Competitive endo- and exo-cyclic C-N fission in the hydrolysis of N-aroyl β-lactams

Article abstract of DOI:10.1139/v05-153

The balance between endo- and exo-cyclic C-N fission in the hydrolysis of N-aroyl β-lactams shows that the difference in reactivity between strained β-lactams and their acyclic analogues is minimal. Attack of hydroxide ion occurs preferentially at the exocyclic acyl centre rather than that of the β-lactam during the hydrolysis of N-p-nitrobenzoyl β-lactam. In general, both endo- and exo-cyclic C-N bond fission occurs in the alkaline hydrolysis of N-aroyl β-lactams, the ratio of which varies with the aryl substituent. Hence, the Bronsted β-values differ for the two processes: -0.55 for the ring-opening reaction and -1.54 for the exocyclic C-N bond fission reaction. For the pH-independent and acid-catalysed hydrolysis of N-benzoyl β-lactam, less than 3% of products are derived from exocyclic C-N bond fission.

Full text of DOI:10.1139/v05-153

1432
Competitive endo- and exo-cyclic C–N fission in
the hydrolysis of N-aroyl ␤-lactams1
Wing Y. Tsang, Naveed Ahmed, Karl Hemming, and Michael I. Page
Abstract: The balance between endo- and exo-cyclic C–N fission in the hydrolysis of N-aroyl β-lactams shows that
the difference in reactivity between strained β-lactams and their acyclic analogues is minimal. Attack of hydroxide ion
occurs preferentially at the exocyclic acyl centre rather than that of the β-lactam during the hydrolysis of N-p-nitro-
benzoyl β-lactam. In general, both endo- and exo-cyclic C–N bond fission occurs in the alkaline hydrolysis of N-aroyl
β-lactams, the ratio of which varies with the aryl substituent. Hence, the Brønsted β-values differ for the two processes:
–0.55 for the ring-opening reaction and –1.54 for the exocyclic C–N bond fission reaction. For the pH-independent and
acid-catalysed hydrolysis of N-benzoyl β-lactam, less than 3% of products are derived from exocyclic C–N bond fission.
Key words: β-lactams, hydrolysis, linear free energy relationships, strain.
Résumé : La balance entre la fission C–N endo- et exocyclique lors de l’hydrolyse de N-aroyl $-lactames montre que
la différence de réactivité entre des lactames tendus et leurs analogues acyles est minimale. L’attaque par l’ion hy-
droxyde se fait préférentiellement au niveau du centre acyle exocyclique plutôt que sur celui du $-lactame lors de
l’hydrolyse du N-p-nitrobenzoyl $-lactame. En général, il se produit des fissions des liaisons C–N tant endo- que exo-
cycliques lors de l’hydrolyse alcaline des N-aroyl $-lactames et le rapport des deux varie en fonction de la nature du
substituant. En conséquence, les valeurs $ de Brønsted diffèrent pour les deux processus étant égales à –0,55 pour la
réaction d’ouverture de cycle et de –1,54 pour la réaction de fission de la liaison C–N exocyclique. Les hydrolyses in-
dépendantes du pH et catalysées par les acides du N-benzoyl $-lactame fournissent moins que 3 % de produits dérivés
de la fission de la liaison C–N exocyclique.
Mots clés : $-lactames, hydrolyse, relations linéaires d’énergie libre, tension.
[Traduit par la Rédaction] Tsang et al. 1439
Introduction
lactam (2), which contains an endocyclic and an exocyclic
acyl centre both of which are potential sites for nucleophilic
attack. The hydrolysis of 2 could occur by two routes. For
example, the hydroxide ion could attack the exocyclic car-
bonyl to generate benzoic acid and the intact β-lactam
(Scheme 1, route a). The β-lactam could then undergo fur-
ther reaction with the hydroxide ion to give a ring-opened β-
amino acid. Alternatively, the hydroxide ion could attack the
endocyclic β-lactam carbonyl to give a ring-opened product
3 (Scheme 1, route b). Many would expect this to be the
dominant reaction pathway based on the belief that ring
strain increases the reactivity of β-lactams. We were inter-
ested in exploring the balance between endo- and exo-cyclic
C–N fission as a function of substituents in the aroyl group,
and herein report further evidence that the difference in re-
activity between strained β-lactams and their acyclic ana-
logues is minimal.
β-Lactam antibiotics such as penicillins (1) and cephalo-
sporins are often considered unusually reactive when com-
pared with normal amides because most of their biologically
important reactions involve the opening of the highly
strained four-membered ring (1, 2). The effect of ring strain
on the reactivity of β-lactams may be demonstrated by com-
paring the rates of the reactions of monocyclic β-lactams
with their acyclic analogues. For example, and perhaps con-
trary to expectation, the rate of alkaline hydrolysis of β-lac-
tams is, at most, 100-fold greater than that of an analogous
acyclic amide. The second-order rate constant for the hy-
droxide ion catalysed hydrolysis of N-methyl β-lactam is
only threefold greater than that for N,N-dimethyl acetamide
(3).
Incorporating a benzamide as a potential leaving group
into a simple azetidin-2-one gives the imide N-benzoyl β-
Experimental section
Received 3 March 2005. Published on the NRC Research
Press Web site at http://canjchem.nrc.ca on 11 November 2005.
