J. Sep. Sci. 2013, 36, 1455–1462 1455

Peilong Wang1,3

Hongxia Zhu2

Wei Zhang1,3

Zhihua Ye1,3

Ruohua Zhu2

Xiaoou Su1,3∗

1Key Laboratory of Agro-productSafety and Quality, Ministry ofAgriculture, Beijing,P. R. China

2Department of Chemistry,Capital Normal University,Beijing, P. R. China

3Institute of Quality Standards &Testing Technology forAgro-Products, ChineseAcademy of AgriculturalSciences, Beijing, P. R. China

Received October 31, 2012Revised January 29, 2013Accepted January 30, 2013

Research Article

Synthesis of ractopamine molecularlyimprinted membrane and its application inthe rapid determination of three �-agonistsin porcine urine samples

A novel molecularly imprinted membrane (MIM) with ractopamine (RAC) as the templateand the hydrophilic PVDF membrane as the support was synthesized for the selectiveabsorption of RAC and its structure analogues. The absorption behavior and selectivity of theMIM were studied. The experimental results showed that the MIM had the good selectivityto three �-agonists including RAC, RIT, and formoterol (FOM) than that of nonimprintedmembrane. The adsorption capacity for three compounds was above 1.88 �g/cm2 of permembrane. Based on the clean-up and enrichment of porcine urine samples with theMIM, a sensitive determination method of three �-agonists in porcine urine samples byusing MIM followed ultra performance chromatography coupled MS/MS detection wasdeveloped. The LOD and LOQ for RAC, RIT, and FOM were below 0.006 and 0.02 ng/mL,respectively. The mean recoveries, repeatability, and reproducibility of three compounds inporcine urine samples varied from 67.9 to 86.3%, from 3.3 to 10.8%, and from 5.3 to 8.5%,respectively. The presented method was applied to test 50 real porcine urine samples. It wasdemonstrated to be more sensitive and robust for the determination of RAC, RIT, and FOMin porcine urine.

Keywords: Molecularly imprinted membrane / Porcine urine / RactopamineDOI 10.1002/jssc.201201014

� Additional supporting information may be found in the online version of this articleat the publisher’s web-site

1 Introduction

Ractopamine (RAC) is a �-adrenergic agonist and a leanness-enhancing feed additive to increase lean muscle growth andprotein accretion [1, 2]. Since the use of clenbuterol in pigfeeds has been strictly prohibited, as a replacement RAC hasbecome one of the most often used additives in pig feeds.However, similarly to clenbuterol, RAC could be accumulatedin pig bodies. Meat products obtained from treated animalsmay pose a potential risk for consumer health, resulting insome harmful effects on human health including headache,dizziness, chest distress, heart palpitations, and limb numb-ness [2,3]. Therefore, RAC has been banned to use as the feed

Correspondence: Professor Ruohua Zhu, Department of Chem-istry, Capital Normal University, Beijing 100048, P. R. ChinaE-mail: zhurh@mail.cnu.edu.cnFax: +86-10-68903047

Abbreviations: AA, acrylamide; AIBN, 2,2-azo-bis-iso-butyronitrile; FOM, formoterol; MAA, methyl methacrylicacid; MET, metoprolol; MIM, molecularly imprinted mem-brane; MIP, molecularly imprinted polymer; NAD, nadolol;NIM, nonmolecularly imprinted membrane; RAC, rac-topamine; RIT, ritodrine

additive in European Union and China [4]. Recently, somealternative growth promoters such as ritodrine (RIT) and for-moterol (FOM) that also belong to �-adrenergic agonist andof the similar structure and function to RAC were found inpork or porcine urine samples. It is of great significance todevelop a convenient and selective method for the extractionand determination of RAC and its analogues in animal urinesamples to ensure the food safety.

To monitor and control illegal use of RAC and its ana-logues in animals breeding, many methods for the deter-mination of these drugs in biological matrixes includingHPLC [5, 6], GC-MS [7], HPLC-MS [8, 9], electrochemical de-tection [10, 11], colloidal gold immuno-assay [12], and ELISA[13, 14] have been developed. Because biological samples arevery complex, several sample pretreatment methods were re-quired. To analysis of RAC, clean-up methods usually usedinclude liquid–liquid extraction [15], SPE [4, 9], and ion-pairsupercritical fluid extraction [16] for samples of animal urine,meat, liver, etc.

