The enzymatic resolution of an α-fluoroamide by an acylase

Article abstract of DOI:10.1016/S0022-1139(99)00284-5

The acylase from Aspergillus melleus was able to hydrolyse the amide bond of (S)-phenylalanine-N-2-(R,S)-fluoropropionamide and discriminate the diastereoisomers such that the (S,S)-diastereoisomer was hydrolysed by an order of magnitude faster than the (S,R)-diasteroisomer. The origin of the kinetic discrimination is attributed to both binding and kinetic effects.

Full text of DOI:10.1016/S0022-1139(99)00284-5

Journal of Fluorine Chemistry 102 (2000) 235±238
The enzymatic resolution of an a-¯uoroamide by an acylase
John W. Banksa, David O'Haganb,*
aChemical Technology Department, Searle, Division of Monsanto, Whalton Rd., Morpeth, NE61 3YA, UK
bDepartment of Chemistry, Science Laboratories, University of Durham, South Road, Durham, DH1 3LE, UK
Received 30 June 1999; received in revised form 19 July 1999; accepted 3 August 1999
Dedicated to Professor Paul Tarrant on the ocassion of his 85th birthday
Abstract
The acylase from Aspergillus melleus was able to hydrolyse the amide bond of (S)-phenylalanine-N-2-(R,S)-¯uoropropionamide and
discriminate the diastereoisomers such that the (S,S)-diastereoisomer was hydrolysed by an order of magnitude faster than the (S,R)-
diasteroisomer. The origin of the kinetic discrimination is attributed to both binding and kinetic effects. # 2000 Elsevier Science S.A. All
rights reserved.
Keywords: 19F-NMR; Acylase; Aspergillus melleus, a-¯uoroamide; Enzyme resolution
It is part of the culture of bio-organic chemistry that the
¯uorine atom is considered to have a steric in¯uence just
slightly greater than that of a hydrogen atom [1]. However,
electronically ¯uorine and hydrogen are clearly very differ-
ent. It is also part of the culture of bio-organic chemistry that
lipase and acylase enzymes distinguish substrates through
binding interactions which are largely steric in nature. Steric
models for the common hydrolytic enzymes have been
developed where `small', `medium' and `large' pockets
have been identi®ed which de®ne the chirality of the pre-
ferred enantiomer of a given substrate enantiomer [2,3]. In
general, these models have served well as predictive tools.
On the other hand, if an a-¯uoroester or a-¯uoroamide is
presented as a substrate to a hydrolytic enzyme, then the
compound is chiral by virtue of the substitution of a hydro-
gen by a ¯uorine. It is now not so obvious, on the basis of
established steric models, to predict the outcome of such a
kinetic resolution, as the ¯uorine and hydrogen are both
`small' and a distinction based on steric considerations
cannot be made, which predicts how these groups will best
®t the enzyme surface.
mer and on work up the (R)-ester was isolated in 99% ee
after 60% conversion, as illustrated in Fig. 1. Such a dis-
crimination cannot easily be accounted for on the basis of a
steric difference between hydrogen and ¯uorine, and we
have previously suggested that a stereo-electronic effect is
responsible for this discrimination [1,6]. The enzyme attacks
the ester, via a serine alkoxide nucleophile and becomes
acylated. The preferred transition state has the serine alk-
oxide attacking the carbonyl anti to the vicinal ¯uorine
atom. The overlap of the C±F s* orbital with the carbonyl
LUMO stabilises this molecular orbital in transition state A
relative to transition state B, and the electron density of the
serine nucleophile can donate into the C±F s* orbital only in
transition state A, lowering the energy of anti approach to
the ¯uorine atom. Also, the electrostatic repulsion between F
and O is minimised in transition state A. Calculations
1
suggest up to a 2.5 kcal mol difference in the energy of
these transition states, and there is a suf®cient energy
difference to account for the observed kinetic resolution.
