Sven Pokoj a,1,Iris Lauer a,*,1,Kay Fötisch a ,Martin Himly b ,Adriano Mari c ,Ernesto Enrique d ,Maria del Mar San Miguel-Moncin e ,Jonas Lidholm f ,Stefan Vieths a ,Stephan Scheurer a

a

Division of Allergology,Paul-Ehrlich-Institut,Langen,Germany

b

Division of Allergy and Immunology,Department Molecular Biology,University of Salzburg,Salzburg,Austria c

Center for Clinical and Experimental Allergology IDI-IRCCS,Rome,Italy d

Hospital General de Castellon,Allergy Division,Castellon,Spain e

Department Pius Hospital de Valls,Tarragona,Spain f

Phadia AB,Research and Development,Uppsala,Sweden

a r t i c l e i n f o Article history:

Received 5June 2009

and in revised form 10August 2009Available online 3September 2009Keywords:

Recombinant allergen Hazelnut Cor a 8

Non-specific lipid-transfer proteins nsLTP

Heterologous over-expression

a b s t r a c t

Non-specific lipid-transfer proteins (nsLTP)from food and pollen are clinically important allergens,espe-cially in patients recruited from the Mediterranean area.For the use of recombinant nsLTPs in allergy diag-nosis and preclinical allergy studies the preparation of nsLTPs in a properly folded and biologically active form is required.Using hazelnut nsLTP (Cor a 8)as a model allergen,heterologous over-expression in Esch-erichia coli and Pichia pastoris was compared.Recombinant Cor a 8derived from E.coli and P.pastoris was purified by IMAC and SEC or ammonium sulphate precipitation followed by IEC and SEC,respectively.The recombinant proteins were characterized with regard to IgE-binding by immunoblotting and ELISA,struc-ture by N-terminal sequencing,CD-spectroscopy and LS and to their biological activity using an in vitro basophil histamine release assay.Purification of hazelnut nsLTP from bacterial lysate under native condi-tions resulted in a low yield of Cor a 8.In addition,the preparation contained non-IgE-reactive aggrega-tions besides the IgE-reactive monomer.In contrast,the yield of rCor a 8produced in P.pastoris was approximately 270-fold higher and impurities with oligomers have not been detected.Purified mono-meric Cor a 8from bacteria and yeast showed similar IgE-antibody reactivity and secondary structures,and both were capable of inducing histamine release from basophils.In summary,P.pastoris is superior to E.coli as expression system for the production of large quantities of soluble,properly folded,and bio-logically active rCor a 8.

Ó2009Elsevier Inc.All rights reserved.

Introduction

Allergies are inadequate immune responses against otherwise harmless environmental antigens,such as proteins from pollens,animals or foods,resulting in elevated level of allergen-specific IgE antibodies.In the effector phase of the type I allergic reaction,IgE antibodies,which are bound to high affinity IgE-receptors on the surface of mast cells and basophils,are cross-linked by the aller-gen.This event leads to activation of effector cells and results in the release of preformed and de novo synthesized mediators.The phys-iological activity of these mediators determines the extent of clini-cal symptoms of the allergic reaction.During the past two decades most of the clinical relevant allergens have been identified,and for the majority of them cDNA or at least amino acid sequence informa-tion is available.Purified recombinant allergens can be applied to improve allergy diagnosis and finally probably the specific immu-notherapy (SIT)maybe replacing crude allergen extracts in the long-term.The component-resolved diagnosis (CRD),utilizing sin-gle allergen molecules for serological testing,allows the investiga-tion of the sensitization profile across geographical regions or across selected patient collectives,for instance between patients with mild or aggressive symptoms,adults and children,or patients with persistent allergies versus patients outgrowing their allergies [1].Since IgE reactivity for most allergens is determined by confor-mational epitopes the application of correctly folded molecules is essential to enable an optimal diagnostic value.Recently it was shown by CRD that the geographical distribution of sensitization to allergenic non-specific lipid-transfer proteins (nsLTP)is predom-inant in the Mediterranean area compared to Northern Europe [2].Cor a 8from hazelnut is a member of the nsLTP protein family and has been described as a major food allergen in Spain [3].LTPs are small basic proteins (91–95amino acids,approximately 9kDa)

1046-5928/$-see front matter Ó2009Elsevier Inc.All rights reserved.doi:10.1016/j.pep.2009.08.014

*Corresponding author.Address:Paul-Ehrlich-Institut,Division of Allergology,Paul-Ehrlich Str.51-59,D-63225Langen,Germany.Fax:+49(0)6103771258.

E-mail address:lauir@pei.de (I.Lauer).1

Both authors contributed equally to this study.Protein Expression and Purification 69(2010)

68–75

Contents lists available at ScienceDirect

Protein Expression and Purification

j ou r na l h om ep a ge :w w w.e lse vi e r.c om /lo c at e /yp re

p

Only small amounts of nsLTPs can usually be obtained from nat-ural sources,especially in case of pollen[14]and nuts[15].In order to provide sufficient amounts of high quality material,Cor a8was previously expressed as recombinant protein in Escherichia coli[3]. Initially,E.coli was used for the production of nsLTPs from parie-taria pollen[16],cherry[17]and tomato[18].One disadvantage of using bacterial expression systems is the frequently observed misfolding of target proteins,formation of insoluble aggregates such as inclusion bodies,and/or oligomerization which can lead to a biologically inactive protein.To overcome these problems,co-don optimization and the usage of E.coli origami cells[19],which are capable to form disulfide bonds due to mutations of glutathi-one and thioredoxin reductases is recommended[20].The proteins are frequently expressed as fusion-proteins to alleviate subsequent purification and to enhance solubility.However,the cleavage of the fusion part by specific proteases is often incomplete or unspecific, the latter leading to fragmentation of the protein[21].To improve the expression of Cor a8we have chosen the yeast Pichia pastoris as host organism.The expected advantages are the enhancement of the expression level,the capability to introduce post-transla-tional modifications,such as disulfide bond formation or glycosyl-ation,and,depending on the chosen signal peptides,the secretion of the target protein into the supernatant of the culture[21].Sev-eral allergens(e.g.,Ole e1[22],Ole e6,Ole e10,Bla g4)from var-ious protein families have already been expressed in P.pastoris,as well as some allergenic nsLTPs from ragweed[23],orange[24], strawberry[25],apple[26],peach[26],wheat[27],carrot[28] and hazelnut[29].

