Trace analysis of aminoglycoside antibiotics in bovine milk by MEKC with LIF detection

Juan Manuel SerranoManuel Silva

Department of Analytical Chemistry,University of Cordoba,Cordoba, Spain

Received March 29, 2006Revised May 13, 2006Accepted June 28, 2006

Research Article

Trace analysis of aminoglycoside antibioticsin bovine milk by MEKC with LIF detection

This work describes a straightforward and sensitive method for the multi-residueanalysis of aminoglycoside antibiotics (kanamycin B, amikacin, neomycin B and par-omomycin I) in bovine milk samples. The method involves the pre-capillary derivati-zation of antibiotics with sulfoindocyanine succinimidyl ester (Cy5) and their separa-tion and determination by MEKC with LIF detection. The optimum procedure includesa derivatization step of the antibiotics at 257C for 30 min and direct injection for MEKCanalysis, which is performed in about 20 min by using borate buffer (35 mM; pH 9.2)with 55 mM SDS as an anionic surfactant and 20% ACN as the organic modifier.Under these conditions, dynamic ranges of 10–500 mg/L and RSDs (within-day pre-cision) from 3.8 to 5.3% were obtained. These results indicate that the proposedMEKC-LIF method is useful as a selective and sensitive tool for the determination ofthese antibiotics and surpasses other reported electrophoretic alternatives. Finally,the method was successfully applied to bovine milk samples after a simple solid-phase extraction clean-up and preconcentration procedure. The aminoglycosideswere readily detected at 0.5–1.5 mg/kg levels with average recoveries ranging from89.4 to 93.3%.

Keywords: Aminoglycoside antibiotics / Bovine milk samples / Cy5 / Diode LIF /MEKC DOI 10.1002/elps.200600184

1 Introduction

Aminoglycosides encompass a group of water-solublebroad-spectrum antibiotics that have been widely used inboth human and veterinary medicine for their bactericidalactivity against Gram-positive and Gram-negative organ-isms [1]. Monitoring aminoglycoside concentrations isimportant in drug pharmacokinetic studies, as well as inthose concerning the clinical efficacy and side effects ofthese drugs. Thus clinical chemotherapy with these anti-biotics is frequently associated with oto- and nephrotoxi-city, and therefore careful monitoring of blood levels is

required especially when therapy is of long duration. Inveterinary medicine, aminoglycosides are used for thetreatment of disease and as dietary supplements of manyspecies of animals destined for human feeding [2]. Hence,the use of these antibiotics in dairy cattle may result indrug residues in their milk, especially if they are not usedaccording to label directions. Various methods are avail-able for the determination of aminoglycoside antibiotics,such as microbiological methods, enzyme immu-noassays and chromatographic analyses. Microbiologicalassays are inexpensive and simple, but time consumingand nonspecific [3], enzyme immunoassays also lackversatility in routine multicomponent chemical analysisapplications [3, 4], and therefore LC is the most populartechnique for the analysis of these compounds [3, 5–7]. Inrecent years, CE is increasingly being viewed as an alter-native to LC for the determination of aminoglycosideantibiotics [3, 8, 9]. When compared with LC, CE offerssignificant benefits such as higher resolution, smallersample requirements and shorter analysis time.

Correspondence: Professor Manuel Silva, Department of AnalyticalChemistry, Marie-Curie Building (Annex), Rabanales Campus, Uni-versity of Cordoba, E-14071 Cordoba, SpainE-mail: [email protected]: 134-957-218614

Abbreviations: Cy5, sulfoindocyanine succinimidyl ester; MRL,maximum residue limit

