The relative catalytic efficiency of β-lactamase catalyzed acyl and phosphyl transfer

Article abstract of DOI:10.1006/bioo.2000.1192

Phosphonamidates which bear a simple resemblance to penicillin type structures have been synthesised as potential inhibitors of β-lactamases: -ethyl N-(benzyloxycarbonyl) amidomethyl phosphonyl amides, PhCH₂OCONHCH₂P(O)(OEt)NR₂, the amines HNR₂ being L-proline, D-proline, L-thiazolidine, and o-anthranilic acid. The proline derivatives completely and irreversibly inactivated the class C β-lactamase from Enterobacter cloacae P99, in a time-dependent manner, indicative of covalent inhibition. The inactivation was found to be exclusive to the class C enzyme and no significant inhibition was observed with any other class of β-lactamase. The anthranilic acid derivative exhibited no appreciable inactivation of the β-lactamases. The phosphonyl proline and phosphonyl thioproline derivatives were separated into their diastereoisomers and their individual second order rate constants for inhibition were found to be 7.72 ± 0.37 and 8.3 × 10^-2 ± 0.004 M^-1 s^-1 for the L-proline derivatives, at pH 7.0. The products of the inhibition reaction of each individual diastereoisomer, analyzed by electrospray mass spectroscopy, indicate that the more reactive diastereoisomers phosphonylate the enzyme by P-N bond fission with the elimination of proline. Conversely, gas chromatographic detection of ethanol release by the less reactive proline diastereoisomer suggests phosphonylation occurs by P-O bond fission. The enzyme enhances the rate of phosphonylation with P-N fission by at least 10⁶ compared with that effected by hydroxide-ion. The pH dependence of the rate of inhibition of the β-lactamase by the more reactive diasteroisomer is consistent with the reaction of the diprotonated form of the enzyme, EH₂, with the inhibitor, I (or its kinetic equivalents EH with IH). This pH dependence and the rate enhancement indicate that the enzyme appears to use the same catalytic apparatus for phosphonylation as that used for hydrolysis of β-lactams. The stereochemical consequences of nucleophilic displacement at the phosphonyl centre are discussed.

Full text of DOI:10.1006/bioo.2000.1192

Bioorganic Chemistry 29, 77–95 (2001)
doi:10.1006/bioo.2000.1192, available online at http://www.idealibrary.com on
The Relative Catalytic Efficiency of ␤-Lactamase
Catalyzed Acyl and Phosphyl Transfer
Martin J. Slater, Andrew P. Laws, and Michael I. Page1
Department of Chemical & Biological Sciences, The University of Huddersfield,
Queensgate, Huddersfield, HD1 3DH United Kingdom
Received July 28, 2000
Phosphonamidates which bear a simple resemblance to penicillin type structures have been
synthesised as potential inhibitors of ␤-lactamases: -ethyl N-(benzyloxycarbonyl) amidomethyl
phosphonyl amides, PhCH2OCONHCH2P(O)(OEt)NR2, the amines HNR2 being
L-proline, D-
proline, -thiazolidine, and o-anthranilic acid. The proline derivatives completely and irrevers-
L
ibly inactivated the class C ␤-lactamase from Enterobacter cloacae P99, in a time-dependent
manner, indicative of covalent inhibition. The inactivation was found to be exclusive to the
class C enzyme and no significant inhibition was observed with any other class of ␤-lactamase.
The anthranilic acid derivative exhibited no appreciable inactivation of the ␤-lactamases. The
phosphonyl proline and phosphonyl thioproline derivatives were separated into their diastereoi-
somers and their individual second order rate constants for inhibition were found to be
7.72 Ϯ 0.37 and 8.3 ϫ 10Ϫ2 Ϯ 0.004 MϪ1 sϪ1 for the
L-proline derivatives, at pH 7.0. The
products of the inhibition reaction of each individual diastereoisomer, analyzed by electrospray
mass spectroscopy, indicate that the more reactive diastereoisomers phosphonylate the enzyme
by P-N bond fission with the elimination of proline. Conversely, gas chromatographic detection
of ethanol release by the less reactive proline diastereoisomer suggests phosphonylation occurs
by P-O bond fission. The enzyme enhances the rate of phosphonylation with P-N fission by
at least 106 compared with that effected by hydroxide-ion. The pH dependence of the rate of
inhibition of the ␤-lactamase by the more reactive diasteroisomer is consistent with the reaction
of the diprotonated form of the enzyme, EH2, with the inhibitor, I (or its kinetic equivalents
EH with IH). This pH dependence and the rate enhancement indicate that the enzyme appears
to use the same catalytic apparatus for phosphonylation as that used for hydrolysis of ␤-lactams.
The stereochemical consequences of nucleophilic displacement at the phosphonyl centre are
discussed.
᭧ 2001 Academic Press
It is generally accepted that nucleophilic substitution at acyl centres proceeds
through the formation of an unstable tetrahedral intermediate (TI) (1). The reaction
pathway thus involves a change in geometry and the conversion of the carbonyl
carbon from three to four coordination. Furthermore, it is assumed that there is
some preferential direction of nucleophilic attack such that the incoming nucleophile
approaches at approximately the tetrahedral angle to the carbonyl group (2).
By contrast, the associative mechanism for phosphyl group transfer involves a
1 To whom correspondence and reprint requests should be addressed. Fax: 01484472182. E-mail:
m.i.page@hud.ac.uk.
77
0045-2068/01 $35.00
Copyright ᭧ 2001 by Academic Press
All rights of reproduction in any form reserved.

78
SLATER, LAWS, AND PAGE
pentacoordinate intermediate with trigonal bipyramidal geometry (1,3). Here an ini-
tially four coordinate and tetrahedral phosphorus centre is converted to a five coordi-
nate one and, in general, it is assumed that the preferential pathway involves the
nucleophile taking up the apical position and the leaving group departing from an
apical position of the trigonal bipyramidal intermediate (TBPI) (1,3).
Nucleophilic attack at both carbonyl and phosphyl centres is often facilitated by
general base catalysis when covalent bond formation to the incoming nucleophile
generates an acidic center in the intermediate. Similarly if the leaving group is very
basic, bond fission and expulsion of the leaving group may be assisted by general
acid catalysis. Given the preferential geometrical requirements for incoming nucleo-
philes and departing leaving groups, there should be a favoured relative positioning
of the general base and general acid catalysts in an enzyme catalyzed reaction (Scheme
1). In principle, the general base, which accepts a proton from the attacking nucleo-
phile, is not necessarily the same residue, which then acts as a general acid to donate
a proton to the leaving group. Conceptually this is neater and appears to be the case
for some serine proteases such as ␣-chymotrypsin. However, this may not always be
the situation.
It is often assumed, but with little actual supporting evidence, that enzymes catalyze
SCHEME 1.

