ISSN 0003�6838, Applied Biochemistry and Microbiology, 2011, Vol. 47, No. 3, pp. 321–326. © Pleiades Publishing, Inc., 2011.Original Russian Text © M.A. Burkin, I.A. Galvidis, 2011, published in Prikladnaya Biokhimiya i Mikrobiologiya, 2011, Vol. 47, No. 3, pp. 355–361.

321

INTRODUCTION

Neomycin is a complex of aminoglycoside antibi�otics (neomycins A, B, and C) produced by Streptomy�ces fradiae. Neomycin B (NM), which accounts forover 90%, is the most active and practically significantcomponent, whereas the proportion of neomycin A(neamine) does not exceed 1%. These compounds arepresent in commercial preparations of the antibiotic inthe same proportion. Neomycin exhibits bactericidalactivity with respect to Gram�positive and Gram�nega�tive microorganisms both at the stage of reproductionand at rest. The group of MN�sensitive bacteria includesStaphylococci, Pneumococci spp., Listeria, Escherichia,Salmonella, anthrax pathogen, etc. In veterinary medi�cine, NM is used for treatment and prevention of gas�trointestinal infections as well as pulmonary, suppura�tive�septic, and skin diseases. Similarly to other ami�noglycosides, NM is a neurotoxic and nephrotoxicagent [1]. For the regulation of the antibiotic residuesin foodstuff that provoke the selection of microorgan�isms resistant to this drug, many countries establishedmaximum residue limit (MPL) of NM content infoodstuff of animal origin. In the EU countries, thisindex, determined by the marker compound neomy�cin B, is 500 µg/kg for meat, fat, liver and eggs;1500 µg/kg for milk; and 5000 µg/kg for kidneys [2].In the Russian Federation, the regulated level of thisantibiotic has not yet been established. The residual

NM concentration in milk recommended byFAO/WHO experts should not exceed 500 µg/kg [3].

Microbiological tests, which are commonly used todetect the antimicrobial activity in objects and whichare based on the suppression of antibiotic�sensitivemicroorganism growth, allow NM detection at con�centrations as low as 500 µg/kg. However, these testsare time�consuming, nonspecific, and unsuitable incases of combination antibiotics. Analytical methodsthat are recommended for the identification of thisdrug in animal products are based on HPLC with flu�orescence detection and liquid chromatography with adetection limit of 100 µg/kg (www.emea.europe.eu;www.codexalimentarius.net). Laborious and expen�sive chromatographic methods limit their use formass�scale analysis of samples. The attractiveness ofenzyme�linked immunosorbent assay (ELISA) forNM detection is determined by its simplicity, specific�ity, sensitivity (much higher than the above�men�tioned methods [4–6]), and convenience of its use forscreening studies.

The purpose of the work was to develop an indirectcompetitive ELISA as a simple and convenientmethod suitable for implementation of screening teststo detect NM residues in milk.

Development and Application of Indirect Competitive Enzyme Immunoassay for Detection of Neomycin in Milk

M. A. Burkin and I. A. GalvidisMechnikov Research Institute of Vaccines and Sera, Russian Academy of Medical Sciences, Moscow, 105064 Russia

e�mail: burma68@yandex.ruReceived February 9, 2010

Abstract—As a result of immunization of rabbits with neomycin B (NM) conjugated to sodium periodate�oxidized (SP) transferrin, polyclonal antibodies were generated and used to develop an indirect competitiveenzyme�linked immunosorbent assay (ELISA) of NM. Several heterologous conjugates, namely, glutaralde�hyde (GA)�polymerized NM, gelatin–ribostamycin (sp), and gelatin–NM (ga) were used as coating antigensin different ELISA variants for quantification of NM in milk. These variants were characterized by differentdynamic ranges and detection limits of 1.0, 0.1, and 0.01 ng/ml, respectively. Maximum residue level (MRL)of this antibiotic in milk accepted in the EU can be detected without any special pretreatment at a 100�foldsample dilution in the least sensitive assay variant. The mean recovery rate from NM�spiked milk containing1.5–10% fat was 111.7% and ranged from 84 to 125.2%. We found that 57 of 106 tested milk samples con�tained NM at concentrations higher than 100 ng/ml. In ten percent of cases (11/106), the residual level ofthis antibiotic was greater than 500 ng/ml. In one case, the MRL was exceeded (1690 ng/ml). The assay devel�oped in this study is specific shows no cross�reactivity with other veterinary aminoglycosides, has a good sen�sitivity reserve, and can serve as an effective tool to monitor the NM content in milk stuff.

