use of a hepta-tyr glycopeptide antibiotic as chiral selector in capillary electrophoresis

1742 S . Fanali ef a/. Electrophoresis 1998, 19, 1742-1751

Salvatore Fanali' Zeineb Aturki' Claudia Desiderio' Alessandra Bossi' Pier Giorgio Righetti'

'Istituto di Cromatografia del C.N.R., Area della Ricerca di Roma, Monterotondo Scalo (Roma), Italy 'Department of Agricultural and Industrial Biotechnology, University of Verona, Verona, Italy

Use of a Hepta-tyr glycopeptide antibiotic as c h i d selector in capillary electrophoresis

A new glycopeptide antibiotic, MDL 63,246 (Hepta-tyr), of the teicoplanin family, has been evaluated in capillary electrophoresis for the resolution of chiral compounds of pharmaceutical and environmental interest. Electro- phoretic separations were carried out in a polyacrylamide-coated capillary using the partial filling-counter current mode with aqueous-organic buffers in the pH range 4-6. Experimental parameters affecting resolution, such as antibiotic concentration, buffer pH, organic modifier type and capillary temperature, were studied. The Hepta-tyr antibiotic exhibited a high enantiorecognition capability towards the studied compounds at very low concentrations (1-2 mg/mL). The optimum experimental conditions were achieved by using a buffer at pH 5 containing acetonitrile at 25°C.

1 Introduction The separation of chiral compounds is an interesting topic of research in analytical chemistry, especially in the pharmaceutical, biomedical and environmental fields where the use of pure enantiomeric forms is being widely adopted. As a consequence of the above, highly efficient and fast analytical methods for monitoring, e.g., stereoselective synthetic processes, different biological activities, metabolism and pharmacokinetics of the two enantiomers, chiral purity control of drugs, etc., are nowadays continuously requested. The search for power- ful new chiral selectors is therefore a challenging topic of research. Several analytical techniques have been used for the separation of enantiomers; among them, thin- layer chromatography (TLC), high performance liquid chromatography (HPLC), gas chromatography (GC) and capillary electrophoresis (CE) are excellent tools for the analysis of such isomers [l-151. CE can offer several advantages over the other methods, namely: very high efficiency of separation and resolution power, short analysis time, feasibility, versatility, reduced sample prepara- tion and low cost. The use of minute quantities of reagents and samples allows us to utilize expensive power- ful chiral selectors and to strongly reduce the environmental chemicals pollution impact.

As documented by the wide number of chiral analyses reported in the literature, the direct separation method has been shown to be very useful. In fact, compared with the indirect method where the enantiomers are irre- versibly converted into diastereoisomers by chemical de- rivatization procedures, in the direct method the chiral compounds are injected in the CE system unmodified, and during the electrophoretic run they interact differ- ently with a chiral environment by forming labile and

Correspondence: Dr. Salvatore Fanali, Consiglio Nazionale delle Ricerche, Istituto di Cromatografia-Area della Ricerca di Roma, P.O. Box 10-00016 Monterotondo Scalo, Rome, Italy (Tel: +39-6-9062-5328; Fax: +39-6-9062-5849; E-mail: [email protected])

Abbreviations: BGE, background electrolyte; B.R., Britton/Robinson buffer; Hepta-tyr, Hepta-tyrosine (MDL 63,246) glycopeptide antibiotic

Keywords: Antibiotics / Chiral separations / Enantiomers / Drugs / Herbicides / Capillary electrophoresis

reversible diastereoisomeric complexes. With this method the chiral selector can simply be dissolved in the buffer or bound to the capillary wall or included into a gel [ 14, 16-18]. According to the physico-chemical properties of the compounds to be analyzed, several resolution mechanisms and chiral selectors can be chosen, e.g., inclusion-complexation (cyclodextrins, crown-ethers), ligand exchange (metal amino acid complexes), optical micelle solubilization (bile salts, SDS-digitonin), ion-pair interaction and affinity interactions (proteins, antibiotics), etc. [14, 161.

Macrocyclic antibiotics have recently been introduced by Armstrong and co-workers in CE for the successful resolution of several classes of racemic compounds [19]. Other macrocyclic antibiotics have been used as chiral selectors in CE for the resolution of both anionic, cationic and neutral compounds, including glycopeptides (vancomycin and analogues, ristocetin A, teicoplanin) [20-221, ansamycins (rifamycin B and SV) [23, 241 and aminoglycosides (fradiomycin, kanamycin, streptomycin) [25]. All these chiral agents possess several functional and ionizable groups which are responsible for their stereoselective resolution capability and for their charge, which can be both positive or negative, depending on solution pH. Macrocyclic antibiotics have been demon- strated to be excellent chiral selectors exhibiting a very high resolution power at very low concentration towards a wide variety of chiral compounds. The enantioselective mechanism is based on both primary (charge-charge) and secondary (hydrogen, dipole-dipole, hydrophobic and n-n, steric repulsions) interactions [21].

The use of macrocyclic antibiotics shows some limita- tions mainly due to both adsorption on the capillary wall and strong absorption in the UV regions commonly used in CE for detection. However, the use of coated capillaries, in which the electroosmotic flow is almost absent, can be particularly advantageous for reducing the adsorption phenomena and for improving the detection sensitivity by filling only the effective length of the capillary with the chiral selector (partial filling method - counter current process). In fact, due to the opposite charge of the antibiotics with respect to the analytes, the chiral selector will migrate far from the detector cell so that the analyte can be monitored with the highest sensitivity.

