Syntheses and hydrolysis of basic and dibasic ampicillin esters tailored for intracellular accumulation

Article abstract of DOI:10.1016/S0968-0896(00)00255-8

Readily hydrolysable basic and dibasic esters of ampicillin were synthesised by alkylation of the carboxylate function of ampicillin to obtain prodrugs that may accumulate in cells and allow for an intracellular delivery of ampicillin (Fan et al., Bioorg. Med. Chem. Lett. 1997, 7, 3107). We found that the β-lactam ring cleavage and the hydrolysis of the ester function were competitive reactions. The prerequisite for biological activity of compounds of this type is therefore that ester hydrolysis proceeds faster than ring opening. Some synthesised compounds show promise as prodrugs since they displayed a reasonable stability and regenerate large quantities of bioactive ampicillin in broth.

Full text of DOI:10.1016/S0968-0896(00)00255-8

Bioorganic & Medicinal Chemistry 9 (2001) 493±502
Syntheses and Hydrolysis of Basic and Dibasic Ampicillin Esters
Tailored for Intracellular Accumulation
Isabelle Paternotte,a,b Hua Juan Fan,a Pascal Screve,a Michel Claesen,a
Paul M. Tulkensb and Etienne Sonveauxa,*
a
 Â
Unite de Chimie Pharmaceutique et de Radiopharmacie, Universite Catholique de Louvain,
Avenue E. Mounier 73 p.b. 7340, B-1200 Bruxelles, Belgium
b
Â
Unite de Pharmacologie Cellulaire et Moleculaire, Universite Catholique de Louvain,
Â
Â
Avenue E. Mounier 73 p.b. 7340, B-1200 Bruxelles, Belgium
Received 29 May 2000; accepted 4 October 2000
AbstractÐReadily hydrolysable basic and dibasic esters of ampicillin were synthesised by alkylation of the carboxylate function of
ampicillin to obtain prodrugs that may accumulate in cells and allow for an intracellular delivery of ampicillin (Fan et al., Bioorg.
Med. Chem. Lett. 1997, 7, 3107). We found that the b-lactam ring cleavage and the hydrolysis of the ester function were competitive
reactions. The prerequisite for biological activity of compounds of this type is therefore that ester hydrolysis proceeds faster than ring
opening. Some synthesised compounds show promise as prodrugs since they displayed a reasonable stability and regenerate large
quantities of bioactive ampicillin in broth. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
slowly hydrolysing esters give ring-opened derivatives in
phosphate bu€er at pH 7.4.
We described in a previous short communication1 the
discovery of the cellular accumulation and intracellular
antibacterial activity of basic ester prodrugs of ampicillin.
These were original observations for b-lactams since,
contrary to macrolides, ¯uoroquinolones and ansamy-
cins, these antibiotics do not accumulate into cells and are
therefore largely inecient against most forms of intra-
cellular infection.2,3 We observed that important points
to get intracellular accumulation of b-lactams were (i)
the esteri®cation of the carboxylate function and (ii) the
presence of a protonable amino group in the antibiotic.
We describe here the synthetic procedures giving access
to monobasic prodrug esters of ampicillin which feature
a marked cellular accumulation (the amino function is
situated in the sidechain of the drug). Moreover, we
illustrate in the particular case of ampicillin a general
procedure allowing one to modify, in a bioreversible
way, the carboxylate function of a drug into a basic ester
(an amino function being present in the alcoholic part of
this ester). Finally, we stress that the four-membered ring
of b-lactam antibiotics is weakened by esteri®cation.
Very labile esters could be promising prodrugs whereas
Results and Discussion
The context of the research: Design of esters of ꢀ-lactam
antibiotics
Benzylpenicillin methyl ester has been the most studied
compound. Early works (reviewed by Hamilton-Miller4)
showed that it was inactive in vitro as well as in vivo, at
least in man. The interest in penicillin esters thus faded
until double esters (like pivampicillin 1; Scheme 1)
proved to be active in man. While benzylpenicillin
methyl ester was not a substrate for esterases in larger
mammals, its acetoxymethyl double ester was enzyma-
tically hydrolysed and the generated benzylpenicillin
hydroxymethyl ester spontaneously decomposed to
benzylpenicillin and formaldehyde5 (Scheme 2).
It was discovered later on that human serum degraded
benzylpenicillin methyl ester by opening of the b-lactam
ring. Indeed, the ester was inactive not because its car-
boxylate function could not be regenerated, but because
its b-lactam ring was cleaved ®rst.6 Ring opening and ester
cleavage are competitive reactions. The quickest process
wins the race. Physico-chemical studies in bu€ers at
*Corresponding author. Fax: +32-2-764-7363; e-mail: son-
veaux@cmfa.ucl.ac.be
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(00)00255-8

494
I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
physiological pH7À9 established the kinetic network
presented in Scheme 3. The pH of 7.4 is situated in the
base-catalysed domain of both ester hydrolysis and b-lac-
tam ring opening.10 The productive pathway (kp) liberates
the active antibiotic while the non-productive pathway
(knp) yields inactive decomposition products. The gener-
ated b-lactam antibiotic itself eventually decomposes to
ring-opened compounds (kd). Page's team11 found that,
for benzylpenicillin methyl ester, the sum kp+knp was
sixteen times larger than kd. The rate constant kp being
small in this case, this means that knp equalled 16 kd. In
other words, the b-lactam ring of the ester was sixteen
times more labile than that of the parent drug. This proved
to be true even when kp was larger than knp, i.e. in the case
of ecient ester prodrugs. For example, pivampicillin
and bacampicillin featured knps respectively twelve and
14 times larger than kd.7,8
Why is the b-lactam ring of penicillins weakened by
esteri®cation of the carboxylate function? The ester
group is more electron-withdrawing than the carboxy-
late group, which increases both the electrophilicity of
the b-lactam carbonyl carbon and the leaving-group
ability of the b-lactam amino group.7,12 This electron-
withdrawing e€ect is re¯ected in the shift of the amine
pKa from 5.2 for benzylpenicilloic acid to 3.2 for its
methyl ester.11
Useful esters of b-lactam antibiotics have thus to feature
the following kinetic characteristics: kp>knp=12±16 kd.
The kp/knp competition has to be pushed in the good
direction by synthesising so called `activated esters', i.e.
labile esters. In the basic pH range, they have to be at
least 20±30 times more labile than the b-lactam ring of
the parent compound. However, a decent half-life of the
ester prodrugs in the serum has to be maintained. These
opposite requirements assign intrinsic limits to our
approach.