Electrophoresis 2006, 27, 4703–4710 4703

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

4704 J. M. Serrano and M. Silva Electrophoresis 2006, 27, 4703–4710

Despite the better scope of CE with respect to LC, little hasbeen published on the determination of aminoglycosidesby CE, which might be attributed to the difficulty in theirquantification using a conventional spectroscopic detec-tor. In fact, these antibiotics lack a strong chromophoricmoiety, and direct absorbance or fluorescence detection isnot possible. Thus, the first work on the determination ofaminoglycoside antibiotics by CZE uses indirect detectionat low pH under reversal conditions [10]. Although the sen-sitivity achieved is suitable for the analysis of pharmaceu-tical samples, some of the closely related species could notbe separated. In order to carry out direct UV detection,several alternatives have been reported such as the for-mation of UV-absorbing negatively charged complexesbetween the hexoxe ring of the aminoglycoside and theborate ion [11], the use of a polymer-coated capillary and anon-absorbing running buffer [12], or a Tris buffer withsodium pentanesulfonate as an anionic surfactant [13]. Ingeneral, the main drawback of these methods is again theirlow sensitivity since the LOQs were ca. �50 mg/mL, andtherefore they have been applied to the analysis of bulkpharmaceuticals and their formulations.

Two alternatives have been reported to improve sensitivityin the CE determination of aminoglycoside antibiotics. Oneconsisted of both off-line [14–16, 19, 20] and in-capillary[20] derivatization with 1,2-phthalic dicarboxaldehyde(OPA)/thiol, separation of the derivatives by CZE [14–16,18, 20] or by MEKC [17, 18] and direct UV detection. LODsin the 0.5–2.0 mg/mL range are obtained and therefore themethods are only suitable for the analysis of aminoglyco-sides in bulk pharmaceutical samples. However, sensi-tivity can be improved by using SPE and on-line field-amplified sample stacking, as is the case in the currentdetermination of kanamycin in serum samples [16]. Onlyone reference has been found on the use of fluorescencedetection for the determination of aminoglycoside antibi-otics, namely, the MEKC determination of amikacin inhuman plasma by using 1-methoxycarbonyl-indolizine-3,5-dicarbaldehyde as labeling reagent. Despite thepotentially higher sensitivity of this detection technique,an LOD of only 0.5 mg/mL can be achieved for this antibi-otic [21]. Electrochemical detection is another choice toincrease the sensitivity for the determination of theseantibiotics by CZE [22–24]; however, the methods provideLODs similar to those achieved by using OPA. Recently,potential gradient detection has been proposed for thedetermination of gentamycin components by CZE [25];again, the sensitivity provided by the method only allowsits application to pharmaceutical samples. For these rea-sons, it is clear that the determination of aminoglycosidesby CE demands more sensitive approaches without det-riment to the resolution in order to extend this techniqueto the analysis of more complex analytical samples.

This work develops a simple method for the sensitivesimultaneous determination of aminoglycoside antibiot-ics by MEKC with fluorescence detection. As these ana-lytes have no native fluorescence, they were derivatizedwith sulfoindocyanine succinimidyl ester (Cy5), a veryuseful probe for obtaining amine fluorescent derivativesfor LIF detection by using low-cost diode laser as anexcitation source. Although the background emissionprovided by the excess of Cy5 yielded a small number ofinterfering peaks in the electropherograms, the Cy5-deri-vatized antibiotics overlapped both each other and thebackground emission peaks in the zone modality, andtherefore a BGE containing SDS and an organic modifiersuch as ACN was used to overcome this shortcoming.Finally, the method was applied to the determination ofaminoglycoside antibiotics in bovine milk samples withhigh sensitivity (at mg/kg level) requiring minimal samplepreparation.

