Process research and development for a tetrazole-based growth hormone secretagogue (GHS) pharmaceutical development candidate

Article abstract of DOI:10.1021/jo9003508

(Chemical Equation Presented) BMS-317180 (1) is a potent, orally active agonist of the human growth hormone secretagogue (GHS) receptor. This manuscript details the process research and development efforts that enabled the synthesis of the phosphate salt of 1 on a multi-kilogram scale. Key considerations in the development of this process focused on safe execution and the requirement for telescoped synthetic transformations (i.e., without isolation of intermediate products) to contend with a lack of suitably crystalline products.

Full text of DOI:10.1021/jo9003508

Process Research and Development for a Tetrazole-Based Growth
Hormone Secretagogue (GHS) Pharmaceutical Development
Candidate
Akin H. Davulcu,* Douglas D. McLeod, Jun Li, Kishta Katipally, Adam Littke,
Wendel Doubleday, Zhongmin Xu, Cary W. McConlogue, Chiajen J. Lai, Margaret Gleeson,
Mark Schwinden, and Rodney L. Parsons, Jr.*
Process Research and DeVelopment, Bristol-Myers Squibb Company, One Squibb DriVe,
New Brunswick, New Jersey 08903
akin.daVulcu@bms.com; rodney.parsons@bms.com
ReceiVed February 20, 2009
BMS-317180 (1) is a potent, orally active agonist of the human growth hormone secretagogue (GHS)
receptor. This manuscript details the process research and development efforts that enabled the synthesis
of the phosphate salt of 1 on a multi-kilogram scale. Key considerations in the development of this
process focused on safe execution and the requirement for telescoped synthetic transformations (i.e.,
without isolation of intermediate products) to contend with a lack of suitably crystalline products.
Introduction
that demonstrate the beneficial effects of hormone replacement
therapy.2 However, administration of recombinant hGH has
significant disadvantages. First, hGH therapy is accompanied
by a number of serious side effects, including edema, arthralgia,
and carpal tunnel syndrome. Second, intravenous (IV) dosing
at a frequency of 3-5 times per week presents compliance issues
among the patient population. Lastly, hGH therapy is very
expensive.3 The identification of the growth hormone secreta-
gogue (GHS) receptor triggered a paradigm shift in this field
by advancing the notion that a small-molecule GHS agonist
might restore normal pulsatile GH secretion by the pituitary
gland while avoiding the side effects associated with IV
administration of hGH.4 This advance resulted in the identifica-
tion of several small-molecule GHS agonists, including BMS-
317180 (1), a potent, orally effective agent that has been
advanced into clinical development by the Bristol-Myers Squibb
Company.
Over the next 25 years, the worldwide population of
individuals over 65 years of age is expected to double, resulting
in an elderly population of ca. 1 billion people.1 By 2050, it is
projected that 10% of the worldwide population will be at least
65 years of age. Current estimates suggest that 30-60% of
elderly individuals suffer from frailty or other disorders (e.g.,
decline in musculoskeletal function and metabolic profile,
interference with normal sleep cycles, and development of
cardiovascular risk factors) associated with age-related functional
decline. These health issues have a profound negative impact
on these individuals’ ability to carry out many of the activities
of daily life. Contemporary medicine associates many of these
pathological conditions with human growth hormone (hGH)
deficiency that results from neuroendocrine dysregulation; this
notion is corroborated by clinical studies with recombinant hGH
(1) (a) Li, J.; Chen, S. Y.; Li, J. J.; Wang, H.; Hernandez, A. S.; Tao, S.;
Musial, C. M.; Qu, F.; Swartz, S.; Chao, S. T.; Flynn, N.; Murphy, B. J.;
Slusarchyk, D. A.; Seethala, R.; Yan, M.; Sleph, P.; Grover, G.; Smith, M. A.;
Beehler, B.; Giupponi, L.; Dickinson, K. E.; Zhang, H.; Humphreys, W. G.;
Patel, B. P.; Schwinden, M.; Stouch, T.; Cheng, P. T. W.; Biller, S. A.; Ewing,
W. R.; Gordon, D.; Robl, J.A.; and; Tino, J. A. J. Med. Chem. 2007, 50, 5890,
and references cited therein. (b) Strobl, J. S.; Thomas, M. J. Pharmacol. ReV.
1993, 46, 1.
(2) Jørgensen, J. O. L.; Christiansen, J. S. Lancet 1993, 341, 1247.
(3) Lehrman, S. Nature (London) 1993, 364, 179.
(4) (a) Bowers, C. Y.; Reynolds, G. A.; Durham, D.; Barrera, C. M.; Pexxoli,
S. S.; Thorner, M. O. J. Clin. Endocrinol. Metab. 1990, 70, 975. (b) Reichlin,
S.; Saperstain, R.; Jackson, I. M. D.; Boyd, A. E.; Patel, Y. Annu. ReV. Physiol.
1976, 38, 389. (c) Gertz, B. J.; Barrett, J. S.; Eisenhandler, R.; Krups, D. A.;
Wittreich, J. M.; Seibold, J. R.; Schneider, S. H. J. Clin. Endocrinol. Metab.
1993, 77, 1393.
4068 J. Org. Chem. 2009, 74, 4068–4079
10.1021/jo9003508 CCC: $40.75 2009 American Chemical Society
Published on Web 04/24/2009

Tetrazole-Based Growth Hormone Secretagogue
SCHEME 1. Retrosynthetic Analysis for 1
SCHEME 2. Discovery Synthesis of 1•HCl
Results and Discussion
Discovery group had demonstrated that 2 was a stable, crystal-
line intermediate, our synthetic efforts quickly focused on this
target.
Consideration of structure 1 from the process research and
development perspective revealed two primary challenges. Given
the fact that tetrazoles are generally regarded as energy-rich,
thermodynamically unstable compounds, the first challenge
involved the identification of safe, scaleable chemistry for the
synthesis of this heterocyclic core. Many of the known synthetic
methods for tetrazole formation require the use of azide reagents,
and while employing such chemistry on a large scale is feasible,
execution is oftentimes far from routine, generally requiring
stringent safety and engineering controls. As such, avoidance
of azide chemistry was an important practical goal for us. The
second challenge derived from the reactive primary amine and
alcohol functionalities displayed by the molecule. Orchestrating
a sequence that would contend with their intrinsic reactivity
while minimizing reliance on protecting groups would be critical
to defining a practical route. Tactical and strategic issues were
also foreseen with regard to the resident stereogenic center.
Strategically, options employing asymmetric synthesis versus
exploiting the rich “chiral pool” would require thorough
evaluation. Tactically, the chemistry would require fine-tuning
to contend with the stereochemical lability (e.g., racemization
upon exposure to base, as was observed with 1) of the
intermediates encountered throughout the sequence. In planning
the path forward, a route that would intersect crystalline
intermediates (whereupon isolation would provide the op-
portunity for purity control) was regarded as ideal. Toward this
end, retrosynthetic analysis of 1 (Scheme 1) suggested that late-
stage construction of the amide and carbamate linkages would
lead to tetrazole fragment 2 as a key intermediate. Since our
The synthetic approach employed by our Discovery col-
leagues is detailed in Scheme 2.5 In their preparation, com-
mercially available N-Boc-O-benzyl-D-serine (3) was coupled
with ethanolamine via the intermediacy of an isobutyl chloro-
formate-derived mixed carbonic anhydride, followed by O-
acylation of the derived 2-hydroxyethyl amide, affording acetate
4 in ca. 90% overall yield. The conversion of 4 to tetrazole 5
was carried out according to the method of Duncia and co-
workers, wherein exposure of 4 to a mixture of trimethylsilyl
azide (1.0 equiv), triphenylphosphine (1.0 equiv), and diethyl
azodicarboxylate (DEAD, 1.0 equiv) in THF, followed by the
addition of equivalent supplemental quantities of reagents after
24 and 48 h, afforded tetrazole 5 in 85% yield after chromato-
graphic purification.6 Removal of the Boc protecting group was
effected through the action of HCl in dioxane, and subsequent
coupling with N-Boc dimethylglycine under EDAC/HOAt-
mediated amidation conditions, followed by exposure of the
resulting amide to aqueous lithium hydroxide, afforded the
tetrazole core 6 in 71% yield over the three-step sequence. Next,
treatment of alcohol 6 with p-nitrophenyl chloroformate afforded
carbonate 7 in 83% yield after chromatographic purification.
Subsequent reaction with 4-amino-1-butanol formed the pen-
ultimate 8 (not shown), and final exposure to HCl in dioxane
yielded 1 as the corresponding HCl salt. Although this sequence
(5) See ref 1a.
(6) Duncia, J. V.; Pierce, M. E.; Santella, J. B., III. J. Org. Chem. 1991, 56,
2395.
J. Org. Chem. Vol. 74, No. 11, 2009 4069