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
79
reactions by an exquisite positioning of the catalytic groups (4). If this were the case
then it is doubtful if an enzyme with a primary function, say, as a catalyst for acyl
transfer could be an effective catalyst for phosphyl transfer because of the geometrical
differences just described.
Herein, we report that a serine enzyme—␤-lactamase class C—is, in fact, also an
extremely efficient catalyst for phosphyl transfer. The pH dependence of the kinetic
parameters indicate that similar catalytic machinery is used for both types of reactions.
A preliminary description of the inactivation of the ␤-lactamase by phosphonylation
has been reported (5).
MATERIALS AND METHODS
Enzymes. The class C ␤-lactamase of Enterobacter cloacae P99 was obtained from
the Centre of Applied Microbiology and Research (Porton Down, UK). The class A
and B ␤-lactamases from Bacillus cereus were gifts from M. Galleni (University of
Liege, Belgium).
Starting compounds were purchased from Aldrich/Sigma Chemical Co. All solvents,
unless stated otherwise, were used as purchased without purification. Ethanol free
chloroform was prepared by shaking with 5% (v/v) conc H2SO4 and then washing
with distilled water. The chloroform layer was then separated, dried (CaCl2) and then
distilled with exclusion of moisture from P2O5. Triethylamine was usually distilled
over KOH. Thionyl chloride was distilled under reduced pressure with exclusion of
moisture, at least twice, before use.
Benzyl penicillin was kindly donated by Glaxo Wellcome and cephaloridine was
kindly donated by SmithKline Beecham.
HPLC. All HPLC analyses were performed using a Lichrosorb 7 micron reverse
phase RP 18 column (250 ϫ 4 mm). A standard gradient elution programme was
used involving elution from 100% aq. ammonium acetate (0.1% w/v) to 100% acetoni-
trile over a 30 min period.
Diethyl N-(phenylacetyl) amidomethyl phosphonate (1). This compound was syn-
thesized from diethyl N-(phthalimido)-methylphosphonate (20 g, 67 mM) using the
method of Bartlett (6). Initial N-deprotection with hydrazine monohydrate (3.78 ml,
71 mM) gave the crude amine (diethyl aminomethylphosphonate, 11.5 g), which was
condensed with phenylacetyl chloride (11.11 ml, 84 mM) to give the product (3) in
55% yield (10.6 g) as a yellow oil. HPLC (220 nm): 16.1 min. TLC (Rf in EtOAc):
0.23 1H (CDCl3) ␦ 1.26 (t, J 7.1 Hz, 6H, CH3CH2OP), 3.60 (s, 2H CH2Ph), 3.68 (dd,
JP-CH 11.96 Hz, JNH-CH 6.00 Hz, 2H, CH2P), 4.06 (quin., JP-OCH 7.1 Hz, JCH 7.1 Hz,
3
4H, CH3CH2OP), 6.0 (broad s, 1H, NH), 7.30 (m, 5H, ArH).
Diethyl N-(benzyloxycarbonyl) amidomethyl phosphonate (2). This compound was
synthesized as described above except that, in this case, the amine (diethyl aminometh-
ylphosphonate, 11.25 g) was condensed with benzyl chloroformate (12 ml, 84 mM),
to give the product (4), in 60% yield (12.2 g) as an oil. HPLC (257 nm): 17.3 min,
1
TLC (Rf in EtOAc): 0.26. H (CDCl3) ␦ 1.29 (t, J 7.2 Hz, 6H, CH3CH2OP), 3.61
(dd, JP-CH 11.24 Hz, JNH-CH 6.04 Hz, 2H CH2P), 4.10 (quin., JP-OCH 7.2 Hz, J 7.34
Hz, 4H, CH3CH2OP), 5.11 (s, 2H OCH2Ph), 5.30 (broad s, 1H, NH), 7.35 (m, 5H,
ArH). IR (liquid film): 3266 cmϪ1(NH), 1719 cmϪ1 (CO).

80
SLATER, LAWS, AND PAGE
Ethyl N-(benzyloxycarbonyl)amidomethyl phosphonate (3). Diethyl (N-benzyloxy-
carbonyl) amidomethyl phosphonate (9.43 g, 31 mM) was dissolved in 62 ml of
dioxane 3 equivalents of sodium hydroxide were added (23 ml of a 4 M solution)
and the resulting emulsion was left to stir overnight at room temperature. The reaction
mixture was diluted with 23 ml of distilled water and then washed with two 56-ml
portions of chloroform to remove unreacted starting material. A further 112 ml of
chloroform were added to the aq. layer before acidification with 51 ml of H2SO4
(1M). The organic layer was separated and the aq. layer extracted with two 140-ml
portions of chloroform. The combined portions of chloroform were dried (anhyd.
MgSO4) and rotary evaporated under reduced pressure to yield an oil, which solidified
on standing to give product as a white powder in 93.5% yield. Mpt 98.7–101.3ЊC
(lit. 106–106.56). HPLC (257 nm): 9.2 min. TLC (Rf in EtOAc): 0.4. 1H (CDCl3) ␦
1.30 (t, J 7.1 Hz, 3H, CH3CH2OP), 3.62 (d, JP-CH 11.48 Hz, 2H, CH2P), 4.11 (quin.
JP-OCH 7.26 Hz, JCH 7.1 Hz, 2H, CH2OP), 5.12 (s, 2H, CH2Ph), 5.61 (broad s, 2H,
NH/POH), 7.33 (m,35H, ArH).
Ethyl N-(benzyloxycarbonyl)amidomethylphosphonyl L-proline (benzyl ester)
(4). Benzyl L-proline was obtained by adding satd. sodium bicarbonate to a solution
of the hydrochloride salt (3 g in 37 ml water/50 ml EtOAc) at 0ЊC. The aq. layer
was quickly extracted with three 75-ml portions of ethyl acetate, which were then
combined, dried (MgSO4), and rotary evaporated to give the desired amine (2.4 g)
as a liquid. The phosphonate monoester (3) was converted to the corresponding
phosphonyl chloride by the treatment of a solution of the monoester (1 g, 3.7 mM
in 7 ml CHCl3) with thionyl chloride (0.26 ml, 3.7 mM). After the removal of solvent,
residual HCl and SO2, the resulting oil was dissolved in chloroform (7 ml), and then
condensed with benzyl proline (0.9 g in 7 ml CHCl3) at 0ЊC, using triethylamine
(0.93 ml) as a base. The product was obtained as a mixture of two diastereoisomers
in the form of a thick yellow oil (1.4 g, 83%). Yields were typically in excess of
70% (1.2 g) after purification by column chromatography on silica 60 gel. HPLC
(220nm): 22.85 and 23.13 min. (2\3 ratio of two diastereoisomers). TLC (Rf in
EtOAc): 0.26 and 0.33. 13C (CDCl3) ppm 15.90, 16.00 (CH3CH2OP), 24.78, 24.88
(Pro ␥ CH2), 30.74, 30.84 (Pro ␤ CH2), 36.0 (d, JC-P 148.9 Hz, CH2P), 45.72, 45.80
(Pro ␦ CH2N), 59.20, 60.00 (Pro ␣ CH), 60.42, 60.52 (CH3CH2OP), 66.41 (CH2 Bzl),
66.66 (CH2 Cbz), 127.57–128.27 (ArCH ϫ 10), 135.22, 135.45 (quat. Cbz), 136.40
(quat. Bzl), 156.31 (CO Cbz), 173.94, 174.06 (CO Bzl). IR (liquid film): 3235 cmϪ1
(NH), 1727, 1716 cmϪ1Ϫ (CO). GCMS: m/z 461 (M ϩ1), 325 (M -Z), 256 (M -
Pro(OBzl)), 204 (Pro(OBzl)). The two diastereoisomers were separated by column
chromatography on silica 60 gel using gradient elution (from 100% diethyl ether
to100% ethyl acetate). The less polar diastereoisomer, major component (4a) was
1
eluted first. H (CDCl3) ␦ 1.24 (t, J 6.96 Hz, 3H, CH3CH2OP), 1.87 (m, 3H, Pro ␤
CH ϩ ␥ CH2), 2.13 (m, 1H, Pro ␤ CH), 3.18 (m, 1H, Pro ␦ CH), 3.35 (m, 1H, Pro
␦ CH), 3.62 (ABCX, J 6.06 Hz, J 11.35 Hz, JP-CH 6.8 Hz, 1H, CHAP), 3.80 (ABCX,
J 8.0 Hz, J 11.35 Hz, JP-CH 15.8 Hz, 1H, CHBP), 4.10 (m, 2H, CH2OP), 4.27 (dt, J
3.35 Hz, JP-CH 7.64 Hz, 1H, Pro ␣ CH), 5.06 (AB, J 12.4 Hz, 1H, Cbz CH), 5.09
(AB, J 12.4 Hz, 1H, OBzl CH), 5.12 (AB, J 12.4 Hz, 1H, cbz CH), 5.13 (AB, J 12.4
Hz, 1H, OBzl CH), 5.55 (m, 1H, NH), 7.32 (m, 10H, ArH). The more polar, minor
component (4b) was eluted second. 1H (CDCl3) ␦1.2 (t, J 6.98 Hz, 3H, CH3CH2OP),

