Isolation and structure elucidation of the major degradation products of cefaclor formed under aqueous acidic conditions

Article abstract of DOI:10.1021/js960427x

The aqueous acidic degradation of the oral cephalosporin cefaclor was investigated. A number of degradation products were isolated and characterized. The degradation products can be loosely classified into three categories: thiazole derivatives, pyrazine

Full text of DOI:10.1021/js960427x

RESEARCH ARTICLES
Isolation and Structure Elucidation of the Major Degradation Products of
Cefaclor Formed under Aqueous Acidic Conditions
X
S
G
TEVEN W. BAERTSCHI, DOUGLAS E. DORMAN, JOHN L. OCCOLOWITZ, MONTE W. COLLINS, LARRY A. SPANGLE
,
REGORY A. STEPHENSON
,
AND
LESLIE J. LORENZ
Received October 15, 1996, from the Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285.
X
Final revised manuscript received January 27, 1997.
Accepted for publication February 7, 1997 .
this point, ∼85% of the cefaclor had degraded as determined by
Abstract
0 The aqueous acidic degradation of the oral cephalosporin
analytical rp-HPLC (Figure 1). The compounds formed under these
conditions were isolated and purified by the preparative HPLC
procedures described in the next section. The degraded solution also
contained a precipitate that was isolated by filtration and subse-
quently identified as elemental sulfur.1
cefaclor was investigated. A number of degradation products were
isolated and characterized. The degradation products can be loosely
classified into three categories: thiazole derivatives, pyrazine derivatives,
and simple hydrolysis or rearrangement products. Degradation pathways
Unbuffered Water DegradationsCefaclor (95 g) was dissolved in
are proposed that involve (1) hydrolysis of the
â-lactam carbonyl with
1
9 L of deionized water, stored in a glass container with a loose
subsequent rearrangement, (2) ring contraction of the six-membered
cephem nucleus to five-membered thiazole derivatives through an
episulfonium ion intermediate, and (3) attack of the primary amine of the
phenylglycyl side chain on the “masked aldehyde” at carbon-6 to form
fluorescent substituted pyrazines.
aluminum foil cover, and allowed to degrade at room temperature
for 9 months prior to isolation of the degradation products. At this
point, >90% of the cefaclor had degraded as determined by analytical
rp-HPLC (Figure 2). During this degradation, the acidity of the
solution changed from pH ∼6 to pH 2.75. Two compounds formed
under these conditions were isolated and purified by the preparative
HPLC procedures described (11 and 13). The degraded solution also
contained a precipitate that was identified as elemental sulfur.
Introduction
Degr a d a tion a t p H 5.5sCefaclor (200 g) was dissolved in 20 L
of water, the solution was stored in a glass container with a loose
aluminum foil cover, and the pH of the solution was adjusted to 5.5
with 5 N NaOH (the pH 5.5 degraded solution). The pH was adjusted
daily to ∼5.5 with 5 N NaOH to counteract the tendency of the
solution to become more acidic. The pH was maintained between 4.8
and 6.0 during the course of the degradation. After 3 months at room
temperature, cefaclor was shown by HPLC analysis to be ∼87%
degraded (Figure 3). The solution was fractionated by the preparative
HPLC procedures described in the next section. Two fractions, those
containing 8 and 14, were further purified and characterized. In
addition to these compounds, elemental sulfur was present as a
precipitate in the solution. The relative amount of elemental sulfur
was higher in the pH 5.5 solution than in the acidic solutions in which
cefaclor was degraded.
Knowledge of the structures of degradation products and
the degradation pathways by which they arise are an impor-
tant part in the understanding of the chemistry, stability, and
activity of drug products. For this reason, we investigated
the degradation pathways of the oral cephalosporin cefaclor
(
1) and recently characterized a novel aqueous degradation
1
product (2). The structure of 2 revealed a prominent acidic
degradation pathway involving a ring-contraction from a six-
membered cephem ring to a five-membered thiazole ring. This
new pathway of cefaclor degradation is distinct from the
previously established degradation pathway leading to 2-hy-
droxy-3-phenyl pyrazine (3).2
Examination of acidic aqueous degradations (degradations
in 0.1 N HCl and in unbuffered water) of cefaclor by high-
performance liquid chromatography (HPLC) reveals numerous
degradation products (see Figures 1-3). This report details
the isolation and characterization of the major degradation
products of cefaclor arising under acidic aqueous conditions,
along with proposals for pathways by which they form.
Evidence is presented for a generalized pathway to fluorescent
degradation products.
HP LCsAnalyticalsThe analytical scale rp-HPLC system utilized
a photodiode array detector to detect eluting peaks (deuterium lamp;
wavelength accuracy, (1.0 nm adjusted to the line spectrum of the
deuterium lamp at 656.1 nm; UV detection from 200 to 400 nm at
4
-nm resolution; 1-s interval between acquisition of successive spectra;
Waters 990, Milford, MA). The HPLC was run in a gradient mode
from 0 to 100% solvent B in 30 min and held at 100% solvent B for
1
0 min. Solvent A was an aqueous solution containing 2.4 g/L NaH2-
PO4 and 1 mL of H3PO4 per liter. Solvent B was a mixture of 60%
acetonitrile and 40% solvent A (v/v). The flow rate was 1.0 mL/min.
An octyldecyl silane (ODS) rp-HPLC column was used (4.6 × 250 mm,
Experimental Section
The solutions of cefaclor described were periodically monitored by
HPLC for degradation. The degradation profile was examined to
ensure that the major degradation product levels were increasing so
that isolation by preparative HPLC would be facilitated.
Degr a d a tion of Cefa clor s0.1 N HCl DegradationsCefaclor (95
g) was dissolved in 19 L of 0.1 N HCl and stored in a glass container
with a loose aluminum foil cover at room temperature. The solution
was allowed to degrade for 2 months prior to isolation of the
degradation products by preparative reversed-phase (rp)-HPLC. At
5
µM, YMC, Wilmington, NC).
PreparativesThe degraded solutions of cefaclor just described were
separated into 20 fractions by an automated HPLC and fraction
collection system. The sample solutions were placed in a pressurized
container, and aliquots of the sample were automatically injected onto
the column through a 10-mL sample loop. The system used a C-18
semipreparative rp-HPLC column (20 × 250 mm, 5 µM, YMC,
Wilmington, NC). The HPLC was run in a gradient mode from 0 to
1
00% solvent B in 35 min and held at 100% solvent B for 7 min.
X
Abstract published in Advance ACS Abstracts, April 1, 1997.
Solvent A was an aqueous solution of 0.1% acetic acid, and solvent B
5
26 / Journal of Pharmaceutical Sciences
S0022-3549(96)00427-3 CCC: $14.00
© 1997, American Chemical Society and
American Pharmaceutical Association
Vol. 86, No. 5, May 1997

