Op

JD

a

ARRAA

KALSWS

1

tiadebbtatpotst

p

0h

Analytica Chimica Acta 731 (2012) 32– 39

Contents lists available at SciVerse ScienceDirect

Analytica Chimica Acta

jou rn al hom epa ge: www.elsev ier .com/ locate /aca

ptimization of antibiotic analysis in water by solid-phase extraction and higherformance liquid chromatography–mass spectrometry/mass spectrometry

ohn L. Zhou ∗, Khalid Maskaoui, Adeleye Lufadejuepartment of Applied Science, London South Bank University, 103 Borough Road, London SE1 0AA, UK

r t i c l e i n f o

rticle history:eceived 26 February 2012eceived in revised form 2 April 2012ccepted 6 April 2012vailable online 25 April 2012

eywords:ntibioticsC–MS/MS

a b s t r a c t

This paper describes the development of an optimized method based on solid-phase extraction (SPE)followed by liquid chromatography–electrospray ionization tandem mass spectrometry (LC–MS/MS) forthe simultaneous analysis of ten antibiotic compounds including tetracyclines, sulfonamides, macrolidesand quinolones. LC–MS/MS sensitivity has been optimized by alterations to both LC and MS operations. Ofthe two high resolution columns tested, Waters Symmetry C18 endcapped and Agilent Zorbax Bonus-RP,the latter was found to show better performance in producing sharp peaks and clear separation for mostof the target compounds. Optimization of the MS fragmentation collision and cone energy enhanced thepeak areas of the target analytes. The recovery of the target compounds from water samples was most

olid-phase extractionater

ewage

efficient on Waters Oasis HLB SPE cartridge, while methanol was shown to be the most suitable solventfor desorbing the compounds from SPE. In addition, acidification of samples prior to SPE was shownto enhance the recovery of the compounds. To ensure a satisfactory recovery, the flow rate through SPEshould be maintained at ≤10 mL min−1. The method was successfully applied to the analysis of antibioticsfrom environmental water samples, with concentrations being <LOD in tap water, between <LOD to28 ng L−1 in river water and between <LOD to 230 ng L−1 in sewage effluent.

© 2012 Elsevier B.V. All rights reserved.

. Introduction

Of current concern worldwide are antibiotic compounds dueo their widespread use in large quantities [1–3] and potential tonduce bacterial resistance [4]. In addition, antibiotics are regardeds “pseudopersistent” contaminants due to their continual intro-uction to ecosystems [5,6]. The occurrence of antibiotics in thenvironment has therefore received considerable attention. It haseen demonstrated that antibiotics are, in general, poorly absorbedy the human body, and thus are excreted either unchanged orransformed, via urine and faeces [7]. Several studies have shown

relationship between local sales of human pharmaceuticals andheir concentrations detected in the influents of sewage treatmentlants (STPs) [8,9]. The removal efficiency of antibiotics in STPs isften compound dependent [1]. As a result, it is widely reportedhat the raw and inadequately treated effluents from STPs are aignificant point source of antibiotics and other pharmaceuticals in

he aquatic environment [1,2,10,11].

To protect our aquatic environment such as rivers from beingolluted by antibiotic residues, regular monitoring is essential,

∗ Corresponding author. Tel.: +44 2078157933; fax: +44 2078157699.E-mail address: zhouj9@lsbu.ac.uk (J.L. Zhou).

003-2670/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.aca.2012.04.021

which is itself dependent on the availability of reliable and sensi-tive analytical methods. Furthermore, water pollution is often dueto a mixture of pollutants, as in the case of antibiotics, arising fromthe simultaneous occurrence of different classes of antibiotics inwastewater [12]. So far the majority of published methods for theanalysis of antibiotics are focused on specific therapeutic classes[2,13], which can become time consuming and resource intensiveif they are to be extended for environmental applications studyingmultiple residues. Developing a multi-residue method for antibi-otics is therefore rewarding by generating a significant amount ofdatasets which are essential in the assessment of aquatic toxicityof mixtures, i.e. the so-called mixture effects [14].

Currently there have been many methods published for theanalysis of antibiotics, based on gas chromatography–mass spec-trometry (GC–MS) [15] and liquid chromatography–tandem massspectrometry (LC–MS/MS) [16,17], although LC–MS/MS is morewidely adopted due to its advantages over GC–MS including bet-ter selectivity [13] and compatibility with antibiotics with lowvolatility and high polarity. In addition, simultaneous analysis ofseveral groups of compounds with different physicochemical char-acteristics generally requires a careful selection of experimental

conditions. For that purpose it was crucial to optimize a number ofoperating parameters that may improve the chromatographic reso-lution, sensitivity and precision for the target antibiotics in order to

J.L. Zhou et al. / Analytica Chimica Acta 731 (2012) 32– 39 33

Table 1The different types of SPE cartridges being studied for concentrating antibiotics from water.

