A mouse model of lupin allergy

ORIGINAL ARTICLE

A mouse model of lupin allergyN. E. Vinje�,w, S. Larsenz and M. Løvik�,‰�Division of Environmental Medicine, Norwegian Institute of Public Health, Oslo, Norway, wNational Veterinary Institute, Oslo, Norway, zDepartment of Production

Animal Clinical Services, Norwegian School of Veterinary Science, Oslo, Norway and ‰Department of Cancer Research and Molecular Medicine, Faculty of Medicine,

Norwegian University of Science and Technology, Trondheim, Norway

Clinical &Experimental

Allergy

Correspondence:Nina E. Vinje, Division of EnvironmentalMedicine, Norwegian Institute of PublicHealth, PO Box 4404 Nydalen, NO-0403Oslo, Norway.E-mail: [email protected] this as: N. E. Vinje, S. Larsen andM. Løvik, Clinical & ExperimentalAllergy, 2009 (39) 1255–1266.

Summary

Background Lupin has been introduced as a new food ingredient in an increasing number ofEuropean countries, resulting in reports of allergic reactions mostly due to cross-reactions inpeanut-allergic individuals. Some cases of primary lupin allergy have also been reported.Objective The aim of our study was to develop a food allergy model of lupin in mice withanaphylaxis as the endpoint and further, to develop an approach to estimate the allergen doseinducing maximal sensitization using a statistical design requiring a limited number ofanimals.Methods Mice were immunized by intragastric gavage using cholera toxin as an adjuvant. Atwo-compartment response surface design with IgE as the main variable was used to estimatethe maximal sensitizing dose of lupin in the model. This estimated dose was further used toevaluate the model. The mice were challenged with a high dose of lupin and signs of ananaphylactic reaction were observed. Antibody reactions (IgE and IgG2a), serum mast cellprotease [mouse mast cell protease-1 (MMCP-1)] and ex vivo production of cytokines (IL-4,IL-5 and IFN-g) by spleen cells were measured. An immunoblot with regard to IgE bindingwas also performed.Results The dose that elicited the maximal sensitization measured as IgE was 5.7 mg lupinprotein per immunization. Mice that received this dose developed anaphylactic reactionsupon challenge, IgE against several proteins in the lupin extract, and high levels of MMCP-1,and showed a general shift towards a T-helper type 2 response. Post-challenge serum MMCP-1levels corresponded to the seriousness of the anaphylactic reactions.Conclusion We have established a mouse model with clinical symptoms of lupin allergy, withan optimized dose of lupin protein. A statistical design that can be used to determine anoptimal immunization dose with the use of a minimum of laboratory animals is described.

Keywords allergy, anaphylaxis, legumes, lupin, mouse modelSubmitted 5 May 2008; revised 8 December 2008; accepted 10 March 2009

Introduction

Food allergy is a serious and apparently growing problemin the western world. The prevalence varies considerablybetween different studies [1], but overall a prevalence ofabout 1–3% in the general population and 3–8% amongchildren is the consensus [2, 3].

The most important allergens are peanuts, tree nuts,soy, milk, fish, shellfish, wheat and egg. However, with therapid introduction of novel foods and new food ingredi-ents in traditional foods, the number of allergenic foods isalso rising. Assessment of whether a novel food is aller-

genic is difficult in the absence of validated animalmodels. It will often take some time from when a novelfood is introduced until allergic episodes occur, becausethe food must be used for a period of time for an allergy todevelop. In contrast, when reactions to food are caused bycross-reactivity with food ingredients already in commonuse, reactions may occur as soon as the ‘new’ food isintroduced. In both cases, there will be delays in reportingbecause of lack of awareness that the novel food is aproblem.

Lupin is a legume, like peanut and soy, which arecommon allergens. Some of the species from the Lupinus

Experimental Models of Allergic Disease

doi: 10.1111/j.1365-2222.2009.03269.x Clinical & Experimental Allergy, 39, 1255–1266

�c 2009 The AuthorsJournal compilation �c 2009 Blackwell Publishing Ltd

mailto:[email protected]





family known as sweet lupines (Lupinus albus, Lupinusluteus and Lupinus angustifolius) have been consumedtraditionally in the southern part of Europe as a supple-ment in cooking or as a snack. When ground into flour,lupin is a protein-rich supplement with good bakingqualities [4]. Lupin consumption has been increasing inseveral European countries, and allergic reactions to lupinhave occurred following its introduction into processedfoods in the late 1990s [5–10]. The major part of thereported cases seems to be caused by an allergic cross-reaction between peanut and lupin. The prevalence ofprimary allergy to lupin is unknown, but appears tobe low [4]. There are extensive in vitro cross-reactivitiesbetween members of the legume family, but this cross-reactivity has only been estimated to be of clinicalrelevance in about 5% of legume-allergic individuals[11, 12]. Concerning lupin, on the other hand, theclinically relevant cross-reactivity with peanut has beenreported in two studies to be 30% [7] and 68% [7, 13].In a new study of six lupin-allergic patients, threeshowed evidence of clinical or laboratory cross-reactivitywith other legumes [14]. However, in a UK studyonly two of 47 peanut-allergic children had an allergicreaction to lupin upon challenge [15]. Likewise, aNorwegian study concluded that children with sensitiza-tion to lupin are not very likely to have clinical lupinallergy [16].

To study the sensitization process and the allergicreaction to foods in detail, animal models are needed.There are a number of such models in different animalspecies [17–19], but models in mice and rats are mostwidely used. So far, it seems that every type of foodrequires its own model with regard to immunizationprotocols, although the models can be based on the sameprinciples.

The main aim of the present paper was to establish amouse model of primary lupin allergy. The model is basedon the peanut allergy model published by Li et al. [20]that uses an intragastric immunization protocol withcholera toxin as an adjuvant. This cholera toxin modelof food allergy has also been used successfully in modelsof cows milk allergy [21] and buckwheat allergy [22].Different allergens will need different doses to elicit amaximal IgE response. Our first goal was to establish thedose of lupin that gives the highest IgE response inthe mice and to develop an approach for a generalapplication that could minimize the number of animalsneeded for establishing a dose–response relationship. Theestablished dose of lupin was then used in a new experi-ment to further characterize the model of primary lupinallergy and to compare it with the human situation. Amain focus in this process was the anaphylactic reaction,determined both clinically and biochemically by measur-ing the serum mast cell protease [mouse mast cell pro-tease-1 (MMCP-1)].

Materials and methods

Animals

Twenty-nine female inbred C3H/HeJOlaHsd-Lpsd mice(Harlan UK Ltd, Oxon, UK) were used in the first twoexperiments and, due to an availability problem, 30 femaleinbred C3H/HeJ mice (Jackson Laboratories, Bar Harbor,Maine, USA) were used in the last experiment. The micewere about 5 weeks old at the start of the immunizationprocess. A total of 21 female Sprague–Dawley rats,150–200 g (Taconic M&B A/S, Ry, Denmark), were used toperform the passive cutaneous anaphylaxis tests (PCA-test).