Activated esters are usually obtained by enhancing the
leaving group capability of the alcohol moiety by the
mesomeric e€ect (e.g. p-nitrophenol) or the inductive
e€ect (e.g. tri¯uoroethanol). Aromatic esters of b-lactam
antibiotics being dicult to synthesise, we considered
ampicillin esters of the type AMPI-COOCH2X. There
are severe limitations on X. It cannot be a carbonyloxy
group (e.g. as in pivampicillin 1), owing to the required
resistance to human serum esterases. Compound 5 should
be an exception (as stated before, the methyl ester of
benzylpenicillin is not a good substrate for human
esterases). The X function cannot be either an alkoxy
R0O- or a carbonylamino R0CON(R00)- group because
a-carboxymethyl ethers and a-carboxymethylamides
spontaneously cleave by an SN1 mechanism that does
not require base catalysis, just like a-chloromethyl
ethers. This uncatalysed process renders impossible the
controlled release of ampicillin from a stable prodrug
solution.13,14 To avoid a chloromethyl ether type of
®ssion, the lone pair of X has to be strongly involved in
mesomery. That is why we considered imide functions
(X=N(COR)2), i.e. compounds 2±4. This type of ester
was known to be much more rapidly hydrolysed than an
ethyl ester at pH 7.4.15
Scheme 1. Discussed ampicillin basic prodrugs.
Scheme 2. Degradation path of acyloxymethyl esters of penicillins in
human serum.
Compounds 1±3 and 5 are basic owing to the amino
function present in the phenylglycine sidechain of
ampicillin. Compounds 4a,b bear an additional amino
function in the alcoholic part of the ester. Attachment
of this alcoholic part to b-lactam antibiotics other than
ampicillin would thus render them basic as well. Com-
pounds 4a,b were therefore synthesised to enlarge the
Scheme 3. Kinetic network describing the hydrolysis of penicillin
esters at pH 7.4.

I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
495
scope of the work. Moreover, if a monobasic molecule
can theoretically be concentrated a hundred times in
acidic organelles, dibasic ones could reach a theoretical
maximum enrichment of ten thousand, based on the pH
e€ect only.16 The structural characteristics we wanted
for the basic alcoholic part of these esters were (i) the
absence of chiral centre (to avoid epimerisation prob-
lems when working with chiral synthons or the obten-
tion of a mixture of diastereoisomeric prodrugs when
using racemic synthons), (ii) the lack of protection of
the amino function (protections are dicult to remove
in the presence of the b-lactam ring) and (iii) a good
lipophilicity (to enhance the rate of cellular penetra-
tion). These requirements were met with 4a,b.
energy content due to the steric clash between the imide
carbonyl and the phenyl ring, impairing mesomery. By the
way, the methylene of the imide ring of the E isomer sits in
the deshielding area of the phenyl substituent, so that its
chemical shift goes up by 0.3±0.4 ppm relative to the
corresponding methylene of 10, the synthon used to
obtain 3. Moreover, the ole®nic hydrogen of 7a or 7b
sits in the deshielding area of the carbonyl anisotropy
double cone: the chemical shift of this hydrogen (7.53
ppm for 7a and 7.64 for 7b) ®ts well with the value cal-
culated by additive rules (7.2 ppm for the E isomer versus
6.7 for the Z one20).
The N-hydroxymethyl function of 8a,b, obtained by
condensation with formaldehyde, is incompatible with a
free amino group, because it would promote the ionisation
of the OH group and the expulsion of the imide anion.21
That is why the hydrobromide was engaged in the con-
densation with formaldehyde, instead of the free amine.
Synthesis. The synthesis of 3 and 4a,b are sketched in
Scheme 4. A key reagent was the phosphorus ylid
obtained by Michael addition of triphenyl (or tributyl)
phospine on maleimide.17 The double bond of 7a and
7b, obtained by the Wittig reaction, has the E con®gura-
tion. Stabilised ylids usually indeed give the thermo-
dynamically more stable ole®n.18,19 The Z con®guration
of these succinimide derivatives would have a higher
The bulky diisopropylamino function was used to prevent
auto-alkylation of 9a. We observed indeed that N-bromo-
methylphthalimide quantitatively alkylated triethylamine
Scheme 4. Synthesis of 4a: [a] (1) Cs2CO3, DMF; (2) Cl±CH2±CH2±N[CH(CH3)2]2; [b] (1) HBr; (2) aqueous CH2O; [c] PBr3; [d] see Scheme 5.

496
I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
(DMF, room temperature, 24 h), while diisopropylethyl-
amine did not react. We recognised afterwards, when
synthesising 9b, that the presence of two isopropyl groups
on the amine function was not an absolute requirement, as
long as the amine remained protected by protonation till
the coupling with ampicillin. During the coupling itself, a
competition was established between autoalkylation
and alkylation of the carboxylate function of ampicillin
but this latter reaction proved to be the fastest.
of similar magnitude. The half-lives of compounds 1±5
(=0.693/(kp+knp)), in phosphate bu€er, pH 7.4, 37 ꢀC,
were determined by HPLC (Table 1). The hydrolysis was
base-catalysed above pH 5.5 and followed ®rst-order
kinetics. The rate increased with the electron-with-
drawing power of the alcoholic part of the ester (compare
2 with 3), in agreement with the base-catalysed mechan-
ism. The HPLC trace of the mixture obtained after
pivampicillin 1 and prodrugs 2 and 3 stayed in the buf-
fer during 1 half-life is shown in Figure 1(a). The corre-
lation between the decrease in concentration of the
prodrugs and the increase in concentration of released
ampicillin is also shown (Fig. 1(b)). Ampicillin itself was
only marginally degraded in these conditions (t1/2 of
ampicillin was 2144 min). The kp:knp competition was in
favour of ampicillin for quickly hydrolysing compounds
1, 2, 5 (ratios of ampicillin versus ring opened products
of 66:34, 68:32 and 82:18, respectively). Less readily
hydrolysing compounds, 3 and 4a, generated a lower
amount of ampicillin (ratios of 29:71 and 41:59, respec-
tively: knp>kp). The addition of the released formalde-
hyde to ampicillin, giving an imidazolidinone ring,25 was
not an important side reaction in our conditions.
The reaction of N-bromomethylimides with ampicillin
(on way to 2±4) or of diiodomethane with ampicillin (on
way to 5) required the protection of the amino function
of the phenylglycine sidechain. This was done by in situ
condensation with benzaldehyde (Scheme 5). We veri-
®ed that alkylation in these conditions took place on the
carboxylate function of ampicillin (and not on its amino
function) by N-acylating compound 2 with acetic anhy-
dride in 90% yield. The acylated derivative featured two
1
amide NH protons in H NMR(the 6-CHN H and the
10-CHNH; see Scheme 1 for atom numbering).