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
81
1.82 (quin., J 7.4 Hz, 2H, Pro ␥ CH2), 1.97 (ABC2X, J 4.49 Hz, J 6.70 Hz, J 11 Hz,
1H, Pro ␤ CH), 2.15 (ABC2X, J 8.28 Hz, J 8.28 Hz, J 11 Hz, 1H, Pro ␤ CH), 3.23
(m, 2H, Pro ␦ CH2), 3.60 (ABCX, J 5.68 Hz, JP-CH 12.57 Hz, J 15.4 Hz, 1H, CHP),
3.73 (ABCX, J 7.57 Hz, JP-CH 10.58 Hz, J, 15.4 Hz, 1H, CHP), 3.91 (ABC3X, J 7.39
Hz, JP-OCH 7.4 Hz, J 10.08 Hz, 1H, CHOP), 4.08 (ABC3X, J 7.26 Hz, JP-OCH 7.4 Hz,
J 10.08 Hz, 1H, CHOP), 4.45 (dt, J 3.52 Hz, JP-CH 8.59 Hz, 1H, Pro ␣ CH), 5.09
(AB, J 12.28 Hz, 1H, OBzl CH), 5.12 (s, 2H, cbz CH2), 5.18 (AB, J 12.28 Hz, 1H,
OBzl CH), 6.0 (m, 1H, NH), 7.31 (m, 10H, ArH).
Ethyl N-(benzyloxycarbonyl)amidomethylphosphonyl
D-proline (methyl ester)
(5). Methyl -prolate was obtained by treating a suspension of
D
D-proline (590 mg)
in ethanol (10 ml) at 0ЊC with ethereal diazomethane. Rotary evaporation of the
solvent under reduced pressure gave the desired amine as a colourless liquid (471.5
mg) in 71.2% yield. The ester amine (471.5 mg) was condensed with the phosphonyl
chloride (obtained from 0.52 g, 1.9 mM, of the monoester 3, as before), using the
method described previously, to give a yellow oil (620 mg) in 84% yield. A mixture
of the two diastereoisomers was obtained (560 mg) in 76.5% yield (after purification
by column chromatography on silica 60 gel, as described previously). The ratio of
the less polar isomer (5a) relative to the more polar isomer (5b) was found to be 2\1
by 1H nmr. HPLC (220nm): 17.4 min. TLC (Rf in 10% EtOH/EtOAc): 0.31 and 0.37.
13C (CDCl3) ppm 16.24, 16.33 (CH3CH2OP), 25.08, 25.33 (Pro ␥ CH2), 31.03, 31.13
(Pro ␤ CH2), 37.40, 37.76 (doublet ϫ 2 , JC-P 148.7, 158.00 Hz, CH2P), 46.00, 46.07
(Pro ␦ CH2N), 52.08, 52.34 (Pro OCH3), 60.09, 60.16 (Pro ␣ CH), 60.36, 60.74
(CH2OP), 66.78, 67.04 (CH2 Cbz), 127.91–128.45 (ArCH ϫ 3), 136.36, 136.66 (quat.
1
Cbz), 156.47, 156.55 (CO Cbz), 174.92, 175.05 (CO OCH3) 5b. H (CDCl3) ␦ 1.3
(t, J 7.1 Hz, 3H, CH3CH2OP), 1.86 (m, 4H, Pro ␤ CH2 ϩ ␥ CH2), 3.18 (m, 1H, Pro
␦ CH), 3.32 (m, 1H, Pro ␦ CH), 3.61 (ABCX, J 6.08 Hz, J 11.32 Hz, JP-CH 6.8 Hz,
1H, CHAP), 3.67 (s, 3H, OCH3), 3.82 (ABCX, J 8.10 Hz, J 11.32 Hz, JP-CH 16.88
Hz, 1H, CHBP), 4.13 (m, 2H, CH2OP), 4.21 (dt, J 4.05 Hz, JP-CH 6.80 Hz, 1H, Pro
␣ CH), 5.05 (AB, J 12.33 Hz, 1H, Cbz CH), 5.13 (AB, J 12.33 Hz, 1H, Cbz CH),
1
6.06 (m, 1H, NH), 7.32 (m, 5H, ArH) 5a. H (CDCl3) ␦ 1.25 (t, J 8.3 Hz, 3H, CH3
CH2OP), 1.83 (quin., J 7.0 Hz, 2H, Pro ␥ CH2), 1.95 (ABC2 X, J 4.5 Hz, J 6.75 Hz,
J 12.5 Hz, 1H, Pro ␤ CH), 2.38 (ABC2X, J 8.1 Hz, J 8.1 Hz, J 12.5 Hz, 1H, Pro ␤
CH), 3.01 (m, 2H, Pro ␦ CH2), 3.69 (s, 3H, OCH3), 3.93 (m, 2H, CH2P), 4.11 (m,
2H, CH2OP), 4.39 (dt, J 4.05 Hz, JP-CH 8.1 Hz, 1H, Pro ␣ CH), 5.10 (s, 2H, cbz
CH2), 6.4 (m, 1H, NH), 7.33 (m, 5H, ArH).
Ethyl N-(benzyloxycarbonyl)amidomethylphosphonyl
L-thioproline (methyl ester)
(6). -Thioproline methyl ester (1.14 g) was obtained from 1 g of the corresponding
L
acid. The amine (1 g in 10 ml CHCl3) was condensed with the phosphonyl chloride
(obtained from 1.55 g of the monoester (3). An unresolved mixture of the pure
diastereoisomers (1:1 ratio) was obtained in 30% yield (300 mg), after separation by
column chromatography on silica 60 gel. HPLC (220 nm): 18.20 min. TLC (Rf in
1
10% EtOH/EtOAc): 0.39 and 0.48. H (CDCl3) ␦ 1.26 (t, J 7.4 Hz, 3H, CH3CH2OP,
6a), 1.27 (t, J 7.4 Hz, 3H, CH3CH2OP, 6b), 2.96 (ABX, J 6.75 Hz, J 10.8 Hz, 1H,
␤ CHS, 6b), 3.16 (ABX, J 6.8 Hz, J 10.8 Hz, 1H, ␤ CHS, 6b), 3.18 (ABX, J 6.8
Hz, J 10.8 Hz, 1H, ␤ CHS, 6a), 3.28 (ABX, J 2.2 Hz, J 10.8 Hz, 1H, ␤ CHS, 6a),
3.48 (ABCX, J 4.7 Hz, JP-CH 10.8 Hz, J 15.5 Hz, 1H, CHP), 3.69 (s, 3H, CH3OP,