Chart 1sa0.1 N HCl degradation product. bDeionized water degradation product. cpH 5.5 degradation product.
consisted of a mixture of 60% acetonitrile and 40% solvent A (v/v).
The flow rate was 10 mL/min. Fraction collector advance times were
determined manually. A total of 78, 117, and 45 preparative
injections were made for fractionation of the 0.1 N HCl, the unbuf-
fered, and the pH 5.5 degraded solutions, respectively. In each case,
the solvent was removed by lyophilization.
1. All data in this table were measured in the dimethyl sulfoxide/
trifluoroacetic acid (DMSO/TFA) solvent system described later.
2-Hydroxy-3-phenylpyrazine (3)sCompound 3 was obtained from
synthesis as described previously.1
D-Phenylglycine (4)sCompound 4 was obtained from the Aldrich
Chemical Company, Milwaukee, WI.
Selected fractions were further purified by rp-HPLC prior to
spectroscopic analysis. These fractions were re-dissolved in a solution
of 0.1% acetic acid and acetonitrile and purified by rp-HPLC (20 ×
D-Phenylglycyl-glycine (5)st-Butyloxycarbonyl-D-phenylglycine (1
mmol) and 1-hydroxybenzotriazole (1 mmol) were dissolved in 50 mL
of dichloromethane at room temperature. One equivalent of dicyclo-
hexylcarbodiimide (206 mg) was added, and the solution was stirred
for 2 h. A white precipitate was formed and removed by filtration.
t-Butylglycinate (1 mmol) was then added and the solution was
allowed to stir overnight at room temperature. The solution was then
successively washed with 0.1 N HCl, NaHCO3 (0.02 g/mL), 1 N NaOH,
and 0.1 N HCl. The organic phase was separated and then evaporated
under reduced pressure. The reaction product was dissolved in a
solution of TFA (5 mL) and triethylsilane (1 mL). After 1 h, the TFA/
triethylsilane was removed under reduced pressure at room temper-
ature. Addition of ∼50 mL of diethyl ether yielded a white solid,
2
50 mm column as already described) using appropriate isocratic
solvent systems ranging between 5 and 80% solvent B. In this second
purification step, fractions of interest were collected manually. This
collection was guided by observation of the UV response at 220 nm.
The collected fractions were immediately cooled to 0 °C by placing
them on ice. The solvent was removed by lyophilization. The purities
of these fractions were assessed by analytical rp-HPLC already
described.
Sou r ces of Kn ow n Com p ou n d ssChemical shifts of carbons and
protons of the known compounds described next are tabulated in Table
Journal of Pharmaceutical Sciences / 527
Vol. 86, No. 5, May 1997

described) and analyzed by probe electron-impact mass spectroscopy
EI-MS), fast atom bombardment MS (FAB-MS), and proton nuclear
(
1
magnetic resonance spectroscopy ( H NMR). FAB-MS showed an
MH of m/ z 209; the highest mass ion observed by EI-MS was m/ z
+
1
90 (208-18, ring-closure and dehydration on the heated probe). The
sample showed major fragments at m/ z 147, 132, 118, 106, 104, 91,
9, and 77. For NMR assignments see Table 1.
-Hydroxy-3-phenyl-5-benzylpyrazine (13)sThe preparation of 13
7
2
3
was performed according to the procedure of Barbhaiya. Cephalexin
100 mg, Eli Lilly and Company) in distilled water (100 mL) was
(
treated (10 min, room temperature) with NaOH solution (50 mL, 1
N). Then HCl (50 mL, 1 N) was added, followed by Sorensen’s citrate
buffer (300 mL) containing benzaldehyde (1.0% v/v). The pH of the
buffer was adjusted to 5.0 and then the buffer was heated at 100 °C
for 30 min, cooled, and extracted repeatedly with ethyl acetate. The
combined organic layers were evaporated, and the product was
purified by crystallization from ethyl acetate. For NMR assignments
see Table 1.
∆2
-Cefaclor (6)sCefaclor‚N,N-dimethylformamide disolvate (250 g,
obtained from Eli Lilly and Company) was added to a 5-L 3-necked
round-bottomed flask containing 2500 mL of dimethylformamide and
fitted with a mechanical paddle stirrer. After stirring for 15 min,
Figure 1srp-HPLC chromatogram (UV, 220 nm) of a solution of cefaclor stored
for 2 months at room temperature in 0.1 N HCl. See Experimental Section for
chromatographic details.
2
.5 eq (298 mL) of bis(trimethylsilyl)acetamide (BSA) was added,
followed by addition of 2.5 eq of triethylamine (TEA, 170 mL) and
stirring for 10 min. At this point, the solution was a ∼50:50 mixture
2
of tetramethylsilane (TMS)-protected cefaclor and ∆ -cefaclor. Hy-
drolysis of the TMS-protecting groups and precipitation of cefaclor
was accomplished by rapid addition of 250 mL of 6 N HCl, which
lowers the pH to ∼5.2. After 30 min, the pH dropped to ∼4.1, and
was adjusted to ∼5.7 via dropwise addition of TEA. The temperature
was reduced to 10 °C by submerging the flask in an ice bath, and the
2
cefaclor precipitate was removed by filtration. ∆ -Cefaclor was
crystallized from the solution by adding two volumes of acetone. The
crystalline precipitate was recovered by filtration and washed with
acetone, cold water, and acetone. Solvent removal was accomplished
by placing the product under reduced pressure at room temperature
for ∼72 h. Overall yield was: 19% (47.5 g). HPLC analysis of the
final product revealed <1% contamination by the parent cefaclor.
FAB-MS indicated the protonated molecular ion to be m/ z 368. FAB-
MS/MS measurements of m/ z 368 revealed a fragmentation pattern
2
similar to that of cefaclor. The UV spectrum of ∆ -cefaclor has a
characteristic “double-humped” appearance (λmaxvalues at ∼240 and
13
2
60 nm, Figure 4). The fully assigned 1H and C NMR spectra (see
1
Table 1) were consistent with the proposed structure. In the H NMR
spectrum, the resonances of the protons at sites 6, 7, and 9 had the
characteristic three-spin pattern of â-lactams, and H-2 and H-4 were
Figure 2srp-HPLC chromatogram (UV, 220 nm) of a solution of cefaclor stored
in deionized water for nine months at room temperature. See Experimental Section
for chromatographic details.
2
4
coupled by 1.5 Hz, as is typical of ∆ -cephalosporins.
2
∆ -Cefaclor dihydrate was crystallized by evaporation from an
acetonitrile/water mixture in the monoclinic space group C2 (No. 5)
with unit cell parameters and calculated volume of a ) 26.939 (5), b
)
7.551 (1), c ) 9.704 (2), â ) 109.23 (2)°, and V ) 1863.9. The
-
3
calculated density is 1.44 g‚cm for Z ) 4 and a formula weight of
4
03.84 The data were collected on a Enraf-Nonius CAD4 diffracto-
meter using CuKR radiation and a graphite monochromator. A total
of 2099 reflections were collected, of which 2053 were unique. The
5
structure was solved by direct methods using the SIR92 program.
The remaining atoms were located in succeeding difference Fourier
6
syntheses. SHELX-93 was used for structure refinement. The
-
3
highest peak in the final difference Fourier map was 0.71 e‚A
.
Anisotropic temperature factors were assigned to all nonhydrogen
atoms, but the hydrogens were kept isotropic. Full-matrix least-
squares refinement of all nonhydrogen coordinates and thermal
parameters gave final factors R ) 5.2% and Rw ) 14.3%. Complete
tables of the fractional coordinates, bond angles, and bond lengths
are provided in the supplement.
Mea su r em en t of Sp ectr a sUV SpectrasUV spectra were ac-
quired by the HPLC photodiode array-UV system (already described)
as the peaks eluted off the analytical column.
IR SpectrasFT-IR spectra of the samples were obtained with a
SpectraTech IR PlanII FT-IR microscope in conjunction with a Nicolet
model 60 SXB FT-IR spectrometer. Typically, several particles of the
sample totaling <50 µg were placed on a 13-mm diameter potassium
bromide disk. The particles were then flattened with a microspatula
to provide a sample of suitable thickness.
Figure 3srp-HPLC chromatogram (UV, 220 nm) of a solution of cefaclor stored
at room temperature for 2 months at a pH maintained between 4.8 and 6.0. See
Experimental Section for chromatographic details.
which was collected by filtration and washed with diethyl ether. The
total weight of recovered product was 120 mg (0.37 mmol) for a 37%
overall yield. The product was purified by rp-HPLC (as already
Mass SpectrasFAB-MS and FAB-MS/MS data were obtained with
a VG ZAB-3 (B1EB2) triple-sector mass spectrometer or a VG ZAB-
5
28 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