2 Materials and methods

2.1 Reagents

All chemicals and solvents used were of analytical-reagent grade and Milli-Q water was used throughout.Neomycin sulfate (minimum 85% neomycin B, remain-der is neomycin C) was purchased from Sigma (Sigma-Aldrich Química, Madrid, Spain), whereas paromomycinI sulfate (�98% purity), kanamycin B sulfate (�99%purity) and amikacin sulfate (�99% purity) wereacquired from Fluka (Sigma-Aldrich Química). Standardsolutions containing 100 mg/mL of each antibiotic wereprepared by dissolving the required amount in Milli-Qwater and storing at 47C in a refrigerator. Standardmixtures were prepared by dilution of the correspondingstandard solutions with the derivatization buffer asrequired. Cy5 of over 97% purity was purchased fromAmersham Biosciences Europe (Barcelona, Spain). A2.5-mM stock labeling solution was made by dissolving1 mg of the chemical in 500 mL of chromatographicgrade DMF (Merck, Darmstadt, Germany), and storingin a freezer. Other solvents and chemicals were pur-chased from Romil Chemicals (Cambridge, UK) andMerck, respectively.

2.2 Derivatization procedure

To 90 mL of the standard solution containing a mixture ofthe antibiotics at a concentration between 10 and 500mg/L and prepared in the derivatization buffer (10 mMsodium borate at pH 8.3, adjusted with 50 mM hydro-chloric acid), 10 mL of 400 mM Cy5 solution in DMF was


Electrophoresis 2006, 27, 4703–4710 CE and CEC 4705

added in a 1.5-PTFE vial. The mixture vial was capped,homogenized and allowed to react in the dark at roomtemperature for 30 min. After derivatization, the vialswere kept at 2207C until analysis.

2.3 MEKC

Experiments were carried out using a Beckman P/ACE5500 CE system (Beckman, Fullerton, CA, USA) equip-ped with a Beckman Laser Module 635-nm LIF detectorand interfaced with a PC-Pentium 75-MHz compatiblecomputer. System Gold software (Beckman) was usedfor control of the instrument and data collection. MEKCseparation was performed on fused-silica capillaries(50 mm id, total length 57 cm, effective length 50 cm)purchased from Beckman. Samples were introducedinto the capillary by hydrodynamic injection, using apressure of 0.5 psi (3.45 kPa) for 5 s. The separation wasperformed at 25.0 6 0.17C using a running voltage of15 kV. At the beginning of each session, the capillarywas rinsed with 0.1 M NaOH for 5 min, followed by5 min with Milli-Q water, and then equilibrated with theBGE (35 mM sodium borate adjusted to pH 9.2, 55 mMSDS and 20% ACN) for 2 min. Between runs, the capil-lary was rinsed with 0.1 M NaOH for 5 min followed by5 min with Milli-Q water. Post-run analysis of data wasperformed using the System Gold chromatography datasystem.

2.4 Milk sample preparation

Milk samples (10 mL) spiked with 1.0 mL of water con-taining 20–1000 ng of antibiotics were added to 50-mLcentrifugation tubes. The spiked samples were allowedto air (2 h) at room temperature with shaking every30 min for homogeneity. They were then vortex-mixedwith 10 mL of 30% m/v TCA adjusted to pH 2.0 with0.5 M hydrochloric acid and the resulting mixture cen-trifuged for 20 min at 5000 rpm. The supernatant wasseparated by filtration and the residue washed with theTCA solution and filtrated again. The resulting solutionwas loaded (flow rate, 5.0 mL/min) onto an SPE col-umn packed with 200 mg of weakly Acidic AmberliteIRC-50 (Fluka) exchange material. Initially the columnwas conditioned by passing the TCA solution for10 min at a flow rate of 0.5 mL/min. The analytes wereeluted in countercurrent with 1 mL of 20 mM sodiumborate adjusted to pH 11.5 with sodium hydroxide.Finally, the eluate was diluted with Milli-Q water andadjusted to pH 8.3 in a 2.0-mL volumetric flask, afterwhich 90 mL of this solution was derivatized as statedin Section 2.2.

