Acid-catalyzed degradation of clarithromycin and erythromycin B: A comparative study using NMR spectroscopy

Article abstract of DOI:10.1021/jm9904811

One of the major drawbacks in the use of the antibiotic erythromycin A is its extreme acid sensitivity, leading to degradation in the stomach following oral administration. The modern derivative clarithromycin degrades by a different mechanism and much more slowly. We have studied the pathway and kinetics of the acid-catalyzed degradation of clarithromycin and of erythromycin B, a biosynthetic precursor of erythromycin A which also has good antibacterial activity, using ¹H NMR spectroscopy. Both drugs degrade by loss of the cladinose sugar ring and with similar rates of reaction. These results suggest that erythromycin B has potential as an independent therapeutic entity, with superior acid stability compared with erythromycin A and with the advantage over clarithromycin of being a natural product.

Full text of DOI:10.1021/jm9904811

J . Med. Chem. 2000, 43, 467-474
467
Acid -Ca ta lyzed Degr a d a tion of Cla r ith r om ycin a n d Er yth r om ycin B: A
Com p a r a tive Stu d y Usin g NMR Sp ectr oscop y
Mohd N. Mordi,† Michelle D. Pelta,‡ Valerie Boote,‡ Gareth A. Morris,*,‡ and J ill Barber*,†
School of Pharmacy and Pharmaceutical Sciences and Department of Chemistry, University of Manchester,
Oxford Road, Manchester M13 9PL, U.K.
Received September 21, 1999
One of the major drawbacks in the use of the antibiotic erythromycin A is its extreme acid
sensitivity, leading to degradation in the stomach following oral administration. The modern
derivative clarithromycin degrades by a different mechanism and much more slowly. We have
studied the pathway and kinetics of the acid-catalyzed degradation of clarithromycin and of
erythromycin B, a biosynthetic precursor of erythromycin A which also has good antibacterial
1
activity, using H NMR spectroscopy. Both drugs degrade by loss of the cladinose sugar ring
and with similar rates of reaction. These results suggest that erythromycin B has potential as
an independent therapeutic entity, with superior acid stability compared with erythromycin A
and with the advantage over clarithromycin of being a natural product.
Sch em e 1. Decomposition Pathway for Erythromycin A
In tr od u ction
in Acidic Aqueous Solution2
Erythromycin A (1) is an important antibiotic of the
macrolide class and has been in clinical use for some
50 years. Despite long success, particularly in the
treatment of deep-seated infections, it has a number of
problems in use. Compliance in taking this drug is
relatively poor. In children, this results from the vile
taste; in adults severe gastric disturbance proves intol-
erable in a minority of patients.1 In addition, erythro-
mycin and its metabolites are processed by the same
cytochrome P450 enzymes as a number of other drugs.
Saturation of these enzymes can lead to overdose of
theophylline, and even of common hay fever remedies,
if they are taken with erythromycin. All these problems
are exacerbated by the very large doses of erythromycin
A required for effective antimicrobial action. Four doses
of 250 mg or 500 mg per day for 7 or 14 days is typical.
The large doses are necessary because of the extreme
acid sensitivity of the drug. Below about pH 6.5 it is
degraded to anhydroerythromycin (2), as shown in
Scheme 1.2,3
Clarithromycin (3) is a modern derivative of eryth-
romycin A with an improved range of antibacterial
action and, in particular, with greatly improved acid
stability. Daily oral doses of this drug are typically a
quarter of the corresponding erythromycin doses, and
the course is often shorter. Chemically, clarithromycin
differs from erythromycin A only in the substitution of
6-OMe for 6-OH, but this gives two important changes
in properties. First, clarithromycin is more hydrophobic
than erythromycin A. This is the probable reason for
its higher activity against certain Gram-negative organ-
isms such as Haemophilus ducreyi.4 Second, it is unable
to cyclize in a 6,9-direction (and empirically does not
cyclize in a 12,9-direction5) and so cannot degrade in
acid via Scheme 1. The drug is much more stable to acid
than is erythromycin A6 and is even used as a compo-
nent of triple therapy against Helicobacter pylori.7 The
mechanism of acid-catalyzed degradation of clarithro-
mycin in aqueous solution has been explored6 with the
result shown in Scheme 2; clarithromycin degrades
below pH 3 at 37 °C to 5-O-desosaminyl-6-O-methyl-
erythronolide A (4), with loss of the cladinose sugar.6
Erythromycin B (5) is a biosynthetic precursor of
erythromycin A, lacking only the 12-OH group.8 It is
present as a minor component of commercial erythro-
mycin. Erythromycin B has comparable antibacterial
activity to erythromycin A9 but has never been exploited
as an antibacterial agent in its own right. The structure
* Correspondence to Jill.Barber@man.ac.uk or G.A.Morris@man.ac.uk.
† School of Pharmacy and Pharmaceutical Sciences.
‡ Department of Chemistry.
10.1021/jm9904811 CCC: $19.00 © 2000 American Chemical Society
Published on Web 01/25/2000

468 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3
Mordi et al.