DOI: 10.1134/S0003683811030045

322

APPLIED BIOCHEMISTRY AND MICROBIOLOGY Vol. 47 No. 3 2011

BURKIN, GALVIDIS

MATERIALS AND METHODS

In this study, we used neomycin B trisulfate fromHuashu Pharmaceutical Corporation (China) and ami�kacin sulfate (AM) from Eczacibasi Ilac San ve Tic A.S.(Turkey). Sulfates of kanamycin (KM), sisomycin(SSM), gentamicin (GM), streptomycin (SM), ribosta�mycin (RS), and apramycin (AP); human transferrin(TR); horseradish peroxidase (EC 1.11.1.7); and o�phe�nylenediamine were purchased from Sigma (UnitedStates). Other reagents used in the study were neamine(neomycin A, NA; European Pharmacopoeia), skimmilk powder (70166, Fluka, Switzerland), netilmicin(NTM, Schering�Plough Labo H.B., Belgium), andtobramycin (TM, Lilly France S.A., France). Donkeyantiserum to rabbit immunoglobulins, bovine serumalbumin (BSA), and gelatin (Gel) were produced inRussia. The object of the study was milk (106 samples)with different fat content (0.1–35%), produced in theEuropean part of Russia and near abroad and sold inretail outlets in 2008–2009.

Synthesis of conjugated antigens. TR(sp)�NM × 30,TR(sp)�NM × 100. An aqueous solution containing20 mg of TR (2 × 130 nmol, 3 ml) was supplementedwith 17 mg of crystalline sodium periodate (SP) andincubated under stirring on a magnetic stirrer for15 min. The oxidized glycoprotein was dialyzedagainst 10 mM acetate buffer (pH 5.0) for 1 day at 4°С.The resultant dialysate was divided into two equal por�tions supplemented with 3.6 and 11.8 mg of NM in1.2 ml of 0.05 M carbonate–bicarbonate buffer (CBB,pH 9.5), which corresponded to 30� and 100�foldmolar excess, and incubated with stirring for 2 h.Then, the reaction mixtures were supplemented with100 µl of sodium borohydride (2 mg/ml) and incu�bated with stirring for another 2 h. Thus obtained con�jugates were subjected to exhaustive dialysis against0.15 M phosphate buffer (pH 7.0).

Poly�NM�0.3 (ga), Poly�NM�1 (ga), Poly�NM�3(ga). Three portions of a solution containing 36.4 mgNM (40 µmol) in 1 ml of CBB were mixed with 53,160, and 480 µl of 2.5% glutaraldehyde solution (GA;0.3�, 1�, and 3�fold molar excess relative to NM) andincubated with stirring overnight at 4°С. The reactionmixtures, which acquired a yellowish tint whose inten�sity was proportional to the amount of GA, weremixed with 50, 100, and 300 µl of NaBH4 (4 mg/ml)and incubated with stirring for 2 h. The unreactedingredients were removed by exhaustive dialysis(membrane pore size, ~12000 Da; Sigma, UnitedStates). The resultant polymeric antigen was added toequal volume of glycerol and stored at –15°С.

Gel�NM × 5(ga), Gel�NM × 15(ga), Gel�NM ×25(ga). Aqueous solutions of gelatin (4 mg Gel,0.025 µmol) in 1 ml was supplemented with 22.7, 68.2,and 113.6 µl (125, 375, and 625 nmol) of NM solution

(5 mg/ml; 5�, 15�, and 25�fold molar excess, respec�tively) and 30 µl of a freshly prepared 2.5% GA solu�tion. After incubation at room temperature for 2 hwith stirring, each reaction mixture was mixed with100 µl of sodium borohydride (1 mg/ml), incubatedwith stirring for 2 h, and dialyzed for 2 days against 5 lof distilled water.