Molecular imprinting has been known as a polymeriza-tion technique to prepare synthetic polymers with recognition

∗Additional correspondence: Professor Xiaoou Su,E-mail: suxiaoou@caas. net.cn

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

1456 P. Wang et al. J. Sep. Sci. 2013, 36, 1455–1462

sites for a given molecule. Molecularly imprinted polymer(MIP) has been used in different ways such as SPE [17], chro-matographic separation materials [18], recognition units [19]for the extraction and detection of RAC, and relative com-pounds from complex biological samples. Wang et al. [20]prepared a RAC–MIP layer on the surface of silica gel with asol-gel process. An online SPE-HPLC with the MIP modifiedsilica gel as the extraction material was established and ap-plied to the analysis of RAC in pork. The LOD for RAC wasobtained as 4.6 ng/mL. Zhang et al. [21] synthesized the MIPusing RAC as the template and acryl amide as the monomerby using bulk polymerization. The binding capacity of thepolymer toward RAC was about 2.57 mg of RAC per gram ofpolymer. In the spiked samples with different spiked levels,the mean recoveries of analyte on the MIP were >90%. Tanget al. [22] prepared a novel MIP by bulk polymerization forthe separation and concentration of RAC by a covalent im-printing approach. The results of the competitive adsorptiontest showed that the MIPs had specific recognition ability forthe analyte RAC. Under the optimum experimental condi-tions, the LOD of proposed method was 0.01 ng/mL. Xu et al.[23] coated the stir bar with RAC–MIP. The thickness of theMIP coating on the stir bar could be controlled and the coatedMIP stir bar was highly selective for the extraction of clen-buterol, RAC, and isoxsuprine. The same research group pub-lished their work recently on the synthesis and application ofmagnetic RAC–MIP beads [24]. The binding sites of MIPand nonmolecularly imprinted polymer were measured, andthe MIP beads showed the high selectivity for RAC, isox-suprine, and fenoterol. Moreover, the magnetic beads madethe sample clean-up procedure more practical. Furthermore,the MIP was used as sample preparation in the rapid screenmethod for RAC. Wang et al. [25] coated MIP against RAC onthe stir bar to selectively extract and purify RAC in real sam-ples. Based on the electro-chemiluminescence signal changes

of Ru(bpy

)2+3

–2-(dibutylamino) ethanol system, a detectionmethod for RAC in pork has been established with the de-tection limit is 3.5 × 10−12 mol/L. The MIP-coated stir barsimply the sample preparation procedure in complex sampleand the developed method was better than other previouslyreported methods on detection limit. In addition, the MIPwas used as recognition materials in the developed sensor torapidly detect RAC in complex biological samples [26, 27].

Taking the commercially available microporous mem-brane as the support, MIP layer can be grafted on the surfaceof the membrane to form molecularly imprinted membrane(MIM) [28]. Cellulose acetate [29], nylon [30], and PVDF [31]were used as the supporting membrane. The commercialmembrane filters have the uniform size and composition,well-controlled porosity, good chemical stability, and resistantto the certain organic solvents. MIMs can keep the advan-tages of the commercial membranes mentioned above andprovide good selectivity. Comparing with the MIP of bulkpolymerization, the MIMs are more convenient and robustto operate in the clean-up process without sieving MIP andpacking the cartridge. Meanwhile, the MIM could provide thelarger specific surface area and less consumption of chemi-

cal reagents. Renkecz et al. [32] synthesized MIPs in 24-wellglass fiber membrane filter plates to obtain a novel type ofSPE device for the clean-up of propranolol, but the selectivityof the synthetic membrane was poor. So far the MIP modifiedthe porous membrane for selective extraction of RAC and itsanalogues in complex samples were not reported.

In this work, the hydrophilic PVDF membrane was se-lected as the support membrane and RAC was used as thetemplate molecule. By thermal polymerization, the RAC–MIP layer was cross-linked on the surface of the PVDF mem-brane. The synthesized RAC–MIM showed the good selec-tivity for RAC and its structure analogue RIT and FOM. TheRAC–MIM could be directly used for extraction and clean-up of target analytes in complex sample matrix without fur-ther process. With ultra-performance liquid chromatography(UPLC)-MS/MS detection, RAC, RIT, and FOM in porcineurine samples could be detected in sub ng/mL. The sensitiv-ity and recovery of the developed method for determinationof RAC, RIT, and FOM in porcine urine samples were verysatisfactory.

2 Experimental procedures

2.1 Chemicals and regents

Hydrophilic PVDF milli pore filters (� 13 mm) with poresize of 0.45 �m were purchased from Agela Technolo-gies (Tianjin). Nylon (� 13 mm, 0.45 �m) and PVDFmembranes (� 13 mm, 0.45 �m) were obtained fromBeijing Dilang Bio-Chem Technologies. RAC were purchasedfrom J & K Reagent (Beijing, China). RIT, FOM, metopro-lol (MET), and nadolol (NAD) were from Dr. Ehrenstor-fer, Augsburg, Germany. Methacrylic acid (MAA), acryli-camide (AAM), and ethylene glycol dimethacrylate were pur-chased from Sigma-Aldrich (St. Louis, MO, USA). 2,2-Azo-bis-iso-butyronitrile (AIBN), which was recrystallized withmethanol before use, were purchased from Fu Chen Chemi-cal Reagent Factory (Tianjin, China). Methanol, chloroform,ACN, glacial acetic acid, acetone, methylbenzene with ana-lytical grade were purchased from Beijing Chemical Works(Beijing, China). Methanol used as HPLC mobile phase wasof HPLC grade and obtained from Merck (Darmstadt, Ger-many). Ultra pure water (18.2 M�/cm) obtained from a Mil-lipore ultra pure water system (Billerica, MA, USA) was usedto prepare buffers and aqueous solutions.

