Purification of erythropoietin from human plasma samples using an immunoaffinity well plate

Journal of Chromatography B, 878 (2010) 2117–2122

Contents lists available at ScienceDirect

Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb

Purification of erythropoietin from human plasma samples using an immunoaffinity well plate J. Mallorquí a,b , E. Llop a , C. de Bolòs c , R. Gutiérrez-Gallego a,b , J. Segura a,b , J.A. Pascual a,b,∗ a b c

Bioanalysis Group, Neuropsychopharmacology Research Program, IMIM-Hospital del Mar. PRBB, Barcelona, Spain Department of Experimental and Health Sciences, Pompeu Fabra University, PRBB, Barcelona, Spain Gastric Carcinogenesis Group, Cancer Research Program, IMIM-Hospital del Mar. PRBB, Barcelona, Spain

a r t i c l e

i n f o

Article history: Received 21 April 2010 Accepted 20 June 2010 Available online 1 July 2010 Keywords: Erythropoietin Doping Human Plasma Immunoaffinity N-Glycolyl-neuraminic acid

a b s t r a c t A method is described to isolate human erythropoietin (hEPO) from plasma using an EPO-specific immunoaffinity micro well plate (IAP). The operating conditions of the method (binding, blocking and elution) were optimised to avoid isoform discrimination and cross-contamination with other glycoproteins. The overall hEPO recovery was ca. 56% and significant clean-up for plasmatic hEPO was achieved. Polyvinylpyrrolidone (PVP) was used as a blocking reagent and elution took place at pH 11.0. Under these conditions all isoforms from recombinant human EPOs (rhEPOs) and analogues were uniformly recovered guaranteeing lack of discrimination. The resulting procedure allowed isolating erythropoietin from plasma in conditions amenable to hEPO analysis by other techniques such as SDS-PAGE or IEF. Moreover, avoiding contamination with other glycosylated material allowed the identification in human plasma samples of the non-human N-glycolyl-neuraminic acid (Neu5Gc) using HPLC-FLD. Neu5Gc is present as 1–2% of the sialic acid content in rhEPO so this approach could be used to unequivocally detect abuse of rhEPOs or analogues as part of the doping control. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Human erythropoietin (hEPO) is the hormone responsible for stimulating the production of red blood cells. It is a glycoprotein, mainly synthesised in the kidney, and consists of a single 165 amino acid polypeptide chain with two disulfide bonds between cysteine residues 7-161 and 29-33. It also contains three N-linked (Asn-24, 38, 83) and one O-linked (Ser-126) glycosylation sites. The resulting carbohydrates content accounts for roughly 40% of the total molar mass of the glycoprotein (∼30 kDa) [1,2]. Since 1989 a series of recombinant preparations of hEPO and analogues are available for clinical use, all containing different glycosylation profiles resulting in different pharmacokinetic prop-

Abbreviations: hEPO, human erythropoietin; NESP, novel erythropoiesis stimulating protein; rhEPO, recombinant human EPO; CERA, Continuous Erythropoietin Receptor Activator; uhEPO, urinary human EPO; Neu5Gc, N-glycolyl-neuraminic acid; IAP, immunoaffinity plate; IAC, immunoaffinity column; ELISA, enzyme linked immunosorbent assay; mAb, monoclonal antibody; HPLC-FLD, high performance liquid chromatography with fluorescence detection; HRP, horseradish peroxidase; SDS-PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis; IEF, isoelectric focusing; BSA, bovine serum albumin; PVP, polyvinylpyrrolidone; TFA, trifluoroacetic acid; WADA, World Anti-Doping Agency. ∗ Corresponding author at: Bioanalysis Group, Neuropsychopharmacology Research Program, IMIM-Hospital del Mar. PRBB, Dr. Aiguader, 88, 08003-Barcelona, Spain. Tel.: +34 933 160 500; fax: +34 933 160 467. E-mail address: [email protected] (J.A. Pascual). 1570-0232/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jchromb.2010.06.025

