Comparison Ofa Competitive ELISA with an HPLC Method for the Determination of Aflatoxin M 1 in Turkish White, Kasar and Tulum Cheeses

Eur Food Res Technol (2006) 223: 719-723 DOI 10.1007 /s00217-006-0258-4

ORlGlNAL PAPER . • No r,:o : :::::·.·.·.·_·_·_·_·_·_·__ ·_·_·.-.-:::::::::::·. f

Hilal Colak • Hamparsun Hampikyan • Beyza Ulusoy • üzer Ergun Comparison ofa competitive ELISA with an HPLC method for the determination of aflatoxin M 1 in Turkish White, Kasar and Tulum cheeses

Rcceived: 1 November 2005 / Revised: 27 Decembcr 2005 / Accepted: 4 January 2006 / Published online: 1 February 2006 © Springer-Verlag 2006

Abstract AFMl is a hydroxylated metabolite produced dairy products such as cheese, which represents an im- when ruminants ingest contaminated feed with AFB 1 and portant risk factor for consumers, in particular for infants transferred to dairy products such as cheese, which rep- and young children (6, 7]. Therefore, many countries have resents an important risk factor for consumers. Turkish established tolerance limits for AFMl in milk and dairy White, Kashar and Tulum cheeses are traditional cheese products (Table 1). types in Turkey. This study was planned to compare the In Turkey, 40-50 cheese varieties are known but only performance of a competitive ELISA for the determina- three of them have national and economic value: Turk- tion of AFM1 in Turkish White, Kasar and Tulum cheeses ish White (named as Beyaz Peynir originally), Kasar (like against a new HPLC method. For this purpose, different Kashkaval or Kasseri cheeses) and Tulum (ripened in goat AFMl concentrations (50-400 ng/kg) were added to the skin bags or plastic material) cheeses [9]. Therefore, in cheese samples and the toxin levels were determined by our country, there are various studies on AFM 1 in differ- ELISA and HPLC. Also, both methods were performed in ent cheese types especially in White, Kasar and Tulum 24 real samples obtained from different markets. In con- cheeses and different results were reported by a number of clusion, the results obtained by ELISA in this study were researchers [ 10-15]. related to those by HPLC for AFMl. Many analysis methods of AFMl in milk and dairy products are available in the literature. For aflatoxin deterrnina- Keywords AFMl · ELISA · HPLC · Cheese ·Tulum· tion, the routinely used methods are based on thin layer Kasar chromatography (TLC), high performance liquid chromatography (HPLC) and enzymed-linked immunosorbent assay (ELISA) [ 16]. Gilbert and Anklam [ 17] reported that lntroduction TLC methods have rather high coefficients of variations and have been used only where at aflatoxin contarnina- (B 1, B2, G 1 and G2) are produced by at least tion levels are higher than the current control limits. Other three species of Aspergillus: A. fiavus, A. parasiticus, A. chromatographic methods such as HPLC are very expen- nomius. These toxic metabolic products have sub-acute sive and time consuming techniques. Therefore, current and chronic effects such as, liver cancer, chronic hepati- immunological methods are preferred to chromatographic tis, jaundice, hepatomegaly and cirrhosis in humans [ 1- methods in routine analysis and survey studies [10]. ELISA 3]. Of ali mycotoxins, aflatoxin B 1 (AFB 1) is considered is a fairly sensitive and rapid method which is used in de- to be the most toxic compound. lnternational Agency for termination of AFMl in milk and dairy products and suit- Research on Cancer (IARC) included AFB 1 as primary able routine diagnostic application of atlatoxin detection and aflatoxin Ml (AFMl) as secondary groups of carcino- [10, 18, 19]. Surveys have been carried out in our country genic compounds [4]. AFMl is a hydroxylated metabo- to verify the occurrence of AFMl in cheese by ELISA lite produced when ruminants ingest contaminated feed (Ridascreen AFMl test kit, Germany) in with AFB 1 [5]. it is excreted in milk and transferred to general. The aim of this study is to compare the performance of H. Colak (81) • I-I. Hampikyan • B. Ulusoy • O. Ergun a competitive ELISA for the deterrnination of AFMl in Faculty of Veterinary Medicine, Department of Food Hygiene Turkish White, Kasar and Tulum cheeses against a new and Technology, Istanbul University, HPLC method described by Manetta et al. [20] and to eval- Avcilar, Istanbul 34320, Turkey e-mail: [email protected] uate the results of extraction procedure for both ELISA and Tel.: +90-212-473-70-70(17181) HPLC whether effected from various cheese types which Fax.: +90-212-473-72-41 are produced by different techniques. ...