0 WILEY-VCH Verlag GmbH, 69451 Weinheim, 1998 0173-0835/98/1010-1742 $17.50+.50/0

Electrophoresis 1998, 19, 1742-1751 CE chiral separations via a Hepta-tyr antibiotic 1743

This paper reports the use of a new small glycopeptide antibiotic of the teicoplanin family, the compound MDL 63,246 (Hepta-tyr), as chiral selector in CE for the resolution of chiral compounds of pharmaceutical and environmental interest. This antibiotic was recently synthesized by Malabarba et al. [26] and purified by Righetti’s group (271 b y isoelectric focusing under operating conditions compatible with the physico-chemical properties of this chiral selector. The enantiorecognition capability of Hepta-tyr has been tested by CE in a polyacrylamide- coated capillary using the partial filling - counter current method. The effect of several experimental parameters such as antibiotic concentration, buffer pH, organic solvent and capillary temperature on chiral resolution and migration separation factor of studied racemic compounds was investigated.

2 Materials and methods

2.1 Chemicals and reagents

Phosphoric acid (85% w/w), acetic acid, boric acid and sodium hydroxide were purchased from Carlo Erba (Milan, Italy). Methanol, n-propanol, 2-propanol, acetone, acetonitrile, all of analytical grade, were purchased from BDH Laboratories Supplies (Poole, Dorset, Eng- land). 2-2-2-trifluoroethanol was from Fluka (Buchs, Switzerland). The Hepta-tyr (MDL 63,246) antibiotic was synthesized at the Lepetit Research Center (Gerenzano, Italy). Racemic ibuprofen, indoprofen, fenoprofen, ketoprofen and flurbiprofen were from Sigma (St. Louis, MO, USA); racemic suprofen, carprofen, cicloprofen and naproxen were kindly provided by Dr. Cecilia Bartolucci (Istituto di Strutturistica Chimica, C.N.R., Monterondo Scalo, Rome, Italy). Racemic warfarin was from Aldrich (Steinheim, Germany). Racemic acenocoumarol was extracted with acetone from the pharmaceutical prepara- tion commercially, available. D,L-Loxiglumide was kindly supplied by Rotta Research Laboratories S.p.A. (Monza, Italy). Haloxyfop, fluazifop, fenoxaprop, flamprop, diclofop, mecoprop, fenoprop and dichlorprop racemic standard compounds were purchased from Lab-Service Analytica (Anzola Emilia, Bologna, Italy). Doubly dis- tilled water was used for preparing the solutions.

2.2 Apparatus and procedures

A Biofocus 3000 automated CE apparatus (Bio-Rad, Her- cules, CA, USA) equipped with a multi-wavelength vari- able UV detector was used for all experiments. A polyacrylamide-coated capillary, prepared in our laboratory according to Schutzner et al. [28], of 35 cm total length (effective length, 30.5 cm) X 50 pm ID was inserted into the user assembler cartridge. The temperature of both the cartridge and the carousel was kept constant at 25°C. Samples were injected by low pressure application at the cathodic end of the capillary by setting to 10 p.s.i-s ins- trument value, corresponding to 34.475 kPa X 2 s. Con- stant applied voltage was -20 kV (negative polarity). The buffers used for different experiments (50 mM), pH range 4-6, were Britton/Robinson (B.R.) buffers com- posed of equal concentrations of boric, acetic and phosphoric acids. The 50 mM buffers were prepared from a

concentrated 150 mM B.R. solution, titrated with sodium hydroxide to the desired pH before the final dilution. The running background electrolyte (BGE) was supplemented with acetonitrile (ACN) to a final composition of 50 mM B.R./ACN (80:20 v/v). Between runs the capillary was rinsed at high pressure with a running buffer containing 40°/o ACN for about 100 s followed by the standard running buffer (containing 20% of ACN) for 100 s. The next step was the introduction of the buffer containing the antibiotic (the Hepta-tyr chiral selector) at low pressure in order to control the process and fill the capillary up to the detector cell (partial filling method). Acenocoumarol, warfarin, dichlorprop and mecoprop stock solutions (1 mM) were prepared in acetone. All other standard-compound concentrated solutions were dissolved in methanol. The stock loxiglumide drug solution was prepared in an ACNIwater (60:40 v/v) solvent mixture. The concentrated solutions were diluted before injection with ten times diluted buffer (5 mM) to a final concentration of 0.1 mM with the exception of cycloprofen, ketoprofen, acenocoumarol, warfarin, loxiglumide, dichlorprop, fenoprop, flamprop and fluazifop, which were diluted to 0.05 mM (injection solution).

