) pentanedioate 12, which was used to introduce an orthogonally protected chiral glutarate arm to the 1,4,7,10-tetraazacyclododecane (cyclen) nucleus in high optical purity. Cyclen derivative (R)-t-Bu₄-DOTAGA, 4, a key intermediate for the manufacture of a magnetic resonance imaging (MRI) candidate, was produced with high chemical (≥95%) and optical (ee ≥ 97%) purity. The process developed was successfully applied to the kilogram-scale cGMP synthesis of (R)-t-Bu₄-DOTAGA.

Full text of DOI:10.1021/op8002932

Organic Process Research & Development 2009, 13, 535–542
Development of a Multigram Asymmetric Synthesis of 2-(R)-2-(4,7,10-Tris
tert-Butylcarboxymethyl-1,4,7,10-tetraazacyclododec-1-yl)-pentanedioic Acid,
1-tert-Butyl Ester, (R)-tert-Bu4-DOTAGA1
Stuart G. Levy,† Vincent Jacques,* Kevin Li Zhou, Shirley Kalogeropoulos, Kelly Schumacher, John C. Amedio,‡
Jonathan E. Scherer, Steven R. Witowski,§ Richard Lombardy,⊥ and Karsten Koppetsch¶
EPIX Pharmaceuticals, Inc, 4 Maguire Road, Lexington, Massachusetts 02421, and Johnson-Matthey Pharma SerVices, 25
Patton Road, DeVens, Massachusetts 01432, U.S.A.
Abstract:
we have used ligand 4, bearing a free carboxylate side chain as
a handle for conjugation, as a precursor to more highly
functionalized gadolinium chelate diagnostic agents.
A process for the multigram asymmetric synthesis of the
chiral tetraazamacrocycle 2-(R)-2-(4,7,10-tris tert-butylcar-
boxymethyl-1,4,7,10-tetraazacyclododec-1-yl)-pentanedioic acid,
1-tert-butyl ester ((R)-tert-Bu4-DOTAGA, 4) has been
devised and demonstrated. The nine-step synthesis
features an improved synthesis of 2-(S)-5-oxotetrahy-
drofuran-2-carboxylic acid, tert-butyl ester 8, the
precursor to the novel alkylating agent (S)-5-benzyl
1-tert-butyl 2-(methylsulfonyloxy)pentanedioate 12, which
was used to introduce an orthogonally protected chiral
glutarate arm to the 1,4,7,10-tetraazacyclododecane
(cyclen) nucleus in high optical purity. Cyclen deriva-
tive (R)-t-Bu4-DOTAGA, 4, a key intermediate for the
manufacture of a magnetic resonance imaging (MRI)
candidate, was produced with high chemical (g95%)
and optical (ee g 97%) purity. The process developed
was successfully applied to the kilogram-scale cGMP
synthesis of (R)-t-Bu4-DOTAGA.
Results and Discussion
Discovery Synthesis of 4. The initial synthesis of optically
enriched 4 was based on a preparation described in the literature
(Scheme 1).3 A sample of 4 prepared using this synthetic route
was converted to its (R)-(+)-R-methylbenzylamide 5 for ster-
eochemical analysis by proton NMR. The proton NMR
spectrum of this material showed that 5 prepared via this route
was a 2:1 (R,R):(R,S) mixture of diastereomers, indicating that
partial racemization of the alkylating agent or racemization
during the alkylation reaction had occurred. It was clear that a
key intermediate, such as 4, would need to be prepared in a
more optically homogeneous form for use in the manufacture
of a drug candidate. Additionally, the single literature reference
describing the synthesis of 4 reported its preparation on a scale
of only 500 mg, without an evaluation of the optical purity of
key intermediates or of the product. In order to support our
MRI diagnostic development program, it was necessary to
devise a synthetic route, which would produce 4 in multigram
and kilogram quantities, having greatly improved enantiomeric
excess (ee) and operational efficiency.
The basis for a more stereoselective asymmetric synthesis
of 4 was found in the literature.4 At Bristol-Myers-Squibb, Kang
and co-workers reported the use of the triflate ester of benzyl-
(L)-lactate to prepare DO3MA (2,2′,2′′-(1,4,7,10-tetraazacy-
clododecane-1,4,7-triyl)tris-propanoic acid) by alkylation of a
cyclen derivative. It was found that, whereas alkylation of
N-formyl cyclen with 2-chloropropionic acid required temper-
atures of 85-90 °C and a 12 h reaction time in aqueous base,
use of benzyl (L)-2-trifyloxy propionate alkylated this cyclen
derivative at ambient temperature, in the presence of diisopro-
pylethylamine, within 2.5 h (Scheme 2). The enhanced reactivity
of the triflate relative to the chloride derivative allowed the
reaction to proceed under far milder conditions, with the
expectation that the possibility of racemization would be
minimal.
Introduction
Macrocyclic chelates of Gd(III) provide an advantage over
open-chain chelates in that the macrocycle enhances the kinetic
inertness and thermodynamic stability of the coordination
complex.2 There are several approved, commercially available
MRI contrast agents based on tetraazamacrocyclic ligands
(Figure 1): Gd-DOTA (Dotarem), 1, Gd-HPDO3A (ProHance),
2, and Gd-DO3AB (Gadovist), 3. These commercial contrast
agents exemplify the utility of tetraazamacrocycles, specifically
cyclen (1,4,7,10-tetraazacyclododecane), in their application to
MRI. In our continuing efforts to improve MRI technology,
* Author to whom correspondence may be sent. E-mail: vjacques@
epixpharma.com.
† Magen Biosciences, 100 Beaver St., Suite 101, Waltham, MA 02453.
‡ Ziopharm Oncology, 1 First Avenue, Building #34, Navy Yard Plaza,
Boston, MA 02129.
§ Ironwood Pharmaceuticals, 320 Bent St., Cambridge, MA 02141.
⊥ Johnson-Matthey Pharma Services.
¶ Sirtris, A GSK Company, 200 Technology Drive, Cambridge, MA 02139.
(1) During the preparation of this manuscript, a patent was submitted and
issued covering the chemistry described in this publication: (a) Amedio,
J. C.; Caravan, P. D.; Jacques, V.; Zhou, K. L.; Levy, S.; Kalog-
eropoulos, S.; Greenfield, M. WO2005/001415 (AN 2005 14659),
2005.
