Electrophoretic behavior and pKa determination of quinolones with a piperazinyl substituent by capillary zone electrophoresis

Electrophoretic behavior and pK

a

determination of quinolones with a

piperazinyl substituent by capillary zone electrophoresis

Ching-Erh Lin

, Yan-Jr Deng

, Wei-Ssu Liao

, Shao-Wen Sun

,

Wann-Yin Lin

, Chia-Chong Chen

a_{Department of Chemistry, National Taiwan University, Taipei, Taiwan} b_{School of Pharmacy, National Taiwan University, Taipei, Taiwan}

Abstract

Electrophoretic behavior and pKadetermination of six quinolones with a piperazinyl substituent, together with two quinolones without a

piperazinyl substituent and 1-phenylpiperazine, were investigated by capillary zone electrophoresis. The results indicate that quinolones with a piperazinyl substituent involve three protonation/deprotonation equilibria. The results also suggest that the contribution of the zwitterionic species of these quinolones to the effective mobility may not be neglected. This is probably due to a slightly incomplete protonation of the piperazinyl moiety in the pH range of 6.0–8.0, compared with the complete dissociation of the carboxylic group. Consequently, the zwitterionic species of ciprofloxacin, in particular, is slightly negatively charged. With the aid of computer simulation, three pKavalues were determined

for quinolones with a piperazinyl substituent, thus allowing us to rationalize precisely the influence of pH on the electrophoretic behavior of these compounds.

© 2004 Elsevier B.V. All rights reserved.

Keywords: pKadetermination; Quinolones; Antibiotics; Capillary zone electrophoresis

1. Introduction

Quinolones are used extensively as either antibacterial agents or antibiotics in both human and veterinary practices. As their antibacterial activity is pH-dependent and acid dis-sociation constants can be a key parameter for understanding chemical phenomena such as biological activity, biological transports, and drug delivery[1,2], a good knowledge of the protonation equilibria of quinolines is essential for a better understanding of their activity.

In recent years, the influence of pH on the electrophoretic behavior of quinolones in water and hydro–organic media has been examined by capillary electrophoresis (CE)[3–8], in which a single protonation equilibrium was considered for the piperazinyl substituent of those quinolones. As two pro-tonation/deprotonation equilibria are involving in the piper-azinyl substituent of these compounds[9,10], two pKavalues

∗_{Corresponding author. Tel.: +886 2 23635357; fax: +886 2 23636359.}

E-mail address: celin@ntu.edu.tw (C.-E. Lin).

are expected to associate with the piperazinyl substituent. Therefore, it is thought that the protonation/deprotonation equilibria of those compounds are not accurately described in the literature. In view of the zwitterionic species of p-hydroxyphenylalanine being slightly negatively charged at pH 7.0 because of the incomplete compensation of the neg-ative charge of the carboxylic group by the positive charge of the protonated amino group [11], it is of interest to find out whether the net charge of the zwitterionic species of quinolones with a piperazinyl moiety deviates from zero, be-cause the dissociation of the carboxylic group may compar-atively be stronger than the protonation of the piperazinyl moiety in the pH range 6.0–8.0.

Capillary zone electrophoresis (CZE) has proven to be a convenient and useful method for precise pKadetermination [2,12–23]. This method is based on the measurement of the electrophoretic mobility of the solute as a function of buffer pH and the analysis of the mobility curve using a mobility equation which describes appropriately the migration behav-ior of the solute in the pH range studied.

0021-9673/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2004.08.069

In this study, the influences on the electrophoretic behav-ior of six quinolones with a piperazinyl substituent, together with two quinolones without a piperazinyl substituent and 1-phenylpiperazine, were investigated and their pKa values

were determined by CZE. The mobility equations which may properly describe the electrophoretic behavior of quinolones with a piperazinyl substituent were examined and accord-ingly, the pKa values of six quinolones studied were

deter-mined. It is hoped that the protonation/deprotonation equilib-ria involving in these compounds can be better understood.