3 Results and discussion

Figure 1 shows the structures of the aminoglycosideantibiotics studied in this work, which correspond to twoof their representative families. Thus, glycosylation of the4- and the 6-hydroxyl groups of the 2-deoxystreptaminecore with various aminosugars yields the kanamycinfamily (kanamycin B and amikacin), whereas glycosyla-tion of the 4- and 5-hydroxyl groups characterizes theneomycin family (neomycin B and paromomycin I). Ascheme of their labeling reactions with Cy5 is also in-cluded. The number in parenthesis in the antibiotic struc-tures denotes the respective pKa of the amine groupreported in the literature, which were determined by 15NNMR spectroscopy for paromomycin I and neomycin B[26], by CE-MS/MS for amakacin [27] and by potentio-metric titration for kanamycin B [28].

3.1 Optimization of the derivatizationconditions

The pH and the concentration of the derivatization boratebuffer are two key variables affecting the labeling reactionand fluorescence quantum yields. Thus, the pH stronglyinfluences the basicity of the amine groups of the antibi-otics and the hydrolytic degradation of the Cy5, and bothvariables are closely related to the possible formation ofnegatively charged complexes between the tetra-hydroxyborate ion and hydroxyl groups of aminoglyco-side [29]. Thus, the effect of the pH and the concentrationof the borate buffer were studied over the range 7.5–9.0and 5–100 mM, respectively, and the results are shown inFig. 2. As can be seen in Fig. 2A, the apparent pH meas-ured in the derivatization medium (a mixed aqueous-organic solution containing 10% of DMF, see Section 2.2)was found to markedly affect the peak height, providing apH of ca. 8.8 as the maximum fluorescence signal in allcases, and therefore it was selected for further experi-ments. By comparing the pKa values for the amine groupsof the antibiotics (see Fig. 1) and the selected pH, it canbe inferred that the majority of the amine groups in theantibiotics are deprotonated, and therefore the depend-ence observed at pH over 8.8 can be ascribed to thehydrolytic degradation of the Cy5 [30]. As can be seen inFig. 2B, the concentration of the derivatization buffer alsoexerts a significant influence on the LIF signal. In fact, atborate buffer concentrations over 10 mM, the analyticalsignal strongly decreases, which can be ascribed to theformation of aminoglycoside-borate complexes. As stat-ed above, these complexes have a negative charge andtherefore an electrostatic repulsion can be expected intheir reaction with the label, with the subsequent de-crease in the analytical signal. Based on these results, the



Figure 1. Structures of the fouraminoglycoside antibioticsassayed and scheme of thelabeling reaction with Cy5.Number in parentheses denotesthe respective acid dissociationconstant of the amine groups.

pH in the derivatization medium was adjusted by adding90 mL of 10 mM borate buffer (pH 8.3) to the PTFE reac-tion vial.

Time and temperature, two important variables in thelabeling reaction, were investigated over the ranges 25–557C and 20–75 min, respectively. As in a previous studyon labeling phosphorus-containing amino acid herbicideswith Cy5 [30], the maximum analytical signal wasobtained at room temperature with 30 min of reactiontime. Finally, the influence of the concentration of thederivatized reagent was studied at levels up to 250 mM,which correspond to the addition of 10 mL of 2.5 mM Cy5in the reaction vial (see Section 2.2). The signal increasedon increasing the concentration up to ca. 35 mM and thenleveled off up to the maximum concentration tested. Aconcentration of 40 mM was selected as optimal, whichprovides a mole ratio of Cy5 to analytes in the reactionmixture that ranged from ca. 10 to 400 (see dynamic linearranges shown in Table 1).