(3) (a) Eisenweiner, K.; Powell, P.; Ma¨cke, H. R. Bioorg. Med. Chem.
Lett. 2000, 2133. (b) Ma¨cke, H. R.; Eisenweiner, K.; Powell, P. Patent
WO 02/024235 A3 (AN 2002 240598), 2002.
(4) Kang, S. I.; Ranganathan, R. S.; Emswiler, J. E.; Kumar, K.;
Gougoutas, J. Z.; Malley, M. F.; Tweedle, M. F. Inorg. Chem. 1993,
32, 2912.
(2) Jacques, V.; Desreux, J. F. Synthesis of MRI Contrast Agents II.
Macrocyclic Ligands. In The Chemistry of Contrast Agents in Medical
Magnetic Resonance Imaging; Merbach, A. E., Toth, E., Eds.; John
Wiley and Sons: New York, NY, 2001; pp 157-191.
10.1021/op8002932 CCC: $40.75  2009 American Chemical Society
Published on Web 03/20/2009
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
535

Figure 1. Approved, commercially available MRI contrast agents based on 1,4,7,10-tetraazacyclododecane ligands, 1-3, and structure
of protected tetraazamacrocyclic ligand 4.
Scheme 1. Initial discovery synthesis of ((R)-DOTAGA), 4, using a literature procedure and conversion to
r-methylbenzylamide 5
Scheme 2. Synthesis of DO3MA as described by Kang and co-workers4
Based on the above reasoning, it was expected that alkylation
of cyclen with the triflate ester derived from (2)-(S)-1-tert-butyl-
2-hydroxy-5-benzyl glutarate 9, (Scheme 3) would result in 4
with significantly increased optical purity, as compared with 4
prepared using the literature route. The improved asymmetric
synthesis of 4 is depicted in Scheme 3. The commercially
available (S)-5-oxotetrahydrofuran-2-carboxylic acid 6 was
converted to its acid chloride 7 using Vilsmeier reagent
generated in situ from oxalyl chloride and catalytic DMF.
Treatment of 7 with tert-butanol and 2,6-lutidine5 gave the crude
ester, 8, which was purified by gradient elution from a silica
plug (86% yield). The lactone was then hydrolyzed using one
equivalent of 1 N potassium hydroxide, and the resulting
carboxylate salt was alkylated with benzyl bromide in DMF,6
giving the crude orthogonally protected diester, which was
purified using column chromatography to give 9 in 82% yield.
The triflate ester7 of 9 was then generated in situ, at -40 °C,
and added to a cooled solution of cyclen in dichloromethane,
giving crude 10 after aqueous workup. Previous results indicated
that the triflate was not sufficiently stable to be isolated. Fully
alkylated cyclen 11 was then prepared (48% yield for three
steps) by reacting crude 10 with tert-butyl bromoacetate.
Hydrogenation to remove the benzyl protecting group gave 4
in 95% yield. A sample of 4 prepared via this route was
converted to 5, which was analyzed using proton NMR.
(5) Schmidt, U.; Braun, C.; Sutoris, H. Synthesis 1996, 223.
(6) (a) Lactone hydrolysis: Shin, I.; Lee, M.; Lee, M.; Jung, M.; Lee, W.;
Yoon, J. J. Org. Chem, 2000, 65, 7667. (b) Benzyl bromide alkylation:
Hoffmann, M.; Wasielewski, C. Roczniki Chem. 1975, 49, 151.
(7) The motivation for the generation and use of the triflate ester of 11
was based on results reported in ref 2.
(8) (a) Doolittle, R. E.; Heath, R. R. J. Org. Chem. 1984, 49, 5041. (b)
Vigneron, J. P.; Meric, R.; Larcheveque, M.; Debal, A.; Lallemend,
J. Y.; Kunesch, G.; Zagatti, P.; Gallois, M. Tetrahedron 1984, 40,
3521. (c) Gringore, O. H.; Rouessac, F. P. Organic Syntheses 1985,
63, 121. (d) Okabe, M.; Sun, R. C.; Tam, S. Y. K.; Todaro, L. J.;
Coffen, D. L. J. Org. Chem. 1988, 53, 4780. (e) Zamzow, M.; Hocker,
H. Macromol. Chem. Phys. 1994, 195, 2381. (f) Cai, X.; Chorghade,
M. S.; Fura, A.; Grewal, G. S.; Jauregi, K. A.; Lounsbury, H. A.;
Scannel, R. T.; Yeh, C. G.; Young, M. A.; Yu, S.; Guo, L.; Moriarty,
R. M.; Penmasta, R.; Rao, M. S.; Singhal, R. K.; Song, Z.; Staszewski,
J. P.; Tuladhar, S. M.; Yang, S. Org. Process Res. DeV. 1999, 3, 73.
(9) Lactone carboxylic acid 6 was chosen as the optically enriched starting
material for 9. It was necessary to determine whether stereochemical
integrity was maintained throughout the synthesis of 4, starting from
6. Analysis of 4 was done using HPLC, with a chiral stationary phase.
The chromatographic resolution of 4 was not sufficient to obtain
baseline resolution of the peaks corresponding to the enantiomers of
4 and allow a precise determination of optical purity. An analysis of
a diastereomeric derivative of 4 prepared via the triflate ester of 9
using proton NMR was then undertaken. The diastereomer (R,R)-5
was prepared by reacting 4 with (R)-(+)-R-methyl benzylamine in
the presence of DIC and HOBt, as shown in Scheme 1 (see
Experimental Section).
536
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

Scheme 3. Second-generation asymmetric synthesis of 4
Scheme 4. Optimized multigram asymmetric synthesis of 4
Preparation of 4 via the triflate route gave product having ee )
97 ( 2% (HPLC).9 Thus, use of the triflate ester derived from
9 to alkylate cyclen resulted in the first highly stereoselective
asymmetric synthesis of 4.
