The combination of hydroformylation and biocatalysis for the large-scale synthesis of (S)-allysine ethylene acetal

Article abstract of DOI:10.1021/op100258j

The compound, (S)-2-amino-5-[1,3]dioxolan-2-yl-pentanoic acid [(S)-allysine ethylene acetal], is a key intermediate in a number of angiotension-I converting enzyme (ACE) and neutral endopeptidase (NEP) inhibitors currently in clinical trials. Through a co

Full text of DOI:10.1021/op100258j

Organic Process Research & Development 2011, 15, 284–290
The Combination of Hydroformylation and Biocatalysis for the Large-Scale Synthesis
of (S)-Allysine Ethylene Acetal
Christopher J. Cobley,*,† Christopher H. Hanson,‡ Michael C. Lloyd,*,† and Shaun Simmonds†
Chirotech Technology Ltd., Dr Reddy’s Laboratories (EU) Ltd., Unit 162 Cambridge Science Park, Cambridge CB4 0GH U.K.,
and Dr. Reddy’s Laboratories (EU) Ltd., Steanard Lane, Mirfield, West Yorkshire WF14 8HZ, U.K.
Wei Jun Peng
Nantong Cellulose Fibers Co., LTD, 27 East Zhongxiu Road, Nantong, Jiangsu, China
Abstract:
material was key to developing a cost-effective and scalable
process to (S)-allysine ethylene acetal. We envisaged using
crotonaldehyde as a low-cost and readily available starting
material for our synthesis (Scheme 1) and transforming it to
the desired glutaraldehyde monoacetal (2) in two steps: first
conversion into the corresponding ethylene acetal (1), followed
by hydroformylation using a rhodium catalyst. Such a trans-
formation requires the catalysis of two processes: (i) initial
isomerisation of (1) to give the terminal double bond and (ii)
linear selective hydroformylation. The use of hydroformylation
for the production of pharmaceutical intermediates and fine
chemicals has been a technology area under active development
within Chirotech for several years.4 We have a number of
catalytic systems available to us for this transformation and
considerable experience in taking such reactions to the manu-
facturing plant. Having synthesised aldehyde (2), this can be
readily transformed Via a Strecker reaction into aminonitrile
(3). Subsequent conversion into either 4a or 4b yields two
potential substrates for bioresolution (Scheme 1).
The compound, (S)-2-amino-5-[1,3]dioxolan-2-yl-pentanoic acid
[(S)-allysine ethylene acetal], is a key intermediate in a number of
angiotension-I converting enzyme (ACE) and neutral endopepti-
dase (NEP) inhibitors currently in clinical trials. Through a
combination of our hydroformylation and biocatalysis technologies
we have developed an efficient five-step synthetic route to this
material starting from crotonaldehyde. The development of this
process, leading to a large-scale commercial manufacturing
campaign, is described in this paper.
Introduction
(S)-2-Amino-5-[1,3]dioxolan-2-yl-pentanoic acid [(S)-all-
ysine ethylene acetal] is a key intermediate in a number of
angiotension-I converting enzyme (ACE) and neutral endopep-
tidase (NEP) inhibitors currently in clinical trials (Figure 1).
The synthesis of racemic allysine ethylene acetal was initially
reported Via a laborious eight-step chemical synthesis from 3,4-
dihydro-2H-pyran.1 A wide variety of enzyme classes have been
utilised in the preparation of (S)-allysine ethylene acetal,
including phenylalanine dehydrogenases, amidases, D-amino
acid oxidases, and hydantoinases.2 In the majority of cases, these
approaches utilise glutaraldehyde as a key starting material and
involve selective monoacetalation of the dialdehyde. Unfortu-
nately, this step is not particularly selective, and isolated yields
are typically reported as less than 50%.2e,3 An alternative
bioresolution was sought that could be operated without
infringement of existing intellectual property.
Previous work in our laboratories led to the development of
a thermophilic L-acylase from archaeon Thermoccocus litoralis,5
which has been shown to catalyse the bioresolution of a wide
variety of N-acyl amino acids. Although the L-acylase bioreso-
lution can only generate a maximum 50% yield of (5), the
process does offer the opportunity to recycle the unwanted
enantiomer Via an intermediate azlactone (Scheme 2).
Results and Discussion
Stage 1: Acetal Formation. Crotonaldehyde ethylene acetal
(1) was synthesised by following a literature procedure with
minor modifications.6 The literature procedure used benzene
The development of an alternative higher-yielding method
of producing glutaraldehyde monoacetal from a cheap starting
* Authors for correspondence. E-mails: ccobley@drreddys.com; mlloyd2@
drreddys.com.
(4) (a) Cobley, C. J.; Gardner, K.; Klosin, J.; Praquin, C.; Hill, C.; Whiteker,
G. T.; Zanotti-Gerosa, A.; Petersen, J. L.; Abboud, K. A. J. Org. Chem.
2004, 69, 4031–4040. (b) Cobley, C. J.; Klosin, J.; Qin, C.; Whiteker,
G. T. Org. Lett. 2004, 6, 3277–3280. (c) Briggs, J. R.; Klosin, J.;
Whiteker, G. T. Org. Lett. 2005, 7, 4795–4798. (d) Axtell, A. T.;
Cobley, C. J.; Klosin, J.; Whiteker, G. T.; Zanotti-Gerosa, A.; Abboud,
K. A. Angew. Chem., Int. Ed. 2005, 44, 5834–5838. (e) Cobley, C. J.;
Froese, R. D. J.; Klosin, J.; Qin, C.; Whiteker, G. T. Organometallics
2007, 26, 2986–2999. (f) Thomas, P. J.; Axtell, A. T.; Klosin, J.; Peng,
W.; Rand, C. L.; Clark, T. P.; Landis, C. R.; Abboud, K. A. Org. Lett.