Davulcu et al.
SCHEME 3. Routes to 1,5-Disubstituted Tetrazoles
SCHEME 4
more, it represents a straightforward combination of chemistries
that are well-precedented on scale. It was recognized that the
potential for hydrazoic acid formation did still exist in this
chemistry (vide infra), but we were confident, based on literature
precedent, that the relative stoichiometries of hydrazine, sodium
nitrite, and HCl could be adjusted to cope with this issue.12
However, it was evident that implementation of this chemistry
toward our target would almost certainly require protection of
the primary hydroxyl group, since the conditions required to
activate the amide would, in all likelihood, affect this hydroxyl
group preferentially. Consequently, the incipient intermediate
would be expected to undergo facile cyclocondensation to the
corresponding oxazoline. This line of reasoning raised a question
that would prove vital to our process development effort: Would
hydrazine react with an oxazoline in a manner analogous to
the known reaction between hydrazine and an imidoyl ester to
afford an amidrazone (e.g., 15 f 14)? If so, this would address
both of the issues associated with the projected amide to
tetrazole conversionsamide activation and hydroxyl protec-
tionsin an efficient and powerfully simplifying manner. We
were unable to find a precedent for this approach in the literature,
so we set out to investigate it in the laboratory and were
delighted to find that oxazoline 16 was smoothly converted to
the corresponding amidrazone 17 upon exposure to hydrazine
hydrochloride in methanol at 0 °C (Scheme 4). With this key
result secured, our efforts were directed toward the development
of this chemistry into a scaleable process.13
met the Discovery material requirements, consideration from
the process perspective revealed significant concerns. First, the
sequence required extensive use of column chromatography,
primarily due to the noncrystalline nature of the intermediates.
In order to maximize throughput, eliminating the requirement
for chromatographic purification would be a key objective. The
successful achievement of this objective would rely on the
identification of crystalline intermediates along the sequence
that could serve as purity control points. Second, the tetrazole-
forming chemistry employed by the Discovery group was found
to be highly exothermic and thus presented safety concerns for
pilot-scale manufacture. Auxiliary concerns regarding scaleup
of this reaction were (1) sourcing and safe handling of
trimethylsilyl azide and (2) efficient removal of the triph-
enylphosphine oxide and diethyl hydrazodicarboxylate byprod-
ucts formed in the reaction. Indeed, laboratory evaluation of
this reaction revealed that several chromatographic separations
were required to deliver 5 at an acceptable purity level.
First Generation Tetrazole Process. Given the aggressive
development timeline associated with this program, we opted
to take advantage of the pool of chiral substances to supply our
starting material. From previous Discovery efforts, N-Boc-O-
Benzyl-D-serine 3 was identified as a commercial (albeit
expensive at ca. $11000/kg) starting material, available in bulk
quantity, and served as our point of departure. Conversion to
the corresponding 2-hydroxyethyl amide 18 was accomplished
using the Discovery conditions (Scheme 5). After aqueous
workup, the resulting dichloromethane stream was subjected to
methanesulfonyl chloride (MsCl) in the presence of triethyl-
amine to provide 19. The requirement for excess MsCl stemmed
from the fact that the isobutanol byproduct from the amidation
chemistry remained in solution, consuming 1 equiv of the
reagent. Upon reaction completion, the mixture was subjected
to aqueous sodium hydroxide and catalytic tetrabutylammonium
hydroxide (under biphasic conditions), thereby effecting cyclo-
condensation to the corresponding oxazoline 16. The resulting
dichloromethane solution was subjected to solvent exchange
distillation to methanol and subsequently treated with hydrazine
hydrochloride at 0 °C, thus affording the amidrazone 17. Sodium
nitrite was introduced to the solution, and the resulting slurry
was treated with methanolic HCl at -10 °C, initiating the
nitrosation/tetrazole-forming chemistry. Upon reaction comple-
tion, the methanol solution was subjected to solvent exchange
distillation to toluene. An aqueous workup, followed by
crystallization from a toluene/heptanes mixed solvent system,
afforded tetrazole 2 in ca. 75% isolated yield. This process
worked well on the laboratory scale (ca. 100 g input), but further
Given the issues associated with this tetrazole synthesis,
coupled with a lack of suitably crystalline intermediates in the
sequence, we sought to define a new synthesis that would be
better suited for scale up. A survey of the literature reveals that
most approaches to 1,5-disubstituted tetrazoles (e.g., 13) rely
on the facile electrocyclization of iminoazides (Scheme 3).7
Representative examples include the reactions of imidoyl esters
or triflates (9),8 amidines (10),9 and imidoyl halides (11)10 with
sodium azide to afford the corresponding iminoazide (12), which
undergoes a rapid electrocyclization to yield the corresponding
1,5-disubstituted tetrazole (13). From the process development
perspective, these approaches are hampered by the necessary
use of sodium azide (or azide salts, in general). A conceptually
similar approach that does not require azide salts involves the
treatment of activated imidoyl derivatives (15) with hydra-
zine, thereby affording the corresponding amidrazone 14.
Exposure to a nitrosating agent (e.g., N2O4, HONO) then gives
rise to the aforementioned iminoazide, in turn affording tetrazole
13 via electrocyclization.11 From the process perspective, this
is compelling because it avoids the use of azide salts. Further-
(7) Wittenberger, S. J. Org. Prep. Proced. Int. 1994, 26, 499, and references
cited therein.
(8) Thomas, E. W. Synthesis, 1993, 767.
(9) Gaponik, P. N.; Grigorev, Y. V.; Karavai, V. P. Khim. Geterotsikl. Soedin.
1985, 566.
(10) (a) von Braun, J.; Rudolf, W. Chem. Ber. 1941, 74, 264. (b) Harvill,
E. K.; Herbst, R. M.; Schreiner, E. C.; Roberts, C. W. J. Org. Chem. 1950, 15,
662.
(11) Temple, C., Jr.; McKee, R. L.; Montgomery, J. A. J. Org. Chem. 1962,
27, 1671.
4070 J. Org. Chem. Vol. 74, No. 11, 2009