2.2 Instrumentation

Waters E2695 HPLC (including Waters 600 pump, Waters2996 photo-electricity diode array detector, Waters 2998 fluo-rescence detector, and empower chromatographic work sta-tion) was used for the evaluation of RAC–MIM. A XTerraMS C18 column (150 × 3.9 mm, 5 �m, Waters) was usedfor the separation. The column temperature was kept at 28�Cduring the separation. The mobile phase was prepared as1.0 mmol/L NH4Ac solvent in ACN (82:18, v/v). The flow

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

J. Sep. Sci. 2013, 36, 1455–1462 Sample Preparation 1457

rate was 0.8 mL/min and sampling volume 5.0 �L. The fluo-rescence detector was used to determine the RAC, RIT, andNAD, the excitation and emission wavelengths were 275 and306 nm for RAC and RIT and 230 and 300 nm for NAD. ForFOM and MET, the UV detector was used and measurementwavelength was 212.0 and 224.0 nm, respectively.

The determination method for RAC, RIT, and FOM wasdeveloped on a Waters Acquity UPLC coupled Xevo TQ tan-dem MS (Milford, MA, USA). An acquity BEH C18 LC sepa-ration column (100 × 2.1 mm, 1.7 �m particle size) was usedto separate target analytes and the column temperature main-tained at 25�C. The mobile phase consisted of solvent A (0.1%formic acid in water) and solvent B ( ACN). The flow rate ofthe mobile phase was 0.3 mL/min. The gradient procedurewas as follows: the initial composition was 100% A. A gra-dient elution was performed where phase B was increasedlinearly to 10% at 1.0 min, and then increased to 30% at5.0 min, then decreased to 10% at 6.0 min, finally returned tothe initial composition at 7.0 min and hold 2.0 min to equili-brate the chromatographic separation column. The determi-nation of target analytes was carried out on Xevo TQ tandemMS platform fitted with ESI probe operated in the positiveion mode and the target analytes was monitored under multireaction monitoring (MRM) mode. The optimal parametersare as followed: capillary voltage, 2800 V; ion source tempera-ture, 150�C; desolvation gas temperature, 450�C; desolvationgas flow rate, 800 L/h. Argon was used as the collision gas,and the collision cell pressure was 4 mbar. Other parameterswere shown in Supporting Information Table 1. Instrumentcontrol, data acquisition, and data processing were carriedout with Masslynx V 4.1 software (Waters, MA, USA).

A series heating and drying oven (DHG-9023A, YihengInstrument, Shanghai) was used for thermal polymerization.An S-4800 scanning electron microscope (Hitachi, Japan) anda Bruker tensor 27 infrared spectrophotometer (Ettlingen,Germany) were used for examining the surface structure andrecording Fourier transform infrared spectroscopy spectra.A centrifugal machine from Sigma (Germany) was used toeliminate the proteinic floc in urine. The SPE experimentwas carried out on the SPE device (Supelco, VisiprepTMDL,USA) equipped with a vacuum pump.

2.3 Procedures for the preparation of molecularly

imprinted membrane

Hydrophilic PVDF milli-pore filters were activated as fol-lowing: membranes were rinsed first with pure water andACN subsequently, and then soaked in a 0.15 mol/L AIBNACN solution for 20 min. Finally, membranes were takenout and dried. The activation membranes should be quicklyimmerged in the mixture of molecular imprinting solution.

Prepolymer solution for molecularly imprinting was pre-pared with 0.0052 g RAC, 10.43 �L MAA, and 0.3 mL redis-tilled chloroform dissolved in 0.06 mL methanol. The solu-tion was mixed thoroughly and kept at ambient temperaturefor 1 h in order to carry out fully prepolymerization. Then

0.1178 mL ethylene glycol dimethacrylate and 0.0060 g AIBNwere added. After they were dissolved completely, pretreatedPVDF membranes were immerged in the solution. The so-lution was deoxygenized in an ultrasonic bath and kept theprepolymerization for 2 h. After the prepolymerization, mem-branes were transferred one by one into a glass flask with asealed stopper and then the polymerization was carried outunder a nitrogen atmosphere at 65 in an oven for 24 h. Whenthe reaction finished, the templates were removed by 10% v/vacetic acid and methanol repeatedly until the fluorescence ofRAC could not be detected. The synthesized MIM were rinsedwith pure water to neutral and then with acetone that allowedthe MIM drying at ambient quickly. The prepared MIMs werekept in a desiccator for use.

