The self-association of the drug acemetacin and its interactions and stabilization with beta-cyclodextrin in aqueous solution as inferred from NMR spectroscopy and HPLC studies.

Article abstract of DOI:10.1016/S0008-6215(02)00174-X

Strongly concentration dependent, (1)H NMR chemical shifts of the non-steroidal anti-inflammatory drug acemetacin sodium salt (sodium [[1-(4-chlorobenzoyl)-5-methoxy-2-methylindol-3-yl]acetoxy]acetate), were observed in aqueous solution. Self-titration and nOe experiments, point to a self-association model where stacking takes place via the indole portion of the drug. In addition, conformational isomerism (atropisomerism) of the anti to syn form was confirmed. Further increase of the concentration eventually led to stable chemical shifts and nearly simultaneous appearance of microcrystals. In the presence of betaCD, 1:1 inclusion complexation occurred through the p-chlorobenzoyl part of the drug, whereas with excess betaCD the indole part seemed to participate to a minor degree. The anti isomer is suggested to be involved in the inclusion process. In addition, aggregation of acemetacin was also evident, as competing with the conformational and inclusion equilibria. The present case demonstrates that many competitive processes are simultaneously active in a seemingly simple system. The measurements were strongly dependent upon the pH and use of buffered solutions was mandatory. Finally, for the quantitative analysis of acemetacin in the presence of betaCD, a special HPLC method was developed. The stability of the drug, studied by the identification of the degradation products and the pseudo-first order rate of hydrolysis, was found to be unaffected by the presence of betaCD.

Full text of DOI:10.1016/S0008-6215(02)00174-X

Carbohydrate Research 337 (2002) 1387–1395
www.elsevier.com/locate/carres
The self-association of the drug acemetacin and its interactions
and stabilization with b-cyclodextrin in aqueous solution as
inferred from NMR spectroscopy and HPLC studies
Dimitris Zouvelekis,a Konstantina Yannakopoulou,a,* Irene M. Mavridis,a
Ekaterini Antoniadou-Vyzab
aInstitute of Physical Chemistry, NCSR ‘‘Demokritos’’ Aghia Paraske6i, Gr-15310, Athens, Greece
bDepartment of Pharmacy, Uni6ersity of Athens, Zografou, Gr-15771, Athens, Greece
Received 6 May 2002; accepted 19 June 2002
Abstract
Strongly concentration dependent, 1H NMR chemical shifts of the non-steroidal anti-inflammatory drug acemetacin sodium salt
(sodium {[1-(4-chlorobenzoyl)-5-methoxy-2-methylindol-3-yl]acetoxy}acetate), were observed in aqueous solution. Self-titration
and nOe experiments, point to a self-association model where stacking takes place via the indole portion of the drug. In addition,
conformational isomerism (atropisomerism) of the anti to syn form was confirmed. Further increase of the concentration
eventually led to stable chemical shifts and nearly simultaneous appearance of microcrystals. In the presence of bCD, 1:1 inclusion
complexation occurred through the p-chlorobenzoyl part of the drug, whereas with excess bCD the indole part seemed to
participate to a minor degree. The anti isomer is suggested to be involved in the inclusion process. In addition, aggregation of
acemetacin was also evident, as competing with the conformational and inclusion equilibria. The present case demonstrates that
many competitive processes are simultaneously active in a seemingly simple system. The measurements were strongly dependent
upon the pH and use of buffered solutions was mandatory. Finally, for the quantitative analysis of acemetacin in the presence of
bCD, a special HPLC method was developed. The stability of the drug, studied by the identification of the degradation products
and the pseudo-first order rate of hydrolysis, was found to be unaffected by the presence of bCD. © 2002 Elsevier Science Ltd.
All rights reserved.
Keywords: b-Cyclodextrin; Self-association; Acemetacin; NMR; HPLC; Multiple equilibria
1. Introduction
their cavity, thus frequently resulting in greater stabi-
lization, increased aqueous solubility, controlled release
and other desirable properties of the included com-
pounds.2 No reference is found in the literature regard-
ing the interaction/inclusion of acemetacin with
cyclodextrins, whereas numerous studies regarding
preparation, stabilization and potential pharmaceutical
applications of the corresponding inclusion complex
with indomethacin have appeared,3 only a few4 dealing
with the structural characterization of the complex in
solution. We have been interested in the complexation
of acemetacin with bCD, given the practical importance
of formulating NSAIDs with cyclodextrins,5 as reflected
by the growing number of patent applications5c on this
subject. We have focused on the structure and degrada-
tion pathways of the bCD/acemetacin complex, since
Acemetacin
({[1-(4-chlorobenzoyl)-5-methoxy-2-
methylindol-3-yl]acetoxy}acetic acid (1)) is a non-
steroidal anti-inflammatory drug (NSAID), whose
biological action is a consequence of its in situ hydroly-
sis to indomethacin ({[1-(4-chlorobenzoyl)-5-methoxy-
2-methylindol-3-yl]acetic acid (2)), also
a known
NSAID, that is, 1 is a pro-drug of 21 (Scheme 1).