Tetrazole-Based Growth Hormone Secretagogue
SCHEME 5. First-Generation Process for the Preparation of 2
scaleup was hampered by four main issues. First, the isobutanol
byproduct generated in the amidation step was not separable
from 18 either through extraction or azeotropic distillation and
was thus carried through the entire sequence. Second, while an
organic base (e.g., triethylamine) was required for the mesylation
reaction (18 f 19), its presence resulted in downstream issues.
Specifically, organic bases catalyzed the dimerization of amidra-
zone 17 to the corresponding triazole 20 and tetrazine 21
impurities (Figure 1). These dimeric impurities led to coloration
of the isolated product, and since their levels increased with
time, they represented a major liability for scaleup. To com-
plicate matters further, the extraction of triethylamine at the
oxazoline stage was precluded by the facile racemization of 16
upon exposure to acid. In fact, 16 was found to racemize simply
on standing in dichloromethane, presumably due to the residual
HCl found in many common chlorinated solvents. Third, it was
found that the diazotization reaction exhibited a substantial
exotherm after an induction period, with much of the heat
evolution occurring during the addition of the last 20-30% of
the methanolic HCl solution. Closer examination (vide infra)
revealed the basis for this delayed exotherm. Fourth, and most
troubling, was the fact that the yield for the sequence became
progressively worse as the reaction scale was increased, with
ca. 30% isolated yields at 1 kg (input) scale and <20% isolated
yields at 2.5 kg (input) scale. The root cause for decreasing
yield upon scaleup was traced to two factors, both of which are
related to the inherent time delays associated with running at
large scale. Normal time lapses in processing led to racemization
of 16 at a higher level than was anticipated from laboratory
runs, resulting in entrainment of the desired compound in the
crystallization liquors as the corresponding racemate. Also, the
solvent exchange distillation from methanol to toluene post-
nitrosation required much more time on a multi-kilogram scale
than was expected (based on laboratory data). This, in turn, led
to substantial decomposition of the product via cleavage of the
Boc protecting group, with subsequent loss of the water-soluble
amine hydrochloride to the aqueous layer during workup.
Addressing the Issues and Refining the Process. Although
significant progress had been made toward the goal of
achieving a safe, azide-free synthesis of tetrazole 2, a robust,
scaleable process was not yet in place. In order to bring this
objective to fruition, the aforementioned issues would have
to be addressed. From a safety perspective, the delayed
exotherm observed in the nitrosation chemistry was a major
concern. However, closer examination revealed that the
underlying cause was quite straightforward and could be
traced back to the triethylamine that was implemented in the
mesylation chemistry. As mentioned earlier, this base could
not be extracted from the product stream, and as a result, it
was simply carried through the entire sequence. Hence, during
the course of the methanolic HCl addition at the nitrosation
stage, the first portions of the HCl charge served simply to
engage the triethylamine in acid-base chemistry, with a
minimal associated heat signature. Once the triethylamine
was neutralized, excess HCl lowered the apparent pH of the
reaction medium. Eventually, significant levels of HONO
(pKa ) 3.2) were accumulated, which initiated the highly
exothermic tetrazole formation (Figure 2). This thermochemi-
cal issue, coupled with the base-catalyzed dimerization of
amidrazone 17 to form impurities 20 and 21, highlighted the
need to purge triethylamine from the process. This was
achieved by modifying the route such that the mesylation
reaction was avoided. Specifically, ethanolamine was replaced
with chloroethylamine hydrochloride in the first step of the
sequence (Scheme 6). In doing so, the derived amide 22
FIGURE 1. Triazole (20) and tetrazine (21) impurities resulting from
amidrazone dimerization.
J. Org. Chem. Vol. 74, No. 11, 2009 4071

Davulcu et al.
FIGURE 2. Depiction of the relationship between reaction pH and rate of diazotization.
SCHEME 6. Second-Generation Process for the Preparation of 2
contained the requisite leaving group at the ꢀ position, thus
heralding a smooth intramolecular cyclocondensation to the
oxazoline 16 upon exposure to aqueous NaOH/n-Bu4NOH.
By removing the mesylation chemistry, this process modi-
fication allowed for more expedient access to 16 and provided
a way to eliminate triethylamine from the sequence. It should
be noted that although chloroethylamine is a strong vesicant,14
the corresponding hydrochloride salt is safe to handle and is
widely available in bulk quantity. During the development
of this process, we were cognizant of the rapid conversion
of chloroethylamine to ethanolamine under basic aqueous
conditions.15 Indeed, the simple addition of water as the first
step of the reaction workup served to reduce residual
chloroethylamine to levels that were not detectable by GC
analysis.
To contend with the persistent isobutanol byproduct that
formed in the first-generation amidation reaction, we opted to
replace isobutyl chloroformate with ethyl chloroformate as the
(12) (a) Perrott, J. R.; Stedman, G.; Uysal, N. J. Chem. Soc., Dalton Trans.:
Inorg. Chem. 1976, 2058. (b) Doherty, A. M.; Howes, K. R.; Stedman, G. Naji,
M. Q. J. Chem. Soc., Dalton Trans.: Inorg. Chem. 1995, 3103.
(13) The scope/generality of this chemistry has not been evaluated.
(14) Ward, K., Jr. J. Org. Chem. 1935, 57, 914.
(15) (a) Chan, S. C.; Leh, F. Aust. J. Chem. 1966, 19, 2271. (b) Cohen, B.;
van Artsdalen, E.; Harris, J. J. Am. Chem. Soc. 1952, 74, 1875.
4072 J. Org. Chem. Vol. 74, No. 11, 2009