Gel�RS× 30(sp), Gel�RS × 100(sp). An aqueous RSsolution containing 4.2 mg of RS in 0.5 ml(9.25 µmol) was mixed with 4.1 mg of NaIO4 in 1 ml ofwater (19 µmol) and incubated at room temperaturefor 15 min with stirring. Aqueous solutions containing4 mg of Gel (0.025 µmol) in 1 ml were supplementedwith 30� and 100�fold molar excess (0.75 and 2.5 µmol)of oxidized RC in 120 and 300 µl, respectively. Afterincubation at room temperature for 1 h with stirring,each reaction mixture was mixed with 100 µl ofsodium borohydride (2 mg/ml), incubated for another1 h, and then dialyzed. The dialysates were mixed withan equal volume of glycerol and stored at –15°С in theform of solutions with a concentration of 1 mg/ml(protein).

Immunization procedure and antiserum preparation.Dark�colored rabbits weighing 2.5–3 kg were injectedsubcutaneously into the dorsal region (10–15 points)with TR�NM × 100(sp) emulsified in complete Fre�und’s adjuvant at a dose of 100 µg. Repeated injectionswere performed with aqueous solutions of the immu�nogen at the same dose. Blood was collected 7 daysafter each injection from the marginal ear vein. Serumwas separated, mixed with an equal volume of glycerol,and stored at –15°С until use.

Immune response assessment and ELISA optimization.The development of the immune response in animalswas assessed by the interaction of antisera in the indi�rect ELISA with conjugated antigens adsorbed on 96�well plates (Costar, United States). For this purpose,wells were filled with 0.2 ml of a conjugate solution inCBB in a concentration range of 0.05–1.5 µg/ml(1/1000–1/30 000 for Poly�NM (ga)) and incubatedat 4°С for 16 h. The plates were then washed four orfive times with 0.15 M PBS (pH 7.5) supplementedwith Tween�20 (PBS�T), after which the wells werefilled with 0.1 ml PBS�T and antiserum in serial dilu�tions in the same buffer supplemented with 1% BSA.After 1 h incubation, plates were washed again andwells were filled with 0.2 ml of anti�rabbit IgG anti�bodies conjugated with horseradish peroxidase. After1 h incubation and subsequent washing, wells werefilled with 0.2 ml of a substrate solution containing0.4 mg/ml о�phenylenediamine and 0.005% H2O2 in0.15 M citrate–phosphate buffer (pH 5.0). After 45 minincubation, the enzymatic reaction was stopped byadding 50 µl of 4 M sulfuric acid, and the plates werephotometered at 492 nm.

APPLIED BIOCHEMISTRY AND MICROBIOLOGY Vol. 47 No. 3 2011

DEVELOPMENT AND APPLICATION OF INDIRECT COMPETITIVE ENZYME 323

The immunoreagent concentration ratios obtainedby such “checkerboard” titration that ensured thereaction absorbance between 0.8 to 1.2 were used incompetitive assay.

Indirect competitive ELISA. The conditions ofanalysis did not differ from those described above. Theconjugated antigens were used for coating on plates attheir optimal concentrations. After incubation andsubsequent washing, the antiserum (0.1 ml) in appro�priate dilutions prepared in PBS�T containing 1%BSA was added to the wells simultaneously with anti�biotic solutions (1000–0.1 and 0 ng/ml) or samples inPBS�T or a sample diluted in the same buffer. Thefinal stages of the reaction were carried out asdescribed above.

The level of antibody binding to the wells with thezero NM concentration was taken as 100% and used tocalculate the percentage of antibody binding for eachconcentration of the compound.

The variant that ensured the highest sensitivity ofNM detection was selected among the coating anti�gens with different hapten load and used in furtherstudies.

Cross�reactivity of antibodies was determined asthe ratio between the NM concentration and respec�tive concentrations of RS, NA, GM, SSM, TM, KM,SM, and AP that inhibited binding by 50% (IC50) andwas expressed in percent. Graphical and statisticalanalysis of results was performed using Origin 8.0 soft�ware.

Preparation of samples and the recovery test. Withrespect to the fact that ingredients of milk had amarked effect on the antibody–antigen interaction,which was expressed in the signal reduction (the so�called “matrix effect”), we used skim milk powder tosimulate such interference. A solution of this reagent,prepared in accordance with the manufacturer’s rec�ommendations and corresponding to normal milk bythe degree of expression of the matrix effect, served asa medium for preparation of standard solutions.

Six different NM�negative (according to prelimi�nary tests) milk samples with 1.5–10% fat contentwere spiked with NM (final concentrations, 10000,1000, and 100 ng/ml). Sample preparation mean a dilu�tion with PBS�T by a factor of 100, and the correspon�dence of the antibiotic level to 100, 10, and 1 ng/ml wasdetermined by ELISA. Degree of recovery was deter�mined by the ratio of the found value (averaged for dif�ferent samples) to the added NM concentration.