Ta ble 1. Solubilities of Erythromycins A and B and
Clarithromycin in Deuterated Phosphate Buffer (50 mM) at
Apparent pH 7.0 and 25 °C
Sch em e 2. Decomposition of Clarithromycin in Acidic
Aqueous Solutiona
drug
solubility (mM)
erythromycin A
erythromycin B
clarithromycin
53.4
19.9
6.3
An NMR-Ba sed Com p a r ison of th e Acid -Ca ta -
lyzed Degr a d a tion of Er yth r om ycin B a n d Cla r i-
th r om ycin . The aim of this study was to compare the
pathway and kinetics of clarithromycin and erythromy-
cin B decomposition under conditions which mimic the
stomach (pH 2.5 and 37 °C). NMR spectroscopy was the
preferred method because of the possibility of determin-
ing the pathways and the kinetics simultaneously. The
two drugs were therefore degraded in the NMR tube at
37 °C and apparent pH 2.5 in Britton-Robinson buffer11
over 48 h. Experiments were also carried out at 55 °C
for clarithromycin in order to obtain good data for the
slower phase of the degradation.
Some technical difficulties had to be overcome before
these experiments could be implemented successfully.
For sensitivity reasons it was necessary to work with
solutions of the two drugs close to saturation under
experimental conditions, leading to small amounts of
transient precipitation during the early stages of the
degradations. This caused severe disturbance of field
homogeneity, rendering conventional NMR data acqui-
sition methods unusable. A recently developed method
for rapid and accurate automatic field homogeneity
adjustment12 was therefore used to maintain good field
homogeneity throughout the experiments, using alter-
nating periods of shimming and data acquisition. Re-
sidual shimming errors were corrected using the FIDDLE
algorithm13 for reference deconvolution.
a
After ref 6.
of the drug suggests that it should possess some of the
advantageous properties of clarithromycin. The lack of
a 12-OH group is expected to increase hydrophobicity
and will prevent 12,9-cyclization and the formation of
inactive compounds analogous to anhydroerythromycin
A. Erythromycin B also shares some of the advantages
of erythromycin A. It is a fermentation product and can
be produced relatively cheaply and with less adverse
effect on the environment than the semisynthetic
clarithromycin.
(i) A Revised Decom p osition P a th w a y for Cla r i-
1
th r om ycin . Figure 1 shows the region of the H NMR
spectrum of clarithromycin corresponding to methyl 15
as it changed over 48 h at 55 °C. It can be seen that the
signals corresponding to clarithromycin (labeled C)
disappear over less than 3 h, to be replaced by a second
and then a third set of signals (labeled F and G). This
shows that Scheme 2, proposed by Nakagawa et al.,6 in
which the macrolide degrades to a single product (4), is
incomplete under these conditions.
We now report a comparative study of the acid-
catalyzed degradation of erythromycin B and clarithro-
mycin, with a view to determining whether erythromy-
cin B has potential as an independent therapeutic
entity.
Resu lts
The component corresponding to methyl signal F was
identified by NMR and mass spectrometry. The degra-
dation was run for 2 h, to the point where the concen-
tration of F was maximized. A two-dimensional diffusion-
ordered (DOSY14,15) spectrum of the mixture was then
acquired.16 In a 2D DOSY spectrum, NMR signals are
separated according to chemical shift in one dimension
and according to their self-diffusion coefficients in the
other. The degradation was, of course, continuing during
the DOSY experiment, which affected the signal-to-noise
ratio slightly. (The alternative strategy, that of quench-
ing the reaction by the addition of base, would have
caused a more severe reduction in the signal-to-noise
ratio because of the poor solubility of clarithromycin in
aqueous media at neutral pH.) The DOSY spectrum
showed clearly that the signals for the cladinose ring
exhibit much faster diffusion than those due to the
macrolide and the desosamine sugar, demonstrating
that cladinose had been hydrolyzed from the macrolide.
Solu bility Mea su r em en ts. Clarithromycin is poorly
soluble in aqueous media;5 this property is expected to
contribute to its activity against Gram-negative organ-
isms. Erythromycin A, which has better water solubility,
is almost inactive against most Gram-negative organ-
isms, but it has better activity than clarithromycin
against Haemophilus influenzae.10 The relative solubili-
ties of the three drugs at physiological pH were assessed
by preparing saturated solutions in D2O-based phos-
phate buffer containing 3 mM 2,2,3,3-tetradeuterio-4,4-
dimethyl-4-silapentanoic acid (TSP) and measuring the
1
relative integrals in the H NMR spectrum.
Table 1 shows that erythromycin B is of aqueous
solubility intermediate between erythromycin A and
clarithromycin. It is likely therefore to have a spectrum
of action broadly similar to the other macrolides, with
good activity against Gram-positive bacteria, but very
limited activity against Gram-negative bacteria.

Degradation of Clarithromycin and Erythromycin B
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3 469
Clarithromycin was allowed to degrade at pH 2.5 and
1
55 °C for 4 days, and the H NMR spectrum run. (The
low field region of the spectrum is shown in the
Supporting Information.) Extrapolation of the 48 h time
course16 revealed that by this point the F and G species
were close to equilibrium, with approximately equal
concentrations. An olefinic signal at 6.05 ppm follows
the same time course as methyl G, and, on extreme
resolution enhancement, shows quartet structure with
a long range coupling constant of ca. 1 Hz. Electrospray
mass spectrometry showed signals at m/z 590 (due to
4) and 554. The MH+ signal at m/z 554 suggests that 4
loses two molecules of water to give an olefinic species,
with NMR signals G. Further signals, which grow more
slowly than G, can be seen at 6.01 and 5.88 ppm; at the
end of the experiment these were approximately 20 and
10% of the main olefinic signal, respectively.