2007, 9, 2665–2668.
† Chirotech Technology Ltd.
‡ Dr. Reddy’s Laboratories (EU) Ltd.
(1) Rumbero, A.; Mart´ın, J. F.; Lumbreras, M. A.; Liras, P.; Esmahan, C.
Bioorg. Med. Chem. 1995, 3 (9), 1237–1240.
(2) (a) Patel, R. N. AdV. Synth. Chem. 2001, 343, 527–546. (b) Vo-Quang,
Y.; Marais, D.; Vo-Quang, L.; Le Goffic, F.; Thie´ry, A.; Maestracci,
M.; Arnaud, A.; Galzy, P. Tetrahedron Lett. 1987, 28 (35), 4057–4060.
(c) Tanaka, A.; Doya, M.; Uchiyama, T. Tahara, T. U.S. Patent
6,174,711, 2001. (d) Taoka, N.; Matsumoto, T. Inoue, K. U.S. Patent
6,174,707, 2001. (e) Boeston, W. H. J.; Broxterman, Q. B. Plaum,
M. J. M. U.S. Patent 6,133,002, 2000.
(5) Taylor, I. N.; Brown, R. C.; Bycroft, M.; King, G.; Littlechild, J. A.;
Lloyd, M. C.; Praquin, C.; Toogood, H. S.; Taylor, S. J. C. Biochem.
Soc. Trans. 2004, 32 (2), 290–292.
(3) Panella, L.; Aleixandre, A. M.; Kruidhof, G. J.; Robertus, J.; Ben, L.;
Feringa, B. L.; de Vries, J. G.; Minnaard, A. J. J. Org. Chem. 2006, 71
(5), 2026–2036.
(6) Lu, T.-J.; Yang, J.-F.; Sheu, L.-J. J. Org. Chem. 1995, 60, 2931–2934.
284
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development
10.1021/op100258j  2011 American Chemical Society
Published on Web 11/17/2010

Figure 1. (S)-Allysine ethylene acetal and examples of its reported uses as a chiral intermediate.
Scheme 1. Synthetic route to (S)-allysine ethylene acetal from crotonaldehyde
Scheme 2. Racemisation of (R)-N-benzoyl allysine ethylene acetal
as an azeotroping solvent to remove water. We found that
cyclohexane and toluene both worked as well although the
relative rates under the same reaction conditions using the
different solvents were not compared. The literature procedure
reported 76% crude GC yield with complete conversion of
crotonaldehyde in one hour. We achieved about 83% conversion
of crotonaldehyde in 12 h. After removing the excess ethylene
glycol by phase separation, neutralisation with NaHCO3, and
evaporation of volatiles under reduced pressure, crude product
was obtained as an orange free-flowing oil. Toluene proved to
be a slightly beneficial reaction solvent as it helped to remove
any unreacted crotonaldehyde prior to purification. Distillation
under reduced pressure produced >99% pure product in 54%
yield. The lower yield and longer reaction time were most likely
due to the much larger scale (about 140 times that described in
the literature), which required longer time to remove water by
azeotrope.
Stage 2: Hydroformylation. After an initial ligand screen
(see Supporting Information), Biphephos was chosen as the
optimum ligand on the basis of desired product yield, catalyst/
product separability, and overall process economics.
Investigation of the process parameters highlighted both
higher temperatures and lower pressures to be beneficial for
increased reaction rates and increased regioselectivity in favor
of the required linear aldehyde.
over the corresponding branched regioisomers (∼n:iso, 15:1).
In addition to the branched aldehydes several other byproducts
were observed in a total of up to 4%. These resulted from olefin
reduction yielding 2-propyl-1,3-dioxolane and isomerisation of
1 to 2-propylidene-1,3-dioxolane. Formation of the branched
aldehydes was of a greater concern due to the fact that the
branched aldehydes were more difficult to separate from the
linear aldehyde.
Vacuum distillation of the product from the reaction solution
was investigated and found to be ineffective due to formation
of heavies during the distillation in the presence of the catalyst.
Because the next step was to convert the linear aldehyde to
aminonitrile in an aqueous solution, on-scale an aqueous
extraction process was developed to isolate the aldehydes in
water, eliminating the need for vacuum distillation. The phase
separation was achieved by first removing the majority of the
THF and then adding hexane to the remaining reaction solution.
The resulting solution was extracted with an aqueous 0.1 M
NaHCO3 solution three times, with the total volume of water
used being calculated so as to obtain a solution containing
approximately 25 wt % aldehydes. A weak sodium bicarbonate
solution was used to ensure continued stability of the acetal
product if the solution were to be stored for prolonged periods
of time prior to performing the downstream chemistry. Residual
THF partitioned mostly in the aqueous phase; however, some
THF was extracted into the hexane phase, which enabled the
efficient removal of ligand and catalyst into this organic phase.