Cartridge Abbreviation Description Manufacturer

DSC-C18 (0.5 g, 3 mL) SDC Polymerically bonded octadecyl SupelcoDSC-Si (0.5 g, 3 mL) SDS Unbonded acid washed silica sorbent SupelcoDSC-SCX (0.5 g, 3 mL) SCX Aliphatic sulfonic acid, Na+ counterion SupelcoDSC-SAX (0.5 g, 3 mL) SAX Quaternary amine, Cl− counterion SupelcoStrata X-CW (0.5 g, 6 mL) SCW Polymeric weak cation PhenomenexStrata SDB-L (0.2 g, 3 mL) SL Styrene–divinylbenzene polymeric PhenomenexChromabond-Easy (0.2 g, 6 mL) CE Bifunctionally modified

polystyrene–divinylbenzene adsorbent resinMacherey-Nagel

Chromabond-C18 Hydra (0.5 g, 6 mL) CH Octadecyl-modified silica Macherey-NagelChromabond-Drug (0.2 g, 3 mL) CD Modified silica Macherey-Nagel

cyl

droxyivinyl

eTgo

sbtwdcptt

2

2

dRoaaswsospc

TP

Isolute C18 (1 g, 6 mL) IC OctadeIsolute C18/ENV+ (0.4 g, 6 mL) IE C18 hyOasis HLB (0.2 g, 6 mL) Oasis Poly(d

nsure their accurate detection at varying concentrations in water.he analytical methods established will be an important tool touide water companies in their efforts to removal antibiotics andther similar emerging pollutants in STPs.

The aim of this study was therefore to develop a method for theimultaneous determination of 4 therapeutic classes of antibioticsy liquid chromatography–electrospray tandem mass spectrome-ry (LC-ESI–MS/MS) in positive ionization (PI) mode. The objectivesere to ensure the separation of the compounds by varying the gra-ient and composition of the mobile phase and using different LColumns, to optimize the sensitivity of the method by adjustingarameters such as collision energy in tandem MS, and to improvehe recovery during SPE. The method was then ready to be used forhe analysis of environmental water samples.

. Experimental

.1. Reagents and materials

All the solvents used including methanol, acetonitrile andicloromethane, were of distilled-in-glass grade purchased fromathrburn Chemicals Ltd., Walkerburn, Scotland. Formic acid wasf HPLC grade. The antibiotic compounds including tetracyclinend oxytetracycline (tetracyclines), sulfathiazole, sulfamethazinend sulfadiazine (sulfonamides), erythromycin, roxithromycin andpiramycin (macrolides), ofloxacin and norfloxacin (quinolones)ere purchased from Sigma, Dorset, UK. These compounds were

elected based on their production and use worldwide, frequency

f occurrence in the environment, different classes (tetracyclines,ulfonamides, macrolides and quinolones) and hence differenthysicochemical properties [18], and different removal efficien-ies in STPs which depend on their physiochemical proprieties and

able 2hysicochemical properties and LC–MS/MS conditions for the analysis of antibiotics by M

Compound CAS number Formula RT (min) MW pKa

TetracyclinesTetracycline 60-54-8 C22H24N2O8 11.61 444 3.3

Oxytetracycline 79-57-2 C22H24N2O9 11.13 460 3.27SulfonamidesSulfathiazole 72-14-0 C9H9N3O2S2 10.64 255 7.2

Sulfamethazine 57-68-1 C12H14N4O2S 12.33 278 7.59Sulfadiazine 68-35-9 C10H10N4O2S 8.99 250 6.36MacrolidesErythromycin–H2O 114-07-8 C37H65NO12 14.24 715 8.88Roxithromycin 80214-83-1 C41H76N2O15 14.51 836 9.17Spiramycin 8025-81-8 C43H74N2O14 12.40 842

QuinolonesOfloxacin 82419-36-1 C18H20FN3O4 11.07 361

Norfloxacin 70458-94-7 C16H18FN3O3 11.23 319

International sorbent technologylated polystyrene–divinylbenzene International sorbent technologybenzene-co-N-vinylpyrrolidone) Waters

wastewater treatment technology adopted. Erythromycin–H2O,a major degradation product of erythromycin, was obtained byadjusting the pH of an erythromycin solution to 3.0 using 3 MH2SO4, following a method described by McArdell et al. [7].After 4 h of stirring at room temperature complete conversionto erythromycin–H2O was achieved. Stable isotope-labelled inter-nal standards (13C-phenacetin, diuron-d6) were obtained fromCambridge Isotope Laboratories, USA. Separate stock solutions(1000 mg L−1) of individual compounds and internal standardswere prepared by dissolving an appropriate amount of eachsubstance in methanol, which were further diluted before use.Glass fibre filters (GF/F, 0.7 �m) were obtained from Whatman(Maidstone, UK). Stock solutions (1000 mg L−1 in methanol) forall standard substances were prepared, and stored at −20 ◦C.Ultrapure water was supplied by Maxima Unit from USFElga, UK.

2.2. Sample extraction by SPE

The target compounds were extracted from water samples bydifferent SPE cartridges (Table 1). The compounds (100 ng each)were spiked in triplicate in 1 L of ultrapure water for the recov-ery test. All the cartridges were first conditioned with 10 mL ofmethanol, followed by ultrapure water (3 × 5 mL) at a rate of1–2 mL min−1. Then, water samples were extracted at a flow rateof 1–20 mL min−1. After the extraction, the cartridges were driedunder vacuum for 30 min, with the analytes being eluted to 20 mLvials from the sorbents with 10 mL of solvents (e.g. methanol)at a flow rate of 1 mL min−1. The solvents were blown down to

0.2 mL under a gentle flow of nitrogen at less than 40 ◦C. In addi-tion, the effects of water properties (e.g. pH, salinity) and SPE flowrate on antibiotic recovery were assessed. The method developedwas further verified by checking recoveries in natural matrices,

RM using Agilent Zorbax Bonus-RP column.