Morimoto et al.5 showed that it is possible to prepare
4 by acid treatment of clarithromycin in nonaqueous
conditions, that prolonged treatment of 4 with acid
yields 6, and that 6 is capable of isomerization to 7. Our
NMR and mass spectral data are consistent with these
reactions also occurring in aqueous acid, with clarithro-
mycin 3 losing cladinose to form 4, dehydrating to give
6, and then isomerizing further, perhaps to 7; the third
olefinic signal suggests still further reaction. The signals
labeled G would therefore correspond to 6. In a further
experiment, 5-O-desosaminyl-6-O-methylerythronolide
A (4) was synthesized and incubated for 48 h at 55 °C
in the same Britton-Robinson buffer at apparent pH
2.5. The NMR spectrum obtained at the end of the
experiment was similar to that described above. The
revised pathway for the degradation of clarithromycin
in aqueous acid is shown in Scheme 3.
Several weak signals were observed at different
stages of the degradation, suggesting the presence of
small amounts of further species. A weak signal (ca. 5%)
at 4.85 ppm followed the same time course as methyl
C, suggesting a species in equilibrium with clarithro-
mycin on the time scale of this experiment; this may be
the elusive 9,12-hemiketal (analogous to 1b). Anomeric
equilibrium gives rise to two H1′′ signals of the free
cladinose at 5.1 and 4.96 ppm following its hydrolysis
from 3; the â:R ratio is ca. 6:1. The slow production of
second and third olefinic signals suggests that, on a time
scale of weeks, degradation to further products occurs.
(ii) An a lysis of Ra tes of Rea ction in th e Decom -
p osit ion of Cla r it h r om ycin . The principal rates of
reaction in Scheme 3 were determined by iterative
nonlinear least-squares fitting of integrated signal
intensities extracted from the 55 °C proton NMR
spectra. For each point on the time course, the spectrum
was baseline corrected and the methyl region was
F igu r e 1. Degradation of clarithromycin in deuterated Brit-
ton-Robinson buffer at apparent pH 2.5 and 55 °C. The high
field region of the 400 MHz NMR spectrum, corresponding to
CH3-15, is shown over the time course of the experiment (48
h).
The signals labeled F in Figure 1 therefore correspond
to 5-O-desosaminyl-6-O-methylerythronolide A (4). This
conclusion was supported by electrospray mass spec-
trometry; the major peak in the spectrum of the mixture
appeared at m/z 590, corresponding to an MH+ ion
consistent with compound 4 and in agreement with
Nakagawa et al.6
The material corresponding to methyl signal G was
also identified by NMR and mass spectral methods.
Sch em e 3. Revised Pathway for the Decomposition of Clarithromycin in Acidic Aqueous Solution

470 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3
Mordi et al.
Ta ble 2. Kinetic Parameters Obtained by Fitting
Experimental Signal Intensities for the Degradation of
Clarithromycin at 55 °C and 37 °C and Apparent pH 2.5a
In Figure 2 the signals due to erythromycin B disap-
pear, with concomitant emergence of a second set of
signals (marked J ). The product is essentially stable
over this time scale in these reaction conditions. The
DOSY spectrum of the mixture was similar to that
obtained during clarithromycin degradation and is
shown in Figure 3. Resonances assigned to the cladinose
ring (H1′′, H2′′, H8′′) showed considerably faster diffu-
sion than those due to the macrolide ring and the
desosamine sugar, showing that the cladinose sugar had
been hydrolyzed from the macrolide. This result was
confirmed by electrospray mass spectrometry of the
mixture, in which m/z 561 (MH+) indicated the presence
of 5-desosaminyl erythronolide B as the principal com-
ponent of the mixture. This compound, 8, was now
synthesized, isolated, and characterized. The full as-
signment of its 1H NMR spectrum is given in the
Supporting Information. The change in J 4,5 from 7 to 8
Hz to 2.6 Hz on loss of cladinose indicates a significant
conformational change in this part of the macrolide ring.
The same reduction in J 4,5 is observed in 3-mycaro-
sylerythronalide B (a derivative of erythromycin B in
which the desosamine sugar is missing). This suggests
that interactions between the two sugars constrain the
conformation of the macrolide ring in the C2-C6 region
(Mordi and Barber, unpublished data). Conformational
analysis of these derivatives is in progress.
55 °C
37 °C
kCF
kFG
kGF
2.51 ( 0.02 × 10-2 min-1
2.28 ( 0.02 × 10-4 min-1
2.40 ( 0.09 × 10-4 min-1
2.229 ( 0.004 × 10-3 min-1
1.38 ( 0.03 × 10-5 min-1
a
Uncertainties quoted are twice the standard deviations esti-
mated in the Levenberg-Marquardt fit.
subjected to least-squares fitting using singular value
decomposition17 with a set of three basis functions. The
basis functions contained the isolated methyl triplet
signals C, F , and G, respectively, and were obtained by
taking linear combinations of experimental spectra,
adjusting the coefficients by eye to minimize cross-
contamination of the signals. This fitting procedure
gives close to optimum accuracy for the signal intensity
and offers a significant improvement over simply mea-
suring peak heights. The relative signal intensities were
normalized to remove the effects of small drifts in
spectrometer sensitivity, and finally the small baseline
error in the intensities for C was corrected using the
data for times greater than 250 min. The signal intensi-
ties thus obtained were fitted by the Levenberg-
Marquardt algorithm using the reaction scheme
kCF
kFG
Integration of the signals due to 5 and 8 over the time
course of the acid-catalyzed degradation, and compari-
son with the standard, revealed that these two com-
pounds did not account for all the macrolide present,
especially in the early stages of the reaction. Traces of
a third compound could be detected in the spectra (see
Figure 2, in which these signals are marked K), but it
could not be identified from these spectra alone. A clue
to the identity of this intermediate came from the
methyl 19 signal. During the course of the experiment
this signal collapsed from a doublet to a singlet owing
to incorporation of deuterium at C8. This suggested that
erythromycin B enol ether (9) was a likely candidate
for the intermediate compound.