Tetrazole-Based Growth Hormone Secretagogue
coupling reagent. This modification led to the generation of
ethanol, a byproduct that was easily removed by aqueous
extraction. Although isobutyl chloroformate is generally re-
garded as a superior coupling reagent, due to an intrinsically
higher level of chemoselectivity in the reaction of the amine
with the derived mixed carbonic anhydride intermediate, it was
determined that control of two key reaction variables, temper-
ature and base charge, could effectively mitigate the chemose-
lectivity and racemization issues commonly associated with the
use of ethyl chloroformate in this arena. Specifically, by
employing a temperature profile wherein the reaction was first
aged at -30 °C for 45 min, then warmed to 0 °C over 30 min,
the extent of racemization was controlled to e2%. Furthermore,
adducts arising from nucleophilic attack at the undesired
carbonyl of the mixed carbonic anhydride intermediate were
maintained at nondetectable levels. Given our laboratory data
which showed that the mixed carbonic anhydride intermediate
suffered racemization primarily via base catalysis, it was
important that this intermediate had only minimal contact with
base. Hence, the N-methylmorpholine charge was split, with
1.1 equiv introduced at the activation step, and the balance
introduced concurrent with the chloroethylamine hydrochloride
charge. Following an aqueous workup to remove ethanol,
N-methylmorpholine and excess chloroethylamine (vide supra),
the resulting dichloromethane stream was subjected to base-
induced cyclocondensation (conditions identical to the first-
generation process), thus affording a dichloromethane solution
of oxazoline 16 in greater than 95% solution yield. Recall that
16 is subject to facile acid-catalyzed racemization and was found
to undergo substantial stereochemical degradation upon standing
in dichloromethane for only a few hours. In order to perform
the required solvent exchange to methanol, a process that would
require ca. 15 h on a pilot scale, we found it beneficial to add
solid sodium bicarbonate to the product stream prior to
distillation. Presumably by quenching residual acid present in
the solution, this base allowed for successful solvent exchange
with only minimal (e2%) racemization. Due to the intrinsically
low solubility of sodium bicarbonate in methanol, the base could
be conveniently removed at the end of the solvent exchange by
simple filtration. Having advanced to this stage of the telescoped
sequence with a high level of enantiomeric purity (g96% ee)
and no residual organic base, we were delighted to find that
subsequent treatment with hydrazine hydrochloride to form
amidrazone 17 proceeded in essentially quantitative solution
yield, with dimeric impurities 20 and 21 controlled at e1.5%
(total). The amidrazone solution was then treated with sodium
nitrite (1.75 equiv), and the resulting slurry was metered into a
cold solution of methanolic HCl (2.0 equiv), resulting in efficient
conversion to tetrazole 2. Notably, the strategy of expunging
residual base from the reaction stream, coupled with an inverse
addition protocol, resulted in a smooth, addition-controlled
exotherm. Furthermore, the relative stoichiometries of hydrazine
hydrochloride, sodium nitrite, and hydrochloric acid served to
preclude hydrazoic acid formation. In preparation for solvent
exchange to toluene, the methanol solution of 2 required
neutralization in order to avoid losses associated with cleavage
of the Boc protecting group. Cognizant of the potential
racemization of 2 under basic conditions, sodium acetate was
selected as a neutralizing agent, thus quenching the HCl in
solution while simultaneously eliminating any potential for
excursion into the high pH regime. Subsequent aqueous workup,
followed by crystallization from toluene:heptanes, afforded a
75% isolated yield of tetrazole 2 in high chemical [>99.5%
HPLC area, > 97% (wt/wt)] and enantiomeric (>99.5%) purity.
Importantly, this process proved to be robust and practical, with
no erosion of yield or quality when employed on pilot scale
(ca. 40 kg input).
Route-Scouting and Securing a Viable Synthetic End-
game. With a suitable process for the preparation of 2 in hand,
laboratory research efforts shifted toward defining a viable
synthetic endgame. Preliminary route-scouting efforts demon-
strated that while advancing 2 via initial installation of the
carbamate side chain was synthetically feasible, it led to a series
of intermediates 23, 24, and 8 that were noncrystalline (Scheme
7). This was a concern because it provided no recourse for purity
control/upgrade by recrystallization until the final stage of the
synthesis (i.e., recrystallization of 1 as its phosphate salt).
Alternatively, conversion of 2 via initial installation of the
N-Boc-dimethylglycine side chain led to a crystalline intermedi-
ate, 6, and while the corresponding carbamate 8 was not
crystalline, the known crystallinity of 1·H3PO4 suggested the
possibility of telescoping the conversion of 6 to 1 without
isolating 8. Given that this latter option (alternative B) presented
two isolable, crystalline intermediates (versus one in alternative
A), it was selected for further development and optimization.
The first step in the advancement of 2, removal of the Boc
protecting group to afford 26, was complicated by the formation
of tert-butyl impurity 27 at levels of ca. 3% (Scheme 8). Given
that 26 was not crystalline and telescoping the product stream
into the subsequent reaction (rather than isolation) was antici-
pated, efforts were focused on controlling purity through reaction
optimization. From a three-factor (equivalent amounts of HCl,
EtOH, and water content), two-level statistical design of
experiment (DOE) analysis,16 water content was found to have
the strongest influence on the formation of 27. Application of
this knowledge led to the design of a reaction medium
comprising 1:2 (vol/vol) ethanol/water containing 4 equiv of
HCl. Gratifyingly, implementation of these reaction conditions
on a kilogram scale afforded practically quantitative conversion
to 26, with levels of 27 maintained at e0.5% (HPLC area). A
variety of amidation protocols were evaluated for the conversion
of 26 to 6. It was found that the mixed carbonic anhydrides
(derived by treating 28 with either ethyl or isobutyl chlorofor-
mate) led to unacceptable levels of the corresponding carbamate
impurities upon reaction with 26. Likewise, evaluation of other
commonly employed coupling reagents (e.g., DMT-MM or
MsCl) led to similarly unacceptable results. However, applica-
tion of the ubiquitous EDAC/HOBt amide bond-forming
methodology proved quite satisfactory, affording 6 in both high
yield and purity, enabling a fully telescoped conversion of 2 to
6. Thus, the acidic stream of 26 from the preceding Boc
deprotection reaction was adjusted to pH 8.3-8.6 with aqueous
potassium phosphate, and subsequently extracted with ethyl
acetate. The resulting extract was subjected to azeotropic
distillation to render it dry, and then introduced directly into
the EDAC/HOBt mediated coupling chemistry (Scheme 9).
Workup and crystallization from a ternary solvent system
(EtOAc/MTBE/n-heptane) afforded high purity 6 in 79%
isolated yield on kilo scale. The major process-related impurities
present in the isolated material were the tert-butyl analogue 29
(vide supra), at ca. 0.1%, and the acylated analogue 30, derived
(16) Box, G. E. P.; Hunter, W. G. ; Hunter, J. S. Statistics for Experimenters;
Wiley: New York, 1978. Carlson, R. Design and Optimisation in Organic
Synthesis; Elsevier: Amsterdam, 199.
J. Org. Chem. Vol. 74, No. 11, 2009 4073

Davulcu et al.
SCHEME 7
SCHEME 8
from EDAC/HOBt mediated esterification at the primary
hydroxyl group of 6, at ca. 0.1%.
The most expedient solution to this problem involved transient
protection of the hydroxyl group present in 31. The trimethylsilyl
(TMS) protecting group was preferred since removal could be
accomplished under very mild conditions, ideally during the
reaction workup. Laboratory evaluation of this strategy was
encouraging. Deployment of the TMS-protected nucleophile
4-(trimethylsilyloxy)butan-1-amine (34) effectively suppressed
the formation of the pseudodimer 33, and protodesilylation could
be accomplished by exposure to a dilute aqueous hydrochloric
In order to secure the carbamate-linked aminobutanol side
chain of the target, 1,1-carbonyldiimidazole (CDI) was selected
as the reagent of choice. However, we found that implementation
of unprotected 4-amino-1-butanol (31) in the coupling reaction
resulted in unacceptable levels of the pseudodimer 33, presum-
ably via reaction of the free hydroxyl of the desired product 8
with the imidazolide intermediate 32 (Scheme 10).
4074 J. Org. Chem. Vol. 74, No. 11, 2009