3.2 MEKC separation conditions

To attain the best separation between the peaks fromlabeled aminoglycoside antibiotics and those provided bythe unreacted Cy5, different electrophoretic conditionswere tested using CZE and MEKC. Initially, CZE studieswere carried out by using borate buffer as BGE and bychanging the pH (8.0–10.0) and the buffer concentration(10–100 mM). From the electropherogram shown inFig. 3A (35 mM borate at pH 9.0), it can be inferred that byusing this simple separation mode of CE, based mainly onthe differences between charge-to-mass ratios of thederivatives, it is impossible to resolve the addressedmixture: the kanamycin B and paromomycin I peaksmigrate as a single peak and the excess of Cy5 overlapswith that from the neomycin B derivative. In order toimprove resolution, the most frequently used organicmodifiers, such as methanol, 2-propanol and ACN, wereadded to the BGE at a concentration range between 5



Figure 2. Effect of (A) pH and(B) borate buffer concentrationon Cy5 labeling efficiency forparomomycin I (s), kanamycinB (h), amikacin (d) and neomy-cin B (n). All other conditions asin Section 2.

Table 1. Features of the calibration plots and analytical figures of merit for the determination oflabeled aminoglycoside antibiotics with Cy5 by MEKC and diode laser detection

Aminoglycoside Linear range(mg/L)

Regression equationa) LOD(mg/L)

RSD(%)b)

Kanamycin B 10–250 H = (0.1 6 0.2) 1 (0.91 6 0.02)6C 3.0 4.2Amikacin 15–250 H = (0.2 6 0.4) 1 (0.63 6 0.03)6C 4.5 4.5Neomycin B 20–500 H = (0.2 6 0.4) 1 (0.46 6 0.03)6C 6.0 5.3Paromomycin I 10–250 H = (0.1 6 0.2) 1 (1.02 6 0.02)6C 3.0 3.8

a) H, peak height (in mV); C, analyte concentration (in mg/L); intercept and slope 6 SD (n = 10)b) n = 11 at a aminoglycoside concentration of 50 mg/L (within-day precision)

and 20% each. From experimental results it can be con-cluded that only ACN at 20% provided a slightlyenhancement in the selectivity, since the kanamycin Band paromomycin I peaks were partially resolved (seeFig. 3B).

To further enhance selectivity, several MEKC alternativeswere studied based on the use of anionic (SDS), non-ionic(Triton X-100) and mixed (SDS/Triton X-100) micelles aswell as microemulsions (SDS/octane/1-butanol), in allcases in the absence and in the presence of ACN as theorganic modifier. After many experiments, and as can beseen in Fig. 3D, a BGE composed of 35 mM borate(pH 9.2), 55 mM SDS and 20% of ACN can fully separatethe four labeled analytes from the peaks caused by theCy5 excess. Microemulsion electrokinetic chromatogra-

phy (MEEKC) provides poorer results (see Fig. 3C) be-cause it requires higher analysis times (elution window28–30 min) and some overlapping is observed betweenpeaks; however, as is expected, the results obtained byMEEKC are better than those of the CZE.

The remarkable effect of ACN content and borate con-centration in the BGE for the MEKC separation of theselabeled aminoglycosides is noteworthy, because a de-crease in the content of ACN or an increase in the borateconcentration provides an important loss in resolution. Infact, at the apparent pH measured in the BGE (pH 9.8),the unlabeled amine groups of the aminoglycosides arefully deprotonated and therefore the derivatives initiallyhave a negative charge supplied by the Cy5 moiety, al-though, due to the presence of borate in the BGE, the



Figure 3. Separation of Cy5-derivatized aminoglycosides by using various CE modes: (A) CZE separation, BGE: 35 mMborate (pH 9.0); (B) CZE separation in the presence of an organic modifier, BGE: 35 mM borate (pH 9.0) and 20% ACN; (C)MEEKC separation, BGE: 35 mM borate (pH 9.2), 110 mM SDS; 0.8% n-octane, 6.6% 1-butanol and 20% ACN; and (D)MEKC separation, BGE: 35 mM borate (pH 9.2), 55 mM SDS and 20% ACN. Peak assignment: 1, kanamycin B; 2, amika-cin; 3, neomycin B; and 4, paromomycin I. Further explanations can be found in the text.

formation of negatively charged complexes between thehydroxyl groups of the hexoxe rings of the aminoglyco-sides and borate ion can be assumed with the sub-sequent increase in the negative charge of derivatives. Inthese conditions, an electrostatic repulsion between theanionic derivatives and the SDS micelles can be inferred.However, ACN, due to its solvating capacity, might pre-vent the formation of these complexes (the concentrationof borate in the BGE is only 35 mM), which could favor theinteraction between the labeled aminoglycosides and theSDS micelles.