erties. They have been marketed as epoetin alpha, beta, omega, delta (dynepoTM ), darbepoetin alpha (NESP, aranespTM ), pegserpoetin alpha (CERA, mirceraTM ), amongst others [3]. Besides, rhEPO and analogues are known to be misused by athletes to increase oxygen transport in an attempt to improve endurance capacity. This is considered a doping practice and as such these substances are prohibited in sport since their development [4]. The method currently used in doping control to differentiate between exogenous and endogenous hEPO in urine is based on their distinct behaviour when submitted to isoelectric focusing (IEF) [5]. Samples (20 mL of urine) have to be concentrated down to ca. 20 ␮L in order to be applied to the IEF gel. This is routinely done by filtering through a 30 kDa MWCO device. Besides, given the great variety of recombinant preparations currently available, additional characterisation may also be obtained through SDS-PAGE analysis where endogenous hEPO and its recombinant counterparts would show a differing behaviour owing to changes in their molecular weight. An immunopurification step using a commercially available ELISA kit has already been used for this purpose [6,7]. On the other hand, another way to unequivocally differentiate endogenous hEPO from rhEPOs and analogues is through sialic acid analyses [8]. It has already been proven that rhEPO produced in CHO (Chinese Hamster Ovary) and BHK (Baby Hamster Kidney) cells contain small amounts of N-glycolyl-neuraminic acid (Neu5Gc). This is a sialic acid that cannot be endogenously produced by humans since we lack the corresponding enzyme (CMP-Neu5Ac hydroxylase) [9,10]. There-

2118

J. Mallorquí et al. / J. Chromatogr. B 878 (2010) 2117–2122

fore, the detection of Neu5Gc in hEPO present in human urine or plasma would constitute an absolute proof of the exogenous origin of the hormone. The development of procedures to purify hEPO or analogues from plasma in order to make them compatible with the analysis by IEF and SDS-PAGE could be a valuable analytical approach from an anti-doping perspective. In particular, plasma hEPO concentrations are an order of magnitude higher than in urine and CERA, for example, can almost exclusively be detected in plasma. Different immunoaffinity approaches for the purification of hEPO from plasma or serum have already been tried. In 2001, Skibeli et al. purified hEPO, using magnetic beads coated with a polyclonal rabbit anti-hEPO antibody. The recovery of rhEPO from plasma was around 20% [11] and the resulting purified serum hEPO showed an IEF profile very shifted towards basic pI values as compared to its urinary counterpart (uhEPO) [12]. In 2007, Lasne et al. [13] applied an immunoaffinity column prior to the IEF analysis of hEPO from serum. They showed a serum hEPO profile slightly more basic than the regular uhEPO. In this paper, a study is presented on the development of an immunopurification method using in-house prepared microwell plates coated with a commercially available monoclonal anti-hEPO antibody, establishing the binding, blocking and elution conditions to avoid isoform selectivity and cross-contamination with other biological materials, particularly glycoproteins. Recovery and clean-up were thoroughly investigated. The resulting procedure is intended to be used for isolation of erythropoietin from plasma prior to its analysis by either IEF or SDS-PAGE as well as for glycan analyses [14], with particular emphasis on the detection of its nonhuman Neu5Gc content. The procedure was successfully applied to the analysis of plasma samples from subjects suspected of having used rhEPO.