Synthesis
W.Y. Tsang, N. Ahmed, K. Hemming, and M.I. Page.2
Department of Chemical and Biological Sciences, The
University of Huddersfield, Queensgate, Huddersfield,
HD1 3DH, UK.
N-Aroyl β-lactams were prepared according to the follow-
ing general procedure: To a stirred solution of azetidin-2-one
(0.5 g, 7.03 mmol) in dry dichloromethane (DCM, 20 mL)
was added 4,4-dimethylaminopyridine (0.1 g, 0.82 mmol) at
–78 °C, and a solution of aroyl chloride (1.57 g, 8.46 mmol)
1This article is part of a Special Issue dedicated to organic
reaction mechanisms.
in dichloromethane (10 mL) dropwise over
Triethylamine (0.98 mL, 7.02 mmol) was added dropwise
5 min.
2Corresponding author (e-mail: m.i.page@hud.ac.uk).
Can. J. Chem. 83: 1432–1439 (2005)
doi: 10.1139/V05-153
© 2005 NRC Canada

Tsang et al.
1433
Scheme 1.
(CDCl3) δ: 7.95 (2H, d, J = 6.68 Hz), 7.44 (2H, d, J =
8.66 Hz), 3.77 (2H, t, J = 5.5 Hz, CH2N), 3.12 (2H, t, J =
5.61 Hz, CH2CO). 13C NMR (CDCl3) δ: 165.01 (C=O),
163.84 (C=O), 139.51 (quaternary carbon), 131.14 (CH),
130.1 (quaternary carbon), 128.44 (ArCH), 36.73 (CH2N),
34.99 (CH2CO). HR-EI-MS calcd. for C10H8NO2Cl:
210.0316 [M + H]+; found: 210.0316.
N
O
O
Ph
2
Exocyclic
C-N fission
Endocyclic
C-N fission
(a)
(b)
1-(4′-Nitrobenzoyl)-1-azetidin-2-one
-
-
OH
OH
Yield: 1.23 g (80%); mp 130 to 131 °C. IR υmax (cm–1,
CHCl3): 3020, 1787, 1686, 1599, 1523, 1397, 1309, 1252,
1
1204, 992. H NMR (CDCl3) δ: 8.33 (2H, d, J = 8.75 Hz),
-
-
O2C
HN
8.14 (2H, d, J = 8.74 Hz), 3.85 (2H, t, J = 5.58 Hz, CH2N),
3.2 (2H, t, J = 5.57 Hz, CH2CO). 13C NMR (CDCl3) δ:
164.16 (C=O), 163.79 (C=O), 150.29 (quaternary carbon),
137.32 (quaternary carbon), 130.76 (CH), 123.28 (CH), 37.0
(CH2N), 35.5 (CH2CO).
O
O
O
+
NH
Ph
O
Ph
3
-
-
OH
OH
Kinetic procedures
Standard UV spectroscopy was carried out on a Cary 1E
UV–vis spectrophotometer (Varian, Australia) equipped with
a 12-compartment cell block. The instrument was used in a
double beam mode, allowing six reaction cells to be fol-
lowed in a single run. The cell block was thermostatted us-
ing a Peltier system.
-
O
O
-
O2C
+
NH2
Ph
The pH measurements were made with a φ40 pH meter
(Beckman, Fullerton, California) using a semi-micro calo-
mel electrode (Beckman). A calibration of the pH meter was
carried out at 30 °C using pH 6.99 0.01, 4.01 0.02, or
over 10 min forming a white precipitate. The reaction mix-
ture was stirred at –78 °C for 1 h and for a further 24 h at
ambient temperature. DCM (10 mL) was added to the reac-
tion mixture, and the solution was washed with water
(15 mL) and saturated brine (2 × 15 mL). The organic layer
was dried over anhydr. Na2SO4, and the solvent was re-
moved under reduced pressure by rotary evaporation at
30 °C to yield a pale yellow oil, which was purified by col-
umn chromatography.
9.95
0.02 calibration buffers. For solution pH ≥ 3 and
≤ 11, the pH was controlled by the use of ≤0.2 mol/L buffer
solutions of formate (pKa 3.75), ethanoate (pKa 4.72), MES
(pKa 6.1), MOPS (pKa 7.2), TAPS (pKa 8.4), CAPSO (pKa
9.6), and CAPS (pKa 10.4). Buffer solutions were prepared
by partial neutralization of their sodium salts to the required
pH. Hydroxide ion concentrations were calculated using pKw
(H2O) = 13.83 at 30 °C (4).
In all experiments, temperatures were maintained at 30 °C
and ionic strength at 1.0 mol/L with AnalaR grade KCl un-
less otherwise stated. AnalaR grade reagents and deionized
water were used throughout. Organic solvents were glass-
distilled prior to use and stored under nitrogen.