Table 1s13C and 1H NMR Chemical Shifts for Isolated Degradation Productsa
site:
4
8b
3
14
5
10c
11
12
13d
9e,f
7e
n.o.
3.71,3.56
n.o.
2.96
n.o.
2.96
n.o.
n.o.
8.57
n.o.
6
2g
2
3
4
n.o.
25.50
2.96
25.50
.96
n.o.
.96
118.23
6.75
113.83
2.96
54.02
4.87
167.28
52.64
5.18
59.94
5.62
163.37
9.64
28.15k
3.48
5
.11
j
n.o.
147.12
2
2.96
h
n.o.
142.72
2
4.87
i
4
-COOH
n.o.
n.o.
163.22
6
126.80
7.40
122.20
7.26
137.73
3.82
31.04
3.68
n.o.
122.60
7.43
140.43
7.74
122.75
7.18
139.07
7.74
164.24
5.18
8.85
7
123.09
44.24
41.06
3.82
170.56
8.85
n.o.
7.74
n.o.
40.65
4.54
7.47
3.94,3.74
3.82
4.57
n.o.
h
8
9
0
1
165.85
10.77
174.24
61.23
8.13
n.o.
n.o.
167.86
55.73
5.01
h
1
1
155.28
151.57
5.16
166.89
167.79
55.47
5.05
155.57
147.82
5.05
154.12
129.00
5.05
n.o.
n.o.
5.05
155.85
148.30
5.05
155.85
148.13
5.05
n.o.
168.09
55.50
5.06
58.60
4.87
8.61
138.86
126.85
7.36
128.58
7.36
n.o.
5.16
∼
4.9
1
1-NH
8.40
136.11
126.74
9.5
n.o.
n.o.
n.o.
8.84
n.o.
h
i
1
2
136.06
128.24
8.31
127.76
7.44
129.36
7.44
134.1
128.12
7.54
136.28
127.82
8.21
128.03
7.34
129.00
7.34
135.87
127.82
8.25
128.31
7.43
129.20
7.43
n.o.
n.o.
8.22
n.o.
7.43
n.o.
7.43
136.08
128.16
8.30
127.82
7.43
129.13
7.40
136.17
127.73
8.29
128.18
7.38
128.96
7.29
133.53
127.94
7.54
133.62
128.17
7.54
1
3,17
7
.38
128.68
.38
128.29
.38
7.4
−
7.25
7.25
7.25
i
14,16
128.82
7.42
129.24
7.45
n.o.
128.88
7.44
129.44
7.44
129.45
7.44
128.92
7.44
7
7.4−
j
1
5
127.90
7.36
n.o.
7
7.4−
a
Each cell of the table contains the carbon (top line) and proton (bottom line) chemical shift(s); unless otherwise noted, all spectra were measured in DMSO-d
6
/TFA;
b
compounds are numbered to reflect their most probable mechanistic origin fom cefaclor, which is numbered as shown in 1; n.o., not observed. Measured in
perdeuteromethanol.
9.72/2.64; 7
41.60/- ; 3
c
d
e
f
7
′
: 15.84/2.14; 8
′
: n.o. CH
2
: 35.60/3.88; ipso:137.36/-; ortho:128.84/7.38; meta:128.63/7.35; para: 126.84/7.26. Measured in DMSO-d
6
. 6
′
:
:
h
i
i
j
g
2
1
′
:
37.26/3.43; 9
146.69/- ; 4
′
:
n.o.; 10
′
:
170.92; 11
-COOH: 163.01/-; 6
′
:
57.58/4.56; 11
′
-NH
3
:
n.o.; 12
′
:
138.81 ; 13
′
,17′
:
127.06/7.4; 14
′
,16′
:
128.01/7.4; 15
′
:
127.66/7.4.
2′
h
j
k
i
h−k
′
:
′
:
28.58/3.48 ; 4
′
′
:
151.22/8.91.
Assignments within a single column may be interchanged.
Figure 4sUV spectra of cefaclor degradation products obtained on the photodiode array-UV detector as the compounds eluted from the column.
2
“
SE (BE) two-sector mass spectrometer. Samples were dissolved in
Magic Bullet” (a 5:1 solution of dithiothreitol:dithioerythritol in
methanol) and bombarded with 8 KeV xenon atoms in the ZAB-3 or
Cs ions, with a net energy of 12 KeV, in the ZAB-2SE. For MS/MS
experiments with the ZAB-3, the precursor ions were selected with
B1 and collisionally activated (50% attenuation using He collision gas)
Journal of Pharmaceutical Sciences / 529
Vol. 86, No. 5, May 1997