82
SLATER, LAWS, AND PAGE
6b), 3.7 (s, 3H, CH3OP, 6a), 3.80 (m, 3H, CHP ϩ CH2P), 4.19 (m, 6H, CH2OP ϩ
␣ CH), 3.38 (ABX, J 5.4 Hz, J 8.1 Hz, 1H, ␦CH,6b), 4.49 (ABX, J 5.4 Hz, J 8.1
Hz, 1H, ␦CH, 6a), 4.84 (ABX, J 5.4 Hz, J 8.1 Hz, 1H, ␦CH, 6b), 4.95 (ABX, J 5.4
Hz, J 8.1 Hz, 1H, ␦CH, 6a), 5.06 (AB, J 12.5 Hz, 1H, cbz CH, 6b), 5.10 (s, 2H, Cbz
CH2, 6a), 5.13 (AB, J 12.5 Hz, 1H, Cbz CH, 6b), 6.2 (m, 1H, NH 6b), 6.35 (m, 1H,
NH, 6a), 7.34 (m, 10H, ArH). 13C (CDCl3) ppm 16.09, 16.19 (CH3CH2OP), 34.66,
34.73 (Thiopro NCH2S), 37.66, (Thiopro ␤ CH2), 37.40, 37.72 (doublet ϫ 2 , JC-P
153.92,154.60 Hz, CH2P), 48.06, 48.13 (Thiopro CH2N), 52.58, 52.75 (Thiopro
OCH3), 61.35, 61.45 (CH2OP), 62.90, 62.99 (Thiopro CH), 69.42 (CH2 Cbz), 128.03–
128.44 (ArCH ϫ 4), 136.45, 136.53 (quat. Cbz), 156.49, 156.57 (CO Cbz), 171.66,
171.70 (CO OCH3).
Ethyl N-(benzyloxycarbonyl)amidomethylphosphonyl anthranilate (methyl ester)
(7). This compound was synthesised from methyl anthranilate (0.56 ml) and the
phosphonyl chloride (obtained from 1g of the monoester (3)) as described previously.
The product was obtained as a white solid in 68% yield, after purification by column
chromatography on silica 60 gel Mpt. 91–92ЊC. HPLC (220 nm): 21.34 min. TLC
(Rf in EtOAc): 0.29. 13C (CDCl3) ppm 16.19, 16.28 (CH3CH2OP), 38.43 (d, JC-P
143.66 Hz, CH2P), 52.04 (OCH3), 61.67, 61.77 (CH2OP), 67.00 (CH2Cbz), 114.32,
114.44, 117.70, 117.75 (ArCH Anth. C4), 120.54 (ArCH Anth C3), 127.90–128.35
(ArCH Cbz), 131.23 (ArCH Anth. C2), 134.56 (ArCH Anth. C5), 136.17 (ipso Cbz),
1
143.94 (C CO), 156.04, 156.13 (CO Cbz), 168.68 (CO Cbz). H (CDCl3) ␦ 1.37 (t,
J 7.09 Hz, 3H, CH3CH2OP), 3.76 (ABX, JP-CH 10.37 Hz, JNH-CH 6.02 Hz, 2H, CH2P),
3.90 (s, 3H, OCH3), 4.18 (m, 2H, CH2OP), 5.00, (AB, J 14.8 Hz, 1H, CH2Ph), 5.05
(AB, J 14.8 Hz, 1H, CH2Ph), 5.15 (broad s, 1H, NH), 6.95 (t, J 7.35 Hz, 1H, (C4)-
ArH), 7.33 (m, 5H, Cbz ArH), 7.41 (t, J 7.82 Hz, 1H, (C3)-ArH), 7.60 (d, J 8.33
Hz, 1H, (C2)-ArH), 7.97 (d, J 8.00 Hz, 1H, (C5)-ArH), 9.20 (d, JP-NH 8.00 Hz, 1H,
PNH). IR (liquid film): 3226.4 cmϪ1 (NH CO), 1717.2 cmϪ1 (CO Cbz), 1682.6
cmϪ1 (CO ester). Elemtl. anal.: expect -C(51.93), H(5.05), N(3.19), found -C(51.68),
H(5.08), N(3.09).
Ammonium salt of ethyl N-(benzyloxycarbonyl)amidomethylphosphonyl L-proline
(8). This compound was synthesized from the diester (4) by hydrolysis of a 0.1M
solution in a mixture of 50% acetonitrile/water (v/v), with 3 equiv. of lithium hydrox-
ide. Benzyl alcohol was removed by extraction with diethyl ether followed by freeze-
drying to give the lithium salts. HPLC (220 nm): 10.62 and 11.24 min (in the ratio
3\2). 13C (D2O) ppm 14.87, 14.95 (CH3CH2OP), 24.38, 24.49 (Pro ␥ CH2), 31.08,
31.21 (Pro ␤ CH2), 35.81 (doublet, JC-P 147.67 Hz, CH2P), 46.27, 46.35 (Pro ␦
CH2N), 61.02, 61.11 (Pro ␣ CH), 61.34, 61.48 (CH2OP), 66.45, 66.56 (CH2Cbz),
127.03–128.13 (ArCH ϫ 3), 135.77 (ipso. Cbz), 157.50 (CO Cbz), 181.39 (CO pro).
1H (D2O) ␦ 1.24 (t, J 7.0 Hz, 3H, CH3CH2OP, 8b), 1.30 (t, J 7.0 Hz, 3H, CH3CH2OP,
8a), 1.83 (m, 6H, Pro ␥ CH2, and ␤ CH), 2.13 (m, 2H, Pro ␤ CH), 3.26 (m, 4H, Pro
␦ CH2), 3.62 (d, JP-CH 10.2 Hz, 2H, CH2 P, 8a), 3.67 (d, JP-CH 10.2 Hz, 2H, CH2P,
8b), 3.99 (dt, JP-NCH 13.5 Hz, 2H, Pro ␣ CH), 4.11 (quin., J 7.0 Hz, JP-OCH 7.0 Hz,
4H, CH2OP), 5.15 (s, 4H, cbz CH2), 7.42 (m, 10H, ArH). The two diastereoisomers
were separated by HPLC using a Dynamax 60 A reverse phase C 18 column (25 ϫ
2.14 cm, flow rate 12 ml/min) and eluting with 13% acetonitrile–87% (v/v) of an
aqueous ammonium acetate solution (1% w/v). The ammonium acetate solutions