2. Experimental 2.1. Apparatus

All CE separations were performed on a Beckman P/ACE System MDQ equipped with a photodiode array detector for absorbance measurements at 214 nm (Beckman Coulter, Fullerton, CA, USA). Uncoated fused-silica capillary pur-chased from Polymicro Technologies (Phoenix, AZ, USA) was used. The dimensions of the capillary were 50.2 cm× 50␮m i.d. The effective length of the capillary was 40 cm. The CE system was interfaced with a microcomputer and a laser printer. System Gold software of Beckman was used for data acquisition. For pH measurements, a pH meter (Suntex Model SP-701, Taipei, Taiwan) with a precision of±0.01 pH unit was used.

2.2. Chemicals and reagents

Ofloxacin, norfloxacin, enoxacin, and flumequine were obtained from Sigma (St. Louis, MO, USA). Nalidixic acid and 1-phenylpiperazine were purchased from Aldrich–Sigma (St. Louis, MO, USA). Lomefloxacin, ciprofloxacin, and pipemidic acid were obtained from local pharmaceutical sup-pliers. All other chemicals were of analytical reagent grade from various suppliers. Deionized water was prepared with a Milli-Q system (Millipore, Bedford, MA, USA).

Standard solutions of quinolones and 1-phenylpiperazine at a concentration of 10␮g/mL were prepared by dissolving analytes in an aqueous solution containing 10% (v/v) ethanol. The pH of a phosphate buffer was adjusted to the desired pH value by monitoring the pH of the solution with a pH meter while mixing various proportions of 50 mM trisodiumphos-phate solution with the same concentration of phosphoric acid (50 mM). All buffer solutions, freshly prepared weekly and stored in a refrigerator before use, were filtered through a membrane filter (0.22␮m).

2.3. Electrophoretic procedure and operating conditions

When a new capillary was used, the capillary was washed 30 min with 1.0 M NaOH solution, followed by 30 min with deionized water at 25◦C. Before each injection, the capil-lary was prewashed for 3 min with running buffer. After each

injection, the capillary was postwashed for 3 min with deion-ized water, 3 min with 0.1 M NaOH, and 5 min with deiondeion-ized water to maintain proper reproducibility of run-to-run injec-tions. Sample injections were done in a hydrodynamic mode over 5 s under a pressure of 1.0 psi at 25◦C. The measure-ments were run at least in triplicate to ensure reproducibility. The detection wavelength was set at 214 nm and a voltage of 20 kV was applied. Peak identification was conducted by spiking with the analyte to be identified. Ethanol was used as a neutral marker. The relative standard deviation of migration time is less than 0.6% (n = 5).

2.4. Mobility calculations

The electrophoretic mobility of analytes was calculated from the observed migration times with the equation:

µep= µ − µeo=LdLt V
1 tm− 1 teo  (1) whereµepis the electrophoretic mobility of the analyte tested,

µ is the apparent mobility of each quinolone, µeois the elec-troosmotic mobility, tm is the migration time measured

di-rectly from the electropherogram, teo is the migration time

for an unchanged solute, Lt is the total length of capillary,

Ldis the length of capillary between injection and detection,

and V is the applied voltage.

3. Results and discussion

3.1. Influence of pH on electrophoretic behavior

Fig. 1depicts the structures of eight quinolones studied (six with a piperazinyl moiety and two without the piper-azinyl substituent). Fig. 2shows the variation of the elec-trophoretic mobility of six quinolones with a piperazinyl moiety, together with flumequine, nalidixic acid, and 1-phenylpiperazine, as a function of buffer pH in the range 3.0–11.7. For clarity, the variations of the electrophoretic mo-bility of six quinolones with a piperazinyl substituent, i.e., ofloxacin and ciprofloxacin, enoxacin and norfloxacin, and lomefloxacin and pipemidic acid, are shown inFig. 2A, B and C, respectively. Depending on the involvement of one, two, or three protonation/deprotonation equilibria, the mo-bility curves of quinolones without a piperazinyl group, 1-phenylpiperazine, and quinolones with a piperazinyl moiety exhibit different shapes. The sigmoidal shape of the mobility curves with negative values of the electrophoretic mobility were observed for flumequine and nalidixic acid, because these two compounds possess only one ionizable group. The shape of the mobility curve of 1-phenylpiperazine clearly dicates that two protonation/deprotonation equilibria are in-volved. Thus, in addition to a singly protonated species, there should exist a doubly protonated species. Quinolones with a piperazinyl substituent exhibit their substituent curves in a much complicated shape because these compounds possess

Fig. 1. Structures of the eight quinolones studied.