The animals were housed, four to eight mice or two tothree rats per cage, on NESTPAK bedding (Datesand Ltd,Manchester, UK) in type III macrolon cages in filter cabinets(Scantainers, Scanbur AS, Nittedal, Norway). They wereexposed to a 12-h/12-h light/dark cycle (30–60 lx in cages),room temperature of 21�2 1C and 35–75% humidity. Theywere given pelleted food (RM1, SDS, Essex, UK) and tapwater ad libitum. Before entering the experiments, theanimals were allowed to rest for at least 1 week.

The experiments were performed in conformity with thelaws and regulations for experiments with live animals inNorway, and were approved by the Norwegian AnimalResearch Authority under the Ministry of Agriculture.

Chemicals and reagents

The National Veterinary Institute of Norway provided theprotein extract of lupin. Briefly, seeds of L. angustifoliuswere mechanically homogenized, and then 2 g of thesample was extracted in 10 mL 0.1 M Tris/0.5 M glycinebuffer at pH 8.7 overnight on a shaking water-bath at45 1C. After the extraction was completed, the sample wasspun at 39 200 g at 4 1C for 25 min. The total proteinconcentration of the extract was measured by Lowry’smethod [23]. The absorbance was read by a 1420 VICTOR(2) multilabel plate counter (Wallac, Turcu, Finland) with a690 nm filter, and the total protein content was deter-mined relative to the BSA standard.

Cholera toxin from Vibrio cholerae, azide free (CAS no.9012-63-9; EMD Biosciences Inc., San Diego, CA, USA),was purchased from Quadratech Diagnostics Ltd (Epsom,Surrey, UK). Ovalbumin (OVA grade VII) and Evans Blue(CAS no. 314-13-6) were provided by Sigma Aldrich (StLouis, MO, USA). Amersham Pharmacia Biotech, Uppsala,Sweden provided the chemicals and equipment for SDS-PAGE and immunoblot, unless stated otherwise.

Immunization protocol

The lupin extract was diluted in Hank’s balanced saltsolution (HBSS; PAA Laboratories GmbH, Linz, Austria).The immunization followed the protocol published by

�c 2009 The AuthorsJournal compilation �c 2009 Blackwell Publishing Ltd, Clinical & Experimental Allergy, 39 : 1255–1266

1256 N. E. Vinje et al

Li et al. [20] with some modifications. In the surfaceresponse part of the study, immunizations were performedon days 0, 1, 2, 7 and 21. The mice were exsanguinated onday 28 (Fig. 1). In the final experiment, we allocated themice into five cages (A–E) of six mice each. The mice inthree of the cages were treated with 5.7 mg lupin extractplus 10 mg cholera toxin per immunization; in cage fourthe mice were treated with 10 mg cholera toxin only (sham)and the mice in the fifth cage were left untreated (naıve).On day 27, two mice from each cage were exsanguinated(Fig. 1). The remaining mice received an additionalimmunization on day 28. One mouse from each cage (atotal of three immunized mice and two controls) waschallenged on day 35 with an intraperitoneal (i.p.) injec-tion of 5.7 mg lupin (anaphylaxis control), while the othermice were challenged on the same day by gavage with atotal dose of 24.2 mg lupin divided into two gavages given

30 min apart. Exsanguination was then performed 30 minafter the last gavage, or when a serious anaphylacticreaction was observed.

Assessment of clinical anaphylactic reactions

Anaphylactic symptoms were evaluated for 30 min afterthe second peroral challenge on day 35. The mice thatreceived an i.p. challenge were exsanguinated 10–15 minafter the challenge because of ethical considerations dueto the seriousness of the reactions. We used the scoringsystem described by Li et al. [20]: 0, no symptoms; 1,scratching and rubbing around the nose and head; 2,puffiness around the eyes and mouth, diarrhoea, pilarerecti, reduced activity and/or decreased activity with anincreased respiratory rate; 3, wheezing, laboured respira-tion, cyanosis around the mouth and tail; 4, no activityafter prodding or tremor and convulsion; and 5, death. Ingeneral, we assigned a score of 1 to a mild reaction, 2–3 amoderate reaction and 4–5 a serious reaction.

Passive cutaneous anaphylaxis

The amount of lupin-specific IgE antibodies was deter-mined by the heterologous mouse rat PCA-test [24–27].The technique is illustrated in Fig. 2. Briefly, 100 mL serumfrom immunized mice, 0.9% NaCl (negative control) andpositive control serum were administered intradermallyon each side of the shaved back of anaesthetized rats. Foranaesthesia, we used isofluran gas. The positive serum wasfrom a pooled serum sample previously shown to have ahigh OVA-specific IgE level by an ELISA protocol [28].

0 7 14 21 28 35Time(days)

Surfaceresponse

experiments

Finalexperiment

GGG G G

GGG GBG B BGE

C/E

E

Fig. 1. Experimental design. Time line for the experiments and thetreatments given: G, intragastric gavage; B, blood sampling; E, exsan-guination and C, allergen challenge.

(a) (b)

(c) (d)

Fig. 2. The passive cutaneous anaphylaxis test (PCA-test). Photographs illustrating the procedure of the PCA-test in rat. (a) Intradermal injection ofmouse serum in the skin. (b) i.v. injection of allergen and Evans Blue solution. (c) The outside of the skin at the time of measurement provides a crudeindication of the size of reactions and is a guide to where the different sera were injected. (d) Measurements of the size of weal in the inverted skin.


Mouse model of lupin allergy 1257

Twenty-four hours after deposition of serum samples,1 mL of a 0.9% saline solution containing 100 mg lupinextract, 100 mg of OVA and 4.5 mg Evans Blue wasadministered intravenously in a tail vein. One hour later,the rats were killed, and the results were read as size indiameter of the dot in the skin, or endpoint dilution (titre).

To monitor how the levels of lupin-specific IgE anti-bodies developed during the experiment, we analysed apooled serum from each cage (A–E) at days 6, 13, 27 and35. We also analysed individual samples drawn on the dayof exanguination (either day 27 or 35).

Splenocyte preparation

The spleens were excised and brought to the laboratory inHBSS. Spleen cells were isolated by crushing the spleensthrough a steel wire mesh. The cell suspension was thenaspirated several times in a 5 cc syringe with a 21 G needleto break up clumps and transferred to glass tubes contain-ing HBSS. After spinning the suspension at 420 g for10 min, the cells were resuspended in 1 mL completeculture medium (RPMI 1640 with L-glutamine, supple-mented with 10% fetal bovine serum and 1% streptomy-cin/penicillin). The cell concentration was measured usinga Coulter Counter Z1 (Beckman Coulter Inc., Miami, FL,USA). 5�106 (1 mL) cells were added per well on 24-wellcell culture plates (Costar 3524, Corning Inc., Corning, NY,USA). Cells from each spleen were incubated either with orwithout 50 mg lupin extract or with 3 mg ConA (positivecontrol), and cultured at 37 1C and 5% CO2 for 48 and 72 h.The plates were centrifuged at 310 g for 5 min, and thesupernatants were collected and stored at �80 1C untilcytokine ELISA analyses.