1-(4′-Methoxybenzoyl)-1-azetidin-2-one
Yield: 0.6 g (42%); mp 125–127 °C. IR υmax (cm–1,
CHCl3): 3020, 3009, 2975, 2912, 2842, 1784, 1668, 1606,
1
1325, 1259, 1195, 1108, 1028. H NMR (CDCl3) δ: 7.99
(2H, d, J = 8.88 Hz), 6.92 (2H, d, J = 8.97 Hz), 3.85 (3H, s,
CH3), 3.77 (2H, t, J = 5.44 Hz, CH2N), 3.02 (2H, t, J =
5.45 Hz, CH2CO). 13C NMR (CDCl3) δ: 165.23 (C=O),
163.92 (C=O), 132.57 (quaternary carbon), 131.92 (ArCH),
123.76 (quaternary carbon), 113.948 (ArCH), 55.21 (CH3),
36.45 (CH2N), 34.44 (CH2CO). HR-EI-MS calcd. for
C11H11NO3: 206.0812 [M + H]+; found: 206.0812.
Reactions studied by UV spectrophotometry were com-
menced by injections (20 µL) of acetonitrile stock solutions
2 × 10–2 mol/L of the substrate into the cells containing
preincubated buffer (2.0 mL). Final reaction cells contained
≤1% acetonitrile (v/v). The pH of the reaction cells was mea-
sured before and after each kinetic run at 30 °C; kinetic runs
experiencing a change >0.05 units were rejected. Reactant
disappearance or product appearance were followed at
absorbance change maxima for individual compounds. The
solubility of compounds was ensured by working within the
linear range of absorbance in corresponding Beer–Lambert
plots. Reaction concentrations were generally within the
range of 2 × 10–5 to 2 × 10–4 mol/L. Pseudo-first-order rate
constants from exponential plots of absorbance against time
or gradients of initial slopes were obtained using the Cary
Win UV kinetics application (Version 02.00(26)). The pH–
rate profiles were modelled to theoretical equations using
the Scientist program (V2.02, Micromath Software Ltd,
Saint Louis, Missouri).
1-Benzoyl-1-azetidin-2-one
Yield: 0.91 g (65%). IR υmax (cm–1, CHCl3): 3020, 1786,
1
1673, 1327, 1298, 1216, 1192. H NMR (CDCl3) δ: 8.00
(2H, d, J = 7.43 Hz), 7.61 (1H, t, J = 7.29 Hz), 7.49 (2H, t,
J = 7.83 Hz), 3.81 (2H, t, J = 5.49 Hz), 3.14 (2H, t, J =
5.49 Hz). 13C NMR (CDCl3) δ: 166.26 (C=O), 163.95
(C=O), 133.19 (PhCH), 131.83 (quaternary carbon), 129.69
(PhCH), 128.12 (PhCH), 36.77 (CH2), 35.05 (CH2).
1-(4′-Chlorobenzoyl)-1-azetidin-2-one
Yield: 0.85 g (58%). IR υmax (cm–1, CHCl3): 3020, 1788,
1673, 1593, 1404, 1324, 1284, 1217, 1093. 1H NMR
© 2005 NRC Canada

1434
Can. J. Chem. Vol. 83, 2005
Fig. 1. The pH–rate profile for the hydrolysis of N-benzoyl β-
lactam (2) in 1% acetonitrile – water (v/v) at 30 °C with
1.0 mol/L ionic strength.
HPLC analysis
HPLC analysis was carried out on the Beckman System
Gold HPLC with the 127 solvent module running a linear
gradient program from solvent A (94% water, 5% aceto-
nitrile, and 1% formic acid) to solvent B (20% water, 79%
acetonitrile, and 1% formic acid) over a period of 30 min,
using a RP-18 (5 µm) LiChrospher® 100 column, and moni-
tored by the 168 diode array detector at 228 nm.
0
-1
-2
-3
-4
-5
-6
-7
Results and discussion
The hydrolysis of N-benzoyl ␤-lactam
The rate of hydrolysis of 2 was studied by UV spectro-
photometry in buffered solutions of various pH with 1%
acetonitrile – water (v/v) and 1.0 mol/L ionic strength (KCl)
at 30 °C. The hydrolysis of 2 shows simple first-order kinetics
to give an observed rate constant (kobs). The pseudo-first-
order rate constant (kint) for hydrolysis, obtained by extrapo-
lating kobs to zero-buffer concentration, shows acid-catalysed,
pH-independent, and alkaline hydrolysis terms, the relative
importance of which vary with pH (Fig. 1). The pseudo-
first-order rate constant (kint) has a first-order dependence on
acid and hydroxide ion concentration, as indicated by the
slope of unity on the acid and basic limb, respectively, of the
plot of the logarithm of the rate constant (kint) against pH.
The pseudo-first-order rate constant can be fitted to eq. [1]
using the Scientist program:
0
2
4
6
8
10
12
pH
after mixing, shows no starting material but two pairs of
distinctive triplets in the methylene region. One pair of the
triplets belongs to the simple unsubstituted azetidin-2-one,
identified by spiking the reaction mixture with the com-
Rate
[1]
= kint = ko + kH[H+] + kOH[OH−]
[β-lactam]
1
pound. For the other pair of triplets, two-dimensional H,
13C HMBC NMR shows a cross peak between the exocyclic
carbonyl carbon and one of the triplets, which indicates that
the exocyclic carbonyl was still attached to the ring skeleton,
i.e., the β-amido acid hydrolysis product 3.
The rate constants for acid-catalysed hydrolysis (kH), pH-
independent hydrolysis (ko), and alkaline hydrolysis (kOH) of
2 are kH = 3.19 × 10–5 (mol/L)–1 s–1, ko = 8.92 × 10–7 s–1,
and kOH = 11.2 (mol/L)–1 s–1. The latter is in the range com-
monly observed for the alkaline hydrolysis of imides (5).