A concern was also the possible epimerisation of the
phenylglycine sidechain (at 10-C) due to the transient
formation of its imine derivative by condensation with
That b-lactam ring opening was a major degradation
pathway was shown by LC±MS±MS (Tables 3±5).
1
benzaldehyde, but both the H and 13C spectra corre-
sponded to a single diastereoisomer. Ampicillin epimers
at 10-C are distinguishable even when using a 60 MHz
instrument.22 In fact, aldehydes and ketones are known
to react with ampicillin not to give an imine but an
imidazilidinone ring,23,24 so that the a-hydrogen of the
phenylglycine sidechain is not acidi®ed, as should be the
case if an imine was formed.
Table 1. Half-lives (min) of compounds 1±5
Compound
Half-lifea (min)
Pivampicillin 1
2
92
62
3
4a
5
191
143
26
Hydrolysis of the esters. Imidomethyl esters of ampi-
cillin are biologically active in cellular culture, as repor-
ted in our previous short communication.1 This means
that, at least in this medium, kp was larger than knp, or
aPhosphate bu€er (0.02 M, NaCl 0.15 M, ethanol 1% v/v, pH 7.4,
37 ꢀC).
Scheme 5. Synthesis of ampicillin esters 2±4.

I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
497
Samples of 1, 2 and 3 were kept in an ethanol:water
mixture (1:4, v/v) for 15 days at room temperature, and
then analysed. The obtained compounds are sketched in
Scheme 6. Besides the starting prodrugs and ampicillin,
5-R and 5-S ampicilloic acid esters 13 were detected,
resulting from the opening of the b-lactam ring by water
and epimerisation via an intermediate imine (equili-
brium [a]). 5-R, 5-S ampicilloic acid esters 15 (pathway
[b]) resulted from the decarboxylation of 13 (see refs 26
and 27 for mechanistic details). The ampicilloic acid
disymmetric double esters 16, resulting from the b-
lactam ring opening by ethanol, and the piperazine-2,5-
diones 14 were also found. The rearrangement of
ampicilline itself in a piperazine-2,5-dione is a known
reaction.28
The hydrolysis of active esters and the ring opening of
b-lactams are subject to general acid±base and nucleo-
philic catalysis.10 Enzymes may also accelerate both reac-
tions. To obtain a ®rst information on the usefulness of
our compounds in terms of e€ective release of the parent
antibiotic, we measured their antibacterial potency in
tryptic soy broth, in comparison with ampicillin. Table 2
gives their minimum inhibitory concentrations (MICs)
against S. aureus in this culture medium. The determina-
tion entails a 24 h incubation at 37ꢀC, which should allow
both productive (kp) and non-productive disappearance
of the prodrugs (knp). The MICs of 2 and 5 were of the
same order of magnitude as that of ampicillin, indicating
that a large part of these prodrugs followed a productive
degradation pathway. Conversely, slower hydrolysing
esters 3 and 4a displayed a lower activity than ampicillin,
suggesting that, for these compounds, knp indeed over-
whelmed kp. Results along the same line were obtained
after ageing of the solutions. The MICs had a tendency
to increase, probably due to the spontaneous degradation
of the released ampicillin.
Figure 1. Hydrolysis of pivampicillin (1), 2 and 3 (1 mM) in phosphate
bu€er (0.02 M, NaCl 1.5 M, pH 7.4, 1% ethanol v/v) at 37 ꢀC. (a)
HPLC traces of the reaction mixtures after ca. a half-life of each
compound (UV detection at 220 nm, uncorrected for di€erences on
molar adsorptivity coecients). a. is ampicillin; i.s. is the internal
standard, benzamide; 13 (two epimers) and 14 are identi®ed in Scheme
6. (b) Evolution of the relative molar concentrations (%) of the start-
ing compound and of released ampicillin (a.) as a function of time.
Perpectives
The method we used to synthesise basic and dibasic
esters of ampicillin is general and could be applied to
other penicillins and cephalosporins. The ester function
of the most promising b-lactam prodrugs is more labile
than their four-membered ring. Further work is in pro-
gress to fully characterise the cellular accumulation of
these esters and to check their fate in human blood as
well as their activity in vivo.
Table 2. Minimum inhibitory concentrations (MICs) of compounds
1±5
Compound
MICa
MICb
mM
mg/mL
mM
mg/mL
Ampicillin
Pivampicillin 1
2
3
4a
5
0.3Æ0.1
0.5Æ0.1
0.4Æ0.1
1.2Æ0.3
1.2Æ0.4
0.2Æ0.1
0.1Æ0.05
0.5Æ0.1
0.6Æ0.1
0.5Æ0.1
1.2Æ0.3
1.2Æ0.4
0.4Æ0.1
0.2Æ0.05
0.3Æ0.05
0.3Æ0.05
0.7Æ0.2
1Æ0.3
0.25Æ0.05
0.2Æ0.05
0.7Æ0.2
1Æ0.3
0.15Æ0.05
0.3Æ0.05
aAgainst Staphylococcus aureus using freshly prepared solutions. The
MIC is de®ned as the lowest concentration of antibiotic giving no
visible bacterial growth (naked-eye examination) after 24 h incubation
at 37 ꢀC in tryptic soy broth 37 ꢀC (initial pH 7.4), inoculum 106 CFU/
mL, dilutions tested: 0.05; 0.075; 0.1; 0.15; 0.2; 0.25; 0.3; 0.4; 0.5; 0.7;
1; 1.5 mg/mL.
Scheme 6. b-Lactam ring opening products identi®ed by LC±MS±MS
after prodrugs 1, 2 or 3 (0.2 mg/mL) were kept 15 days at room tem-
perature in an ethanol:water mixture (1:4). Rhas the same meaning as
in Scheme 1.
bSame determination using an aged solution (24 h standing in tryptic
soy broth (pH 7.4) at 37 ꢀC).

498
I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
Experimental
General
Solvents and reagents. Unless otherwise stated, com-
mercially available solvents and reagents were used.
4-[3-(Dimethyl-amino)propoxy]benzaldehyde was from
Aldrich Co.
IR spectroscopy. A Perkin±Elmer 457 spectrophotometer
was used.
NMR spectroscopy. NMRspectra were recorded on a
Bruker AC-300 instrument. Chemical shifts are reported
in d values (ppm) with TMS as an internal reference.
Splitting patterns are indicated as s (singlet), d (doublet),
t (triplet), q (quartet), m (multiplet). For the 13C data,
only selected signals are reported. The numbering of the
carbons is indicated in Scheme 1.
Scheme 7. Principal fragments observed by LC±MS±MS.