Finally, the electropherogram in Fig. 3D also shows thatthe kanamycin family of antibiotics assayed (kanamycinB and amikacin) provide two peaks with different sensi-tivity in terms of relative peak heights. Taking intoaccount the purity of the corresponding standards (both�99%, see Section 2.1), these additional peaks can beassigned to the possible formation of two derivatives,the mono- and di-Cy5-labeled aminoglycosides, sincevarious amine groups in the structure of the antibioticcan interact with Cy5 and only one peak is observed

when the labeling reaction is carried out in an excess ofthe antibiotic. Although it is difficult to assign the mobilityof each derivative to a particular structure, it can beassumed that di-substituted derivatives interact to alesser degree with the micelle from electric (highernegative charge) and steric effects, which provideshorter migration times, as is the case of kanamycin B.As a result, the peak with higher migration time (thepossible mono-Cy5-labeled kanamycin B) has highersensitivity and therefore it was used for quantification.Regarding amikacin, due to the fact that both peaksshowed similar sensitivity, the first one was used forquantification for reasons of simplicity.

3.3 Analytical performance characteristics

Under optimum conditions, the performance and reliabil-ity of the proposed method were assessed by determin-ing its analytical figures of merit such as sensitivity (slopeof the calibration graph), linear range, LODs and precision(RSD) for aminoglycosides. Table 1 gives the equations



for the standard curves (correlation coefficients rangingfrom 0.9975 to 0.9992) obtained by plotting the peakheight against analyte concentration, the LODs definedas the minimum analyte concentrations providing elec-trophoretic signals three times higher than peak-to-peaknoise, and the RSDs obtained by measuring 11 samplesspiked with each antibiotic at 50 mg/L concentration level(within-day precision). As can be seen, the proposedmethod allows the determination of these antibiotics atvery low levels (mg/L), all with good precision, viz., RSDsfor peak height from 3.8 to 5.3%.

3.4 Determination of antibiotics in bovine milksamples

The widespread use of antibiotics in dairy cattle man-agement may result in the presence of their residues inmilk, which may cause allergic reactions, interfere withstarter cultures for cheese and other dairy products, orindicate that the milk may have been obtained from ananimal with a serious infection [31]. In order to protect thesafety of the consumer, maximum residue limits (MRLs)for residues of veterinary drugs in foodstuffs of animalorigin have been established by international organiza-tions; for example, the European Medicines Agency [32]imposed MRL values ranging from 100 to 500 mg/kg foraminoglycoside antibiotics. Taking into account the highsensitivity provided by the proposed Cy5 method, itsperformance was assessed for the determination of ami-noglycoside antibiotics in bovine milk samples, whichhave not yet been reported with CE. From the aminogly-cosides studied in this work, kanamycin B, neomycin Band paromomycin I were analyzed in these samples,whereas amikacin was not assayed because this antibi-otic is not included in the list provided by the EuropeanMedicines Agency [32]. The determination of these ami-noglycosides offers additional interest on account of thecross-resistance action between kanamycin, neomycinand paromomycin [32].

As the milk is a complex matrix due to the presence of alipid emulsion, precipitation and separation of the pro-teins are necessary prior to the determination of antibiot-ics. TCA and hydrochloric acids are the typical pre-cipitating agents used for this purpose followed by aclean-up of the supernatant on a C18 SPE column [31]. Asstated in Section 2.4, TCA was used as the precipitatingagent and the antibiotics were cleaned up from thesupernatant on-line by using an amberlite IRC-50 weaklyacidic cation exchanger column instead of the classic C18

one. This SPE system considerably simplifies the overallanalytical procedure since the antibiotics can easily beeluted with 20 mM sodium borate (pH 11.5), which, after itadjustment to pH 8.3, can be directly derivatized with

Cy5. In addition, the proposed procedure provided anenrichment factor of ca. 5.0 with respect to the optimizedone in Milli-Q water.