Cyclodextrins (Scheme 2) are a family of cyclic
oligosaccharides able to include organic molecules in
* Corresponding author. Tel.: +3010-6503796; fax: +
3010-6511766
E-mail address: dyanna@chem.demokritos.gr (K. Yan-
nakopoulou).
0008-6215/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(02)00174-X

1388
D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
nucleation, leading to formation of crystals, was
recognized.
The present results can be regarded as clarifying and
complementing previous studies on the related complex
of indomethacin (2) with bCD. In the most recent
report,4a the subject of structure of the possible com-
plexes was comprehensively treated. However in gen-
eral,4 experiments regarding stoichiometry have been
partly performed, the effect of pH has not been ad-
dressed, neither has the self-association of the drug
itself been investigated, a factor known6 to alter the
mole-ratio and continuous variation curves signifi-
cantly, as we also demonstrate in the current case.
Scheme 1. Numbering scheme and conformational isomerisa-
tion of acemethacin (1) and indomethacin (2).
2. Results and discussion
1
NMR studies.—During preliminary H NMR experi-
ments with acemetacin sodium salt—the free acid was
very insoluble in water—it became evident that there
existed a strong dependence of the drug’s chemical
shifts on its concentration. Therefore, we had to exam-
ine this process by separate experiments.
Scheme 2. Side (left) and top (middle) view of bCD and of the
repeating unit (right) with numbering.
Concentration dependence of chemical shifts of
acemetacin in D2O. Self-titration of the sodium salt of 1
from 0.36 to 9.75 mM revealed a pronounced shielding
for all protons of the molecule. In order to ensure that
acemetacin sodium would remain ionized throughout
the experiment, we used a solution buffered at pH 8.2.
The shape of the titration curves in Fig. 1 shows that
the chemical shifts of all protons decrease uniformly
upon increasing the concentration of 1. The aromatic
proton H-7 is the most shielded and the aliphatic
methylene protons H-12 are the least shielded. The
curves can be grouped into two categories, one of the
most affected protons (H-7, H-6, H-16, H-18 and H-15,
H-19) and one of the least affected protons (H-4 and
aliphatic Me, MeO, H-11, and H-12). The experimental
curves do not have the typical hyperbolic shape of
binding, as systems with p–p aromatic stacking7 but
these are the crucial parameters towards sustained re-
lease and optimized bioavailability formulations of the
drug. The present article presents a complete study on
the stoichiometry, the association constant, the effect of
pH, and the structural characterization of the inter-
molecular interactions of the sodium salt of 1 with bCD
in aqueous solution, using NMR spectroscopy. The
chemical stability of the complex in solution, on the
other hand, is evaluated using HPLC and the rates of
hydrolysis at several temperatures are measured. For
this reason, a specific validated and stability indicating
HPLC method for the determination of 1 in the pres-
ence of bCD has been developed. In the course of the
NMR experiments, the enormous dependence of the
chemical shifts of 1 on its concentration, and the in situ
Fig. 1. Chemical shift displacements (DlH=linitial−lfinal) of acemetacin sodium salt during self-titration at pH 8.2, 300 K (500
MHz) (left) and fitting to a sigmoidal-type equation (right), see text.

D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
1389
The very strong shielding effect on H-7, H-6 and
H-16, H-18, H-15, H-19 (the top group of curves)
indicates changes involving the aromatic rings (indole
and p-chlorobenzoyl). This could be attributed to the
continuous shifting of a conformational equilibrium
(Scheme 1), such as that reported for indomethacin,4a
towards the syn conformer1 (the indole ring becoming
nearly perpendicular to p-chlorobenzoyl ring to maxi-
mize =CꢀH···p interactions, thus resulting in mutual
shielding), although this motion is not expected to
affect all protons in the same way. After a critical
concentration, the syn conformer aggregates fast in a
fashion reminiscent of crystal packing, and produces
the first crystals. The ongoing conformational equi-
librium in solution was detected through intermolecular
interactions, as shown next.