In this paper, we report the ®rst study on the enzymatic
resolution of an a-¯uoroamide with an acylase. We were led
to this investigation after a recent evaluation of the preferred
conformation of N-alkyl a-¯uoropropionamides [7]. A num-
ber of years ago, both electron [8] and neutron diffraction [9]
analyses have demonstrated that the major conformer of
mono¯uoroacetamide (FCH2C(O)NH2) has the C±F bond
anti to the C=O bond and syn to the N±H bond of the amide
in the solid state as illustrated in Fig. 2. The energy
difference between this and the next favoured syn confor-
mation was calculated to be 7.5 kcal mol 1. In our recent
Examples of successful kinetic resolutions with a-¯uor-
oesters are rare [4,5], however, the following one [4] serves
to illustrate that certain enzymes can distinguish ¯uorine and
hydrogen at the stereogenic centre.
Lipase P30 was able to selectively hydrolyse the (S)-
enantiomer of ethyl 2-¯uorohexanoate over the (R)-enantio-
* Corresponding author. Tel.: ‡44-191-374-2000.
E-mail address: david.o'hagan@durham.ac.uk (D. O'Hagan).
0022-1139/00/$ ± see front matter # 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 0 2 2 - 1 1 3 9 ( 9 9 ) 0 0 2 8 4 - 5

236
J.W. Banks, D. O'Hagan / Journal of Fluorine Chemistry 102 (2000) 235±238
Fig. 1. Rationale for the stereoselectivity observed during the lipase mediated hydrolysis of a racemic a-fluoroester.
This substrate was chosen as the diastereoisomers are
distinguised only by the ¯uorine atom at the stereogenic
centre. Incubation of this substrate with the acylase from
Aspergillus melleus resulted in a slow hydrolysis. The
relative rates of hydrolysis of the two diastereoisomers
could be followed analytically, in a very straightforward
manner by 19F-NMR, as each diastereoisomer had a unique
and resolvable 19F-NMR resonance ((S,S) 182.1 ppm and
(S,R) 181.67 ppm). The enzymatic reaction was moni-
tored through time in an NMR tube, and the rate slowed
down considerably as the reaction approached 50% con-
version. The reaction was stopped at 57% conversion where
the ratio of isomers was 98 : 2, representing a diastereo-
meric excess of the residual substrate of 96% de. The major
diastereoisomer was (S,R), the diastereoisomer which is
hydrolysed more slowly by the enzyme (Scheme 2).
In order to gain some insight into the origin of the
observed selectively, Km and Vmax data for each diaster-
eoisomer were obtained in separate experiments. This
required the preparation of each diastereomer in an enan-
tiomerically enriched form. This was achieved by preparing
(S)- and (R)-2-¯uoropropionic acids directly from (S)- and
(R)-alanines by diazotisation and in situ treatment with HF-
pyridine [10]. The diazotisation reaction is reported to
proceed with predominant retention of con®guration
[11,12]. In the event, it emerged that the product 2-¯uor-
opropionic acids were generated in an enantiomeric excess
of 80% ee from the enantiomerically pure alanines. This
enantioselectivity was established by recording 19F-NMR
of (S)-2-(diphenylmethyl)pyrrolidine salts of the resultant
Fig. 2. Conformation of fluoroacetamide.
study [7], we have investigated the preferred conformation
of N-alkyl a-¯uoropropionamides by both X-ray crystal-
lography and ab initio calculations. It is found that the trans
conformation is again favoured in the solid state for these
amides where the C±F bond lies anti to the C=O bond as
shown in Fig. 3. Calculations revealed that this was the only
minimum on the rotational energy pro®le (around the FC±
C(O) bond) and that the barrier to rotation was large at
8.0 kcal mol 1. The ¯uorine atom and not the methyl group
dictates the conformation of these amides.
It follows that with such a conformational preference,
where the C±F and N±H bonds are syn and co-planar,
enantiomers of a-¯uoropropionamides will necessarily have
to locate their methyl groups in opposite spacial locations
around the stereogenic centre (Fig. 3). This is anticipated to
impose signi®cantly different diastereomeric interactions
during the binding of such enantiomers to an enzyme. With
this in mind, we have synthesised (S)-phenylalanine N-
(R,S)-2-¯uoropropionamide as a substrate for acylase
hydrolysis as shown in Scheme 1.
Scheme 1. (i) NaNO2, HF pyridine, 14%; (ii) SOCl2, 83%; (iii) Aq.
NaOH, 16%.
Fig. 3. Conformation of a-fluoropropionamides.