In the present study we systematically assessed the quality of recombinant nsLTP derived from P.pastoris and E.coli by biophys-ical and immunochemical methods.Moreover,the expression level of Cor a8in P.pastoris was improved by a BioFloÒ110fermenter system.

Materials and methods

Expression in E.coli and purification of rCor a8

The plasmid construct pET16b-Cor a8(Cor a8,GenBank Acces-sion No.AF329829)[3](restriction sites:50-NdeI and30-XhoI,N-terminal His6-Tag)was transformed in E.coli BL21(DE3)star cells (Invitrogen,Karlsruhe,Germany).Four liter LB-medium containing 50mg/L ampicillin were inoculated with a pre-culture(200mL) and grown for approximately1h at37°C using two benchtop fer-menters(2L each)(MoBiTec GmbH,Göttingen,Germany).The expression was induced with1mM IPTG(isopropyl-b-D-thiogalac-topyranoside)at OD600=0.6and the cell culture was grown for4h at37°C and430rpm.After centrifugation(2200g,30min,4°C), cells were resuspended in20mL lysis buffer(50mM Na2HPO4, 500mM NaCl,2mM imidazole,pH8.0+375U benzonase)and fro-zen atÀ70°C.Cells were subjected to two cycles of thawing and freezing in liquid nitrogen and centrifuged at12,000g for30min at4°C.Hexahistidine(His6)-tagged rCor a8from supernatant was bound to3mL Ni–NTA–Agarose(Qiagen,Hilden,Germany) and step-wise eluted by an increasing imidazole gradient in lysis buffer(starting with25mM,50,100,250,500mM,1M imidazole, 10mL each).The elution of His6-tagged Cor a8(approximately 12kDa)was monitored by SDS–PAGE and CBB staining.Fractions containing oligomeric rCor a8were pooled and concentrated 6-fold using ultrafiltration units(Vivaspin20,MWCO3.500Da, Vivascience,Göttingen,Germany).Thefinal purification step was performed by size-exclusion chromatography(SEC)with a HiLoad 16/60Superdex75prep grade column(Amersham Biosciences, Uppsala,Sweden)in20mM MOPS,500mM NaCl,pH7.4.Elution of rCor a8was monitored by SDS–PAGE and CBB staining.Frac-tions containing monomeric rCor a8were pooled.Protein content was determined using a BCA assay(Pierce,Cheshire,UK). Cloning of Cor a8cDNA in P.pastoris

Cloning of the Cor a8cDNA into yeast expression vector pPICZ a A and transformation of the construct pPICZ a A-Cor a8into P.pastoris X-33cells(Invitrogen GmbH)was previously described [29].For protein expression,positively selected clones were streaked on fresh yeast extract peptone dextrose sorbitol plates containing100l g/mL Zeocin TM,incubated at37°C and subse-quently stored at4°C until inoculation.Positive Zeocin TM resistant clones were identified as Mut S strain resulting in only one single PCR product of700–800bp using50AOX1and30AOX1primers.This PCR product represents the integrated sequence of Cor a8,the a-factor signal sequence and theflanking DNA sequences. Expression in P.pastoris and purification of rCor a8

Expression and purification of P.pastoris-derived Cor a8was performed as described[29]with some modifications.Briefly,a po-sitive clone with the Mut S phenotype was used for the pre-culture. Protein expression was performed using a BioFloÒ110fermenter (New Brunswick Scientific Co.,New Jersey,USA).The fermentation process was regulated to30°C,pH5.0,and30%dissolved oxygen (DO),adjusted by stirring(100–500rpm)and aeration(1–10L/ min).The induction process was prolonged for2days to obtain higher yields.For purification,one out of4L of culture supernatant was concentrated by a two step ammonium sulphate(AS)2precip-itation,in contrast to ultrafiltration in the previous study[29].Most of the proteins from the supernatant precipitated at50%saturation of AS.In the second step rCor a8was precipitated at80%saturation of AS and afterwards reconstituted in desalted water.Further purifi-cation was performed by a two step FPLC method(IEC,SEC).Protein content was determined using a BCA assay(Pierce).Purified rCor a8 from P.pastoris was analyzed on a492Procise sequencer(Applied Biosystems,Foster City,CA,USA)in pulse-liquid mode by Edmann degradation to determine the N-terminal sequence.

2Abbreviations used:AS,ammonium sulphate;BMGY,buffered glycerol-complex medium;CBB,coomassie blue brilliant;CD,circular dichroism;LS,light scattering; HR,histamine release;IEC,ion exchange chromatography;IMAC,immobilized metal affinity chromatography;MW,molecular weight;nsLTP,non-specific lipid-transfer protein;SEC,size-exclusion chromatography;YPD,yeast extract peptone dextrose medium;YPDS,YPD supplemented with1M sorbitol.

S.Pokoj et al./Protein Expression and Purification69(2010)68–7569Competitive ELISA for quantification of rCor a8

Wells of a96-well-plate were coated with100l L rCor a8from E.coli(6ng/mL)and incubated for1h.Blocking of the solid phase was performed using PBS with0.05%Tween20and1%bovine ser-um albumin for30min at room temperature.Binding of a rabbit-anti-Cor a8(50l L,1:10,000,Phadia),raised against Cor a8from E.coli,was inhibited by simultaneous incubation with50l L of the Cor a8-containing samples(3-fold serial dilutions starting from1:500,over night at4°C).Subsequently,the wells were incu-bated with100l L goat-anti-rabbit-IgG conjugated with horserad-ish peroxidase(Sigma–Aldrich GmbH,Taufkirchen,Germany)for 1h at room temperature.For staining100l L of a dilution from TMB(3,30,5,50-Tetramethylbenzidine)in citrate buffer(210mM, pH3.95)was used.After incubation for5to20min at room tem-perature,the reaction was stopped using50l L of25%sulphuric acid.The OD was measured at450nm with an ELISA-Reader (Molecular Devices GmbH,Ismaning,Germany).The Cor a8con-centration was calculated on the Cor a8standard curve(mono-meric rCor a8from E.coli,100ng/mL–0.14ng/mL).