Fig. 2. Variations of the electrophoretic mobility of eight quinolones and 1-phenylpiperazine as a function of buffer pH in the range 3.0–11.7 using a phosphate buffer (50 mM). Capillary, 50.2 cm× 50 ␮m i.d; sample concentration, 10 ␮g/mL; detection wavelength, 214 nm. Curve identification: 1, ofloxacin ( ); 2, enoxacin (
); 3, norfloxacin (); 4, lomefloxacin (); 5, ciprofloxacin (䊉); 6, pipemidic acid (); 7, flumequine (); 8, nalidixic acid (); 9, 1-phenylpiperazine ().

a piperazinyl substituent and a carboxylic group. Three pro-tonation/deprotonation equilibria are expected and the elec-trophoretic mobilities of these compounds span from positive to negative values. As one of the inflection point is obscurely observed, these mobility curves can only be analyzed with the aid of computer simulation based on a mobility equation involving three protonation/deprotonation equilibria which will be discussed later inSection 3.2.

The fact that the negative values of the electrophoretic mo-bility of quinolones with a piperazinyl group are relatively less negative than those of the fully dissociated species of quinolones without a piperazinyl group, such as flumequine and nalidixic acid, in the pH region 5.0–10.0 may suggest that, in addition to the dissociated carboxylic group, there exists a protonated moiety in quinolones with a piperazinyl substituent. Thus, a zweitterionic species (HB+A−) exists and a deprotonation equilibrium occurs between the species HB+A− and BA− for quinolones with a piperazinyl sub-stituent in the pH region 7.0–10.0. On the other hand, in comparison with the nil values of the electrophoretic mo-bility of a fully protonated species of quinolones without a piperazinyl substituent, such as flumequine or nalidixic acid, at pH < 3.5, that the positive values of the elec-trophoretic mobility of quinolones with a piperazinyl group is relatively much larger than the negative values (actu-ally the absolute values) of the electrophoretic mobility, at pH > 9.5, clearly indicate the existence of a doubly proto-nated species of quinolones with a piperazinyl substituent (H2BAH2+).

3.2. Determination of pKavalues

3.2.1. Mobility equation involving one protolytic equilibrium

As shown inFig. 1, flumequine and nalidixic acid possess only one ionizable functional group. Thus, due to the

disso-Table 1

The pKavalues and limiting mobility data of quinolones and piperazine derivatives evaluated according toEq. (4) Analytes Literature valuesa pKavalues Limiting mobilityb

pKa pKa3 pKa2 pKa1 µBA µH+BA− µHBAH+ µH2BAH2+

Quinolones(with a piperazine substituent)

Ofloxacin (1) 8.11 6.05 8.20± 0.02 6.20± 0.03 5.20± 0.06 −1.45 ± 0.01 −0.27 ± 0.01 0.95± 0.02 1.86± 0.01 Enoxacin (2) 8.50 6.00 8.80± 0.02 6.25± 0.05 5.05± 0.08 −1.60 ± 0.02 −0.43 ± 0.02 1.00± 0.03 2.00± 0.02 Norfloxacin (3) 8.38 6.22 8.45± 0.03 6.25± 0.04 5.00± 0.10 −1.55 ± 0.02 −0.38 ± 0.02 0.95± 0.03 1.95± 0.02 8.22 5.94 Lomefloxacin (4) – – 9.00± 0.03 6.25± 0.05 5.00± 0.10 −1.48 ± 0.01 −0.36 ± 0.02 0.86± 0.02 1.78± 0.01 Ciprofloxacin (5) 8.24 5.86 8.95± 0.04 6.35± 0.07 5.05± 0.15 −1.54 ± 0.02 −0.65 ± 0.03 0.95± 0.03 1.89± 0.02 8.62 6.09 Pipemidic acid (6) 8.18 5.42 8.90± 0.04 6.15± 0.06 5.25± 0.12 −1.61 ± 0.02 −0.60 ± 0.03 0.90± 0.03 1.95± 0.02 pKa pKa µA−