To determine the effect of sterilization or pasteur�ization temperature on the content of NM in milk, anumber of milk samples with known NM concentra�tion were heated for 5 min in boiling water bath andthen subjected to ELISA testing.

RESULTS AND DISCUSSION

Absence of the characteristic pattern in the UVspectrum of NM did not allow us to evaluate the syn�thesized immunogen spectrophotometrically. Wedetected no evident differences in the electrophoreticmobility between TR(sp)NM × 30 and TR(sp)NM ×100 indicating an increase in the hapten density on thecarrier (data not shown). Nevertheless, to immunizerabbits, we used TR(sp)�NM × 100, the immunogenwith a higher hapten/carrier ratio by syntesis. Unlikeother authors who used glutaraldehyde and carbodi�imide condensation [4, 6], we prepared immunogenby the periodate method, which is also based on thereaction between protein and six possible aminogroups in the case of NM or four amino groups in thecase of NA. For this reason, the location of the haptenon the carrier, as in the above�mentioned papers,could vary.

The presence of more than one NH2 group in theNM molecule allowed us to consider the possibility ofpolymerization of the molecule by cross�linking withdialdehyde. Moreover, the GA: NM ratio determinedthe polymer type (linear or branched). The interactionof antibodies with Poly�NM�0.3(ga) was noticeablyweaker than with two other polymeric antigens.Apparently, this preparation had an insufficient degree ofpolymerization, and part of oligomers was eliminatedduring dialysis. We assumed that a triple excess of GAwith respect to NA in the case of Poly�NM�3(ga) shouldlead to branching of the polymer and, respectively,provide a greater masking of determinants in the anti�biotic molecule. However, this only slightly influencedthe efficiency of its binding to antibodies. The bestanalytical characteristics were exhibited by the variantwith immobilized poly�HM�1(ga). An equimolar ratioof the cross�linking agent and NM is a prerequisite forthe formation of a linear polymer, and the degree ofpolymerization (n) could be represented by the for�mula:

n ≥ 12000 Da/MW [NM–GA],where 12000 Da is the dialysis membrane cut off andMW [NM–GA] is the molecular weight of the penta�nal NM (699). Thus, the estimated degree of polymer�ization n was ≥ 17.

As a result of selection of optimum concentrationsof immunoreagents and competitive assays, the coatingantigens that ensured more sensitive NA determinationwere selected from the series of other conjugates with dif�ferent ratios of reagents used in the synthesis. The anti�gens that met this requirement were Gel�NM × 5 (ga)with the lowest hapten load and Gel�RS × 100(sp). Acharacteristic feature of the latter, unlike the previousconjugates, is the heterologous hapten, which hadthree out of four NM glycoside moieties (Fig. 1) and astrictly defined orientation on the carrier. The treat�ment of RS with NaIO4 led to oxidation of vicinal

324

APPLIED BIOCHEMISTRY AND MICROBIOLOGY Vol. 47 No. 3 2011

BURKIN, GALVIDIS

hydroxyl groups located on the rybosyl moiety of themolecule and their subsequent interaction with theamino groups of protein.

The analysis of the immune response dynamicsshowed that the titer of the blood serum increased bythe third bleeding and it could ensure a sufficientlyhigh level of assay sensitivity. Thus, the working dilu�tion of antiserum in using (ga)�conjugates was1/3000–1/7500. The using of Gel�RS × 100(sp)

required a higher concentration of antibodies (1/300–1/500).

The standard curves of ELISA systems designed onthe basis of selected pairs of immunoreagents areshown in Fig. 2. As was shown in [4], the sensitivity ofassay can be significantly affected by the structuralheterology (the type of hapten and the method of itsconjugation with the carrier) of the immobilized anti�gen compared with the immunogen. This study, whichincluded alternative approaches to designing immu�noreagents, is not an exception to this observation. Forexample, the limit of NM detection for the variantGel�RS × 100(sp)– ELISA was 0.1 ng/ml, and itsdinamic range was 0.1–100 ng/ml. An order of magni�tude less sensitivity was ensured by the polymeric anti�gen (1–1000 ng/ml), and the immobilized Gel�NM ×5(ga) allowed detection of 0.01 ng/ml NM with adinamic range of 0.01–10 ng/ml.