3 Results and discussion The MDL 63,246 Hepta-tyr antibiotic is a semisynthetic small glycopeptide antibiotic structurally related to a class of amides of 34-de(acetylglucosaminyl)-34-deoxy teicoplanin. It was recently purified by Righetti’s group 1271 and shows a unique antibiotic activity against Entero- coccus fecalis and E. faecium bacteria strains, which was not exhibited by vancomycin or teicoplanin. On the basis of the excellent results previously obtained with glycopeptide antibiotics as chiral selectors in CE, the Hepta-tyr enantiorecognition capability was here tested for the enantiomeric separation of drugs and molecules of environmental interest. Figure 1 shows the chemical structure of the Hepta-tyr antibiotic. The Hepta-tyr ab- breviation is related to the presence of a stretch of seven tyrosine residues in the heptapeptide backbone. As can be observed, the peptide part of the molecule does not contain any free carboxylic group and, as a consequence, it exhibits a basic character with an experimentally obtained isoelectric point (pl) of 8.62 [27]. As described in a recent paper [27], a drawback in the use of Hepta-tyr antibiotic is its poor solubility in aqueous solutions, which is strongly dependent on buffer pH. In fact, the pH range 5-10, which is often used in CE for the analysis of acidic compounds, is critical and therefore the addition of organic modifiers or detergents to the buffer can improve the solubility of the chiral selector. This characteristic of low solubility in aqueous solutions is a common feature of all teicoplanin antibiotics. Another disadvantage in using Hepta-tyr, which is currently exhibited by all macrocyclic antibiotics, is the intense W absorption which leads to a strong decrement in the method detection sensitivity. Figure 2 shows the UV spectrum of Hepta-tyr obtained, on line, by analyzing the antibiotic by CE in 50 mM B.R. buffer at pH 5.

At low antibiotic concentrations the direct UV detection of analytes is possible at wavelengths higher than

1744 S. Fanali e f a/. Electrophoresis 1998, 19, 1142-1151

0.0500

0.0398

0.0296

a 4 E

0.0194

0.0092

-0.0010

Y = NH -N(CH,),

I I , I I I I I I I I I 12!3-263- zoo 220 240 260 280 300

nm

Figure 2. W spectra of Hepta-tyr antibiotic (195-300 nm) obtained in 50 mM B.R. buffer, pH 5, 20% acetonitrile.

.NH-CH,

Figure 1. Chemical structure of Hepta-tyr (MDL 63,246) glycopeptide antibiotic. Modified from [27].

254 nm. However, this condition does not represent the best solution for achieving the highest sensitivity and thus the use of partial filling - counter current methods in coated capillaries can be a good approach for method optimization. The partial filling method, which was introduced by Valtcheva et al. [29] when using an enzyme (cellobiohydrolase I) as chiral selector in CE, has been extensively adopted with vancomycin antibiotics for improving the detection sensitivity [30-331. This method consists of filling only part of the capillary with the BGE containing the chiral selector, thus leaving the detector path free of UV-absorbing buffer species. When the voltage is applied, due to the opposite charge, the chiral selector migrates away from the detector cell and the analytes can be detected at the highest sensitivity. Due to the inverse migration of analytes and chiral agent this method is also called counter current process.

Twenty racemic acidic compounds with diverse molec- ular structures belonging to different classes of drugs and herbicides (for their chemical structures see Fig. 3) were selected in order to study the enantiorecognition capability of this new chiral selector. Several physico- chemical parameters, such as the antibiotic concentration, the buffer pH, the temperature and the addition of several organic solvents to the BGE, were investigated with the aim of finding the optimum experimental conditions and of outlining how to overcome the drawbacks in using Hepta-tyr in CE. Preliminary experiments were devoted to finding the best organic solvents for improving the antibiotic solubility in aqueous buffer. Based on Hepta-tyr solubility data, several organic solvents, e.g.,


Drugs I y 3

CI m / /

Carprofen

FenOBrOfen

H

Suprofen

CHCOOH

Cidopmfen

&? CHCOOH

Flurbiprofen

$HI CHCOOH @a

Woprofen

CH,O m;,Po,, Naoroxen

Herbicides

CI

Haloxyfop acid Fluazifop acid

Diclofop acid Fenoxaprop acid

Flampmp acid

Mecoprop Fenopmp Dichlorprop Loxiglumide

q H OH 0 No, q H O OH -

CH,COCH, CH,COCH,

Acenowumarol Warfarin Figure 3. Chemical structures of the analyzed compounds.

methanol, 2- and n-propanol, acetonitrile, acetone and trifluoroethanol, were mixed with the B.R. buffer at pH 5 at 20% volume aliquots and the chiral selector was dissolved. The mixture of B.R. buffer at pH Uacetoni- trile (80:20 v/v) proved to be the most effective solvent, allowing us to dissolve reasonable amounts of Hepta-tyr (up to 3 mg/mL).

Different capillary washing procedures between runs were tested in order to avoid peak tailing, high noise, lack of detection sensitivity, and long migration times. After several attempts, the best procedure for removing the adsorbed antibiotic and for restoring the initial capillary conditions was to flush the capillary with the BGE solution containing 40% of acetonitrile at high pressure for 100 s followed by column conditioning with the running buffer supplemented with 20% acetonitrile (100 s at high pressure). The buffer containing the chiral selector was flushed at low pressure (5 p.s.i., corresponding to 34.475 kPa) for the time necessary to reach the detector cell. The introduction of the BGE/chiral selector was controlled according to the change of buffer viscosity when the antibiotic concentration was increased.