Table 3. LC±MS±MS determination of the products of solvolysis of
1 (15 days, rt, ethanol:water mixture, 1:4 v/v)
Product
Retention Epimerc [M+H]+
timeb
Main
fragmentd
HPLC analysis. Chromatographies were performed on
a Gilson apparatus, injection loop of 100 mL, using a
precolumn ®lled with reversed phase C18 powder and a
reversed phase C18 column (Adsorbosphere Alltech,
15cmÂ4.6 mm) at a ¯ow rate of 1 mL/min. Eluent A
(acetonitrile:methanol:aqueous bu€er, 10:10:80) was pas-
sed during 2 min, then a linear gradient of eluent B (same
components, 50:20:30) in eluent A from 0 to 100%, over
5 min. The washing with 100% eluent B was maintained
for 20 min. The aqueous bu€er was an ammonium for-
mate bu€er, 0.02 M (adjusted at pH 5 with formic acid).
The UV detection was performed at 220 nm. The "220s
were determined: ampicillin, 6675; pivampicillin 1, 6625;
Ampicillin
5.6
13.4
17.1
21.3
21.7
27.0
Ð
14.2
Ð
Ð
23.0
Ð
350
482
464
464
438
510
[A+H]+
160
13a
14a
1
[M+H±CO2]+ 438
[A+H]+
[A+H]+
[E]+
274
274
278
237
15a
16a
[B+2H]+
aSee Scheme 6, with R=R of 1 (see Scheme 1).
bIn min. See Experimental.
cRetention time of the epimer (see Scheme 6).
dSee Scheme 7.
Table 4. LC±MS±MS determination of the products of solvolysis of
2 (15 days, rt, ethanol:water mixture, 1:4 v/v)
compound 2, 43305; 3, 12526; 4a, 16802 l molÀ1 cmÀ1
.
Retention Epimerd [M+H]+
timec
Main
fragmente
The hydrolysis of the samples (100 mL) was stopped
by addition of 7 mL of HCl, 0.5 M (end pH=5). The
samples were placed in dry ice (for 3 h maximum) till the
injection was performed.
Product
Ampicillin
5.4
7.5
Ð
8.1
10.0
12.6
Ð
350
324
396
527
509
509
483
555
[A+H]+
[M+H±NH3]+ 307
[B+2H]+
237
160
15a
16a
13b
14b
2
9.4
12.0
15.7
18.4
18.8
21.7
[M+H±CO2]+ 483
LC±MS±MS analysis. These were run on a Thermo
Separation Product AS-3000 chromatograph (column:
Adsorphosphere Alltech, 25 cmÂ2.1 mm), coupled with
a TSQ-7000 mass spectrometer working in the APCI
mode (atmospheric pressure chemical ionisation). A
linear gradient of eluent B in eluent A was used, from 0
to 100% over 10 min (¯ow rate 0.2 mL/min). The eluent
and bu€er were the same as above. We realised ®rst an
LC±MS experiment to set the parameters for the LC±
MS±MS analysis, performed the same day.
[A+H]+
[A+H]+
[D]+
319
319
336
319
Ð
19.5
Ð
15b
16b
[A+H]+
aSee Scheme 6, with R=H.
bSee Scheme 6, with R=R of 2 (see Scheme 1).
cIn min. See Experimental.
dRetention time of the epimer (see Scheme 6).
eSee Scheme 7.
Ampicillin phthalimidomethyl ester, hydrochloride (2). A
mixture of ampicillin trihydrate (2 g, 5 mmol), potas-
sium bicarbonate (0.5 g, 5 mmol), benzaldehyde (1 mL,
10 mmol) and DMF (50 mL) was stirred at 0 ꢀC for 12 h
to give an imidazolidinone (see Scheme 5), not isolated.
Anhydrous magnesium sulfate (1.2 g, 10 mmol, heated
at 800 ꢀC for 24 h) was added to trap water, and, after a
few min, a further equivalent of potassium bicarbonate
(0.5 g, 5 mmol) and N-(bromomethyl)phthalimide (1.2 g,
5 mmol). After stirring 24 h at 0 ꢀC, the mixture was
poured into cold water (80 mL) and extracted with ethyl
acetate (3Â60 mL). The organic extract was washed with
brine (3Â50 mL), dried over magnesium sulfate and ®l-
tered. The ®ltrate, after evaporation under vacuum, was
Table 5. LC±MS±MS determination of the products of solvolysis of
3 (15 days, rt, ethanol:water mixture, 1:4 v/v)
Product
Retention Epimerc [M+H]+
timeb
Main
fragmentd
Ampicillin
5.6
14.0
18.5
22.5
23.4±25
27.7
Ð
14.2
Ð
350
559
541
541
515
587
[A+H]+
160
13a
14a
3
[M+H±CO2]+ 515
[A+H]+
[A+H]+
[D]+
351
351
336
351
Ð
15a
16a
Ð
[A+H]+
aSee Scheme 6, with R=R of 3 (see Scheme 1).
bIn min. See Experimental.
cRetention time of the epimer (see Scheme 6).
dSee Scheme 7.

I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
499
stirred at À15 ꢀC for 20 min in a 1:1 mixture of acetoni-
trile and water (40 mL) adjusted at pH 2.5 with 0.5 M
HCl. Water was added (40 mL) and the pH was adjus-
ted to 5.0 by adding potassium bicarbonate. This solu-
tion was freed of acetonitrile by evaporation under
vacuum, washed with ethyl acetate (4Â50 mL) and satu-
rated with sodium chloride. The formed precipitate was
collected by ®ltration and washed with dichloromethane
and ether. Yield 1.9 g, 76%, mp 159±160 ꢀC. Elemental
Exact mass: calcd for C10H13NO2, 179.094629; found,
179.094564.
Cyclohexylidene-N-hydroxymethylsuccinimide (11).
A
mixture of cyclohexylidenesuccinimide (1 g, 6 mmol), 20%
aqueous formaldehyde (10 mL, 66mmol) and DMF (four
drops) was heated at 100 ꢀC for 10min with stirring,
cooled and left to stand at room temperature. The white
needles formed were collected, washed with water and
dried over P2O5 (yield: 1.17g, 100%). Mp 157±159 ꢀC. 1H
.
analysis C25H25ClN4O6S H2O: calcd C, 53.33, H, 4.83,
N, 9.95, Na, 0.0; found C, 53.38, H, 4.94, N, 9.79, Na,
NMR(CDCl ) d=3,69 (t, J=7.8Hz, 1H, OH), 5.05 (d,
3
1
0.045. H NMR(CD OD): d=1.39 and 1.41 (2Âs, 6H,
J=7.8 Hz, 2H, N±CH2±O). 13C NMR(CDCl 3): d=29.7
(imide ring CH2), 61.9 (N±CH2±O). Exact mass: calcd
for C11H15NO3, 209.105194; found, 209.105114.