ICP analysis of both phases showed that >98% of the rhodium
could be recovered in this manner. This aqueous extraction
process also worked very well in isolating and purifying the
aldehydes, since any heavies produced stayed with the catalyst
Ultimately, tandem isomerisation/hydroformylation of cro-
tonaldehyde ethylene acetal (1) was efficiently achieved in a
50 wt % THF solution at 80 °C, 3 bar Syn gas (CO/H2 1:1),
and molar substrate to catalyst ratio of 4000:1 (0.025 mol %)
(Scheme 3). Under these conditions, glutaraldehyde monoeth-
ylene acetal (2) was obtained with a high degree of selectivity
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
285

Scheme 3. Hydroformylation of crotonaldehyde ethylene acetal
Scheme 4. Strecker reaction/formation of amino nitrile
process. The Strecker reaction was successfully carried out in
our laboratories at multihundred gram scale in 5-L reactors and
was subsequently transferred to the manufacturing plant where
five batches were successfully carried out in 10,000-L reactors,
generating multiton quantities of 3.
Stage 4: Synthesis of N-Acetyl Allysine Ethylene Acetal
(4a). Since the key bioresolution step in the synthesis of (S)-
allysine ethylene acetal (5) utilises an L-acylase enzyme, it was
initially envisaged that N-acetyl allysine ethylene acetal (4a)
would be an ideal substrate for the bioresolution. Synthesis of
4a was accomplished by treating 3 with sodium hydroxide in
50% aqueous ethanol and heating the resulting stirred solution
at 70 °C for 16 h (Scheme 5). The racemic amino acid product
was then acetylated in situ with acetic anhydride. Unfortunately,
it was discovered that 4a was highly soluble in water, and
consequently, isolation Via organic extraction only yielded
55-70% product.
Stage 5: Bioresolution of N-Acetyl Allysine Ethylene
Acetal (4a). Test bioresolutions were carried out using 4a and
a proprietary thermophilic L-acylase from archaeon Thermoc-
cocus litoralis. These experiments were carried out at pH 7 and
60 °C with a substrate concentration of 25 g/L (Scheme 6).
Within 24 h, conversion had exceeded 40%, and (S)-allysine
ethylene acetal was produced in greater than 99% ee.
Interestingly, it was noted that branched N-acetyl amino acid
impurities were not accepted as substrates by the enzyme.
Consequently, the bioresolution offered a potential route for
removing the branched impurities that had been produced by
the hydroformylation step. Unfortunately, these initial results
were obtained with an extremely high enzyme loading of 4658
units per gram of substrate. A series of bioresolutions were
carried out using isolated 4a in order to further improve enzyme
loading and substrate concentration. The results from these
experiments (Table 1) showed that the bioresolution can be
successfully carried out at greater than 80 g/L substrate
concentration with an enzyme loading of 1 mL (∼2000 units)
L-acylase solution per gram of substrate. Bioresolutions reached
greater than 40% conversion within 24 h, generating (S)-allysine
ethylene acetal with ee greater than 99%. Unfortunately,
isolation of product from starting material proved to be very
difficult as a result of the high water solubility of the N-Ac
material. An additional concern was the high enzyme loading
used in the bioresolution, which raised the question of the
solution and also enhanced the n:iso ratio (up to 24:1) by
preferentially extracting some of the branched aldehydes into
the organic phase. This process was scaled up with multiple
batches being performed in a 300-L pressure vessel that had
been appropriately modified for hydroformylation. In each case,
the reaction was found to be reproducible with negligible
variation in both isolated yield and regioselectivity.
Stage 3: Strecker Reaction. The mixture of linear and
branched aldehydes (15:1) isolated from lab-scale hydroformyl-
ations were reacted with potassium cyanide in the presence of
35% ammonia solution to generate a 75% yield of predomi-
nantly linear amino nitrile product (3) (Scheme 4). No signifi-
cant change in the linear:branched ratio was observed during
this step.
Since the product 2 from stage 2 was isolated as an aqueous
solution, it was decided that the process could be streamlined
by using this aqueous solution directly in the Strecker reaction.
The aldehyde content of the aqueous solution was determined
by GC analysis to be between 20 and 30 wt %, and the progress
of the reaction was also monitored by GC. Initial Strecker
reactions were carried out over 18-20 h with an aldehyde
concentration of 6 wt %; however, subsequent development
work confirmed that by using an aldehyde concentration of ∼25
wt % the reaction was complete within 5 h. Temperature control
of the reaction during addition of 2 was found to be unnecessary.
An endotherm resulted from the dissolution of ammonium
chloride and potassium cyanide in ammonia solution resulting
in a reactor temperature of 13 °C. Slow addition of the aqueous
aldehyde solution meant that reaction temperature remained
below 30 °C. During laboratory-scale preparation of 3, potas-
sium cyanide was used as the cyanide source for the reaction;
however, during the transfer of the process to the manufacturing
plant it was necessary for reasons of cost, safety, and ease of
operation to use a 30 wt % aqueous solution of sodium cyanide.