Nonmolecularly imprinted membranes (NIM) were pre-pared by the same procedure but without template in thesynthesis.

2.4 The static and kinetic adsorption test

The static adsorption of MIM and NIM were examined byputting one piece of membrane in 1 mL RAC, FOM, and RITacetone solution with different concentrations and shakingfor 1 h at room temperature. Then membranes were rinsedwith pure water and eluted by 2 mL 10% v/v acetic acid.The concentration of elution was detected by fluorescence forRAC and RIT or UV absorbance for FOM in order to calculatethe equilibrium adsorption quantity (Q). Data were acquiredin replicates of three with RSD <5%.

The kinetic adsorption experiment was conducted as fol-lows: one piece of the MIM was added into 1 mL 10.0 �g/mLRAC of acetone solution for the adsorption in ambient tem-perature, 10 �L solution was taken at different time intervalsof 5, 10, 20, 30, 60, and 90 min and the concentration ofunbound compounds at the different absorption time weremeasured. Data were acquired in replicates of three with RSD<5%.

2.5 Sample pretreatment

The porcine urine samples were collected from a pig farmnear Beijing, China. The freezing porcine urine samples werethawed at 4�C and centrifuged at 10 000 rpm for 8 min in acentrifugal machine to eliminate the proteinic floc in urine. Atotal of 4 mL porcine urine was drawn. After adjustion of pH5.0 and the addition of 40 �L �-glucuronidase/acryl sulfatase,which are mainly used in urine samples to release glucuronicand sulphate conjugates of �-agonists, the urine sample wasset to incubate at 37�C for 4 h.

2.6 Sample preparation using molecularly imprinted

membrane

The pH value of porcine urine samples was adjusted to 9.0,2.0 mL toluene were added into the porcine urine samples,

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

1458 P. Wang et al. J. Sep. Sci. 2013, 36, 1455–1462

extracted in an ultrasonic bath for 20 min and then cen-trifuged at 5000 rpm for 5 min. One milliliter supernatantwas taken out of the centrifuge tube and 1.0 mL toluene wascomplemented. The sample was extracted again and super-natants were combined. Toluene of extraction solution wasremoved with a stream of nitrogen and the residues weredissolved in 1 mL of acetone. The MIM filter was placed in adismountable filter holder and the filter holder was mountedon the SPE device. The MIM filter was conditioned with1 mL acetone and 1 mL water, and the above acetone solutionswere loaded on the MIM filter. The MIM were washed with1 mL water and 1 mL methanol. Finally, the target compoundswere eluted by 2 mL 10% v/v acetic acid. The extracts wereevaporated with N2 at 60�C, and reconstituted in 1 mL mobilephase, then determined by UPLC-MS/MS. In the above pro-cedure, the flow rate was controlled at the rate of 0.2 mL/minunder the vacuum of −9 kPa.

2.7 Sample preparation using mix cation exchange

cartridge

The SPE procedure on mix cation exchange cartridge wascarried out as follows. The columns were conditioned by thefollowing sequence: 3 mL methanol, 3 mL water. After appli-cation of the samples, the columns were washed with 3 mLwater and 3 mL methanol and eluted with 3 mL methanolcontaining 5% ammonia. The extracts were evaporated withN2 at 60�C, and reconstituted in 1 mL mobile phase.

3 Results and discussion

3.1 Preparation of the molecularly imprinted

membrane

The hydrophilic PVDF was selected as support membrane.In order to remove residual substances and activate the hy-drophilic functional group on the membrane surface, theconditioning of the membrane was tried in different waysincluding soaked in NaOH, pure water, organic solvents, andAIBN ACN solution. It was found that the PVDF membranewas rinsed first with water and ACN and then soaked in AIBNsolution was benefit for the next reaction.

According to the molecular structure of the RAC, twofunctional monomer methyl methacrylic acid (MAA) andacrylamide (AA) were compared. After adsorption experi-ments, it was founded that the effect of MIM prepared byMAA was better than that by AA, so MAA was selected touse as the functional monomer. It seems that the functionalgroup –COOH of monomer MAA provide stronger noncova-lent force for the template molecular than AA. The results weobtained agreed with reports from Feas et al. [33].