log Kow Water solubility(mg L−1)

Precursor ion(m/z)

Quantification ions(m/z)

−1.3 231 445 427(100) −0.9 313 461 426(100), 43(30)

0.05 373 256 156(100) 0.89 1500 279 156(100), 204(20) −0.09 77 251 92(100)

3.06 1.44 716 539(100), 522(70) 2.75 837 158.2(100), 697(30)

843 231(100), 422(10)

−0.39 28,300 362 318(100)−1.03 178,000 320 302(100)

3 Chimic

wpcl

2

aat(cewltoe7w

tecdawpiPsaa(pm

3

3

LwLwWcAbgtpsZatpessawa

4 J.L. Zhou et al. / Analytica

here environmental water samples (1 L) were filtered throughre-combusted GF/F filters and spiked with 100 ng of the targetompounds. These samples were then extracted using SPE and ana-yzed by LC–MS/MS.

.3. LC–MS/MS analysis

The LC separation was performed using a Waters 2695 HPLC sep-rations module (Milford, MA, USA). Two LC columns were tested:n Agilent Zorbax Bonus-RP column (2.1 mm i.d. × 150 mm, par-icle size 5 �m), and a Waters Symmetry C18 endcapped column2.1 mm i.d. × 100 mm, particle size 3.5 �m). The mobile phase wasomposed of eluent A (0.1% formic acid in ultrapure water), andluent B (acetonitrile with 0.1% formic acid). The best separationas achieved with a gradient starting with 100% of eluent A, fol-

owed by a 3-min gradient to 90% of eluent A, then a 2-min gradiento 80% of eluent A, and a further 2-min gradient change to 78.5%f eluent A. By 10-min, the mobile phase was changed to 75% ofluent A, which was followed by 4 steps of 5-min gradient to 79%,9.5%, 80% and 100% of eluent A, respectively. The injection volumeas 10 �L.

The tandem MS analyses were carried out on a Micromass Quat-ro triple-quadrupole mass spectrometer equipped with a Z-spraylectrospray interface. The analyses were done in the PI mode for allompounds. The temperatures of the electrospray source block andesolvation were 100 and 300 ◦C, respectively. The capillary volt-ge was 3.0 kV. Nitrogen as both nebulizing gas and desolvation gasas set at 25 and 630 L h−1, respectively. Following the selection ofrecursor ions by the first quadrupole mass analyzer, the collision-

nduced dissociation was performed by argon at 3.6 × 10−3 mbar.roduct ions were obtained at a series of collision energies andelected according to the fragmentation that produced a usefulbundance of product ions. The optimal collision energy, cone volt-ge and transitions chosen for the multiple reaction monitoringMRM) experiment were optimized. A dwell time of 100 ms per ionair was adopted. The mass spectrometer was operated in MRMode with unit mass resolution on both mass analyzers (Table 2).

. Results and discussion

.1. Chromatogram and mass spectra

During method development for multi-residue analysis usingC–MS/MS, the primary drive is to enhance detection sensitivityith a good separation of the compounds, by alterations to both the

C and the MS operations. A series of experiments were conductedith different mobile phase gradients on two HPLC columns, aaters Symmetry C18 endcapped column which has been used suc-

essfully for pharmaceuticals analysis in our early work [19] and angilent Zorbax Bonus-RP column. The latter used a unique com-ination of densely reacted sterically protected, diisopropyl-C14roups covalently bonded through embedded amide functionalityo an ultrapure (>99.995% SiO2; Type B) Zorbax Rx-SIL silica sup-ort. This special silica support is designed to reduce or eliminatetrong interaction of basic and other highly polar compounds. Theorbax Bonus-RP column shows different selectivity from totallylkyl or aryl stationary phases and can be a preferred alterna-ive to such phases for separating basic, acidic and other highlyolar compounds by reverse-phase LC. The Waters Symmetry C18ndcapped column was able to separate all the compounds rea-onably well (Fig. 1(a)). However, the separation in terms of peak

hape had to be improved, especially for norfloxacin, spiramycinnd erythromycin-H2O. In comparison, the antibiotic compoundsere better separated by the Zorbax Bonus-RP column (Fig. 1(b))

lthough for tetracycline and oxytetracycline no improvement was

a Acta 731 (2012) 32– 39

shown. Overall, the Zorbax Bonus-RP column offered better perfor-mance over the Waters Symmetry C18 endcapped column for thegroup of target analytes.

Reverse phase HPLC for antibiotics analyses are predominantlyperformed on C18 columns [13], e.g. Kim and Carlson [20] usedan XTerra MS C18 Column (2.1 mm × 50 mm, 2.5 �m particlesize, endcapped), Chang et al. [21] used a UPLC BEH column(100 mm × 2.1 mm, 1.7 �m particle size), while Xu et al. [22]adopted an ODS-P column (4.6 mm × 250 mm, 3.5 �m particlesize). There are, however, situations in which a conventionalC18 column produces less than optimal chromatography. Due totheir hydrophobic nature, C18 columns have little retention forhydrophilic compounds, while basic compounds often exhibitpeak tailing and highly aqueous conditions can cause inconsistentretention or even phase collapse. Although Chang et al. used ashorter separation time of 6.5 min during the analysis of antibi-otics, they reported peak interference which was especially seriousfor environmental samples [21]. Our method has been optimizedwithin a longer run time of 23 min for ten compounds (Fig. 1),with the aim to allow more compounds to be added to the currentmixture, for a good separation and detection. Moreover, as the tentarget compounds were from four families, this gradient elutionmethod was a good compromise for suitable sample throughputwhile ensuring that all of the compounds were detected in onerun. In comparison, others have adopted a separate gradient foreach antibiotic family [20].