The reaction was now followed by electrospray mass
spectrometry, sampling at 5 min intervals. The early
accumulation of a compound with an MH+ at m/z 700
could be seen, followed by its disappearance between
50 and 250 min. Erythromycin B enol ether has a
molecular weight of 699 Da, so this experiment provided
further support for its presence in the reaction mixture.
C
8 F y z G
kGF
with the program Mathematica 3.018 on an Apple Power
Macintosh model 7600/132 computer, using the rate
constants and initial time as variable parameters. The
fitting was repeated for data obtained at 37 °C, but with
the reverse reaction kGF omitted from the reaction
scheme since this reaction is not significant over the
time scale of the experiment. Indeed the presence of the
reverse reaction kGF as 55 °C was detectable only by
kinetic analysis. Direct evidence could, in principle, be
obtained by measuring the incorporation of deuterium
at C8 and C10 of 4. The equilibrium is, however, so slow
that further degradation takes place during the several
days required to establish even a modest incorporation
of deuterium. Table 2 shows the kinetic parameters and
estimated standard errors obtained from the fitting
process at the two temperatures. The experimental and
calculated time courses of the integrals of signals C, F ,
and G at the two temperatures were compared,16 and
the quality of the fit obtained was very satisfactory.
The key step in terms of the use of clarithromycin as
a drug is the initial loss of cladinose. At 37 °C, this is a
first-order process with a rate constant of 2.23 × 10-3
min-1, corresponding to a half-life for the drug of 310
min.
Erythromycin B enol ether (9) was synthesized by the
1
literature method8 and fully characterized. Its H NMR
assignments at apparent pH 7.4 are shown in Support-
ing Information. It was then incubated at pH 2.5 and
55 °C in the NMR spectrometer. The 1H NMR spectrum
obtained was identical with that obtained when eryth-
romycin B was treated in the same way. This demon-
strates that erythromycin B enol ether is readily
converted to erythromycin B at this pH, strongly sup-
porting the suggestion that 9 is formed during the
degradation of erythromycin B to 8.
The data obtained by monitoring the degradation of
erythromycin B at 37 °C were now subjected to the same
type of kinetic analysis as the clarithromycin data. The
erythromycin B spectra showed poorer signal-to-noise
ratios in the methyl region than clarithromycin, but the
8′′ methyl region of the spectrum, obscured in the
(iii) Mech a n ism of Acid -Ca ta lyzed Degr a d a tion
of Er yth r om ycin B. A mechanism for the degradation
of clarithromycin in acid has been published,6 but that
of erythromycin B has not appeared in the literature.
Initially, therefore, we set out to determine the pathway
of acid-catalyzed degradation of erythromycin B.
A solution of erythromycin B in D2O-based Britton-
Robinson buffer (containing TSP as a standard) was
degraded in the NMR spectrometer at 37 °C. The
1
reaction was followed by measuring H NMR spectra
at 10 min intervals. A typical time course is shown in
Figure 2.

Degradation of Clarithromycin and Erythromycin B
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3 471
F igu r e 2. Degradation of erythromycin B in deuterated Britton-Robinson buffer at apparent pH 2.5 and 37 °C. Signals due to
the degradation products are marked J and K.
clarithromycin samples, was well-resolved. The concen-
trations of the three species 5, 9, and 8 were therefore
followed using the peak heights of the methyl 8′′ signals
rather than the fitted methyl multiplet integrals (which
showed very similar time courses but more scatter).
Experimental data were subjected to reference decon-
volution, resolution enhancement, and baseline correc-
tion, and methyl 8′′ peak heights were measured. The
peak heights were multiplied by the experimental line
widths to determine the relative integrals and normal-
ized to remove the effects of any spectrometer gain drift.
The 8′′ methyl peak intensities for the two anomers of
cladinose were combined to give a single value.
the simple biexponential behavior expected for two
sequential first-order reactions. The source of the
unusual behavior here is the incorporation of deuterium
from the solvent D2O at C8 of erythromycin B, by the
reverse reaction 9 f 5. Only a single methyl 8′′ signal
is seen for 5, deuteriation at C8 having a negligible
effect on the chemical shift of this methyl group.
Because the dehydration of 5 to the enol ether 9 requires
the breaking of the C-H bond at C8, there is a
substantial primary kinetic isotope effect on the reaction
5 f 9. Thus at the onset of reaction, 5 rapidly forms 9.
The reverse reaction then forms 8-deuterio erythromy-
cin B, 5D, which dehydrates about five times more slowly
than the parent compound 5, rapidly pushing the
limiting position of the equilibrium between 5 and 9
back toward 5. Both compounds 5 and 5D hydrolyze at
the same rate to 8 (which is, of course, partially
The time courses of the resultant three signals for
species 5, 9, and 8, shown in Figure 4, show some
interesting features. The rapid initial fall in 5, and
concomitant rapid growth in 9, are not consistent with

472 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3
Mordi et al.