3
2ÂCH3), 4.40 (s, 1H, 3-CH), 5.07 (s, 1H, 10-CH), 5.46
and 5.57 (2Âd, J=4.2 Hz, 2H, 5- and 6-CH), 5.73 and
5.78 (2Âd, J=10.6 Hz, 2H, diastereotopic H0s of O±CH2±
N), 7.47 (m, 5H, C6H5), 7.88 and 7.94 (2Âm, 4 H, phtha-
Cyclohexylidene-N-bromomethylsuccinimide (12). A sus-
pension of the former compound (0.4 g, 2 mmol), phos-
phorus tribromide (0.52 g, 2 mmol) and toluene (2 mL)
was re¯uxed for 4 h. The solution was left to settle and
the supernatant was poured out. The insoluble residue
was triturated with hot toluene (1±5 mL). The solid
obtained after evaporation under vacuum of the com-
bined toluene solutions was triturated with dry ether
(3 mL). The solution was ®ltered and the ®ltrate evapo-
limido group). 13C NMR(CD OD): d=27.13 and 31.21
3
(2ÂCH3), 57.53 (6-C), 60.31 (10-C), 62.61 (2-C), 65.56 (5-
C), 68.72 (3-C), 71.32 (O±CH2±N), 167.92, 168.12, 168.85
and 173.62 (carbonyls). IR(KBr): 1790cm À1 (ester CˆO
stretch.), 1775 (b-lactam CˆO stretch.), 1750, 1720 and
1705 (imide asym. and sym. CˆO stretch.+sidechain
amide CˆO stretch.).
N-Acetylampicillin, phthalimidomethyl ester. Triethyl-
amine (0.4 mL, 2.8 mmol) and acetic anhydride (0.3 mL,
3 mmol) were added to a suspension of 2 (1 g, 2 mmol)
in CH2Cl2 (20 mL) at 0±5 ꢀC. The reaction mixture was
stirred at 0±5 ꢀC for 4 h, washed with water, dried over
magnesium sulfate and ®ltered. Ether was added to give a
colourless solid. The solid was collected by ®ltration,
washed with ether and dried over P2O5 in vacuo (yield,
0.9 g, 90%), mp 126±128 ꢀC. Elemental analysis C27H26
rated in vacuo ꢀto leave a yellowish solid (yield 0.35 g,
N±CH2±Br). 13C NMR(CDCl 3): d=30.73 and 31.74
(imide ring CH2 and N±CH2±Br). Exact mass: calcd for
C11H14NO2Br, 271.020790; found, 271.021280.
1
70%). Mp 110 C. H NMR(CDCl ) d=5.28 (s, 2H,
3
Ampicillin cycohexylidenesuccinimidomethyl ester, hydro-
chloride (3). The same procedure was used as for com-
pound 2. The yield was 60%. Mp 176±177 ꢀC. Elemental
.
.
N4O7S 0.75 H2O: calcd C, 57.49, H, 4.91, N, 9.93; found
analysis C27H33ClN4O6S 1.5 H2O: calcd C, 53.68, H, 6.01,
1
C, 57.48, H, 4.84, N, 9.93. 1H NMR (CDCl3): d=2.01 (s,
3H, CH3CO), 5.55 (q, J1=4.2 Hz, J=8.7 Hz, 1H, 6-CH),
5.64 (d, J=6.9 Hz, 1H, 10-CH), 6.85 (d, J= 6.9 Hz, 1H,
10-CHNH), 7.00 (d, J=8.7 Hz, 1H, 6-CHNH). 13C NMR
(CDCl3): d=23.08 (CH3CO), 166.36, 166.55, 169.77,
N, 9.27; found C, 53.46, H, 5.89, N, 9.25. H NMR
(CD3OD): d=1.38 and 1.41 (2Âs, 6H, 2ÂCH3), 1.65 (m,
6H), 2.26 (m, 2H) and 3.03 (m, 2H) (cyclohexylidene
substituent), 3.36 (s, 2H, imide ring CH2), 4.38 (s, 1H, 3-
CH), 5.13 (s, 1H, 10-CH), 5.46 (d, J=4.2 Hz, 1H) and
5.57 (m, 3H) (5-CH, 6-CH and O±CH2±N), 7.45±7.53
À1
169.98 and 172.78 (carbonyls). IR(KBr): 1780 cm
(ester+b-lactam CˆO stretch.), 1755, 1730 and 1650
(2Âm, 5H, C6H5). 13C NMR(CD OD): d=26.99 and
3
(imide+amides CˆO stretch.).
27.14 (2ÂCH3), 57.50 (6-C), 60.03 (10-C), 62.86 (2-C),
65.57 (5-C), 68.69 (3-C), 71.31 (O±CH2±N), 168.15,
168.92, 169.62, 173.64 and 174.78 (carbonyls). IR(KBr):
1800±1750 cmÀ1 (several unresolved bands) and 1700
(carbonyls CˆO stretch.), 1655 (CˆC stretch.).
Cyclohexylidenesuccinimide (10). Maleimide (1.0 g,
10mmol) and tri-n-butylphosphine (2.1 mL, 10mmol) in
glacial acetic acid (15 mL) were heated at 80ꢀC for 15min
and then left to stand for 1 h. The solvent was evaporated
at 50ꢀC (10 mm Hg) and the residue coevaporated with
toluene (2Â20mL), leaving the ylid as a pink oily residue.
Cyclohexanone (5 mL, 50 mmol) and K2CO3 (1.38 g,
10 mmol) were added. The mixture was heated at 100ꢀC
for 4 h and the excess of cyclohexanone was removed
under vacuum (1 mm Hg) to give a red residue. This resi-
due was partitioned between water and CH2Cl2. The
organic extract was dried (MgSO4) and evaporated. The
residue was crystallised in C2H5OH (yield 0.34 g, 20%).
4-(2-Diisopropylaminoethoxy)benzaldehyde (6a). 2-Di-
isopropylamino-ethyl chloride hydrochloride (3.2 g,
16 mmol) was partitioned between 5% Na2CO3 ans
CH2Cl2. The organic phase was dried on MgSO4, ®ltered
and evaporated, leaving the free amine (2.2 g, 84%).
4-Hydroxybenzaldehyde (1.65 g, 13.5 mmol) was dis-
solved in absolute methanol (120mL, dried on Mg).
Caesium carbonate (2.28 g, 7 mmol) was added. After stir-
ring for 20 min at room temperature, the solvent was eva-
porated under vacuum (40 ꢀC, 3 h). The residue was
dissolved in DMF (dried on CaH2), and the mixture was
evaporated again to remove all traces of methanol. A sec-
ond volume of DMF was added (80 mL), followed by di-
isopropylaminoethyl chloride (2.2 g, 13.5 mmol). The
mixture was heated at 65 ꢀC for 24 h, then ®ltered. The ®l-
trate was concentrated and distilled to give a colourless
Mp 166 ꢀC. H NMR(CDCl ) d=1.65 (m, 6H), 2.16 (t,
1
3
J=6Hz, 2H) and 2.98 (t, J=6.3Hz, 2H) (cyclohexylidene
substituent), 3.25 (s, 2H, imide ring CH2), 8.62 (1H, NH).