2. Experimental 2.1. Chemicals and reagents Reference preparation of rhEPO (equimolar mixture of epoetin alpha and beta) was obtained from the European Pharmacopoeia Commission Biological Reference Preparation (BRP) batch no. 2. Darbepoetin alpha or NESP (aranespTM ) was obtained as the pharmaceutical preparation from Amgen (syringe containing 10 ␮g of NESP in 0.4 mL solution). Pegserpoetin alpha or CERA (mirceraTM ) was obtained as the pharmaceutical preparation (syringe containing 200 ␮g in 0.3 mL solution) from Roche. Human urinary EPO (uhEPO) was purchased from the National Institute for Biological Standards and Control (NIBSC). Proteasefree Tris, glycine, phosphate-buffered saline (PBS), Tween 20, polyvinylpyrrolidone (PVP) average Mw 40 kDa, bovine serum albumin (BSA), 1,2-diamino-4,5-methylenedioxy-benzene (DMB) and 2-mercaptoethanol were from Sigma. CompleteTM , protease inhibitor cocktail, was from Roche. Steriflip filters (0.22 ␮m), Ultrafree-MC (MWCO 50 kDa), Amicon Ultra-15 and Ultra-4 (MWCO 30 kDa), Durapore (0.65 ␮m) and Immobilon-P (0.45 ␮m) membranes were from Millipore. ELISA assay kit (Quantikine IVD Erythropoietin), anti-hEPO monoclonal antibody (clone 9C21D11), anti-hEPO monoclonal antibody (clone AE7A5) and polyclonal anti-hEPO antibody (AB-286-NA) were from R&D Systems. EPO Chemiluminescent Immunoassay (Immulite 1000 EPO) and diluent were from DPC. Urea, gelbond film, electrode paper and DTT were from Amersham-Pharmacia. Acrylamide-Bis (97:3, w/w) and SDS were from Merck. Ampholytes Servalyt 2-4 and 4-6 were from Serva. TEMED was from BioRad. Supersignal West Femto and biotinconjugated goat-anti-mouse IgG (H + L) were from Pierce. Streptavidin-HRP was from Biospa. Plates (96 wells EIA/RIA high

binding) were from Costar. Sialic acid reference panel was from Glyco. N-glycolyl-neuraminic acid (Neu5Gc) was from Calbiochem. All other chemicals were of the highest purity commercially available. 2.2. Plasma samples Collection of lithium heparin plasma samples from healthy volunteers (control plasmas) was approved by the Local Ethics Committee. Concentrations ranged 9–18 mIU/mL. Plasma samples (CPD) from three subjects, with high hEPO concentrations (87, 75 and 34 mIU/mL), suspected of having used rhEPO were obtained from the doping control laboratory of Barcelona. Samples were kept at −20 ◦ C until analysis. 2.3. Optimisation of blocking, binding and elution conditions on the immunoaffinity plate The efficiency of three different blocking reagents was studied as follows: the monoclonal anti-hEPO antibody (clone 9C21D11) in 50 mM NH4 HCO3 pH 9.0 was applied to three 96 well plates (50 ␮L/well) at different concentrations (0, 0.025, 0.05, 0.1, 0.25 and 0.5 ␮g/mL). After incubation for 1 h at 37 ◦ C wells were washed three times with 300 ␮L of 10 mM Phosphate-Buffered Saline pH 7.4 (PBS). Then, the different blocking reagents (1% BSA, 1% Gelatine or 1% PVP, all dissolved in PBS) were added at 300 ␮L/well, one to each plate. After incubation overnight at 4 ◦ C the reagent was removed and the plates were ready to use. To evaluate the blocking capacity in each case, 50 ␮L/well of anti-mouse IgG antibody-HRP conjugate (1/1000 in PBS) were added and incubated for 1 h at 37 ◦ C. Then, wells were washed three times with 300 ␮L of 0.1% Tween 20 in PBS at pH 7.4 (PBS-Tween) and 50 ␮L of substrate solution (from the Quantikine EPO kit) were added to each well and kept at room temperature (RT) for 20 min. Finally 25 ␮L stop solution (also from the Quantikine EPO kit) was added to each well and the absorbance was read at 450 nm within 15 min. Additionally, in order to further study the blocking efficiency of PVP, a second experiment was done in which plates were first blocked with 1% PVP in PBS and then, after removing the reagent, 50 ␮L of the 0.5 ␮g/mL solution of the mAb 9C21D11 were applied to each well. After incubation and washing, the ELISA test was continued as described in the previous experiment expecting to get only background signals if the blocking was efficient enough. The binding conditions of the primary antibody to its substrate (hEPO) were studied using an immunoaffinity plate (IAP) prepared following the established conditions as follows: A 10 ␮g/mL solution of the mAb 9C21D11 in 50 mM NH4 HCO3 , pH 9.0 (100 ␮L/well) was applied to a 96 well micro plate and incubated for 1 h at 37 ◦ C. Plates were then washed three times with 300 ␮L PBS and blocked with 1% PVP in PBS (300 ␮L/well) overnight at 4 ◦ C with shaking. In parallel, rhEPO standard curves were generated using our plates (IAP) and commercial Quantikine plates under different binding conditions as follows: 100 ␮L of rhEPO standard solutions (5, 20, 50, 100 and 200 mIU/mL) was added to each well. Plates were incubated either for 1 h at RT or overnight at 4 ◦ C both with shaking. Then the rest of the ELISA assay procedure was followed. Solutions were removed and wells were incubated with 200 ␮L of rabbit anti-rhEPO polyclonal antibody conjugated to horseradish peroxidase for 1 h at 37 ◦ C. Wells were washed four times with 300 ␮L of washing solution and 200 ␮L of substrate solution were added and incubated at RT for 20 min. Finally, 100 ␮L of stop solution were added and plates were read at 450 nm within 15 min. Three standard curves were generated and compared. All reagents for the ELISA assay were those from the Quantikine kit from R&D systems.