Quinolones (without a piperazine)

Flumequine (7) 6.61 6.50 6.35± 0.01 −1.97 ± 0.02

Nalidixic Acid (8) 5.95 6.01 6.00± 0.01 −1.97 ± 0.02

pKa2 pKa1 pKa2 pKa1 µBH+ µ_BH

22+ Piperazine derivatives

1-Phenylpiperazine – – 8.80± 0.02 6.30± 0.02 1.68± 0.02 2.58± 0.03

Piperazine 9.73 5.33 – – – –

a _{Literature pKavalues of quinolones obtained from[3–5,7,8]; literature values of piperazine obtained from[9].} b _{Mobility in unit of /10}−4_cm2_V−1_s−1_.

ciation of the carboxylic group, only one pKavalue could be

determined for these two compounds. The determination of the pKavalues of these compounds is quite straightforward,

as described previously [11–14,19,20]. The pKa values of

nalidixic acid and flumequine determined are 6.00 and 6.35, respectively. These pKavalues were in good agreement with

the literature values[3–5,7,8]. The results reflect that the pKa

values attributable to the ionization of the carboxylic moiety are around 6.00–6.35.

3.2.2. Mobility equation involving two protolytic equilibria

Two pKavalues were determined for 1-phenylpiperazine

which involved two protonation/deprotonation equilibria. The variation of the electrophoretic mobility of this com-pound as a function of buffer pH (shown inFig. 2) could be described y the following equation[2,10,16]:

µeff =[H3O+]

2

µBH22++ Ka1[H3O+]µBH+

[H3O+]2+ Ka1[H3O+]+ Ka1Ka2

(2)

whereµBH+andµ_BH22+, respectively, are the electrophoretic mobilities of the first and second protonated species of the solute. The two pKa values and two limiting mobilities of

1-phenylpiperazine were then determined by adjusting the trial values of these four parameters and by curve-fitting the experimental mobility data with the predicted mobil-ity curve as a function of buffer pH through the utiliza-tion of Microcal Origin software until the best fit was ob-tained. For illustration,Fig. 3shows the best fit of the mo-bility curve of 1-phenylpiperazine. As can be seen, good agreement between the predicted and observed mobility

Fig. 3. The agreement between the predicted mobility curves (represented by dashed lines) and observed mobility curves (shown by data points) for 1-phenylpiperazine ().

curves were obtained. The two pKavalues determined for

1-phenylpiperazine are 6.30 and 8.80 (seeTable 1). These two pKa values are qualitatively agreeable with those of

piper-azine reported in the literature, which are 5.33 and 9.73[9]. The results clearly reveal that two protonation/deprotonation equilibria are involved in the piperazinyl moiety of a compound.

Fig. 4. Protonation/deprotonation equilibria of quinolones with a piperazinyl substituent.

3.2.3. Mobility equations involving three protolytic equilibria

It should be pointed out that the mobility equation de-rived previously by Barbosa et al.[3]is good for describing the variation of the electrophoretic mobility of quinolones involving two protonation/deprotonation equilibria, but not appropriate for describing the electrophoretic mobility of quinolones with a piperazinyl substituent. As two protona-tion/deprotonation equilibria are involved in piperazine[9]

and in the piperazinyl moiety of compounds such as pheno-thiazines with a piperazinyl group[10], it is thought that three protonation/deprotonation equilibria should be involved for quinolones with a piperazinyl substituent.

Fig. 4 shows the schematic diagram of the protona-tion/deprotonation equilibria of quinolones with a piper-azinyl substituent. As indicated, three charged species des-ignated as H2BAH2+, HBAH+ and BA−, together with a

zwitterionic species, designated as HB+A−, exist in the electrophoretic system. The effective electrophoretic mo-bility of each individual quinolone with a piperazinyl sub-stituent is mainly contributed from two positively charged species and one negatively charged species, because zwitte-rionic species (HB+A−) is usually assumed to have zero net charge.