Tetrazole-Based Growth Hormone Secretagogue
SCHEME 9
SCHEME 10
SCHEME 11
acid solution. The preparation of 34 from 31, employing
hexamethyldisilizane (HMDS) and catalytic chlorotrimethylsi-
lane, had been described in the literature,17 but the procedure
was found to be difficult to implement on scale for a number
of reasons. First of all, catalyst loadings were not specified in
the original paper. To address this, we carried out a series of
experiments which showed that catalyst loadings as low as 0.3%
gave acceptable results. Reaction stalling, resulting in only
partial protection of 31, was another vexing problem. We found
that stalling could be remedied through the introduction of
supplemental hexamethyldisilizane charges during the course
of the reaction. The observation that conversion was directly
proportional to the initial HMDS charge led us to suspect that
reaction stalling might be due to loss, through volatilization, of
the intermediate silylating agent (trimethylsilylamine), suggest-
ing that the problem could be overcome by simply performing
the reaction in the absence of a nitrogen sweep. Gratifyingly,
this modification did resolve the stalling issue, decreasing the
residual levels of 31 to e0.1%. Furthermore, the resulting
process proved to be robust and reproducible, with typical yields
of 96% and purities ranging from 89-96% (by GC analysis)
on a kilogram scale (Scheme 11). Also, the fact that crude 34
could be employed directly in the subsequent step without the
need for purification was of great practical value and proved to
be a key factor in enabling a the CDI-mediated synthesis of 8.
Considering the lack of crystallinity in 8, the plan from the
outset was to telescope a stream containing 8, without isolation,
into the Boc cleavage reaction to furnish 1. While 1 was also
noncrystalline, its phosphate salt was found to be crystalline,
and we anticipated that isolation of the phosphate salt would
be accompanied by purity upgrade. Unfortunately, screening
of a number of conditions showed that isolation of 1·H3PO4
(17) Mormann, W.; Leukel, G. Synthesis 1988, 990, 992.
J. Org. Chem. Vol. 74, No. 11, 2009 4075

Davulcu et al.
SCHEME 12. Telescoped Preparation of 1 ·R-Mandelate
SCHEME 13. Conversion of 1 ·R-Mandelate to 1·H3PO4
was accompanied by only minimal purity upgrade, inadequate
to purge the impurities present in the material to acceptable
levels. Thus, an automated, high-throughput salt screen was
initiated, with the goal of identifying a “bridging salt” that would
confer an element of purity control at this late stage of the
process. In this screen, a total of 60 acids were screened in five
different solvent systems. This screen revealed the 1·R-
mandelate and 1·glycolate salts as two previously unidentified
crystalline forms. Further experimentation showed that the
mandelate salt, when isolated from ethyl acetate/2-propanol,
provided the desired level of purity/impurity and crystallization
control with optimal mass recovery. This was a key finding and
proved to be instrumental in enabling the endgame of this
synthetic sequence, culminating in the isolation of 1·H3PO4
(Scheme 12 and 13). In the process, a solution of 6 in isopropyl
acetate was first treated with CDI, eliciting conversion to the
corresponding imidazolide (32), followed by introduction of
crude 34, to afford 35. Washing of the reaction stream with 0.5
N hydrochloric acid induced protodesilyation, thus giving rise
to 8. The resulting organic solution was subjected to vacuum
distillation, thereby effecting solvent exchange to ethanol, and
subsequently treated with concentrated hydrochloric acid at 60
°C to effect removal of the Boc protecting group. Upon reaction
completion, the ethanolic solution of 1·HCl was diluted with
ethyl acetate, then neutralized with potassium phosphate solu-
tion, and the derived organic layer was distilled to an end point
of e1% EtOH (GC) and e1000 ppm water [Karl Fischer (KF)
titration]. Introduction of 2-propanol, followed by R-mandelic
4076 J. Org. Chem. Vol. 74, No. 11, 2009