J. Mallorquí et al. / J. Chromatogr. B 878 (2010) 2117–2122

In order to study the performance of different elution buffers, 6.4 mL plasma samples (control plasma from a subject not using any rhEPO product) and samples spiked with 1 ng/mL rhEPO, NESP or CERA were applied to the IAP at 100 ␮L/well and incubated overnight at 4 ◦ C with shaking. Then, wells were washed twice with 300 ␮L PBS-Tween and hEPO was eluted with 250 ␮L of either 0.4 M glycine supplemented with 6 M urea and adjusted to a final pH 11.0 or 0.7% acetic acid (resulting pH 2.8). Eluates were concentrated by ultra-filtration through Amicon-15. Pellets were frozen in liquid nitrogen, lyophilised and analysed following the routine IEF method for uhEPO [5]. The software GASEPO version 2.0 (2010 Seibersdorf Labor GmbH) was used for gel image processing. The conditions finally used for immunopurification using IAP were as follows: Samples (100 ␮L/well) were applied to the IAP (microtiter plate coated with a 10 ␮g/mL solution of mAb 9C21D11 and blocked with 1% PVP in PBS) and incubated overnight at 4 ◦ C with shaking. Then, wells were washed twice with 300 ␮L PBSTween. Erythropoietin was eluted with 0.4 M glycine, 6 M urea, pH 11.0 (for IEF method) or acetic acid 0.7%, pH 2.8 (for sialic acid analyses of rhEPO). Eluates were collected and kept at 4 ◦ C. 2.4. Efficiency of clean-up and hEPO recovery Initial plasma samples to be purified as well as different fractions through the application of the procedure (unbound, first and second washings and the elution fractions) were ultra-filtered through Amicon-15 to obtain a pellet of 350 ␮L. Total protein content and hEPO concentration in each sample were measured by a Bradford test using a nanophotometer (Nanodrop, Thermo scientific) and IMMULITE 1000 (chemiluminescent immunoassay), respectively. 2.5. Sialic acid analysis Immunopurified dry plasma extracts were hydrolysed in 200 ␮L 0.1 M TFA for 1 h at 80 ◦ C and filtered through Ultrafree-MC and the filtrate lyophilised. Derivatization was performed by reconstituting the sample in 8 ␮L of water and adding 8 ␮L of DMB solution (7 mM DMB in 1.4 M aqueous acetic acid containing 18 mM sodium hydrosulphite and 1 M ␤-mercaptoethanol) and incubating 2 h at 50 ◦ C. DMB derivatives of sialic acids were analysed by reversed phase capillary-HPLC on a Zorbax SB-C18 (150 mm × 0.3 mm, 3.5 ␮m) column using ACN:H2 O (20:80) as isocratic mobile phase at a flow rate of 4 ␮L/min. Chromatographic run was completed within 35 min. Chromatographic analyses were performed on an Agilent 1100 series capillary instrument equipped with a JASCO micro21FP capillary fluorescence detector (ex = 373 nm and em = 448 nm). 2.6. Application to human plasma samples Plasma samples from four subjects, one control subject not using any rhEPO product and three subjects suspected of having used rhEPO according to their very high plasmatic concentrations were analysed following the procedure. Plasma samples (6.4 mL) were immunopurified using the developed method as described above. Eluates were ultra-filtered through Amicon-15 and pellets were frozen in liquid nitrogen and lyophilised. Finally samples were re-suspended in 20 ␮L of H2 O and analysed by the IEF method [5]. For the analysis of Neu5Gc the immunopurified eluates were directly lyophilised and subject to the analytical procedure for sialic acids as described above. 3. Results and discussion The development of a purification method for hEPO based on immunoaffinity micro well plates (IAP) is an attractive alternative to other approaches [11–13] because plates are disposable, easy to