13C NMR(CDCl ): d=29.45 (imide ring CH2), 116.41
3
and 158.53 (ole®nic Cs), 170.24 and 173.90 (imide
carbonyls). IR(KBr): 3170cm À1, 3060 (NH stretch.),
1750, 1720 and 1710 (CˆO stretch.), 1655 (CˆC stretch.).

500
I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
liquid (yield 2.4 g, 71%). Bp 137±138 ꢀC/0.9 mm Hg. H
Ampicillin, [4-(2-diisopropylaminoethoxy)benzylidene]suc-
cinimidomethyl ester, hydrochloride, (4a). Ampicillin tri-
hydrate (0.24 g, 0.6 mmol) was alkylated as described
for 2, i.e. protected by benzaldehyde and esteri®ed. The
ester was extracted in ethyl acetate and then depro-
tected. When the aqueous deprotection mixture (freed
from acetonitrile by partial evaporation) was saturated
with sodium chloride, a yellow gum appeared. It was
washed with dry ether and dried over P2O5 (0.3 g, 65%).
The solid so formed was triturated with isopropanol
(dried over MgSO4) and the mixture was ®ltered on an
HPLC ®lter (Millipore, HVHP, 0.45mm) to remove
sodium chloride. Isopropanol was partially evaporated
under vacuum. A suspension was obtained, that was again
®ltered on an HPLC ®lter. This solution was evaporated,
giving a solid that was dried over P2O5. Mp 170±172 ꢀC.
1
NMR(CDCl ): d=0.96 (d, J=6.5 Hz, 12 H, 4ÂCH3),
3
2.76 (t, J=7.3 Hz, 2H, CH2±N), 2.96 (m, 2H, CH±N),
3.88 (t, J=7.3 Hz, 2H, CH2±O), 6.91 and 7.74 (2Âd,
J=8.7 Hz, 4 H, C6H4), 9.79 (s, 1 H, CHO). 13C NMR
À1
(CDCl3): d=190.58 (CHO). IR(®lm): 1690 cm
stretch.). Exact mass: calcd for C15H23NO2, 249.172879;
(CˆO
found, 249.173264.
[4 - (2 - Diisopropylaminoethoxy) benzylidene] succinimide
(7a). Maleimide (1.6 g, 16 mmol) and triphenylphos-
phine (4.2 g, 16mmol) were stirred in glacial acetic acid
(30 mL) at 70ꢀC for 0.5 h. Ether was added. The ylid
precipitated as a white solid. It was dried in vacuo over
P2O5 (yield: 5.1 g, 14mmol, 89%). This ylid, DMSO
(30 mL) and 4-(2-diisopropylamino-ethoxy)benzalde-
hyde (3.5 g, 14 mmol) were stirred at 80 ꢀC for 3 h. A red
coloration developed. DMSO was removed under
reduced pressure at 60 ꢀC, leaving a red glue. Trituration
with dry ether gave a solid that was ®ltered and crystal-
.
Elemental analysis C36H47Cl2N5O7S 2.8 H2O: calcd C,
53.04, H, 6.50, N, 8.59; found, C, 53.04, H, 6.38, N,
1
8.49. H NMR(CD OD): d=1.61±1.66 (multiplet, 18
3
H, 6ÂCH3), 3.88 (t, J=4.7 Hz, 2H, OCH2CH2N), 3.92
(d, J=1.8 Hz (allylic coupling), 2H, succinimide ring
CH2), 4.05 (m, J=6.5 Hz, 2H, isopropyl CHs), 4.60 (s,
1H, 3-CH), 4.64 (t, J=4.7 Hz, 2H, OCH2CH2N), 5.31
(s, 1H, 10-CH), 5.67 and 5.77 (2Âd, J= 4.0 Hz, 2H, 5-
and 6-CH), 5.86 (s, 2H, OCH2N), 7.33 (d, J=8.8 Hz,
2H) and 7.44±7.62 (8H, 3 multiplets) (C6H5, C6H4 and
lised in CH2Cl2 (yield: 3.5 g, 76%). Mp 160±162 ꢀC. H
1
NMR(CDCl 3): d=1.04 (d, J=6.5 Hz, 12H, 4ÂCH3),
2.82 (t, J=7.3 Hz, 2H, CH2±N), 3.04 (m, J=6.5 Hz, 2 H,
CH±N), 3.56 (d, J=2.3 Hz, 2H, imide ring CH2), 3.93 (t,
J=7.3 Hz, 2H, CH2±O), 6.94 and 7.40 (2Âd, J=8.8 Hz,
4H, C6H4), 7.53 (t, J=2.3 Hz, 1H, ole®nic H). 13C NMR
(CDCl3): d=35.09 (imide ring CH2), 121.06 and 134.87
(ole®nic carbons), 171.26 and À1174.23 (imide carbonyls).
ole®nic CH). 13C NMR(CD OD): d=17.38 and 19.09
3
(diastereotopic isopropyl CH3s), 27.18 and 31.22
(2ÂCH3 on the thiazolidine ring), 34.86 (imide ring
CH2), 57.53 (6-C), 60.34 (10-C), 63.09 (2-C), 65.61 (5-
C), 68.74 (3-C), 71.36 (O±CH2±N), 168.11, 168.91,
171.19, 173.63 and 174.83 (carbonyls). IR(KBr) 1775
cmÀ1 and 1710 (CˆO stretch.), 1640 (CˆC stretch.).
IR(KBr) 1755 and 1705 cm
(CˆO stretch.), 1645
(CˆC stretch.). Exact mass: calcd for C19H26N2O3,
330.194343; found, 330.194602.
[4-(2-Diisopropylaminoethoxy)benzylidene]-N-hydroxy-
methylsuccinimide, hydrobromide (8a). The above
described compound (2 g, 6 mmol) was dissolved by
shaking 2 h at room temperature with one equivalent of
HBr in water (70 mL). After ®ltration, water was eva-
porated under vacuum. The obtained white solid was
washed with a large quantity of acetone (dried over
K2CO3). It was dissolved in water (7 mL) at 80 ꢀC. An
equivalent of aqueous formaldehyde (0.486 g, 37%,
6 mmol) was added and the mixture was heated at
100 ꢀC for 20 min. Evaporation under vacuum gave a
white solid that was crystallised in ethanol. The crystals
were ®ltered, washed with acetone and dried over P2O5
(yield: 1.3 g, 50%). Mp 207±208 ꢀC. 1H NMR
([D6]DMSO): d=4.83 (2H, N±CH2±O), 6.35 (1H, OH).