It was also of great importance to select an appropri-ate porogenic solvent in the synthesis of MIM to keep thegood permeability. The porogenic reagent should be care-fully selected otherwise the interaction between RAC and

MAA would be disturbed. Three different mixture solvents at1:10 v/v were studied, the amount of RAC adsorbed on theMIM was used as a key parameter to evaluate the effect ofdifferent porogenic solvent. It is indicated that the MIM us-ing methanol/chloroform mixture as porogenic solvent couldprovide better adsorption capacity (1.71 �g RAC/cm2) and se-lectivity than that of other porogenic solvent. The compositionof the solution of methanol mixed with chloroform was fur-ther examined among 1:5, 1:7, and 1:10. As a result, optimalvolume ratio of methanol and chloroform was 1:5 at which thedifference of the adsorption to RAC between MIM and NIMwas very obvious, the ratio of adsorption amount of MIM andNIM for RAC typically was 2.65.

The different molar ratios of template, monomer, andcross-linker at 1:4:20, 1:6:30, 1:8:40, and 1:12:60 were studiedin detail. The concentration of RAC solution for the absorp-tion was 5.0 �g/mL. Accordingly, the volume of elution was2 mL. Each ratio was three times parallel determined withRSD ≤ 5%. The results indicated that the MIM could pro-vide the better selective absorption than that of the NIM andhad higher absorption quantity when the ratio of template,monomer and cross-linker of 1:8:40.

The IR spectra of the blank PVDF, RAC–MIM, and NIMwere measured. All of the three samples revealed the C-F vi-brations in 1295 and 1149 cm−1. The MIM and NIM revealed–C=O infrared absorptions near 1725 and 3000 cm−1, indicat-ing that the molecular imprinted layer had been cross-linkedon the membrane. The infrared absorbance of the MIM washigher than that of NIM.

The SEM was used to examine the microstructures of theblank PVDF and RAC–MIM membrane. In the SEM of blankPVDF (Fig. 1A), the thin fibers of the PVDF membrane couldbe observed clearly. The fibers of the MIM became broader af-ter polymerization as shown in Fig. 1B. Although porogenicsolvents were used, there were still some bulk material re-mained on the surface of the membrane. As the result, theflow resistant of MIM was larger than that of the blank PVDFmembrane. The thickness of the MIM was measured. The av-erage thickness of the blank PVDF membrane and the MIMwas about 105 and 184 �m, respectively.

3.2 Adsorption performance of molecularly

imprinted membrane

Pure water, methanol, and acetone were used for prepareRAC solution for the study of extraction performance andthe results demonstrated that acetone was the optimal sol-vent for the extraction of three compounds. In the followingexperiments, acetone was used as extraction solvent.

The kinetic adsorption of MIM for RAC was studied ac-cording to Section 2.4 to explore the adsorption rate of thesynthetic materials. The kinetic curve was ploted and the ki-netic curve showed that adsorption equilibrium of MIM couldbe reached at about 10 min or so.

The adsorption capabilities of RAC–MIM and NIMwere examined with RAC, RIT, and FOM standard acetone

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

J. Sep. Sci. 2013, 36, 1455–1462 Sample Preparation 1459

Figure 1. Scanning electron microscope photograph of the blankPVDF membrane (A) and the MIM (B).

solution, the binding isotherm was shown in SupportingInformation Fig. S1. The range of concentration was 0.1–10 �g/mL for RAC and RIT, the concentration range of FOMwas from 0.5 to 10 �g/mL due to the low responsive sig-nal of FOM on the HPLC detection. The extraction time wasabout 10 min. The results indicated that the extraction yieldincreased along with the increase of concentration and theadsorption capacity reached saturation above the concentra-tion of 3.0 �g/mL. The adsorption capacity of the RAC–MIMfor RAC, RIT, and FOM was estimated about 2.56, 2.03, and1.88 �g/cm2, respectively. In addition, the adsorption yieldsof MIM were obviously higher than that of NIM and the ad-sorption capacity was above 3.0 times of the MIM over theNIM. This distinction could be attributed to the differenceof extraction mechanism between MIM and NIM. The MIMpossessed the specific imprinted sites that arranged regularlyon the surface of MIM due to the imprinting effect of RACtemplate, so the MIM exhibited higher affinity to the templateand the analogues.

The selective adsorption of RAC–MIM and NIM for RAC,RIT, FOM, MET, and NAD were investigated. The experi-ments were conducted according to Section 2.3. In order to

Table 1. Selective adsorption of MIM and NIM for the differentcompounds at the level of 2.5 �g/mL in acetone (n = 6)

Compounds MIM (�g/piece) NIM (�g/piece)