2119

Fig. 1. Direct ELISA standard curves prepared by coating three immunoaffinity plates with increasing concentrations of the mAb 9C21D11. Plates were blocked with either 1% BSA (white bar), 1% PVP (gray bar) or 1% gelatine (black bar) in PBS. Absorbance (450 nm) was developed after direct incubation with anti-mouse IgG antibody-HRP conjugate (1/1000 in PBS) and addition of the corresponding substrate.

use and amenable to the simultaneous processing of a great number of samples. Although they can be commercially obtained as part of ELISA kits, purification can be affected by factors like binding capacity, cross-contamination with other matter present in the wells (e.g. blocking materials) and by the elution conditions chosen. When 200 ␮L of an elution buffer (either at pH 2.8 or pH 11.0) were applied to a commercial ELISA plate and incubated for two min, the amount of proteins present in the eluate, as determined by a Bradford test accounted for ca. 8 ␮g protein eluted from each well. Considering that the amount of antibody that can be bound to a well is around 0.3–0.5 ␮g [15], it became clear that a large amount of other proteic material was eluted from the well. Most probably proteins from the complex mixture used as blocking reagent. Those proteins may result in the sample being incompatible with further analytical purposes, as sialic acid analyses. 3.1. IAP purification method development Blocking, binding and elution conditions were studied to develop an immunopurification method with optimal hEPOantibody binding, ensuring no isoform discrimination and avoiding cross-contamination with other glycoproteic material. First, in order to test if any protein-free blocking conditions could be used, as efficiently as the well established protein-based reagents (e.g. gelatine, animal sera, etc.), we studied the behaviour of a usual blocking reagent (1% gelatine in PBS) with another protein (1% BSA in PBS) and a non-proteic reagent (1% PVP in PBS). When building ELISA tests on plates where those reagents were used for blocking, virtually the same results were obtained (Fig. 1). Accordingly, PVP was chosen as the blocking reagent in order to avoid contamination. The blocking efficiency of PVP was confirmed when two ELISA tests were done blocking the wells with 1% PVP in PBS before or after coating with the antibody. Results confirmed that when wells were pre-blocked with PVP, no antibody (nor any other further reagent) was bound hence resulting in very low signals in the ELISA tests (data not shown). Second, in order to study the hEPO-antibody binding, two different incubation conditions were tested. Incubations overnight at 4 ◦ C with shaking resulted in a 100% increase of rhEPO bound, as compared to the regular 1 h at RT. The resulting ELISA calibration curve is shown in Fig. 2 as compared to the calibration curve of the commercially available Quantikine EPO ELISA kit run in parallel. The monoclonal anti-hEPO antibody clone 9C21D11 and overnight incubations were used throughout the study in order to maximize

2120

J. Mallorquí et al. / J. Chromatogr. B 878 (2010) 2117–2122

Fig. 2. Sandwich ELISA standard curve generated for rhEPO using our immunoaffinity plate (gray solid line), incubating the EPO samples overnight at 4 ◦ C as compared with the commercial QuantikineTM kit (black dashed line).

the capacity of the method. Attempts to use the monoclonal antihEPO antibody clone AE7A5 (used in the current IEF method for hEPO as it has shown to have the best sensitivity for blotting purposes [16,17]) or the polyclonal anti-hEPO antibody (AB-286-NA), in the selected conditions, resulted in a much lesser sensitivity (data not shown). This is not surprising since, clone AE7A5 had already shown to perform very badly when immobilised on a surface [18]. Finally, elution was studied considering that the elution buffer shall be able to disrupt the antigen–antibody binding while not degrading the analyte and producing (in principle) a uniform, nonselective, elution of all isoforms either from endogenous hEPO, rhEPOs or analogues like NESP or CERA. Interestingly, it was found that when elution was performed at an acidic pH (0.7% acetic acid, pH 2.8) basic isoforms were preferentially eluted, as tested by IEF. As a consequence the overall IEF profile changed with an apparent shift towards the cathode and more importantly, the loss of those very acidic isoforms, i.e. the whole NESP profile, for example [19]. The enrichment in basic bands (those present in rhEPO or CERA) could be exploited for the analysis of sialic acids where the elution with acetic acid simplified the further processing of the sample.