Ex vivo cytokine production and quantification byenzyme-linked immunosorbent assay

The levels of IL-4, IL-5 and IFN-g cytokines in the cellsupernatants were determined by sandwich ELISA accord-ing to the protocols provided by the manufacturer. MouseDuoSets (R&D Systems Inc., Minneapolis, MN, USA) wereused for detection. The detection limit was 15.625 pg/mLin the IL-4 assay, and 31.25 pg/mL in the IL-5 and IFN-gassays.

Measurement of immunoglobulin G2a

Polystyrene microtitre plates (9018, Corning Costar EIA,Sigma Aldrich) were coated with 100 mL of 5 mg/mL lupinextract in 0.05 M carbonate/bicarbonate buffer, pH 9.6.The coated plates were incubated for 1 h at room tempera-ture and thereafter at 14 1C overnight. After washing with0.1 M Tris-HCl buffer pH 7.4 containing 0.05% Tween 20(Tris/Tw), the plates were incubated with 100 mL blockingbuffer [Tris/Tw with 1% bovine serum albumin (BSA)] for

1 h. After another washing with Tris/Tw the plates wereincubated with a lupin-specific standard IgG2a serum,negative control serum and the sera to be tested diluted inBSA/Tris/Tw, 100 mL per well. To prepare the IgG2astandard serum we immunized BALB/c mice i.p. withlupin extract together with CpG–ODN 1826 (sequence:50-TCCATGACGTTCCTGACGTT-30), a known T-helpertype 1 (Th1)-adjuvant. The standard was diluted by afactor of 2 in the range from 1 : 800 to 1 : 51 200. A pooledserum from normal mice was used as a negative control,yielding levels below the detection limit. Duplicates of a1 : 10 dilution of the sera to be tested were used, while thestandard serum was applied in triplicate. Because of smallamounts of serum obtained from a few of the mice, somesamples were tested in a 1 : 20 dilution. The plates wereincubated for 2 h at room temperature, followed byincubation at 14 1C overnight. After a new wash cycle,the plates were incubated with 100 mL biotin-labelled ratanti-mouse IgG2a antibody (clone R19-15, BD Bios-ciences, Franklin Lakes, NJ, USA) diluted 1 : 500 in Tris/Tw for 2 h. Again the plates were washed and thenincubated for 1 h with 1 : 50 dilutions in 0.05 M Tris/HClof complexes of avidin/streptavidin and biotinylatedalkaline phosphate (StrepABComplex AP, DAKO, Glostr-up, Denmark). After a final wash cycle, the plates wereincubated with 100 mL p-nitrophenyl phosphate (Sigma104 Phosphate Substrate, Sigma Aldrich) in 10% dietano-lamide buffer pH 9.8. The plates were read at 405 nm in anMRX microtitre plate reader (Dynatech Laboratories,Chantilly, VA, USA) after 12 min of incubation. Antibodyconcentrations are given in kilo arbitrary units (kAU)/mL.Kilo arbitrary units per millilitre was determined from thedilution of the standard that gave an OD value of 0.8. Thiswas the 1 : 25 600 dilution and the value was set to25.6 kAU/mL. The other values were then calculatedfrom this value so that 1 : 51 200=12.8 kAU/mL and1 : 12 800 = 51.2 kAU/mL. The non-specific binding waslow, as demonstrated by levels of IgG2a below the detec-tion limit in the negative control. The specificity of theIgG2a antibody has previously been tested in our lab, andfound to be satisfactory. The detection limit of the assaywas 12.8 kAU/mL.

Measurement of serum mouse mast cell protease-1

Serum levels of MMCP-1 were determined using an ELISAkit (Moredun Scientific Ltd, Scotland, UK). The ELISA wasperformed according to the manufacturer’s instructions.The detection limit of the MMCP-1 assay was 0.25 ng/mL.

Sodium dodecyl sulphate polyacrylamide gelelectrophoresis and Western blot

SDS-PAGE was performed with ExcelGel SDS gradient8–18 Precast gel (Amersham Pharmacia Biotech). A fish



extract was used as a negative control. It was made fromcod fillet using an extraction method similar to that forthe lupin extract. The extract was provided by TheNational Veterinary Institute of Norway. Samples of lupinand fish extracts were diluted in a reducing buffercontaining b-mercaptoethanol to a concentration of2 mg/mL, and boiled for 3 min. Ten microlitres of sampleor Rainbow marker (RPM800 and RPM2107, AmershamPharmacia Biotech) was loaded on the application pieces.Electrophoresis was performed at 50 mA for about 1 h. Thegels were either silver stained or electroblotted for sub-sequent determination of IgE binding. The proteins wereblotted onto nitrocellulose paper (pore size 0.22 mm, GEWater & Process Technologies, Trevose, Pennsylvania,USA) at 200 mA for 1 h. The nitrocellulose was blockedwith tris buffered saline (TBS) with 5% skimmed milk for1 h at room temperature before incubating overnight at4 1C with sera from mice immunized with lupin, shamimmunized, naıve or OVA-immunized mice. All sera werediluted 1 : 100. After washing, the membranes were in-cubated with a biotin-labelled conjugate (anti-mouse IgE)diluted 1 : 50 for 1 h at room temperature. Then StreptAB-Complex/HRP (DAKO A/S) was added for a new incuba-tion of 1 h. Western Lightning ChemiluminiscenceReagent Plus (PerkinElmer Life Sciences, Boston, MA,USA) was used as a substrate for detection. The blot wasdeveloped and analysed using a Kodak Image Station2000R (Eastman Kodak Company, Rochester, NY, USA).

Experimental design

The first two experiments were performed as a two-compartment response surface design with 15 and 14mice in the respective compartments. Lupin-specific IgEwas used as the measured outcome. All sera were tested at1 : 4 dilutions and a PCA reaction of o3 mm was con-sidered to be negative. A PCA reaction of 15 mm was usedas a cut-off line in the calculations shown in Fig. 3a toensure a high IgE response. In the first experiment thedoses were equally distributed from 0.1 up to 10.0 mg.Serum from each mouse was tested on two different rats,

and the mean of the two results was used in furtheranalyses. The results from the first experiment were usedto determine a range of doses that would give a reaction of

(b)

0

8 10 126420

5

10

15

20

25

30

35

Dosage (mg/mouse)

Siz

e of

rea

ctio

n (m

m)

yB=15.01+0.96x–0.31(x–4.19)2

yB

max =

5.74(a)

0

0

5

42

15

10

20

25

30

35

6 8 10 12Dosage (mg/mouse)

Siz

e of

rea

ctio

n (m

m)

yA=19.49+0.32x–0.31(x–4.2)2

yA

max =

4.73

x2 =

9.10

x1 =

0.36

(c)

2

25 30 3520151050

0

4

6

8

10

12

Dos

age

(mg

/mou

se)

Size of reaction (mm)

yC=5.58–0.008x–0.018(x–15.67)2

7.0

3.9 yC

max =

15.2

5.74

Fig. 3. Two-compartment surface response design. Each square repre-sents the passive cutaneous anaphylaxis (PCA) result of one mouse. Thepolynomial line (yA, yB and yC) in each graph is determined by all thepoints of the plot. The line y = 15 in plot (a) is the cut-off line. Plot (a)shows the results from the first compartment (n = 15). The doses ofcompartment number two were concentrated around the intersectionsbetween yA and the cut-off line (x1 and x2), and the yAmax. Plot (b) showsthe results from both compartments of the design (n = 29). The dosagethat gives the largest PCA-reaction is given by yBmax = 5.74 mg/mouse.Plot C shows the approximation of the 95% confidence interval of theoptimal dose, found by exchanging the y-axis with the x-axis from plot(b) and creating a new polynomial line with its confidence interval (thinlines). The approximated confidence interval for the dose 5.74 mg/mouseis h3.9–7.0i.