720

Table 1 lnternational Country Raw milk (µ,g/kg) Dairy products ( µ,g/kg) legislation on AFMI in milk and dairy products for human European Union 0.05 0.05 consumption [8] Austria 0.05, O.Ol (pasteurized infant mille) 0.02 (butter), 0.25 (cheese), 0.4 (powdered mille) France 0.05, 0.03 (for children <3 years) Switzerland 0.05 0.025 (milk whey and products), 0.25 (cheese), 0.02 (butter) Bulgaria 0.50 0.10 (powdered milk) Rumania o o Czech Republic 0.50 USA 0.50 Brazil 0.50 (fluid milk), 5.0 (powdered milk) Argentina 0.05 0.50 (milk products) Honduras 0.05 0.25 (cheese) Nigeria 1 Egypt o o Turkey 0.05 0.25 (cheese)

methanol, 0.5 mi buffer phosphate (Na2HPO4: 1.427% Materials and methods (w/v), KH2PO4: 0.907 (w/v), ratio 8:2, pH: 7.2), 1 mi n- heptane and was mixed thoroughly. After centrifugation for Chemicals 15 min at 2,700 rpm, 100 µl of the methanolic phase was AFM I standard and pyridinium hyrobromide perbro- brought to 10% methanol content by addition of 400 µl mide (PBPB) were obtained from Sigma. The stock so- Ridascreen Buffer 1. An aliquot (100 µl per well) of this lution for AFMl (Sigma-A 6428) was prepared in a solution was used in the test. methanol/chloroform mixture (81: 19, v/v) and the concen- The AFMl standards and test samples in duplicate were tration of 50 mg/ml was kept frozen at - 20°C. Before added to the wells of micro-titer plate pre-coated with an- using, it was diluted in methanol/chloroform (1:1, v/v) at tibodies for AFMl and incubated at room temperature in proper concentrations [19]. Ali other chemicals and sol- dark for 60 min. After the washing step, AFMl Peroxidase vents were of analytical or HPLC grade and supplied by conjugate (supplied with the kit) was added to the wells and plate was incubated again for 60 min at room temperature Merek. m dark. The unbound conjugate was removed during washing. Subsequently, 50 µ! each of substrate (urea peroxide) and chromogen (tetramethyl-benzidine) were added to the Sample preparation wells and incubated for 30 min in dark. The reaction was Experiments were performed on AFMl-free White, Kasar stopped by 100 µl of 1 mol/1 H2SO4 and the absorbance and Tulum cheeses (three determinations at each con- was measured at 450 nm in ELISA plate reader (ELX centration levels have been performed). Recoveries were 800, Bio-tek Inst.). The mean values of the absorbance determined by spiking of cheese samples at 50-400 ng/kg for the standards and the samples were evaluated accord- (0.2-1.6-fold the maximum residue limit for AFMl in ing t_o the Rl-DAVIN.EXE computer program prepared by cheese set in Turkish Food Codex). The AFM 1 concentra- The mean lower detection limit was 50 ng/kg for cheese. tions were directly added to the triturated cheese samples and the toxin levels were detem1ined by ELISA and HPLC. Twenty-four real samples (eight White, eight Kasar and eight Tulum cheeses) which were obtained from different HPLC markets were alsa analyzed for AFMI by both methods. AFMl concentrations of cheese samples were determined by HPLC using post-column derivatisation with PBPB ac- ELISA cording to a new method described by Manetta et al. [20]. 1O g of spiked cheese samples were extracted with 50 mi of Extraction of AFMl and ELISA test procedure were dichlormethane/acetone mixture (l: 1, v/v) by Ultra Turrax performed according to the instructions of the test kit (Janke & Kunkel) with addition of sodium chloride (10 g). (Ridascreen Art No.: Rl 101, R-Biopharm) manual [21]. After centrifugation (3000 x g for 10 min), 10 mi of the or- Spiked cheese samples (2 g each) were homogenized ganic extract were evaporated to dryness under a stream of (Ultraturax, Janke & Kunkel) and extracted with 40 mi nitrogen. The residue was dissolved in methanol (0.5 mi) dichloromethane. The suspension was filtered and then and O.Ol mol/1 sodium phosphate-buffered saline (PBS), 1O mi of the extract was evaporated at 60 °C under pH 7.2-7.4 (20 mi) and n-hexane (10 ml) were added. Af- nitrogen. The oily residue was redissolved in 0.5 mi ter shaking, the Iower la yer was quantitatively collected and 721