in the second field-free region. The products were separated by a
constant B2/E linked-scan. For MS/MS experiments using the ZAB-
2
SE, the ions leaving the ion source were collisionally activated with
helium in the first field-free region and the products were separated
with a constant B/E linked-scan.
NMR Spectras H and 13C NMR spectra were recorded on Bruker
AC-250 or AMX-500 spectrometers. Spectra were recorded in DMSO-
d6 and referenced to internal TMS on the bases of the chemical shifts
of residual solvent peaks. In some cases the peaks of the initial
spectrum recorded were broad, and in these cases the addition of a
trace of TFA sharpened the spectra significantly, presumably due to
the conversion of a mixture of protonation states to the TFA salt. The
use of this solvent system also permitted the observation of the
protons at exchangeable sites, such as ammonium ions. The multi-
plicities of the carbon resonances were determined by the DEPT
1
7
method. Carbon-proton correlations were detected by two-dimen-
sional heteronuclear experiments designed to detect correlations due
8
9
to coupling through one or more than one chemical bond. Chemical
shift data for all compounds for which NMR spectra were measured
1
appear in Table 1. H NMR spectra were modeled using the Bruker
program PANIC.
Der iva tiza tion ssDerivatizations with diazomethane were ac-
1
complished as previously described.
Results
Elu cid a tion of Str u ctu r essBecause all the compounds
isolated from these degradations are related structurally,
some generalities can be stated about their properties and
spectra. For example, many of the compounds isolated in this
research include the phenylglycyl substructure. This sub-
structure was found to be associated with two fragments in
the FAB-MS/MS spectrum:
Figure 5sIllustration of the structural possibilities as determined with the indicated
substructural units and the computer program GENOA. From the indicated
substructures, GENOA constructed 10 possible structures. The assumption that
the molecule retains the 4-carboxylthiazol-5-yl substructure reduced this list to
three candidates.
generally group together at about δ 7.45. In the mass spectra,
this substructure is often found to lose a benzonitrile frag-
ment.11
Several of the degradation products included thiazole rings,
which had characteristic UV absorbances near 245 nm (see
2
, 7, 8, and 15, Figure 4).12 Other spectroscopic properties
associated with thiazoles are three carbon chemical shifts, one
near δ 164 (C6) and two between δ 140 and 145 (C2 and C3).
Thiazoles with a proton on the carbon between the two
The first fragment results from benzylic cleavage to give a
fragment at m/ z 106. The second fragment results from
cleavage of the amide bond with loss of the phenylglycyl group
from the protonated molecular ion to yield a fragment m/ z
M - 132. In the UV spectrum, the phenylglycyl group
contributes primarily to end absorbance, but also includes a
weak phenyl chromophore at ∼260 nm (see UV spectra of 4,
heteroatoms, as occurs in the case of 2, are characterized by
1
low field methine resonances in both the 13C and
1
H NMR
spectrum (∼δ 151 and 8.7, respectively).
As we elucidated these structures, we attempted to ratio-
nalize the formation of each with plausible mechanistic
degradation pathways. These pathways were helpful in the
proposal of the structures of new compounds as they were
isolated and characterized spectroscopically, or to propose final
details of the molecular structure of the degradation products
isolated previously. To allow the direct comparison of NMR
chemical shifts throughout the series of isolated products
(Table 1), all compounds are numbered in a way that reflects
their proposed mechanistic origin from cefaclor, which is
numbered as shown in 1.
5
, and 14, Figure 4). In the NMR spectra, the phenylglycyl
group contributes peaks due to the phenyl ring and to a
methine with carbon and proton shifts at approximately δ 55.5
and 5.05, respectively (see Table 1). The last resonance is
+
typically coupled to the three-proton resonance of the NH
3
1
group near δ 8.7 when the H NMR spectrum is measured in
DMSO with added TFA (see Experimental Section). Correla-
tion of the carbon and proton resonances through their two-
Th ia zole-Con ta in in g P r od u ctssThe UV spectra of four
of the aqueous degradation products were similar to the UV
and three-bond couplings9
,10
can be used to extend this
substructure further. For example, the amide carbonyl
spectrum of 2 (see Figure 4) and other thiazole-containing
1
resonance, which can be identified by its correlations to H-11
structures.12 This similarity was interpreted to indicate of
+
and the protons of the NH
3
group, is also generally correlated
the presence of the thiazole substructure.
to the amide NH and the proton(s) on carbon 7. This permits
the definition of connectivity between the phenylglycine
moiety and the remainder of the molecule.
Another common type of degradation product encountered
in this research contained the 2-hydroxy-3-phenylpyrazine
substructure.2 The UV spectra of these compounds are quite
distinctive, showing significant absorbances in the regions
Compound 7sAs shown in Figure 1, 7 is a relatively minor
component of the degraded solution, and it has a UV spectrum
very similar to that of 2 (λmax ∼243 nm). The amount of
compound isolated was only sufficient to permit measurement
of the 1H NMR and DEPT7 spectra. These NMR spectra
(Table 1) indicated the presence of a phenyl group, thus
suggesting the presence of a phenylglycyl substructure. This
conclusion was confirmed by the presence of an m/ z 106 peak
in the mass spectrum. The presence of a sharp singlet at δ
2
40-260 and 330-370 nm (see UV spectra of 3, 11, and 13,
1
Figure 4). These compounds are also characterized by
H
1
NMR spectra including a two-proton multiplet near δ 8.3,
assigned to the ortho protons of the phenyl ring. The other
aromatic protons, including any on the pyrazine ring, all
8.57 in H NMR spectrum implied the presence of a thiazole
with a methine group between the two heteroatoms, as in the
1
1
case of one of the thiazole rings of 2. The H NMR spectrum
5
30 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

Scheme 1sGeneralized acidic degradation pathways of cefaclor.
also showed the presence of a methine (δ 4.57) attached to a
methylene (δ 3.71, 3.56). Finally, the methylation experi-
ments discussed later indicated the presence of two carboxylic
acids. Hence, the spectroscopic data indicated the partial
structures shown at the top of Figure 5.
ment at m/ z 106 persists in the dimethyl derivative, whereas
m/ z 217 increases 28 amu to m/ z 245 upon methylation with
diazomethane, indicating the presence of two carboxylic acid
groups in this fragment.
Compound 8sThe mechanistic origins of 2 and 7 (see
Scheme 2) suggested that an intermediate thiazole structure,
such as 8, might be formed during the acidic degradation of
cefaclor. Examination of the UV spectra of the degradation
products (pH 5.5 conditions) obtained from HPLC analysis
revealed a minor degradation product eluting at ∼9.7 min
with a spectrum consistent with a thiazole substructure (λmax
These substructures account for all the atoms in the
molecular formula of the structure, and were used in conjunc-
1
3
tion with GENOA to obtain the results shown in part in
Figure 5. Initial structure generation led to 10 candidate
1
structures. Assuming, based on mechanistic arguments, that
one of the carboxylic acid groups would be attached adjacent
to the nitrogen of the thiazole ring, the list of candidate
structures is reduced to 7, 7a , and 7b, as shown in Figure 5.
Considering the initial connectivity of cefaclor and plausible
degradation pathways, structures 7a and 7b are not feasible.
On this basis, and in conjunction with the MS/MS results
described later, the structure was assigned as 7. Assignments
for several of the key fragments from the FAB-MS/MS of the
protonated molecular ions of 7 and its dimethyl ester are
shown below (dimethyl ester fragments denoted by *):
2
43 nm; see Figure 4). This product was isolated, and analysis
+
5 6 3
by FAB-MS revealed a molecular formula of C H NO S (MH ,
found, m/ z 160.0081; calc, 160.0068). This molecular formula
is consistent with thiazole structure 8. Derivatization with
+
diazomethane increased the MH to m/ z 174 (+14 amu),
indicating the presence of one carboxyl group. The largest
peak in the FAB-MS/MS of the parent compound occurred at
m/ z 142, implying facile loss of water, as might be expected
via lactonization. The 1H NMR spectrum acquired in deu-
terated methanol showed singlets at δ 8.85 (1H) and δ 5.11
(
2H), which are reasonable chemical shifts for the protons of
1
2
the proposed structure.
Therefore, the structure of this
compound is proposed to be 8.
Compound 15sThe fourth thiazole-containing degradation
product had a UV λmax of 245 nm (see Figure 4). This product
was shown by mass spectrometry to have a molecular formula
+
of C18
H
19
N
3
O
6
S (found for MH , m/ z 406.1065; calc, 406.1072)
and a phenylglycyl moiety (m/ z 106). From the FAB-MS/MS
experiment, other fragments were detected at m/ z 319, 305,
and 273. The fragment at m/ z 305 is analogous to a fragment
in the mass spectrum of another thiazole-containing degrada-
tion product, 2.1 The fragment at m/ z 319 suggests an
The presence of the m/ z 106 and 217 (M - 132) fragments
confirms the presence of the phenylglycyl moiety. The frag-
Journal of Pharmaceutical Sciences / 531
Vol. 86, No. 5, May 1997