3.1 Effect of antibiotic concentration

On the basis of preliminary experiments, the 50 mM B.R. buffer pH 5 , containing 20% v/v of acetonitrile was selected as BGE to study the effect of the antibiotic concentration on the enantiomeric resolution of the racemic compounds described in Fig. 3. Considering the high enantioselectivity exhibited by glycopeptide antibiotics at very low levels, Hepta-tyr was added to the buffer in the concentration range of 0.5-3 mg/mL. Table 1 sum- marizes the obtained results on migration time, resolution and migration time separation factor a. When the antibiotic concentration was increased from 1 to 3 mg/mL, longer migration times were generally recorded for almost all the analyzed compounds. At 0.5 mg/mL concentration level, unexpected long migration times were observed. These results have been confirmed by several experiments and therefore 0.5 mg/mL seems to be a critical antibiotic concentration. It is hypothesized that this antibiotic, belonging to the teicoplanin family group, can form micellar aggregates with a CMC value close to 0.5 mg/mL. Below this concentration the monomeric form of the antibiotic could be prevalent, leading

1746 S . Fanali ef a/. Electrophoresis 1998, 19, 1742-1751

Table 1. Effect of Hepta-tyr concentration on resolution, migration separation factor (a) and migration timesa'

Hepta-tyr buffer concentration Compounds 0 mg/mL 0.5 mg/mL 1.0 mg/mL 2.0 mg/mL

-

Drugs Carprofen

Acenocoumarol

Cycloprofen

Fenoprofen

Flurbiprofen

Ibuprofen

Indoprofen

Ketoprofen

Loxiglumide

Naproxen

Suprofen Warfarin

Herbicides Dichloprop

Diclofop

Fenoprop

Fenoxaprop

Flamprop

Fluazifop

Haloxyfop

Mecoprop

6.90

7.24

6.30

5.97

5.15

6.34

6.39

6.01

9.70

6.09

5.69 9.32

4.46

5.67

4.69

5.85

5.58

5.58

6.02

4.52

9.50 9.96

10.12

8.05

1.60

7.07

8.40

8.10 8.26 7.42 8.24

11.45 11.76 6.97 7.04 7.55

11.03

5.23

7.33

6.04

7.42

7.19 7.28 6.73

7.24

-

0.55 1.05 -

-

-

-

-

< 0.50 1.02 2.52 1.11 0.80 1.03

<0.50 1.01 - -

-

-

-

-

<0.50 1.01 -

-

-

Tm 1

TlllZ

7.11 8.27 8.31 8.79 6.54 6.92 6.02 6.29 5.72 5.83 6.38 6.42 7.15 7.58 6.43 7.18 8.34 8.93 5.92 6.24 6.35 8.65 8.88

4.43 4.58 6.76 7.71 4.84 4.94 5.89

5.30 5.55 5.30

6.39

4.57 4.67

R a

1.04 1.16 0.91 1.06

<0.50 1.06 0.75 1.04

<0.50 1.02

<0.50 1.01 0.88 1.06 4.36 1.12 1.26 1.07 0.88 1.05

0.71 1.03

0.98 1.03 0.66 1.14 0.55 1.02

-

-

1.17 1.05 -

-

0.87 1.02

8.70 13.78 8.68 9.09 6.83 7.18 6.40 6.73 7.08 1.79 7.15 7.34 9.50 9.95 7.48

11.71 8.74

6.54 6.95 8.40 8.38 8.64

4.91 5.35 7.77 8.51 4.89 5.04 7.02 7.20 5.95 6.15 5.50

7.30

4.60 4.85

4.64 1.58 1.05 1.05 0.58 1.05 1.27 1.05 1.38 1.10

<0.50 1.03

<0.50 1.05 7.18 1.56 -

1.34 1.06

0.86 1.03

2.04 1.08

<0.50 1.09 0.96 1.03

<0.50 1.02 3.31 1.13

-

-

-

1.08 1.05

3.0 mg/mL

8.65 1.30 9.19 1.06

11.98 0.82 12.26 1.02 1.35 1.22 7.80 1.06

11.45 0.76 11.71 1.02 7.70 <0.50 8.07 1.05

12.36 <0.50 12.69 1.03 11.07 1.18 11.65 1.05 9.50 -

6.64 1.66 7.19 1.08 9.10 - 8.76 1.26 9.14 1.04

5.25 2.55 5.70 1.08

11.90 <0.50 12.28 1.03 5.39 1.21 5.64 1.05 8.34 <0.50 8.55 1.02 6.61 5.77 8.03 1.21 6.63 0.66 6.77 1.02 7.60 <0.50 7.80 1.02 4.80 1.54 5.03 1.05

a) Experimental conditions: 50 mM B.R. buffer, pH 5/ACN (80:20 v/v). Applied voltage -20 kV. Capillary: 35 cm total length (30.5 cm effective length) X 50 pm ID, polyacrylamide coated. Capillary temperature: 25OC. Sample injection: by pressure using 5 p.s.i. (34.475 Kpa) X 2 s. Sam- ples solutions were 0.1 mM with the exception of cycloprofen, ketoprofen, acenocoumarol, warfarin, loxiglumide dichlorprop, fenoprop, flamprop and fluazifop, which were injected at 0.05 mM level. Partial filling at 5 psi X 32-36 s.