The fractions of the unsubstituted azetidin-2-one and the
β-amido acid 3 in H2O, calculated using the ratio of their
NMR signal integrals, are 0.19 and 0.81, respectively. The
product ratio was the same when the reaction was carried
out at pH 10.50, maintained by pH-stat. This indicates that
the rates of hydrolysis of the β-lactam and the exocyclic am-
ide have the same dependence on the hydroxide ion concen-
tration. As the relative rates between two competing
pathways in a parallel reaction are proportional to their
product distribution (7), the second-order rate constant for
the alkaline hydrolysis of the β-lactam of 2 (kOenHdo) is
9.07 (mol/L)–1 s–1 and that for the alkaline hydrolysis of the
exocyclic amide (kOexHo) is 2.13 (mol/L)–1 s–1 (Table 1).
The hydrolysis of N-benzoyl β-lactam shows no deute-
rium incorporation in the hydrolysis product β-amido acid,
indicative of a direct addition–elimination mechanism of
acyl transfer. The hydrolysis of N-acyl β-lactams thus occurs
by competitive attack at the endo- and exo-cyclic carbonyl
centres. The fact that exocyclic attack is competitive with
endocyclic is yet another example of the surprisingly small
difference in reactivity between β-lactams and acyclic
amides (2, 6). If the rate-limiting step is attack of hydroxide
ion on the β-lactam carbonyl carbon to form a tetrahedral in-
termediate, then the release of strain in the four-membered
ring would not be realized in the transition state, although
one may have anticipated a preferential conversion of a
three-coordinate to a four-coordinate carbon in the four-
Alkaline hydrolysis of N-benzoyl β-lactam
The site of nucleophilic attack by the hydroxide ion on 2
is not obvious because the rates of alkaline hydrolysis of β-
lactams with weakly basic leaving groups are similar to
those for analogous acyclic amides (2, 6). The rate of alka-
line hydrolysis was also studied by pH-stat to monitor the
progress of the reaction by titrating against a NaOH solu-
tion. One equivalent of NaOH to the β-lactam concentration
was consumed for the overall reaction, and the calculated
second-order rate constant (kOH) for the alkaline hydrolysis
of 2 from a single pH-stat experiment, carried out at
pH 10.50, was very similar to that determined by UV
spectrophotometry. At the end of the reaction, the solution
was analysed by HPLC with a UV detector. The presence of
benzoic acid is clearly demonstrated as a new signal. No
benzoic acid would be formed from the hydrolysis of the β-
amido acid 3 under these conditions (Scheme 1, route b).
Any hydrolysis to give benzoic acid via route b would also
consume more than 1 equiv. of hydroxide ion to the β-lactam
concentration in contrast to 1 equiv. as seen in the pH-stat
experiment. The benzoic acid must then be formed from the
alkaline hydrolysis of the exocyclic amide (Scheme 1, route a).
Competing exo- and endo-cyclic C–N bond fission was also
confirmed by studying the hydrolysis reaction in 1.0 mol/L
1
NaOD–D2O by H NMR. The spectrum, taken immediately
© 2005 NRC Canada

Tsang et al.
1435
Table 1. The second-order rate constants for the hydrolysis at the endo- and exo-cyclic carbonyl
centres of N-benzoyl β-lactam (2) in water and deuterium oxide with 1% acetonitrile (v/v) and
1.0 mol/L ionic strength (KCl) at 30 °C.
Observed rate
constants
Rate constants for
endocyclic C–N fission
Rate constants for
exocyclic C–N fission
kOH ((mol/L)–1 s–1)
kOD ((mol/L)–1 s–1)
SKIEa
11.2
15.1
2.13
2.87
0.74
9.07
12.2
0.74
aSKIE (solvent kinetic isotope effect).
Scheme 2.
membered ring. If the rate-limiting step is the breakdown of
the tetrahedral intermediate or if the reaction is concerted,
then this appears to be another example of the reluctance of
four-membered rings to undergo facile opening (2, 8).
k1
-O
N
N
O
O
k-1
HO
O
The apparent second-order rate constant (kOapDp) for the al-
kaline hydrolysis of 2 in D2O, using pKwD2O = 14.699 (4) and
pD = pH meter reading + 0.40 (9), is 15.1 (mol/L)–1 s–1. The
fraction of the unsubstituted azetidin-2-one and the β-amido
Ph
Ph
-
HO
k2
1
acid 3 obtained from the H NMR signal integrals of the
products of the reaction in 1.0 mol/L NaOD–D2O is identi-
cal to that obtained in 1.0 mol/L NaOH–H2O. Therefore, the
second-order rate constant for the alkaline hydrolysis of the
exocyclic amide (kOexDo) is 2.87 (mol/L)–1 s–1 and that for the
alkaline hydrolysis of the β-lactam (kOenDdo) is 12.2 (mol/L)–1 s–1
(Table 1). As the ratio between the alkaline hydrolysis of the
exocyclic amide and the β-lactam is the same in water and in
deuterium oxide, the two reactions have, within experimen-
tal error, the same solvent kinetic isotope effect (kOH/kOD) of
0.74 (Table 1). The inverse solvent isotope effect is compati-
ble with a rate-limiting formation of the tetrahedral interme-
diate or its breakdown by anion expulsion (Scheme 2).