Fig. 4. Total protein content (black line) and hEPO amounts (bars) present in each fraction through the immunopurification procedure: initial plasma sample, unbound fraction (plasma removed after incubation), first wash, second wash and elution fraction.

Elution at basic pH (0.4 M glycine, 6 M urea, pH 11.0) resulted in a profile identical to the applied standard, showing no isoform discrimination. In order to check for the effect of the presence of urea (amenable to the further IEF analysis) on the elution behaviour, the same buffer composition was kept but taking it to pH 2.8. The elution showed the same skewed behaviour as with acetic acid showing that pH plays a key role. Fig. 3 shows the results of analysing human plasma samples immunopurified using the method developed and eluted under basic and acidic conditions. Reference preparations of rhEPO, NESP and CERA as well as the analysis of plasma samples spiked with these compounds are shown. 3.2. Evaluation of clean-up efficiency and recovery Total protein content in plasma is very high (60–80 g/L). Conversely, hEPO is present in very low concentration (30–170 ng/L). After IAP purification, using basic pH elution, the protein content of the eluate was reduced three thousand times. From 403 mg of total protein present in 6.4 mL plasma, 387 mg (96.0%) were found in the unbound fraction and another 12.8 mg (3.2%) in the combined washing fractions. However total protein content in the eluate was

Fig. 3. Isoelectric focusing (IEF) analysis of reference preparations of hEPO products as well as human plasma samples immunopurified using the newly developed procedure. Left panel: IEF gel images. Right panel: densitometric profiles of the corresponding IEF lanes. (a) Reference preparation of uhEPO from NIBSC; (b) mixture of the Biological Reference Preparation (BRP) of rhEPO plus NESP; (c) CERA (mirceraTM ) from Roche; (d) control plasma eluted under basic conditions (pH 11.0), (e–g) control plasma spiked with 1 ng/mL rhEPO, 1 ng/mL NESP and 1 ng/mL CERA, respectively, eluted under basic conditions (pH 11.0). (h) Control plasma showing isoform discrimination after elution under acidic conditions (pH 2.8). Horizontal lines divide lanes in basic (upper), endogenous (middle) and acidic (lower) areas according to the WADA’s Technical Document TD2009EPO.

J. Mallorquí et al. / J. Chromatogr. B 878 (2010) 2117–2122

2121

Fig. 5. Isoelectric focusing (IEF) analysis of different rhEPO products as well as human plasma samples immunopurified using the newly developed procedure. Left panel: IEF gel images. Right panel: densitometric profiles of the corresponding IEF lanes. (a) Reference preparation of uhEPO from NIBSC; (b) mixture of the Biological Reference Preparation (BRP) of rhEPO plus NESP; (c) control plasma (14 mIU/mL); (d–f) plasma samples from subjects suspected of having used rhEPO (87, 75 and 34 mIU/mL respectively). Horizontal lines divide lanes in basic (upper), endogenous (middle) and acidic (lower) areas according to the WADA’s Technical Document TD2009EPO.