Cross�reactivity of various aminoglycosides withthe antibodies was assessed by the ratio of half�maxi�mal inhibitory concentrations IC50 (NM)/IC50 (ami�noglycoside) × 100%. The table shows the values thatreflect the competitive ability of these antibiotics interms of considered ELISA variants and illustrate theassay selectivity with respect to NM.

Weak recognition of NA and RS, analogues that arefragments of the NM molecule, in Poly�NM�1(ga)�ELISA and Gel�NM × 5 (m)�ELISA can be due to theblockade of amino groups (four out of six) in theneamine moiety of NM in the immunogen as a resultof conjugation. Thus, the major part of the pool ofserum antibodies was apparently generated against theNM fragment that is absent in NA and RS. The use of

OO

NH2

NH2 O

HOHO

HO

OH

O

O

H2N

NH2

HOHO

O OHNH2

NH2

OHO

OH

O

O

H2N

NH2

HOHO

O OHNH2

NH2

HOO

ONH2

OH

NH2

HO

H2NHO

HO

NH2

Neomycin BRibostamycin

Neamine(neomycin A)

Fig. 1. Chemical structure of aminoglycosides of the neomycin group.

100

00.01

20

40

60

80

10001001010.1

123

Binding, %

ng/ml

Fig. 2. Standard curves for NM determination in ELISAvariants based on antibodies to TR(sp)�NM and differentimmobilized conjugates: (1) Poly�NM(ga), (2) Gel�RS(sp), and (3) Gel�NM(ga). Each point represents themean value (n = 3), the error corresponds to the standarddeviation.

APPLIED BIOCHEMISTRY AND MICROBIOLOGY Vol. 47 No. 3 2011

DEVELOPMENT AND APPLICATION OF INDIRECT COMPETITIVE ENZYME 325

a solid phase with the immobilized Gel�RS × 100(sp)allowed us, owing to certain orientation of the haptenon the carrier, to use in the reaction only those anti�bodies that were generated against the fragment com�mon for NA, RS, and NM. This significantlyincreased the competitive ability of these compounds,which reached 36.2% for NA and 74.1% for RS. Com�pared with the two previous ELISA variants, Cross�reactivity of other 2�deoxystreptamine aminoglyco�sides (KM, NM and GM, SSM) also slightlyincreased; however, its level remained low (less than0.4%). Thus, the quantification of NM in all ELISAsystems was selective and did not depend on the pres�ence of other aminoglycoside drugs (GM, KM, SM,and AP) that are used in veterinary practice, whoseactivity was more than 100 times lower than the activ�ity of NM. Such a stringent specificity of antibodies(i.e., the absence of Cross�reactivitys with structuralanalogues) was also observed in other studies [4, 5] andcan be explained by the peculiarities of immunogendesign, when determinants common for some ami�noglycosides were shielded.

Since the MRL of NM in milk is high (1500 µg/kg)and the ELISA developed by us allows determinationof NM at concentrations as low as 0.01–1 ng/ml, forthe convenience of testing milk samples in all subse�quent experiments we used the less sensitive variantPoly�NM�1(ga)–ELISA. Samples were diluted 100 timeswith PBS�T buffer. At this dilution, the MRL fell inthe middle of the standard curve (15 ng/ml) and cor�responded to 50% binding (Fig. 2). Thus, under theseconditions, the degree of contamination of milk canbe recorded in a broad range, one order of magnitudeabove and below the maximum value. Other, moresensitive variants can be used for analyzing objectswith a prominent matrix effect, which can be elimi�nated only at additional sample dilution.

To test samples artificially contaminated with anti�biotic several milk samples were preliminarily provedto be NM�free. In addition, we estimated the necessityof milk defatting at the stage of sample preparationusing milk samples with 1.5, 2.5, 3.2, 4.0, and 10% fatcontent. ELISA of these milk samples (n = 6) contain�ing 10000, 1000, and 100 ng/ml NM revealed no sig�nificant differences between samples with different fatcontent. Thus, a 100�fold dilution of milk, regardlessof its fat content, ensured 84 to 125.2% analyte recov�ery (on average, 111.7%) without sample pretreat�ment.