to different interaction equilibria with the analytes. It is clear that these interactions are strong but, on the other hand, less stereoselective. Teicoplanin is well known to form micelles with a CMC value of 0.18 mM in unbuf- fered solutions [21], a phenomenon which is influenced by the buffer composition, the pH and the temperature. In a recent paper Wan and Blomberg [34], while studying the resolution of DL-peptides in CE by teicoplanin, and monitoring the current vs. antibiotic concentration, observed a drop in current value when the antibiotic was used at a level very close or below its CMC value. As depicted in Fig. 4, the curve of the current versus Hepta- tyr concentration shows a drop in the current at 0.5 mg/mL concentration level, which corresponds to 0.277 mM. This result is consistent with the hypothesis that, when Hepta-tyr is at 0.5 mg/mL concentration, it reaches its critical micellar concentration value under

our experimental conditions. Furthermore, when preparing Hepta-tyr solutions, the formation of micellar dis- persions has been observed.

It can be additionally observed that, concomitant with the slight increase of migration times observed when raising the concentration of the antibiotic, an increment of enantiomeric resolution was recorded for such compounds as ketoprofen, flamprop, dichlorprop, naproxen and mecoprop. This effect is in accord with the general behavior of macrocyclic antibiotics, which, by combining high stereoselectivity with reduced analysis times, have been defined by Vespalec et al. as quasi ideal chiral selectors [35]. When using the Hepta-tyr chiral selector the characteristic effect of glycopeptide antibiotics in selec- tively modifying the migration time of only one enan- tiomer was not observed. In this specific case the migra-

Electrophoresis 1998, 19, 1742-1751 CE chiral separations wio a Hepta-tyr antibiotic 1747

25 1 23 4

.17 1 15

0 0.5 1 2 3 hepta-tyr concentration

(mglml) Figure 4. Dependence of current on Hepta-tyr concentration in B.R. buffer, pH 5/ACN (80:20 v/v). For experimental conditions see Table 1.

tion times of both enantiomers were affected by the stereoselective complexation.

All twenty studied racemic compounds were separated in their enantiomers with the exception of suprofen; however, baseline separation was not achieved for cycloprofen, ibuprofen, indoprofen, diclofop, fenoxaprop and haloxyfop. Fenoprofen, flurbiprofen and ketoprofen showed a maximum resolution value at 2.0 mg/mL of Hepta-tyr concentration, which was particularly high for ketoprofen (R = 7.18). Indoprofen and diclofop, which were only partially resolved, showed a maximum of R at 1.0 mg/mL. Particularly high resolution was observed for the above-cited ketoprofen (R = 7.18), carprofen (R = 4.64), dichlorprop ( R = 2.55) and flamprop ( R = 5.77). The migration separation factor a, which was calculated as the enantiomers migration time ratio (Tm2/Tm1), fol- lows almost the same trend of resolution. When resolution increases, the a value rises concomitantly. However, among the pharmaceuticals, only the compounds which exhibited a very high R value showed a noticeable increase of a, namely carprofen (a = 1.58, R = 4.64) and ketoprofen (a = 1.56, R = 7.18). For herbicide compounds the hehavior of a was almost the same. For flamprop free acid a strong increase of a was recognized in accord with an R increase, but a in this case did not reach a relevant high value, as for caprofen and ketoprofen compounds, although a quite high R value was obtained (R = 5.77). For an example, see Figs. 5 a and b, which show the electropherograms of the chiral resolution of dichlorprop and naproxen, respectively, at different antibiotic concentrations.

3.2 Effect of buffer pH

The ionization state of both analytes and chiral selectors are pH dependent and, as a consequence, when this parameter changes, different mobilities and stereoselective

interactions can be observed. When a new ionizable chiral selector is tested for CE enantiomeric separation, the buffer pH is an important parameter to be studied in order to investigate the effect on the enantiomeric separation and to obtain additional information about the resolution mechanism. The antibiotic stability is also strongly influenced by pH; in fact, working in the pH range 4-7 is generally suggested to avoid bond hydro- lysis and macrocyclic ring opening [20]. In the specific case of the Hepta-tyr antibiotic the pH is also important for its solubility, which is very reduced at a pH higher than 5 and lower than 10 [27]. For antibiotics which form micellar aggregates, such as teicoplanin, the pH strongly influences the process, which is usually favored at low pH [36]. On the basis of the results obtained when studying the effect of antibiotic concentration on resolution, 1 mg/mL was the antibiotic buffer concentration selected to investigate the effect of pH on enantiomeric separations. In fact, at this Hepta-tyr buffer content almost all the compounds were partially or completely resolved.

Figure 6 depicts the plots of R versus pH for all the studied compounds which exhibited adequate resolution. Flurbiprofen and ibuprofen compounds have been omit- ted due to the very low resolution value obtained at 1 mg/mL Hepta-tyr buffer content. With the exception of flamprop, all the studied herbicides showed a maximum of resolution at pH 5. Flamprop enantiomeric resolution exhibited a high increase from pH 4 to 5, reaching a resolution of 1.23, which remained almost constant at all the additional pH values studied. The same herbicide trend was observed for some drugs, namely, warfarin, acenocoumarol, fenoprofen and ketoprofen, which exhibited the highest resolution at pH 5 and almost no resolution at the other pHs studied. A different behavior was recorded for cycloprofen and carprofen, which exhibited higher enantiomeric separation when the pH was increased. Resolution of the indoprofen compound has the unique characteristic of dis- playing an almost constant value in the range of pH 4-5, followed by a decrease at pH 6.