The change in cyanide source had no detrimental effect on the
286
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Scheme 5. Hydrolysis and acetylation of amino nitrile
Scheme 6. L-Acylase-catalysed bioresolution of N-acetyl
were successfully hydrolysed; since solutions of 3 were highly
alkaline, it was possible to reduce the quantity of sodium
hydroxide from 4.2 mol equiv to 1.5 mol equiv. The use of
ethanol in the hydrolysis step was rapidly eliminated during
development work, and hydrolysis of 3 was found to be
complete after heating at 70 °C for 16 h. A key issue identified
during development was the need to remove all ammonia from
the reaction mix prior to the benzoylation step. Any residual
ammonia present was found to react with benzoyl chloride and
generate benzamide impurity. The benzoylation was eventually
carried out at 10 °C, and the addition of benzoyl chloride was
at such a rate as to ensure that the reaction temperature did not
exceed 30 °C; the pH could be maintained between 7 and 9 by
the addition of 46/48 wt % sodium hydroxide solution. The
benzoylation was found to reach completion within 4 h, and
product was isolated by acidification of the aqueous reaction
mixture, extraction into ethyl acetate, and removal of solvent
under reduced pressure to yield 4b as a white solid. Analysis
of the isolated product revealed that it was essentially free of
any of the branched impurities that had been produced during
the hydroformylation step. Unfortunately, during scale-up it was
noted that 4b exhibited poor solubility in ethyl acetate and had
a tendency to precipitate out of solution post extraction. Whilst
this issue was not a problem at lab scale, it did present a
significant issue at manufacturing scale. To overcome the
problem, an alternative isolation procedure was developed,
involving acidification of the reaction mixture to pH 4 and
recovery of the precipitated product by filtration. The resulting
filter cake was washed with ethyl acetate and water in order to
remove any potential impurities and dried in Vacuo to remove
traces of solvent. A subsequent production campaign was run
in five batches in 10,000-L vessels.
Stage 5: Bioresolution of N-Benzoyl Allysine Ethylene
Acetal (4b). A trial bioresolution was carried out on 500-mg
scale at a substrate concentration of 89 g/L of N-benzoyl allysine
ethylene acetal and an enzyme loading of 657 units per gram.
The bioresolution proved to be successful, yielding (S)-allysine
ethylene acetal with an ee > 99%, at a conversion of ap-
proximately 50% within a 19 h reaction time (Scheme 8). Initial
trial experiments also demonstrated that the bioresolution of
4b could be accomplished at significantly lower enzyme
loadings than for the bioresolution of 4a (Table 1). Furthermore,
through the use of 4b we observed a cleaner and more effective
separation of product from starting material, allowing for the
potential recycle and racemisation of (R)-4b Via an intermediate
azlactone (Scheme 2).
allysine ethylene acetal
Table 1. Optimisation of L-acylase catalysed bioresolution of
4a and 4b
substrate
enzyme
ee
reaction
time
loading loading (units/ conversion (S)-5
(g/L)
g substrate)
(%)
(%)
(h)
4a
4a
4a
4a
4b
4b
4b
4b
23
77
4658
4529
2531
2329
657
57
45
45
41
50
50
48
49
49
>99
>99
>99
>99
>99
>99
>99
>99
24
19
19
19
18
18
4
58
83
89.5
261
201
265
97
97
5
Scheme 7. Hydrolysis and benzoylation of amino nitrile
economical viability of a process using this resolution. As a
result of these concerns, alternative bioresolution substrates were
considered. Past experience had shown us that the Thermoc-
cocus litoralis L-acylase exhibits a preference for N-benzoyl
substrates over N-acetyl substrates, and consequently, the
preparation of 4b was undertaken.
Stage 4: Synthesis of N-Benzoyl Allysine Ethylene Acetal
(4b). Synthesis of 4b was accomplished by treatment of 3 with
sodium hydroxide in 50% aqueous ethanol and by heating the
resulting stirred solution at 70 °C for 16 h, followed by
benzoylation of the resulting amino acid under Schotten-
Baumann conditions with benzoyl chloride (Scheme 7). Product
was isolated by acidification of the aqueous reaction mixture,
extraction into ethyl acetate and removal of solvent under
reduced pressure to yield 4b as a white solid.
The high yields and crystalline nature of 4b offered a good
opportunity to streamline the process and carry out stages 3
and 4 of the synthesis in one pot. The crystalline nature of the
product meant that any impurities carried through from earlier
stages could potentially be removed. Consequently, laboratory
experiments were carried out to validate this proposed approach.
Aqueous solutions of amino nitrile 3 from the Strecker reaction
Significant optimisation work on the bioresolution step was
undertaken, resulting in an improvement in substrate concentra-
tion from 89 g/L to over 260 g/L and a reduction in enzyme
loading from 657 units per gram of substrate to 57 units per
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
287

Scheme 8. L-Acylase-catalysed bioresolution of N-benzoyl allysine ethylene acetal
gram of substrate. The bioresolution was successfully carried
out at multihundred gram scale in a 5-L reactor using these
optimum conditions, leading to complete resolution within 18 h.