RAC 2.21 ± 0.15 0.84 ± 0.09RIT 2.12 ± 0.17 0.69 ± 0.10FOM 1.82 ± 0.12 0.57 ± 0.08NAD 0.53 ± 0.09 0.63 ± 0.07MET 0.75 ± 0.10 0.79 ± 0.13

avoid saturation of the RAC–MIM and competitive adsorp-tion [34], the extraction solutions were prepared individuallywith the concentration of 2.5 �g/mL. The results of the se-lectivity for RAC–MIM are shown in Table 1. The resultsindicated that the RAC–MIM could selectively extract RAC,RIT, and FOM, the adsorption amounts of RAC, RIT, andFOM with RAC–MIM were about 2.6 times higher than thatof NIM. But for the MET and NAD which were different instructure and chemical properties with RAC, were showedpoorer adsorption capacity on the RAC–MIM. The selectiveadsorption study showed that the RAC–MIM possess the spe-cific sites and cavities to selectively recognize the templateand its analogues. The specific recognition sites and cavitiesof RAC–MIM could reciprocal complementary and interac-tion to the template functional groups. Because there wereno such specific sites and cavities in the NIM, the adsorptioncapacity of RAC–MIM and NIM for MET and NAD had nosignificant difference.

3.3 Selection of clean-up protocols

The clean-up procedures of MIM for the RAC were exam-ined. The MIM filter was conditioned with 1 mL water and1 mL acetone to swell the cavity on the MIM. The washingsolution was 1 mL water and 1 mL methanol to wash theimpurity of sample matrix and clean the target analytes. Thewashing procedures could eliminate the interference of sam-ple matrix effectively and ensure the recovery of RAC, RIT,and FOM. The elution solvent was 10% acetic acid aqueous.Different volume of 10% v/v acetic acid was used. Accordingto the experiment results, 2 mL of volume for the eluting wasenough.

3.4 Optimization of instrumental conditions

In this study, the UPLC-MS/MS was used to determine theRAC, RIT, and FOM in the porcine urine samples after se-lective extraction using RAC–MIM. The positive ESI+ sourcewas selected due to its sensitivity and robustness and thecharacteristic of target analytes. The Working solutions ofRAC, RIT, and FOM (500 ng/mL) were directly infused toMS to optimize the detection parameters and to select theappropriate diagnostic ions. The optimal MS/MS parameters

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

1460 P. Wang et al. J. Sep. Sci. 2013, 36, 1455–1462

Figure 2. MRM chromatogram of RAC, RIT, and FORM in MIMextracts (A) and MCX extracts (B).

including cone voltage and collision energy are shown inSupporting Information Table S1. The composition of mo-bile phase including methanol/water containing 0.1% formicacid and ACN/water containing 0.1% formic acid were opti-mized to achieve the better peak shape and maximum re-sponse value. The results indicated that the mobile phase ofACN/water containing 0.1% formic acid was more satisfac-tory for the peak shape and signal strength of target analytes,and the signal strength of target analytes could be increased>10%. So the mobile phase of ACN/water containing 0.1%formic acid was used in the whole study.

3.5 Molecularly imprinted membrane clean-up

To evaluate the performance of the RAC–MIM for thecleaning-up of RAC, RIT, and FOM in the porcine urine sam-ples, the comparison of mixed cation mode (MCX) SPE car-tridge and RAC–MIM was conducted. The LC-MS/MS chro-

Table 2. Performance of the developed method based on theRAC–MIM

Compounds Monitor Linear R2 LOD LOQion (m/z) ranges (ng/mL) (ng/mL) (ng/mL)

RAC 302 >164 0.05–10.0 0.997 0.003 0.01RIT 288 >121 0.05–10.0 0.993 0.006 0.02FOR 345 >149 0.05–10.0 0.991 0.003 0.01

Table 3. Recoveries of target analytes in porcine urine samples(n = 6)

Compounds Added Measured Recoveries CVr CVR(ng/mL) (ng/mL) (%) (%) (%)

RAC 0.05 0.0341 68.2 7.0 8.20.10 0.0776 77.6 4.6 7.70.15 0.1260 84.0 3.6 5.3

RIT 0.05 0.0368 73.6 10.8 7.30.10 0.0785 78.5 3.3 6.70.15 0.1295 86.3 4.7 5.9

FOR 0.05 0.0351 70.2 6.7 5.60.10 0.0679 67.9 6.6 6.90.15 0.1158 77.3 5.1 8.5

matograms are shown in Fig. 2, it is obvious that a clear en-hancement of response value of target analytes is obtained byusing the RAC–MIM (Fig. 2A) instead of MCX-SPE cartridge(Fig. 2B). Moreover, the sample extracts using RAC–MIMshowed to be clean with a very low content of contaminatingcompounds. The satisfactory cleaning effect of RAC–MIMmight be attributing to the high selectivity of RAC–MIM forRAC, RIT, and FORM. Nielen et al. [4] have developed asensitive and robust method for RAC and other �-agonistsin porcine urine by using commercial mix mode columncleaning followed UPLC-MS/MS detection with LOD <0.28ng/mL. However, the enrichment factor was 7 and injectionvolume was 50 �L in the reported method. Compared withthe reported method, the advantages of selectivity and sensi-tivity of the presented method based on RAC–MIM clean-upcombined with UPLC-MS/MS detection for RAC, RIT, andFORM in complicated porcine urine samples were obvious.