In the case that HB+A−possesses no net charge, the effec-tive electrophoretic mobility of these quinolone compounds is described by the equation as derived for phenothiazines with a piperazinyl group at low pH[10]. This mobility equation can also be obtained from a generalized mobility equation involving multiprotolytic equilibria[16]and is given by:

µeff = αH2BAH2+µH2BAH2++αHBAH+µHBAH++αBA−µBA−

= [H3O+] 3 µH2BAH2++ Ka1[H3O+] 2 µHBAH+ + Ka1Ka2Ka3µBA− [H3O+]3+ Ka1[H3O+]2+ Ka1Ka2[H3O+] + Ka1Ka2Ka3 (3)

whereµ and α represent the limiting electrophoretic mo-bility and mole fraction, respectively, of a particular charged species of quinolones and Karepresents the dissociation

con-stants of the charged species. Ka1, Ka2 and Ka3refer to the

dissociation constants for the first deprotonation of the piper-azinyl substitutent, the dissociation of the carboxylic group,

and the second deprotonation of the piperazinyl substitutent, respectively, of the solutes, while pKa1< pKa2< pKa3.

On the other hand, based on the fact that the ampho-teric compounds such as 4-hydroxyphenylalanine are slightly charged at pH 7.0 [11], it may be reasonable to assume that HB+A−possesses a slightly negative charge at pH 7.0. Besides, no significant difference in the mobility curves of quinolones studied were observed with or without the addi-tion of neutral marker in the electrophoretic system. Thus, the contribution of the mobility of HB+A− to the overall effective electrophoretic mobility of these quinolones is not completely negligible. This is caused by a stronger dissoci-ation of the carboxyl group as compared to the protondissoci-ation of the amino group. Thus, the mobility equation can be ex-pressed as:

µeff = αH2BAH2+µH2BAH2++ αHBAH+µHBAH+

+ αHB+A−µHB+A−+ αBA−µBA− = [H3O+]3µ_H2BAH2++ Ka1[H3O+]2µHBAH+ + Ka1Ka2[H3O+]µHB+A−+ Ka1Ka2Ka3µBA− [H3O+]3+ Ka1[H3O+]2+ Ka1Ka2[H3O+] + Ka1Ka2Ka3 (4) Accordingly, the migration behavior of quinolones with a piperazinyl substitutent can then be predicted, once the pKa values and limiting electrophoretic mobility are

known.

Although the analysis of the mobility curves of quinolones with a piperazinyl substitutent according toEq. (4)is not so straightforward because one of the inflection points of the mo-bility curve is obscurely observed, with the aid of computer simulation, the difficulty can be overcome and the inflection points can be determined for compounds with two or more consecutively close pKa values. For instance, the mobility

curve of tyrosine exhibits only one inflection point, but two pKavalues were determined at 8.94 and 9.99[16]; only one

inflection point appeared in the mobility curve of isophthalic acid, but two pKavalues were determined at 3.63 and 4.58[2].

Hence, we believe that, with the aid of computer simulation and by consulting the pKa values of compounds possessing

a piperazinyl substituent, cloudiness of the determination of pKavalues can be clarified.

Table 2

The pKavalues and limiting mobility data of quinolones and piperazine derivatives evaluated according toEq. (3)

Analytes pKavalues Limiting mobilitya

pKa3 pKa2 pKa1 µBA µHBAH+ µH2BAH2+

Quinolones(with a piperazine substituent)

Ofloxacin (1) 8.00± 0.03 5.85± 0.05 5.20± 0.08 −1.45 ± 0.01 1.00± 0.03 1.85± 0.01 Enoxacin (2) 8.30± 0.03 5.75± 0.06 5.15± 0.10 −1.60 ± 0.02 1.10± 0.03 2.00± 0.02 Norfloxacin (3) 8.10± 0.04 5.85± 0.08 5.10± 0.15 −1.55 ± 0.02 1.00± 0.03 1.95± 0.01 Lomefloxacin (4) 8.55± 0.04 5.80± 0.10 5.15± 0.15 −1.50 ± 0.01 0.90± 0.02 1.78± 0.01 Ciprofloxacin (5) 8.25± 0.05 5.75± 0.15 5.15± 0.20 −1.54 ± 0.02 0.95± 0.03 1.88± 0.02 Pipemidic acid (6) 8.35± 0.04 5.80± 0.10 5.25± 0.20 −1.63 ± 0.02 0.95± 0.03 1.95± 0.02 a _{Mobility in unit of /10}−4_cm2_V−1_s−1_.