Circular dichroism(CD)spectroscopy

CD spectra of rCor a8from E.coli(oligomeric and monomeric,5 and7.5l M)and P.pastoris(25l M)were recorded in 10mM KH2PO4/K2HPO4buffer(pH7.4)on a Jasco J-810S spectropo-larimeter(Jasco,Gross-Umstadt,Germany)with constant N2flush-ing at20°C.The measurement was performed in a quartz glass cuvette with a step width of1nm and a band width of1nm.The spectral range was180–260nm at50nm/min.Ten scans were accu-mulated and buffer spectra were subtracted.The results are ex-pressed as mean residue molar ellipticity[H]MRW:[H MRW]= 100Âh[mdeg]/N[number of amino acid residues]Âc[mM]Âd [cm].Observed ellipticity h[mdeg];N,number of amino residues; d,optical path length[cm]and c,protein concentration[mM].

Light scattering analysis

For dynamic light scattering(DLS)recombinant Cor a8samples (oligomer from E.coli:750l g/mL in PBS,monomer from yeast: 1mg/mL in20mM MOPS,500mM NaCl)were centrifuged at 14,000g for10min before measuring.The solvent settings for water were used.Data were accumulated for10Â10s and the cor-relation function wasfitted into the combined data curve,from which the intensity distribution was calculated.The intensity dis-tribution was weighted statistically by mass using the mass model for proteins(OmniSize TM software package,Viscotek Corp.)display-ing molecular size R H[nm]and polydispersity(%RSD).

Static light scattering analysis was performed using a 10Â300mm Superdex75GL column(GE Healthcare)on a HP1100analytical chromatography system(Hewlett–Packard, San Jose,CA,USA)at0.4mL/min in20mM MOPS pH7.4,0.5M NaCl(HPSEC–TDA analysis)with the Cor a8oligomer(0.725mg/ mL;100l L),the Cor a8monomer(0.03mg/mL;100l L)from E.coli and the Cor a8monomer from yeast(1mg/mL;100l L). Using a combination of the built-in UV detector measuring at 280nm and a sequential refractive index(RI),intrinsic viscosity (IV),and right-angle light scattering detection(TDA302,Viscotek Corp.,Houston,TX,USA)the MW of eluting peaks can be deter-mined.Detector calibration was performed using bovine serum albumin from Sigma(A7638)weighed out at1.0mg/mL. Patients’sera

For immunological characterization of the recombinant Cor a8 preparations,40sera from patients with a positive case history of immediate-type reactions to hazelnut(e.g.,oral allergy syndrome (OAS),urticaria,angiooedema,or anaphylaxis occurring repeatedly after isolated ingestion of hazelnuts),either with a positive re-sponse to hazelnut extract in skin prick tests(SPT)or specific IgE levels to hazelnut P0.35kU A/L were included.Thirty-six patients were recruited at the Allergy Department of the Institute Universitari Dexeus in Barcelona(Spain),one patient at the Hospital General de Castellon(Spain)and three patients from the Center for Clinical and Experimental Allergology in Rome(Italy). Ethical approval was obtained by the local committees.SPTs were performed according to the recommendations of the European Academy of Allergology and Clinical Immunology(EAACI)[30] and were considered positive when the wheal area was greater than7mm2(diameter P3mm).All patients reported a concomi-tant peach allergy and had specific IgE levels P0.35kU A/L to peach extract.Thirty-five out of36tested were sensitized to the LTP homologue from peach,Pru p3(0.35–49.99kU A/L=Immuno-CAP TM class1–4).

Electrophoresis and immunoblotting

Proteins were separated by SDS–PAGE(17.5%)[31]under non-reducing conditions and gels were stained with CBB.Immunoblot-ting was performed as previously described[29]using a polyclonal rabbit antibody(CE-Immundiagnostika,Eschelbronn,Germany) (1:20,000)raised against rCor a8derived from yeast.Antibody detection was performed using goat-anti-rabbit-IgG-Biotin(Dako Deutschland GmbH,Hamburg,Germany)and streptavidin-AP (CALTAG Laboratories GmbH,Hamburg,Germany).In addition,a mouse-anti-His6-Tag(Qiagen,Hilden,Germany)(1:500)was used for the detection of Cor a8derived from E.coli,followed by a rab-bit-anti-mouse-IgG-AP(Sigma–Aldrich GmbH,1:750).Visualiza-tion of bound antibodies was carried out with nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate(NBT/BCIP) according to the manufacturers’instructions(Biorad,Munich, Germany).

Quantification of specific Immunoglobulin E

Quantification of Cor a8-specific IgE was performed using the ImmunoCAP TM system(Phadia).The ImmunoCAP TM is a commercial in vitro test,afluorescence enzyme immunoassay,which measures the concentration of circulating,allergen-specific IgE antibodies in human serum or plasma in the range of<0.35–100kU A/L(CAP TM classes0–6),thus,providing a highly sensitive and specific mea-surement of the sensitization to the allergen.Experimental Immu-noCAP tests carrying rCor a8preparations from yeast and E.coli were prepared by covalently coupling5l g allergen to the solid phase[32].

Basophil histamine release

The in vitro basophil histamine release assay was performed with re-sensitized basophils as previously described[33]with sev-eral modifications[34].For stimulation of the cells,the three dif-ferent recombinant Cor a8preparations and hazelnut extract [35]were diluted10-fold starting at100l g/mL for hazelnut ex-tract and10l g/mL for purified rCor a8.BSA served as negative control.Released histamine was measured by a commercial com-petitive enzyme immunoassay(Immunotech,Marseille,France). After subtraction of the spontaneous release,the allergen-induced histamine release was calculated as percent of the total amount of histamine determined after lysis of the basophils by two times freezing and thawing.A histamine release of more than10%was considered positive.