Tetrazole-Based Growth Hormone Secretagogue
chloroformate (16.3 kg, 150 mol, 1.1 equiv) was subsequently
introduced while maintaining the reaction temperature below -30
°C. The charge was chased with 10 L of dichloromethane, and the
resulting white slurry was aged at -30 ( 3 °C for 50 min. Next,
N-methylmorpholine (26.0 kg, 257 mol, 1.9 equiv) was charged
while maintaining the reaction temperature below -30 °C. Upon
completion of the addition, the mixture was cooled to -40 °C, and
milled chloroethylamine hydrochloride (17.3 kg, 149 mol, 1.1 equiv)
was added in one portion via the reactor manway, causing the
internal temperature to rise to approximately -28 °C. Once the
addition was complete, the reaction temperature was adjusted to
-30 °C and the slurry was agitated for 45 min. The slurry was
then warmed to 0 °C over the course of 40 min and aged at said
temperature until complete consumption of starting material was
achieved [specification: [3] e 2.0% (HPLC area)]. The reaction
was quenched at 0 °C with deionized water (140 L), and the
resulting mixture was warmed to 20 °C over 30 min. Agitation
was stopped, the mixture was allowed to separate, and the resulting
aqueous (top) layer was discarded. The derived organic layer was
treated with 0.5 N hydrochloric acid solution (140 L) and agitated
for 30 min. Again agitation was stopped, the mixture was allowed
to separate, and the resulting aqueous (top) layer was discarded. In
the same manner, the organic layer was washed with deionized
water (140 L), and the resulting aqueous (top) layer was discarded.
Step 2: Cyclocondensation (22 f 16). The dichloromethane
solution (from step 1) was treated with deionized water (97 L),
15% (wt/wt) sodium hydroxide solution (105 L), and 40% (wt/wt)
tetrabutylammonium hydroxide solution (4.4 kg, 17 mol, 0.05
equiv), and the resulting biphasic mixture was stirred at 20 °C for
10.5 h. Upon complete consumption of 22 [specification: [22] e
1.0% (HPLC area)], agitation was stopped, the layers were allowed
to separate, and the aqueous (top) layer was discarded. The
dichloromethane layer was then washed with 5% (wt/wt) sodium
bicarbonate solution (224 L), followed by 5% (wt/wt) sodium
chloride solution (224 L), and the spent aqueous (top) phases were
discarded. The derived dichloromethane solution was then treated
with solid sodium bicarbonate (7.9 kg) and distilled under 200 -
500 mbar pressure, while maintaining the internal temperature e35
°C, to a final volume of 1.2 - 1.5 L per kg of N-Boc-O-benzyl-
D-serine (3) input. Methanol was added (100 L) and the distillation
was continued at 50-500 mbar pressure, while maintaining the
internal temperature e35 °C, to an end point of e1.0% (wt/wt)
CH2Cl2 (determined by GC). Finally, the concentrate was diluted
with methanol (80 L) and filtered for use in the next step.
Step 3: Amidrazone Formation (16 f 17). To an 800 L
Hastalloy reactor were charged hydrazine monohydrochloride (12.1
kg, 176.6 mol, 1.30 equiv) and methanol (520 L), and the stirred
solution was cooled to 0 °C. The methanolic solution of 16 (product
stream from step 2) was then transferred into the hydrazine
monohydrochloride solution, while maintaining the reaction tem-
perature below 5 °C, and the resulting solution was aged at 0 °C
for 1.5 h, effecting complete conversion to amidrazone 17
[specification: [16] < 1.0% (HPLC area)].
Step 4: Nitrosation (17 f 2). To a 2000 L Hastalloy reactor
was charged methanol (400 L), which was cooled to 0 °C and
treated with acetyl chloride (21.3 kg, 271 mol, 2.0 equiv) while
maintaining the reaction temperature between 0 and 5 °C. The
resulting solution was aged at 0 °C for 30 min and then cooled to
-50 °C. Next, sodium nitrite (16.4 kg, 238 mol, 1.75 equiv) was
charged to a solution of 17 (product stream from step 3) contained
in an 800 L reactor at 0 °C. The resulting mixture was agitated for
15 min at 0 °C, then cooled to -30 °C and metered into the 2000
L Hastalloy reactor containing methanolic HCl, while maintaining
the reaction temperature e -10 °C. Once this transfer was
complete, the resulting mixture was stirred at -10 °C for 20 min.
HPLC analysis of the reaction mixture at this point indicated e0.1%
(area) 17 remaining. The enantiomeric purity of 2 (in solution) was
ca. 97% ee, and the pH of the reaction mixture was ca. 1 (by
water-wet pH paper). To neutralize the reaction mixture, anhydrous
acid, furnished the 1·R-mandelate salt in 83% overall isolated
yield for the four-step sequence. The expected purity upgrade
was realized, evidenced by assay values of 99.5% (HPLC area),
99.2% (wt/wt), and >99.9% ee.
With high-purity 1·R-mandelate salt in hand, all that
remained was conversion to the desired phosphate salt 1·H3PO4
(note that mandelic acid is not a pharmaceutically preferred
counterion). Thus, 1·R-mandelate was partitioned between
ethyl acetate and aqueous potassium phosphate, and the resulting
free-base solution was rendered anhydrous through azeotropic
distillation. After filtration to remove insoluble contaminants,
1·H3PO4 seed crystals were charged, followed by aqueous
phosphoric acid, at 50 °C. During laboratory development, this
process was found to have two major problems. One was the
tendency to form the O-acylated impurity 36 (not shown), via
acid-catalyzed esterification, during the crystallization. This side
reaction was minimized through the careful quantitation of 1
and implementation of a substoichiometric (0.95 equiv.) charge
of H3PO4. The second problem was ascribed to the exceedingly
low solubility of 1·H3PO4 in the crystallization medium. As a
consequence of the high level of supersaturation, the addition
of phosphoric acid resulted in uncontrolled crystallization on
the reactor walls. To contend with this issue, we opted to charge
1·H3PO4 seed crystals prior to the slow, controlled introduction
of phosphoric acid. Demonstration of this protocol at kilo scale
was met with success, resulting in an 81% isolated yield of
1·H3PO4 with greater than 99.5% purity and only 0.03% of
the O-acylated impurity 36.
Conclusion
In summary, this manuscript discloses the development and
execution of a robust, efficient and practical synthesis of the
growth hormone secretagogue 1·H3PO4. Expansion of the
established amidrazone formation/diazotization methodology to
include oxazolines as competent electrophiles eliminated the
need for azide salts and allowed for the development of a safe,
scaleable synthesis of the tetrazole core (2). This process
employed a complex, four-step telescoped sequence that enabled
the preparation of 2 from N-Boc-O-benzyl-D-serine in 75% yield
with only one isolation. The ambident reactivity of 4-amino-
1-butanol in the carbamoylation of 6 led to the implementation
of the labile trimethylsilyl protecting group, solving a critical
chemoselectivity issue while only minimally impacting process
throughput. The endgame of the synthesis also leveraged
extensive telescoping to contend with a lack of crystalline
intermediates. In order to more fully control the quality attributes
(e.g., color, purity, etc.) of the final product, an isolation was
added to the endgame sequence, exploiting an intermediate
R-mandelate salt as a chemical purity gatekeeper. A highly
optimized sequence of eight chemical operations furnished the
active pharmaceutical intermediate (API) 1·H3PO4 in 53%
overall yield (from 2).
Experimental Procedures
Telescoped Preparation of 1-(ꢀ-Hydroxyethyl)-5-(r-R-tert-
butyloxycarbonylamino-ꢀ-benzyloxyethyl)tetrazole (2). Step
1: Amidation (3 f 22). To an 800 L Hastalloy reactor were
charged N-Boc-O-senzyl-D-serine (3) (40.0 kg, 135 mol, 1.0 equiv)
and dichloromethane (200 L). The resulting solution was cooled
to -35 °C, N-methylmorpholine (15.0 kg, 148 mol, 1.1 equiv) was
introduced while maintaining the reaction temperature below -30
°C, and the charge was chased with 10 L of dichloromethane. Ethyl
J. Org. Chem. Vol. 74, No. 11, 2009 4077