still 0.13 mg, much higher than the amount of EPO present (Fig. 4). The recovery of EPO from human plasma when using the basic pH conditions was around 56% (Fig. 4). The behaviour of the calibration curve between 5 and 200 mIU/mL showed that recovery is constant

for the concentration range studied without a significant impact of a matrix effect for plasma. Our experience showed that hEPO recovery when it was immunopurified using an IAC column (Affigel-HZ from BioRad) and the same antibody (clone 9C21D11) at 1 mg/mL was also around 50% using elution under the same basic conditions (data not shown), but there was a significant portion of the analyte that was not eluted from the column. By reusing such IAC columns, the increasing amounts of hEPO retained eventually ended up being eluted together with the corresponding elution fraction of a sample. Using the IAP system developed all hEPO not eluted was found in the unbound fraction (44%) with no hEPO detectable in any of the washing fractions. When eluting at acidic pH, recovery was around 25%. In those conditions, 50% of the hEPO was found in the unbound fraction while another 25% remained bound to the antibody. Considering the loss of acidic isoforms from hEPO or analogues eluted at acidic pH, the hEPO more strongly bound to the antibody must be enriched in those more acidic isoforms, showing the magnitude of the isoforms selectivity. Other authors using different immunopurification approaches (e.g. magnetic beads) described recoveries for hEPO around 20% while showing very basic IEF profiles. It is possible that those low recoveries were the result of isoform selectivity, contributing to the peculiar profiles shown [11]. On the other hand, this selectivity has proved to be useful in identifying the compound present resulting in a potentially faster and cheaper screening procedure for rhEPO [19].

3.3. Implementation for human plasma samples

Fig. 6. HPLC-FLD analysis (excitation 373 nm, emission 448 nm) of the DMBderivatized sialic acid residues derived from hEPO immunopurified from control plasma (black thin line); control plasma spiked with Neu5Gc (gray thin line); plasma samples from subjects suspected of having used rhEPO (black and gray thick lines); standard sialic acid (SA) mixture.

In Fig. 3, it was shown how control plasma samples and those spiked with reference substances could be analysed using the procedure. After ensuring no isoform discrimination, we could see that plasma hEPO may show a profile slightly more basic than the regular urinary hEPO as also described for serum hEPO [13]. Despite the

2122

J. Mallorquí et al. / J. Chromatogr. B 878 (2010) 2117–2122

slight basic shift, plasma hEPO still failed to comply with the identification criteria for rhEPO as described in WADA’s TD2009EPO [17] thus resulting in no false “positive” evaluation. When the same control plasma was spiked with 0.3 or 1 ng/mL of rhEPO, the increase in the intensity of the bands corresponding to rhEPO made the profile comply with the identification criteria for rhEPO (Fig. 3). Also, when control plasma samples were spiked with NESP, its four bands appeared in the acidic area of the IEF profile. Control plasma samples spiked with CERA showed the particular CERA profile in the basic area with some bands approximately co-localised with those defined by rhEPO and others interspersed amongst them [17]. The procedure was also applied to real samples, some control and others from individuals suspected of having used rhEPO (suspicious samples) owing to their elevated concentrations of hEPO in plasma. Fig. 5 shows the IEF profiles obtained after immunopurification. The three suspicious samples with elevated plasma hEPO concentrations resulted in profiles compatible with the presence of rhEPO. When these samples were immunopurified using the basic and acidic conditions, the profiles obtained were identical, since all bands present are in the basic area for which acidic conditions showed good recoveries [20]. Taking profit of the fact that the plates used did not contain any non-human proteic material that could be eluted from the wells, a control plasma sample and samples from two of the subjects suspected of using rhEPO were analysed for the presence of Neu5Gc the non-human sialic acid only present in the recombinant glycosylated preparations. Fig. 6 shows the results obtained. Sialic acid analyses of immunopurified samples indicated, that control plasma used as a negative control did not contain any Neu5Gc while the peak corresponding to Neu5Gc clearly appeared in the suspicious samples. 4. Conclusions In this study, a hEPO-specific immunoaffinity procedure was developed and used to purify and concentrate hEPO in human plasma samples. The overall hEPO recovery was 56% while the content of other proteins was reduced by three orders of magnitude. Elution showed to be very sensitive to pH. Under acidic conditions (pH 2.8) there was an obvious discrimination in favour of the more basic isoforms while under basic conditions (pH 11.0) there was no discrimination. Ensuring no isoform discrimination renders the IAP a valuable tool for the treatment of plasma, complementary to the current detection method (IEF) in doping control of rhEPOs or analogues. Besides, avoiding cross-contamination with other non-