The structure of the aggregated species in the solu-
tion was investigated by 2D ROESY spectra acquired
under conditions identical to the corresponding spectra
of 1 with bCD (to be shown later) and at a concentra-
tion corresponding to the beginning of the plateau in
Fig. 1. The spectra obtained mapped various interac-
tions among the protons (Fig. 2). First of all interac-
tions between the p-chlorobenzoyl protons (H-15,
H-19, but also H-16, H-18) with both the methyl group
Me-10 (Fig. 2) and H-7 (not included) demonstrate the
(anticipated) coexistence of the fast exchanging syn and
anti conformers in solution. Further, the methylene
protons of C-12 correlate with H-4, weakly with H-15,
H-19 and H-16, H-18, but also with the methyl group
at C-10 (not included in Fig. 2) indicating a motion of
the side chain in space. What cannot really be at-
tributed to the intramolecular proximity of the protons
but to the formation of aggregates, is the observed
correlation of the methoxy group with both sets of
p-chlorobenzoyl ring protons. In addition, correlation
was observed between the protons of Me-10 with dia-
metrically opposite aromatic protons (H-7, H-6, H-4)
of the indole ring. It seems, therefore, that the aggrega-
tion involves a head-to-tail stacking of two indole rings
belonging to adjacent molecules. Towards this model,
also point the detectable correlations of MeO with
Me-10 and with H-12 (not in Fig. 2). The above are in
agreement with what was observed in the crystal struc-
ture of indomethacin,12 in which the indole plane was
found almost perpendicular to the p-chlorobenzoyl ring
of the same molecule, and overlapping with the acetic
acid side-chain of an inverted indole ring of another
molecule, all in the syn configuration (Scheme 1), this
isomer being both the more stable4a,12 and the bioactive
one.13 Unfortunately, we were unable to obtain crystals
Fig. 2. Partial ROESY spectrum of acemethacin sodium salt
(1.8 mM, D2O, pH 8.2, 300 K, 500 MHz).
rather a sigmoidal shape, with a slowly growing part up
to about 2 mM (ꢀsixfold concentration increase) fol-
lowed by a fast rise and a turning point at about 6 mM.
The plateau following that point has a non-zero slope.
Increasing difficulty was encountered during collection
of the last data points due to the appearance of tiny
crystals in the NMR tube upon standing, which had to
be dissolved by vigorous shaking before acquisition of
the next spectrum. We, therefore, suggest that the
molecule participates in a self-aggregation process.
Upon increasing the concentration, a cooperative
event8,9 sets on, which results not only in dimers but in
higher aggregates and finally in crystals. In other
words, during the self-titration we are actually witness-
ing the nucleation and crystallization process. A similar
shape of the titration curves has also been reported for
the self-aggregation of chlorpromazine10 and some
naphthalene bis(dicarboximide)s11 both in aqueous so-
lution, probed by NMR spectroscopy. Quantitative
treatment using the entire range of the concentration
dependence data (see Eq (2) in the Experimental), with
the assumption that dimerization prevails and higher
aggregates are negligible, did not result to any reason-
able fit. This is expected since the crystals formed
during the final points of the self-titration cannot be
dimers. More reasonable fitting, however, was obtained
using the first seven data points giving the dimerization
constant, K=210922, with R2=0.986. On the other
hand, fitting of all but the last two data points to the
sigmoidal Eq (3) provided excellent fit (R2=0.996).
The data were also plotted according to the Hill
equation8 and afforded a linear graph (R2=0.961) with
slope, i.e., the ‘‘Hill coefficient’’ h=2.5, indicating pos-
itive cooperativity during the process.8
1 The two distinct conformers should be strictly referred to
as atropisomers, since the amide linkage is not coplanar with
the indole ring,12 however we maintain the above symbolism
to conform with previous articles.4a,12

1390
D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
dependence of its chemical shifts on the concentration
of the (unknown) free 1. In addition, complexation
might take place preferentially through one of the
conformers of the drug, and this would also contribute
to alterations of the chemical shifts of 1. Fig. 3 shows
the binding curves and illustrates the behavior of the
cavity protons H-5% and H-3% in the two solutions.