Process Development and Optimization of the Synthesis
of 4. The optimized laboratory multigram asymmetric synthesis
of 4 is shown in Scheme 4. Numerous issues were identified
for optimization of the synthesis of 4 for scale-up and eventual
cGMP manufacture, including sourcing or production of 6,
generation of highly optically pure 8, and improvement of the
alkylation of cyclen with an orthogonally protected electrophilic
(S)-2-hydroxyglutarate equivalent to produce 10.
Conversion of (L)-Glutamic Acid to 6 and 8. Lactone acid
6 was identified as a very expensive starting material. With the
knowledge that there was ample literature precedent for the
preparation of this compound,8 it was decided that 6 would be
prepared from (L)-glutamic acid, via diazotization chemistry.
Using a reported procedure8e (Scheme 4), (L)-glutamic acid was
converted to 6 by treatment with sodium nitrite in the presence
of excess HCl in 48% yield, based on 100 g of starting material
in the initial experiment, thus demonstrating the potential for
large-scale preparation. Practical difficulties with this chemistry
were also identified. The aqueous reaction medium and the
water solubility of the product necessitated the distillation of a
large volume of water in order to isolate the product. Addition-
ally, the water solubility of 6 made crystallization (initially from
ether) unpredictable and very slow (24-48 h at -20 °C).
Significant improvements to the synthesis of 6 occurred after
an analytical method was developed to assess the enantiomeric
excess (ee) of samples of crude and purified 6 (chiral GC). It
was discovered that crude 6 could be reproducibly prepared
with ee g 93%. The ability to prepare and carry forward 6 in
crude form simplified the workup and isolation and improved
the contained yield. The stereoselective conversion of (L)-
glutamic acid to 6 depended critically on the rate of addition
of the aqueous sodium nitrite solution to the acidic (L)-glutamic
acid solution, and on maintaining an internal temperature range
(0-5 °C) of the reaction mixture during this addition. An
acceptable rate of addition was 1.5-3.0 mL/min (a total addition
time of 3-5 h on a scale of hundreds of grams). If the sodium
nitrite solution was added too rapidly (e.g., > 100 mL/h on a
scale of 200 g starting (L)-glutamate), the quality and ee of the
product decreased significantly, regardless of the temperature
at which the reaction mixture was maintained. A significant
amount of nitric oxide was produced, even at temperatures
within the recommended range of 0-5 °C, and very low yields
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
537

Table 1. Effect of base, temperature and solvent on
alkylation of cyclen with 12
HPLC area %
conversion to 10
temperature, °C in 24 h (48 h)
Figure 2. 2-(S)-2-Hydroxyglutaric acid, dibenzyl ester, the
entry solvent
base
CH2Cl2 K2CO3
DMF K2CO3
major impurity present in crude 9.
1
2
3
4
5
6
7
8
9
25
25
12
29
45 (67)
37 (57)
30 (49)
30 (47)
80
(20-30%) were obtained. Crude 6 was carried forward to the
acid chloride intermediate 7, at which point it was purified by
vacuum distillation.8a A temperature range of 0-5 °C was
acceptable for the reaction of 7 with tert-butanol to form the
tert-butyl ester. After conversion to 8, the crude material was
purified by filtration through a plug containing a layer each of
activated carbon, flash silica, and Celite, eluting with ethyl
acetate. The filtrate was then concentrated, and the crude solid
was recrystallized from a mixture of ethyl acetate and hexanes.
This purification procedure resulted in reproducible production
of 8 in isolated yields of 55-60%, having chemical purity
g99% and ee g 99%. This result was key to the development
of a practical kilogram-scale synthesis of 4, since stereochem-
istry is fixed at the formation of 6 and optical purity is a key
specification for 4. It was subsequently found that the recrys-
tallization step was responsible for most of the enhancement
in the optical purity of 8, as compared with that of 6, since use
of the crude acid chloride 7 to prepare 8 gave purified material
having ee identical to that obtained from distilled acid chloride.
Conversion of 8 to 9. Operationally, the chemistry used to
convert 8 to 9 was quite straightforward. The major question
concerned identification of reaction end points (i.e., reaction
rate and the introduction of an in-process control), side reactions,
and optimization of the chromatographic purification of 9. An
HPLC study indicated that the hydrolysis of the lactone was
not complete after 1 h at ambient temperature, the time
described for the reaction in the discovery synthesis and in
literature reports.6a Since the workup consisted of removal of
water in Vacuo at elevated temperature, it is likely that the
reaction was completed during concentration of the reaction
mixture. In situ monitoring of the hydrolysis by infrared
spectroscopy showed that it was complete after 2 h at 50 °C.
During this interval, no hydrolysis of the tert-butyl ester was
apparent. However, the major impurity from the subsequent
alkylation reaction of the potassium carboxylate intermediate
13 was a dibenzylated hydroxy diester 14 (Figure 2). The above
result indicates that formation of the dicarboxylate was occurring
during the concentration of the reaction mixture in Vacuo at
elevated temperature, or that the tert-butyl ester was transes-
terified to the benzyl ester under the alkylation conditions.
One operational difficulty was the solubility of 13, product
of alkaline hydrolysis of 8. Compound 13 was obtained as a
crystalline solid after the removal of water from the lactone
hydrolysis reaction mixture. The potassium carboxylate salt was
not soluble in DMF, the solvent used for the subsequent
reaction. After benzyl bromide was added to DMF, 13 adhered
to the bottom and walls of the reaction vessel. In the presence
of benzyl bromide, the carboxylate salt dissolved as it reacted.
This created problems on the kilogram scale, since the reaction
vessel was not identical to the flask used to concentrate the
reaction mixture, and there was not a suitable organic solvent
to transfer the 13 to a reaction vessel. It was unlikely that the
presence of a small amount of water would affect the outcome
CH3CN K2CO3
CH3CN i-Pr2NEt
CH3CN Et3N
25
25
25
CH3CN 2,6-lutidine
25
DMF
CH3CN K2CO3
CH3CN K2CO3
K2CO3
50
50
95
72
100
of the alkylation reaction, so 13 was dissolved using a small
amount of water and transferred to the reaction vessel, then
diluted with DMF prior to the addition of benzyl bromide. The
results for the reaction containing water were comparable to
the results obtained for the original reaction run in dry DMF.
On further scale-up, the hydrolysis reaction mixture was
concentrated to leave an aqueous solution of 13, which was
diluted with DMF for the subsequent reaction.