J.W. Banks, D. O'Hagan / Journal of Fluorine Chemistry 102 (2000) 235±238
237
Scheme 2
2-¯uoropropionic acids in CDCl3 [13]. The enantiomeri-
cally enriched 2-¯uoropropionic acids were converted to
their acid chlorides after treatment with thionyl chloride and
were then coupled directly to (S)-phenylalanine. This gen-
erated diastereomerically enriched compounds of 80% de, a
value which was readily determined by integration of the
two ¯uorine signals in the 19F-NMR spectrum, associated
with each diastereoisomer. In the case of the (S,R) stereo-
isomer, the material was stereochemically puri®ed by an
initial short bio-transformation with the A. melleus acylase,
to hydrolyse the 10% (S,S) isomer and the diastereomeri-
cally pure (S,R) substrate was recovered as its sodium salt.
The (S,R) stereoisomer was enantiomerically pure within
the limits of 19F-NMR detection. For the (S,R) stereoisomer,
improvement of the 80% de was not so readily achievable
and this material was used directly for the kinetic study
(Table 1).
Initial reaction rates (n) at different substrate concentra-
tions [S] were determined by 19F-NMR analysis, and then
plotted as Lineweaver±Burke plots (1/n versus 1/[S]) to
determine Vmax and Km for each substrate [14]. It emerged
that the (S,S) substrate has a Km of 16.26 mM and a Vmax of
0.92 mM min 1, whereas the (S,R) substrate had a Km of
40.34 mM and a Vmax of 0.20 mM min 1. Km is a measure
of substrate binding and V max a measure of the rate of
product formation (enzyme-substrate complex breakdown).
The overall ef®ciency of the enzyme for each substrate can
be estimated by comparing V/K values for each substrate.
These values emerge as V/K ˆ 0.0566 for the (S,S) sub-
strate, and V/K ˆ 0.005 for the (S,R) substrate. The study
predicts one order of magnitude difference in the rates of
hydrolysis in favour of the (S,S) substrate and the experi-
mental value of 96% de for the residual (S,R) diastereoi-
somer in the previous study is consistent with this analysis.
The Km values reveal that there is a differential binding
af®nity of the substrates by the A. melleus lipase. Also the
Vmax values indicate a differential rate of product formation,
which is perhaps related to the relative rates of enzyme
acylation/deacylation.
In summary, the A. melleus lipase can mediate an ef®cient
kinetic resolution between diastereoisomers of (S)-pheny-
lalanine-N-(R,S)-2-¯uoropropionamide. Although, acylases
have hardly been explored in such reactions, this initial
study suggests that they offer an excellent method for
preparing enantiomerically enriched 2-¯uorocarboxylic
acids and amides. The preferred conformation of such
amides, where the C±F and C=O bonds orientate anti to
each other, predicts a discernable binding difference
between the enzyme and each diastereomeric substrate.
Although, the data reveal that both substrates have Km
values in the high mM range, and are poor binders, a two
fold difference in binding af®nity emerges. Perhaps less
obviously, the Vmax data revealed a four fold difference
between the diastereoisomers, and re¯ects the difference in
the rate of product formation (or enzyme-substrate break-
down). Perhaps this latter difference lies in off-loading the
2-¯uoroacyl ester-enzyme complex by nucleophilic attack
by water, a process which should be susceptible to the
stereoelectronic effect discussed above for the lipase hydro-
lysis of a-¯uoroesters [1,6].
The overall kinetic pro®le of such enzyme resolutions are
complex and dif®cult to deconstruct as there are various
factors operating at different stages along the enzyme
reaction course, however, it emerges from this study that
conformational and stereoelectronic effects can combine to
result in highly stereoselective enzymatic resolutions with
mono-¯uorinated ester and amide substrates.
1. Experimental
FTIR spectra were recorded on a Nicolet Magna IR 550
Spectrometer. NMR spectra were obtained on a Jeol EX
270 MHz spectrometer in CDCl3 or D2O. Chemical shifts
1
are quoted relative to TMS for H- and 13C-NMR spectra
and 19F-NMR chemical shifts are quoted as negative relative
to ¯uorotrichloromethane. Solvents were dried and distilled
prior to use. Reactions requiring anhydrous conditions were
carried out under an atmosphere of nitrogen.