Davulcu et al.
sodium acetate (26.0 kg, 316 mol, 2.3 equiv) was charged, and the
agitated mixture was warmed to 20 °C to afford a pH of 5-6 (by
water-wet pH paper). Toluene (600 L) was then charged, and the
mixture was concentrated to ca. 530 L via reduced-pressure
distillation, maintaining the internal temperature at e40 °C. A
second portion of toluene (600 L) was added, and again the mixture
was concentrated to ca. 530 L under identical conditions. Subse-
quent GC analysis of the filtered reaction mixture indicated e3.0%
(vol/vol) methanol, thus verifying that the solvent exchange was
complete. The reaction temperature was adjusted to 20 °C, and the
mixture was treated with deionized water (320 L). After 15 min,
agitation was stopped, the mixture was allowed to separate for 30
min, and the aqueous (bottom) layer was discarded. The organic
layer was subsequently treated with 5% (wt/wt) sodium bicarbonate
solution (320 L) and agitated for 15 min. Again, agitation was
stopped, the mixture was allowed to separate for 30 min, and the
aqueous (bottom) layer was discarded. The organic layer was finally
treated with 5% (wt/wt) sodium chloride solution (320 L) and
agitated for 15 min. Again, agitation was stopped, the mixture was
allowed to separate for 30 min, and the aqueous (bottom) layer
was discarded. The resulting toluene solution was concentrated to
ca. 266 L via reduced-pressure distillation, while maintaining the
internal temperature e35 °C, to a KF titration end point of e500
ppm H2O. The mixture was clarified by filtration, and the resulting
solution was transferred to an 800 L Hastalloy reactor. With
agitation, the solution was warmed to 50 °C and then treated with
n-heptane (180 L) while maintaining the internal temperature
between 45 and 50 °C, yielding a homogeneous solution (GC
analysis of the mixture indicated a 56:44 toluene/n-heptane (vol/
vol) composition). The solution was cooled to 20 °C over 3 h, and
spontaneous crystallization commenced at ca. 35 °C. The slurry
was further cooled to 0 °C over 1 h, aged for an additional hour,
and then discharged to a Nutsche filter. The resulting wetcake was
washed twice with heptanes (2 × 60 L) and subsequently dried
under vacuum (35 °C, 25 mmHg) to 0.02% LOD, furnishing 36.9
kg of 2 (75% yield) as a white, crystalline solid, mp 77.1 °C with
97.8% HPLC area purity, 99.3% (wt/wt) assay purity, and enan-
tiomeric purity of 99.4% ee: [R]25D +17.2 (c 1.03, MeOH); 1H NMR
(400 MHz, DMSO-d6) δ 7.44 (app d, J ) 8.5 Hz, 1 H), 7.25-7.35
(m, 5 H), 5.28 (app q, J ) 6.4 Hz, 1 H), 5.11 (t, J ) 5.5 Hz, 1 H),
4.51-4.55 (m, 4 H), 3.93 (app t, J ) 8.9 Hz, 1 H), 3.79-3.85 (m,
2 H), 3.68-3.75 (m, 1 H), 1.37 (s, 9 H); 13C NMR (100 MHz,
DMSO-d6) δ 155.4, 154.8, 138.0, 128.2, 127.5, 127.4, 78.8, 72.1,
69.5, 59.4, 49.5, 44.3, 28.1; IR (KBr pellet) νmax 3488 (br), 3325,
2983, 2938, 2883, 1681, 1527, 1458, 1369, 1337, 1279, 1159. Anal.
Calcd for C17H25N5O4: C, 56.18; H, 6.93; N, 19.27. Found: C, 56.29;
H, 7.04; N, 19.40.
Telescoped Preparation of (S)-tert-Butyl-1-(2-(benzyloxy)-1-
(1-(2-hydroxyethyl)-1H-tetrazol-5-yl)ethylamino)-2-methyl-1-
oxopropan-2-ylcarbamate (6). Step 1: Deprotection (2 f 26).
To a 100 L glass-lined reactor were charged 2 (2.41 kg), SDA 3A
ethanol (3.21 kg), water (8.09 kg), and concentrated hydrochloric
acid (2.62 kg of a 37% solution). The resulting mixture was heated
at 60 °C for 1.5 h, at which point HPLC analysis indicated complete
consumption of 2. Upon cooling to 20 °C, the pH of the batch was
adjusted to ca. 8 by the addition of 2.0 M potassium phosphate
solution, and the resulting mixture was extracted with ethyl acetate
(54.1 kg). The derived organic phase was concentrated to ca. 55 L
via reduced pressure distillation, while maintaining the internal
temperature e40 °C, to a KF titration end point of 0.8% H2O, and
then cooled to 20 °C.
Step 2: Amidation (26 f 6). To the stirred solution of 26 in
ethyl acetate (product stream from step 1) were introduced HOBt
(180 g), N-Boc-2-aminoisobutyric acid (1.40 kg), EDAC (1.37 kg),
and DMF (3.41 kg). The resulting mixture was aged at 20 °C for
4.5 h, at which point HPLC analysis indicated complete consump-
tion of 26. The mixture was diluted with water (12.2 kg) and ethyl
acetate (54.2 kg). After 15 min, agitation was stopped, the mixture
was allowed to separate for 30 min, and the aqueous (bottom) layer
was discarded. The derived organic layer was successively washed
with 0.1 N HCl solution (12.0 kg), 5% (wt/wt) NaHCO3 solution
(24.0 kg), and water (2 × 24.0 kg). The derived solution was then
concentrated to ca. 65 L via reduced pressure distillation, while
maintaining the internal temperature e40 °C, to a KF titration end
point of e0.2% H2O. The mixture was clarified by filtration,
concentrated to 10.0 L, and diluted with MTBE (7.12 kg). While
maintaining the batch temperature between 45 and 50 °C, n-heptane
(11.85 kg) was introduced, followed by a portion of 6 seed material
(24..0 g, 1%). The batch was aged at 50 °C for 2 h, resulting in a
thin slurry. Additional n-heptane (7.9 kg) was charged, and the
slurry was aged for 1 h at 50 °C and then cooled to 20 °C over
2 h. The product was collected by filtration, washed with a solution
of n-heptane (1.64 kg) and MTBE (1.78 kg), and subsequently dried
under vacuum (50 °C, 25 mmHg), affording 2.35 kg of 6 (79.3%
yield) as a white, crystalline solid, mp ) 110 °C, with 99.6% purity
(HPLC area): [R]25D +12.4 (c 1.02, CH2Cl2); 1H NMR (300 MHz,
CDCl3) δ 7.35-7.18 (m, 5H), 5.66 (q, J ) 6.0 Hz, 1 H), 4.94 (s,
1 H), 4.46-4.60 (m, 4 H), 3.95-4.0 (m, 4 H), 3.83 (m, 1 H), 3.49
(t, J ) 6.3 Hz, 1 H), 1.43 (s, 3 H), 1.40 (s, 3 H), 1.33 (s, 9 H); 13
C
NMR (75 MHz, CDCl3) δ 174.4, 154.1, 154.0; 136.5, 128.0, 127.6,
127.3, 80.6, 73.4, 70.2, 61.0, 56.8, 50.6, 43.9, 28.5, 25.9, 25.7; IR
(KBr pellet) νmax 3401 (br), 2981, 1709, 1668, 1542, 1509, 1459,
1366, 1276, 1186, 1160, 1122, 1083. Anal. Calcd for C21H32N6O5:
C, 56.23; H, 7.19; N, 18.73. Found: C, 56.19; H, 7.25; N, 18.82.
Preparation of 4-(Trimethylsilyloxy)butan-1-amine (34). To
a 4 L glass reactor were charged 4-amino-1-butanol (31) (1.64 L,
1.59 kg, 17.8 mol, 1.0 equiv) and 1,1,1,3,3,3-hexamethyldisilazane
(2.1 L, 1.61 kg, 9.95 mol, 0.56 equiv). With agitation, chlorotri-
methylsilane (6.78 mL, 5.8 g, 0.053 mol, 0.003 equiv) was
introduced via syringe; the resulting mixture was warmed to 80 °C
over the course of 90 min and subsequently aged at said temperature
for ca. 3 h. GC analysis of the reaction mixture at this point
indicated 0.049% (area) 31 remaining (specification: [31] e 0.25%).
The mixture was cooled to 20 °C and discharged to a 20 L carboy,
affording 2.98 kg of crude, unpurified 34 [97.4% yield (corrected
for purity)] as a light yellow oil. GC analysis of the product stream
indicated purity of 94.5% (area), with 0.024% (area) residual 31.
Spectral data were consistent with those reported in the literature.
Telescoped Preparation of 1 ·R-Mandelate. Step 1: Car-
bamate Formation (6 f 8). To a 100 L glass-lined reactor were
charged CDI (1,1′-carbonyldiimidazole) (910 g, 5.61 mol, 1.1 equiv)
and isopropyl acetate (2.45 kg). To a 20 L carboy were charged 6
(2.1 kg, 4.68 mol, 1.0 equiv) and isopropyl acetate (15.3 kg). To
the agitated CDI slurry was introduced the solution of 6, and the
resultant mixture was aged for 2 h at 22-25 °C, effecting complete
consumption of starting material [specification: [6] e 1.0% (HPLC
area)]. Next, neat 34 (1.13 kg, 7.0 mol, ca. 1.5 equiv) was introduced
to reaction mixture, followed by an isopropyl acetate (1.0 kg) rinse,
and the resulting solution was aged at 22-25 °C until the
imidazolide intermediate was reduced to a level of e0.5% (HPLC
area). Upon completion, 0.5 N HCl (17.32 kg) was introduced to
the mixture. After 30 min, agitation was stopped, the mixture was
allowed to separate for 30 min, and the aqueous (bottom) layer
was discarded. Two additional 0.5 N HCl treatments were carried
out in an identical manner, followed by HPLC analysis to verify
that desilylation was complete [completion criteria: [35] e 0.5%
(HPLC area); result: 0.3%]. The organic layer was successively
washed with 5.0% (wt/wt) potassium phosphate solution (15.0 kg)
and 5.0% (wt/wt) NaCl solution (15.0 kg). The derived solution
was concentrated to ca. 8 L via reduced-pressure distillation,
maintaining the internal temperature e50 °C, then subjected to a
constant-volume distillative solvent exchange to ethanol, to a GC
end point of e0.1% (vol/vol) isopropyl acetate in ethanol.
Step 2: Deprotection (8 f 1). To the filtered reaction stream
from step 1 were charged water (25.2 kg) and 37% aqueous
hydrochloric acid solution (1.85 kg), and the resulting mixture was
heated 60 °C for 3.25 h, at which point HPLC analysis indicated
complete consumption of 8. Upon cooling to 20 °C, the pH of the
4078 J. Org. Chem. Vol. 74, No. 11, 2009