However, during transfer of the process to the manufacturing
plant, it was decided to carry out the reaction at the higher
enzyme loading of 97 units per gram in order to benefit from
the quicker reaction time (5 h cf. 18 h). Isolation of amino acids
from aqueous solutions is frequently problematic and often
results in poor recoveries with the product contaminated with
inorganic salts. Initial attempts to recover (S)-allysine ethylene
acetal from the bioresolution mixture involved filtration through
a plug of Celite to remove any precipitated protein material
followed by concentration of the aqueous solution to ap-
proximately one-third volume. Addition of two volumes of
ethanol led to precipitation, and after cooling to 10 °C, the
product was recovered by filtration, washed with further cold
ethanol, and dried in Vacuo to give a 40% isolated yield.
Reproducing this yield proved difficult, and unacceptably high
salt content resulted in the evaluation of alternative antisolvents
to aid product recovery. Isopropanol was briefly looked at, but
the recovered precipitate contained significant contamination
with N-benzoyl allysine ethylene acetal and benzoic acid. A
switch to 1,2-dimethoxyethane (DME) was the key break-
through and marked a switch to isolating material Via recrys-
tallisation rather than a precipitation approach. The aqueous
bioresolution solution was partially concentrated under reduced
pressure and stirred at 60 °C. DME (2 volumes) was slowly
added taking care to ensure that no product precipitation
occurred during the addition. The resulting solution was
gradually cooled to 10 °C and left stirring for 12 h in order to
allow product to crystallise. (S)-allysine ethylene acetal was
subsequently isolated as a white crystalline solid in good yield
(>35%) and excellent chemical and optical purity (>99%).
campaign that produced multihundred kilograms of this valuable
chiral intermediate.
Experimental Section
Materials and Methods. All the materials were purchased
from commercial suppliers and used without further purification.
L-Acylase enzyme solution was produced via a generic fed-
batch fermentation and subsequent standard downstream pro-
cessing.5 Optical rotations were measured on a Perkin-Elmer
341 polarimeter, and data are reported as [R]25D (concentration
in g/100 mL, solvent). 1H and 13C NMR spectra were recorded
on a Bruker Avance 400 spectrometer. Chemical shifts are
reported in ppm relative to the residual deuterated solvent. GC
analysis was performed on a Perkin-Elmer Autosystem XL
equipped with a FID detector. HPLC analysis was carried out
on an Agilent 1100 system equipped with a UV detector.
Crotonaldehyde Ethylene Acetal (1). A 3-L three-neck
flask was fitted with a Dean-Stark trap, an overhead stirrer, a
temperature probe, and a gas inlet on the condenser. The flask
was then placed under an atmosphere of nitrogen by slowly
purging for 30 min. Tartaric acid (6.41 g, 42.7 mmol) was then
charged to the flask. Crotonaldehyde (599.1 g, 8.55 mol) and
toluene (700 mL) were then transferred into the flask followed
by ethylene glycol (636.7 g, 10.26 mol). No exotherm was
observed. The solution was heated to reflux using an oil bath
at 120 °C (internal temperature of bulk ) 102 °C) until
collection of the aqueous phase slowed significantly (ap-
proximately 12 h). A total of 165 g of aqueous phase was
collected (GC analysis showed this to contain small quantities
of crotonaldehyde, toluene, and ethylene glycol). Stirring was
stopped to allow the solution to settle into two phases, a
colourless top phase containing the product and a yellow bottom
phase containing most of the excess ethylene glycol and tartaric
acid. After removal of the yellow bottom phase and back
extraction with toluene (2 × 100 mL) to extract any residual
product, the toluene washes were combined with the top organic
phase. Analysis of this solution by GC prior to distillation
showed crotonaldehyde conversion to be approximately 83%.
Solid NaHCO3 (10.1 g, 120 mmol) was added to ensure
neutralisation of any residual acid prior to distillation. (NB: This
is essential to prevent the appearance of crotonaldehyde Via the
back reaction during distillation.) The above procedure was
repeated twice, and the batches were combined prior to
fractional distillation. Initial volatiles (unreacted crotonaldehyde
and toluene, together with approx 10% w/w of the desired
acetal) were removed on a rotary evaporator to leave a crude
orange-coloured solution. Distillation was performed under
reduced pressure in a 3-L round-bottomed flask using a glass
column packed with stainless steel Knitmesh (diameter ) 30
mm, height ) 325 mm). Distillation of 1308 g of the crude
reaction mixture gave an initial fraction consisting of mainly
toluene with a trace amount of crotonaldehyde and the desired
Conclusions
A robust five-step process for the production of (S)-allysine
ethylene acetal has been developed using a combination of two
catalytic technologies, namely rhodium-catalysed tandem isom-
erisation/hydroformylation and bioresolution. Starting from
crotonaldehyde, the synthesis involved acetal formation with
ethylene glycol followed by linear selective isomerisation/
hydroformylation to yield predominantly glutaraldehyde mo-
noethylene acetal. After an aqueous extraction to remove and
recover the transition metal catalyst, a Strecker reaction followed
by hydrolysis and benzoylation afforded the appropriate N-
substituted racemic amino acid. Bioresolution of this racemate
using a thermophilic L-acylase from archaeon Thermoccocus
litoralis not only enabled separation of the linear product from
the corresponding branched isomers but ultimately provided
material with >99% ee in >99% purity. The development of
this commercial process led to a large-scale manufacturing
288
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

product (168.1 g, boiling point: 38 °C at 72 mbar, oil bath temp
) 60 °C, internal bulk temperature ) 43 °C). After discarding
a small forerun (∼5 g), the final product was isolated as the
next fraction (820 g, boiling point: 66-74 °C at 70 mbar, oil
bath temp ) 95 °C, internal bulk temperature ) 82 °C),
analysed by GC and 1H NMR, and found to contain the desired
acetal as a mixture of cis- and trans-isomers. The distillation
was stopped with 309.9 g of an orange residue remaining that
GC showed to contain undistilled product and unidentified
heavies. 1H NMR for the major trans-isomer (400 MHz, CDCl3)
δ 5.83 (dq, JH-H ) 15.5, 6.6 Hz, 1H). 5.38 (ddq, JH-H ) 15.5,
8.7, 1.6 Hz, 1H), 5.05 (d, JH-H ) 8.7, 1H); 3.71-3.92 (m, 4H,),
1.63 (dd, JH-H ) 6.6, 1.6 Hz. 3H). GC sample preparation: Dilute
2 µL to 1 mL with acetonitrile. DB5 30 m × 0.25 mm ×0.25
µm, oven program 60 °C for 8 min, then 10 °C/min to 250 °C
(hold 3 min). Retention times (relative response factors):
crotonaldehyde 2 min (1.0), ethylene glycol 2.2 min (0.4),
toluene 2.9 min (1.7), trans-acetal 5.4 min (0.9), cis-acetal 5.5
min (0.9).