Table 1
Kinetic parameters for the hydrolysis of (S)-phenylalanine-N-2-fluoropro-
pionamides with Aspergillus melleus. The data are an average of two
separate studies in each case
Substrate
diastereomer
Km (mM)
Vmax
(mM min )
V/K
1.1. Preparation of (R,S)-2-fluoropropanoic acid
1
(S,S)a
(S,R)
16.26 Æ 1.84
40.34 Æ 3.21
0.92 Æ 0.07
0.20 Æ 0.02
0.0566
0.0050
(R,S)-Alanine, (8.6 g, 96 mmol) was added to a stirred
solution of pyridinium hydrogen ¯uoride (100 ml, 70% w/
w) in a Te¯on bottle. The solution was cooled to 08C and
a 80% de.

238
J.W. Banks, D. O'Hagan / Journal of Fluorine Chemistry 102 (2000) 235±238
sodium nitrite (22.6 g, 232 mmol) was added in ®ve equal
portions over 1 h. The reaction mixture was stirred at 08C
for 1 h and was then warmed to room temperature and then
stirred overnight. The reaction was quenched by addition to
ice water (100 ml). The aqueous solution was extracted into
diethylether (3 Â 100 ml), washed with sodium chloride
solution (3 Â 100 ml, 5% w/v) and dried over anhydrous
sodium sulphate. The organics were ®ltered and the solvent
was evaporated. The residue was distilled (508C, 23 mbar)
to afford the product (1.4 g, 14%) as a colourless oil.
dH(270 MHz, CDCl3) 5.12 (1H, dq, J 48.5, 6.9, CFH),
1.61 (3H, dd, J 23.5, 6.9, Me); dC(67.9 MHz, CDCl3)
85.1 (d, J 182.7), 18.2 (d, J 22.3); dF(254.2 MHz, CDCl3)
184.76 (dq, J 48.3, 23.5, CFH).
183.6, CFH), 88.3 (d, J 183.5, CFH), 52.6, 52.4, 37.4, 37.24,
18.3 (d, J 21.9), 18.1 (d, J 20.9); dF(254.2 MHz, CDCl3)
182.79 (ddq, J 49.2, 24.7, 4.2, CFH), 182.92 (ddq, J
49.8, 25.2, 4.3, CFH).
1.4. Acylase enzyme reaction
(S)-Phenylalanine-N-2-(R,S)-¯uoropropionamide
was
dissolved in 4 M NaOH (three equivalents) and the solution
adjusted to pH ˆ 7 with aq. HCl and then diluted to a known
volume of phosphate buffer (pH ˆ 7), to give a 50 mM
concentration of substrate in solution. Each diastereoisomer
was readily observable by 19F-NMR. dF(254.2 MHz,
CDCl3) 181.67 (dq, J 48.7, 25.2, CFH) (S,S)-diastereoi-
somer, 182.10 (dq, J 48.5, 25.2, CFH), (S,R)-diastereoi-
somer. The above solution of sodium (S)-phenylalanine-N-
2-(R,S)-¯uoropropionamide (0.2 ml) (for a 10 mM reaction)
and phosphate buffer, (pH ˆ 7, 0.6 ml) was placed in a
5 mm NMR tube. A solution (0.2 ml of a 25 mg/ml,
0.5 units/mg) of the A. melleus acylase (Sigma Chem.) in
phosphate buffer (pH ˆ 7) was added to initiate the reaction
and the reaction maintained at 258C and monitored by 19F-
NMR.
1.2. Preparation of (R,S)-2-fluoropropanoyl chloride
Thionyl chloride, (2.8 g, 23.5 mmol) was added dropwise
to heat (R,S)-2-¯uoropropanoic acid, (1.00 g, 10.9 mmol) at
room temperature. The reaction was heated to re¯ux for
1
60 min and a sample removed for H- and 19F-NMR ana-
lysis to con®rm the absence of starting material. The product
was distilled (bp. 568C) directly from the reaction mixture at
atmospheric pressure to give the product as a colourless oil
(1.0 g, 83%). dH(270 MHz, CDCl3) 5.14 (1H, dq, J 48.6,
7.0, CFH), 1.66 (3H, dd, J 22.8, 6.8, Me); dC(67.9 MHz,
CDCl3) 172.5 (d, J 27.9), 90.3 (d, J 194.5), 17.6 (d, J 21.9);
dF(254.2 MHz, CDCl3) 171.4 (dq, J 49.1, 23.0, CFH).