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
83
containing the pure diastereoisomers were freeze-dried and then chromatographed a
second time, eluting with 10% acetonitrile–90% (v/v) distilled water, to remove
ammonium acetate. The resulting fractions were freeze dried to yield white hygro-
scopic powders (8a) and (8b) as their ammonium salts. HPLC at 220nm confirmed
the presence of the two diastereoisomers without contamination by either the other
isomer or any other impurity. FABMS (8a): m\z 393, 20 (MNa+), 372, 3 (MH+),
256,10 (M Ϫ Pro), 116, 14 (ProH+), 91, 100 (PhCH2 ). ESMS (8a): mz 415, 10 (MH+,
+
Na2 ), 409, 5 (MK+), 393, 100 (MNa+), 371, 5 (MH+). ESMS (8b): m/z 409, 100
+
(MK+), 393, 25 (MNa+), 371, 20 (MH+), 256, 12 (M Ϫ Pro), 116, 22 (ProH+).
Ammonium salt of ethyl N-(benzyloxycarbonyl)amidomethylphosphonyl L-thiopro-
line (9). The lithium salt of this compound was synthesized from (6) as previously
described. HPLC (220 nm): 11.3 and 11.64 min (in the ratio 1.1:1, (9a,9b)). The
two diastereoisomers were separated by HPLC and the products were obtained as
hygroscopic powders as their ammonium salts, which showed no contamination from
either the other isomer or any other impurity. ESMS (9a): m/z 427, 90 (MK+), 389,
15 (MH+), 256, 65 (M Ϫ thiopro), 145, 100. ESMS (9b): m/z 427, 100 (MK+), 411,
40 (MNa+), 389, 20 (MH+), 256, 85 (M Ϫ thiopro), 145, 30.
Lithium salt of ethyl N-(benzyloxycarbonyl) amidomethylphosphonyl anthranilate
(10). The lithium salt of this compound was synthesized from (7) as previously
described. HPLC (220 nm): 11.79 min. 1H (D2O) ␦ 1.34 (t, J 7.0 Hz, 3H, CH3CH2OP),
3.81 (d, JP-CH 9.4 Hz, 2H, CH2P), 4.21 (ABX J 7.0 Hz, JP-OCH 7.0 Hz, 2H, CH2OP),
5.00 (s, 2H, cbz CH2), 7.04 (t, J 7.4 Hz, 1H, (C4)-ArH), 7.41 (m, 7H, Anth. (C2-
C3) and Cbz ArH), 7.92 (d, J 8.03 Hz, 1H, (C5) ArH).
Kinetics. Some of the phosphonamidate salts were hygroscopic and determination
of their concentration by weighing was not possible. An HPLC calibration curve for an
authentic sample of the phosphonate monoester (3) was obtained using the conditions
described previously. The concentration of the phosphonamidate salt stock solution
was then determined from its complete hydrolysis in HCl and measuring the concentra-
tion of (3) produced.
␤-Lactamase activity was determined against benzylpenicillin by a spectrophoto-
metric method (7). Concentrations of the Ent. cloacae P99 ␤-lactamase were also
determined spectrophotometrically, using an extinction coefficient at 280 nm, 7.1 ϫ
104 MϪ1 cmϪ1 (8). Concentrated stock solutions of the phosphonamidates were pre-
pared in 20 mM MOPS buffer at pH 7.5.
For the kinetic measurements required for the pH-rate profiles the buffer systems
were HCl, formate, acetate, Mes, Mops, Hepes, Tapso, Ches, and NaOH. In all cases,
an ionic strength of 1.0 M (KCl), and 30ЊC was used. Rates of enzyme inactivation
were determined from measurements of enzyme activity against benzylpenicillin as
a function of time. In the general procedure, enzyme (ca. 5 ␮M) and the inhibitor
(20 ␮M to 20 mM) were incubated together in buffer at the required pH. Aliquots
of the reaction mixture were withdrawn at relevant time intervals and immediately
assayed for enzyme activity. The activity of samples of enzyme without inhibitor was
also routinely monitored as a control. Pseudo-first order rate constants of inactivation
were determined from the Enzfitter programme and second-order rate constants were
then obtained from the slopes of plots of the first-order rate constants against inhibi-
tor concentration.

84
SLATER, LAWS, AND PAGE
RESULTS AND DISCUSSION
The major cause of the resistance of some bacteria to the normally lethal action
of ␤-lactam antibiotics is the ability of the bacteria to produce ␤-lactamase enzymes,
which catalyze the hydrolysis of the ␤-lactam of penicillins (11) (Scheme 2) and
cephalosporins. The most prevalent and clinically important ␤-lactamases are the
class A and C type, which are both serine enzymes and act by a mechanism involving
the formation of a relatively unstable acyl enzyme (EA) intermediate, Eq. [1] (9).
k
k
k
1
2
3
E ϩ S s ES → EA → E ϩ P
[1]
k
Ϫ1
Although there have been several crystal structures of ␤-lactamases and their
derivatives reported (10–12) and although the nature and degree of the conservation
of the amino acid residues in and near the active site are known (13) there is little
detailed evidence of the groups presumed to be involved with the necessary proton
transfer steps (14,15). Comparison with serine proteases would indicate the need for
general acid–base catalysis (16). The main contenders for these roles are, in class A
␤-lactamase, the carboxyl group of Glu-166 and the amino group of Lys-73 (9). In
class C ␤-lactamase there is no equivalent glutamate residue but tyrosine-150 may
take its role and some unusual solvent kinetic isotope effect studies indicate that the
tyrosine may have a low pKa of ca. 6 (17). It is assumed that this pKa corresponds
to that of the general base catalyst required for proton removal from serine 64 and
used in the formation of the tetrahedral intermediate leading to acylation (10,17,18).
Site-directed mutagenesis of Tyr-150 has not led to firm conclusions although it
appears to be essential for hydrolysis of the best substrates (19).
We have previously demonstrated that the class C ␤-lactamase from Enterobacter
cloacae P99 is inactivated by the phosphonamidate (13) (R ϭ PhCH2O) (5). Our
preliminary report indicated that the enzyme was phosphonylated in a process analo-
gous to the enzyme catalysed acylation by the penicillins (Scheme 3). We are interested
in the detailed mechanism of this inactivation and the geometrical differences in the
catalytic machinery required for acyl and phosphyl transfer. It is necessary to demon-
strate that the enzyme increases the rate of phosphonylation reaction. It is well known
that enzymes can catalyse the same reaction of a variety of substrates and even different
SCHEME 2.

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
85
reactions with alternative substrates (20). However, demonstrating the efficiency of
the catalysis when the substrate is modified is not straight-forward. Modification of
the substrate structure can affect the free energies of both the initial reactant state
and the transition state, whereas the observed differences in rate constants for the
enzyme catalysed reaction only reflect the difference in energies between these two
states. Importantly, changes in the activity of an enzyme catalysed reaction can result
from differences in intrinsic “chemical” effects within the substrate as well as from
differences in binding interactions between the substrate and enzyme (20). Different
chemical structures can affect the ease of bond-making and -breaking by classical
electronic factors such as inductive, resonance and steric effects. However, the free
energy of activation of an enzyme-catalyzed reaction is also affected by the favourable
binding energies between the protein and substrate substituents not directly involved
with the reaction site. It is therefore important to separate these two effects before
conclusions about the efficiency of enzyme catalysis can be made. We have suggested
(21) that an “enzyme rate-enhancement factor” (EREF) can be evaluated by dividing
the second-order rate constant for the enzyme catalysed reaction, kcat/Km, by that for
hydrolysis of the same substrate catalysed by hydroxide ion, kOH. To demonstrate
enzyme “catalyzed” phosphonylation, it is therefore necessary to compare the relative
“intrinsic” reactivities of the two “substrates"—the penicillin (11) and the phospho-
namidate (13) toward hydrolysis.
The phosphonamidate (13) has two asymmetric centres and exists as four diastereoi-
somers. Reactions of these diastereoisomers at the chiral active site of the enzyme
would be expected to show differences in reactivity and perhaps even reaction type
such as P-O or P-N bond fission. Any such observed discrimination would also be
indicative of the enzyme using its catalytic machinery for phosphonylation.
(i) Hydrolysis of the phosphonamidate (8). The pH rate profile for the hydrolysis
of the phosphonamidate (8) is shown in Fig. 1. There are two acid catalyzed reactions
corresponding to the reaction of the phosphonamidate with an undissociated and
dissociated carboxylic acid. The inflection point corresponds to the expected pKa of
the carboxylic acid, 3.86 Ϯ 0.1. At pHs above this pKa, there is a pH independent
SCHEME 3.