were mixed, allowed to separate into a triphasic mixture
consisting of a top hexane phase, a middle organic phase, and
a lower aqueous phase. The lower aqueous phase was separated
off. The remaining top two organic phases were then extracted
twice more with further quantities of 0.1 M NaHCO3 (aq) (2 ×
370 g), and the aqueous phases were combined. GC analysis
of this aqueous phase showed this to contain 25 wt % aldehyde
content.
2-Amino-5-[1,3]dioxolan-2-yl-pentanenitrile (3). Into a
10-L jacketed vessel was placed ammonium chloride (92.86 g,
1.74 mol) and potassium cyanide (113.05 g, 1.74 mol) in water
(2.25 L) and 35% aqueous ammonia (2 L). The solution was
stirred under a nitrogen atmosphere and cooled to below 0 °C.
To the cooled solution was slowly added the crude reaction
mixture of 4-[1.3]-dioxolan-2-yl butanal (2) (250 g, 1.74 mol)
over a period of 60 min, taking care that the temperature
remained below 0 °C. After the addition was complete, the
reaction mixture was allowed to slowly warm up to room
temperature and was left stirring overnight. After overnight
stirring the reaction was halted and extracted with ethyl acetate
(3 × 2.5 L). The combined organics were then dried over
magnesium sulphate and concentrated under reduced pressure
to yield 3 as a yellow oil (216 g, 73%). 1H NMR (400 MHz,
CDCl3) δ 4.88 (t, J ) 4.4 Hz, 1H), 4.00-3.92 (m, 2H),
3.91-3.81 (m, 2H), 3.69 (t, J ) 6.8 Hz, 1H), 1.85-1.76 (m,
2H), 1.76-1.54 (m, 4H); 13C NMR (100 MHz, CDCl3) δ
122.41, 104.34, 65.32, 43.70, 35.15, 33.33, 20.26.
N-Acetyl Allysine Ethylene Acetal (4a). Into a 50-mL
round-bottom flask was placed 3 (725 mg, 4.3 mmol) in 50%
aqueous ethanol (20 mL), together with sodium hydroxide (720
mg, 18 mmol, 4.2 equiv). The reaction mixture was stirred
continuously and heated under refluxing conditions for 4 h. After
this time the mixture was cooled to below 5 °C by means of a
salt-ice bath, and acetic anhydride (490 µL, 5.11 mmol, 1.2
equiv) was added to the stirred solution. The reaction mixture
was allowed to warm up to room temperature and was left
stirring overnight. The mixture was then acidified to pH 3 with
6 M hydrochloric acid solution and extracted with ethyl acetate
(3 × 50 mL). The combined extracts were dried over magne-
sium sulphate and concentrated under reduced pressure to yield
530 mg (55%) of the N-acetyl allysine ethylene acetal (4a) as
an orange oil. 1H NMR (400 MHz, CD3OD) δ 4.84 (t, J ) 4.6
Hz; 1H), 4.37 (dd, J ) 5.2, 9.2 Hz, 1H), 4.00-3.89 (m, 2H),
3.89-3.79 (m, 2H), 2.01 (s, 3H), 1.95-1.83 (m, 1H), 1.81-1.59
(m, 3H), 1.59-1.43 (m, 2H); 13C NMR (100 MHz, CD3OD) δ
175.94, 173.78, 105.74, 66.26, 54.08, 34.72, 32.91, 22.71, 21.80.
(S)-Allysine Ethylene Acetal (5) Wia 4a. Into a 50-mL
jacketed vessel was placed 4a (500 mg, 2.16 mmol) dissolved
in distilled water (20 mL). The solution was continuously stirred
and heated to 60 °C; the pH of the solution was adjusted to 7
by dropwise addition of 1 M sodium hydroxide solution.