The effect of pH observed for Hepta-tyr was different from those usually obtained with glycopeptide macrocyclic antibiotics where the maximum resolution was in general obtained at the lowest pHs. When pH increased from 4 to 5, the anionic analytes were more dissociated and thus stronger electrostatic interactions with the posi- tively charged Hepta-tyr could occur, thus leading to higher resolutions. When the pH was further increased, the drop of Hepta-tyr antibiotic solubility at pH 6, even when organic solvent was present in the buffer, was probably responsible for the reduction of enantiorecognition which could be caused by a different structural con- formation of both the monomeric and micellar antibiotic forms. Note that the pH influences the self-association of antibiotics, which form micelles by modifying the monomeric and the micellar concentrations and by altering the charge-charge interactions between the chiral selector and the analyte. Fig. 7 shows the enantiomeric separation of the racemic mixture of dichlorprop obtained at different pHs with 1 mg/mL of Hepta-tyr antibiotic.

1748 S . Fanali el a).

a 0.018

- 0.013

0.008

0.003

-0.002

E 8

4 5 6 migration time (min)

0.5 rng/rnl

0.008

g 0.005 c?

n

E

z 0.002

-0.001

0.5 mglml

p 0.007 Y) 0.005

0.003 5

0.001

-0.001

e 0

2


1 mglml

0.008

g 0.005 c?

n

f

2 0.002


1 mglml

-0.002 5 6 7 migration time (min)

Electrophoresis 1998, 19, 1742-1751

3 mglml

0.011 1 3 mglml

6 7 8 migration time (rnin)

Figure 5. Effect of Hepta-tyr concentration on enantiomeric resolution of (a) dichlorprop and (b) naproxen. Experi- mental conditions as in Table 1.

3.3 Effect of capillary temperature

The temperature influences the mobility of both the analyzed compounds and the chiral selector together with the buffer viscosity and the stability of antibiotic solution. As previously described, the antibiotic stability is strongly dependent on temperature. Lower temperatures are usually recommended for maximum efficiency, resolution and antibiotic solution stability [19, 21, 22, 361. Figure 8 shows the effect of capillary temperature on enantiomeric resolution obtained with B.R. buffer, pH 5 , containing 1 mg/mL of Hepta-tyr for a limited range of substances representative of the different classes of compounds. The range of temperatures studied was restricted (20-30°C) according to the physico-chemical properties of antibiotics. For the majority of the studied compounds, the maximum enantiomeric resolution was obtained at 25"C, with the exception of fenoprofen and warfarin, which exhibited a slightly different R value between 20 and 25°C. A different behavior was observed for flamprop and ibuprofen, which showed a decrease in R when the temperature increased.

3.4 Effect of organic solvent

The addition of organic solvents to the separation buffer can have different effects on enantiorecognition, depending on the type of antibiotic. The role of organic solvent in modifying the resolution and migration times is influenced by several effects, such as a change of buffer viscosity, modulation of analyte and chiral selector mobility, modification of the stereoselective interaction and binding constant (competition for hydrophobic interactions), etc. Furthermore, the organic solvents seem to

enhance the charge-charge interactions which are at the basis of enantiorecognition for this class of chiral selectors and also to break up micellar aggregates, thus increasing the monomer concentration of the chiral selector [36]. The study of the effect of the organic solvent on resolution is particularly interesting for the Hepta-tyr antibiotic due to its low solubility in aqueous buffers and probably to the formation of micellar aggregates. For this purpose different organic solvents, namely: acetonitrile, acetone, trifluoroethanol, n-propanol and 2-propanol, have been separately added as 20 O/o volume aliquots to 50 mM B.R., pH 5 , containing 1 mg/mL of Hepta-tyr. Table 2 shows the results for a restricted number of analytes chosen as representative of the main classes of compounds investigated. Methanol was not studied due to the very low Hepta-tyr solubility in this solvent.

The addition of the above-described organic solvents to the buffer modified the buffer viscosity, forcing the filling time of chiral selector into the capillary, a particularly long procedure when using the n- and 2-propanol, to be prolonged. Good Hepta-tyr solubility was obtained with acetonitrile, acetone and n-propanol while 2-propanol and trifluoroethanol required a long vortex time to completely dissolve the antibiotic. As can be observed in Table 2, the organic solvent showed different effects for enantiomers of profens according to the variety of chemical structures in this group. Profens were generally better resolved with other solvents than with acetonitrile; however, migration times were longer and usually broader peaks were observed. Only warfarin showed the highest resolution with trifluoroethanol as buffer addi- tive, although migration times were almost three times

Electrophoresis 1998, 19, 1742-1751 CE chiral separations via a Hepla-tyr antibiotic 1749

a 4.5

4

3.5

3

2.5

R 2

1.5

1

0.5

0 3 5 4 4.5 5 5.5 6

DH

b 16 1

1

R 0 8

0 6

04

0 2

0 35 4 4 5 5 5 5 6

PH

PH 4 0.0143

- 0.0093

v) 0

c 2 0.0043

-0.0007 3 4 5

5

C

PH 5

0.0083 I

Figure 6. Effect of buffer pH on enantiorneric resolution of the studied compounds. BGE: 50 mM B.R. buffer, containig 1 mg/mL of Hepta-tyr antibiotic. For other experimental conditions see Table 1.