86
SLATER, LAWS, AND PAGE
FIG. 1. A plot of the first order rate constant, kobs/sϪ1, for the hydrolysis of the phosphonamidate
(8) as a function of pH at 30ЊC and I ϭ 1.0M (KCl).
and a hydroxide ion dependent reaction of the phosphonamidate with the carboxylic
acid residue ionized. At very high pH, the rate again starts to fall off which is
presumably the result of ionisation of the side chain amido group. The overall rate
of hydrolysis is therefore given by Eq. [2], where kOH is the second order rate constant
for the hydroxide ion catalyzed hydrolysis and ko is the pH-independent first-order
rate constant for hydrolysis of (8) with the carboxylic acid ionised; kH and kHH, are
the second-order rate constants for the acid catalysed hydrolysis for the phosphonami-
date (8) with a dissociated and undissociated carboxyl group, respectively; ka1 is the
dissociation constant for the carboxylic acid and Ka2 is the dissociation constant of
the amido side chain.
(H+)
Ka2 ϩ (H )
kobs ϭ kOH(OHϪ)
ϩ kO ϩ kH(H+)
+
΂
΃
Ka1
H+
ϩ kHH(H+)
[2]
ͫ ͬ
Ka1 ϩ H+
+
΂ ΃
Ka1 ϩ H
Below pH 11 the reaction occurs with P-N fission and the hydrolysis products are
the monoester (14) and proline (Scheme 4) as expected from previous studies of the
hydrolysis of phosphonamidates (3).
The value of kH, the acid-catalyzed hydrolysis of the phosphonamidate with a
dissociated carboxylic acid, is 1.36 dm3 molϪ1 sϪ1, which is 200-fold greater than
kHH (6.59 ϫ 10Ϫ3 dm3 molϪ1 sϪ1)—the acid-catalyzed hydrolysis of the same compound

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
87
SCHEME 4.
with an undissociated carboxylic acid. It is possible that this rate difference indicates
neighboring group participation by the carboxylate anion or the kinetically equivalent
intramolecular general acid catalysed mechanism of water attack on the phosphonami-
date with an undissociated carboxylic acid (Scheme 8). The rate of the acid catalyzed
hydrolysis of the phosphonamidate with the carboxyl group esterified as the benzyl
ester is also about 100-fold less than that for the carboxylate anion derivative and is
similar to that for the phosphonamidate with an undissociated carboxylic acid. How-
ever, the simplest explanation for these rate differences is that the acid catalyzed
hydrolysis of the phosphonamidate occurs by initial N-protonation followed by attack
of water on the phosphonyl centre (3,22) (Scheme 5). The N-protonated phosphonami-
date (15) is stabilized by the negatively charged carboxylate anion relative to the
undissociated carboxylic acid, giving rise to a higher concentration of the conjugate
acid and thus an overall faster rate of hydrolysis. The proline nitrogen is about 2 pKa
units less basic with an undissociated or esterified carboxylic acid. The rate difference
of 200 between the two species gives a crude estimate of a Bronsted ␤1g of ca.1.0
which is indicative of a positively charged nitrogen in the transition state.
The pH-independent hydrolysis occurs with predominant P-N fission and is very
–
–
–
–
SCHEME 5.

88
SLATER, LAWS, AND PAGE
slow (ko ϭ 5.29 ϫ 10Ϫ8 sϪ1). This pathway probably involves rate-limiting breakdown
of the intermediate (16). Above pH 11 the rate of hydrolysis of the phosphonamidate
is first order in hydroxide ion (Fig. 1) and the second-order rate constant for the
hydroxide ion catalysed hydrolysis, kOH, is 6.79 ϫ 10Ϫ5 dm3 molϪ1 sϪ1. In this region
P-O fission occurs to expel ethanol—again as expected from previous studies of
phosphonamidates (3) (Scheme 6). The basic limb of the pH-rate profile (Fig. 1)
indicates a slight decrease in the dependence upon hydroxide ion at very high pH
which is compatible with a reduced rate of hydrolysis of the phosphonamidate with
its amide side chain ionised (pKa2 ϭ 14.1).
In summary, the phosphonamidate (8) is a relatively stable derivative with a reactiv-
ity similar to that predicted from previous studies (3), and any reaction of it with
enzymes is likely to require the protein to lower the activation energy.
(ii) Enzyme inactivation. Both diastereoisomers of the phosphonamidate of L-
proline (8) completely and irreversibly inactivate the class C ␤-lactamase from Entero-
bacter cloacae (P99) in a time-dependent manner. There is an exponential decrease
in enzyme activity, measured by its hydrolysis of benzylpenicillin, as a function of
time, from which an apparent first order rate constant, kobs, for inactivation could be
obtained. The experiments were conducted with excess phosphonamidate relative to
enzyme concentration and there was no evidence of a zero order decline in enzyme
activity, which would be indicative of saturation of the enzyme by the phosphonamidate
(8). The pseudo first-order rate constants for inactivation show a first order dependence
on the concentration of the phosphonamidate (8) (Fig. 2). The slopes of these graphs
give the second-order rate constants, ki, for inactivation. Although there is no discern-
ible difference in the chemical reactivity of the two diastereoisomers towards alkaline
hydrolysis, for the enzyme reaction they do show different rates of inactivation, the
two values of ki being 7.72 dm3 molϪ1 sϪ1 and 8.3 ϫ 10Ϫ2 dm3 molϪ1 sϪ1 at pH 7.0
SCHEME 6.

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
89
FIG. 2. A plot of the observed first order rate constant for the inactivation of the class C E.cloacae
␤-lactamase by the phosphonamidate (8) at pH 7 as a function of the concentration of the inactivator.
and 30ЊC. The 93-fold difference in reactivity is therefore good evidence of selectivity
in the reaction of the enzyme with the two phosphonamidates (8) and indicative of
specific interactions between the inactivators and the protein. Further evidence of
selectivity is seen from the fact that the phosphonamidate (8) is not a significant
inhibitor of either the class A ␤-lactamase or the class B zinc dependent ␤-lactamase
from B.cereus.
The time-dependent inactivation is indicative of covalent bond formation between
the enzyme and the phosphonamidate. For the most active diastereoisomer this was
confirmed by electrospray mass spectrometry (ESMS). The positive ion ESMS mass
transformed spectrum of the native P99 ␤-lactamase gave a molecular mass Mr of
39,202 Ϯ 5.3. After incubation of the enzyme with the more reactive phosphonamidate
(8) the molecular mass Mr increased to 39,442 Ϯ 3.1, a mass shift of ϩ 240, which
suggests that the enzyme had been phosphonylated by displacing proline from the
phosphonamidate. The calculated mass for a phosphonylated enzyme formed by
displacing proline would be 39,459; that for an enzyme-phosphonamidate complex
would be 39,573 and for that formed by displacing ethanol it would be 39,527. The
discrepancy of 17 in the mass could be experimental error or indicative of the
accompanying loss of, say, ammonia from glutamine. It seems clear that inactivation
of the enzyme occurs by formation of a 1:1 covalently bound enzyme-inactivator
complex in which a proline residue has been displaced by a nucleophilic group on
the enzyme—presumably the active site serine. There was no evidence from gas
chromatography that any ethanol was displaced as a result of the reaction between
␤-lactamase and the more reactive phosphonamidate.
Conversely, the less reactive diastereoisomer reacts with ␤-lactamase by displacing
ethanol, which could be detected by gas chromatography. The observation that the
enzyme reacts with one diastereoisomer by displacing proline and with another by