L-Acylase enzyme solution (1 mL, 2329 units) was then added
to the reaction, and the mixture was left stirring for 24 h at 60
°C. The reaction mixture was then acidified to pH 3 with 6 M
hydrochloric acid solution and extracted with ethyl acetate (3
× 25 mL) in order to remove residual N-acetyl allysine ethylene
acetal. The pH of the aqueous phase was readjusted to 7.3 using
5 M sodium hydroxide solution, was concentrated to one-quarter
4-[1.3]-Dioxolan-2-yl Butanal (2). Crotonaldehyde ethylene
acetal (1) (533 g, 4.67 mol) was charged to a 2-L Parr pressure
vessel fitted with a glass liner, double impeller, injection port,
bursting disk, pressure relief valve, and pressure gauge. THF
(533 mL) was added and the vessel assembled. The reaction
solution was deoxygenated by pressurising with nitrogen (6 bar),
stirring vigorously (1000 rpm), and venting. This procedure was
repeated five times. In the meantime, Rh(CO)2(acac) (301 mg,
1.17 mmol) and Biphephos (1.10 g, 1.40 mmol) were charged
to a glass liner of a 50 mL Parr vessel and the vessel assembled.
This was purged 6 times by pressurising to 4 bar with Syn gas
(CO/H2 1:1) and venting. Degassed THF (10 mL) was added
through a septum and the vessel charged to 10 bar with Syn
gas. This was heated to 40 °C for 30 min with rapid stirring
and then the activated catalyst solution removed into a syringe
and transferred to the 2-L Parr vessel. The 2-L Parr vessel was
then charged to 3 bar with Syn gas, heated to 80 °C, and stirred
vigorously (1000 rpm), maintaining the desired pressure with
constant addition of Syn gas until gas uptake ceased. The vessel
was cooled to room temperature, purged with nitrogen, and
sampled. 1H NMR spectroscopy and GC analysis showed the
reaction to be complete (<1% starting material), yielding a
mixture of the desired linear aldehyde product (approximately
90%), the corresponding branched regioisomers (6%), 2-propyl-
1,3-dioxolane and 2-propylidene-1,3-dioxolane (∼4% in total).
The solution was concentrated on a rotary evaporator at 45 °C
1
to give a clear, mobile, yellow oil. H NMR for the linear
product (400 MHz, CDCl3) δ 9.71 (s, 1H), 4.80 (d, JH-H ) 3.3
Hz. 1H), 3.76-3.94 (m, 4H), 2.45 (m, 2H), 1.51-1.74 (m, 4H).
13C NMR for the linear product (100 MHz, CDCl3) δ 202.0,
103.9, 64.5, 43.4, 32.8, 16.3. GC sample preparation: Dilute 2
µL to 1 mL with acetonitrile. DB5 30 m × 0.25 mm ×0.25
µm, Oven program 60 °C for 8 min, then 10 °C/min to 250 °C
(hold 3 min). Retention times: THF 1.9 min, trans-acetal 5.4
min, cis acetal 5.5 min, branched aldehydes 13.1 and 13.2 min,
linear aldehyde (2) 14.4 min.
In preparation of the aqueous solution, hexane was added
(171 g) to the concentrated reaction mixture (438 g) together
with a 0.1 M aqueous solution of NaHCO3 (370 g). The phases
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
289

volume and diluted with isopropanol (20 mL). The resulting
white precipitate was recovered by filtration and dried overnight
in a vacuum oven, yielding 150 mg (36%) of (S)-allysine
ethylene acetal. The product was analysed by HPLC on chiral
stationary phase (Chirex (D)-penicillamine 50 mm × 4.6 mm;
5 µm; 30 °C; 0.5 mL/min methanol/2 mM copper sulphate
pentahydrate in water 5:95; 254 nm). Retention times: (S)-
allysine ethylene acetal ) 14 min and (R)-allysine ethylene
(600 mL) and 2 M sodium hydroxide solution (725 mL) giving
a pH of 7.5. The reaction mixture was continuously stirred, and
L-acylase enzyme solution (19 mL, 44,251 units) was added.
Stirring was continued at 65 °C for 5 h after which time NMR
analysis indicated that the reaction had reached ∼50% conver-
sion. Concentration under reduced pressure removed water (600
mL). The remaining aqueous phase was returned to the 5-L
vessel and heated to 60 °C, and DME (1.5 L) was then added
via a dropping funnel to the reaction mix at 60 °C. Precipitation
started to occur, and the mixture was stirred at this temperature
for 40 min. The stirred solution was cooled to 10 °C and was
left stirring overnight at this temperature. The resulting pre-
cipitated amino acid was recovered by filtration, washed with
further DME (400 mL), and dried in Vacuo to yield 105.2 g
(36%) of white crystalline solid. Mp >200 °C dec; [R]25D +3.7°
(c 1, H2O); >99% ee; 1H NMR (400 MHz, D2O) δ 4.84 (t, J )
4.8 Hz, 1H), 4.06-3.97 (m, 2H), 3.97-3.83 (m, 2H), 3.63 (t,
J ) 6.2 Hz; 1H), 1.99-1.80 (m, 2H), 1.78-1.63 (m, 2H),
1.59-1.34 (m, 2H); 13C NMR (100 MHz, D2O) δ 175.00,
104.42, 64.85, 54.86, 32.46, 30.52, 19.32.
Racemisation of (R)-N-Benzoyl Allysine Ethylene Acetal
(4b). Into a two-neck 100-mL round-bottom flask was placed
(R)-N-benzoyl allysine ethylene acetal (5 g, 17 mmol) and
sodium acetate (200 mg, 2.4 mmol) in ethyl acetate (25 mL).