0.0053

0.0023

-0.0007 3 4 5

3 a

0.008 I 0.003

-0.002 4 5 6

migration time (min) migration time (min) migration time (min)

Figure 7. Electropherograms of the enantiomeric separation of dichlorprop racemic mixtures obtained at different pHs. Antibiotic concentration, 1 mg/mL. For other experimental conditions see Table 1 and Fig. 5 .

longer than with other organic modifiers and a reduced Hepta-tyr stability was observed: in fact, every two electrophoretic runs the antibiotic solution needed to be re- plenished with a freshly prepared one. The enantiomers of loxiglumide were separated only when acetonitrile

was present in the BGE; although the addition of other organic solvents produced longer migration times, and thus probably stronger interactions, resolution was not obtained. All the racemic herbicides exhibited a maximum enantiomeric resolution with 20 O/o of n-propanol

1750 S. Fanali et al. Electrophoresis 1998, 19, 1742-1751

1 4

1.2

1

0.8

R

0.6

0.4

0.2

0

,. . . .

17 5 20 22 5 25 27 5 30 32.5

temperature ('C)

1 t f s n o p m l e n -ibupofen -e-napmxen t w r t a n n . . + . .dchbrprOp . . 0 . .Rsrnpmp I

Figure 8. Effect of capillary temperature on enantiomeric resolution of the studied compounds. For experimental conditions see Table 1.

and a particularly high R value was obtained for flamprop ( R = 3.47) but the analysis time increased. Further- more, 2-propanol and trifluoroethanol strongly absorbed at the operating UV wavelengths, leading to a decrement in detection sensitivity.

The addition of acetone to the buffer produced R values comparable to those obtained with the acetonitrile for almost all the studied compounds, but migration times were always longer, the solvent was more volatile and a loss of efficiency was recorded for profens. Therefore acetonitrile seemed to offer the best results by combining good solubilizing power and high enantiomeric resolution with the shortest analysis time. Finally, the effect of acetonitrile concentration was investigated. The addition of 10% acetonitrile to the buffer was not con- venient due to very frequent drops in current during

voltage application, probably due to the antibiotic precipitation caused by the decreased solubility. Although, by using 30% of acetonitrile in the BGE, the Hepta-tyr solubility was improved, the resolution of all analyzed compounds was completely lost.

4 Concluding remarks

From the results it can be concluded that the Hepta-tyr antibiotic exhibited a high enantiorecognition capability towards the studied analytes, showing high resolution at low concentrations (1-2 mg/mL; 0.55-1.11 mM). Some physico-chemical parameters, such as buffer pH, capillary temperature and buffer composition, should be carefully controlled in order to achieve the best enantiomeric resolution. Buffer pH 5 and 25°C were the optimum experimental conditions. Because of the low solubility of Hepta-tyr in aqueous solution, the addition of organic solvent to the buffer is necessary. Among the organic solvents tested, acetonitrile seemed to offer the best enantiomeric separations in the shortest analysis time. Due to the intense UV absorption, the use of the partial filling method in coated capillary is strongly recommended. Adsorption of Hepta-tyr on the capillary has been observed even in coated capillaries. Optimization of capillary washing procedures is necessary to effective- ly remove the adsorbed Hepta-tyr on the capillary wall and to avoid antibiotic precipitation.

P.G.R. is supported by a grant from M.U.R.S.II: (Roma, Italy, quota 40%), from C.N.R. (P.R Biotecnologie) and from POP Calabria 94-98. Z.A. is supported by a fellow- ship from C.N.R. (Comitato Ambiente e Habitat, Roma, Italy). The authors are grateful to Mrs. G. Caponecchi and M. Cristalli, Istituto di Cromatografia (C.N.R., Monteli- bretti) for their technical assistance.

Received January 2, 1998

Table 2. Effect of different organic solvents on enantiomeric resolution and migration times of selected samples')

Sample 20% Acetonitrile 20°/0 Acetone 20% 2,2,2-Trifluor- 20% n-Propanol 20% 2-Propanol ethanol

Tm1 R Tmz

Drugs Fenoprofen 6.02 0.75 7.40 0.80 14.59 0.50 13.90 1.77

6.29 7.73 15.14 15.05 Ibuprofen 6.38 <0.50 1.72 <0.50 13.06 < O S O 13.65 - 15.83 0.88

6.42 8.16 13.42 16.67 Naproxen 5.92 0.88 7.58 1.88 12.17 0.75 13.13 1.04 14.90 -

6.24 8.52 12.98 13.75 Warfarin 8.65 0.71 9.15 0.76 20.93 1.32 17.20 - 18.83 <0.50

8.88 9.48 22.75 19.19 1.26 11.24 - - 18.50 - nd Loxiglumide 8.34

8.93 Herbicides Dichloprop 4.43 0.98 5.18 0.74 6.20 1.18 8.74 1.28 11.16 1.83

Fenoxaprop 5.89 - 7.73 1.05 11.49 1.42 13.75 1.68 nd

- -

- nd

4.58 5.28 6.49 9.11 12.14

8.44 12.55 14.94

5.55 7.29 13.21 12.46

-

Flamprop 5.30 1.17 6.52 2.08 11.81 1.33 11.01 3.47 11.90 -

a) BGE: 50 mM B.R. buffer, pH 5, containing 1 mglmL of Hepta-tyr antibiotic. For other experimental conditions see Table 1.