Finally, to assess matrix effects, bovine milk sampleswere fortified according to the procedure described inSection 2.4. Thus, concentrations of aminoglycosidesbetween 2.0 and 100 mg/kg were spiked to milk samplesand the corresponding calibration plots and extractionrecoveries determined. As can be seen from Fig. 4, theantibiotics were resolved without matrix effect; also, theywere quantitatively recovered with the following averagevalues: 90.1 6 6 (kanamycin B), 93.3 6 7 (neomycin B)and 89.4 6 6 (paromomycin I). These recoveries are bet-ter than those reported in the literature using the C18 SPEcolumn for the clean-up step [31], which is remarkableconsidering the smaller concentration levels of antibioticsused in this study. On the other hand, the proposed pro-cedure allows decreasing of the LODs with respect tothose reported in Table 1 due to the involved pre-concentration step, and therefore LODs over the range0.5–1.5 mg/kg can be achieved. In summary, these resultstestify to the good performance of the proposed methodin the determination of aminoglycosides in this type ofsample.

Figure 4. Electropherograms for aqueous standards(dotted line) and spiked milk sample (solid line) with 5 mg/kg of aminoglycoside. Peak assignment: 1, kanamycin B;3, neomycin B; and 4, paromomycin I. Conditions asdescribed in Section 2.



4 Concluding remarks

In this work, MEKC is shown to be a powerful analyticaltechnique for the sensitive determination of aminoglyco-sides in bovine milk samples at levels clearly lower thanthose specified by the legislation. The simplicity of theproposed method makes it suitable for the routine analy-sis of residues of these antibiotics in milk samples. Thefollowing conclusions can be drawn from its performance:(i) it extends the scope of CE for the determination ofaminoglycoside antibiotics, since little work has beendone in this field; (ii) it develops a derivatization schemefor labeling aminoglycoside antibiotics with Cy5, whichhas not been used previously to label this kind of antibi-otic; (iii) it uses diode laser as an excitation source be-cause it provides significant features in contrast to otherlaser systems [33], such as long operating lifetimes andlow maintenance costs; (iv) it becomes a useful choice forthe determination of aminoglycoside antibiotics in a greatvariety of real samples because of the LODs obtained atlow or sub- microgram-per-liter level (lower than those ofexisting electrophoretic alternatives); and (v) to ourknowledge, this is the first report on the analysis of theseantibiotics by CE in this kind of sample, which is possiblethanks to the high sensitivity and selectivity provided bythe method proposed.

The authors gratefully acknowledge the financial supportprovided by the Spanish Department of Research of theMinistry of Education and Science under the BQU2003–03027. FEDER also provided additional funding.

5 References

[1] Chambers, H. F., in: Hardman J.G., Limbird, L. E., Gilman, A.G. (Eds.), Goodman and Gilman’s The Pharmacological Basisof Therapeutics, 11th ed., McGraw-Hill Inc., New York 2006.

[2] Salisbury, C. D. C., in: Oka, H., Nakazawa, H., Harada, K.,MacNeil, J. D. (Eds.), Chemical Analysis for Antibiotics Usedin Agriculture. AOAC International, Toronto 1995.

[3] Stead, D. A., J. Chromatogr. B 2000, 747, 69–93.[4] Hage, D. S., Anal. Chem. 1999, 71, 294R–304R.[5] Isoherranen, N., Soback, S., J. AOAC Int. 1999, 82, 1017–

1045.[6] Leveque, D., Gallion-Renault, C., Monteil, H., Jehl, F., J.