Thus, H-5% is shielded more strongly than H-3% upon
increasing the concentration of the guest drug. The
corresponding smooth curves (Fig. 3), show binding
that reflects a stoichiometry close to 1:1. The peaks of
the drug became broadened, from the stoichiometry 1:1
and over, probably due to the formation of aggregates
of 1. Indeed, crystallization of 1 out of the solution
occurred after addition of about 1.5 equivalents of the
drug. The curves corresponding to the unbuffered titra-
tions (pH 6.0–6.8) show a step at the mole ratio 1:1,
followed by a plateau above 2:1 and the overall dis-
placements are greater. Apparently the pH variation,
and thus the presence of a non-negligible amount of the
free acid, is sufficient to cause anomalies in the appear-
ance of the curve (by, e.g., altering the self-association
process), and therefore we regard the unbuffered curves
unreliable. Fitting of the experimental curve obtained
at pH 8.2 with the equation describing the 1:1 model18
(this stoichiometry also suggested by the Job plots,
shown below) gives a binding constant K11=1100 (9
31) M−1 and intrinsic DlH-3% 0.2012 (90.0013) ppm
with R2=0.9991 at 300 K. This K11 value is more than
four times higher that the roughly calculated dimeriza-
tion constant reported above, meaning that complexa-
tion with bCD prevails, until the concentration of 1
increases (past the 1:1 point), and then the aggregation
process manifests itself again, in agreement with the
spectral observations.
of good quality for X-ray analysis to verify the struc-
ture of the proposed ‘‘pre-crystalline’’ state.
The strong concentration dependence of the chemical
shifts of a substance has been reported for several types
of small heterocyclic molecules such as quinolines, in-
doles, benzofuranes, isoquinolines and acridine in vari-
ous
organic
solvents,14
and
of
conjugated
phenylacetylene macrocycles7 in chloroform. In
aqueous solutions, there are also examples of self-ag-
gregation processes evaluated through NMR chemical
shift measurements, such as the pre-micellar organiza-
tion of chlorpromazine (a phenothiazine derivative),10
the self-stacking of naphthalene diimides,11 and the
color stabilizing self-aggregation of the flavylium
cation.15 The present report of the similar behavior of a
commonly used drug enforces the notion that self-asso-
ciation of small organic molecules in aqueous solutions
at sub-millimolar concentrations is a widely-spread phe-
nomenon that should be taken into account when
trying to assess intermolecular interactions, evaluate
binding constants in host–guest systems or, in general,
use chemical shifts as a measure of molecular recogni-
tion phenomena. On the other hand, the formation of
aggregates of acemetacin, and very likely many other
NSAIDS (e.g., indomethacin) in aqueous solution,
could also affect their physiological behavior, especially
in cases where the latter is concentration dependent.16
The above observations prompted us to further study
the complexation of 1 with bCD, and allowed to ex-
plain the unusual curves obtained during Job plots.
Molar ratio titrations. In order to examine the inclu-
sion of the acemetacin molecule inside the bCD host,
titration of a bCD solution with solid acemetacin
sodium salt was carried out in both buffered ([bCD]=
3.1 mM, pH 8.2) and unbuffered ([bCD]=3.0 mM)
solutions and the chemical shift displacement of the
bCD cavity protons H-5% and H-3% (Scheme 2) was
monitored. In the buffered solution, only the anionic
form of 1 is present during the process, given that the
pKa of acemetacin is 3.91 at 20 °C.17 The reverse exper-
iment (titration of an aqueous solution of 1 sodium
salt) with solid bCD and monitoring of the chemical
shift alterations of 1) would be unreliable given the
serious aggregation of acemetacin and the resulting
An ESI-mass spectrum of the unbuffered solution
showed
a weak molecular ion corresponding to
([bCD·acemetacinNa]+Na)+ at m/e 1596.7. A fairly
strong peak (13%) corresponding to ([acemetacin·
acemetacin]+Na)+ at m/e 853 indicated the strong
self-association of the guest.
Job plots. Continuous variation (Job) plots9 were
calculated for buffered and unbuffered drug/bCD mix-
tures are shown in Fig. 4. The curves that are produced
1
Fig. 3. Titration H NMR data of bCD (3.1 mM) aqueous solutions with solid acemetacin sodium salt, where DlH%=lfree
lcomplex (bCD protons). For unbuffered D2O solutions, pH 6.0–6.8; for buffered D2O solutions, pH 8.2.
−

D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
1391
Fig. 4. Job plots of unbuffered (left) and buffered (pH 8.2, right) D2O solutions of bCD/acemetacin sodium salt.
primarily intramolecular nOes, and weak dipolar corre-
lation with H-5% and probably with H-3% of the host.