Since we used pasteurized and sterilized ratherthan unprocessed milk samples, we assessed the ther�mal stability of NM. Heating in a boiling water bathfor 5 min had no effect on the level of NM detection;i.e., the immunochemical activity of NM did notchange. Therefore, the antibiotic did not undergo deg�radation during milk processing, and NM can beimmunochemically detected in milk both before andafter its thermal processing.

A variety of milk samples (n = 106) differing in fatcontent, thermal processing method, and other nutri�tional criteria, as well as in the time and place of man�ufacturing, was an adequate real object for assessmentof the test relevance and verification of its analyticalcharacteristics. ELISA revealed NM in more than halfof the samples (57 out of 106) at concentrations higherthan 100 ng/ml (Fig. 3). The level of contaminationwith NM, except for one sample (1690 ng/ml), waslower than the MRL established in EU (1500 µg/kg).However, the threshold recommended by WHO(500 µg/kg) [3] was exceeded in 10% of cases (11 out of106). An important advantage of the developed testconsists in the possibility to analyze subthreshold con�centrations of the antibiotic. For example, the contentof NM (100–500 ng/ml) in 46 milk samples was belowthe sensitivity threshold of microbiological tests.

Cross�reactivity of antibodies in different ELISA variants

Immobilized antigen

IC50, ng/ml Relative cross�reactivity, %

NM NM NA RS GM SSM TM KM SM AP

Poly�NM�1(ga) 15.4 100 1.3 3.6 <0.4 <0.4 <0.4 <0.4 <0.4 <0.4

Gel�RS×100(sp) 2.9 100 36.2 74.1 0.1 0.2 0.3 0.4 <0.07 <0.07

Gel�NM×5(ga) 0.3 100 0.4 0.2 <0.02 <0.02 <0.02 <0.02 <0.02 <0.02

326

APPLIED BIOCHEMISTRY AND MICROBIOLOGY Vol. 47 No. 3 2011

BURKIN, GALVIDIS

Thus, on the basis of neomycin�specific antibodiesand a number of antigens adsorbed onto polystyreneplates, we developed ELISA systems for quantitativedetection of the antibiotic. Due to significant reservesof sensitivity and the detection limit (0.01 ng/ml),which is comparable and even exceeds some describedsystems [4–6] and commercial test systems [7;www.biokits.com], the maximum residue level of NMcould be detected even in the least sensitive variant with�out any special pretreatment of samples at a 100�folddilution of milk. The study of milk contamination withthis aminoglycoside using the developed assay allowedus to obtain the first results that are worth specialattention and may serve as a guide for considering theissue on the introduction of the regulation of NM inthe Russian Federation; the method of measurementproposed in this study can be used as a tool for moni�toring NM content.

REFERENCES

1. Kovalev, V.F., Volkov, I.B., and Violin, B.V., Antibiotiki,sul’fanilamidy i nitrofurany v veterinarii (Antibiotics,Sulfonamides, and Nitrofurans in Veterinary Medi�cine), Moscow: Agropromizdat, 1988.

2. Council Regulation (EEC) No. 2377/90, Official J.European Union, 1990, vol. L224, pp. 26–27.

3. Gigienicheskie trebovaniya bezopasnosti i pishchevoitsennosti pishchevykh produktov (Hygienic Safety andNutritional Value of Foods), Moscow: FGUP InterSEN, 2002.

4. Chen, Y., Shang, Y., Wu, X., Qi, Y., and Xiao, X., FoodAgric. Immunol., 2007, vol. 18, no. 2, pp. 117–128.

5. Jin, Y., Jang, J.W., Lee, M.H., and Han, C.H., Clin.Chim. Acta, 2006, vol. 364, nos. 1–2, pp. 260–266.

6. Loomans, E.M.G., van Wiltenburg, J., Koets, M., andvan Amerongen, A., J. Agric. Food Chem., 2003, vol. 51,no. 3, pp. 587–593.

7. Diana, F., Paleologo, M., and Persic, L., Food. Add.Contam., 2007, vol. 24, no. 12, pp. 1345–1352.

1000

100

10000

100

10 20 30 40 50 60 70 80 90 100Sample no.

NM, ng/ml

<100

500*

1500*1

2

Fig. 3. Neomycin (NM) content in milk samples according to the results of ELISA.1—NM MRLs in milk according to EU regulations (1500 µg/kg);2—NM MRLs in milk recommended by the FAO/WHO (500 µg/kg).