Sch em e 4. Degradation of Erythromycin B in Acidic
Aqueous Solution
F igu r e 3. DOSY spectrum of the products of the degradation
of erythromycin B in Britton-Robinson buffer at apparent pH
2.5 and 37 °C for 24 h.
Ta ble 3. Kinetic Parameters Obtained by Fitting
Experimental Signal Intensities for the Degradation of
Erythromycin B at 37 °C and Apparent pH 2.5a
k59
1.17 ( 0.05 × 10-2 min-1
5.55 ( 0.19 × 10-2 min-1
2.36 ( 0.19 × 10-3 min-1
2.50 ( 0.01 × 10-3 min-1
k95D
k5D9
k58
a
Uncertainties quoted are twice the standard deviations esti-
mated in the Levenberg-Marquardt fit.
F igu r e 4. Seventeen hour time course of the degradation of
erythromycin B in deuterated Britton-Robinson buffer at
apparent pH 2.5 and 37 °C. Experimental points are indicated
by open circles, calculated fit by solid lines.
an extra pathway is added from the enol ether (9) to
5-desosaminyl erythronolide B (8). (The descladinose
enol ether would be a transient intermediate.) Attempts
to fit the experimental data using a reaction scheme
which includes this pathway also gave acceptable, albeit
slightly poorer, fits for values of the rate constant k98
less than or equal to k58. The presence or absence of this
pathway has implications for the behavior of erythro-
mycin B in H2O as opposed to D2O, since in H2O, the
unfavorable isotope effect will no longer act to reduce
the concentration of the enol ether (9), emphasizing the
importance of any side reactions involving 9. The half-
life for erythromycin B reduces from 235 to 200 min as
k98 changes from zero to k58. However, in medicinal
terms, the enol ether 9 is expected to serve as a reservoir
for 5. The half-life of (erythromycin B + enol ether)
varies from 335 min (k98 ) 0) to 277 min (k98 ) k58).
Thus, whether or not the extra pathway 9 f 8 is
significant, the half-life of erythromycin B under condi-
tions mimicking those in the stomach is similar to that
of clarithromycin.
deuterated), since this reaction does not involve the
C-H bond at C8. The net result is a steep initial decline
in 5 and growth in 9, which rapidly level off, while the
growth in the product 8 remains close (but not identical)
to that expected for single-exponential conversion of
reactants to product.
The measured signal intensities were again fitted by
the Levenberg-Marquardt algorithm with rate con-
stants and initial time as variable parameters, but this
time using reaction Scheme 4, with k58 ) k5D8. Because
of the complexity of the reaction scheme it was neces-
sary to reduce the size of the data set used in the fitting,
so only 43 of the 84 time points available were used.
Table 3 shows the kinetic parameters, including the
ratio RHD of the dehydration rate constants k59 and k5D9
,
and estimated standard errors obtained from the fitting
process. Once again, the quality of the fit, as evidenced
both by the statistics and by Figure 4, is gratifying.
As with all analyses of kinetic data, there is an
infinite set of possible models which will fit the experi-
mental data, which the combined application of chemi-
cal knowledge and Occam’s Razor generally reduces to
a single kinetic scheme. Here, however, there remains
at least one plausible variation on Scheme 4, in which
Th e Acid -Ca ta lyzed Degr a d a tion of Er yth r om y-
cin A. In a control experiment, erythromycin A was
dissolved in Britton-Robinson buffer at apparent pH
2.5 and incubated in the NMR spectrometer at 37 °C.
Under these conditions, the degradation of erythromycin
A to anhydroerythromycin was complete before a spec-

Degradation of Clarithromycin and Erythromycin B
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3 473
Electrospray-ionization mass spectra were acquired using a
PLATFORM mass spectrometer. The sample solution (10 µL)
was analyzed by ESI-MS using acetonitrile or 0.3% aqueous
trifluoroacetic acid as the running solvent. Data were analyzed
using the program PLATFORM (Micromass) with a Masslynx
data system. All degradation experiments were carried out in
deuterated Britton-Robinson buffer11 (40 mM) unless other-
wise stated. Fine adjustments to the pH of the buffer were
made using sodium deuterioxide or deuterium chloride solu-
tions in D2O.
trum could be acquired, that is, within 10 min of the
beginning of the experiment.
Discu ssion
Erythromycin A is one of the most important and
successful drugs of all time. It has excellent activity
against a wide range of Gram-positive pathogens. It is
particularly effective because it inhibits bacterial pro-
tein synthesis, a metabolic process which is necessary
for cellular survival, not just for cell division. The
second-generation macrolides, clarithromycin and azithro-
mycin, have slightly different spectra of antibacterial
activity from erythromycin A,4,7,10,19 and azithromycin
shows activity against Toxoplasma gondii in experi-
mental animals.20 The prediction of complex in vivo
characteristics, such as metabolism and tissue absorp-
tion, from in vitro chemical data is in general a perilous
undertaking. Among such similar compounds, however,
it is fair to point to the large differences in acid stability
as being crucial to the success of the modern macrolides.
Acid stability is a key factor in allowing shorter courses
and lower doses of these drugs to be given, reducing
adverse reactions in both adults and children.