These weak interactions suggest that there could be a
second 1:1 complex present in the solution involving
inclusion of the indolyl part, since there is an excess of
bCD available. It is worth noting that there is only one
intramolecular nOe peak present between Me-10 and
H-15, H-19 whereas in Fig. 2 (acemetacin alone) all
by monitoring the bCD cavity protons, in either
buffered or unbuffered solutions provide 1:1 stoi-
chiometry. The protons of acemetacin, however, behave
inhomogeneously in both cases. These anomalies in the
chemical shifts of 1 arise from a combination of
changes due to de-aggregation, inclusion and possibly
conformational syn to anti isomerism, therefore no
conclusions can be drawn from the curves of 1. How-
ever, one observes once more that without buffers the
results can be erroneous, by comparing for example, the
behavior of H-6 and H-7 in the buffered and un-
buffered solutions.
Intermolecular nOe data. The results presented so far
demonstrate that inclusion does take place in solution,
that the complex formed is most probably 1:1, and that
it is not clear which part (indolyl- or p-chlorobenzoyl)
is inserted in the cavity and how. This information was
derived from 2D ROESY spectra of the complex. The
intermolecular interactions were mapped with this ex-
periment at different bCD/acemetacin mole ratios. At
mole ratio 3:1 (excess bCD, Fig. 5), we observe very
strong interactions of both sets of the p-chlorobenzoyl
protons with H-3% and H-5% of bCD, indicating inser-
tion of this part into the cavity. On closer inspection,
we see that this is strictly true for H-15, H-19 whereas
H-16, H-18 show a large cross peak that culminates at
(H-6, H-6%)% of the primary side of bCD, suggesting
deep insertion of the p-chlorobenzoyl part from the
secondary side of the host. The indolyl protons, show
Fig. 5. Partial ROESY spectrum of bCD (6.6 mM) with
acemetacin sodium salt (2.2 mM), D2O, pH 8.2, 300 K, 500
MHz.

1392
D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
(not shown) and the appearance of the drug’s spectrum
change. The resonance of H-7 moves close to that of
H-4, the complexation of the p-chlorobenzoyl ring with
bCD is strong but the respective cross peaks of the
indolyl protons are diminished, and only H-7 shows
interactions with H-3% and H-5%. This could be due to
either minimization of the second complex mentioned
above or possibly interaction of the H-7 with the
p-chlorobenzoyl ring-filled bCD cavity due to this pro-
ton’s proximity with the inserted part. The second
scenario does not justify the (weak) interaction H-7/H-
5%. At even higher concentrations of 1 and at mole ratio
1:3, there is broadening of the aromatic peaks and the
resonances of H-4 and H-7 come together (Fig. 6). The
intramolecular correlations of Me-10 with the aromatic
protons appear now again as in Fig. 2. The aggregation
of the drug is apparently effective again and competes
with the inclusion process resulting into a broadening
of the peaks of 1, not observed in Fig. 2. The bCD
complex is prominent as the strong cross-peaks p-
chlorobenzoyl/H-3%, H-5% show, whereas H-7 interacts
weakly with bCD (less than shown in Fig. 5).
Fig. 6. Partial ROESY spectrum of bCD (3.0 mM) with
acemetacin sodium salt (6.0 mM) D2O, pH 8.2, 300 K, 500
MHz.
The intermolecular interactions, therefore, have
shown that in the solution there exist at least four
different equilibria, i.e., conformer exchange, aggrega-
tion of 1, and two inclusion processes of which the
predominant is bCD including the p-chlorobenzoyl
group (Scheme 3). The percentage of contribution the
second complex (indolyl group included) is small, it
depends on the mole ratio of the components and can
be considered negligible at or below 1:1 ratio [bCD]/[1].
aromatic protons show interactions with Me-10 and in
spite the similar concentration of 1 in the two spectra.
This in turn indicates that complexation with bCD
breaks up the aggregates of the drug and that there is
appreciable presence of the anti-type which is, for steric
reasons, expected to complex preferentially with bCD.
As the mole ratio of the two complexing partners
approaches the value 1:1, the dipolar correlation map
Scheme 3. Equilibria in the aqueous solution of bCD/acemetacin.

D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
1393
verse phase column with 50:5:45 methanol–iso-
propanol–phosphate buffer (pH 4.0) as the mobile
phase, a flow rate of 1.35 mL/min at 35 °C and the UV
detector set at 238 nm. These conditions gave satisfac-
tory results regarding the accuracy of determination of
1 and also the retention times of the potential degrada-
tion products of 1,20 indomethacin and p-chlorobenzoic
acid, in the presence of bCD. The method developed
was checked for RSD (1.10%), precision (98.5%) and
repeatability for the entire range of its linear region
(0.45–4.5 mg/L), all of which were satisfactory. In
addition, the method was characterized by a very good
sensitivity, and a low detection limit (0.18 mg/L) and
quantitation limit (0.54 mg/L).