The determination of the pKa values of these quinolones

requires the initial values of the pKaand the limiting mobility

of the three charged species, and a trial value for the mobility of the zwitterionic species as well, according toEq. (4)or (3). The initial values ofµ_H2BAH2+ andµBA− can be estimated from the mobility curve at pH < 3.5 and pH > 9.5, respectively. The initial value ofµHBAH+estimated is about the half of the initial value ofµ_H2BAH2+. In the case of HB+A−possessing zero net charge, the initial values of pKa3can be estimated

from the pH values corresponding to the half of the limiting mobility of the negatively charged species of each individ-ual quinolone, which is about 8.0–8.5, but when HB+A−is slightly negatively charged, a little higher value is expected for the initial value of pKa3. The pKa1values are expected to

fall in the pH range of 3.7–5.6 for quinolones with a piper-azinyl substituent because the difference between the pKa1

and pKa3values are very likely to fall in the range 2.5–4.4.

This is based on the facts that the difference of the two pKa

values determined for 1-phenylpiperazine is 2.5 and those for piperazine and phenothiazines with a piperazinyl substitutent are in the range of 4.0–4.4[9,10]. Moreover, the pKavalues

of anilinium compounds also fall in the pH range 4.5–5.1

[9]. Furthermore, the two pKavalues of phenothiazines with

a piperazinyl moiety reported are in the ranges of 3.60–3.86

Fig. 5. The agreement between the predicted mobility curve according to (A)Eq. (3)and (B)Eq. (4)(represented by dashed lines) and observed mobility curve (shown by data points) for ofloxacin ( ).

and 7.90–8.15[10]. Judging from the shape of the mobility curves of quinolones with a piperazinyl substitutent shown inFig. 2, the expected range of the pKa1value is further

nar-rowed down to 4.7–5.6 because the inflection point for the determination of pKa1 is very unlikely to occur in the pH

range of 3.0–4.7 or below 3.0. The pKa2values are expected

close to the pKa values of nalidixic acid and flumequine.

Hence, it is reasonable to set the initial value of pKa2in the

range of 5.7–6.5.

The three pKavalues and three or four limiting mobilities

of each individual quinolone with a piperazinyl substitutent were then determined by adjusting the trial values of these six or seven parameters simultaneously and by curve-fitting the experimental mobility data with the predicted mobility curve as a function of buffer pH through the utilization of Microcal Origin software until the best fit was obtained.Tables 1 and 2

list the pKavalues and limiting mobilities evaluated for six

quinolones with a piperazinyl substitutent, according toEqs. (4) and (3), respectively, and those for two quinolones without a piperazinyl substituent, and 1-phenylpiperazine, together with the literature pKavalues of quinolones and piperazine.

In the case that HB+A− possesses zero net charge, the agreement between the predicted mobility curve based on

Fig. 6. The agreement between the predicted mobility curve according to (A)Eq. (3)and (B)Eq. (4)(represented by dashed lines) and observed mobility curve (shown by data points) for ciprofloxacin (䊉).

6.0–10.0 was poor for quinolones with a piperazinyl sub-stituent, except for ofloxacin. The agreement between the pre-dicted and observed mobility curve was fair, because the pKa3

value of ofloxacin evaluated was comparatively lower than those of the others. On the other hand, good agreement be-tween the predicted mobility curves based onEq. (4)and the observed mobility curves were obtained for all six quinolones when HB+A− possessing a slightly negative charge was considered. For illustration,Figs. 5 and 6show the agreement between the predicted mobility curves based on Eqs. (3) and (4)and the observed mobility curves for ofloxacin and ciprofloxacin, respectively. As can be seen, the predicted mobility curves of ofloxacin and ciprofloxacin based on