70S.Pokoj et al./Protein Expression and Purification69(2010)68–75

Results

Low yield of rCor a 8from E.coli purified under native conditions Escherichia coli BL21(DE3)star cells transformed with the con-struct pET16b-Cor a 8were used for expression of rCor a 8.The tar-get protein was expressed with an additional N-terminal His 6-Tag which facilitated further purification by immobilized metal affinity chromatography (IMAC).Fractions eluted with 25–50mM imidaz-ole showed process-related impurities consisting of E.coli proteins (data not shown),whereas the following fractions (100,250,500mM,1M imidazole)contained mainly the target protein ($12kDa)co-eluted with proteins with apparent molecular weights of 24and 36kDa.These compounds are oligomeric forms of rCor a 8(product-related impurities)which are not detectable under reduced conditions using CBB staining.Moreover,the 24kDa band visible under non-reducing conditions was confirmed as a dimeric form of rCor a 8by its reactivity with a mouse-anti-His 6-Tag (data not shown).Fig.1illustrates the step-wise purification of rCor a 8under native conditions.After IMAC,mono-mers,dimers,and trimers are present in the eluted fractions.Fractions (0.1–1M imidazole)enriched with monomeric rCor a 8were pooled (Fig.1,lane 3)and subjected to size-exclusion chro-matography to remove the oligomeric forms of Cor a 8which re-sulted in purified monomeric rCor a 8(Fig.1,lane 4).BCA assay revealed a concentration of 0.25mg/mL,thus approximately 0.3mg of pure recombinant and monomeric Cor a 8can be ob-tained under native conditions out of 1L E.coli culture.High yield expression and purification of rCor a 8in P.pastoris For expression of rCor a 8,a clone was used resembling the phe-notype Mut S which is characterized by a slow utilization of meth-anol [36].Time course experiments revealed that the highest yield of Cor a 8expression was obtained after an induction period to 5days.Quantification by ELISA revealed 217mg rCor a 8per liter out of 4.2L culture volume.For further purification,1L of the supernatant (Fig.2,lane 1)was used.A purity of >95%of mono-meric rCor a 8was obtained after several purification steps indi-cated in Fig.2.SEC was needed to remove remaining impurities

such as large molecules ($60kDa)and low molecular weight med-ium compounds (<10kDa).Finally,82mg (2.25mg/mL)Cor a 8per liter P.pastoris culture was purified using this procedure which equals a recovery rate of 38%.Verification of identity of rCor a 8

The identification of Cor a 8in the preparations derived from E.coli and P.pastoris was performed by a rabbit-anti-Cor a 8-anti-body (Fig.3A–C,lane 9).All preparations showed comparable IgG antibody reactivity.The correct start of the translation of the rCor a 8derived from E.coli was verified by a mouse anti-His 6-Tag anti-body (data not shown).Furthermore,the N-terminus of Cor a 8de-rived from P.pastoris was identified by N-terminal sequencing:Eleven amino acids (EAEA SLTCPQI)were analyzed including four amino acids that are left after cleavage at the kex2-cleavage-site and were not processed by the diaminopeptidase ste13.

Monomeric Cor a 8preparations from E.coli and P.pastoris showed a -helical secondary structures

The secondary structures of recombinant proteins were ana-lyzed by CD-spectroscopy (Fig.4).The spectrum of rCor a 8derived from P.pastoris showed two minima at 208and 222nm typical for a properly folded a -helical protein.In contrast,the oligomeric preparation from E.coli (Fig.1,lane 3–before SEC)showed a spec-trum typical for a mixture of folded a -helical and unfolded,ran-dom-coiled material (minimum at 200nm)visible by the shift of the passage through the x -axis below 200nm.After a further puri-fication step by size-exclusion chromatography,the monomeric rCor a 8from E.coli revealed a CD spectrum similar to that of the protein derived from P.pastoris .The monomeric,SEC-purified rCor a 8from E.coli did not contain unfolded material.

rCor a 8from P.pastoris consisted of >99.9%monomeric molecules In order to analyze the aggregation behavior of main oligomeric fraction of the E.coli -expressed material in comparison to the yeast material dynamic light scattering was performed.The rCor a 8E.coli preparation before SEC seemed to contain a heterologous mixture of approx.one third monomers and small oligomers hav-ing a mean hydrodynamic radius of 2.7nm corresponding to

glob-

Fig.1.Purification of rCor a 8after expression in E.coli analyzed by SDS–PAGE and CBB staining:Whole E.coli cells after 4h of induction (lane 1),cell lysate prepared under native conditions (lane 2),rCor a 8-enriched fraction after IMAC containing dimers and trimers (lane 3),pure monomeric fraction after SEC (lane

4).

Fig. 2.Purification of rCor a 8from P.pastoris .Culture supernatant (lane 1),enrichment of Cor a 8after AS precipitation (lane 2),dialysis (lane 3),IEC (lane 4),and SEC (lane 5).

S.Pokoj et al./Protein Expression and Purification 69(2010)68–7571

ular proteins with a MW of 20–80kDa and approx.two thirds of aggregates with hydrodynamic radii of 4–30nm (mean radius:10nm).The high M W fraction (<200kDa)showed a polydispersity of >60%indicative of a highly heterogeneous population of mole-cules.In contrast,rCor a 8from yeast consisted of >99.9%mono-meric molecules having a mean hydrodynamic radius of 1.7nm corresponding to globular proteins with a M W of approximately 12kDa.

Using HPSEC–TDA analysis we detected four eluting fractions from the Cor a 8multimeric preparation.The first two fractions corresponded to approximately 37%multimers in the range of 200–500kDa (V ret =10.3–12.4mL)and approximately 30%oligo-mers with a MW of 50–100kDa (V ret =12.4–15.8mL)and the last two fractions corresponded to approximately 9%dimers (V ret =12.4–15.8mL)and to 24%monomer of Cor a 8(V ret=17.5–19.6mL)with a M W of 20and 10kDa,respectively.

The amount of one third monomers/dimers versus two third oligomers/multimers in the Cor a 8preparation was similar com-paring both methods.

Performing HPSEC–TDA analysis of rCor a 8monomer prepara-tion (E.coli )one single peak (V ret =18.8mL)was detected coincid-ing with the monomeric peaks in the rCor a 8multimer run.Accordingly,HPSEC–TDA analysis of rCor a 8expressed in yeast a single peak (V ret =17.9mL)was detected having a MW of approx-imately 9kDa.