Table 2 Analytical parameters for the determination of AFM l in spiked White cheeses using ELISA and HPLe Amount added (ng/kg) ELISA HPLe Mean amount Mean ev• Mean amount Mean recovered (ng/kg) recovery (%) recovered (ng/kg) recovery (%) 50 49.4 98.8 ±5.8 10.8 45.0 90.J ± 2.8 0.993 100 99.5 99.5 ±9.2 il.O 89.7 89.7 ±5.l 0.99 1 150 151.0 100.7±7.6 12.0 132.3 88.2± 3.5 0.994 200 202.0 101.0±6.2 11.4 170.0 85.0 ± 5.0 0.998 250 244.0 97.6±8.4 9.2 209.2 83.7±2.l 0.999 300 300.6 100.2±7.3 8.6 258.3 86.l ± 4.8 0.996 350 357.0 102.0± 8.4 9.8 288 .7 82.5 ± 3.8 0.998 400 411.2 102.8 ±9.6 11.6 326.4 81.6± 5.4 0.999 Average 225 226.8 100.3 ±7.8 10.5 189.9 85 .9±4.1 0.996 'CV: Coefficient of variation hR2: Regression

Table 3 Analytical parameters for the determination of AFMlin spiked Kasar cheeses using ELISA and HPLe Amount added (ng/kg) ELISA HPLe Mean amount Mean ev• ------:,--:-----~~-Mean amount Mean R2 b recovered (ng/kg) recovery (%) recovered (ng/kg) recovery (%) 50 48.3 96.7 ± 9.8 l0.2 44.7 89.5 ± 2.1 0.991 100 99.0 99.0± l0.2 11.8 88.2 88.2± 1.8 0.992 150 149.2 99.5 ± 5.4 12.5 130.9 87.3 ± 5.2 0.994 200 198.0 99.0± 8.6 13.4 168.2 84.l ± 2.6 0.996 250 244.0 97.6 ± 9.2 14.0 207.5 83.0 ±4.7 0.997 300 305.4 l01.8 ± 6.3 13.2 255 .6 85.2± 3.4 0.995 350 350.3 100.1 ±7.5 10.9 283.5 81.0 ± 2.5 0.999 400 404.4 101.1 ±6.2 12.8 321.2 80.3 ±4.2 0.998 Average 225 224.8 99.3±7.9 12.3 187.5 0.996 0.995 •ev: of variation b R2 : Regression coefficient cleaned up on C 18 -SPE as described by Manetta et al. [20]. nation of AFMl in three kjnds of Turkish cheese. ELISA Then, AFM 1 eluate was evaporated un der a stream of nitro- and HPLC results of spiked White, Kasar and Tulum cheese gen and the residue dissolved in acetonitrile (200 µ.]). An samples are given in Tables 2-4 respectively. aliquot ( 10 µ.I) of the AFMl extract was analysed by HPLC For ELISA, the average recoveries ± SD were found (Hewlett Packard 1100 HPLC chromatograph, equipped to be 100.3 ± 7.8% for White cheese, 99.3 ± 7.9% for with a Hewlctt Packard 1100 fluorescence detector). Ex- Kasarcheese, 100.7 ± 6.5% forTulurn cheese (Tables 2-4). citation and emission wavelengths were set at 353 and These results are similar with Kaniou-Grigoriadou et al. [8] 423 nm, respectively. The eluate passes through a Supel- who reported mean recovery of AFMl was 108% for feta cosil LC-18 column (5 µ.m particle size, 250 mm x 4.6 mm cheese. According to the kit's manufacturer instructions, I.D.). The mobile phase was acetic acid/acetonitrile/2- the recovery rate in cheese has been stated approximately propanol/water (2:10:10:78, v/v/v/v). lsocratic HPLC was 102%. performed at 1.2 ml/min. As a post-column derivatising For HPLC, the average recoveries ± SD were found agent an aqueous solution of 50 mg/1 PBPB was used. The to be 85.9 ± 4.1 % for White cheese, 84.8 ± 3.3% for quantity of AFMl was determined as stated by Manetta Kasar cheese, 84.7 ± 3.4% for Tulum cheese (Tables 2- et al. [20]. The mean lower detection limit was 5 ng/kg for 4). Manetta et al. [20] reported their mean recovery result chcese. ± SD as 76 ± 4% for cheese by using the same method. Our results were found to be higher than thls finding. The Results and discussion HPLC method used in this study was a new method recently published by Manetta et al. [20]. The authors stated In this study, a competitive ELISA (using Ridascreen test that the method with post-column derivatisation improves kit) anda new HPLC method were compared for determi- AFMl analysis compared with earlier methods and it was 722