The rate-limiting step for the alkaline hydrolysis of acy-
clic amides is normally the breakdown of the tetrahedral in-
termediate with partial protonation on the leaving group amine
nitrogen (6, 10). However, with 2, protonation on the amide
nitrogen is thermodynamically unfavourable. The leaving
group may be expelled in the rate-limiting step as an amide
anion without N protonation 4 and is compatible with the
observed inverse solvent kinetic isotope effect. C–N bond
fission expelling an amide anion from the tetrahedral inter-
mediate would be rate limiting if the expulsion of the amide
anion is slower than the expulsion of an incoming hydroxide
ion in the intermediate, i.e., k2 < k–1 (Scheme 2). The pKa of
amides like benzamide is about 15, so the pKa of the conju-
gate acids of the groups expelled, water and amide, are very
similar. The relative rate of C–O fission expelling alkoxide
anions compared with C–N fission expelling amines of the
same basicity is not easy to predict. Some reactions appear
to favour amine expulsion (11), whereas others have faster
rates for oxygen (2). A large negative charge development
on the leaving group nitrogen is expected for expulsion of an
amide anion. The hydrolysis was further studied by using
linear free energy relationships to elucidate changes in effec-
tive charge in the transition state by monitoring the effect of
changing the aryl substituents on reactivity.
O
- N
O
HN
O
O
OH
O
Ph
Ph
N-aroyl β-lactams were studied as a function of pH in 1%
acetonitrile – water (v/v) with 1.0 mol/L ionic strength (KCl)
at 30 °C by UV spectrophotometry. The pseudo-first-order
rate constants (kint) for the buffer-independent hydrolysis were
obtained by extrapolating the observed rate constant to zero-
buffer concentration. The apparent second-order rate con-
stant (kOapHp) for alkaline hydrolysis was obtained from the
slope of the plot of the pseudo-first-order rate constant (kint)
against the hydroxide ion concentration. Using the fraction
1
of products obtained, as previously described using H NMR,
the apparent second-order rate constants (kOapHp) are then sep-
arated into the second-order rate constants for the alkaline
hydrolysis of the exocyclic amide (kOexHo) and for the alkaline
hydrolysis of the β-lactam (kOenHdo) (Table 2). The product ratios
between the unsubstituted azetidin-2-one and the β-amido
acid hydrolysis product, and hence the relative rates of endo-
and exo-cyclic C–N bond fission, depend on the nature of
the para-substituents in the benzamide residue. Electron-
withdrawing substituents favour attack at the exocyclic
amide, presumably reflecting the relative importance of the
electrophilicity of the carbonyl centre compared with the
nucleofugality of the leaving group. The more electron-
withdrawing is the substituent, the higher the ratio of
exocyclic to endocyclic C–N bond fission and the greater
the formation of the simple azetidin-2-one relative to the β-
amido acid 3 as products. Exocyclic C–N bond fission is the
dominant reaction in the alkaline hydrolysis of N-p-
nitrobenzoyl β-lactam in preference to β-lactam hydrolysis
(Table 2).
The observation that electron-withdrawing substituents
enhance the rate of reaction is compatible with either rate-
limiting formation of the tetrahedral intermediate or its
breakdown with amide anion expulsion occurring in the
Structure–activity relationships for the alkaline
hydrolysis of N-aroyl ␤-lactams
The rates of alkaline hydrolysis of some aryl substituted
© 2005 NRC Canada

1436
Can. J. Chem. Vol. 83, 2005
Table 2. The second-order rate constants (SD 5%) for the alkaline hydrolysis of N-aroyl β-lactams in 1% acetonitrile – water (v/v)
with 1.0 mol/L ionic strength (KCl) at 30 °C.
app
OH
endo
exo
k
Fraction of endocyclic
C–N fission
k
Fraction of exocyclic
C–N fission
k
OH
OH
((mol/L)–1 s–1)
57.7
((mol/L)–1 s–1)
((mol/L)–1 s–1)
31.2
pKa
p-NO2
3.44
0.46
26.5
0.54
p-Cl
p-H
p-OMe
o-COOH
3.99
4.20
4.47
5.28
18.2
11.2
8.08
2.60
0.76
0.81
0.90
1.00
13.8
9.07
7.27
2.60
0.24
0.19
0.10
0.0
4.37
2.13
0.81
—
Note: The pKa of the carboxylic acid corresponding to exocyclic aryl carboxamide is given as an indicator of the inductive effect of the substituent in
the aromatic ring. pKa values are taken from ref. 12.
Fig. 2. Brønsted plot for the second-order rate constant (kOH),
for the alkaline hydrolysis at the endocyclic (᭹) and exocyclic
(᭺) carbonyl centre of N-aroyl β-lactams against the pKa of the
corresponding benzoic acid in 1% acetonitrile – water (v/v) at
30 °C. I = 1.0 mol/L (KCl).
transition state. The difficulty with correlating the reaction
rate to the ionization of the leaving group is the lack of
available ionization constants of substituted benzamides.
The ionization constants of substituted benzoic acids, as ana-
logues of benzamides, were therefore used for the correla-
tion.