15 mm or above in the PCA. The results were plotted and acurve was drawn through the points on the graph, asshown in Fig. 3a. Analysing the first compartment gave usthree focus-points: the top of the curve and the twointersections between the curve and the cut-off line. Inthe second experiment we used five doses of lupinclustered around the lower intersection, five around thetop point and four around the upper intersection. Theoptimal dose of lupin extract was found by combining theresults from both compartments in one graph, as shown inFig. 3b. The top of the curve indicates the dose of lupinextract that will statistically result in the largest PCAreaction. Figure 3c shows the approximation of the 95%confidence interval of the optimal dose, found by ex-changing the y-axis with the x-axis from plot B andcreating a new polynomial line with its confidence inter-val (thin lines).

A final experiment was set up to further characterizethe model using the established optimal dose of lupin. Inthis experiment we measured the lupin-specific IgE titre,and determined the Th1/Th2 balance, using the cytokinesIL-4, IL-5 and IFN-g, the Th2-dependent IgE response andthe Th1-dependent IgG2a antibody response. We used themast cell protease (MMCP-1) as an indicator of the localanaphylactic response in the gut, and the general anaphy-lactic response in the mice after challenge with a highdose of lupin was determined by the described gradedsystem of clinical signs.

Statistical analysis

All results are expressed as mean values with 95%confidence intervals constructed using the Student proce-dure [29]. As an index of dispersion, the standard devia-tion is given. All tests were performed two tailed anddifferences were considered significant if the p-valueswere found to be � 0.05. One-way ANOVA was used tocompare groups. The analyses were performed on log-transformed data. The clinical anaphylactic reaction wasanalysed by contingences table analysis [30].

In order to estimate the dose of lupin eliciting maximalsensitization in the model, non-linear regression analysiswith dosage (mg/mouse) and size of reaction (millimetre)was performed. The data from the first part of the surfacedesign experiment with the size of reaction as the depen-dent variable were used to estimate the doses of lupin thatresulted in allergic reactions. Based on these results thesecond part of the experiment was performed, and thetotal set of data were analysed by the same type ofstatistical model, but now with the dosage of lupin as thedependent variable.

The repeatability of the PCA method was evaluatedusing agreement analysis of the two parallel samples fromeach animal [31, 32]. A repeatable method is a methodthat yields similar results on parallel samples, and the

agreement index estimates the degree of similarity [33].An agreement analysis is a stepwise procedure. First themean difference between the parallel samples (A and B)was calculated and tested against zero using a pairedStudent’s t-test [29]. Thereafter, the regression line for Aand B was drawn and tested to detect significant deviationfrom the line of equality (A = B). The agreement index wasthen determined. A positive index supports agreement,and a value 40.5 indicates good agreement. An agree-ment plot of the differences against the mean of A and Bwas used to spot outliers, defined as differences lyingoutside of the agreement limits (�2 SD). Finally, thecorrelation coefficient between the mean of the twoparallel samples and the absolute value of the differencewas calculated and tested to reveal whether the differencechanges with increasing measurement values [34].

Results

The lupin allergy model

Based on the results obtained from 15 mice in the first partof the experiment, the dose range of lupin resulting in aPCA reaction Z15 mm was found to be between 0.4 and9.1 mg/mouse (Fig. 3a). The dose of lupin extract resultingin the largest allergic reaction was estimated to be 4.7 mg/mouse. By adding the results obtained from the 14 mice inthe second part of the experiment, this estimated doseincreased to 5.7 mg/mouse (Fig. 3b) with a 95% confi-dence interval from 3.9 to 7.0 mg/mouse (Fig. 3c).

The agreement analysis performed to validate the PCAreaction as a method of IgE measurement detected nosignificant difference between the two tests of each serum(P = 0.07) and the linear regression of the first registrationexpressed by the second did not differ significantly fromthe line of equality. The agreement index was estimated tobe 0.61 and thereby classified as good. This is underlinedwith only 3.4% outliers related to the agreement limits.However, the correlation between the mean measurementlevel and the absolute difference between the first and thesecond registration was found to be positive and signifi-cant. This means that the difference between two measure-ments increases with increasing size of reaction (Table 1).

Table 1. Agreement analysis

Parameters PCA reaction (mm)

Mean level (SD) (first and second registration) 16.4 (8.1)Mean difference (SD) (first–second registration) �1.1 (3.2)Agreement limits �6.3Agreement index 0.61% outliers 3.4% (1/29)Correlation between ‘level’ and absolute difference 0.45

PCA, passive cutaneous anaphylaxis; SD, standard deviation.



Anaphylactic reactions

All the immunized mice challenged by intragastric gavageshowed signs of mild to moderate anaphylactic reactions.They all showed symptoms like scratching and rubbing of thenose and head, and some of them were clearly puffy aroundthe mouth and eyes. The mice also displayed signs of generaldiscomfort, like sitting and moving around with arched backsas well as not settling down to rest. The immunized mice thatwere provoked by an i.p. injection of lupin developed seriousto near-fatal reactions, and hardly responded with anyactivity after prodding. Because of ethical considerationsthese mice were exsanguinated 10–15min after challengeand it is not possible to know whether they would have livedafter 30min. None of the sham-treated or naıve mice showedany signs of anaphylaxis.

The local anaphylactic reactions in the gut, as measured byMMCP-1, paralleled the clinical signs of anaphylaxis that weobserved (Fig. 4). The groups of immunized mice that werechallenged by gavage [per os (p.o.)] or by an i.p. injection hadsignificantly higher serum MMCP-1 than the immunizedmice that did not undergo any challenge and the mice thatwere not immunized with lupin (control group). The group ofimmunized, but not challenged, mice was found to besignificantly higher compared with the control group(Po0.0001) and significantly lower than the p.o.-provokedgroup (P= 0.0039). The mean MMCP-1 level of the i.p.-provoked group (14602.7ng/mL) was again 95 times higherthan the p.o.-provoked group (173.2ng/mL).

Antibody and cytokine responses

In the pooled serum samples, there was no detectablelupin-specific IgE on days 6 and 13 of the experiment.