Table 4 Analytical parameters for the determination of AFMI in spiked Tulum cheeses using ELISA and HPLC Amount added (ng/kg) ELISA HPLC Mean amount Mean ev• Mean amount Mean R2 b recovered (ng/kg) recovery (%) recovered (ng/kg) recovery (%) 50 49.6 99.2±7.6 il.O 45 .9 91.8 ± 2.2 0.993 100 98.8 98.8 ±5.4 11.5 90.0 90.0±5.4 0.994 150 150.3 100.2± 8.6 11.9 130.2 86.8 ± 3.1 0.99 1 200 205.6 102.8 ±4.3 12.5 168.4 84.2±4.4 0.996 250 253.5 101.4 ± 7.3 10.5 204.0 81.6 ± 2.0 0.998 300 297.6 99.2 ± 5.2 11.5 250.5 83.5 ±3.8 0.997 350 355.9 101.7 ±6.2 10.0 280.7 80.2 ± 2.8 0.995 400 408.0 102.0±7.4 13.4 318.8 79.7 ± 3.7 0.999 Average 225 227.4 100.7 ±6.5 11.5 186.1 84.7 ±3.4 0.995 •ev: Coef!i cient of variation hR2 : Regression

Table 5 The AFM 1 levels of Sample no Cheese type ELISA HPLC cheese Mean value (ng/kg) Mean value (ng/kg) 1 White cheese ND• ND 2 White cheese 245 .0±5.2 208.2±2.2 3 White cheese ND" 22.l ±2.8 4 White cheese 115.3 ± 3.9 97.7 ±3.0 5 White cheese 68 .7 ±5.7 57 .7±2.4 6 White cheese ND ND 7 White cheese ND ND 8 White cheese 159.5 ±4.6 135.1 ±2.6 9 Kasar cheese ND• ND 10 Kasar cheese 148.2±6.6 124.3 ± 1.9 11 Kasar cheese ND 18.0± 3.3 12 Kasar cheese 79.4±5.9 65.4±2.7 13 Kasar cheese ND• 27 ± 1.8 14 Kasar cheese 95.2±7.4 80.92± 1.5 15 Kasar cheese ND 35 ±2.1 16 Kasar cheese ND ND 17 Tulum cheese ND ND 18 Tulum cheese 210.0 ± 3.9 176.4± 3.8 19 Tulum cheese 195.3 ±6.5 164.0±2.4 20 Tulum cheese ND ND 21 Tulum 93 .2±4.l 77.3 ± 3.3 22 Tulum cheese 158.7 ±7.2 136.4±2.6 23 Tulum cheese 385.4 ±8.6 331.4±3.4 24 Tulum cheese ND •ND: not detected 12 ± 2.0 simple and easy to automate. Therefore, this method was In this study, it was also performed both methods in 24 preferred in this study. Several HPLC methods have been real samples (eight White, eight Kasar and eight Tulum described in the literature to determine AFMl in cheese. cheeses) which were obtained from different markets. The in the first research studies with HPLC, the normal station- results are shown in Table 5. As it can be seen from Table 5, ary phase was used coupled with a detection system for the results of both methods correlate well with each other. UY absorption. Since afl atoxins have flu orescent proper- Immunoassay has been an important tool for aflatoxin ties, it was soon demonstrated that systems for determina- testing since kits for recognizing different mycotoxins tion with fluorescence detection were more sensitive [22]. were commercially available [16]. üne of the well-known Recently, HPLC methods based on solid phase extraction enzyme immunoassays, ELISA, is generally used to or immunoaffinity chromatography in combination with determine AFMl. Several competitive ELISAs for AFMl reversed-phase HPLC and flu orescence detection with or are available commercially. ln the literature, there are without derivatisation [18, 20]. many studies on the presence of AFM 1 in cheese by using 723