Tetrazole-Based Growth Hormone Secretagogue
batch was adjusted to ca. 8 by the addition of 2.0 M potassium
phosphate solution, and the resulting mixture was extracted twice
with ethyl acetate (2 × 15.8 kg). The derived organic phase was
then subjected to distillative solvent exchange to ethyl acetate, while
maintaining the internal temperature e50 °C, to a KF titration end
point of 0.02% H2O.
cooled mixture was transferred to another glass-lined 100 L reactor
through a 5 µm cartridge filter, and the resulting clarified solution
of 1 was adjusted to a final volume of 36 L through the addition of
filtered ethyl acetate.
Step 2: Crystallization and Isolation of 1 ·H3PO4. To the
solution of 1 in ethyl acetate was charged water (0.36 kg, 1.0%
(vol/vol)), and the batch was heated to 50 °C. A phosphoric acid
solution was prepared by combining concentrated phosphoric acid
(0.365 kg, 0.97 equiv based on quantitation of 1), water (0.36 kg),
and ethyl acetate (32.0 kg). With vigorous agitation, ca. 500 mL
of the phosphoric acid solution was introduced to the reactor at a
rate of e100 mL/min, affording a cloudy mixture. A portion of
1·H3PO4 seeds (34 g, 2.0%) was charged, followed by introduction
of the remaining acid solution over ca. 3 h. Upon completion of
the acid charge, the batch was cooled to 20 °C over ca. 1 h,
subsequently aged at said temperature for 1 h, and ultimately
discharged to a Nutsche filter. The resulting wetcake was washed
with ethyl acetate (18.1 kg) and dried under vacuum (50 °C, 25
mmHg) to e1.0% LOD, furnishing 1.8 kg of 1·H3PO4 (81.2%
yield) as a white, crystalline solid, mp ) 104 - 105 °C, with 99.9%
HPLC area purity, 99.2% (wt/wt) assay purity, and enantiomeric
purity of >99.9% ee: [R]25D +3.03 (c 1.01, MeOH); 1H NMR (400
MHz, DMSO-d6) δ 7.24-7.37 (m, 5 H), 7.15 (app t, J ) 5.7 Hz,
1 H), 5.51 (app t, J ) 7.01 Hz, 1 H), 4.65-4.74 (br m, 2 H), 4.54
(br s, 2 H), 4.36-4.44 (m, 1 H), 4.28-4.35 (m, 1 H), 4.00 (dd, J
) 7.49, 9.77 Hz, 1 H), 3.92 (dd, J ) 6.51, 9.60 Hz, 1 H), 3.32-3.38
(m, 2 H), 2.89-2.96 (m, 2 H), 1.43 (s, 3 H), 1.40 (s, 3 H),
1.33-1.41 (m, 4 H); 13C NMR (100 MHz, D2O) δ 172.6, 157.0,
153.9, 136.6, 128.7, 128.4, 128.3, 73.5, 68.8, 62.6, 61.8, 57.7, 47.9,
44.6, 40.8, 29.2, 23.9, 23.7; IR (KBr pellet) νmax 3345 (br, 2943
(br), 1733, 1718, 1692, 1675, 1542, 1470, 1376, 1253, 1199, 1124,
1101; HRMS (ESI-MS) calcd. for C21H34N7O5 464.2616 [(M +
H)+], found 464.2614.
Step 3: Salt Formation (1 f 1·R-Mandelate). After clarifica-
tion, the mixture was treated with a solution of mandelic acid (0.72
kg) in ethyl acetate (7.58 kg) at 50 °C. The resulting slurry was
aged at 50 °C for 1 h, cooled to 10 °C over 1.5 h, and aged for an
additional 1 h at 10 °C. The product was collected by filtration,
washed with ethyl acetate (16.3 kg), and subsequently dried under
vacuum (50 °C, 25 mmHg) to e1.0% LOD, affording 2.38 kg of
1·R-Mandelate (83.1% yield) as a white, crystalline solid, mp )
113.7 °C (dec), with an HPLC area purity of 99.45% and
enantiomeric purity of >99.9% ee: [R]25 -18.12 (c 1.01, H2O);
D
1H NMR (400 MHz, DMSO-d6) δ 1.30 (s, 3 H), 1.32 (s, 3 H),
1.33-1.43 (m, 4 H), 2.83-2.97 (m, 2 H), 3.29-3.40 (m, 2 H),
3.88 (dd, J ) 6.32, 9.71 Hz, 1 H), 3.96 (dd, J ) 7.83, 9.74 Hz,
1H), 4.25-4.42 (m 2 H), 4.49-4.57 (m, 2 H), 4.61-4.76 (m, 3
H), 5.50 (t, J ) 6.34 Hz, 1 H), 7.12-7.41 (m, 10 H); 13C NMR
(100 MHz, DMSO-d6) δ 26.1, 26.2, 27.1, 30.8, 44.3, 47.6, 56.6,
61.3, 62.3, 69.9, 73.1, 73.9, 126.7, 126.8, 127.8, 127.9, 127.9, 128.6,
138.1, 142.9, 154.4, 155.6, 174.2, 174.3; IR (KBr pellet) νmax 3477,
3322 (br), 2944 (br), 1737, 1683, 1637, 1542, 1456, 1366, 1342,
1256, 1193; HRMS (ESI-MS) calcd for C21H34N7O5 464.2616 [(M
+ H)+], found 464.2614. Anal. Calcd for C29H41N7O8: C, 56.57;
H, 6.71; N, 15.93. Found: C, 56.30; H, 6.55; N, 15.86.
Telescoped Preparation of 1 ·H3PO4. Step 1: Free-Base
Formation (1 ·R-Mandelate f 1). To a glass-lined 100 L reactor
were charged 1·R-mandelate (2.37 kg) and water (40.17 kg), and
the resulting mixture was agitated to effect dissolution. Next, ethyl
acetate (14.5 kg) was charged, and the batch pH was adjusted to
8.5-9.0 by the addition of ca. 2.5 kg of a 2.0 M K3PO4 solution.
After 15 min, agitation was stopped, and the phases were allowed
to separate. The organic phase was retained, the aqueous layer was
extracted twice with ethyl acetate (14.5 kg), and the resulting
organic solutions were combined. After washing with 0.075 M
K3PO4 (10.2 kg), followed by 5.0% aqueous NaCl (10.5 kg), the
solution was concentrated via vacuum distillation (batch temperature
35-50 °C, ca. 300 mmHg) to a target volume of ca. 30 L. The
Supporting Information Available: General experimental
1
details, along with copies of H and 13C NMR spectra for all
new compounds. This material is available free of charge via
the Internet at http://pubs.acs.org.
JO9003508
J. Org. Chem. Vol. 74, No. 11, 2009 4079