5
32 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

additional methylene attached to the m/ z 305 fragment, and
together with the other fragments implies the following partial
structure:
presence of a methylene group. On the basis of accurate mass
FAB/MS, the molecular formula of this compound was shown
+
to be C12
H
10
N
2
O
3
(MH : found, 231.0766; calc 231.0770). The
FAB-MS/MS spectrum showed significant fragments at m/ z
85, 157, 104, and 77; the fragment at m/ z 185 is consistent
1
with the attachment of a methylene group to a 2-hydroxy-3-
phenyl pyrazine ring. Thus, structure 10 was proposed. The
fragmentation pathway represented next is consistent with
the proposed structure.
Prior to further spectroscopic characterization, this com-
pound degraded, thus precluding further characterization.
However, upon evaluation of the various proposed degradation
pathways, the structure of the C
see especially Schemes 2 and 5). On this basis, we propose
the structure to be 15. Further work would be needed for
3 3 3
H O group can be proposed
(
Compound 11sThis compound, isolated from the degrada-
tion in unbuffered water, had a long HPLC retention time
(24.7 min), indicating that it was nonpolar relative to most of
the other degradation products. The UV spectrum (Figure
4) contained λmax values at 366 and 256 nm. Thus, while the
UV spectrum of this product is similar to that of 2-HPP, there
is a bathochromic shift of the λmax from the 346 nm absorbance
of 2-HPP. Accurate mass EI-MS measurements indicated a
molecular formula of C H N OS (found 244.0637, calc
conclusive structure elucidation.
“
F lu or escen t Com p ou n d s”sIt is well established that
â-lactam antibiotics containing the phenyl glycine side chain
degrade under certain conditions to yield a highly fluorescent
product, 2-hydroxy-3-phenyl pyrazine (2-HPP, structure 3).14
In our studies of the acidic degradation of cefaclor, numerous
fluorescent products were detected by HPLC with fluorescence
detection. Examination of the UV spectra of these products
using photodiode array UV detection revealed numerous small
peaks (see Figure 1) with spectra similar or identical to that
of 3. Six of these fluorescent degradation products were
isolated and characterized. Two of these (3 and 9) were
reported previously.14 The mechanistic origin of all of these
fluorescent compounds (except for 11) can be rationalized and
generalized on the basis of the pathway shown in Scheme 3.
This mechanism permits assignment of the site of substitution
on the pyrazine ring, which in some cases is not possible on
the available spectroscopic evidence.
1
3
12
2
244.0670). The FAB-MS/MS spectrum (and the EI mass
spectrum) showed fragmentation similar to those of the other
2-HPP derivatives.
These results suggest that the structure of this compound
includes a 2-HPP group, but with substitution of the pyrazine
ring differing from that seen in the other analogs. This
1
conclusion was confirmed by the H NMR spectrum. Whereas
the qualitative pattern of aromatic resonances typical of these
compounds was present, integration showed that only a total
of five aromatic protons were present. 13C NMR spectroscopy
confirms the presence of these five aromatic protons and that
2
Compound 9sAccurate mass FAB/MS of 9 (HPLC retention
they are in the form of the five sp -hybridized methines of a
+
time 16.8 min) revealed an MH of m/ z 349.1664, which is
phenyl group. Thus, all aromatic protons are accounted for
by the one phenyl group, so that the pyrazine ring must be
fully substituted. In addition, the NMR spectra included
resonances of three methylene groups, one of which appeared
consistent with a molecular formula of C20
49.1665). The UV spectrum of this compound (λmax values
of 356 and 253 nm, see Figure 4) is very similar to that of
-HPP, implying a structural relation between the two
21 4 2
H N O (calc
3
1
2
as a singlet (δ 3.75) in the H NMR spectrum and the other
compounds. The presence of a 2-HPP substructure was
two of which were mutually coupled (δ 3.12 and 2.99). These
data account for a total of 11 of the 12 protons of the molecule.
The last proton appeared as a broad singlet below δ 12,
attributable to the aromatic hydroxyl proton.
1
supported by the H NMR spectrum (Table 1).
The 1H NMR spectra indicated that the compound con-
tained two phenyl groups, but that only one of the two phenyl
groups was attached to pyrazine. The mass and NMR spectra
clearly showed the presence of a phenylglycyl moiety in the
molecule, accounting for the second phenyl group. An amide
NH, presumed to be part of the phenylglycyl moiety, was
From the aforementioned data, the three substructures in
Figure 6 can be deduced. These substructures account for all
but the sulfur atom of the molecular formula. As shown by
GENOA, only two structures (11 and 11a ) can result from
assembly of these groups. On the basis of mechanistic
considerations (see Scheme 4), we propose the structure to
be 11.
1
shown by the H NMR spectra to be attached to a methylene
of an ethylene substructure. These proposals were supported
by fragmentations in the mass spectrum of the compound.
The structural information just derived, coupled with
GENOA, leads to two candidate structures which differ only
by the site of attachment of the methylene group to the
pyrazine ring (carbon 6 or 7). On the basis of these results
and mechanistic arguments (see Scheme 3) the structure of
this compound is proposed to be 9. This compound has been
previously isolated from degradation of ampicillin,14 and our
NMR data are in full accord with those published.
Compound 12sCompound 12 has a UV spectrum similar
to that of 2-HPP (λmax values of 353 and 247 nm). Accurate
mass FAB-MS measurements indicated a molecular formula
of C H N O (found for MH+, m/ z 257.0923; calc 257.0926).
1
4
12
2
3
In the 1H NMR spectrum there appeared the pattern of
1
resonances typical of 2-HPP derivatives. The H NMR
spectrum contains a five-proton spin system typical of the
substructure -CH -CH -CH<. From the chemical shifts of the
2
2
Compound 10sThe UV of 10 is similar to that of 2-HPP
protons of the methine (δ 4.12) and the terminal methylene
(δ 4.47, 4.35) of this substructure, it is apparent that these
carbons bear electronegative substituents. Also present in the
(λmax values of 354 and 246 nm, respectively, see Figure 4).
The presence of the 2-HPP substructure was confirmed by the
1
1
H NMR spectrum, which included the relevant pattern of
H NMR spectrum is a broad singlet at δ 7.71. At 80 °C, this
resonances in the aromatic region of the spectrum (vide
supra). The only other peak in the NMR spectrum was a
singlet at δ 3.6 integrating for two protons, indicating the
resonance is sharpened, but not changed in chemical shift.
Furthermore, addition of H
2
2
O to the sample showed that this
resonance was not exchangeable.
Journal of Pharmaceutical Sciences / 533
Vol. 86, No. 5, May 1997

Scheme 3sPathway B: formation of pyrazine derivatives.
The mass spectrum of the compound includes a significant
M-45 fragment, which suggests the presence of a carboxylic
acid or a lactone. The compound was shown to react with
diazomethane slowly with the uptake of ∼1 mol of reagent, a
behavior that we have found to be common with hydroxy-
pyrazines. The sluggish reaction of the compound with
diazomethane and the fact that the same M-45 loss is seen
in the mass spectrum of the methylated derivative indicate
that the compound does not contain a free carboxylic acid. This
degradation product was therefore proposed to include a
lactone substructure.
These substructural constraints, when input into GENOA,
5
34 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