13C NMR([D 6]DMSO): d=33.61 (imide ring CH2),
60.52 (N±CH2±O). Elemental analysis C20H29BrN2O4:
calcd C, 54.42, H, 6.62, N, 6.35; found C, 54.35, H, 6.84,
N, 6.37.
[4 - (3 - Dimethylaminopropoxy)benzylidene]succinimide
(7b). The same procedure was used as for the diisopro-
pyl analogue, except that CHCl3 was used as a solvent
for the Wittig reaction instead of DMSO. Yield: 73%.
ꢀ
1
Mp 187.0±189 C. H NMR(CDCl and CF3COOH):
3
d=2.30 (m, 2H, NCH2CH2CH2O), 3.03 (d, J=5.0 Hz,
6H, 2ÂCH3), 3.45 (m, 2H, CH2±N), 3.68 (s, 2H, imide
ring CH2), 4.16 (t, 2H, CH2±O), 6.91 and 7.43 (2Âd,
J=8.5 Hz, 4H, C6H4), 7.64 (s, 1H, ole®nic CH). 13C
NMR(CDCl and CF3COOH): d=34.79 (imide ring
3
CH2), 120.47 and 136.67 (ole®nic carbons), 173.11
À1
and 176.04 (imide carbonyls). IR(KBr): 1740 cm
and 1700 (CˆO stretch.), 1650 (CˆC stretch.). Exact
mass: calcd for C16H20N2O3, 288.147393; found,
288.146918.
[4 - (3 - Dimethylaminopropoxy)benzylidene] - N - hydroxy-
methylsuccinimide, hydrobromide (8b). The HBr salt of
the former compound was obtained by dissolution in tri-
¯uoroacetic acid, treatment with one equivalent of con-
centrated aqueous HBr and evaporation. The residual
solid was washed with ether. The same reaction procedure
was then used as for the diidopropyl analogue. Yield
[4-(2-Diisopropylaminoethoxy)benzylidene]-N-bromome-
thylsuccinimide, hydrobromide (9a). A mixture of the
above described compound (0.6g, 1.36 mmol), ClCH2±
CHCl2 (3.5mL, dried over P2O5) and PBr3 (1.2 mL,
12.7mmol) was heated at 120 ꢀC for 4 h. The hot mixture
was ®ltered and evaporated under reduced pressure,
leaving a yellow solid that was dried over P2O5. It was
crystallised in CH2Cl2 (yield: 0.4 g, 59%). Mp 207±
209 ꢀC. 1H NMR(CDCl 3): d=5.34 (s, 2H, N±CH2±Br).
13C NMR(CDCl 3): d=30.89 and 34.07 (imide ring CH2
and N±CH2±Br).
59%. Mp 198 ꢀC. H NMR([D ]DMSO): d=4.86 (d,
1
6
J=6.5Hz, 2H, N±CH2±O), 6.38 (t, J=6.5 Hz, 1H, OH).
13C NMR([D 6]DMSO): d=33.86 (imide ring CH2), 60.73
(N±CH2±O). Elemental analysis C17H23BrN2O4: calcd
C, 51.14, H, 5.81, N, 7.02; found C, 51.04, H, 5.97, N,
7.09.

I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
501
[4-(3-Dimethylaminopropoxy)benzylidene]-N-bromomethyl-
Acknowledgements
succinimide, hydrobromide (9b). The procedure was the
same as for the diisopropyl analogue, except that
acetonitrile (dried over P2O5) was used as a solvent
instead of trichloroethane. Yield 69%. Mp 224±225 ꢀC.
1H NMR(CDCl 3): d=5.35 (s, 2H, N±CH2±Br). 13C
NMR(CDCl 3): d=30.90 and 34.12 (N±CH2±Br and
imide ring CH2).
I. Paternotte had a fellowship of the Belgian Fonds
pour la Formation a la Recherche dans l'Industrie et
dans l'Agriculture. This work was supported by the
Belgian Fonds National de la Recherche Scienti®que
(grant no. 1.5113.00), the Belgian Fonds de la
Â
Recherche Scienti®que Medicale (grant no. 3.4516.94)
and the Fonds Special de Recherche from the Universite
 Â
Ampicillin [4 - (3 - dimethylaminopropoxy)benzylidene]-
succinimidomethyl ester, hydrochloride (4b). Ampicillin
trihydrate (0.27 g, 0.67 mmol) was alkylated as described
for 4a and the product was isolated in the same way. Yield
62%. Mp 188±190 ꢀC. Elemental analysis C33H41Cl2N5
Catholique de Louvain. We thank Professor E. De
Ho€mann for his help in mass spectroscopy. We also
thank the Leo Pharmaceutical Products Co., Denmark,
for the gracious gift of pivampicillin.
.
O7S 2H2O: calcd C, 52.24, H, 5.98, N, 9.23; found C,
52.15, H, 5.89, N, 8.85. 1H NMR(CD 3OD): d=1.14 and
1.17 (2Âs, 2Â3 H, 2ÂCH3), 2.28 (m, 2H, OCH2
CH2CH2N), 2.95 (s, 6H, N(CH3)2), 3.38 (t, J= 7.9 Hz,
2H, OCH2CH2CH2N), 3.69 (s, 2H, succinimide ring
CH2), 4.19 (t, J=5.6 Hz, 2H, OCH2CH2CH2N, 4.39 (s,
1H, 3-CH), 5.11 (s, 1H, 10-CH), 5.45 and 5.56 (2Âd,
J=3.9 Hz, 2Â1H, 5- and 6-CH), 5.64 (s, 2H, O±CH2±N),
7.07 (d, J=8.6 Hz, 2H) and 7.45±7.58 (m, 8H) (C6H5,
References and Notes
1. Fan, H. J.; Paternotte, I.; Vermander, M.; Li, K.; Beaujean,
M.; Scorneaux, B.; Dumont, P.; Osinski, P.; Claesen, M.;
Tulkens, P. M.; Sonveaux, E. Bioorg. Med. Chem. Lett. 1997,
7, 3107.
2. van den Broek, P. J. Rev. Infect. Dis. 1989, 11, 213.
3. Tulkens, P. M. Eur. J. Clin. Microbiol. Infect. Dis. 1991, 10,
100.
4. Hamilton-Miller, J. M. T. Chemotherapia 1967, 12, 73.
5. Agersborg, H. P. K.; Batchelor, A.; Cambridge, G. W.;
Rule, A. W. Brit. J. Pharmacol. 1966, 26, 649.