All of the precursor ions in PI mode were the result of theprotonation of the intact, uncharged molecule [M+H]+ (Table 2).Following the selection of the precursor ions, product ions weremonitored at different collision energies (5–45 eV) in MRM for opti-mization. The collision energy for the individual compounds thatgave the peak with the largest area was chosen as the optimumvalue (Fig. 2). The results show that the optimum collision energywas 15 eV for tetracycline, oxytetracycline, ofloxacin, sulfathia-zole, sulfamethazine and erythromycin–H2O, 20 eV for norfloxacin,25 eV for sulfadiazine and roxithromycin, and 35 eV for spiramycinwhich has the highest molecular weight among all the target antibi-otics. Others such as Xu et al. [22] used 25 eV higher energy thanwhat was used in this study.

In addition, cone voltage was optimized for maximum sensi-tivity, by analyzing a standard containing the target compoundsand internal standards. The cone voltage for the individual com-pounds that gave the peak with the largest area was chosen as theoptimum (Fig. 3), which was determined to be 30 V for ofloxacin,norfloxacin, spiramycin, sulfamethazine and roxithromycin, 25 Vfor sulfadiazine and sulfathiazole, 20 V for oxytetracycline, 15 V fortetracycline, and 10 V for erythromycin-H2O.

3.2. Relative response factor (RRF)

In our method, internal standards 13C-phenacetin and diuron-d6were used for quantifying the concentrations of all target com-pounds based on the use of relative response factor (RRF):

RRF = Response of a compound/amount of a compoundResponse of an internal standard/amount of an internal standard

(1)

values of RRF are crucial when analysing environmental samples,as they are used to trace the potential loss or recovery of targetcompounds during the whole or part of a procedure [23]. Ideally RRFshould be a constant, but in reality it may change with time whilerunning environmental samples. To ensure high quality of data, RRFwas calculated for each of the antibiotics. The results show that theRRF of all compounds was stable throughout the duration of 60 h,

with relative standard deviation (RSD) values being less than 10%for all compounds except for oxytetracycline at 17%, for norfloxacinat 16%, for erythromycin–H2O at 14% and for spiramycin at 20%.From RRF stability for the LC–MS/MS response, it can be concluded

J.L. Zhou et al. / Analytica Chimica Acta 731 (2012) 32– 39 35

Time2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD MIxture (10) Roxithromycin 2: MRM of 15 Channels ES+ 837 > 158.2

5.22e7

14.35

STD Mixture (9) Erythromycin-H2OH2O

2: MRM of 15 Channels ES+ 716 > 539

3.65e514.12

STD MIxture (8) Sulfamethazine 2: MRM of 15 Channels ES+

279 > 1562.90e6

13.82

STD Mixture (7) Sulfathiazole 2: MRM of 15 Channels ES+ 256 > 156

4.19e613.36

STD Mixture (6) Sulfadiazine 2: MRM of 15 Channels ES+ 250.9 > 92.2

5.54e5

11.61

Time

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD Mixture (5) Spiramycin 1: MRM of 6 Channels ES+ 843 > 231

2.42e410.09

STD Mixture (4) Norfloxacin 1: MRM of 6 Channels ES+ 320 > 302

7.39e59.90

STD Mixture (3) Tetracycline 1: MRM of 6 Channels ES+ 445 > 427

6.68e59.73

STD Mixture (2) Oxytetracycline 1: MRM of 6 Channels ES+ 461 > 426

2.59e59.69

STD Mixture (1) Ofloxacin 1: MRM of 6 Channels ES+ 362 > 318

7.73e69.28

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (10) Roxithromycin 2: MRM of 15 Channels ES+ 837 > 158.2

5.52e614.51

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (9) Erythrom ycin-H2 O 2: MRM of 15 Channels ES+ 716 > 539

2.11e5

14.24

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (8) Spiramycin 1: MRM of 6 Channels ES+ 843 > 231

6.19e312.40

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (7) Sulfamethazine 2: MRM of 15 Channels ES+ 279 > 156

1.72e6

12.33

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (6) Tetracycline 1: MRM of 6 Channels ES+ 445 > 427

2.18e611.61

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (5) Norfloxacin 1: MRM of 6 Channels ES+ 320 > 302

5.95e511.23

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (4) Oxytetracycline 1: MRM of 6 Channels ES+ 461 > 426

8.82e411.13

Time2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (3) Ofloxacin 1: MRM of 6 Channels ES+ 362 > 318

1.57e611.07

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (2) Sulfathiazole 2: MRM of 15 Channels ES+ 256 > 156

7.19e6

10.64

Time2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

%

0

100

STD C18 (1) Sulfadiazine 2: MRM of 15 Channels ES+ 250.9 > 92.2

1.57e6

8.99

metr

tpi

3

ot

TE

2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50

Fig. 1. Chromatograms for MRM of antibiotics using (a) Waters Sym

hat the selected internal standards combined with their RRF valuesrovide a suitable means for calculating antibiotic concentrations

n environmental samples.