Chromatogr. A 1998, 815, 163–172.

[7] Tawa, R., Matsunaga, H., Fujimoto, T., J. Chromatogr. A1998, 812, 141–150.

[8] Hernandez, M., Borrull, F., Calull, M., Trends Anal. Chem.2003, 22, 416–427.

[9] Kaale, E., Van Schepdael, A., Roets, E., Hoogmartens, J.,Am. Lab. 2002, 34, 22, 24, 26.

[10] Ackermans, M. T., Everaerts, F. M., Beckers, J. L., J. Chro-matogr. 1992, 606, 229–235.

[11] Flurer, C. L., J. Pharm. Biomed. Anal. 1995, 13, 809–816.[12] Calcara, M., Enea, V., Pricoco, A., Miano, F., J. Pharm.

Biomed. Anal. 2005, 38, 344–348.[13] Yeh, H. H., Lin, S. J., Chou, C. A., Chen, S. H., Electropho-

resis 2005, 26, 947–953.[14] Kaale, E., Van Schepdael, A., Roets, E., Hoogmartens, J., J.

Chromatogr. A 2001, 924, 451–458.[15] Kaale, E., Van Schepdael, A., Roets, E., Hoogmartens, J.,

Electrophoresis 2002, 23, 1695–1701.[16] Long, Y. H., Hernandez, M., Kaale, E., Van Schepdael, A. et

al., J. Chromatogr. B 2003, 784, 255–264.[17] Kaale, E., Leonard, S., Van Schepdael, A., Roets, E. et al., J.

Chromatogr. A 2000, 895, 67–79.[18] Wienen, F., Holzgrabe, U., Chromatographia 2002, 55, 327–

331.[19] Kaale, E., Long, Y. H., Fonge, H. A., Govaerts, C. et al. Elec-

trophoresis 2005, 26, 640–647.[20] Kaale, E., Van Schepdael, A., Roets, E., Hoogmartens, J.,

Electrophoresis 2003, 24, 1119–1125.[21] Oguri, S., Miki, Y., J. Chromatogr. B 1996, 686, 205–210.[22] Fang, X. M., Ye, J. N., Fang, Y. Z., Anal. Chim. Acta 1996,

329, 49–55.[23] Voegel, P. D., Baldwin, R. P., Electroanalysis 1997, 9, 1145–

1151.[24] Yang, W. C., Yu, A. M., Chen, H. Y., J. Chromatogr. A 2001,

905, 309–318.[25] Yuan, L. L., Wei, H. P., Li, S. F. Y., Electrophoresis 2005, 26,

196–201.[26] Malvika, K., Barbieri, C. M., Kerrigan, J. E., Pilch, D. S., J.

Mol. Biol. 2003, 326, 1373–1387.[27] Kane, R. S., Glink, P. T., Chapman, R. G., McDonald, J. C. et

al., Anal. Chem. 2001, 73, 4028–4036.[28] Jezewska, M., Bal, W., Kozlowski, H., Inorg. Chim.

Acta1998, 275–276, 541–545.[29] Hoffstetter-Kuhn, S., Paulus, A., Gassmann, E., Widmer, H.

M., Anal. Chem. 1991, 63, 1541–1547.[30] Orejuela, E., Silva, M., Electrophoresis 2005, 26, 4478–4485.[31] Schenck, F. J., Callery, P. S., J. Chromatogr. A 1998, 812,

99–109.[32] Committee for Medicinal Products for Veterinary Use, Euro-

pean Medicines Agency, London 2005.[33] Mank, A. J. G., Lingernan, H., Gooijer, C., Trends Anal.

Chem. 1996, 15, 1–11.


Trace analysis of aminoglycoside antibiotics in bovine milk by MEKC with LIF detection

Documents

Transcript of Trace analysis of aminoglycoside antibiotics in bovine milk by MEKC with LIF detection