Conversion of 9 to 10. The in situ generation of the triflate
ester of 9 and its conversion to monoalkylated cyclen 10 were
identified as key steps at the initial scale-up stage. There were
a number of drawbacks to using the triflate: (a) the stability of
the triflate was unknown, but probably marginal, (b) the triflate
was generated in situ and added to cyclen, thus affording little
control over the quality and purity of the triflate used, and (c)
the reaction was run at -40 °C, and this is not generally a
desirable temperature for scale-up or pilot-plant production. In
order to avoid these issues, an alternative alkylating agent,
mesylate 12, was studied.
Mesylate esters are less reactive than the corresponding
triflates. By one estimate, a triflate ester is 40,000 times more
reactive than its corresponding mesylate ester.10 As expected,
the reaction of 12 with cyclen was slow at ambient temperature
using three different solvents and four different bases (Table
1). As indicated in Table 1, the alkylation reaction proceeded
most rapidly in acetonitrile.
As shown in Table 1, the conversion of 12 to 10 at ambient
temperature was low under all conditions examined. The set
of conditions giving the highest conversion of 12 to 10 at
ambient temperature was entry 3, using acetonitrile as the
solvent and potassium carbonate as the base. The alkylation
reaction run at 50 °C in acetonitrile in the presence of potassium
carbonate (entry 8, Table 1), resulted in 95% conversion of 12
to 10 after 24 h. Running the reaction at 100 °C in acetonitrile
using potassium carbonate as the base resulted in lower
conversion of 12 to 10, due to decomposition.
It is clear that alkylation of cyclen with the mesylate 12 gives
superior results, in comparison with the alkylation of cyclen
with the triflate. The use of the mesylate is advantageous in
that (1) it is stable, up to approximately 50 °C, (2) it can be
isolated in crude (yet highly pure) form via aqueous workup,
and (3) its reaction with cyclen is controllable, giving 11 after
(10) (a) Crossland, R. K.; Wells, W. E.; Shiner, V. J., Jr. J. Am. Chem.
Soc. 1971, 93, 4217. (b) Streitweiser, A., Jr.; Wilkins, C. L.; Kiehlman,
E. J. Am. Chem. Soc. 1968, 90, 1598.
538
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

Scale-up of the palladium-catalyzed hydrogenation to cleave
the benzyl protecting group was straightforward. This hydro-
genation was performed on up to a 50-g scale in a Parr shaker
apparatus. Prewetted Pd-C was used in order to minimize fire
hazard during the transfer of catalyst. It was crucial that residual
methanol in 4 was minimal, since the next step was the
conversion to various activated esters. Coevaporation of 4 with
acetonitrile was necessary to ensure that any remaining methanol
was removed (confirmed by proton NMR). The yield of 4 was
85-90%, in 95% purity by HPLC area. The stereochemical
integrity of the chiral center in the side chain was maintained
with minimal or no racemization, as determined by HPLC of
the derivative 5 (96.8 ( 2.0% ee, based on HPLC analysis).
Preparation for cGMP Kilogram-Scale Manufacture of
4. The reduction to practice of the above process at an external
vendor confirmed that the most problematic aspects were related
to the conversion of (L)-glutamic acid to 6. The yield of 8 was
diminished compared to that originally observed, most likely
due to prolonged heating of the reaction mixture from the
formation of 6 during the removal of water and HCl, and also
to decomposition during vacuum distillation of crude 7. Since
analysis of the process did establish that virtually all of the
enhancement in ee observed in the conversion of 6 to isolated
8 was attributable to the purification using a carbon/silica/Celite
plug, followed by recrystallization, the vacuum distillation of
7 was not necessary in order to obtain 8 with acceptable purity
and ee. Subsequent to the reduction to practice, crude 6 was
converted to crude 7, which was used directly in the preparation
of 8. Intermediate 8 was then purified and isolated as described
above. Aside from lower yields, the remainder of the reduction
to practice synthesis of 4 proceeded as described for the
multigram asymmetric synthesis of 4. Intermediate 9 was
converted to 10 using the mesylate chemistry previously
described. The loss of significant amounts of 10 subsequent to
aqueous washes during workup became apparent during the
early scale-up of this chemistry. The volumes of water and brine
used, relative to the large volume of ethyl acetate of the workup
solution, were determined by screening of workup conditions
and monitoring for the presence of residual cyclen using LC/
MS and HPLC. Monoalkylated cyclen 10 prepared in the
reduction to practice had a purity of 77% with 0.4% residual
cyclen. Alkylation of cyclen with 12, followed by alkylation
of 10 with tert-butylbromoacetate provided 11 in 56% yield,
based on 9. Hydrogenation of 11 to give 4 was performed with
Pd-C in a mixture of water and ethanol. The yield was 92%,
based on mass. The product was 97.6% pure and had ee )
97.4%.
Figure 3. Structures of the 1,4- and 1,7- bis-alkylated cyclen
impurities.
alkylation with tert-butyl bromoacetate in 84% yield (10-g
scale), considerably improved over the yield of 11 (48%)
obtained using the triflate and in identical quality. The significant
improvement in the yield of 11 using 12 over that obtained
using the triflate of 9 is most likely attributable to the increase
in stability of the alkylating agent, and thus the actual quantity
of alkylating agent available to alkylate cyclen.
The monoalkylation of cyclen was carried out by adding
the mesylate 12 to a suspension of 2 equiv of cyclen in
acetonitrile preheated to 50 °C in the presence of 1 equiv of
potassium carbonate. Excess cyclen was necessary in order for
the reaction to proceed optimally. Less than 2 equiv of cyclen
resulted in the formation of significantly more bis-alkylated
cyclen impurities 15 and 16 (Figure 3). The amounts of 15 and
16 were less than 5% when g2 equiv of cyclen were used.
In the aqueous workup to remove the residual cyclen, it was
discovered that consecutive washes with a small amount of
water and brine in a ratio of 1:5 and 1:15 (v/v), respectively,
relative to the volume of the ethyl acetate solution of the crude
10, removed all detectable residual cyclen and its mesylate salt
to the limits of LC/MS detection (<0.1% of peak area of 10).