3.6 Method validation

The evaluation of the suitability of present method for the de-termination of RAC, RIT, and FORM in porcine urine sam-ple was conducted according to the European CommissionDecision 2002/657/EC [35]. The calibration curves were ob-tained in 0.05–10.0 ng/mL (0.05, 0.1, 0.5, 1.0, 5.0, and10.0 ng/mL) for RAC, RIT, and FOM by plotting the peakarea versus compounds concentration. The correlation co-efficients (R2) of the calibration curves were >0.99 for allcompounds.

Under the optimal instrument conditions, by using theMIM for enrichment, RAC, RIT, and FOM were detected with

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

J. Sep. Sci. 2013, 36, 1455–1462 Sample Preparation 1461

UPLC-MS/MS. The LOD and LOQ were estimated based onthree times S/N and ten times S/N, respectively. The LODsand LOQs of RAC, RIT, and FOM in porcine urine werebelow 0.006 and 0.02 ng/mL, respectively. The LOD of thismethod is better than those of previously reported methodsbase on molecularly imprinted method [21, 22]. The resultsof the developed method performance are summarized inTable 2.

The recovery of RAC, RIT, and FOM in porcine urinesamples were examined at three different concentrations(0.05, 0.10, and 0.15 ng/mL). The repeatability and repro-ducibility were measured on the fortified blank urine sam-ples (n = 6 replicates per concentration level and analyzedin three independent analytical runs) and expressed by coef-ficient of variation (CVr and CVR, respectively). The resultsare shown in Table 3. The mean recoveries, repeatability, and

Figure 3. (A) The MRM chromatogram of blank porcine urinesample. (B) The MRM chromatogram of blank porcine urine sam-ple fortified with RAC, RIT, and FOM at 0.05 ng/mL.

reproducibility varied from 67.9 to 86.3%, from 3.3 to 10.8%(CVr), and from 5.3 to 8.5% (CVR), respectively. These resultsdemonstrated that the developed method is accuracy and ro-bust. The MRM chromatograms of blank urine sample anda fortified urine sample (spiked 0.05 ng/mL) are shown inFig. 3. It could be observed that there were no interferingpeaks from endogenous compounds in porcine urine sampleat the retention times of RAC, RIT, and FOM.

3.7 Application in real samples

The developed method was applied into determination ofRAC, RIT, and FOM in 50 porcine urine samples that frompig farms in Shaanxi province, China. RAC, RIT, and FOMwere not found in these samples. The blank regent, blanksample and spiked sample (0.1 ng/mL) were determinedaccompanied with the real samples to assure the accuracyof the results. In addition, the RAC in above porcine sampleswere determined according to the china national standardmethod [36]. The results for RAC with the developed methodwere accordance with the confirmatory determination resultsby using China agricultural industry standard method.

4 Concluding remarks

In conclusion, a novel MIM with RAC as the template andthe hydrophilic PVDF membrane as the support were pre-pared for the selective extraction and enrichment of three�-adrenergic agonists. The prepared MIMs showed the goodselection and absorption capacity for RAC, RIT, and FOM.With the extraction of the MIM, the samples of porcine urinecould be effectively cleaned and good recoveries for threecompounds were achieved. Based on the MIM extraction, aUPLC-MS/MS method can be successfully used to confir-matory analysis trace RAC, RIT, and FOM in the real urinesamples and sensitivity is satisfactory. The MIM revealed theadvantages of the commercial membrane filters and the highselectivity.

The authors would like to thank the Special Fund for Agro-scientific Research in the Public Interest (No. 201203088 andNo. 201203094), Natural Science Foundation of China (No.31201832), and Excellent Personnel Program of Beijing Munici-pal Government (20081D0501600194) for financially supportingthis research.

The authors have declared no conflict of interest.

5 References

[1] Xiong, Y. L., Gower, M. J., Li, C., Elmore, C. A., Cromwell,G. L., Lindemann, M. D., Meat Sci. 2006, 73, 600–604.

[2] Carry, S. N., Hamilton, D. N., Miller, K. D., Schroeder, A.L., Duenas, D. F., Killefer, J., Ellis, M., McKeith, F. K., MeatSci. 2009, 81, 533–539.

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

1462 P. Wang et al. J. Sep. Sci. 2013, 36, 1455–1462

[3] Kuiper, H. A., Noordam, M. Y., Dooren-Filpsen, M. M. H.,Schilt, R., Roos, A. H., J. Anim. Sci. 1998, 76, 195–207.

[4] Nielen, M. W. F., Lasaroms, J. J. P., Essers, M. L., Anal.Bioanal. Chem. 2008, 391, 199–210.