However, at day 27 a specific IgE response in the miceimmunized with lupin extract was detected. This responseappeared somewhat reduced by day 35, but this reductionwas not statistically significant (Fig. 5). In samples fromindividual mice that were immunized with lupin, specificIgE was detected in most but not all sera, and especiallyafter the challenge on day 35 we were not able to detectspecific IgE in several sera. In mice exsanguinated onday 27, we could not detect IgE in two of six sera, whilethe remaining four showed PCA reactions from 12 to24 mm [group mean 11.4 mm (SD 9.7)]. In mice exangui-nated on day 35, we could not detect IgE in five of ninesera, while the remaining sera showed PCA reactions thatvaried from 3.5 to 18.5 mm [group mean 5.7 mm (SD 7.9)].There was no detectable lupin-specific IgE in mice of thesham group or the naıve group at any point in theexperiment.

The spleen cell culture supernatant concentration of IL-4, IL-5 and IFN-g was determined after 72 h ex vivostimulation with lupin extract (Fig. 6). Both IL-4 and -5were found to be significantly higher for the immunized(but not challenged) group compared with all the othergroups (Po0.05). No significant differences were detectedbetween the control group and the challenged groups forIL-4. For IL-5, the response in the group challenged bygavage was found to be significantly higher comparedwith both the control groups and the i.p. group (Po0.04).Additionally, the level of IL-5 in this group was threetimes higher than the controls. A similar pattern wasfound for IFN-g. The immunized (but not challenged)group showed significantly higher levels compared withall the other three groups (Po0.001), and the groupchallenged by gavage was significantly higher than the

0.1

1

10

100

1000

*

**

Control Immunized(not challenged)

Challenged p.o.

MM

CP

-1 (

ng/m

L)

Fig. 4. Mouse mast cell protease 1 (MMCP-1). Levels of MMCP-1 inserum were measured by ELISA. The mean values (ng/mL) with 95%confidence intervals for the different groups are shown. *Statisticallysignificant difference from control (Po0.0001); **Statistically signifi-cant difference from control and immunized groups (Po0.004).

18

14

12

10

8

4

2

0

Day 6 Day 13 Day 27 Day 35

PC

A-r

eact

ion

(mm

)

16

6

Fig. 5. IgE development. Levels of lupin-specific IgE in serum weremeasured using the heterologous passive cutaneous anaphylaxis (PCA)reaction with serum dilution 1 : 4. IgE was measured in a pooled serumsample from all the mice in each cage that were immunized with lupin(three cages of eight mice). Data are presented as mean of the three cages(line), with the highest and lowest value indicated by separate points toshow the variation (total range). Values below the dotted line areconsidered negative.



control groups (Po0.001). No significant difference wasdetected between the i.p. group and the control group.

IgG2a was significantly increased in the immunizedgroup compared with the control groups (Po0.0001) andincreased further to the group challenged by gavage(P = 0.0007). No significant difference was detected be-tween the controls and the i.p. group (Fig. 7).

Western blotting

In order to further characterize the immune responseagainst the lupin extract, we performed a Western blotwith regard to IgE. We included a fish extract from cod asa negative control. At least four distinct bands in therange from 40 to 70 kDa could be detected in the lupinextract with serum from lupin-immunized mice. One ofthese bands (about 42 kDa) could also be detected withserum from non-immunized mice. No bands could bedetected in the fish extract (Fig. 8).

Discussion

Lupin is a novel food ingredient that has been shown toinduce allergy [9, 35–38] and to cross-react with otherlegumes, especially peanut, resulting in serious allergicreactions in peanut-allergic individuals [5, 6, 8, 14]. Wehave developed a peroral immunization mouse model oflupin allergy, with systemic anaphylaxis as the mainoutcome. To determine the optimal immunization dosewhile minimizing the number of animals needed, a novelstatistical design was used.

Challenged p.o.Immunized(not challenged)

Control

IgG

2a (

AU

/mL)

*

*6000

5000

4000

3000

2000

1000

0

7000

Fig. 7. IgG2a antibodies. Levels of lupin-specific IgG2a antibodies weremeasured by ELISA. The mean values (arbitrary units, AU) with 95%confidence intervals for the different groups are shown. The dotted lineindicates the lower detection limit for the ELISA assay. Readings belowthe detection limit were assigned the same value as the detection limit ofthe assay. *Statistically significant difference from control (Po0.0001).

Challenged p.o.Immunized(not challenged)

Control

IL-4

(pg

/mL)

(a)

50

40

30

20

10

60

70

80

90*


Challenged p.o.

IL-5

(pg

/mL)

(b)

400

300

200

100

0

500

600

*

*


Challenged p.o.

IFN

-γ (p

g/m

L)

(c)100 000

10 000

1000

100

0

*

*

IL-4

IL-5

IFN-γ

Fig. 6. Cytokine production. Ex vivo cytokine production by spleen cellscultured with lupin extract (50 mg/mL) for 72 h and measured by ELISA.The spleens were sampled at day 35. The mean values (pg/mL) with 95%confidence intervals for the different groups are shown. The dotted linesindicate the lower detection limits for the ELISA assays. Readings belowthe detection limit were assigned the same value as the detection limit ofthe assay. *Statistically significant difference from control (Po0.05).



Animal models are needed in food allergy research toinvestigate the allergenicity of foods and food compo-nents, and for mechanistic studies. Key elements in animalmodel development are the species and strain of animals,immunization route, allergen dose and the use of adju-vants [17]. The C3H/HeJ mouse is a strain of mice that is ahigh IgE responder and is prone to develop anaphylaxis.Consequently, the C3H/HeJ is well suited as a modelanimal that is able to mimic the human situation [39, 40].As we are dealing with food, the most natural route ofadministration is the oral route. Animals’ innate tendencyto develop tolerance to ingested antigens represents amajor obstacle in the development of oral food allergymodels [17, 41, 42]. It has been shown that C3H/HeJ micehave a reduced tendency to develop oral tolerance [43].Cholera toxin is often used as an adjuvant to overcomeoral tolerance and to induce a Th2 response in oralimmunization models. It has been found to promote IgEand IgG1 responses and to mediate sensitization forsystemic and intestinal anaphylaxis [44, 45]. For ourmodel, we successfully used cholera toxin as an adjuvantin the C3H/HeJ mouse strain.

A major aspect when applying the cholera toxin modelis to establish the optimal dose for each allergen, and atthe same time keep the amount of animals used at aminimum. We therefore used a sequential two-compart-ment response surface statistical design. Because it ispossible to test doses with small intervals, the two-

compartment response surface design will give a moreaccurate dose than other designs. Each mouse was given adifferent dose equally distributed from 0.1 to 10.0 mglupin protein. In the present experiment, we found theoptimal dose of 5.7 mg to have a relatively broad con-fidence interval (3.9–7.0 mg), where even one of the low-est doses used gave a relatively high IgE level. Foss et al.[46] immunized BALB/c-mice with a dose of 5.0 mg lupin,corresponding closely to what we found to be the optimaldose in our model. These investigators, however, do notprovide any explanation to why they selected this dose.Importantly, since other allergens may have a smallerdose window to induce sensitization, the two-compart-ment response surface design will be a valuable method toprecisely determine the optimal dose for sensitization ofindividual allergens.