The removal of excess cyclen at this point was crucial to the
quality of 11 obtained, since the side product, DOTA tetra-
tert-butyl ester, was impossible to separate from a mixture with
11. This reaction sequence was carried out on a scale of 10-100
g. No differences in extent of reaction or impurity profile were
observed.
Conversion of 10 to 11. The use of mesylate 12 as the
alkylating agent for cyclen allowed more control over the quality
of 10 produced, which in turn resulted in higher yields of 11,
in comparison to the triflate chemistry. Three equivalents of
tert-butyl bromoacetate were used, based on the potency of 10.
If the reaction was run with excess tert-butyl bromoacetate (>3
equivalents), overalkylation gave a quaternary ammonium salt.
The overalkylated impurity could be converted back to the
desired product 11 by heating an ethyl acetate solution at reflux.
The limited solubility of 11 in ethyl acetate allowed us to
isolate and purify the crude material by trituration. Using a ratio
of ethyl acetate to crude starting 10 of 5:1 (v/w), 11 with purity
>95% (HPLC area) was consistently obtained. On g100 g scale,
the yield of 11 (based on 9) was 52-63% for three chemical
transformations.
Scale-Up and Production of 6, 8, and 9. (L)-Glutamic acid
was converted to 9 on a 1-2-kg scale, in preparation for
production of starting material for the kg-scale cGMP manu-
facture of 4. This intermediate scale-up indicated that the time-
consuming removal of water to isolate crude 6 remained an
issue. Scale-up of the conversion of (L)-glutamic acid to 6 to 5
kg revealed a hazard that was not apparent at the lower kg scale.
Concentration of the reaction mixture in Vacuo at 30 °C
produced a large exotherm (a temperature increase of 40-50
°C) once most of the water had been removed, accompanied
by vigorous gas evolution. Additionally, the use of ethyl acetate
Conversion of 11 to 4. The hydrogenation of 11 was initially
done in methanol, at 50 psi in the presence of 10% Pd-C by
mass to give the free carboxylic acid 4 in 98% yield. Experi-
ments were run at either 20 or 40 psi hydrogen pressure, and
both were determined to be complete after less than 1 h.
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
539

as the extraction solvent for the crude concentrated residue of
6 resulted in transesterification of the open-chain 2-hydroxy-
glutaric acid present in the crude mixture, due to prolonged
contact with residual HCl. Based on mass spectral data, the
compound most likely to be responsible for the exotherm was
the nitrite ester of 2-hydroxyglutaric acid. Decomposition of
this compound would form 6 in the presence of acid, and gas
evolution (presumably nitric oxide) would occur. An attempt
was made to dissipate the exotherm by feeding portions of the
partially concentrated reaction to boiling toluene and then
removing the remaining water as an azeotrope. This treatment
failed to decompose the energetic material present in the
mixture, as evidenced by the continued presence of a vigorous
exotherm in a sample analyzed by differential scanning calo-
rimetry (DSC). Unfortunately, this heat treatment also resulted
in the decomposition of a significant amount of crude 6 and in
an unacceptable loss of optical purity (ee ) 72% after treatment
with boiling toluene). For the remaining kilogram-scale batches
of 6, the reaction mixture was concentrated in portions of not
more than 8-10 L apiece, and each residue was set aside for
later pooling in order to avoid thermal decomposition of material
due to prolonged exposure of the accumulated product to heat.
Crude 6 (still containing significant amounts of water, even after
coevaporation with toluene) was then extracted with dichlo-
romethane, dried with sodium sulfate, decanted and carried
forward to the preparation of the acid chloride 7. The purifica-
tion of crude 8 was modified to accommodate the increased
scale of the operations by first filtering through a silica plug
and then passing the filtrate through an in-line cartridge of
activated carbon, followed by partial concentration of the ethyl
acetate solution and dilution with heptane to induce crystal-
lization. The above modifications resulted in the ability to
prepare multikilogram quantities of 8 in g99% ee and g99%
purity. The yield for the conversion of (L)-glutamic acid to 8
on a multikilogram scale was 42%, compared with 56% for
the same series of conversions performed on a 200-g scale. The
scale-up of the conversion of 8 to 9 proceeded with essentially
no issues. The reactions were run at 2.5-3.0-kg scale, giving
a 54% yield of 9 based on assay. Purification of ∼1 kg of crude
9 was done by application to a 6-in. column containing 15 kg
flash silica, and eluting with a gradient starting with pure
heptane, stepping up to a maximum of 25% ethyl acetate/
heptane. Therefore, six column chromatography runs were
required to purify all of the intermediate 9 produced in the
cGMP campaign.
were released for further processing, since it was clear that the
dialkylated cyclen impurity and assay specifications for 10 had
been set too tightly. The yields of 10 for these lots were
72-73%, based on assay. The kilogram-scale preparation of
11 from 10 proceeded smoothly. The yield of 11 for both lots
was 64%, giving a total of 5.6 kg of material.
Kilogram-scale conversion of 11 to 4 using catalytic
hydrogenation proceeded without incident. The hydrogenation
was carried out in a 5-gal stainless steel reactor, requiring four
batches of ∼1.4 kg each of 11. Two lots of 4 were produced,
in 87.6% yield (2.11 kg) and 91.4% yield (2.35 kg), based on
assay. Both lots had ee ) 99.0%, based on HPLC.
Experimental Section
2-(S)-5-Oxotetrahydrofuran-2-carboxylic acid, tert-Butyl
Ester (8). A solution of sodium nitrite (140 g, 2.03 mol) in
water (320 mL) was added over 4 h to a mechanically stirred
mixture of L-glutamic acid (200 g, 1.36 mol), dioxane (150 mL)
and HCl (280 mL) in water (530 mL), maintained at an internal
temperature of 0-5 °C. On completion of the addition, the
mixture was warmed to room temperature (RT) and stirred for
2 h. Solvents were then removed in Vacuo (50-55 °C and 30
Torr). Caution: While concentrating the aboWe aqueous
reaction mixture on a scale of 5 kg, an exotherm (40-50 °C
temperature increase), accompanied by Wigorous gas eWolu-
tion, was obserWed. In subsequent runs, the reaction mixture
was concentrated in smaller portions (< 10 L of reaction
mixture concentrated at a time), which were combined in
CH2Cl2 before being carried forward to the acid chloride. The
residue was coevaporated with toluene (2 × 500 mL) to remove
additional water, then ethyl acetate (1 L) was added, followed
by sodium sulfate (100 g), and the mixture was stirred for 0.5 h.