It would clearly be advantageous to have available
an acid-stable macrolide with a therapeutic profile
similar to that of erythromycin A. Erythromycin B has
been known for almost as long as erythromycin A, but
until very recently it has not been available in large
quantity. (Interest in the large scale production of
macrolides other than erythromycin A has been stimu-
lated by the identification and modification of the genes
responsible for macrolide synthesis in actinomycetes.21)
In vitro studies have shown that this drug has a similar
therapeutic range and activity to erythromycin A.9 Our
solubility data suggest that an exhaustive survey of the
therapeutic profile would lead to a position intermediate
between erythromycin A and clarithromycin.
Structurally, erythromycin B differs from erythromy-
cin A only in having no hydroxyl group at C12. It is
therefore unable to cyclize in a 12,9-direction and unable
to form a spiroketal analogous to 2. The formation of 2
from 1 is a facile process in aqueous acid, as indicated
by the failure even to obtain an NMR spectrum of
erythromycin A (1) under near-physiological conditions
mimicking the stomach. We have now demonstrated
that erythromycin B is indeed much more stable in acid
than is erythromycin A. Like clarithromycin, it degrades
in acid by loss of the cladinose sugar. The rate constants
for loss of cladinose from clarithromycin and erythro-
mycin B have been determined as 2.23 × 10-3 min-1
and 2.50 × 10-3 min-1, respectively. These rates are
very similar, as would be expected for processes involv-
ing the same mechanism, and indicate that erythromy-
cin B has comparable stability in acid to clarithromycin.
Both drugs show considerably higher stability than
erythromycin A. Like clarithromycin and other modern
macrolides such as azithromycin, erythromycin B is
expected to survive oral administration without elabo-
rate formulation. Depending upon the results of in vivo
studies, it could therefore find use in the clinic as an
inexpensive, but acid-stable, macrolide antibiotic.
Aqu eou s Solu bilities of Er yth r om ycin A, Cla r ith r o-
m ycin , a n d Er yth r om ycin B. The aqueous solubilities of 1,
3, and 5 were determined at room temperature (25 °C) in
deuterated phosphate buffer (50 mM, apparent pH 7.0) with
a known amount of TSP (3 mM) as an internal standard. An
excess of each compound was added separately to the phos-
phate buffer, and the suspensions were continuously stirred
for 24 h. The pH of each suspension was determined at
intervals during the solubilization and adjusted if necessary.
After 24 h, the suspensions were filtered through filter paper
1
and analyzed by H NMR spectroscopy.
Mea su r em en t of th e Tim e Cou r se of th e Degr a d a tion
of 5 by ESI-MS. The degradation reaction of 5 (4 mM) in
deuterated Britton-Robinson buffer (apparent pH 2.5) was
monitored at 55 °C. At intervals of 0.5 min, 50 µL of the
degradation mixture was aliquoted into 0.5 mL of cooled boric
acid buffer (pH 7.5), to quench the degradation reaction and
also to dilute the mixtures for mass spectrometry. Samples
were analyzed by ESI-MS using 0.3% trifluoroacetic acid in
water as the mobile phase.
Kin etic Mea su r em en ts by NMR Usin g Rea l-Tim e Au -
tosh im m in g. Compounds 3 and 5 were dissolved in deuter-
ated Britton-Robinson buffer at apparent pH 2.5 to give
concentrations of 4 mM, and the solutions were degraded at
55 °C or 37 °C for a total time of 42 h. The experiment was
performed by switching at regular intervals between 1D 1H
NMR spectroscopy, using presaturation for water suppression,
and a deuterium spin-echo-based autoshimming experiment.13
The degradation was performed using an array of 30 transients
for the first 20 spectra, 80 transients for the next 30 spectra,
180 transients for the next 40 spectra, and 380 transients for
each of the remaining 40 spectra. The experiment was ar-
ranged so that the sample was automatically autoshimmed
at the end of each block of transients. Autoshimming was used
to compensate for the local field disturbance caused by
variation in sample composition during the course of the
reaction. The 1H spectra were recorded with a spinning sample,
with a preacquisition saturation delay of 2 s, a 90° pulse of
width 5.2 µs, and a spectral width of 5000 Hz. The spectra
were processed with reference deconvolution using the FIDDLE
algorithm for line shape regularization using the TSP signal
as the reference, with a Gaussian time constant of 0.25 s,
followed by cubic spline baseline correction. The deuterium
spin-echo autoshimming experiment was performed on a
static sample as described in ref 13, with a spectral width of
1000 Hz, a 90° pulse of width 84 µs, and a preacquisition delay
of 3 s.
DOSY Sp ectr oscop y of th e Degr a d a tion P r od u cts of
3 a n d 5. Compounds 3 and 5 were dissolved in deuterated
Britton-Robinson buffer at apparent pH 2.5 to give concentra-
tions of 4 mM, and the solutions were degraded at 55 °C for 2
and 24 h, respectively. The degraded sample of 3 was cooled
1
to ambient temperature and subjected to a H high resolution
DOSY NMR experiment using the GCSTESL pulse sequence.22
Ten spectra were acquired, with gradient pulses of 2 ms
ranging in strength from 1 to 30 G cm-1 and a diffusion delay
of 0.2 s. The degraded sample of 5 was subjected to a 1H high
resolution DOSY NMR experiment at 30 °C using the BPPSTE
pulse sequence.22 Twelve spectra were acquired, with gradient
pulses of 2.9 ms ranging in strength from 1 to 15 G cm-1 with
a diffusion delay of 0.1 s. FIDDLE reference deconvolution was
employed to correct line shapes, using the TSP line as
reference, and baseline errors and systematic errors in ap-
parent gradient strength were corrected. The 2D HR-DOSY
spectra were constructed by taking the first echo spectrum and
distributing the intensities of the individual signals in the
second dimension according to diffusion coefficient, using
Exp er im en ta l Section
Gen er a l P r oced u r es. 1H NMR spectra were acquired
using a Varian INOVA 400 spectrometer operating at 400 MHz
or a Varian Unity 500 spectrometer operating at 500 MHz.