Recombinant Cor a 8from E.coli and from yeast are immune reactive For immunological characterization,immunoblotting was per-formed using sera from hazelnut-allergic patients selected by high volumes available.A polyclonal rabbit antibody raised against rCor a 8from yeast served as positive control and a non-allergic patient as negative control.Eight different patient sera were used for detection of the oligomeric (Fig.3A)and monomeric rCor a 8(Fig.3B)derived from E.coli as well as for detection of rCor a 8de-rived from P.pastoris (Fig.3C).The different mobilities in non-reducing SDS–PAGE between rCor a 8from E.coli and P.pastoris are due to the additional His 6-Tag encoded by the bacterial expres-sion plasmid leading to a slightly higher molecular weight of 2.5kDa.Six sera showed IgE-antibody reactivity to monomeric Cor a 8derived from E.coli ,whereas no IgE reactivity to the

oligo-

Fig.3.Immunological characterization of rCor a 8by immunoblotting using 8hazelnut-allergic patients (lane 1–8),rabbit-anti-Cor a 8-antibody as positive control (lane 9),serum from a non-allergic individual (lane 10),and buffer (lane 11)as negative control.E.coli -derived Cor a 8preparation containing oligomers (A)and monomeric Cor a 8only (B),Cor a 8derived from yeast

(C).

Fig.4.CD-spectroscopy analysis.Secondary structure analysis of oligomeric rCor a 8from E.coli indicates a mixture of folded and unfolded material,whereas the monomeric rCor a 8indicates a properly folded protein.The purified rCor a 8derived from P.pastoris revealed a properly folded a -helical protein.

72S.Pokoj et al./Protein Expression and Purification 69(2010)68–75

mers (Fig.1,lane 3)was detectable (Fig.3A).Patient 1and patient 8showed no reaction to the monomeric rCor a 8from E.coli (Fig.3A +B).The rabbit-anti-Cor a 8-antibody used as positive con-trol was reactive with all Cor a 8preparations recognizing exclu-sively the monomeric form.In case of rCor a 8from P.pastoris seven sera showed a strong IgE-antibody reactivity with purified rCor a 8(Fig.3C).In summary,except for one patient (No.1)who was only reactive with Cor a 8from yeast the IgE-antibody reactivity of E.coli-and P.pastoris -derived rCor a 8was similar.Furthermore,the ImmunoCAP TM (Phadia)was applied to deter-mine the amount of Cor a 8-specific IgE reactive with rCor a 8from yeast and E.coli.31out of 36tested sera from hazelnut-allergic pa-tients recruited from the Mediterranean area showed IgE-antibody reactivity to rCor a 8with CAP TM class from 1to 3(specific IgE:0.35to 17.49kU A /L).Thereby,only one patient was exclusively reactive with the E.coli material whereas two patients recognizing only the yeast Cor a 8(Fig.5).Using both allergen preparations for serolog-ical diagnosis,Cor a 8was determined as major allergen with a fre-quency of sensitization of 86%(31/36).In conclusion,for the majority of patients (28/31)comparable specific IgE concentrations to both Cor a 8preparations were observed.

Recombinant Cor a 8from E.coli and from yeast show biological activity in a functional assay

To confirm the allergenic potency of the different rCor a 8prep-arations an in vitro histamine release (HR)assay was performed using serum from patient 5which was recruited from the Mediter-ranean area and was sensitized to Cor a 8as a representative exam-ple.A functionally active allergen must contain at least two IgE epitopes for cross-linking the allergen-specific cell-bound IgE anti-bodies.Bridging of IgE on the surface leads to mast cell or basophil activation which results in mediator release.After stimulation with hazelnut extract and the rCor a 8preparations,histamine was re-leased in a dose dependent matter (Fig.6).Serum of patient 5showed a strong reactivity and reached values of up to 60%release using extract and the different Cor a 8preparations as antigen.Two further sera were used (patients 2and 8)demonstrating biological activity of the Cor a 8preparations (not shown).No clear differ-ences could be observed for the releases using either E.coli multi-meric,monomeric or yeast P.pastoris -derived Cor a 8,respectively.Nevertheless,the histamine release assay confirmed the allergenic potency for all three rCor a 8preparations.

Discussion

The aim of this study was the comparison of hazelnut non-spe-cific lipid-transfer protein Cor a 8over-expressed either in E.coli or in P.pastoris in regard to their quality and availability for molecular studies of hazelnut allergy.Both preparations were analyzed in re-spect of yield,correct folding,IgE reactivity and capacity to activate effector cells of the allergic reaction.

In general,the amount of allergens purified from natural food sources is limited and natural allergen preparations are difficult to standardize in respect to the content of single allergens in a gi-ven source,due to the variability of modifications or the presence of isoforms.Therefore,single allergen compounds based on recom-binant allergens with defined quality resembling their natural counterparts gain increasing importance for molecular studies in the field of allergy.

Heterologous over-expression in E.coli is applicable as long as post-translational modifications are not required and transgenic proteins reveal no toxic effects for the host cells.Moreover,the application of non-glycosylated recombinant allergens in allergy diagnosis circumvents results frequently obtained by IgE reactivity with glycan structures of natural counterparts.Here,Cor a 8as a model allergen and member of the nsLTP family was selected.nsLTPs,belonging to the prolamine superfamily [37],are character-ized by several intramolecular disulfide bonds,mandatory for a correct folding and formation of conformational IgE epitopes.Cor a 8was cloned into pET16b using a N-terminal His 6-Tag which facilitates further purification by immobilized metal affinity chro-matography (IMAC).After IMAC,we obtained an oligomeric rCor a 8preparation with a yield of approx.1mg/L culture correlating with values for Cor a 8described in the literature (1.5mg/L)[3]and with low expression rates for other nsLTPs using a bacterial expression system [16–18].The lack of ability of E.coli to form disulfide bonds is due to the reducing conditions of bacterial cytoplasma.Furthermore,it has been shown by immunoblot inhi-bition experiments under reducing conditions that E.coli-derived rCor a 8after spontaneous oxidation and refolding on the membrane has IgE-binding properties similar to purified natural Cor a 8[3].In this study,we added a further purification step by size-exclusion chromatography to obtain a pure,monomeric and correctly folded protein from E.coli lysate.With this procedure the final yield of rCor a 8derived from E.coli was very low with 0.3mg per liter of

culture.