Miscella n eou s Str u ctu r essThe structures of six of the
products isolated in this study were established by comparison
with authentic compounds or by reference to previously
published data (five of these products are subsequently
mentioned and the sixth, 13, has already been described).
Tabulated NMR data for these compounds are presented in
Table 1, and UV spectra for 4, 5, and 14 are shown in Figure
4
.
SulfursElemental sulfur that precipitated from the reac-
tion mixtures was identified by mass spectrometry as de-
scribed previously.1
Compound 4sCompound 4 was identified as phenylglycine
on the basis of its molecular formula, as established by mass
spectrometry. Subsequent chromatographic and spectroscopic
comparison to an authentic sample (see Experimental Section)
confirmed this structure assignment.
Compound 5sThe structure of 5 (phenylglycylglycine) was
proposed on the basis of UV, MS, NMR, and chromatographic
retention time by comparison with an authentic sample of the
compound. The authentic sample was prepared as described
in the Experimental Section.
Figure 6sIllustration of the structural possibilities (11 and 11a) as determined
with the indicated substructural units (derived from spectroscopic characterization)
and the computer program GENOA.
result in the construction of only two plausible structures, 12
and 12a . Of these two candidates, 12 appears to be more
1
consistent with the empirical H NMR chemical shifts. The
Compound 14sThe structure of 14 (3-phenyl-2,5-diketopip-
predicted chemical shifts of the methylene protons at sites 2
erazine) was proposed on the basis of its UV and MS spectra,
1
5
and 4 in structure 12 are ∼δ 4, in good agreement with the
empirical data. In 12a , however, the chemical shift of H-2 is
predicted to be at significantly higher field (∼δ 2.5), as in the
case of the analogous site of xanthoquinidin A3.1 The broad
singlet at δ 7.71 can be assigned to H-6 on the pyrazine ring;
this singlet is broadened by hindered rotation about the C2-
C7 bond, which is possibly accentuated through internal
H-bonding from the (protonated) pyrazine nitrogen to the
carbonyl group of the lactone. Thus, we propose the structure
of the degradation product to be 12. The electron-impact mass
spectral (EI-MS) fragmentation assignments of both the
degradation product (RdH) and its monomethyl derivative
1
and confirmed by comparison of the H NMR spectrum to that
18
described previously.
Compound 14 is one of the major
degradation products at pH 5.5 and has an HPLC retention
6
time of ∼13.4 min.
Compounds 16 and 17sOne of the earliest eluting prepara-
tive HPLC fractions contained a total of four compounds, two
of which were identified as 4 and 5 on the basis of MS. Two
other minor compounds coeluted with 4 and 5 and, because
we were unable to separate these minor compounds from the
matrix, their structures (16 and 17) have been proposed based
on MS.
The FAB-MS spectrum of this fraction showed what ap-
3
(RdCH *) are represented as follows:
+
peared to be MH ions at m/ z 152, 168, 209, 225, and 458.
An MS/MS of m/ z 458 showed that this species was an adduct
ion due to the interaction of the m/ z 152 species with the
dispersant used and this ion was not investigated further.
Accurate mass determinations on the other peaks in the FAB-
MS yielded the elemental compositions shown in Table 2. The
peaks at m/ z 152 and 209 can be assigned to the protonated
molecular ions of phenylglycine (4) and phenylglycylglycine
(5), respectively. Those at m/ z 168 and 225 appear to be the
“oxides” of these two compounds.
When the FAB-MS spectrum of this isolate was redeter-
+
mined using deuterated dispersant, the MH ions were shifted
The elemental compositions of all fragment ions shown have
to the masses shown in column 6 of Table 2. These data show
that oxidation of phenylglycine and phenylglycylglycine in-
troduces one additional exchangeable proton in each case,
thereby ruling out the possibility that the amine functions of
these compounds have been oxidized to hydroxylamines.
Additionally, it should be noted that the m/ z 168 and 225
ions lose the elements of ammonia (confirmed by accurate
mass measurements) to form m/ z 151 and 208, respectively,
a result that would require the presence of a simple primary
amine in these molecules.
been confirmed by accurate mass measurements. Finally, this
structure is consistent with the mechanistic proposals de-
scribed next (see Scheme 3).
Compound 13sThis relatively nonpolar product (∼25 min
HPLC retention time) was isolated from the degradation in
water (unbuffered). Compound 13 was postulated to be a
derivative of 3 by its UV spectrum (λmax values of 358 and
2
54 nm, see Figure 4). From accurate mass FAB-MS mea-
+
surements, the molecular formula for MH was determined
to be C17 O (found m/ z 263.1182; calc 263.1184). The
15 2
H N
NMR spectra confirmed that the compound was a 2-HPP
derivative and showed that the structure included a second
phenyl group and a methylene. Because these substructures
account for all the atoms in the molecular formula, the
compound can be represented only as a benzyl-2-hydroxy-3-
phenylpyrazine. Of the two sites of attachment for the benzyl
group (carbons 6 or 7), carbon 7 is selected as the attachment
site on the basis of mechanistic arguments as outlined in
Scheme 3. This conclusion was confirmed by spectroscopic
and chromatographic comparison to a standard.17
The chemically-induced dissociation (CID) mass spectrum
of the m/ z 168 ion taken at 75% primary beam attenuation
(to increase the relative intensity of lower mass fragments)
clearly showed a fragment at m/z 77, indicating that the
phenyl ring was not oxidized. A peak at m/ z 105 in the same
+
spectrum suggests that PhsCtO is a fragment of m/ z 168
and that oxidation occurred on the methine carbon bearing
the amino group. The MS results indicate that the structures
of the species responsible for the ions at m/ z 168 and 225 are
16 and 17, respectively.
Journal of Pharmaceutical Sciences / 535
Vol. 86, No. 5, May 1997