The ethyl acetate solution was decanted and filtered, and the
solids were washed and stirred with additional ethyl acetate (1
L). The combined filtrates were concentrated in Vacuo, and the
resulting residue was further dried under high vacuum (1 Torr).
The mass of crude residue containing 6 was 193 g, exceeding
the mass of the theoretical yield by 16 g, attributable to solvent
trapped by the viscous, syrupy residue. A sample of this residue
was subjected to chiral GC analysis, which indicated the optical
purity of this material to be 94% ee. LC/MS: m/z ) 131 (MH+).
Determination of the ee of crude 6 via chiral GC: ∼25 mg
of crude 6 and ∼50 mg of dicyclohexylcarbodiimide (DCC)
were dissolved in CH2Cl2 (∼1 mL), and methanol (2 drops)
was added; the mixture was stirred 1 h, then filtered through a
0.45 µm Nylon syringe filter; injection volume: 1 µL; assayed
on a 2,6-di-O-pentyl-3-propionyl capillary column, 40 mm ×
0.25 mm; temperature 250 °C; split ratio 100:1; run time 20
min; tR ) 12.1 min ((R)-enantiomer), tR ) 13.7 min ((S)-
enantiomer).
Kilogram-Scale Conversion of 9 to 4. Monoalkylated
cyclen 10 was prepared in two batches from two lots of 9. No
difficulties were encountered, but both lots of 10 were found
to be out of specification (the quantity of the dialkylated cyclen
impurity for one lot was 5.5%, while the release specification
for this intermediate was <5.0%; purity of the second lot was
73.5%, while the release specification for purity of 10 was
g75%). Use tests for the conversion to 11 were conducted with
small samples of each nonconforming lot of 10. Both samples
gave 11 having purity >96.0%, exceeding the specification for
11, set at g95.0%. The major impurities were the dialkylated
cyclen products carried over from 10 (2.5-3.0%). Based on
the results of these use tests, both nonconforming lots of 10
Oxalyl chloride (297 mL, 432 g, 3.40 mol) was added to a
solution of crude 6 (assumed ∼177 g, 1.36 mol) and DMF (2.00
mL) in dry dichloromethane (800 mL) cooled to 0-5 °C in an
ice bath. The mixture was warmed to RT after the addition was
complete and stirred 2 h, then concentrated in Vacuo. The
residue was coevaporated with dichloromethane (1 L), was then
transferred under inert atmosphere to a 500 mL distillation flask,
and was subjected to vacuum distillation (short path still head/
540
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

condenser). The acid chloride 7 distilled at 86-96 °C at 1 Torr.
The mass of distillate obtained was 154 g.
2.59-2.46 (m, 2 H), 2.2-2.1 (two m, 2 H), 1.48 (s, 9 H). 13
C
NMR: δ (ppm) ) 173.9, 173.0, 135.9, 128.6, 128.2, 128.19,
82.7, 69.6, 66.3, 29.8, 29.5, 28.0. Anal. Calcd for C16H22O5: C,
65.29; H, 7.53. Found: C, 65.19; H, 7.74.
Acid chloride 7 (154 g, ∼1.04 mol) in dichloromethane (350
mL) was added over a period of 1.5 h to a stirred solution of
tert-butanol (200 mL, 154 g, 2.08 mol) and 2,6-lutidine (145
mL, 133 g, 1.25 mol) in dichloromethane (800 mL) cooled to
0 °C. On completion of the addition, the mixture was warmed
to RT and stirred overnight. The mixture was then poured into
a separatory funnel and washed with water (2 × 800 mL), 10%
aq citric acid (2 × 800 mL), saturated sodium bicarbonate (2
× 800 mL) and brine (2 × 800 mL), then dried (sodium sulfate),
decanted and concentrated to a dark brown residue, which
crystallized on standing. The crude residue was dissolved in
ethyl acetate (150 mL) and applied to a plug composed of a
wetted bed of Celite (bottom layer, 150 g), flash silica (middle
layer, 150 g) and activated carbon (top layer, Darco, 100 mesh,
150 g). The loaded plug was eluted with ethyl acetate (3 L),
and the filtrate was concentrated to a residue, which was
crystallized by adding ethyl acetate (200 mL) followed by
hexanes (600 mL), to give 134 g of 8 (53%) after drying at
ambient temperature at 1 Torr for 3 h. A second crop of crystals
(7.00 g, 3%) was obtained by concentrating to a residue and
precipitating with hexanes (300 mL), giving a yield of 56% 8,
based on (L)-glutamic acid. The purity of the material was
determined to be 99.7% by HPLC area integration. This material
(2)-(R)-2-(1,4,7,10-Tetraazacyclododec-1-yl-pentanedio-
ic Acid, 1-tert-Butyl-5-benzyl Ester (10). Methanesulfonyl
chloride (38.9 mL, 62.2 g, 0.503 mol) was added to a stirred
mixture of 9 (132 g, 0.449 mol) and Et3N (70.1 mL, 60.0 g,
0.594 mol) in CH2Cl2 (1.5 L) cooled to 0-5 °C in an ice bath.
After the addition was complete, the mixture was warmed to
RT, stirred 0.5 h and checked by HPLC for completeness. After
1 h, water (500 mL) was added, the organic phase was separated
and washed with brine (500 mL) then dried (Na2SO4), decanted
and concentrated in Vacuo to give 12 (163 g), sufficiently pure
to be used in the next step. 1H NMR (CDCl3): δ (ppm) ) 7.35
(m, 5 H), 5.13 (s, 2 H), 4.97 (dd, 1 H), 3.11 (s, 3 H), 2.57-2.14
(m, 4 H), 1.48 (s, 9 H).