Fig.5.Comparison of IgE-binding capacity of rCor a 8derived from E.coli and P.pastoris .Using the ImmunoCAP TM (Phadia),a fluorescence based enzyme immuno-assay,sera from 36hazelnut-allergic patients were

compared.

Fig.6.Histamine release from a hazelnut-allergic patient (patient 5).Hazelnut extract was used as positive and BSA as negative controls.Histamine release was performed with oligomeric and monomeric rCor a 8from E.coli ,and rCor a 8derived from P.pastoris .

S.Pokoj et al./Protein Expression and Purification 69(2010)68–7573

Several reports described a low expression of recombinant ani-mal or plant allergens in E.coli e.g.,for Bla g4(0.25mg/L),a major allergen of cockroach[38],Ole e1(0.2mg/L),a major allergen from olive[22],or the cherry LTP(2–3mg/L)[17].Different approaches to overcome the expression and folding problems are codon opti-mization for E.coli or the use of E.coli origami cells which provide oxidative conditions to form the relevant disulfide bonds in nsLTPs. Both strategies did not improve the accumulation of nsLTPs(not shown).Anti-microbial effects are unlikely as poor expression lev-els were obtained by in vitro translation,in comparison to allergens from distinct protein families(not shown).Possible reasons may include a low efficacy of transcription or translation due to the mRNA structure,e.g.,by impartial accessibility of the ribosome binding sites.

Besides a higher expression level and the correct formation of disulfide bonds,the yeast expression system promises a further advantage:low endotoxin content compared to the bacterial expression system,which represents an essential prerequisite for preclinical immunization studies and functional in vitro assays. The preparation of rCor a8under improved conditions(prolonga-tion of induction in P.pastoris,AS precipitation steps)led to a high-er expression rate and a concentration of217mg rCor a8in1L of supernatant which resulted in a theoretical total yield of900mg purified rCor a8expressed in one batch(4.2L).After several puri-fication steps including AS precipitation,cation exchange,and size-exclusion chromatography,the amount of pure rCor a8obtained from1L of supernatant was82mg.Recombinant Cor a8expressed in P.pastoris using a Mut S clone and the BioFloÒ110fermenter sys-tem resulted in a clearly higher expression yield compared to other yeast-derived allergenic nsLTPs from ragweed(10mg/L culture) [23],peach(20mg/L)[26]and hazelnut(30mg/L)[29].However, there is still a potential for further optimization of the expression, which could be achieved by using optimized feeding strategies,or dissolved oxygen content,as259mg of wheat nsLTP have been ob-tained per liter[27].

Verification of protein identity was achieved by specific antibod-ies using anti-His6-Tag,anti-Cor a8and patient sera.Considering that the IgE reactivity depends on the integrity of conformational epitopes,we can conclude that all monomeric rCor a8preparations are intact and correctly folded.Since we found a truncation of the N-terminus using P.pastoris for over-expression for the nsLTP from lettuce[39],we performed N-terminal sequencing of rCor a 8derived from P.pastoris.In addition to the expected amino acid sequence four additional amino acids(EAEA)were observed, representing two ste13cleavage sites located directly behind the kex2cleavage site.Neither the N-terminal His6-Tag present in rCor a8from E.coli nor the four additional N-terminal amino acids in rCor a8from yeast had an impact on the investigated properties. Moreover,similar secondary structures were detectable for both monomeric Cor a8preparations indicating a correct folding.

The comparison of the IgE-antibody reactivity using8patient sera by immunoblotting revealed nearly the same IgE-binding activity for monomeric Cor a8derived from P.pastoris and E.coli.The dimers and trimers of the oligomeric Cor a8fraction from E.coli showed no IgE-antibody reactivity.Only one patient serum showing IgE reactivity to rCor a8from P.pastoris was not reactive to Cor a8expressed in bacteria.The difference in the IgE-binding epitope of this patient may be due to slight differences on epitopes exposed by the two preparations,which may relevant only for a small minority of patients.However,the comparison of purified rCor a8from both expression systems using the ImmunoCAP TM system operating under native conditions showed comparable binding capacity for the majority of patients. This can be explained by the high amount of allergen(approx-imately5l g/ImmunoCAP TM)used for coupling on the solid phase.

Finally,the biological activity was confirmed by in vitro basophil histamine release.This in vitro assay is a functional test reflecting the pathophysiological mechanism causing type I allergic reactions in patients and thus allows the determination of the allergenic po-tency of proteins.All three rCor a8preparations were capable of inducing histamine release.Using this biological assay,we have to consider that only a dramatic difference in the allergenic po-tency(approximately10-fold)is indicated by a clear shift in the re-lease curves.Taken into account the biological variability the histamine assay is not suitable to differentiate between biological active allergen preparations(monomeric Cor a8)and preparations comprising a certain proportion of unfolded allergen(oligomeric Cor a8).Since the IgE reactivity and biological potency of the nsLTP family is based on conformational epitopes[40]which was demon-strated experimentally,a correct3D structure stabilized by four disulfide bonds seems to be present in recombinant hazelnut nsLTP.

In summary,the production of rCor a8was enhanced by change of the expression system from E.coli to P.pastoris.After purification of rCor a8from both systems we observed a270-fold higher yield using P.pastoris.Recombinant Cor a8from P.pastoris showed no oligomerization and a correct formation of secondary structures, but required a more extensive fermentation process and purifica-tion procedure.Immunological characterization showed almost similar allergenic properties of rCor a8from E.coli and P.pastoris. We conclude that P.pastoris is a highly efficient system for produc-tion of large quantities of soluble,properly folded and biologically active rCor a8.

Acknowledgment

This work was in part funded by European Commission FP6 (Contract16610,NANOSPAD).

References

[1]S.Steckelbroeck, B.K.Ballmer-Weber,S.Vieths,Potential,pitfalls,and

prospects of food allergy diagnostics with recombinant allergens or synthetic sequential epitopes,J.Allergy Clin.Immunol.121(2008)1323–1330.