90
SLATER, LAWS, AND PAGE
displacing ethanol is again indicative of a stereoselective reaction occurring at the
active site.
(iii) Enzyme promoted phosphonylation. It is often assumed that the phosphonyla-
tion of serine enzymes can only occur with organophosphorus compounds with good
leaving groups because these do not require proton transfer from general acid catalysts
to aid departure of the leaving group (23). However, the class C ␤-lactamase is
obviously capable of displacing the proline residue from the phosphonamidate (8)
(Scheme 3) despite the poor leaving group and the presumed need for protonation
of the amine nitrogen.
The phosphonylation of the enzyme displays signs that the reaction takes place at
the active site—other ␤-lactamases are not inactivated by the phosphonamidate and
differential P-N and P-O bond fission occurs with the two diastereoisomers. Phospho-
namidates are relatively chemically stable and the fast reaction with the ␤-lactamase
indicates that the catalytic machinery of the enzyme used for hydrolysis is also used
for phosphonylation. This is confirmed by the enzyme rate enhancement factors
(EREF). The P99 ␤-lactamase catalyses the hydrolysis of benzylpenicillin with an
EREF of 2.6 ϫ 107 (Table 1). The hydroxide ion catalysed hydrolysis of the phospho-
namidate (8) occurs with P-O fission and displacement of ethanol as does the reaction
of the less reactive diastereoisomer with the P99 ␤-lactamase. The EREF value for
enzyme catalysed phosphonylation by P-O fission is obtained from ki/kOH to give
1.2 ϫ 103. However, phosphonylation of ␤-lactamase by the most reactive diastereoi-
somer of the phosphonamidate (8) occurs by P-N fission and displacement of proline.
Because the base hydrolysis of the phosphonamidate (8) occurs with P-O fission, and
as no proline could be detected in the alkaline hydrolysis of the phosphonamidate
(8), the rate constant for hydrolysis by P-N fission must be at least 30 times smaller
than the observed value, i.e., kOH Ͻ 2 ϫ 10Ϫ6 dm3 molϪ1 sϪ1. The calculated EREF
value for phosphonylation of ␤-lactamase is therefore Ͼ 3 ϫ 106. Clearly, the enzyme
is facilitating P-N fission almost as effectively as it does C-N fission in ␤-lactams.
Despite the differences in geometrical requirements for substitution at acyl and phos-
phyl centers and the enormous differences in intrinsic chemical reactivities between
the ␤-lactam in penicillin and the phosphonamidate, the ␤-lactamase enzyme is able
to significantly lower the activation energy for reactions of both compounds.
Further evidence that the phosphonylation process is “catalyzed” by the enzyme
comes from the pH dependence of the rate of inactivation. The dependence of the
rate constant, kcat/Km, for the P99 class C ␤-lactamase catalysed hydrolysis of benzyl-
penicillin on pH (Fig. 3) indicates that there are two ionisations which control hy-
drolytic catalytic activity corresponding to groups of pKa ca 6.1 and 10.1, representing
the two protonic forms of the enzyme, EH2 and EH, respectively (17). The catalytically
important form of the enzyme for hydrolysis is thus EH. The pH-dependence of the
second order rate constant, ki, for the inactivation of ␤-lactamase by the phosphonami-
date(8) (Fig. 3) shows that it also depends on a catalytic group of pKa 6.2, which
suggests that the catalytic machinery used for the hydrolysis of ␤-lactams is also
used for the phosphonylation reaction. However, the pH-rate profile for inactivation
indicates an additional proton is required, i.e., whereas hydrolysis apparently requires
the group of pKa 6 to be in its deprotonated, basic form, phosphonylation apparently
requires this group to be in its protonated, acidic form, although the difference may

␤
TABLE 1
Enzyme Rate Enhancement Factors (EREF) for the P99 Class C ␤-Lactamse Catalyzed Hydrolysis of Benzylpenicillin (11) and the Enzyme Promoted
Phosphonylation of the Phosphonamidate (8)
kOH/dm3 molϪ1 sϪ1
kcat/Km/dm3 molϪ1 sϪ1
EREF
0.15 Ϯ .01
3.97 Ϯ .04 ϫ 106
2.6 ϫ 107
6.79 Ϯ .08 ϫ 10Ϫ5
(P-O) fission)
Ͻ2 ϫ 10Ϫ6
ki/dm3 molϪ1 sϪ1
7.72 Ϯ .15
Ͼ3 ϫ 106
(P-N fission)
(P-N fission)
(P-N fission)
8
Note. kOH is the second order rate constant for the hydroxide ion catalysed hydrolysis, kcat/km and ki the second-order rate constants for enzyme catalyzed
hydrolysis and phosphonylation, respectively, at pH 7.0 and 30ЊC.

92
SLATER, LAWS, AND PAGE
a
SCHEME 7.
be the result of kinetic ambiguity. The rate of hydrolysis of the substrate S is propor-
tional to (EH) (S), where (EH) is the protonic form of the enzyme with the groups
of pKa 6 and 10 in their basic and acidic forms, respectively (Scheme 7). The observed
pH-rate dependence (Fig. 3) for inactivation by the inhibitor, I, indicates a rate
proportional to (EH2) (I), which, for a process of phosphonylation of the active site
serine (Scheme 3) is difficult to interpret mechanistically. Of course, the rate law can
have various kinetic equivalents (Eq. [3]), where ki is the observed second order
rate constant for inactivation, Ka and K1 are the acid dissociation constants for the
phosphonamidate and the enzyme EH2, respectively.
ki
K1
kiKa
K1
Rate ϭ ki(EH2)(I) ϭ
(EH)(I)(H+) ϭ
(EH)(IH)
[3]
Phosphonylation of the enzyme could occur through the protonic form EH with the
group of pKa 6 acting as a general base for serine attack, as is thought to occur in
the hydrolytic reaction (Scheme 8). Breakdown of trigonal bipyramidal intermediate,
FIG. 3. A plot of kcat/Km (left hand axis and open circles O) for the class C E.cloacae ␤-lactamase
catalyzed hydrolysis of benzylpenicillin against pH and ki (right-hand axis and closed squares Ⅲ) against
pH for the inactivation of the enzyme by the phosphonamidate (8).

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
93
however, could occur by proton donation to the proline nitrogen from the solvent
hydronium ion and not from the protein. Only the -proline phosphonamidate (8) is
an effective inhibitor and, surprisingly, the -proline isomer is a much weaker inacti-
L
D
vator by a factor of at least 20-fold. This may reflect the different geometrical binding
requirements of the normal substrate and the phosphonamidate or the possibility of
an inactivator intramolecular catalyzed reaction discussed later. The corresponding
carboxylic acid methyl ester is completely ineffective. which could reflect the normal
requirement of a carboxylate anion for molecular recognition. It could, however,
indicate that the reaction occurs through EH and IH (Eq. [3]), where the latter is the
conjugate acid of the phosphonamidate (8), i.e., with the carboxylic acid group
undissociated. The pKa of this carboxylic acid is 3.86, so, over the pH range studied
for inactivation, the major species present is the carboxylate anion of the phosphonami-
date. The pH dependence for inactivation (Fig. 3) would then be given by Eq. [4].
The true second-order rate constant for phosphonylation
K1
H+
Ka K1 ϩ H+
inact
k
ϭ ki и
и
[4]
obs
would then be given by ki и K1/Ka, which is calculated to be 6.43 ϫ 103 dm3 molϪ1
sϪ1, which in turn would give an EREF of Ͼ 3 ϫ 109. The class C ␤-lactamase
would then indeed be very effective at enhancing the rate of phosphonylation. If the
mechanism does involve the undissociated carboxylic acid of the phosphonamidate
as “substrate” then this is probably the result of substrate-assisted catalysis with the
carboxylic acid acting as an intramolecular general acid catalyst facilitating P-N
fission from the trigonal bipyramidal intermediate (Scheme 8).
The fact that the enzyme can lower the activation energy for the phosphonylation
reaction as well as the acyl transfer reaction of the hydrolysis of the “normal” ␤-
lactam substrates indicates a reasonable degree of flexibilty in the active site. In
SCHEME 8.