The solution was stirred at 80 °C, and acetic anhydride (2.5
mL, 26.5 mmol, 1.5 equiv) was slowly introduced into the
reaction vessel. The reaction mixture was then continuously
stirred at 80 °C for a further 3 h. After this time the vessel was
allowed to cool to room temperature, and the contents were
extracted with saturated sodium bicarbonate solution (20 mL).
To the organic phase was then added 5 M sodium hydroxide
solution (40 mL). The resulting mixture was vigorously stirred
for 1 h, after which time the phases were separated, and the
aqueous phase was acidified to pH 3 with 6 M hydrochloric
acid solution. The acidified solution was then extracted with of
ethyl acetate (2 × 40 mL), combined extracts were dried over
magnesium sulphate, and concentration under reduced pressure
yielded 3.3 g of an off-white solid. The material was then
slurried in MTBE to remove benzoic acid impurities, filtered,
and dried in Vacuo to yield 2.74 g (60%) of racemic N-benzoyl
allysine ethylene acetal.
1
acetal ) 25 min. (S)-allysine ethylene acetal ee ) 99%. H
NMR (400 MHz, D2O) δ 4.84 (t, J ) 4.8 Hz, 1H), 4.06-3.97
(m, 2H), 3.97-3.83 (m, 2H), 3.63 (t, J ) 6.2 Hz; 1H),
1.99-1.80 (m, 2H), 1.78-1.63 (m, 2H), 1.59-1.34 (m, 2H);
13C NMR (100 MHz, D2O) δ 175.00, 104.42, 64.85, 54.86,
32.46, 30.52, 19.32.
N-Benzoyl Allysine Ethylene Acetal (4b). Into a 5-L
jacketed vessel was placed 35% aqueous ammonia solution (860
mL). To the stirred solution was added ammonium chloride
(159.84 g, 2.99 mol) and potassium cyanide (194.6 g, 2.99 mol).
The resulting suspension was left stirring for 10 min at 15 °C.
An aqueous solution of 2 (1760 mL containing 430 g or 2.99
mol aldehyde) was then added dropwise to the reaction mixture
over a period of 45 min. The temperature rose to 30 °C, and
the reaction mixture became homogeneous. The reaction was
left stirring at ambient temperature for 4 h. GC and NMR
analyses confirmed that all starting material had been consumed.
The reaction mixture was transferred into a three-neck, 5-L
round-bottom flask, and sodium hydroxide (220 g, 5.5 mol, 1.8
equiv) was added. The reaction mixture was heated under
refluxing conditions (70 °C pot temperature) for 15 h. NMR
analysis confirmed the formation of amino acid and the absence
of any amino nitrile. Excess ammonia was then removed under
reduced pressure. A total of 1000 mL of ammonia/water was
removed, and no odor of ammonia was detected in the reaction
mix. The concentrated reaction mixture was transferred into a
5-L jacketed vessel and diluted with 500 mL of deionised water,
and the pH was adjusted to 11.5 by addition of concentrated
hydrochloric acid (230 mL). After cooling to below 15 °C,
benzoyl chloride (248 mL, 2.13 mol, 0.71 equiv) was added
dropwise to the vigorously stirred reaction mixture over a period
of 20 min. The pH was maintained between 7 and 9 by multiple
additions of 46-48% sodium hydroxide solution (105 mL). The
reaction temperature rose to 29 °C. The mixture was washed
with 500 mL of ethyl acetate; this wash was discarded. The
aqueous phase was acidified to pH 3.5 with concentrated HCl
(150 mL) and extracted with a total of 2.5 L of ethyl acetate.A
total of 410 g of N-benzoyl allysine ethylene acetal was isolated.
1H NMR (400 MHz, DMSO-d6) δ 12.59 (br s, 1H), 8.59 (d, J
) 7.6 Hz, 1H), 7.89 (d, J ) 6.8 Hz, 2H), 7.59-7.42 (m, 3H),
4.76 (t, J ) 4.6 Hz, 1H), 4.41-4.30 (m, 1H), 3.90-3.79 (m,
2H), 3.79-3.67 (m, 2H), 1.92-1.71 (m, 2H), 1.67-1.33 (m,
4H); 13C NMR (100 MHz, DMSO-d6) δ 174.07, 166.85, 134.33,
131.69, 128.58, 127.79, 103.84, 64.51, 52.90, 33.25, 30.80,
21.03.
Acknowledgment
We thank Cynthia Rand and Ed Daugs for useful discus-
sions, Matthew Bycroft for overseeing enzyme production,
Suzanne Riley and Dean Millaird for analytical support, and
all of the production team in Mirfield for execution of the large-
scale commercial campaign.
Supporting Information Available
Hydroformylation ligand screening data, 1H and 13C NMR
spectra of key products and intermediates. This material is
available free of charge via the Internet at http://pubs.acs.org.
(S)-Allysine Ethylene Acetal (5) Wia 4b. Into a 5-L jacketed
vessel heated to 65 °C was placed N-benzoyl allysine ethylene
acetal (4b) (457.3 g, 1.56 mol) dissolved in a mixture of water
Received for review September 24, 2010.
OP100258J
290
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development