In a traditional block design [47, 48] one would needabout 40 animals to test three different doses of allergen.We used 29 animals to establish the allergen dose (forstatistical reasons the minimum number of animals is 12in each compartment), and then another 30 animals tofurther characterize the model, as the design can consideronly one parameter at a time. In total, this is more animals(59) than the 40 mice needed in the block design where allparameters can be assessed at once. However, the mainadvantage with the two-compartment response surfacedesign is when the dose window is narrow so that a blockdesign may easily fail to identify a dose for sensitizationunless several experiments are performed. If none of thedoses used in the first experiment had been able to inducesensitization, we would have had to reperform the experi-ment with other doses or changes in some of the otherconditions. Compared with a block design, however, wewould only have used 15 animals in the first experimentand not 40.

Although a clinical anaphylactic reaction is the out-come parameter most relevant for the human situation,the provocation of such a reaction is clearly very stressfulfor the mice. Thus, it will be useful to have otherparameters that are less stressful and at the same timecan be used as markers of anaphylaxis. Also, it would bevaluable to supplement subjective observations of clinicalsigns with an objective biochemical parameter. In ourmodel, we found that post-challenge serum MMCP-1levels corresponded to the clinical severity of the anaphy-lactic reactions. A low-dose challenge might then show anelevation in MMCP-1 without eliciting a full-grade clin-ical anaphylaxis. Interestingly, we also observed that theimmunized mice (day 27) showed higher pre-challengeMMCP-1 levels than the control groups. Possibly, mea-surement of MMCP-1 levels may be used as a predictor ofanaphylaxis if one wishes to prevent the mice fromexperiencing a clinical reaction. It is not known whetherthe higher MMCP-1 levels at day 27, 6 days after the lastimmunization, could be related to the immunization alone

31

45

66

97

20.1

14.4

(a) (b)

250

160

75

50

105

35

2530

15

10

S 1 2 S 1 2 3

Fig. 8. Western blot. (a) SDS-PAGE of the lupin extract (1) and the codextract (2) together with the standard, RPM800 (S). The cod extractserved as a negative control in the immunoblot. (b) IgE immunoblotswith serum from lupin-immunized mice. Lane 1, lupin extract; lane 2,cod extract; lane 3, lupin extract incubated with serum from non-immunized mice; S, MW standard, RPM2107.



or whether the soy content of the chow could play a role.As there is extensive serological cross-reactivity betweenlegumes, there is a possibility that the mice after the initialsensitization with lupin could react to the soy in the foodand therefore have a low-grade ongoing allergic responsein the gut. As the group treated with cholera toxin onlydid not have a heightened level of MMCP-1, the choleratoxin treatment alone does not break the establishedtolerance to soy. The serological cross-reaction betweensoy and lupin is possibly also reflected in the results fromthe Western blotting. The lupin-immunized mice have animmune response against several proteins in the lupinextract (Fig. 8). All mice, including the non-immunizedanimals, react to the same protein band (approximately42 kDa) in the lupin extract. Clinically, this reactivity is ofno importance as non-immunized mice do not developanaphylaxis. Pending experiments with a mouse linemaintained on soy-free chow will elucidate this issue. Fosset al. [46] compared mice on a soy-free diet with mice on anormal diet and reported a difference in IgE response whenthey looked at the purified lupin conglutins (a, b, g and d).

When measuring the IgE levels in the mice during orjust after the anaphylactic reaction, we found that abouthalf of the immunized mice tested negative. This may bedue to several factors, one of them being the functionalnature of the PCA test. This test will only give the level offunctional IgE, in contrast to an ELISA, which gives thelevel of IgE almost regardless of affinity and functionalcapacity. Furthermore, as all sera were tested at a 1 : 4dilution, some of the lowest IgE levels may not have beendetected. The most important factor, however, is probablythe fact that an anaphylactic reaction is caused by thebinding of IgE to the mast cells while we only measure thelevels of free IgE molecules in the serum. Mast cells maybe sufficiently sensitized to start a reaction after allergenbinding, while little or no free IgE can be detected inserum [49–51]. A possibly related phenomenon has beenreported in humans, where it has been observed that someindividuals become temporarily IgE test-negative after asystemic anaphylactic reaction, probably because IgE islocated extravascularly and is bound up in immunecomplexes and on mast cells [52, 53]. This could also beone reason why the IgE titre is higher on day 27 (1 weekbefore provocation) compared with day 35 (during orright after anaphylaxis). In the mouse, IgG1 as well as IgEcan be an anaphylactic antibody. The apparent ‘consump-tion’ of IgE during the anaphylactic reaction in this studysupports the notion that the IgE antibody is activelyinvolved in the reaction in our model. Importantly, at notime point was any lupin-specific IgE detected in the twocontrol groups.

In the present food allergy model we can see a mixedTh2/Th1 response. A hallmark of allergic disease is theinfiltration of Th2 in affected tissues [54]. However,factors such as antigen dose and affinity as well as the

use of adjuvant will influence the balance between a Th1and a Th2 response, and often a mixed response will befound [55, 56]. As our main focus was to create a modelwith clinical anaphylaxis, we chose to use the C3H/HeJmouse. Studies have shown that this mouse strain is proneto anaphylaxis; however, it will not always give anoptimal serological response [56–58]. Other mouse strains,like the BALB/c mouse, would probably have shown astrong IgE response and a clear Th2 cytokine profile, butnot a clinical anaphylactic reaction [57](J. Smit, personalcommunication). In humans, there is no clear correlationbetween a serological response and clinical symptoms ofhypersensitivity and an allergy diagnosis requires the pre-sence of clinical symptoms. Therefore, a mouse model with-out anaphylaxis will not be an adequate model of foodallergy. The genetic background of the mouse strain will alsoinfluence whether a response with Th1 components iselicited [59–61]. C3H/HeJ mice lack functional toll-likereceptor 4 (TLR4) and this receptor has been found to benecessary for the optimal development of Th2 responses [62].

The combined Th1/Th2 response in our model is re-flected in the cytokine profile and the development oflupin-specific IgG2a antibodies. Other studies have shownthat serum antibody response patterns and ex vivo cyto-kine responses are not strongly correlated [59, 63]. This isalso reflected in our study, where IgG2a levels are higherin the challenged groups whereas IFN-g is higher in theimmunized group. However, that all cytokines are higherin the immunized group could be due to the test protocol.The spleen cells harvested on day 35 were first challengedin vivo and then in vitro while the cells harvested on day27 were only challenged in vitro and hence may show astronger reaction. Also, the cells on day 35 were harvestedduring an ongoing anaphylactic reaction. Interestingly,when looking at the cytokine profile at day 27 in theexperiment, we can see a much clearer Th2 response withhigh levels of IL-4 and IL-5. This might reflect the timeperiod since the mice received their last booster immuni-zation (day 21). The use of CT as an adjuvant in this modelcould also possibly influence the Th1/Th2 response as itwill induce a local inflammation in the gut and thuscontribute to a Th1-associated response.