Scheme 4sFormation of pyrazine derivatives 11.
Table 2
s
Accurate Mass FAB-MS Measurements on Fraction Containing
lecular attack of the 11′ primary amine on the â-lactam
carbonyl (C8). In a more recent study of the neutral aqueous
degradation of cefaclor, Vilanova et al.24 provided structural
evidence for this intramolecular cyclization pathway. Under
acidic conditions, however, the nitrogen would be protonated,
significantly reducing nucleophilicity. Thus, it is not surpris-
ing that this intramolecular cyclization reaction does not
appear to be the dominant pathway under acidic conditions.
Of all the products isolated in the present study of the acidic
4
, 5, 16, and 17
Found
Mass
Calc.
Mass
Deut. No.
a
Compound
Ion
Composition
Disp
Exch
4
m/z 152 152.0720
168 168.0670
209 209.0947
225 225.0889
208 208.0655
151 151.0400
C
C
C
C
C
C
8
H
10NO
10NO
2
3
152.0712
168.0661
209.0926
225.0610
208.0610
151.0396
156
173
214
231
3
4
4
5
1
5
1
6
8
H
10
10
10
H
H
H
7
13
N
N
2
O
O
3
4
7
13
2
Fragment of 17
Fragment of 16
10NO
4
2
8
H
O
3
degradation of cefaclor, only ∆ -cefaclor (6) retains the â-lac-
tam ring. We therefore propose that acid-catalyzed hydrolysis
of this strained ring is, for most of the degradation pathways,
the first step in the degradation of cefaclor (see Scheme 1).
The various degradation products can be loosely classified into
three categories: thiazole derivatives, pyrazine derivatives,
and simple hydrolysis or rearrangement products.
aDeuterated dispersant: deuterated “Magic Bullet” (dithiothreitol:dithioerythritol
in MeOH).
The isolation of R-hydroxy amino acids such as 16 and 17
seems remarkable considering the expected lability; the loss
of ammonia might be expected to proceed rapidly, leading to
the resulting ketones. The MS results, however, provide
strong evidence supporting these structures. It is noteworthy
that R-hydroxy amino acids have been described previously.1
For example, the 2-hydroxy analog of methionine (also known
as methionine hydroxy analogue) has been studied extensively
Thiazoless8, 7, 15, and 2sThe predominant degradation
product contains a thiazole ring, and a mechanism rational-
9
1
izing the origin of this compound has already been reported.
The isolation of three additional thiazole-containing products
allows us to generalize this mechanism (see Schemes 1 and
2). The critical intermediate for the pathways leading to
thiazole derivatives appears to be (d), which is proposed to
form from episulfonium ion intermediate (c) by loss of a proton
2
0
for use as a methionine substitute in animal feeds.
We
therefore propose that structures 16 and 17 are minor
degradation products indicating the presence of a degradation
pathway involving benzylic oxidation. This benzylic oxidation
pathway is an important part of the mechanistic proposals in
both aqueous conditions (see Scheme 5) and in the solid
state.21
(Scheme 1). Intermediate (d) can continue the degradation
processes in at least two ways. Pathway a (Scheme 2) leads
to cleavage of the molecule into two moieties that may reorient
within a solvent cage, permitting recombination to yield 7.
Escape from the solvent cage leads to the formation of 5 and
8
. Alternatively, intermediate (d) may react as a nucleophile
Discussion
by delivery of the lone pair of electrons on the sulfur atom to
the exocyclic methylene group (see Pathway b, Scheme 2). In
principle, this could lead to reaction with any electrophile in
2
X-r a y Cr ysta l Str u ctu r e of ∆ -Cefa clor sThe ORTEP
2
diagram of ∆ -cefaclor is shown in Figure 7. The absolute
configuration of the C4 stereocenter was determined to be S
the solution. One such electrophile is 8 or its lactonized form
by relation to the known stereochemistry of the cefaclor
molecule. Few single crystal studies have been performed on
-isomers of cephalosporin antibiotics. To our knowledge,
this is the first instance in which the crystal structures of both
the active cephalosporin (i.e., cefaclor ) and its respective ∆ -
isomer can be directly compared. In our degradation studies
of cefaclor (both aqueous and solid state), we observed the
1
(
f) to yield a precursor of compound 2. Another electrophile
that we propose to be present is (h), the reaction of which with
intermediate (d) by pathway b (Scheme 2) yields the structure
proposed for 15. This mechanism can therefore rationalize
the origins of all four of the thiazole-containing products of
these degradations, as well as that of 5.
2
∆
2
2
2
2
Pyrazine Derivativess3, 10, 13, 11, 12, and 9sThe mecha-
formation of only the 4S isomer of ∆ -cefaclor. Further
nistic origin of the fluorescent pyrazine degradation products
from cephalosporins with a phenylglycyl side chain has
already been proposed.1
generalized to explain the formations of some minor fluores-
cent products isolated in this work. The critical intermediate
research would be required to determine whether this selec-
tivity results from kinetic or thermodynamic considerations.
P r op osed Degr a d a tion P a th w a yssPrevious studies on
4,25
In Scheme 3, this mechanism is
2
3
the degradation of cefaclor by Nakashima et al. suggested
that at neutral pH, cefaclor degrades primarily via intramo-
5
36 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

Scheme 5sFormation of reactive “fragments”.
is phenylglycyl-acetaldehyde (i), which can be thought of as
forming from hydrolysis of (e) as shown in Scheme 3. Differ-
ent pathways to (i) can be proposed, but any pathway of
hydrolysis that reveals the “masked aldehyde” at C6 would
suffice. This intermediate can either undergo internal cy-
clization and oxidation leading to 3, or it can undergo an aldol
condensation with any available electrophiles in the solution.
Cyclization, dehydration, and tautomerization lead to the
pyrazine derivatives, which differ only in their substitution
at C7. Alternatively, the condensation with the electrophilic
species could occur after internal cyclization.25 As shown in
Scheme 3, the internal cyclization product can tautomerize
to form (j), which can react with electrophilic species as shown
to form adduct (k); subsequent dehydration and tautomeriza-
tion leads to the substituted pyrazine derivatives. We have
identified five of these substituted pyrazine derivatives (9, 10,
An important part of the generalized pathway to pyrazine
derivatives is the formation of electrophilic degradation
products shown in Schemes 3 and 5. Two products we have
isolated (pyrazine 12 and thiazole 15) appear to be derived
from intermediate (h). A structure similar to intermediate
(h), structure 18, was previously isolated as a major acidic
degradation product of cephalexin.2 In the case of cefaclor,
the formation of intermediate (h) is proposed to result from
episulfide (g) (Scheme 5) via a reductive nucleophilic attack
on thiol (o). The identity of this nucleophile can be proposed
by remembering that elemental sulfur is isolated from these
degradations. Thus, if the nucleophile is hydrogen sulfide,
the presence of which in these degradations is apparent by
its odor, a process is started that can lead to elemental sulfur.
a
Finally, the formation of 10, 13, and 15 by the mechanism
proposed in Schemes 2 and 3 requires the presence of
“electrophilic fragments”, such as glyoxylic acid (l) or phe-
nylglyoxylic acid (m), in the degradation mixtures. The
formation of these proposed intermediates is shown in Scheme
5. Phenylglyoxylic acid (m) could form via benzylic oxidation
of phenylglycine, or through oxidation of the primary amine
of phenylglycine to form hydroxylamine (n), followed by
dehydration and hydrolysis. The proposed formation of these
“electrophilic fragments”, although not supported by direct
1
1, 12, and 13), the formation of four of which can be explained
with this generalized degradation pathway (9, 10, 12, and 13).
The formation of 11 involves a novel degradation pathway that
does not invoke the formation of a C6 aldehyde (see Scheme
4
). An analogous pyrazine derivative, 19, has been identified
as a degradation product of loracarbef, the 1-carba-analog of
cefaclor.26 The proposed degradation pathway to 11 (Scheme
4
) is analogous to that proposed for 19.
Journal of Pharmaceutical Sciences / 537
Vol. 86, No. 5, May 1997