.3. Recovery of antibiotics in water samples by SPE

As SPE is the key step in sample pre-concentration, its recoveryf the antibiotics should be explored fully. In dealing with organicrace analysis, the most commonly used sorbents are porous

able 3ffect of sample acidification on the recovery (average ± standard deviation) of antibiotic

Compound River water River wa

Ofloxacin 28.7 ± 4 61.0 ±

Tetracycline 46.9 ± 1.2 84.7 ±

Oxytetracycline 42.5 ± 3.4 67.2 ±

Norfloxacin 52.3 ± 5 97.4 ±

Spiramycin 19.8 ± 5.1 35.1 ±

Sulfadiazine 73.7 ± 12.7 104.4 ±

Sulfathiazole 87.8 ± 10.5 98.4 ±

Sulfamethazine 49.2 ± 5 75.1 ±

Roxithromycin 103.2 ± 7.7 101.4 ±

Erythromycin–H2O 91.7 ± 11.0 96.4 ±

y C18 endcapped column and (b) Agilent Zorbax Bonus-RP column.

silica particles surface-bonded with C18 or other hydrophobic alkylgroups and polymeric sorbents, such as styrene-divinybenzene andactivated carbon. In addition, hydrophilic functional groups suchas sulfonic acid and N-vinylpyrrolidone are often introduced to thepolymeric sorbents so as to increase water movement, hence mak-

ing the sorbents more compatible with water matrix and moreeffective for sorbing pollutants. In this study, 12 types of car-tridges (Table 1) from 5 different manufacturers were selectedfor the evaluation of extraction efficiency of pharmaceuticals. As

s in river water and STP effluent samples (n = 3).

ter pH 2 Effluent Effluent pH 2

2.4 94.3 ± 5.4 107.7 ± 9.213.0 101.1 ± 17.8 109.6 ± 16.46.3 83.4 ± 7.5 106.1 ± 19.40.7 82.7 ± 8.6 114.5 ± 6.03.0 21.0 ± 15.5 42.2 ± 4.011.6 93.4 ± 11.0 70.7 ± 15.711.4 101.4 ± 10.3 103.8 ± 6.25.4 102.4 ± 21.7 85.3 ± 6.06.7 63.6 ± 5.8 84.2 ± 0.611.1 116.6 ± 14.0 114.1 ± 11.5

36 J.L. Zhou et al. / Analytica Chimica Acta 731 (2012) 32– 39

0

200000

400000

600000

800000

1000000

1200000

1400000

1600000

0

2000000

4000000

6000000

8000000

10000000

12000000

14000000

50454035302520151050

Pea

k a

rea

Pea

k a

rea

Collision energy (eV)

Ofloxacin

Roxithromycin

Sulfathiazole

Sulfamethazine

0

100000

200000

300000

400000

500000

600000

0

50000

100000

150000

200000

250000

300000

350000

400000

45403530252015105

Pea

k A

rea

Pea

k A

rea

Collision energy (eV)

Tetracycline

Oxytetracycline

Norfloxacin

Sulfadiazine

Erythromycin-H2O

Spiramycin

F

miS((Sb6leS(tcatuKapro[

3

eae

0

500000

1000000

1500000

2000000

2500000

3000000

3500000

5045403530252015105

Cone voltage (V)

Pea

k A

rea

0

2000000

4000000

6000000

8000000

10000000

12000000

Pea

k A

rea

Ofloxacin

Sulfathiazole

Sulfamethazine

Roxithromycin

0

100000

200000

300000

400000

500000

600000

5045403530252015105

Pea

k A

rea

0

100000

200000

300000

400000

500000

600000

Pea

k A

rea

Tetracycline

Oxytetracycline

Norfloxacin

Sulfadiazine

Erythromycin-H2O

Spiramycin

ig. 2. The optimal collision energy for MRM of the target antibiotic compounds.

ethanol was suggested to be efficient in eluting polar contam-nants from SPE cartridges [19,24], it was chosen for evaluatingPE performance. As shown in Fig. 4, the recovery of spiramycin1.1–25.2%) was poor from any of the cartridges. Poor recoveries<30%) for all of the compounds were observed on DSC-SAX, DSC-CX and DSC-Si cartridges. DSC-C18 (0.5 g, 3 mL) with polymericallyonded octadecyl, Chromabond-Easy (0.2 g, 6 mL), isolute C18 (1 g,

mL) and C18/ENV+ (0.4 g, 6 mL) showed better but still relativelyow recovery (<60%). For norfloxacin, sulfadiazine, ofloxacin andrythromycin–H2O improved recoveries (>50%) were observed ontrata X-CW (0.5 g, 6 mL), but this cartridge showed poor recovery<40%) for the other compounds such as tetracycline and oxyte-racycline. Of all the cartridges, the Waters Oasis HLB copolymerartridges proved to show the best overall recoveries (51–95%) forll compounds except spiramycin (25%). The Oasis cartridge washerefore chosen as the best SPE sorbent for more detailed eval-ation. A different Oasis product, i.e. Oasis MCX was chosen byasprzyk-Hordern et al. [25] to concentrate personal care productsnd illicit drugs and by Al-Odaini et al. [2] to concentrate humanharmaceuticals and synthetic hormones, respectively from envi-onmental samples although the Oasis HLB has been chosen bythers for endocrine disrupting chemicals and pharmaceuticals10,11,19,24].