human glycoproteins, the IAP allows applying post-immunoaffinity procedures such as the analysis of Neu5Gc in samples from subjects suspected of misusing rhEPOs or analogues. Acknowledgements This project has been carried out with the support of WADA, the Spanish Ministry of Science and Education (DEP2006-56207-C0303) and DIUE de la Generalitat de Catalunya (2009 SGR 492). The authors are thankful to Joan Marc Carbo for his support in image processing. References [1] P.H. Lai, R. Everett, F.F. Wang, T. Arakawa, E. Golwasser, J. Biol. Chem. 261 (7) (1986) 3116. [2] W. Jelkmann, Intern. Med. 43 (8) (2004) 649. [3] I.C. Macdougall, K.U. Eckardt, Lancet 368 (2006) 947. [4] World Anti-Doping Agency (WADA) 2009. Available at http://www.wadaama.org/rtecontent/document/2009 Prohibited List ENG Final 20 Sept 08.pdf (accessed 10 March 2009). [5] F. Lasne, L. Martin, N. Crepin, J. de Ceaurriz, Anal. Biochem. 311 (2002) 119. [6] M. Kohler, C. Ayotte, P. Desharnais, U. Flenker, S. Lüdke, M. Thevis, E. VölkerSchanzer, W. Schänzer, Int. J. Sports Med. 29 (1) (2008) 1. [7] C. Reichel, R. Kulovics, V. Jordan, M. Watzinger, T. Geisendorfer, Drug Test. Anal. 1 (2009) 43. [8] E. Llop, R. Gutiérrez-Gallego, J. Segura, J. Mallorquí, J.A. Pascual, Anal. Biochem. 383 (2) (2008) 243. [9] C.H. Hokke, A.A. Bergwerff, G.W. van Dedem, J. van Oostrum, J.P. Kamerling, J.F. Vliegenthart, FEBS Lett. 275 (1–2) (1990) 9. [10] M. Nimtz, W. Martin, V. Wray, K.D. Klöppel, J. Augustin, H.S. Conradt, Eur. J. Biochem. 213 (1) (1993) 39. [11] V. Skibeli, G.A. Nissen-Lie, A. Noreau, P. Torjesen, P. Hemmersbach, K.I. Birkeland, in: W. Schänzer, H. Geyer, A. Gotzmann, U. Mareck (Eds.), Recent Advances in Doping Analysis, Sport verlag Strauss, Köln, 1998, p. 313. [12] V. Skibeli, G. Nissen-Lie, P. Torjesen, Blood 98 (13) (2001) 3626. [13] F. Lasne, L. Martin, J.A. Martin, J. de Ceaurriz, Int. J. Biol. Macromol. 41 (3) (2007) 354. [14] E. Llop, R. Gutiérrez-Gallego, V. Belalcázar, G.J. Gerwig, J.P. Kamerling, J. Segura, J.A. Pascual, Proteomics 7 (23) (2007) 4278. [15] J.E. Butler, Methods 22 (1) (2000) 4. [16] E. Giménez, C. De Bolós, V. Belalcazar, D. Andreu, E. Borrás, B.G. De la Torre, J. Barbosa, J. Segura, J.A. Pascual, Anal. Bioanal. Chem. 388 (7) (2007) 1531. [17] World Anti-Doping Agency (WADA) 2009. Available at http://www.wada-ama. org/Documents/World Anti-Doping Program/WADP-ISLaboratories/WADA TD2009EPO EN.pdf (accessed 12 March 2010). [18] V. Belalcázar, J. Segura, C. de Bolòs, S.H. Peng, R. Gutiérrez-Gallego, E. Llop, J.A. Pascual, Poster, 15th IFCC-FESCC European Congress of Clinical Chemistry and Laboratory Medicine, Euromedlab2003, Barcelona, 2003. [19] J. Mallorquí, R. Gutiérrez-Gallego, J. Segura, C. de Bolòs, J.A. Pascual, J. Pharm. Biomed. Anal. 51 (1) (2010) 255. [20] J. Mallorquí, J. Segura, C. de Bolòs, R. Gutiérrez-Gallego, J.A. Pascual, Haematologica 93 (2) (2008) 313.