tion pathways. The major solid-state degradation pathways
described by Dorman et al.21 lead to multiple degradation
products, many of which are not formed under aqueous
conditions. Together, the aqueous and solid-state degradation
investigations reveal some of the diverse degradation chem-
istry of the the cefaclor molecule. Degradation pathways can
now be compared with that of loracarbef, the 1-carba analog
of cefaclor.2
6,27
Evaluation of these aqueous and solid-state
degradation pathways reveals the prominent role of the sulfur
atom in the degradation of cefaclor.
References and Notes
1
. Baertschi, S. W.; Dorman, D. E.; Occolowitz, J . L.; Spangle, L.
A.; Collins, M. W.; Wildfeuer, M. E.; Lorenz, L. J . J . Pharm.
Sci. 1993, 82, 622-626.
2
3
4
. (a) Dinner, A. J . Med. Chem. 1977, 20, 963; (b) Tai, J ., Eli Lilly
and Company, unpublished results.
. Barbhaiya, R. H.; Brown, R. C.; Payling, D. W.; Turner, P. J .
Pharm. Pharmacol. 1978, 30, 224-227.
. Demarco, P. V.; Nagarajan, R. In Cephalosporins and Penicil-
lins: Chemistry and Biology; Flynn, E. H., Ed., Academic: New
York; 1972; Chapter 8.
5
6
. Altomare, A.; Cascarano, G.; Giacovazzo, C.; Burla, M. C.;
Polidori, G.; Camalli, M. J . Appl. Crystallogr. 1994, 27, 435.
. Sheldrick, G. M. “SHELXS93: A Program for Crystal Structures
Refinement,” University of Gottingen, Gottingen, Germany,
1
993.
. Doddrell, D. M.; Pegg, D. T.; Bendall, M. R. J . Magn. Reson.
982, 48, 323.
7
8
1
. Derome, A. E. in Modern NMR Techniques for Chemistry
Research, Baldwin, J . E., Ed.; Organic Chemistry Series, Per-
gamon: New York, 1987; Chapter 9.
Figure 7s
ORTEP diagram of ∆2-cefaclor. Ellipsoids are drawn at the 50%
probability level.
9
. Martin, G. E.; Zektzer, A. S. Magn. Reson. Chem. 1988, 26, 640-
6
41.
1
1
0. Marshall, J . L. In Carbon-Carbon and Carbon-Proton NMR
Couplings: Applications to Organic Stereochemistry and Con-
formational Analysis; Marchand, A. P., Ed.; Methods in Stere-
ochemical Analysis; Verlag Chemie International: Deerfield
Beach, FL, 1983; Vol. 2, Chapter 2.
1. Porter, Q. N.; Baldas, J . In Mass Spectroscopy of Heterocyclic
Compounds; Weissberger, A.; Taylor, E. C., Eds.; General
Heterocyclic Chemistry; Wiley-Interscience: New York, 1971,
p 491.
evidence, allows a plausible explanation of the origin of
degradation products 10, 13, and 15.
Simple Hydrolysis or Rearrangement Productss4, 14, 5,
and 6sCompound 4 (phenylglycine) can arise from a simple
hydrolysis of the phenylglycyl-glycine peptide bond. This
hydrolysis can occur directly on cefaclor or any of the degrada-
tion products or intermediates in which the phenylglycyl side
chain is intact. Compound 5 (phenylglycylglycine) can be
formed from cleavage of the C6-C7 bond, presumably after
â-lactam ring-opening. Although Scheme 2 illustrates the
formation of 4 from (d), there are probably multiple “path-
ways” for this simple cleavage. Compound 6 is the result of
a simple tautomerization of the dihydrothiazine double bond.
Compound 14 results from an intramolecular nucleophilic
attack of the 11′-amine on the â-lactam carbonyl (C8) followed
by cleavage of the C6-C7 bond. This nucleophilic attack most
likely occurs when the â-lactam is intact rather than after
â-lactam hydrolysis.23 The fact that this compound is a major
degradation product at pH 5.5, but was not detected in the
acidic degradation solutions is consistent with this hypothesis,
because at lower pH, the amine would be more fully proto-
nated and intramolecular nucleophilic attack on the â-lactam
carbonyl would not occur when the amine is in the protonated
state.
12. Metzger, J . V.; Vincent, E.-J .; Chouteau, J .; Mille, G. In Thiazole
and Its Derivatives, Metzger, J . V., Ed.; Wiley Interscience: New
York, 1979; pp 5-149.
1
3. GENOA is a computer program designed to generate all
candidate structures consistent with the molecular formula and
a given set of substructural constraints; cf. Carhart, R. E., et
al., J . Org. Chem. 1981, 46, 1708-1718.
4. Uno, T.; Masada, M.; Kuroda, Y.; Nakagawa, T. Chem. Pharm.
Bull. 1981, 29, 1344-1354.
1
15. (a) Silverstein, R. M.; Bassler, G. C. Spectrometric Identification
of Organic Compounds, 1st ed.; Wiley: New York, 1963; ap-
pendix B; (b) Smith, S. L. J . Chem. Ed. 1977, 54, 255.
1
6. Tabata, N.; Tomoda, H.; Matsuzaki, K.; Omura, S. J . Am. Chem.
Soc. 1993, 115, 8558.
17. The standard of 13 was prepared from the degradation of
cephalexin in the presence of benzaldehyde; see Experimental
Section for details.
1
8. Kopple, K. D.; Ohnishi, M. J . Am. Chem. Soc. 1969, 91 (4), 962-
9
70.
19. A search of Chemical Abstracts database revealed 15 different
patented R-hydroxy amino acid derivatives.
0. Schreiner, C. L.; J ones, E. E. J . Nutr. 1987, 117 (9), 1541-1549.
1. Dorman, D. E.; Lorenz, L. J .; Occolowitz, J . L.; Spangle, L. A.;
Collins, M. W.; Bashore, F. N.; Baertschi, S. W. J . Pharm. Sci.
2
2
Conclusions
1
997, 86, 540-549 (following paper in this issue).
2
2. Martinez, H.; Byrn, S. R.; Pfeiffer, R. R. Pharm. Res. 1990, 7
In this study of the aqueous degradation of cefaclor, we have
identified more than a dozen degradation products and have
proposed several major degradation pathways. This study,
while extensive, is not an exhaustive examination of the
aqueous degradation pathways of cefaclor. The degradation
pathways of cefaclor under neutral and basic pH conditions
(2), 147-153.
23. Nakashima, E.; Tsuji A.; Nakamura M.; Yamana, T. Chem.
Pharm. Bull. 1985, 33, 2098-2106.
2
2
2
2
4. Vilanova, B.; Donoso, J .; Mu n˜ oz, F.; Garcia-Blanco, F. Helv.
Chim. Acta 1996, 79, 1793-1802.
5. Indelicato, J . M.; Dorman D. E.; Engel, G. L. J . Pharm.
Pharmacol. 1981, 33, 119-121.
6. Baertschi, S. W.; Dorman, D. E.; Spangle, L. A.; Collins, M. S.;
Lorenz, L. J . J . Pharm. Biomed. Anal. 1995, 13, 323-328.
7. Skibic, M. J .; Taylor K. W.; Occolowitz, J . L.; Collins, M. W.;
Paschal, J . W.; Lorenz, L. J .; Spangle, L. A.; Dorman, D. E.;
Baertschi, S. W. J . Pharm. Sci. 1993, 82 (10), 1010-1017.
are complex, essentially unexplored, and remain to be eluci-
dated. The recent study by Vilanova et al.24 describing a
major pathway of cefaclor degradation under neutral condi-
tions provides an excellent start to exploring these degrada-
5
38 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

Su p p or tin g In for m a tion Ava ila blesA description of procedures
for crystal data collection and reduction, tables of crystallographic
data, and an ORTEP drawing of CEFD (17 pages). Ordering
information is given on any current masthead page.
Acknowledgments
The authors gratefully acknowledge Dr. Robin D. G. Cooper for
helpful suggestions regarding the synthesis of phenylglycylglycine,
Mr. J ack Fisher for helpful suggestions regarding the synthesis of 6,
and Dr. J oseph Indelicato and Mr. Gary Engel for providing a sample
of 13. The expert help of Kimmer Smith in carrying out the synthesis
of 6 is appreciated.
J S960427X
Journal of Pharmaceutical Sciences / 539
Vol. 86, No. 5, May 1997