1.6
1.4
1.2
RCONH
S
-O
-O
HO
N
N
Ar
O
N
HO
β(endo) = -0.55
O
O
-
Ph
CO2
1.0
0.8
0.6
0.4
0.2
0
5
4
1
H
O
O
H
'R
NHR
β
(exo) = -1.54
H
OH
O
N
H
H
Ph
O
O
H
O
6
7
-0.2
3.0
3.5
4.0
4.5
5.0
5.5
The Brønsted-type plot of the rate constants for the alka-
line hydrolysis of the β-lactam ring of N-aroyl β-lactams
against the pKa of the corresponding benzoic acids is shown
in Fig. 2. This shows a good linear correlation, the slope of
which gives the Brønsted βlg value for the alkaline hydroly-
sis of N-aroyl β-lactams as –0.55. This indicates that in the
transition state of the rate-limiting step of hydrolysis, the ef-
fective charge on the nitrogen atom in the amine leaving
group becomes more negative relative to the reactant state
by 0.55. The effective charge on the nitrogen of β-lactams is
expected to be very similar to that of acyclic amides of +0.7
(13). There is no data on the effective charge on the nitrogen
in imides. If there is no levelling of the resonance effect,
then presumably the charge could be as high as 1.4. If amide
resonance is reduced, then, effectively, imides could be
treated as amides to which is attached an electron-
withdrawing acyl group not involved in further resonance
stabilization. This would make the effective charge on the
imide nitrogen greater than 0.7 but less than 1.4. A crude es-
timation may be obtained from the difference in σ and σ–
values for a para-acyl substituent, i.e., 0.50 and 0.84, respec-
tively (13). The additional positive charge on the imide
nitrogen due to a second electron-withdrawing but non-
resonating acyl substitutent would be (0.5/0.84) × 0.7 =
pKa
0.42. This would make the effective charge on the imide ni-
trogen +1.1. This will be used in subsequent discussions, but
a slightly smaller or larger value makes little difference to
the conclusions.
The observed Brønsted value of –0.55 for the alkaline hy-
drolysis of the β-lactam ring of N-aroyl β-lactams is compati-
ble with a late transition state for the formation of the
tetrahedral intermediate 5. The effective charge of +0.55 on
N in the transition state structure indicates little or no C–N
bond fission and that the nitrogen atom is still involved in
normal amide resonance with the aroyl residue (Scheme 3).
For comparison, the Brønsted βlg value for the alkaline hy-
drolysis of N-substituted amides is –0.07 (6), which was in-
terpreted in terms of general-acid-catalysed breakdown of
the tetrahedral intermediate owing to partial proton transfer
from water 6. The rates of alkaline hydrolysis of N-aryl β-
lactams generate a Brønsted βlg of –0.44, which was taken as
evidence of rate-limiting formation of the tetrahedral inter-
mediate (2, 14).
© 2005 NRC Canada

Tsang et al.
1437
Table 3. The apparent rate constants (SD 5%) and kinetic parameters for the acid-catalysed hydrolysis, pH-independent hydrolysis,
and alkaline hydrolysis of N-benzoyl β-lactam (2) in 1% acetonitrile – water (v/v) with 1.0 mol/L ionic strength (KCl) at different tem-
peratures.
k ((mol/L)–1 s–1)
20.0 °C
30.0 °C
8.92 × 10–7
11.2
3.19 × 10–5
40.0 °C
1.34 × 10–6
22.2
50.0 °C
2.01 × 10–6
37.7
3.78 × 10–4
60.0 °C
3.53 × 10–6
69.7
1.12 × 10–3
∆S‡ (J K–1 mol–1)
–228
–77.3 4.2
–19.1 6.7
∆H‡ (kJ mol–1)
H2O
HO–
H+
3.77 × 10–7
6.72
9.66 × 10–6
9
40.9 2.8
44.8 1.3
94.5 0.4
Scheme 4.
Scheme 3.
-O
HO
+0.55
Ar
N
δ−
HO-
+1.1
O
N
N
O
Ar
βlg = -0.55
O
Ar
HO
δ−
O
-
O
O
O
+
+
HO-
N
+
NH
O
O
O
Ar
Extrapolating the data for the Brønsted plot for the alka-
line hydrolysis of N-aryl β-lactams to the rate constant for
the alkaline hydrolysis of N-benzoyl β-lactam gives an ap-
parent pKa of –7.5 for the leaving group amine. Formally
this corresponds to the N protonation of benzamide, but as
protonation of amides corresponds to O rather than N
protonation, the predicted pKa for N protonation of an amide
is not unreasonable.
Ar
O
-
HO
O
+
H3O
O
H2O
+
Ar
Ar
formed in the hydrolysis in barely detectable amounts. Less
than 3% of the hydrolysis pathway now occurs by exocyclic
C–N fission. Furthermore, the acetate buffer probably af-
fects the relative rates of hydrolysis of the β-lactam and the
exocyclic amide. Buffer catalysis accounts for 17% of the
observed rate of hydrolysis under the same experimental
conditions as determined from kinetic measurements at dif-
ferent acetate concentrations. If, in an extreme case, the
buffer-catalysed hydrolysis of 2 occurs exclusively at the
exocyclic amide, a minimum of 17% of the product would
be benzoic acid. However, as only a maximum of 3% is at-
tributed to the hydrolysis of the exocyclic amide in the ace-
tate buffer at pH 4, the pH-independent hydrolysis of 2
probably occurs exclusively with endocyclic C–N bond fis-
sion. This is in contrast to the 4:1 ratio of β-lactam and
exocyclic amide hydrolysis seen for the alkaline hydrolysis
of 2. The exocyclic acyl centre must show a greater selectiv-
ity (βnuc) towards nucleophilic attack by water and hydrox-
ide ion, than that of the β-lactam.