In conclusion, we have established a mouse model oflupin allergy using intragastric sensitization and chal-lenge, which reflects the clinical and immunologicalcharacteristics of an immediate-type food allergy in hu-mans. An anaphylactic reaction to lupin after primaryperoral lupin sensitization was achieved. Levels of MMCP-1paralleled the anaphylactic reactions. Lupin-induced aller-gic reactions in this model appear to be IgE-mediated, anda lupin-dependent, presumably T cell-mediated, cytokineresponse was observed. Finally, an experimental designis described that can be used to reduce the number ofanimals needed and to determine the optimal immuniza-tion dose when establishing new food allergy models.



Acknowledgements

We thank Ase Eikeset, Else-Carin Groeng, Bodil Hasselt-vedt, Berit A. Stensby and Astri Grestad for excellenttechnical assistance, and Lena Haugland Moen at theNational Veterinary Institute for providing the food ex-tracts. We also thank Ellen Namork for critically readingthe manuscript and for helpful comments. This study wasfunded by the Research Council of Norway, as part of theStrategic Institute Program (SIP) at the National Veterin-ary Institute entitled ‘A coordinated research program intofood allergen identification, quantification, modificationand in vivo responses’.

References

1 Rona RJ, Keil T, Summers C et al. The prevalence of food allergy:a meta-analysis. J Allergy Clin Immunol 2007; 120:638–46.

2 Kanny G, Moneret-Vautrin DA, Flabbee J, Beaudouin E, MorissetM, Thevenin F. Population study of food allergy in France.J Allergy Clin Immunol 2001; 108:133–40.

3 The European Food Safety Authority, Opinion of the ScientificPanel on Dietetic Products, Nutrition and Allergies on a requestfrom the Commission relating to the evaluation of allergenicfoods for labelling purposes. The EFSA J 2006; 32:1–197.

4 The European Food Safety Authority, Opinion of the ScientificPanel on Dietetic Products, Nutrition and Allergies on a requestfrom the Commission related to the evaluation of lupin forlabelling purposes. The EFSA J 2006; 2005:1–11.

5 Radcliffe M, Scadding G, Brown HM. Lupin flour anaphylaxis.Lancet 2005; 365:1360.

6 Faeste CK, Lovik M, Wiker HG, Egaas E. A case of peanut cross-allergy to lupine flour in a hot dog bread. Int Arch AllergyImmunol 2004; 135:36–9.

7 Moneret-Vautrin DA, Guerin L, Kanny G, Flabbee J, Fremont S,Morisset M. Cross-allergenicity of peanut and lupine: the risk oflupine allergy in patients allergic to peanuts. J Allergy ClinImmunol 1999; 104:883–8.

8 Hefle SL, Lemanske RF Jr, Bush RK. Adverse reaction to lupine-fortified pasta. J Allergy Clin Immunol 1994; 94:167–72.

9 Parisot L, Aparicio C, Moneret-Vautrin DA, Guerin L. Allergy tolupine flour. Allergy 2001; 56:918–9.

10 Moreno-Ancillo A, Gil-Adrados AC, Dominguez-Noche C,Cosmes PM. Lupine inhalation induced asthma in a child. PediatrAllergy Immunol 2005; 16:542–4.

11 Bernhisel-Broadbent J, Sampson HA. Cross-allergenicity in thelegume botanical family in children with food hypersensitivity. JAllergy Clin Immunol 1989; 83:435–40.

12 Bernhisel-Broadbent J, Taylor S, Sampson HA. Cross-allergeni-city in the legume botanical family in children with food hyper-sensitivity. II. Laboratory correlates. J Allergy Clin Immunol1989; 84:701–9.

13 Leduc V, Moneret-Vautrin DA, Guerin L. Allergenicity of lupinflour. Allerg Immunol (Paris) 2002; 34:213–7.

14 Peeters KA, Nordlee JA, Penninks AH et al. Lupine allergy: notsimply cross-reactivity with peanut or soy. J Allergy ClinImmunol 2007; 120:647–53.

15 Shaw J, Roberts G, Grimshaw K, White S, Hourihane J. Lupinallergy in peanut-allergic children and teenagers. Allergy 2008;63:370–3.

16 Lindvik H, Holden L, Lovik M, Cvancarova M, Halvorsen R. Lupinsensitization and clinical allergy in food allergic children inNorway. Acta Paediatr 2008; 97:91–5.

17 Helm RM. Food allergy animal models: an overview. Ann NYAcad Sci 2002; 964:139–50.

18 Knippels LM, VanWijk F, Penninks AH. Food allergy: what do welearn from animal models? Curr Opin Allergy Clin Immunol2004; 4:205–9.

19 Kimber I, Dearman RJ, Penninks AH et al. Assessment of proteinallergenicity on the basis of immune reactivity: animal models.Environ Health Perspect 2003; 111:1125–30.

20 Li XM, Serebrisky D, Lee SY et al. A murine model of peanutanaphylaxis: T- and B-cell responses to a major peanut allergenmimic human responses. J Allergy Clin Immunol 2000;106:150–8.

21 Li XM, Schofield BH, Huang CK, Kleiner GI, Sampson HA. Amurine model of IgE-mediated cow’s milk hypersensitivity. JAllergy Clin Immunol 1999; 103:206–14.

22 Lee SY, Oh S, Lee K et al. Murine model of buckwheat allergy byintragastric sensitization with fresh buckwheat flour extract. JKorean Med Sci 2005; 20:566–72.

23 Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein mea-surement with the Folin phenol reagent. J Biol Chem 2007;1951:265–75.

24 Clausen SK, Sobhani S, Poulsen OM, Poulsen LK, Nielsen GD.Study of adjuvant effect of model surfactants from the groups ofalkyl sulfates, alkylbenzene sulfonates, alcohol ethoxylates andsoaps. Food Chem Toxicol 2000; 38:1065–74.

25 Braga F, Mota I. Homologous passive cutaneous anaphylaxis(PCA) in mice and heterologous PCA induced in rats with mouseIgE. Immunology 1976; 30:655–9.

26 Ovary Z, Caiazza SS, Kojima S. PCA reactions with mouseantibodies in mice and rats. Int Arch Allergy Appl Immunol1975; 48:16–21.

27 Ovary Z. Passive cutaneous anaphylaxis. In: Weir DM, Herzen-berg LA, Blackwell C, eds. Handbook of experimental immunol-ogy, Volume 1: Immunochemistry. Oxford: Blackwell ScientificPublications, 1975; 33.1–33.9.

28 Ormstad H, Gaarder PI, Johansen BV, Lovik M. Airborne housedust elicits a local lymph node reaction and has an adjuvanteffect on specific IgE production in the mouse. Toxicology 1998;129:227–36.