.4. Elution of antibiotics from SPE by different solvents

Following the retention of the target compounds by SPE, it isssential that the adsorbates in SPE columns are desorbed as fullys possible, so as to maximize sensitivity of the analysis. As SPExtraction is a surface phenomenon involving the interaction of

Cone voltage (V)

Fig. 3. The optimal cone voltage for MRM of the target antibiotics.

adsorbates and the adsorbent, the recovery of the target com-pounds by SPE is highly dependent on the polarity of the eluents.Acetone, dichloromethane, ethyl acetate, hexane and methanol aseluents were tested for the elution recovery of antibiotics fromOasis HLB SPE cartridges, which were spiked at 100 ng for each com-pound. Of all the solvents, hexane was found to be the least effectivein the elution and recovery of the target compounds (≤32%), whichis due to the relatively polar nature of these compounds (Fig. 5). Incomparison, dichloromethane, ethyl acetate and acetone producedimproved recoveries for a majority of the compounds, varyingbetween 45% and 71%. More complete recoveries (72–93%) wereachieved by using methanol as the elution solvent. As a result,methanol was adopted as the solvent of choice for the simulta-neous desorption of all antibiotics from Oasis SPE cartridges. Theresults are consistent with the adoption of methanol in the elutionof other trace contaminants such as endocrine disrupting chemicalsfrom SPE [19,24].

3.5. Effect of pH and salinity on antibiotics recovery

Natural waters such as river water and STP effluent can have dif-ferent pH values, and the effect of pH on the extraction efficiency ofantibiotics was studied by adjusting the pH value of water sampleswith dilute solutions of sodium hydroxide and hydrochloric acid.Generally, the dissociation and hence speciation of weakly acidic

compounds in aqueous solutions depends on the solution proper-ties, such as its pH value. Acidification of water solution is likelyto decrease the dissociation of weakly acidic analytes, which maylead to increasing extraction efficiency of the target compounds if

J.L. Zhou et al. / Analytica Chimica Acta 731 (2012) 32– 39 37

he pH

oTetsfitse4ro

Fig. 4. Recovery of antibiotics on different SPE cartridges. T

nly the non-dissociated form binds strongly to the SPE cartridges.he results for river water and STP effluent (Table 3) show that thextraction recovery of most compounds was improved substan-ially by acidifying the samples to pH 2, especially for river wateramples. For compounds with relatively high pKa values such as sul-athiazole and erythromycin–H2O, the improvement in recoverys not significant under sample acidification. The only compoundhat showed relatively poor recovery even after acidification ispiramycin. The results are broadly consistent with those by Liu

t al. [24] and Zhang and Zhou [19] during SPE of 4-tert-octylphenol,-nonylphenol and pharmaceuticals, although the improvement inecovery with a reduction in pH is far more pronounced for antibi-tics.

Fig. 5. Recovery of antibiotics from Oasis SPE cartrid

of solution was not adjusted. SPE flow rate = 10 mL min−1.

Environmental monitoring for organic pollutants covers bothfreshwater and marine water, the latter with characteristicallyhigh salt content. In between freshwater and seawater, estu-arine water has varying amount of saltiness depending on thegeographic locations and tides. In order to widen the applica-bility of our method, the potential effect of salt content on therecovery of antibiotics was assessed by repeating recovery exper-iments at salinity values of 0, 7, 14, 21, and 35 g L−1. The effectof salinity on the extraction efficiency of many antibiotics was

statistically insignificant, when the salinity was varied between0 g L−1 and 21 g L−1. However, when the salinity was increasedfrom 0 g L−1 to 35 g L−1, there was a significant increase (≥12%) inthe recovery of these compounds (P ≤ 0.05). It is well known that

ges by eluting with different organic solvents.

38 J.L. Zhou et al. / Analytica Chimica Acta 731 (2012) 32– 39

0

20

40

60

80

100

120

2520151050

Ofloxacin

Oxytetracycline

Tetracycline

Norfloxacin

Spiramycin

Sulfadiazine

Sulfathiazole

Sulfamethazine

Erythromycin-H2O

Roxithromycin

Flow rate (ml/min)

Rec

over

y (

%)

Fig. 6. Effect of water flow rate on the recovery of antibiotics in SPE. The pH ofsolution was adjusted to 2.

twintprl

3

pbcfarliatwccrp

Table 4Method limit of detection (ng L−1) for the target antibiotics in different watersamples.

Compound Ultrapure water River water Effluent

Ofloxacin 0.3 1.2 2Tetracycline 0.5 1.5 2Oxytetracycline 1.0 1.8 4Norfloxacin 0.4 1.5 3Spiramycin 0.8 3 5Sulfadiazine 0.1 0.4 1Sulfathiazole 0.2 0.4 1

TD

he aqueous solubility of organic compounds in general decreasesith increasing salt concentration, thus their extraction efficiency

n SPE is likely to increase. However, the presence of salts hado significant effect on tetracycline and oxytetracycline. Withhe sustained recovery for saline samples, the method has theotential to be applicable to estuarine and coastal marine envi-onment where our knowledge of antibiotics pollution is veryimited.