[2]E.A.Pastorello,A.M.Robino,Clinical role of lipid transfer proteins in food

allergy,Mol.Nutr.Food Res.48(2004)356–362.

[3]F.Schocker,D.Luttkopf,S.Scheurer,A.Petersen,A.Cistero-Bahima,E.Enrique,

M.Miguel-Moncin,J.Akkerdaas,R.van Ree,S.Vieths,W.M.Becker, Recombinant lipid transfer protein Cor a8from hazelnut:a new tool for in vitro diagnosis of potentially severe hazelnut allergy,J.Allergy Clin.

Immunol.113(2004)141–147.

[4]J.C.Kader,Lipid-transfer proteins in plants,Annu.Rev.Plant Physiol.Plant Mol.

Biol.47(1996)627–654.

[5]R.Asero,G.Mistrello,D.Roncarolo,S.C.de Vries,M.F.Gautier,C.L.Ciurana,E.

Verbeek,T.Mohammadi,V.Knul-Brettlova,J.H.Akkerdaas,I.Bulder,R.C.

Aalberse,R.van Ree,Lipid transfer protein:a pan-allergen in plant-derived foods that is highly resistant to pepsin digestion,Int.Arch.Allergy Immunol.

124(2001)67–69.

[6]A.I.Sancho,N.M.Rigby,L.Zuidmeer,R.Asero,G.Mistrello,S.Amato, E.

Gonzalez-Mancebo,M.Fernandez-Rivas,R.van Ree,E.N.Mills,The effect of thermal processing on the IgE reactivity of the non-specific lipid transfer protein from apple,Mal d3,Allergy60(2005)1262–1268.

[7]S.Scheurer,I.Lauer,K.Foetisch,M.S.Moncin,M.Retzek,C.Hartz,E.Enrique,J.

Lidholm,A.Cistero-Bahima,S.Vieths,Strong allergenicity of Pru av3,the lipid transfer protein from cherry,is related to high stability against thermal processing and digestion,J.Allergy Clin.Immunol.114(2004)900–907.

[8]E.A.Pastorello,C.Pompei,V.Pravettoni,L.Farioli,A.M.Calamari,J.Scibilia,

A.M.Robino,A.Conti,S.Iametti,D.Fortunato,S.Bonomi,C.Ortolani,Lipid-

transfer protein is the major maize allergen maintaining IgE-binding activity after cooking at100degrees C,as demonstrated in anaphylactic patients and patients with positive double-blind,placebo-controlled food challenge results, J.Allergy Clin.Immunol.112(2003)775–783.

[9]E.A.Pastorello,S.Vieths,V.Pravettoni,L.Farioli,C.Trambaioli,D.Fortunato,D.

Lüttkopf,M.Calamari,R.Ansaloni,J.Scibilia,B.K.Ballmer-Weber,L.K.Poulsen,

B.Wüthrich,K.S.Hansen,A.M.Robino,

C.Ortolani,A.Conti,Identification of

hazelnut major allergens in sensitive patients with positive double-blind, placebo-controlled food challenge results,J.Allergy Clin.Immunol.109(2002) 563–570.

[10]V.Niederberger,R.Valenta,Recombinant allergens for immunotherapy.

Where do we stand?,Curr Opin.Allergy Clin.Immunol.4(2004)549–554.

74S.Pokoj et al./Protein Expression and Purification69(2010)68–75[11]E.Enrique,F.Pineda,T.Malek,J.Bartra,M.Basagana,R.Tella,J.V.Castello,R.

Alonso,J.A.de Mateo,T.Cerda-Trias,M.M.Miguel-Moncin,S.Monzon,M.

Garcia,R.Palacios,A.Cistero-Bahima,Sublingual immunotherapy for hazelnut food allergy:a randomized,double-blind,placebo-controlled study with a standardized hazelnut extract,J.Allergy Clin.Immunol.116(2005)1073–1079.

[12]E.Enrique,T.Malek,F.Pineda,R.Palacios,J.Bartra,R.Tella,M.Basagana,R.

Alonso, A.Cistero-Bahima,Sublingual immunotherapy for hazelnut food allergy:a follow-up study,Ann.Allergy Asthma Immunol.100(2008)283–284.

[13]M.D.Spangfort,J.N.Larsen,Standardization of allergen-specific immuno-

therapy vaccines,Immunol.Allergy Clin.North Am.26(2006)191.1vi. [14]I.Lauer,M.S.Miguel-Moncin,T.Abel,K.Foetisch,C.Hartz,D.Fortunato,A.

Cistero-Bahima,S.Vieths,S.Scheurer,Identification of a plane pollen lipid transfer protein(Pla a3)and its immunological relation to the peach lipid-transfer protein,Pru p3,Clin.Exp.Allergy37(2007)261–269.

[15]N.M.Rigby,J.Marsh,A.I.Sancho,K.Wellner,J.Akkerdaas,R.van Ree,A.Knulst,

M.Fernandez-Rivas,V.Brettlova,P.P.Schilte,C.Summer,R.Pumphrey,P.R.

Shewry,E.N.Mills,The purification and characterisation of allergenic hazelnut seed proteins,Mol.Nutr.Food Res.52(Suppl.2)(2008)S251–S261.

[16]G.Duro,P.Colombo, C.M.Assunta,V.Izzo,R.Porcasi,R.Di Fiore,G.

Locorotondo,R.Cocchiara,D.Geraci,Isolation and characterization of two cDNA clones coding for isoforms of the Parietaria judaica major allergen Par j

1.0101,Int.Arch.Allergy Immunol.112(1997)348–355.

[17]S.Scheurer,E.A.Pastorello,A.Wangorsch,M.Kastner,D.Haustein,S.Vieths,

Recombinant allergens Pru av1and Pru av4and a newly identified lipid transfer protein in the in vitro diagnosis of cherry allergy,J.Allergy Clin.

Immunol.107(2001)724–731.

[18]L.Q.Le,Y.Lorenz,S.Scheurer,K.Fotisch,E.Enrique,J.Bartra,S.Biemelt,S.

Vieths,U.Sonnewald,Design of tomato fruits with reduced allergenicity by dsRNAi-mediated inhibition of ns-LTP(Lyc e3)expression,Plant Biotechnol.J.