6. Nielsen, N. M.; Bundgaard, H. J. Pharm. Pharmacol. 1988,
40, 506.
7. Toney-Parker, T. E.; Notari, R. E. Drug Stab. 1995, 1, 40.
8. Fujiwara, H.; Kawashima, S. Chem. Pharm. Bull. 1985, 33,
1202.
C6H4 and ole®nic CH). 13C NMR(CD OD): d=25.67
3
(OCH2CH2CH2N), 27.17 and 31.22 (2ÂCH3), 34.86
(imide ring CH2), 43.66 (N(CH3)2), 56.57 (OCH2CH2
CH2N), 57.52 (6-C), 60.33 (10-C), 63.08 (2-C), 65.59 (5-
C), 66.26 (OCH2CH2CH2N), 68.72 (3-C), 71.35 (O±
CH2±N), 168.11, 168.89, 171.22À, 1173.61 and 174.87
(carbonyls). IR(KBr) 1770 cm
stretch.), 1635 (CˆC stretch.).
and 1710 (CˆO
Diampicillylmethane, dihydrochloride (5). Potassium bi-
carbonate (0.496 g, 4.96 mmol) and benzaldehyde (1 mL,
9.92 mmol) were added to a suspension of ampicillin tri-
hydrate (2 g, 4.96mmol) in DMF (20 mL). The mixture
was stirred at 0±4 ꢀC for 12h. Anhydrous magnesium
sulfate (2.4 g, 19.9 mmol) was added and the mixture was
further stirred for 2±3 h. The reaction mixture was left at
4 ꢀC for 9 days, after the addition of CH2I2 (4 mL,
49.6 mmol) and of a further amount of potassium
bicarbonate (0.496 g, 4.96 mmol). Evaporation of the
mixture after ®ltration gave a yellow solid that was
washed with ether, dried over P2O5, dissolved in ethyl
acetate and submitted to ¯ash-chromatography (100g of
silica, eluent: ethyl acetate, Rf=0.85). The obtained bis-
pyrazolidinone was deprotected at pH 2.5 as described for
2. The precipitation with sodium chloride gave a gum.
After the brine was poured out, the gum was washed with
ether and dried in vacuo over P2O5. The resulting powder
was puri®ed by dissolution in isopropanol as described
for 4a. Yield: 36%. Mp 190±191 ꢀC. Elemental analysis
9. Nguyen, N.-A. T.; Mortada, L. M.; Notari, R. E. Pharm.
Res. 1988, 5, 288.
10. Connors, K. A.; Amidon, G. L.; Stella, V. J. In Chemical
Stability of Pharmaceuticals, a Handbook for Pharmacists; J.
Wiley & Sons: New York, 1986; pp 198±207, 274±283.
11. Proctor, P.; Gensmantel, N. P.; Page, M. I. J. Chem. Soc.,
Perkin Trans. II 1982, 1185.
12. Laws, A. P.; Layland, N. J.; Proctor, D. G.; Page, M. I. J.
Chem. Soc., Perkin Trans. II 1993, 17.
13. Vanwetswinkel, S.; Carlier, V.; Marchand-Brynaert, J.;
Fastrez, J. Tetrahedron Lett. 1996, 37, 2761.
14. Moreira, R.; Calheiros, T.; Cabrita, J.; Mendes, E.;
Pimentel, M.; Iley, J. Pharm. Res. 1996, 13, 70.
15. Saari, W. S.; Freedman, M. B.; Hartman, R. D.; King, S.
W.; Raab, A. W.; Randall, W. C.; Engelhardt, E. L.; Hirsch-
mann, R.; Rosegay, A.; Ludden, C. T.; Scriabine, A. J. Med.
Chem. 1978, 21, 746.
16. de Duve, C. T.; de Barsy, T.; Poole, B.; Trouet, A.; Tulk-
ens, P.; Van Hoof, F. Biochem. Pharmacol. 1974, 23, 249524.
17. Hedaya, E.; Theodoropulos, S. Tetrahedron 1968, 24,
2241.
18. Carey, F. A.; Sundberg, R. J. In Advanced Organic
Chemistry; Plenum Press: New York, 1983; Vol. B, pp 71±73.
19. March, J. In Advanced Organic Chemistry, Reactions,
Mechanisms and Structure; J. Wiley & Sons: New York, 1992;
p 960.
20. Silverstein, R. M.; Clayton Bassler, G.; Morrill, T. C. In
Spectrometric Identi®cation of Organic Compounds; J. Wiley &
Sons: New York, 1991; p 215.
.
C33H40Cl2N6O8S2 1.5H2O: calcd C, 48.89, H, 5.35, N,
10.36; found C, 49.00, H, 5.36, N, 10.09. 1H NMR
(CD3OD): d=1.39 and 1.45 (2Âs, 12 H, 4ÂCH3, 4.41 (s,
2 H, 3-CH), 5.08 (s, 2 H, 10-CH), 5.46 and 5.57 (2Âd,
J=4.0Hz, 2Â2H, 5- and 6-CH), 5.86 (s, 2 H, O±CH2±O,
7.47 (m, 10 H, C6H5). 13C NMR(CD 3OD): d= 27.18 and
31.10 (4ÂCH3), 57.53 (6-C), 60.32 (10-C), 65.47 (5-C),
68.71 and 71.31 (2-C and 3-C), 81.61 (O±CH2±O), 129.47,
21. Johansen, M.; Bundgaard, H. Arch. Pharm. Chem. Sci.
Ed. 1979, 7, 175.
22. Bird, A. E.; Steele, B. R.; Boles, M. O.; Gane, P. A. C. J.
Chem. Soc., Perkin Trans. I 1982, 563.
130.39, 131.14, 133.93 (C6H5), 167.80, 168.94 and
À1
173.60 (carbonyls). IR(KBr): 1770 cm
(ester and b-
lactam CˆO stretch.), 1685 (amide CˆO stretch.).

502
I. Paternotte et al. / Bioorg. Med. Chem. 9 (2001) 493±502
23. Hardcastle Jr., G. A.; Johnson, D. A.; Panetta, C. A. J.
Org. Chem. 1966, 897.
26. Bird, A. E.; Cutmore, E. A.; Jennings, K. R.; Marshall, A.
C. J. Pharm. Pharmacol. 1983, 35, 138.
24. Klixbull, U.; Bundgaard, H. Int. J. Pharmaceutics 1985,
27. Robinson-Fuentes, V. A.; Je€eries, T. M.; Branch, S. K. J.
Pharm. Pharmacol. 1997, 49, 843.
28. Bundgaard, H.; Hansen, J. Int. J. Pharmaceutics 1981, 9,
273.
È
23, 163.
25. Bundgaard, H.; Klixbull, U. Int. J. Pharmaceutics 1985,
27, 175.
È