The mesylate 12 (163 g, 0.431 mol), as a solution in CH3CN
(800 mL), was added over a period of 45 min to a stirred
mixture of cyclen (153 g, 0.888 mol) and potassium carbonate
(61.3 g, 0.432 mol) in CH3CN (2.4 L) preheated to 50 °C,
monitoring by HPLC for completeness. After 19 h, the reaction
mixture was cooled to RT and filtered to remove potassium
salts and cyclen mesylate. The filter cake was then washed with
additional CH3CN (2 × 150 mL), and the filtrate was
concentrated to a red oil in Vacuo and redissolved in ethyl
acetate (4.8 L). The organic solution was washed with water
(300 mL) and brine (100 mL). (If cyclen is detected in the
organic phase after this initial set of washes (LC/MS, m/z )
173, MH+ for cyclen), washing needs to be repeated.) The
organic solution was then separated and dried (Na2SO4) and
then decanted and concentrated in Vacuo (water bath temper-
ature <35 °C) to give 174 g of crude product (92% pure by
1
was determined to have ee g 99% by chiral GC. H NMR
(CDCl3): δ (ppm) ) 4.8 (dd, 1 H), 2.8-2.2 (m, 4 H), 1.5 (s, 9
H). 13C NMR: δ (ppm) ) 176.3, 169.0, 83.0, 76.2, 27.9, 26.8,
25.9. Anal. Calcd for C9H14O4: C, 58.05; H, 7.58. Found: C,
58.31; H, 7.72.
Determination of ee via chiral GC: injection volume: 1 µL
of a 15 mg/mL solution of 8 in CH2Cl2; assayed on a 2,6-di-
O-pentyl-3-propionyl capillary column, 40 mm × 0.25 mm;
temperature 250 °C; split ratio 100:1; run time 60 min. tR )
50.40 min ((R)-enantiomer), tR ) 52.45 min ((S)-enantiomer).
2-(S)-2-Hydroxypentanedioic Acid, 1-tert-Butyl-5-benzyl
Ester (9). A solution of 1 N KOH (357 mL, 0.357 mol) was
added in a single portion to a stirred solution of 8 (66.4 g, 0.357
mol) in THF (250 mL). The internal temperature rose to 37
°C. The mixture was heated and stirred at 40 °C for 2 h, then
concentrated to a solid in Vacuo (55 °C, <30 Torr), and dried
under high vacuum (ambient temperature, < 5 Torr) overnight.
The solid residue was suspended and stirred in DMF (650 mL),
and benzyl bromide (44.6 mL, 64.2 g, 0.375 mol) was added.
After stirring 8 h, the mixture was poured into ice water (1.5
L), then extracted with ethyl acetate (2 × 500 mL), dried
(Na2SO4) and concentrated in Vacuo. The weight of the crude
residue after removing ethyl acetate was 150 g, indicating the
presence of approximately 45 g of DMF and excess benzyl
bromide left to be removed. The remainder of the DMF was
removed in Vacuo using rotary evaporation with a high vacuum
(<1 Torr) pump, leaving 101 g of crude 9. This material was
applied to a plug of flash silica (8 × 25 cm) and eluted with
hexanes (1 L), 1:9 ethyl acetate/hexanes (v/v) (1 L) and 15:85
ethyl acetate/hexanes (v/v) (2 L). Fractions containing the
product were pooled and concentrated to give 79.7 g (76%) of
1
HPLC area integration). H NMR (CDCl3): δ (ppm) ) 7.36
(m, 5 H), 5.8 (s, 2 H), 3.24 (dd, 1 H), 2.65-2.35 (m, 20 H),
1.39 (s, 9 H). LC/MS (electrospray): m/z ) 449 (MH+).
2-(R)-2-(4,7,10-Tris tert-Butyloxycarbonylmethyl-1,4,7,10-
tetraazacyclododec-1-yl)-pentanedioic Acid-1-tert-Butyl-5-
benzyl Ester (11). Crude 10 (prepared from mesylate 12, 174 g,
∼0.388 mol) and potassium carbonate (438 g, 3.18 mol) were
dissolved/suspended in dry CH3CN (2 L), then a solution of
tert-butyl bromoacetate (173.0 mL, 1.17 mol) in CH3CN (432
mL) was added over a period of 1 h. HPLC monitoring
indicated that the reaction was incomplete. An additional portion
of tert-butylbromoacetate (21.6 mL, 28.5 g, 0.146 mol) was
then added, and stirring was continued for 17 h. The mixture
was filtered, and the filtrate was concentrated in Vacuo, dissolved
in ethyl acetate (3 L), and washed with water (2 L), saturated
sodium bicarbonate (1 L), and brine (1 L). After drying
(Na2SO4), the solution was decanted and concentrated to a
volume of 0.75-1 L in Vacuo, at which point a fine, white
precipitate formed. The precipitate was isolated by vacuum
filtration, and the filter cake was washed with ethyl acetate (3
× 300 mL), then dried under high vacuum (<5 Torr) to remove
traces of ethyl acetate to give 223 g (63.0% for three steps,
formation of 12, 10 and 11) of fully alkylated cyclen 11. A
small sample of this material was repurified using preparative
HPLC prior to submission for elemental analysis. The aqueous
1
9 as a light-yellow tinted oil. H NMR (CDCl3): δ (ppm) )
7.35 (m, 5 H), 5.13 (s, 2 H), 4.1 (m, 1 H), 2.88 (d, 1 H),
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
541

eluent for preparative HPLC was 0.1% trifluoroacetic acid
(TFA) in water, resulting in formation of the mono-TFA salt
in situ. 1H NMR (CDCl3): δ (ppm) ) 7.34 (m, 5 H), 5.08 (d,
2 H), 3.39-2.00 (m, 27 H), 1.43 (four s, 36 H). 13C NMR: δ
(ppm) ) 175.0, 174.7, 173.0, 172.84, 172.81, 135.6, 128.6,
128.4, 128.3, 82.6, 81.94, 81.90, 81.88, 66.4, 59.9, 55.9, 55.8,
55.5, 52.7, 52.4, 48.6, 48.5, 48.1, 47.2, 44.3, 32.5, 27.9, 27.8,
27.7, 19.3. Anal. Calcd for C42H70N4O10 ·TFA: C, 56.70; H,
8.00; N, 6.01. Found: C, 56.68; H, 7.96; N, 5.99.