29 Altman DG. Practical statistics for medical research. London:Chapman & Hall, 1991.

30 Agresti A. Categorical data analysis, 2nd Edn. Hoboken, NJ: JohnWiley & Sons Inc., 2002.

31 Altman DG, Bland JM. Measurement in medicine: the analysis ofmethod comparison studies. The Statistician 1983; 32:307–17.

32 Bland JM, Altman DG. Statistical methods for assessing agree-ment between two methods of clinical measurement. Lancet1986; 1:307–10.

33 Aaras A, Veierod MB, Larsen S, Ortengren R, Ro O. Reproduci-bility and stability of normalized EMG measurements on muscu-lus trapezius. Ergonomics 1996; 39:171–85.

34 Kampen AH, Tollersrud T, Larsen S, Roth JA, Frank DE, Lund A.Repeatability of flow cytometric and classical measurement



of phagocytosis and respiratory burst in bovine polymorpho-nuclear leukocytes. Vet Immunol Immunopathol 2004; 97:105–14.

35 Matheu V, de Barrio M, Sierra Z, Gracia-Bara MT, Tornero P,Baeza ML. Lupine-induced anaphylaxis. Ann Allergy AsthmaImmunol 1999; 83:406–8.

36 Quaresma RR, Viseu R, Martins LM, Tomaz E, Inacio F. Allergicprimary sensitization to lupine seed. Allergy 2007; 62:1473–4.

37 Wuthrich B. Anaphylactic reaction to lupine flour because of aprimary sensitization. Allergy 2008; 63:476–7.

38 Wuthrich B, Mittag D., Ballmer-Weber BK. Pizza: a source ofunexpected food allergens – anaphylactic reaction to lupineflour in a pizza dough and in a gingerbread. Allergologie 2004;27:495–502.

39 Li X, Huang CK, Schofield BH et al. Strain-dependent inductionof allergic sensitization caused by peanut allergen DNA immu-nization in mice. J Immunol 1999; 162:3045–52.

40 Lifrani A, Dubarry M, Rautureau M, Aattouri N, Boyaka PN,Tome D. Peanut-lupine antibody cross-reactivity is not asso-ciated to cross-allergenicity in peanut-sensitized mouse strains.Int Immunopharmacol 2005; 5:1427–35.

41 Strobel S. Neonatal oral tolerance. Ann NY Acad Sci 1996;778:88–102.

42 Lamont AG, Mowat AM, Parrott DM. Priming of systemic andlocal delayed-type hypersensitivity responses by feeding lowdoses of ovalbumin to mice. Immunology 1989; 66:595–9.

43 Kiyono H, McGhee JR, Wannemuehler MJ, Michalek SM.Lack of oral tolerance in C3H/HeJ mice. J Exp Med 1982; 155:605–10.

44 Marinaro M, Staats HF, Hiroi T et al. Mucosal adjuvant effect ofcholera toxin in mice results from induction of T helper 2 (Th2)cells and IL-4. J Immunol 1995; 155:4621–9.

45 Snider DP, Marshall JS, Perdue MH, Liang H. Production of IgEantibody and allergic sensitization of intestinal and peripheraltissues after oral immunization with protein Ag and choleratoxin. J Immunol 1994; 153:647–57.

46 Foss N, Duranti M, Magni C, Frokiaer H. Assessment of lupinallergenicity in the cholera toxin model: induction of IgEresponse depends on the intrinsic properties of the con-glutins and matrix effects. Int Arch Allergy Immunol 2006; 141:141–50.

47 Ruxton GD, Colegrave N. Different experimental designs. In:Ruxton GD, Colegrave N, eds. Experimental design for the lifesciences. Oxford: Oxford University Press, 2006; 69–100.

48 Festing MFW, Overend P, Das RG, Borja MC, Berdoy M. Thedesign of experiments. In: Waynforth B, ed. The design of animalexperiments. Reducing the use of animals in research through

better experimental design. London: The Royal Society ofMedicine Press Ltd on behalf of Laboratory Animals Ltd, 2002;38–70.

49 Fish SC, Donaldson DD, Goldman SJ, Williams CM, Kasaian MT.IgE generation and mast cell effector function in mice deficientin IL-4 and IL-13. J Immunol 2005; 174:7716–24.

50 Finkelman FD. Anaphylaxis: lessons from mouse models. JAllergy Clin Immunol 2007; 120:506–15.

51 Lin XP, Magnusson J, Ahlstedt S et al. Local allergic reaction infood-hypersensitive adults despite a lack of systemic food-specific IgE. J Allergy Clin Immunol 2002; 109:879–87.

52 Aalto-Korte K, Makinen-Kiljunen S. False negative SPT afteranaphylaxis. Allergy 2001; 56:461–2.

53 Goldberg A, Confino-Cohen R. Timing of venom skin tests andIgE determinations after insect sting anaphylaxis. J Allergy ClinImmunol 1997; 100:182–4.

54 Kay AB. Allergy and allergic diseases. First of two parts. N Engl JMed 2001; 344:30–7.

55 Rogers PR, Croft M. Peptide dose, affinity, and time of differ-entiation can contribute to the Th1/Th2 cytokine balance. JImmunol 1999; 163:1205–13.

56 Van Wijk F, Hartgring S, Koppelman SJ, Pieters R, Knippels LM.Mixed antibody and T cell responses to peanut and the peanutallergens Ara h 1, Ara h 2, Ara h 3 and Ara h 6 in an oral sensi-tization model. Clin Exp Allergy 2004; 34:1422–8.

57 Morafo V, Srivastava K, Huang CK et al. Genetic susceptibility tofood allergy is linked to differential TH2–TH1 responses in C3H/HeJ and BALB/c mice. J Allergy Clin Immunol 2003; 111:1122–8.

58 Baeza ML, Conejero L, Higaki Y et al. Anisakis simplex allergy: amurine model of anaphylaxis induced by parasitic proteinsdisplays a mixed Th1/Th2 pattern. Clin Exp Immunol 2005; 142:433–40.

59 Nygaard UC, Aase A, Lovik M. The allergy adjuvant effect ofparticles – genetic factors influence antibody and cytokineresponses. BMC Immunol 2005; 6:11.

60 Granum B, Gaarder PI, Eikeset A, Stensby BA, Lovik M. The adjuvanteffect of particles-importance of genetic background and pre-sensitisation. Int Arch Allergy Immunol 2000; 122:167–73.

61 Whitehead GS, Walker JK, Berman KG, Foster WM, Schwartz DA.Allergen-induced airway disease is mouse strain dependent. AmJ Physiol Lung Cell Mol Physiol 2003; 285:L32–L42.

62 Dabbagh K, Dahl ME, Stepick-Biek P, Lewis DB. Toll-like receptor4 is required for optimal development of Th2 immune responses:role of dendritic cells. J Immunol 2002; 168:4524–30.

63 Samuelsen M, Nygaard UC, Lovik M. Allergy adjuvant effect ofparticles from wood smoke and road traffic. Toxicology 2008;246:124–31.



Documents

A mouse model of lupin allergy