94
SLATER, LAWS, AND PAGE
aqueous solution, the generation of a formal negative charge on oxygen in the tetrahe-
dral intermediate is accompanied by a large change in the basicity of the oxygen—a
change in pKa of the corresponding conjugate acids of at least 12 pKa units (Scheme
1). The alkoxide anion must be strongly solvated by H-bonding and many enzymes,
which catalyse acyl transfer reactions, have an “oxyanion hole,” which stabilizes the
presumed tetrahedral intermediate by H-bonding from adjacent peptide links. The
difference in the interaction between the protein and carbonyl oxygen in the initial
state and the oxygen in the tetrahedral intermediate, as a result of the change in
oxygen basicity, makes a major contribution to the lowering of the activation energy
compared with the nonenzyme catalyzed reaction (24). Such an “oxyanion hole”
exists for the serine ␤-lactamases (25) and, in common with serine proteases, one of
the H-bond donors is the peptide N-H of serine residue Ser64 in the class C enzymes.
This ’intramolecular’ H-bond with the active site serine peptide NH presents a fairly
well defined “cyclic” geometry and may be expected to hinder conformational changes
which would be unfavorable for reaction. In the equivalent pentavalent phosphonyl
enzyme intermediate (Scheme 8) the negative charge on the phosphonyl oxygen
presumably takes up an equatorial position so that it is ca. 90Њ to the newly formed
serine O-P bond compared with the approximately tetrahedral angel formed with the
␤-lactam substrate (Scheme 1). Yet despite all these differences, the enzyme is capable
of catalysing the phosphonylation with extreme efficiency, indicative of a flexible
active site, which remains capable of stabilizing all the necessary transition states in
the reaction.
ACKNOWLEDGMENTS
We thank Dr. R.Aplin and Dr. C. Robinson at the University of Oxford for the ESMS studies.
REFERENCES
1. Page, M. I., and Williams, A. (1997) in Organic and Bioorganic Mechanisms, Longman, pp. 134–160.
2. Burgi, H. B., and Dunitz, J. D. (1983) Acc. Chem. Res. 16, 153.
3. Thatcher, G. R. J., and Kluger, R. (1989) Adv. Phys. Org. Chem. 25, 99.
4. Menger, F. M., (1985) Acc. Chem. Res. 18, 128; Menger, F. M., (1993) Acc. Chem. Res. 26, 206;
Daffron, A., and Koshland, D. E. (1971) Proc. Natl. Acad. Sci. USA, 68, 2463; Kirby, A. J., (1997)
Acc. Chem. Res. 30, 290; Kirby, A. J., (1996) Angew. Chem. Int. Ed. Engl. 35, 707.
5. Page, M. I., Laws, A. P., Slater, M. J., Stone, J. R., (1995) Pure Appl. Chem. 67, 711; Laws, A. P.,
Page, M. I., and Slater, M. J. (1993) Bioorg Med. Chem. Lett. 3, 2317.
6. Jacobsen, N. E., and Bartlett, P. A. (1981) J. Am. Chem. Soc. 103, 654.
7. Buckwell, S. C., Page, M. I., and Longridge, J. (1988) J. Chem. Soc. Perkin Trans 2, 1809.
8. Joris, B., De Meester, F., Galleni, M., Reckinger, G., Coyette, J., Fre`re, J.-M., and Van Beeumen, J.
(1985) Biochem. J. 228, 241.
9. Waley, S. G. (1992) in The Chemistry of ␤-Lactams (Page, M. I., Ed.), Ch. 6, Chapman and
Hall, London.
10. Oefner, C., D’Arcy, A., Daly, J. J., Gubernator, K., Charnas, R. L., Heinze, I., Hubschwerfen, C.,
and Winkler, F. K. (1990) Nature, 343, 284.
11. Lobkovsky, E., Moews, P. C., Lin, H., Zhao, H., Fre`re, J.-M., and Knox, J. R. (1993) Proc. Natl.
Acad. Sci., USA, 90, 11257; Jelsch, C., Mourey, L., Masson, J.-M., and Samama, J.-P.(1993) Proteins:
Struct. Funct. Genet. 16, 364; Moews, P. C., Knox, J. R., Dideberg, O., Charlier, P., and Fre`re, J.-
M. (1990) Proteins: Struct., Funct., Genet. 7, 156; Herzberg, O., and Moult, J.-M. (1987) Science
236, 694; Herzberg, O. (1991) J. Mol. Biol. 217, 701; Dideberg, O., Charlier, P., We´ry, J.-P., Dehottay,
P., Dusart, J., Erpicum, T., Fre`re J.-M., and Ghuysen, J.-M. (1987) Biochem J. 245, 911; Jelsch, C.,
Mourey, L., Masson, J.-M., and Samama, J.-P. (1993) Proteins: Struct., Funct., Genet. 16, 364.

␤-LACTAMASE-CATALYZED ACYL AND PHOSPHYL TRANSFER
95
12. Strynadka, N. C. J., Adachi, H., Jensen, S. E., Johns, K., Sielecki, A., Betzel, C., Sutoh, K., and
James, M. N. G. (1992) Nature 359, 700; Chen, C. C. H., Rahil, J., Pratt, R. F., and Herzberg, O.
(1993) J. Miol. Biol. 234, 165.
13. Ghuysen, J.-M. (1991) Annu. Rev. Microbiol. 45, 37; Matagne, A., and Fre`re, J.-M. (1995) Biochim.
Biophys. Acta. 1246, 109; Lamotte-Brasseur, J., Knox, J., Kelly, J. A., Charlier, P., Fonze´, E., Dideberg,
O., and Fre`re, J.-M. (1994) Biotechnol. Gen. Eng. Revs. 12, 189.
14. Laws, A. P., Page, M. I., and Slater, M. J. (1993) Bioorg. Med. Chem. Lett. 2317.
15. Fink, A. F. (1993) Chemtracts - Biochem. Mol. Biol. 3, 395.
16. Buckwell, S. C., and Page, J. I. (1987) Adv. Biosci. 65, 24.
17. Page, M. I., Vilanova, B., and Layland, N. J. (199) J. Am. Chem. Soc. 117, 12092.
18. Lobkovsky, E., Billings, E. M., Moews, P. C., Rahill, J., Pratt, R. F., and Knox, J. R. (1994)
Biochemistry 33, 6762.
19. Dubus, A., Normark, S., Kania, M., and Page, M. G. P. (1994) Biochemistry 33, 8577; Dubus, A.,
Ledent, P., Lamotte-Brasseur, J., and Frere, J.-M. (1996) Proteins : Struct., Funct. and Gen. 25, 473.
20. Page, M. I. (1990) in Comp. Med. Chem., (Sammes, P. G., Ed.), Vol. 2, pp. 45–60, Pergamon,
Oxford; Page, M. I. (1984) in The Chemistry of Enzyme Action (Page, M. I., Ed.), pp. 1–54,
Elsevier, Amsterdam.
21. Laws, A. P., and Page, M. I. (1989) J. Chem. Soc. Perkin Trans. 2, 1577; Laws, A. P., Layland, N.
J., Proctor, D. G., and Page, M. I. (1993) J. Chem. Soc. Perkin Trans. 2, 17.
22. Page, M. I., and Williams, A. (1977) in Organic and Bioorganic Mechanisms, pp. 152–157, Longmans.
23. Kovach, I. M. (1988) J. Enzyme Inhibition 2, 199; Kovach, I. M., and Enyedy, E. J. (1998) J. Amer.
Chem. Soc. 120, 258.
24. see however: Mock, W. L., and Chua, D. C. Y. (1995) J. Chem. Soc. Perkin Trans 2., 2069.
25. Murphy, B. P., and Pratt, R. F. (1988) Biochem. J. 256, 669.