The rate of hydrolysis of 2 was studied at different tem-
peratures by UV spectrophotometry (Table 3). The entropy
of activation (∆S‡) for the pH-independent hydrolysis is
−228 J K–1 mol–1, which is threefold more negative com-
pared with –77.3 J K–1 mol–1 for the alkaline hydrolysis. The
more negative entropy of activation for the pH-independent
hydrolysis compared to that for the alkaline hydrolysis sug-
gests a more ordered system for the pH-independent hydro-
The Brønsted-type plot for the alkaline hydrolysis of the
exocyclic amide of N-aroyl β-lactams against the pKa of the
corresponding benzoic acids is shown in Fig. 2. The plot is
again reasonably linear, but the dependence of the rate on
the pKa is much greater with an apparent Brønsted β of
–1.54. The correlation is intriguing, as the product of the
exocyclic amide hydrolysis and that of ionization of the
carboxylic acid are the same (Scheme 4). The Brønsted
value of –1.54 indicates that the effective charge at the reac-
tion centre in the transition state for the amide hydrolysis is
more negative than that in the reactant imide compared with
the charge in the carboxylate ion relative to that in the
undissociated acid. This indicates that there is a relatively
greater negative charge density in the transition state for the
hydrolysis reaction than that in the carboxylate anion. This
may indicate a rate-limiting formation of the tetrahedral in-
termediate with a late transition state or an early one in a
rate-limiting breakdown because in the tetrahedral interme-
diate the negative charge is more localized. A similar linear
free energy relationship was observed for the rates of alka-
line hydrolysis of esters against the pKa of the product
carboxylic acid, which gave a Brønsted value of –1.3 (15).
The localized negative charge on the oxygen anion in the tet-
rahedral intermediate is more dependent on the stabilizing
effect of electron-withdrawing groups than the more stable
delocalized negative charge in the carboxylate anion.
lysis. The solvent kinetic isotope effect (kH O/kD O) for the
pH-independent hydrolysis of 2 is 1.25, which in2dicates the
transition state in the rate-limiting step involves a degree of
proton transfer (Table 4). The entropy of activation (∆S‡)
and the solvent kinetic isotope effect (kH O/kD O) is compati-
ble with a hydrolysis mechanism involving nucleophilic at-
tack by water that is general-base-catalysed by another water
molecule such as that shown in 7.
2
pH-Independent hydrolysis
The rate of hydrolysis of 2 is pH independent from pH 2–
6 (Fig. 1). A product analysis using HPLC of the reaction at
pH 3.99 (0.1 mol/L acetate buffer), ionic strength 1.0 mol/L
(KCl), and at 60 °C was undertaken. The signal for 2 disap-
peared with time in an exponential manner, but that for ben-
zoic acid, identified by spiking with the compound, was
2
2
© 2005 NRC Canada

1438
Can. J. Chem. Vol. 83, 2005
Table 4. The rate constants for the acid-catalysed hydrolysis (kH), pH-independent hydrolysis (ko), and
alkaline hydrolysis (kOH) of the β-lactam of N-aroyl β-lactams in 1% acetonitrile – water (v/v) with
1.0 mol/L ionic strength (KCl) at 30 °C.
ko (s–1)
kH ((mol/L)–1 s–1)
k
((mol/L)–1 s–1)
endo
pKa
OH
p-NO2
3.44
3.51 × 10–6
1.38 × 10–5
26.5
13.8
9.07
p-Cl
p-H
3.99
4.20
1.23 × 10–6
2.68 × 10–5
8.92 × 10–7 (H2O)
7.16 × 10–7 (D2O)
6.04 × 10–7
3.19 × 10–5
p-OMe
o-COOH
4.47
5.28
7.75 × 10–5
7.27
2.60
Note: The pKa of the carboxylic acid corresponding to exocyclic aryl carboxamide is given as an indicator of the
inductive effect of the substituent in the aromatic ring. pKa values are taken from ref. 12.
The first-order rate constants (ko) for the pH-independent
hydrolysis of aryl substituted N-aroyl β-lactams were deter-
mined (Table 4) and generated a Brønsted βlg value of –0.77.
The Brønsted βlg value for the pH-independent hydrolysis is
slightly more negative than that for the alkaline hydrolysis
(βlg = –0.55) indicating a later transition state probably ow-
ing to water being a much weaker nucleophile than hydrox-
ide ion.
Scheme 5.
Ph
H2O
-
N
O2C
N
H
Ph
N
+
C
Ph
O
OH
O
O
HO
+
catalysed hydrolysis of N-benzoyl β-lactam occurs by an A1
mechanism with O protonation on the exocyclic carbonyl,
C–N bond fission would form a highly unstable acylium ion
and an enol amide (Scheme 5).