Ridascreen AFMl test kit which is a competitive enzyme 3. Simon P, Delsaut P, Lafontaine M, Morele Y, Nicot T (1998) J immunoassay [7, 10-15, 19]. Also, Kaniou-Grigoriadou Chromatogr B 712:95-104 et al. [8] reported that this method is quick, reliable and 4. IARC (1993) Lyon, France 56:245-395 5. Battacone G, Nudda A, Cannas A, Borlino Cappio A, Bomboi cost effective for the of AFM1. Therefore, this G, Pulina G (2003) J Dairy Sci 86:2267-2675 test kit was preferred in our study. The basis of the test is 6. Sibanda L, De Saeger S, Van Petegham C (1999) Int J Food the antigen-antibody reaction. Microbiol 48:203-209 Although there are many studies on the presence of 7. Rastogi S, Dwivedi PD, Khanna SK, Das M (2004) Food Control AFM l in White, Kasar and Tulum cheeses in our coun- 15:287-290 8. Kaniou-Grigoriadou 1, Eleftheriadou A, Mouratidou T, Katikou try, the of extraction procedure for ELISA P (2005) Food Control 16:257-261 has not been tested by researchers. These cheese types 9. Hayaloglu AA, Guven M, Fox PF (2002) lnt Dairy J 12:635-648 have different consistence and production techniques. Our 10. Gurses M, Erdogan A, Cetin B (2004) Turkish J Yet Anim Sci results indicated that the of extraction pro- 28:527-530 11 . Oruc HH, Sona! S (2001) Yet Human Toxicol 43:292-293 cedures for ELISA and HPLC were not affected from 12. Gunsen U, Buyukyoruk I (2002) Yet Human Toxicol 44:289- different cheese consistence as can be seen from Tables 290 2-4. 13. Sarimehmetoglu B, Kuplulu O, Celik H, (2004) Food Control In conclusion, the results obtained by ELISA in this study 15:45-49 14. Yaroglu T, Oruc HH, Tay ar M (2005) Food Control l 6:883-885 were related to those by HPLC for AFMl. This ELISA 15. Aycicek H, Aksoy A, Saygi S (2005) Food Control 16:263-266 method can be considered as an altemative to HPLC, which 16. Lin L, J, Wang P, Wang Y, Chen J (1998) J Chromatogr owing to its advantages (rapid, easy to use, sensitive, ete.) A 815:3-20 can be used for screening in laboratories in which HPLC 17. Gilbert J, Anklam E (2002) Trends Anal Chem 21:468-486 is not available or when a large of samples have to 18. Chiavaro E, Dall 'Asta C, Glavema G, Biancardi A, Gambarelli E, Dossena A (2001) J Chromatogr A 937:31-40 be analyzed. Also, this method is suitable and reliable for 19. Lopez C, Ramos L, Ramadan S, Bulacio L, Perez J (2001) lnt J surveys of AFM I in cheese. Food Microbiol 64:211-215 20. Manetta AC, Di Giuseppe L, Giammarco M, Fusaro l, Simonella A, Gramenzi A, Formigoni A (2005) J Chromatogr A 1083:219- 222 References 21. Anonymous (1999) Enzyme immunoassay for the quantitative analysis of aflatoxin Ml. An. No R 1101. R-Biopharm GmbH, Galvano F, Galofaro V, Galvano G (1996) J Food Prot 59:1079- Darmstadt, Germany 1090 22. Jaimez J, Fenle CA, Vazquez BI, Franco CM, Cepeda A, 2. Kamkar A (2005) Food Control 16:593-599 Mahuzier G, Prognon P (2000) J Chromatogr A 882:1-10