.6. Effect of SPE flow rate on antibiotic recovery

SPE is a key step in the recovery of antibiotics from water sam-les, and often it is also the most time consuming. It is highlyeneficial to understand the kinetics of antibiotic sorption to SPEartridge, so that the relevant samples can be extracted at an asast flow rate as possible. In this work the sorption kinetics weressessed by running recovery experiments under different flowates from 1 mL min−1 to 20 mL min−1. The results (Fig. 6) show thatow flow rates (≤10 mL min−1) favoured the recovery of antibioticsn excess of 70% with the exception of spiramycin, with an over-ll average recovery of >80%. When the flow rate was increased

−1 −1

o more than 10 mL min , e.g. 15 mL min , the average recoveryas reduced to 73.7%. Further increase in flow rate to 20 mL min−1

aused a significant reduction in recovery, e.g. the recover of tetra-ycline was reduced to only 48%. As a result, it is recommended toestrict the flow rates to 10 mL min−1 as a maximum during the SPErocedure.

able 5etection rates and concentrations of the target antibiotics in tap water, river water and

Compound Tap water (n = 28) River water (n

Detection rate (%) Concentration (ng L−1) Detection rate

Ofloxacin 0 <LOD 19

Tetracycline 0 <LOD 27

Oxytetracycline 0 <LOD 43

Norfloxacin 0 <LOD 28

Spiramycin 0 <LOD 9

Sulfadiazine 0 <LOD 17

Sulfathiazole 0 <LOD 37

Sulfamethazine 0 <LOD 10

Roxithromycin 0 <LOD 41

Erythromycin–H2O 0 <LOD 69

Sulfamethazine 0.2 0.5 1Roxithromycin 0.1 0.4 1Erythromycin–H2O 0.1 0.4 1

3.7. Validation and application of the method

The method was validated for linearity, accuracy, precision, andlimit of detection (LOD). The linear range was determined by usingultrapure water spiked with increasing amount of the analytesin the concentration range 0–10,000 ng L−1 and fixed amount ofthe internal standards (100 ng L−1). Calibration curves were pre-pared by plotting the ratio of compound peak area to that of theinternal standard versus the compound concentration. The linearrange was found to be between 10 and 2000 ng L−1, with r2 valuesfrom 0.983 to 0.997. Accuracy was tested by further spiking exper-iments using natural water samples. The recoveries were in therange of 75–98% for all compounds, except for spiramycin at lessthan 40%. Good precision of the measurements was demonstratedby the SD values which were less than 15% for all compounds inboth river water and sewage effluent samples. In addition, intra-and inter-day variability of the method was examined, by ana-lyzing replicates of ultrapure water sample extracts. The resultsfrom those analyses showed that RSD values were in the range of4.3–9.8% for runs between days, and in the range of 3.3–6.6% forruns within the same day, at 100 ng L−1. The LOD, defined as theconcentration that corresponds to three times the standard devi-ation of blanks, was calculated by integrating peak area for eachanalyte in 10 independent analyses of ultrapure water, river waterand effluent as blank. As shown in Table 4, LOD values varied from0.1 to 1 ng L−1 in ultrapure water, 0.4 to 3 ng L−1 in river water,and 1 to 5 ng L−1 in effluent samples; they represent an improve-ment over existing methods, many of which with LOD ≥ 10 ng L−1

as reviewed by Petrovic et al. [13]. Specifically, the LOD values aresimilar to or lower than those reported by Miao et al. [26], Xu et al.[22] and Kasprzyk-Hordern et al. [25], therefore confirming thatgood sensitivity has been achieved.

The method was applied to the analysis of antibiotics in 28 sam-

ples of tap water, 71 samples of river water and 50 samples of STPeffluent from Sussex County, UK, with significantly higher levelsdetected in effluent samples (Table 5). For river water samples, themaximum concentration of sulfathiazole (1 ng L−1) is similar to that

STP effluent samples.

= 71) STP Effluent (n = 50)

(%) Concentration (ng L−1) Detection rate (%) Concentration (ng L−1)

<LOD-12 50 <LOD-28<LOD-5 56 <LOD-17<LOD-3 67 <LOD-34<LOD-23 46 <LOD-77<LOD-4 19 <LOD-7<LOD-3 28 <LOD-6<LOD-1 56 <LOD-5<LOD-2 21 <LOD-5<LOD-5 60 <LOD-11<LOD-28 78 <LOD-230

Chimic

dtfaT((ftm1oaadtwta

4

oceolatetpmae1bswdt

[

[

[[

[

[[[[

[[[[[

[[

J.L. Zhou et al. / Analytica

etected in Japan at 6.6 ng L−1 [21]; and the maximum concentra-ion of erythromycin–H2O (28 ng L−1) is similar to that of 40 ng L−1

ound in River Taff, Wales [25], and of 57 ng L−1 in UK rivers [27]lthough higher than 4 ng L−1 found in the Seine River, France [28].he maximum concentration of erythromycin–H2O in STP effluents230 ng L−1) is also similar to that reported elsewhere in the UK130–180 ng L−1) [25], although higher levels (>800 ng L−1) wereound in effluents from Wales and Canada [25,26]. The concentra-ions of roxithromycin in effluents (<LOD-11 ng L−1) are in the same

agnitude as those detected in effluents from Hong Kong at 2.9 and4.2 ng L−1 [12] and those from Canada at 18 ng L−1 [26]. In addition,ther antibiotics such as norfloxacin and sulfadiazine are presentt similar levels to those found in effluents from Hong Kong [12],lthough the maximum concentration of ofloxacin (556.4 ng L−1)etected in Hong Kong effluent is an order of magnitude higher thanhat found in Sussex effluents. Similar concentration of norfloacinas also found in the Seine River, France [28]. The results confirm

hat the chosen antibiotics are widely used in the UK, and that STPsre a key point source of antibiotics in surface water environment.