The stability of a 0.0106 mM solution of acemetacin
(1) in the presence of bCD was studied under pseudo-
first order reaction conditions (large excess of bCD) at
pH 8.0, and the first-order reaction constant for the
degradation of 1 was measured at different tempera-
tures (Table 1) and compared with the results obtained
without bCD. The table shows that there exists a
satisfactory linearity among the calculated values of k
in both cases (R2=0.9965 for 1 and 0.9981 for bCD/1)
and that the presence of bCD does not alter the rate of
hydrolysis of acemetacin. That is, despite the docu-
mented inclusion that takes place at this pH (K11=
1100 M−1) and with excess bCD (which ensures high
concentration of complex), there is neither stabilization
offered by the host, nor acceleration of its degradation.
The hydrolysis products determined chromatographi-
cally (Fig. 8) were identified using authentic samples, by
NMR spectroscopy and by comparing the results with
the hydrolysis profile of pure indomethacin. The prod-
ucts were 5-methoxy-2-methylindol-3-yl acetic acid (re-
tention time 0.56 min), 5-methoxy-2-methylindol-3-yl
acetic acid carboxymethyl ester (retention time 0.90
min), p-chlorobenzoic acid (retention time 1.15 min),
and indomethacin (2) (retention time 4.35 min). The
above show that hydrolysis affects two functions with
apparently different rates: (i) of the ester bond and (ii)
of the pseudo-amide bond, as previously observed for
some other esters of indomethacin.21
Fig. 7. Non-proportional variation in the UV absorption
(top), and in the HPLC peak area determined using the
UV-detector (bottom) of acemetacin (1) in the presence of
increasing amounts of bCD.
Table 1
First order rate constant (k, s−1) for the hydrolysis of
acemetacin alone and in the presence of bCD at various
temperatures
Temperature (K)
333
3.33
3.50
323
1.33
1.33
308
0.20
0.21
kacemetacin (×10−5
)
kbCD/acemetacin (×10−5
)
Stability studies.—The complex bCD/acemetacin was
studied, regarding its stability, at different temperatures
and under mildly alkaline conditions (pH 8), very close
to that used in the structural studies above. The con-
centration of the drug as well as its products of hydrol-
ysis were determined using HPLC. For these
quantitative measurements, a special stability indicating
and validated chromatographic method was developed.
The UV spectrum of any guest is affected by complexa-
tion with cyclodextrins,19 therefore the accuracy of the
chromatographic determination (using a UV detector
for the HPLC) can be seriously hampered (Fig. 7). For
this reason, conditions suitable for complete de-com-
plexation of 1 from bCD were explored: variations in
the composition of the mobile phase regarding the
organic modifiers, pH effects, types of reverse phase
columns (C-18, phenyl-, amino-) and effect of tempera-
ture in the range of 30–38 °C. The method finally
developed, involved the use of a C-18 Spherisorb re-
3. Conclusions
Acemetacin sodium salt forms aggregates at mildly
alkaline pH, resulting in serious alteration of its chemi-
cal shifts. Self-titration curves and nOe data indicate
that a head-to-tail stacking of the molecules through
their indole parts leads to nucleation and eventually to
crystallization of the drug. In aqueous solution and at
the same pH, bCD forms
a 1:1 complex with
acemetacin, characterized by an association constant of
1100 M−1. Inclusion in the bCD cavity takes place via
the p-chlorobenzoyl moiety, as in the case of the bCD/

1394
D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
indomethacin complex. Under conditions involving ex-
cess bCD, the indole part also forms a complex, which
exists in small concentration. Based on dipolar interac-
tion data, we suggest that the anti conformer is the one
that participates in the formation of inclusion com-
plexes, whereas the syn conformer seems to be the
dominant conformer in solution, in the absence of
bCD. Buffering of the solution for any concentration
dependence study is mandatory for accurate results.
For the quantitative determination of 1 in the presence
of bCD, the special chromatographic method developed
was used to measure the pseudo-first order rate of
degradation of the drug. In spite of the formation of a
fairly strong complex, there is no effect of the bCD host
on the rate of hydrolysis of acemetacin.
chromatography oven (UK) and a Spherisorb S3
ODS2, 4.6 mm×10 cm column. The flow rate was
maintained at 1.35 mL/min and the pressure was 400
psi.
NMR studies.—All samples were allowed to equili-
brate at 300 K for at least 15 min prior to acquisition.
The residual HDO peak of the solvent was used as the
chemical shift reference for the spectra. For the self-ti-
tration studies, the initial concentration of 1 was 0.36
mM and the final 9.75 mM. For the mole ratio studies,
the concentration of bCD was held constant at 3.1 mM
and acemetacin was added in portions up to 14 mM.