Acid-catalysed hydrolysis
The acid-catalysed hydrolysis of 2 in 1.0 mol/L hydro-
chloric acid was also followed by HPLC at 30 °C. The sig-
nal for 2 disappeared with time in an exponential manner,
but that for benzoic acid, identified by spiking with the com-
pound, did not show any significant increase. It is estimated
that benzoic acid formation accounted for less than 3% of
the hydrolysis pathway.
The second-order rate constants (kH) for the acid-
catalysed hydrolysis of the β-lactam of N-aroyl β-lactams in-
crease with electron-donating substituents in the aryl ring of
the amide leaving group (Table 4). The rate constants are
reasonably well-correlated with the pKa of the corresponding
benzoic acid to give a Brønsted βlg value of 0.68. The
change in effective charge of +0.68 indicates a transition
state with significant protonation on the amide leaving
group. There are three possible sites for protonation of 2: the
β-lactam nitrogen is the least basic compared with the β-
lactam and the exocyclic carbonyl oxygens. Protonation on
the exocyclic carbonyl oxygen is probably more favourable
than protonation on the β-lactam carbonyl oxygen, as the
carbonyl oxygen becomes less basic when it is attached to a
four-membered ring system (16). Also, protonation on the
exocyclic carbonyl oxygen may increase the nucleofugality
of the leaving group in β-lactam ring opening.
The second-order rate constants for the acid-catalysed
hydrolysis of 2 shows a greater dependence on temperature
than that for alkaline hydrolysis (Table 3). The entropy
of activation (∆S‡) for the acid-catalysed hydrolysis is
−19.1 J K–1 mol–1, which is threefold less negative compared
with –77.3 J K–1 mol–1 for alkaline hydrolysis. The smaller
entropy of activation for the acid-catalysed hydrolysis may
be indicative of an A1 pathway and a unimolecular break-
down of the conjugate acid of the N-aroyl β-lactam. The
acid-catalysed hydrolysis of the β-lactam in penicillins 1 and
other β-lactams has been suggested to involve an A1 mecha-
nism with protonation on the β-lactam nitrogen leading to
intermediate formation of an acylium ion (6, 16). If the acid-
In conclusion, it has been shown that the alkaline hydroly-
sis of N-aroyl β-lactams occurs at similar rates at the β-
lactam and the exocyclic amide centres. Despite the strain
energy of the four-membered ring, it is not released in the
transition state to lower the activation energy and favour ex-
clusive endocyclic C–N ring fission. The reluctance of four-
membered rings to open rapidly has been noted previously
(2). The competitive rates of hydrolysis at the endocyclic
and exocyclic carbonyl centres provide an opportunity to in-
vestigate the relationship between the specificity and reactiv-
ity of the acyl centres towards enzymes.
Acknowledgment
We thank the University of Huddersfield for financial sup-
port and Andrew Williams for discussion.
References
1. M.I. Page. In The chemistry of β-lactams. Edited by M.I. Page.
Blackie, Glasgow. 1992.
2. M.I. Page. Adv. Phys. Org. Chem. 23, 165 (1987).
3. N.J. Baxter, L.J.M. Rigoreau, A.P. Laws, and M.I. Page. J.
Am. Chem. Soc. 122, 3375 (2000); K. Bowden and K.
Bromley. J. Chem. Soc. Perkin Trans. 2, 2111 (1990).
4. A.K. Covington, R.A. Robinson, and R.G. Bates. J. Phys.
Chem. 70, 3820 (1966).
5. J.T. Edward and K.A. Terry. J. Chem. Soc. 3527 (1957).
6. M.I. Page. Acc. Chem. Res. 17, 144 (1984); P. Proctor, N.P.
Gensmantel, and M.I. Page. J. Chem. Soc. Perkin Trans. 2,
1185 (1982).
7. A.A. Frost and R.G. Pearson. Kinetics and mechanism: a study
of homogeneous chemical reactions. Wiley, New York. 1953.
8. M.I. Page, P. Webster, and L. Ghosez. J. Chem. Soc. Perkin
Trans. 2, 805 (1990).
9. P.K. Glascoe and F.A. Long. J. Phys. Chem. 64, 188 (1960).
© 2005 NRC Canada

Tsang et al.
1439
10. J.J. Morris and M.I. Page. J. Chem. Soc. Perkin Trans. 2, 679
(1980); 685 (1980); R.S. Brown, A.J. Bennet, and H.
Slebocka-Tilk. Acc. Chem. Res. 25, 481 (1992).
11. N. Gravitz and W.P. Jencks. J. Am. Chem. Soc. 96, 499 (1974).
12. E.A. Braude and F.C. Nachod. Determination of organic struc-
tures by physical methods. Academic Press, New York. 1955.
13. M.I. Page and A. Williams. Organic and bio-organic mecha-
nisms. Addison-Wesley Longman, Harlow, UK. 1997.
14. G.M. Blackburn and J.D. Plackett. J. Chem. Soc. Perkin Trans.
2, 1366 (1972).
15. P. Barton, A.P. Laws, and M.I. Page. J. Chem. Soc. Perkin
Trans. 2, 2021 (1994).
16. P. Wan, T.A. Modro, and K. Yates. Can. J. Chem. 58, 2423
(1980).
© 2005 NRC Canada