5 References

[l] Deege, A., Husmann, H., Hubinger, E., Kobor, F., Schomburg, G.,

[2] Granville, C. P., Gehrcke, B., Konig, W. A., Wainer, I. W., J. Chro-

[3] Jung, M., Mayer, S . , Schurig, V., LC. GC-Mag. Separatron. Sci.

[4] Debowski, J., Sybilska, D., Jurczak, J., J. Chromatogr. 1983, 282,

[5] Blaschke, G., J. Liq. Chromatogr. 1986, 9, 341-368. [6] Snopek, J., Jelinek, I., Smolkova-Keulemansova, E., J. Chroma-

[7] Guttman, A., Paulus, A., Cohen, A. S., Grinberg, N., Karger,

[8] Terabe, S., Trends Anal. Chem. 1989, 8, 129-134. [9] Fanali, S., J. Chromatogr. 1989, 474, 441-446.

J. High Resolut. Chromatogr. 1993, 16, 587-593.

matogr. A 1993, 622, 21-31.

1994, 12, 458-466.

83-88.

rogr. 1988, 438, 211-218.

B. L., J. Chromatogr. 1988, 448, 41-53.

[lo] Fanali, S., BoEek, P., Electrophoresis 1990, 11, 757-760. 1111 Schmitt, T., Engelhardt, H., Chromatographra 1993, 37, 475-481. [12] Bechet, I., Paques, P., Fillet, M., Hubert, P., Crommen, J., Electro-

[13] Chankvetadze, B., Endresz, G., Blaschke, G., Electrophoresis 1994,

[14] Fanali, S., J. Chromatogr. A 1996, 735, 77-121. [15] Fanali, S., Desiderio, C., Schulte, G., Heitmeier, S., Strickmann,

D., Chankvedatze, B., Blaschke, G., J. Chromatogr. A 1998, 800,

[16] Terabe, S., Otsuka, K., Nishi, H., J. Chromatogr. A 1994, 666,

[17] Fanali, S . , Cristalli, M., Vespalec, R., BoEek, P., Adv. Electro-

[18] Vespalec, R., BoEek, P., Electrophoresis 1997, 18, 843-852. [19] Armstrong, D. W., Rundlett, K. L., Chen, J. R., Chirality 1994, 6,

phoresis 1994, 15, 818-823.

15, 804-807.

72-82.

295-319.

phoresis 1994, 7, 1-86.

496-509.

[201 Ward, T. J., LC-GC Znt. 1996, 9, 428-435. [21] Gasper, M. P., Berthod, A., Nair, U. B., Armstrong, D. W.,

[22] Desiderio, C., Fanali, S . , J. Chromatogr. A 1998, in press [23] Armstrong, D. W., Rundlett, K., Reid, G. L., Anal. Chem. 1994, 66,

[24] Ward, T. J., Dam, C., Blaylock, A., J. Chromatogr. A 1995, 715,

[25] Nishi, H., Nakamura, K., Nakai, H., Sato, T., Chromatographia

[26] Malabarba, A., Ciabatti, R., Scotti, R., Goldstein, B. P., Ferrari, P., Kurz, M., Andreini, B. P., Denaro, M., J. Antibiotics 1995, 48,

[27] Bossi, A,, Righetti, P. G., Riva, E., Zerilli, L., Electrophoresis 1996,

[28] Schutzner, W., Caponecchi, G., Fanali, S. , Rizzi, A,, Kenndler, E.,

[29] Valtcheva, L., Mohammed, J., Pettersson, G., Hjerten, S . , J. Chro-

[30] Ward, T. J., Dann 111, C., Brown, A. P., Chirality 1996, 8, 77-83. [31] Fanali, S., Desiderio, C . , J. High Resolut. Chromatogr. 1996, 19,

[32] Desiderio, C., Polcaro, C. M., Padiglioni, P., Fanali, S., J. Chroma-

[33] Fanali, S . , Desiderio, C., Aturki, Z., J. Chromatogr. A 1997, 772,

[34] Wan, H., Blomberg, L. G., Electrophoresis 1997, 18, 943-949. [35] Vespalec, R., Billiet, H. A. H., Frank, J., BoEek, P., Electrophoresis

[36] Rundlett, K. L., Gasper, M. P., Zhou, E. Y., Armstrong, D. W.,

Anal. Chem. 1996, 68, 2501-2514.

1690-1695.

337-344.

1996, 43, 426-430.

869-883.

17, 1234-1241.

Electrophoresis 1994, 15, 769-773.

matogr. 1993, 638, 263-267.

322-326.

togr. A 1997, 781, 503-513.

185-194.

1996, 17, 1214-1221.

Chirality 1996, 8, 88-107.

use of a hepta-tyr glycopeptide antibiotic as chiral selector in capillary electrophoresis

Documents