474 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 3
Mordi et al.
(3) Alam, P. S.; Buxton, P. C.; Parkinson, J . A.; Barber, J . Structural
studies on erythromycin A enol ethersfull assignments of the
H-1 and C-13 NMR spectra. J . Chem. Soc., Perkin. Trans. 2
1995, 1163-1167.
(4) Dangor, Y.; Ballard, R. C.; Miller, S. D.; Koornhoof, H. J .
Antimicrobial susceptibility of Haemophilus ducreyi. Antimicrob.
Agent Chemother. 1990, 34, 1303-1307.
(5) Morimoto, S.; Misawa, Y.; Asaka, T.; Kondah, H.; Watanabe, Y.
Chemical modification of erythromycins. Structure and antibac-
terial activity of acid degradation products of 6-O-methyleryth-
romycin A. J . Antibiot. 1990, 43, 570-573.
(6) Nakagawa, Y.; Itai, S.; Yoshida, T.; Nagai, T. Physicochemical
properties and stability in the acid solution of a new macrolide
antibiotic clarithromycin, in a comparison with erythromycin.
Chem. Pharm. Bull. 1992, 40, 725-728.
(7) Hunt, R. H. Eradication of Helicobacter pylori infection. Am. J .
Med. 1996, 100, 42S-51S.
(8) Kurath, P.; J ones, P. H.; Egan, R. S.; Perun, T. Acid degradation
of erythromycin A and erythromycin B. Experientia 1971, 27,
362.
(9) Kibwage, I. O.; Hoogmartens, J .; Roets, E.; Vanderhaeghe, H.;
Verbist, L.; Dubost, M.; Pascal, C.; Petitjean, P.; Levol, G.
Antibacterial activities of erythromycin A, erythromycin B,
erythromycin C and erythromycin D and some of their deriva-
tives. Antimicrob. Agent Chemother. 1985, 28, 630-633.
(10) Liebers, D. M.; Baltch, A. L.; Smith, R. P.; Hammer, M. C.;
Conroy, J . V.; Shayegani, M. Comparative in-vitro activities of
A-56268 (TE-031) and erythromycin against 306 clinical isolates.
J . Antimicrob. Chemother. 1988, 21, 565-570.
(11) Britton, H. T. S.; Robinson, R. A. The use of the antimony-
antimonous oxide electrode in the determination of the concen-
tration of hydrogen ions and in potentiometric titrations the
Prideaux-Ward universal buffer mixture. J . Chem. Soc. 1931,
458-473.
(12) Barjat, H.; Chilvers, P. B.; Fetler, B. K.; Horne, T. J .; Morris,
G. A. A practical method for automated shimming with normal
spectrometer hardware. J . Magn. Reson. 1997, 125, 197-201.
(13) Morris, G. A.; Barjat, H.; Horne, T. J . Reference deconvolution
methods. Prog. Nucl. Magn. Reson. Spectrosc. 1997, 31, 197-
257.
(14) Morris, K. F.; J ohnson C. S. Diffusion-ordered two-dimensional
nuclear magnetic resonance spectroscopy. J . Am. Chem. Soc.
1992, 114, 3139-3141.
Gaussian line shapes with widths determined by the standard
errors obtained from the fitting process.15
Tw o-Dim en sion a l NMR Sp ect r oscop y (TOCSY a n d
DQF -COSY). Solutions of 8 and 9 (4 mM) were prepared in
deuterated buffer at apparent pH 2.5 and 7.4, respectively.
Both TOCSY and DQF-COSY spectra were run at ambient
temperature. For both experiments, the water resonance was
suppressed by presaturation. Data matrices of 2048 × 1024
complex points were acquired. NMR spectra were processed
using negative line broadening and Gaussian multiplication
for enhancement of resolution.
Syn th esis a n d Ch a r a cter iza tion of 5-Desosa m in yl
Er yth r on olid e B (8). Erythromycin B (0.50 g) in 175 mL of
protiated Britton-Robinson buffer (pH 2.5) was degraded at
55 °C for 16 h. The solution was allowed to cool to room
temperature, and its pH was adjusted to 9.5. The aqueous
solution was extracted with chloroform (3 × 175 mL), which
then was concentrated on a rotary evaporator and dried over
anhydrous sodium sulfate. Finally, the organic solvent was
removed in vacuo. The colorless crystalline product was
recrystallized from acetone to give 0.30 g (70% yield) of
5-desosaminyl erythronolide B (8), mp 102-104 °C. Full
1
assignment of the H NMR spectrum was achieved using a 4
mM solution of 8 in deuterated Britton-Robinson buffer
(apparent pH 2.5) at ambient temperature. A TOCSY spectrum
was used to distinguish the different spin groups of 8, and a
DQF-COSY spectrum allowed signals within each spin group
to be distinguished. The isolated methyl signals were assigned
by analogy with the 1H NMR spectrum of erythromycin B
(P. Tyson & J . Barber, unpublished data).