48.4, 48.0, 47.0, 46.97, 46.9, 35.1,27.9, 27.87, 27.8, 22.8, 21.9.
LC/MS (electrospray): m/z ) 805 (MH+). Anal. Calcd for
C43H73N5O9 ·H2O: C, 62.9; H, 9.14; N, 8.52. Found: C, 63.1;
H, 8.90; N, 8.64.
HPLC analysis of the stereochemical composition of 5: 150
mm × 4.6 mm CN column; temperature 30 °C; eluent 75:25
1.0 M K3PO4, pH ) 2.5/acetonitrile; flow rate 1.2 mL/min;
UV 220 nm; tR ) 18.7 min ((R,R)-5), tR) 20.9 min ((S,R)-5,
prepared as a standard).
2-(R)-2-(4,7,10-Tris-tert-Butyloxycarbonylmethyl-1-4-7-
10-tetraazacyclododec-1-yl)-pentanedioic Acid, 1-tert-Butyl
Ester (4). Palladium on carbon (dry, 3.00 g, ≈10% by mass)
was added as a slurry in water (20 mL) to a methanol solution
of 11 (30.0 g, 38.0 mmol, in 200 mL methanol) in a Parr
pressure shaker bottle. The mixture was subjected to three cycles
of vacuum and hydrogen purge, then pressurized to 25-30 psi
hydrogen and shaken for 2 h. HPLC indicated that the only
major species present was the desired product. After evacuating
the system to remove hydrogen, the bottle was opened, and
Celite (15 g) was added. The slurry was filtered through a
methanol-wet bed of Celite (15 g), and the filtrate was
concentrated (bath temperature <35 °C to a light-yellow syrup
in Vacuo. The syrup was coevaporated with acetonitrile (2 ×
150 mL) in order to azeotrope out residual water and to remove
any remaining methanol (confirmed by proton NMR), giving
21.8 g (82%) of an amorphous off-white solid after drying under
high vacuum. HPLC: 95% pure by area integration. 1H NMR
(CDCl3): δ (ppm) ) 3.61-2.0 (6 m, 27 H), 1.45 (3 s, 36 H).
13C NMR: δ (ppm) ) 175.2, 175.0, 172.8, 172.7, 172.6, 82.3,
82.0, 81.94, 81.89, 60.0, 55.8, 55.7, 55.5, 52.6, 52.5, 48.5, 48.4,
48.0, 47.1, 44.1, 27.85, 27.83, 27.79, 20.1. Anal. Calcd for
C35H64N4O10 ·2.7 TFA: C, 47.22; H, 6.93; N, 5.51. Found: C,
47.31; H, 7.04; N, 5.53.
2-(R)-2-(4,7,10-Tris tert-Butylcarboxymethyl-1,4,7,10-tet-
raazacyclododec-1-yl)-pentanedioic Acid, 1-tert-Butyl Ester,
5-(R)-(+)r-Methylbenzylamide (5). (R)-(+)-R-Methylbenzyl-
amine (129 µL, 121 mg, 1.00 mmol) was added to a stirred
mixture of 4 (701 mg, 1.00 mmol), DIC (155 µL, 126 mg, 1.00
mmol) and HOBT (135 mg, 1.00 mmol) in CH2Cl2 (10 mL).
The mixture was stirred 3 h, then diluted with CH2Cl2 (20 mL),
and washed with water (10 mL) and brine (10 mL), dried
(Na2SO4), decanted, and concentrated to a crystalline residue
(757 mg). A portion (250 mg) of this material was purified by
preparative HPLC (25 mm C4 silica, A: 0.1% TFA in water,
B: 0.1% TFA in CH3CN, gradient of 5 to 65% solvent B,
monitoring at 220 nm, tR ) 23 min). 1H NMR (CDCl3): δ (ppm)
) 8.52 (d, 1 H), 7.47-7.10 (m, 5 H), 5.03 (m, 1 H), 3.50-1.70
(m, 26 H), 1.41 (m, 39 H). 13C NMR: δ (ppm) ) 172.9, 172.8,
172.7, 172.5, 172.1, 145.0, 128.3, 126.6, 126.4, 82.1, 82.04,
82.00, 81.9, 60.6, 55.8, 55.5, 52.7, 52.67, 52.5, 49.2, 49.1, 48.5,
Proton NMR Determination of Optical Purity of 5: In
order to prepare (S,R)-5, (S)-4 was synthesized from the
R-enantiomer of 6 (97% ee, purchased from Aldrich) using the
synthetic route used to prepare 4 (Scheme 4). Proton NMR
spectra of crude (R,R)-5 and (S,R)-5 were recorded in CDCl3.
The amide NH signal of 5 (a doublet at approximately 8 ppm,
J ) 7 Hz) showed a difference in chemical shift for the (R,R)-
and (S,R)-diastereomers. Two sets of amide NH peaks were
present in the proton NMR spectrum of a sample containing
both diastereomers of 5. Subsequent to the proton NMR analysis
of optical purity, an HPLC method was developed which
provided accurate analysis of the stereochemical composition
of 5 prepared using the triflate chemistry. The de purity of 5
was determined to be 97 ( 2%, comparable to the ee purity of
the commercial 6 used as the starting material.
Conclusions
A multigram asymmetric synthesis of the tetraazamac-
rocyclic ligand precursor 4 was developed. Stereochem-
istry of 4 originated from (L)-glutamic acid, and stereo-
chemical integrity was maintained throughout the synthesis.
The chemistry developed was successfully extended to the
kilogram-scale manufacture of 4, giving the product in
high chemical (95-99%) and optical (ee ) 97 ( 2%)
purity. The process developed for the preparation of 4
represents a significant improvement over the existing
literature method, as regards scalability, economy, opera-
tional simplicity and stereoselectivity. This process has
been applied to the multikilogram-scale cGMP manufac-
ture of 4.
Supporting Information Available
General experimental details and characterization data for
6, 8, 9, 11, and 4, as well as chiral GC data supporting the ee
reported for 6 and 8 and HPLC data supporting the de reported
for 5 and the ee reported for 4. This material is available free
of charge via the Internet at http://pubs.acs.org.
Received for review November 17, 2008.
OP8002932
542
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development