4(2006)231–242.

[19]T.Lauber,U.C.Marx,A.Schulz,P.Kreutzmann,P.Rosch,S.Hoffmann,Accurate

disulfide formation in Escherichia coli:overexpression and characterization of thefirst domain(HF78)of the multiple Kazal-type inhibitor LEKTI,Protein Expr.Purif.22(2001)108–112.

[20]A.Yadava,C.F.Ockenhouse,Effect of codon optimization on expression levels

of a functionally folded malaria vaccine candidate in prokaryotic and eukaryotic expression systems,Infect.Immun.71(2003)4961–4969.

[21]M.Schmidt,D.R.Hoffman,Expression systems for production of recombinant

allergens,Int.Arch.Allergy Immunol.128(2002)2–270.

[22]S.Huecas,M.Villalba, E.Gonzalez, A.Martinez-Ruiz,R.Rodriguez,

Production and detailed characterization of biologically active olive pollen allergen Ole e1secreted by the yeast Pichia pastoris,Eur.J.Biochem.261 (1999)539–546.

[23]K.M.Hiller,B.C.Lubahn,D.G.Klapper,Cloning and expression of ragweed

allergen Amb a6,Scand.J.Immunol.48(1998)26–36.

[24]O.Ahrazem,M.D.Ibanez,G.Lopez-Torrejon,R.Sanchez-Monge,J.Sastre,M.

Lombardero, D.Barber,G.Salcedo,Lipid transfer proteins and allergy to oranges,Int.Arch.Allergy Immunol.137(2005)201.

[25]L.Zuidmeer,E.Salentijn,M.F.Rivas,E.G.Mancebo,R.Asero,C.I.Matos,K.T.

Pelgrom,L.J.Gilissen,R.Van Ree,The role of profilin and lipid transfer protein

in strawberry allergy in the Mediterranean area,Clin.Exp.Allergy36(2006) 666–675.

[26]A.Diaz-Perales,G.Garcia-Casado,R.Sanchez-Monge, F.J.Garcia-Selles, D.

Barber,G.Salcedo,CDNA cloning and heterologous expression of the major allergens from peach and apple belonging to the lipid-transfer protein family, Clin.Exp.Allergy32(2002)87–92.

[27]C.Klein,F.Lamotte-Guery,F.Gautier,G.Moulin,H.Boze,P.Joudrier,M.F.

Gautier,High-level secretion of a wheat lipid transfer protein in Pichia pastoris, Protein Expr.Purif.13(1998)73–82.

[28]R.Asero,G.Mistrello,D.Roncarolo,S.C.de Vries,M.F.Gautier,C.L.Ciurana,E.

Verbeek,T.Mohammadi,V.Knul-Brettlova,J.H.Akkerdaas,I.Bulder,R.C.

Aalberse,R.van Ree,Lipid transfer protein:a pan-allergen in plant-derived foods that is highly resistant to pepsin digestion,Int.Arch.Allergy Immunol.

122(2000)20–32.

[29]I.Lauer,S.Alessandri,S.Pokoj,A.Reuter,A.Conti,S.Vieths,S.Scheurer,

Expression and characterization of three important panallergens from hazelnut,Mol.Nutr.Food Res.52(Suppl.2)(2008)S262–S271.

[30]S.Dreborg,A.Frew,Position paper:allergen standardization and skin tests.

The European academy of allergology and clinical immunology,Allergy48 (1993)48–82.

[31]U.K.Laemmli,Cleavage of structural proteins during the assembly of the head

of bacteriophage T4,Nature227(1970)680–685.

[32]D.A.Marknell-DeWitt,V.Niederberger,P.Lehtonen,S.Spitzauer,W.R.Sperr,P.

Valent,R.Valenta,J.Lidholm,Molecular and immunological characterization of a novel timothy grass(Phleum pratense)pollen allergen,Phl p11,Clin.Exp.

Allergy32(2002)1329–1340.

[33]B.Kleine-Budde,I,P.G.de Heer,J.S.van der Zee,R.C.Aalberse,The stripped

basophil histamine release bioassay as a tool for the detection of allergen-specific IgE in serum,Int.Arch.Allergy Immunol.126(2001)277–285. [34]K.Foetisch,S.Westphal,I.Lauer,M.Retzek, F.Altmann, D.Kolarich,S.

Scheurer,S.Vieths,Biological activity of IgE specific for cross-reactive carbohydrate determinants,J.Allergy Clin.Immunol.111(2003)8–6. [35]I.Lauer,K.Foetisch,D.Kolarich,B.K.Ballmer-Weber,A.Conti,F.Altmann,S.

Vieths,S.Scheurer,Hazelnut(Corylus avellana)vicilin Cor a11:molecular characterization of a glycoprotein and its allergenic activity,Biochem.J.383 (2004)327–334.

[36]J.M.Cregg,K.R.Madden,K.J.Barringer,G.P.Thill,C.A.Stillman,Functional

characterization of the two alcohol oxidase genes from the yeast Pichia pastoris,Mol.Cell Biol.9(19)1316–1323.

[37]H.Breiteneder,C.Radauer,A classification of plant food allergens,J.Allergy

Clin.Immunol.113(2004)821–830.

[38]L.D.Vailes,M.T.Kinter,L.K.Arruda,M.D.Chapman,High-level expression of

cockroach allergen,Bla g4,in Pichia pastoris,J.Allergy Clin.Immunol.101 (1998)274–280.

[39]C.Hartz,M.M.Miguel-Moncin,A.Cistero-Bahima,K.Fotisch,K.J.Metzner,D.

Fortunato,J.Lidholm,S.Vieths,S.Scheurer,Molecular characterisation of Lac s 1,the major allergen from lettuce(Lactuca sativa),Mol.Immunol.44(2007) 2820–2830.

[40]N.Pasquato,R.Berni,C.Folli,S.Folloni,M.Cianci,S.Pantano,J.R.Helliwell,G.

Zanotti,Crystal structure of peach Pru p3,the prototypic member of the family of plant non-specific lipid transfer protein pan-allergens,J.Mol.Biol.

356(2006)684–694.

S.Pokoj et al./Protein Expression and Purification69(2010)68–7575下载本文