. Conclusions

A method has been developed for the simultaneous analysisf ten antibiotics including tetracycline, oxytetracycline (tetracy-lines), sulfathiazole, sulfamethazine, sulfadiazine (sulfonamides),rythromycin–H2O, roxithromycine, spiramycin (macrolides),floxacin, norfloxacin (quinolones) using SPE–LC–MS/MS. An Agi-ent Zorbax Bonus-RP column has been found to be more suitablend efficient for separating the antibiotics than a Waters Symme-ry C18 endcapped column. The results show that both the collisionnergy and cone voltage are two key parameters in the optimiza-ion of tandem MS analysis of antibiotics. In terms of water samplereparation by SPE, the Oasis HLB cartridge was shown to be theost suitable for antibiotics. To obtain a satisfactory elution of

ntibiotics from SPE, the solvent methanol was found to be the mostffective. In addition, the flow rate through SPE should not exceed0 mL min−1 to ensure sufficient retention of antibiotics on SPEefore elution. Acidification of water samples to pH 2 was demon-

trated to improve the retention of antibiotics on SPE, for both riverater and effluent samples. Several antibiotic compounds wereetected in the UK river and sewage effluent samples, althoughhey were not detected in tap water supply.

[

[[

a Acta 731 (2012) 32– 39 39

Acknowledgement

The project was funded by a Marie Curie Fellowship within the6th Framework Programme of the European Commission (ContractNo. MIF1-CT-2006-021556).

References

[1] K. Kümmerer, Chemosphere 75 (2009) 417–434.[2] N.A. Al-Odaini, M.P. Zakaria, M.I. Yaziz, S. Surif, J. Chromatogr. A 1217 (2010)

6791–6806.[3] Y. Yang, J. Fu, H. Peng, L. Hou, M. Liu, J.L. Zhou, J. Hazard. Mater. 190 (2011)

588–596.[4] G. Dantas, M.O.A. Sommer, R.D. Oluwasegun, G.M. Church, Science 320 (2008)

100–103.[5] B.J. Richardson, P.K.S. Lam, M. Martin, Mar. Pollut. Bull. 50 (2005) (2005)

913–920.[6] M.D. Hernando, M. Mezcua, A.R. Fernandez-Alba, D. Barceló, Talanta 69 (2006)

334–342.[7] C.S. McArdell, E. Molnar, M.J. Suter, W. Giger, Environ. Sci. Technol. 37 (2003)

5479–5486.[8] B. Halling-Sorensen, S. Nors Nielson, P.F. Lanzky, F. Ingerslev, H.C. Holten

Lutzhoft, S.E. Jorgensen, Chemosphere 36 (1998) 357–393.[9] A. Gobel, A. Thomsen, C.S. McArdell, A. Joss, W. Giger, Environ. Sci. Technol. 39

(2005) 3981–3989.10] J.L. Zhou, Z.L. Zhang, E. Banks, D. Grover, J.Q. Jiang, J. Hazard. Mater. 166 (2009)

655–661.11] D. Grover, J.L. Zhou, P.E. Frickers, J.W. Readman, J. Hazard. Mater. 185 (2011)

1005–1011.12] B. Li, T. Zhang, Z. Xu, H.H.P. Fang, Anal. Chim. Acta 645 (2009) 64–72.13] M. Petrovic, M.D. Hernando, M.S. Díaz-Cruz, D. Barceló, J. Chromatogr. A 1067

(2005) 1–14.14] F. Pomati, C. Orlandi, M. Clerici, F. Luciani, E. Zuccato, Toxicol. Sci. 102 (2008)

129–137.15] W. Lin, H. Chen, W. Ding, J. Chromatogr. A 1065 (2005) 279–285.16] M. Gros, M. Petrovic, D. Barceló, Anal. Bioanal. Chem. 386 (2006) 941–952.17] H. Shaaban, T. Górecki, Anal. Chim. Acta 702 (2011) 136–143.18] The U.S. National Library of Medicine, 2012, website www.nlm.nih.gov

(accessed 3.03.2012).19] Z.L. Zhang, J.L. Zhou, J. Chromatogr. A 1154 (2007) 205–213.20] S.-C. Kim, K. Carlson, Environ. Sci. Technol. 41 (2007) 50–57.21] H. Chang, J. Hu, M. Asami, S. Kunikane, J. Chromatogr. A 1190 (2008) 390–393.22] W. Xu, G. Zhang, S. Zou, X. Li, Y. Liu, Environ. Pollut. 145 (2007) 672–679.23] Z. Zhang, D.P. Grover, J.L. Zhou, Monitoring of pharmaceutical residues in

sewage effluents, in: S. Ahuja (Ed.), Handbook of Water Purity and Quality,Academic Press, 2009, pp. 315–342.

24] R. Liu, J.L. Zhou, A. Wilding, J. Chromatogr. A 1022 (2004) 179–189.25] B. Kasprzyk-Hordern, R.M. Dinsdale, A.J. Guwy, Anal. Bioanal. Chem. 391 (2008)

1293–1308.

26] X. Miao, F. Bishay, M. Chen, C.D. Metacalfe, Environ. Sci. Technol. 38 (2004)

3533–3541.27] M.J. Hilton, K.V. Thomas, J. Chromatogr. A 1015 (2003) 129–141.28] Q.T. Dinh, F. Alliot, E. Moreau-Guigon, J. Eurin, M. Chevreuil, P. Labadie, Talanta

85 (2011) 1238–1245.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具