Syn t h esis a n d Ch a r a ct er iza t ion of E r yt h r om ycin B
En ol Eth er (9). Erythromycin B enol ether was prepared by
an adaptation of the published procedure for the preparation
of erythromycin A enol ether.3 Erythromycin B (200 mg) was
dissolved in glacial acetic acid (5 mL) and was allowed to stand
at room temperature for 4 h. Saturated sodium bicarbonate
solution (50 mL) was added, and the enol ether was extracted
using dichloromethane (3 × 100 mL). The organic layer was
concentrated under reduced pressure, washed with saturated
sodium bicarbonate solution, and then was dried over anhy-
drous sodium sulfate. The solvent was removed in vacuo. The
colorless crystalline product was recrystallized from acetone
to give 177 mg of 9 (91% yield), mp128-130 °C (lit. 126-128
(15) Barjat, H.; Morris, G. A.; Smart, S.; Swanson, A. G.; Williams,
S. C. R. High-resolution diffusion-ordered 2D spectroscopy (HR-
DOSY) - A new tool for the analysis of complex mixtures. J .
Magn. Reson. Ser. B. 1995, 108, 170-172.
(16) For figure, please see Supporting Information.
(17) Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vetterling, W.
T. Numerical Recipes in C; Cambridge University Press: Cam-
bridge, 1988.
1
°C23). Full assignments of the H NMR spectra were achieved
in deuteriochloroform and deuterated phosphate buffer (ap-
parent pH 7.4) using TOCSY and DQF-COSY spectra in the
same way as for 8. All spectra were measured at ambient
temperature.
(18) Wolfram. S. The Mathematica Book, 3rd ed.; Wolfram Media/
Cambridge University Press: Cambridge, 1996.
(19) Zuckerman, J . M.; Kaye, K. M. The newer macrolides. Azithro-
mycin and clarithromycin. Infect. Dis. Clin. North Am. 1995, 9,
731-745. (b) Bahal, N.; Nahata, M. C. The new macrolide
anitibiotics: azithromycin, clarithromycin, dirithromycin, and
roxithromycin. Ann. Pharmacother. 1992, 26, 46-55. (c) Meyer,
A. P.; Bril-Bazun, C.; Mattie, H.; Van Den Broek, P. J . Uptake
of azithromycin by human monocytes and enhanced intracellular
antibacterial activity against Staphylococcus aureus. Antimicrob.
Agents Chemother. 1993, 37, 2318-2322.
Ack n ow led gm en t. M.D.P. thanks Infineum Ltd.
and the EPSRC for a CASE studentship. M.N.M. thanks
Universiti Sains Malaysia for a fellowship. This work
was supported by EPSRC Grants GR/K44619, GR/
L17443, and GR/M16863. We thank Dr. Warren Mann
(Abbott International Division, Queenborough, Kent) for
gifts of clarithromycin and erythromycins A and B.
(20) Schwab, J . C.; Cao, Y.; Slowik, M. R.; J oiner, K. A. Localization
of azithromycin in Toxoplasma gondii-infected cells. Antimicrob.
Agents Chemother. 1994, 38, 1620-1627.
1
Su p p or tin g In for m a tion Ava ila ble: Complete H NMR
(21) (a) McDaniel, R.; Thamchaipenet, A.; Gustafsson, C.; Fu, H.;
Betlach, M.; Betlach, M.; Ashley, G. Multiple genetic modifica-
tions of the erythromycin polyketide synthase to produce a
library of novel “unnatural” natural products. Proc. Natl. Acad.
Sci. U.S.A. 1999, 96, 1846-1851. (b) Rowe, C. J .; Cortes, J .;
Gaisser, S.; Staunton, J .; Leadlay, P. F. Contruction of new
vectors for high-level expression in actinomycetes. Gene 1998,
216, 215-223. (c) Stassi, D. L.; Kakavas, S. J .; Reynolds, K. A.;
Gunawardana, G.; Swanson, S.; Zeidner, D.; J ackson, M.; Liu,
H.; Buko, A.; Katz, L. Proc. Natl. Acad. Sci. U.S.A. 1998, 95,
7305-7309.
data for 5-desosaminyl erythronolide B (8) and erythromycin
B enol ether (9) in aqueous buffer and the following figures:
(a) low field region of the 400 MHz 1H NMR spectrum of
clarithromycin after degradation for 4 days at apparent pH
2.5 and 55 °C, (b) 400 MHz DOSY spectrum of clarithromycin
after approximately 2 h degradation in the same buffer at 37
°C, (c) and (d) 48 h time courses of the degradation of
clarithromycin in the same buffer at 55 °C and 37 °C. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(22) Pelta, M. D.; Barjat, H.; Morris, G. A.; Davis, A. L.; Hammond,
S. J . Pulse sequences for high-resolution diffusion ordered
spectroscopy (HR-DOSY). Magn. Reson. Chem. 1998, 36, 706-
714.
Refer en ces
(1) Graham, E. M. Erythromycin. Obstet. Gynecol. Clin. North Am.
1992, 19, 539-549.
(2) Cachet, T. H.; Van den Mooter, G.; Hauchecorne, R.; Vinckier,
C.; Hoogmartens. J . Decomposition kinetics of erythromycin A
in acidic aqueous solutions. Int. J . Pharm. 1989, 55, 59-65.
(23) Krowicki, K.; Zamojski, A. J . Chemical modification of erythro-
mycins. IV. 8-hydroxyerythromycin B. J . Antibiot. 1973, 26,
587-592.
J M9904811