[5] Ying, Y., Pi, X., Cong, H., Zhu, C., Chin. J. Chromatogr.2006, 24, 320–320.

[6] Ying, Y., Pi, X., Wu, P., Chin. J. Anal. Chem. 2006, 34,143–143.

[7] Beatrice, B., Maurizio, F., Claudia, C., Gianfranco, B., J.AOAC Int. 2003, 86, 8–14.

[8] Blanca, J., Munoz, P., Morgado, M., Mendez, N., Aranda,A., Reuvers, T., Hooghuis, H., Anal. Chim. Acta 2005, 529,199–205.

[9] Juan, C., Igualad, C., Moragues, F., Leon, N., Manes, J.,J. Chromatogr. A 2010, 1217, 6061–6068.

[10] Shen, L., He, P., Electrochem. Commun. 2007, 9, 657–662.

[11] Wang, W., Zhang, Y., Wang, J., Meat Sci. 2010, 85,302–305.

[12] Zhang, M., Wang, M., Chen, Z., Fang, J., Fang, M., Liu,J., Yu, X., Anal. Bioanal. Chem. 2009, 395, 2591–2599.

[13] Shelver, W. L., Smith, D. J., Berry, E. S., J. Agric. FoodChem. 2000, 48, 4020–4026.

[14] Shelver, W. L., Smith, D. J., J. Agric. Food Chem. 2002,50, 2742–2747.

[15] Jin, Y. E., Guo, D. H., Zheng, Y., Wang, G. Q., J. Chin.Mass Spectrom. Soc. 2007, 28, 193–201.

[16] Jimenez-Carmona, M. M., Tena, M. T., Luque, C. M. D.,J. Chromatogr. A 1995, 711, 269–276.

[17] Hu, Y., Liu, R., Li, Y., Li, G., J. Sep. Sci. 2010, 33,2017–2025.

[18] He, J., Fang, G., Deng, Q., Wang, S., Anal. Chim. Acta2011, 692, 57–62.

[19] Fang, G. Z., Lu, J. P., Pan, M. F., Li, W., Ren, L., Wang, S.,Food Anal. Methods 2011, 4, 590–597.

[20] Wang S., Liu L., Fang, G. Z., Zhang, C., He, J. X., J. Sep.Sci. 2009, 32, 1333–1339.

[21] Zhang, Q., Su, Y., He, Q., Shen, X., He, L., Zhang, N.,Zeng, Z., J. Sep. Sci. 2011, 34, 3399–3409.

[22] Tang, Y. W., Fang, G. Z., Wang, S., Li, J. L., Anal. Bioanal.Chem. 2011, 401, 2275–2282.

[23] Xu, Z., Hu, Y., Li, G., J. Chromatogr. A 2010, 1217,3612–3618.

[24] Hu, Y., Li Y., Liu R., Tan, W., Li, G., Talanta 2011, 84,462–470.

[25] Wang S., Wei, J., Hao, T. T., Guo, Z. Y., J. Electroanal.Chem. 2012, 664, 146–151.

[26] Zhang, H. C., Liu, G. Y., Chai, C. Y., Sensors Actuators, B2012, 168, 103–110.

[27] Kong, L. J., Pan, M. F., Fang, G. Z., Qian,K., Wang, S., Anal. Bioanal. Chem. 2012, 404,1653–1660.

[28] Sergeyeva, T. A., Brovko, O. O., Piletska, E. V., Piletsky,S. A., Goncharova, L. A., Karabanova, L. V., Sergeyeva,L. M., Elskaya, A. V., Anal. Chim. Acta 2007, 582,1–319.

[29] Sueyoshi, Y., Fukushima, C., Yoshikawa, M., J. Membr.Sci. 2010, 357, 90–97.

[30] Reddy, P. S., Kobayashi, T., Abe, M., Fujii, N., Eur. Polym.J. 2002, 38, 521–529.

[31] Wang, J. Y., Liu, F., Xu, Z. L., Li, K., Chem. Eng. Sci. 2010,65, 3322–3330.

[32] Renkecz, T., Ceolin, G., Horvath, V., Analyst 2011, 136,2175–2182.

[33] Feas, X., Fente, C. A., Hosseini, S. V., Seijas, J. A.,Vazquez, B. I., Franco, C. M., Cepada, A., Mater. Sci. Eng.,C 2009, 29, 398–404.

[34] Hu, X. G., Pan, J. L., Hu, Y. L., Huo, Y., Li, G. K., J. Chro-matogr. A 2008, 1188, 97–107.

[35] European Commission Decision 2002/657/EC, Off. J. Eur.Commun. 2002, L221, 8–36.

[36] Huang, S. X., Wang, B., Cao, Y., Li, D. N., China Ministry ofAgriculture Regulation 1063-3-2008, China AgricultrualPress, Beijing, China 2008.

C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com