Fitting of the mole ratio data points to the theoretical
curve derived for 1:1 complex under fast exchange
conditions17 was carried out using the routine provided
by the program GRAPHPAD PRISM. For the continuous
variation plots, a solution 4.0 mM in 1 and another 4.0
mM in bCD were mixed at different proportions and
the chemical shift displacements were recorded. The
construction of the continuous variation and mole ratio
plots required short 2D COSY experiments (gradient
version) for most of the data points (determination of
chemical shift displacements for overlapping reso-
nances). 2D ROESY spectra were acquired with 1–2 K
size in F2 with 256 experiments, d1=0.8–1 s and
mixing time 350 ms at transmitter attenuation 30 dB.
Processing was carried out using zero-filling in F1 and
the sine-bell squared window function in both
dimensions.
4. Experimental
General.—Acemetacin (Sigma) was converted to its
sodium salt by treatment with 1 equiv of NaOH in
water. The product was dried under diminished pres-
sure and then dissolved in buffered (K2HPO4–
NaH2PO4) D2O (pH 8.2). The HPLC was carried out
using HPLC grade methanol and isopropanol (E.
Merck). Deionized water was purified by a MilliQ
(Millipore, USA) unit. The NMR spectra were acquired
on Bruker instruments AC 250 and DRX 500 MHz.
The ESI mass spectrum was run on an AQA Navigator
Finnigan-Thermoquest instrument. HPLC quantita-
tions were carried out on a Waters LC instrument
consisting of a model 590 pump, a Waters 996 photodi-
ode array UV detector, and a Reodyne 717S injector
fitted with a 20 L loop. The degassed mobile phase of
50:5:45 MeOH–isopropanol–phosphate buffer was
vacuum filtered through a 0.45 m pore-size nylon mem-
brane filter (Millipore, USA). All runs were performed
at constant temperature 35 °C using a Jones column
1H NMR data analysis of self-titration of acemetacin
was carried out using the equation:7
(c/2[m2])+(2[m2]/c)−2=1/2cK
(1)
where c=total concentration of acemetacin sodium
salt, [m2]=the concentration of the dimer, and K=the
dimerization constant. The observed chemical shift,
Dlobs=f(m2)Dl° , where f(m2)=mole fraction of
m2
dimer, can be expressed as Dlobs=2[m2]Dl° /c, there-
m2
fore Eq. (1) becomes
Fig. 8. Chromatogram of 1 and its products of hydrolysis at pH 8, 333 K; detector wavelength 254 nm.

D. Zou6elekis et al. / Carbohydrate Research 337 (2002) 1387–1395
1395
2
DlobsDl°
m2
4. (a) Fronza, G.; Mele, A.; Rendeti, E.; Ventura, P. J. Org.
Chem. 1996, 61, 909–914;
c=
+
(2)
2
2cK (Dl° −Dlobs
)
m2
(b) Djedaini, F.; Lin, S. Z.; Perly, B.; Wouessidjewe, D. J.
Pharm. Sci. 1990, 79, 643–646.
which utilizes the experimental variables c and Dlobs
,
and provides after fitting the values K and Dl° which
m2
5. (a) See special issue: Cyclodextrins in Drug Delivery
Systems; Ad6. Drug Deli6. Re6. 1999, 36, 1;
(b) see special issue: S.T.P. Pharm. Sci. 1999, 9, 225;
(c) Benes, E.; Szejtli, J. Proc. Int. Symp. Cyclodextrins
9th; Santiago de Compostela, Spain, 1998.
6. (a) Djeda¨ıni, F.; Perly, B. Nuclear Magnetic Resonance
of Cyclodextrins, Derivatives, and Inclusion Compounds.
In New trends in Cyclodextrins and Deri6ati6es; Ducheˆne,
D., Ed.; Editions de Sante´: Paris, France, 1991;
(b) Djeda¨ıni, F.; Perly, B. Minutes, Proc. Int. Symp.
Cyclodextrins 5th; Paris, France, 1990; pp. 130;
(c) Eliadou, K.; Yannakopoulou, K.; Rontoyianni, A;
Mavridis, I. M. J. Org. Chem. 1999, 64, 6217–6226.
7. Shetty, A. S.; Zhang, J.; Moore, J. S. J. Am. Chem. Soc.
1996, 118, 1019–1027.
8. Petter, R. C.; Salek, J. S.; Sikorski, C. T.; Kumaravel, G.;
Lin, F.-T. J. Am. Chem. Soc. 1990, 112, 3860–3868.