For the kinetic analysis of the individual diastereoi-
somers, the same conditions were used. For the faster
reacting (S,S)-diastereoisomer the initial reaction rate could
be calculated after 5±10 min by monitoring the integration
of the ¯uorine signals by 19F-NMR. For the slower reacting
(S,R)-diastereoisomer this required several hours.
1.3. Preparation of (S)-phenylalanine-N-2-(R,S)-
fluoropropionamide
References
(S)-Phenylalanine (2.63 g, 16 mmol) was dissolved in
4 M NaOH (12 ml, 48 mmol). The reaction was cooled to
08C and (R,S)-2-¯uoropropanoyl chloride (1.6 g, 14 mmol)
was added dropwise. The reaction was warmed to room
temperature and stirred for 30 min, acidi®ed to pH ˆ 3 with
a HCl solution and the product was extracted into ethyl
acetate (100 ml), washed with brine (3 Â 100 ml, 5% w/v)
and dried over anhydrous sodium sulphate, ®ltered and
evaporated in vacuo to afford the product as an amorphous
white solid (0.6 g, 2.5 mmol, 16%). M.p. 87.0±87.58C;
FTIR 3388 (s), 3335 (s), 2979 (s), 2800 (br), 1729 (s),
1624 (s), 1545 (s), 1227 (s); EI (m/z) 239 (5), 194 (10), 148
(100), 131 (13), 120 (30), 103 (22), 91 (95), 77 (27), 65 (27),
47 (33); dH(270 MHz, CDCl3), 7.23 (5H, m, ArH), 6.92 (1H,
m, NH), 5.03 (1H, dq, J 39.1, 6.8, CFH), 4.90 (1H, over-
lapping dd, J 6.8, 6.8, CH), 3.19 (2H, m, CH2), 1.48 (3H, 2
overlapping dd, J 24.5, 6.8, CH3); dC(67.9 MHz, CDCl3)
174.2, 174.1, 171.2 (d, J 19.0), 171.1 (d, J 20.0), 135.3,
135.3, 129.3, 129.2, 128.6, 128.5, 127.2, 127.2, 88.3 (d, J
[1] D. O'Hagan, H.S. Rzepa, Chem. Commun., (1997) 645.
[2] E.J. Toone, M.J. Werth, B.J. Jones, J. Am. Chem. Soc. 112 (1990)
4946.
[3] K. Hult, T. Norin, Pure Appl. Chem. 64 (1992) 1129.
[4] P. Kalaritis, R.W. Regenye, J.J. Partridge, D.L. Coffin, J. Org. Chem.
55 (1990) 812.
[5] T. Kitazume, K. Murata, T. Ikeya, J. Fluorine Chem. 31 (1986) 143.
[6] D. O'Hagan, H.S. Rzepa, J. Chem. Soc., Perkin Trans., Perlin 1
(1994) 3.
[7] J. Banks, A.S. Batsanov, J.A.K. Howard, D. O'Hagan, H. S Rzepa, S.
Martin-Santamaria, Perlin 2 (1999) 2409.
[8] D.O. Hughes, R.W.H. Small, Acta Cryst. 15 (1962) 933.
[9] G.A. Jeffrey, J.R. Ruble, R.K. McMullan, D.J. DeFrees, J.A. Pople,
Acta Cryst. B37 (1981) 1885.
[10] G.A. Olah, J.T. Welch, Y.D. Vankar, M. Nojima, I. Kerekes, J.A.
Olah, J. Org. Chem. 44 (1979) 3872.
[11] J. Barber, R. Keck, J. Retey, Tetrahedron Letts. 23 (1982) 1549.
[12] R. Keck, J. Retey, Helv. Chem. Acta 63 (1980) 769.
[13] D.J. Bailey, D. O'Hagan, M. Tavasli, Tetrahedron Asymm. 8 (1997)
149.
[14] G. Zubay, Biochemistry, 2nd Edition, Maxwell-Macmillan, 1989.