Eq. (4)agree very well with the observed mobility curves, especially in the pH range of 6.0–10.0. Evidently, the results obtained in this study clearly demonstrate that three proto-nation/deprotonation equilibria are involved for quinolones with a piperazinyl substituent and thatEq. (4)is more appro-priate thanEq. (3)for describing the electrophoretic behavior of quinolones with a piperazinyl substituent. In other words, the results may suggest that the zwitterionic species (HB+A−) of quinolones with a piperazinyl substitutent are slightly negatively charged. In order to obtain further support, experiments for the determination of pKa values of these

quinolones by nuclear magnetic resonance are undertaken.

4. Conclusion

Three protonation/deprotonation equilibria, instead of two, are involved in quinolones with a piperazinyl substituent. With the aid of computer simulation, three pKa values were

determined by CE for quinolones with a piperazinyl sub-stituent through the analysis of mobility curves presented as a function of buffer pH. The zwitterionic species of quinolones with a piperazinyl substituent is slightly negatively charged and the contribution of the zwitterionic species to the

effec-tive mobility may not be neglected. The pKa values

deter-mined allow us to rationalize the influence of buffer pH on the electrophoretic behavior of these quinolones in CZE.

Acknowledgement

We thank the National Science Council of Taiwan for fi-nancial support.

References

[1] C. Hansch, A. Leo, Exploring QSAR, vol. 1, American Chemical Society, Washington, DC, 1995.

[2] S.J. Gluck, K.P. Steele, M.H. Benk¨o, J. Chromatogr. A 745 (1996) 117.

[3] J. Barbosa, D. Barron, E. Jimenez-Lozano, J. Chromatogr. A 839 (1999) 183.

[4] D. Barron, A. Irles, J. Barbosa, J. Chromatogr. A 871 (2000) 367. [5] D. Barron, E. Jimenez-Lozano, J. Barbosa, J. Chromatogr. A 919

(2001) 395.

[6] D. Barron, E. Jimenez-Lozano, A. Irles, J. Barbosa, J. Chromatogr. A 871 (2000) 381.

[7] J. Barbosa, D. Barron, J. Cano, E. Jimenez-Lozano, V. Sanz-Nebot, I. Toro, J. Pharm. Biomed. Anal. 24 (2001) 1087.

[8] J. Barbosa, R. Berges, I. Toro, V. Sanz-Nebot, Talanta 44 (1997) 1271.

[9] D.C. Harris, Exploring Chemical Analysis, Freeman & Co., New York, 2001.

[10] C.E. Lin, W.S. Liao, K.H. Chen, W.Y. Lin, Electrophoresis 24 (2003) 3154.

[11] J.K. Straster, M.G. Khaledi, Anal. Chem. 63 (1991) 2503. [12] J.M. Becker, F.M. Everaerts, M.T. Ackermans, J. Chromatogr. 537

(1991) 407.

[13] J. Cai, J.T. Smith, Z. El Rassi, J. High Resolut. Chromatogr. 15 (1992) 30.

[14] J.A. Cleveland Jr., M.H. Benko, S.J. Gluck, Y.M. Walbrohl, J. Chro-matogr. A 652 (1993) 301.

[15] S.C. Smith, M.G. Khaledi, Anal. Chem. 65 (1993) 193.

[16] Y. Ishihama, Y. Oda, N. Asakawa, J. Pharm. Sci. 83 (1994) 1500. [17] S.J. Gluck, J.A. Cleveland, J. Chromatogr. A 680 (1994) 43.

[18] S.J. Gluck, J.A. Cleveland, J. Chromatogr. A 680 (1994) 49. [19] J.A. Cleveland, C.L. Martin, S.J. Gluck, J. Chromatogr. A 679 (1994)

167.

[20] C.E. Lin, W.C. Lin, W.C. Chiou, J. Chromatogr. A 705 (1995) 325.

[21] C.E. Lin, C.C. Chang, W.C. Lin, E.C. Lin, J. Chromatogr. A 753 (1996) 133.

[22] C.E. Lin, C.C. Chang, W.C. Lin, J. Chromatogr. A 768 (1997) 105.