9. Connors, K. A. Binding Constants; Wiley: New York,
1987.
10. Attwood, D.; Waigh, R.; Blundell, R.; Bloor, D.; The-
vand, A.; Boitard, E.; Dubes, J.-P.; Tachoire, H. Magn.
Reson. Chem. 1996, 32, 468–472.
the limiting chemical shift of the dimer.7
The equation
(Dlobs,final−Dlobs,initial
1+exp[(c%−c)/slope]
)
Dlobs=Dlobs,initial
+
(3)
where c%=the concentration half-way in the titration,
was utilized for the fitting of the self-titration data, and
is a suitably corrected built-in equation for sigmoidal
response in the GRAPHPAD PRISM software.
Stability studies.—Reference solutions of acemetacin,
indomethacin, and p-chlorobenzoic acid were prepared
for quantification of hydrolysis at pH 8. The kinetic
studies were performed at 35, 50 and 60 °C using a
thermostated oven. The samples were analyzed after
careful neutralization using 1 N HCl and vortexed for
10 s.
11. Steullet, V.; Dixon, D. W. J. Chem. Soc., Perkin Trans. 2
1999, 1547–1558.
12. Kistenmacher, T. J.; Marsh, R. E. J. Am. Chem. Soc.
1972, 94, 1340–1345.
13. Gund, P.; Shen, T. Y. J. Med. Chem. 1977, 20, 1146–
1152.
Acknowledgements
We thank Dr L. Leondiadis, NCSR ‘‘Demokritos’’,
for obtaining the ESI-MS.
14. Mitra, A.; Seaton, P. J.; Assarpour, R. A.; Williamson,
T. Tetrahedron 1998, 54, 15489–15498.
15. Houbiers, C.; Lima, J. C.; Macanita, A. L.; Santos, H. J.
Phys. Chem. B 1998, 102, 3578–3585.
References
16. Duffy, C. P.; Elliot, C. J.; O’Connor, R. A.; Heenan, M.
M.; Coyle, S.; Cleary, I. M.; Kavanagh, K.; Verhaegen,
S.; O’Loughlin, C. M.; NicAmhlaoibh, R.; Clynes, M.
Eur. J. Cancer 1998, 34, 1250–1259.
17. Burger, A.; Lettenbichler, A. Pharmazie 1993, 48, 262–
272.
18. Botsi, A.; Yannakopoulou, K.; Perly, B.; Hadjoudis, E. J.
Org. Chem. 1995, 60, 4017–4023.
19. Rozou, S.; Antoniadou-Vyza, E. J. Pharm. Biomed. Anal.
1998, 18, 899–905.
1. Matsuki, Y.; Ito, T.; Kojima, M.; Katsumura, H.; Ono,
H. Chem. Pharm. Bull. 1983, 31, 2033–2038.
2. Szejtli, J. Cyclodextrin Technology; Kluwer Academic:
Dordrecht, 1988.
3. (a) Szejtli, J.; Szente, L. Pharmazie 1981, 36, 694–698;
(b) Hoshino, T.; Tagawa, T.; Hirayama, F.; Otagiri, M.;
Uekama, K. Yakagaru Zasshi 1982, 102, 1184–1190;
(c) Szente, L.; Apostol, I.; Gerloczy, A.; Szejtli, J. Phar-
mazie 1985, 40, 406–407;
(d) Tarimci, N.; Celebi, N. Pharmazie 1988, 43, 323–325;
(e) Croft, A. K.; Easton, C. J.; Lincoln, S. M.; May, M.
L.; Papageorgiou, J. Aust. J. Chem. 1997, 50, 857–860;
(f) Fini, A.; Fernandez-Hervas, M. J.; Holgado, M. A.;
Rodriguez, L.; Cavallari, C.; Passerini, N.; Caputo, O. J.
Pharm. Sci. 1997, 86, 1303–1309;
20. (a) Notarianni, L. J.; Collins, A. J. J. Chromatogr. 1987,
413, 305–308;
(b) Scho¨llnhammer, G.; Dell, H. D.; Doersing, K.;
Kamp, R J. Chromatogr. 1986, 375, 331–338;
(c) Battista, H. J.; Wehinger, G.; Henn, R. J. Chro-
matogr. 1985, 345, 77–89.
(g) Ueda, H.; Ou, D.; Endo, T.; Nagase, H.; Tomono, K.;
Nagai, T. Drug De6. Ind. Pharm. 1998, 27, 863–867.
21. Kahn, A. H.; Jensen, P. B.; Mørk, N.; Bundgaard, H.
Acta Pharm. Nord. 1989, 1, 327–336.