Protein Expression and Purification 73 (2010) 36–45

Contents lists available at ScienceDirect

Protein Expression and Purification

journal homepage: www.elsevier.com/locate/yprep

Expression, purification, cross-reactivity and homology modeling of peanut profilin

Cerrone Cabanos a, Mary Rose Tandang-Silvas a, Van Odijk b, Peter Brostedt c, Akira Tanaka d,
Shigeru Utsumi a, Nobuyuki Maruyama a,*
a
Laboratory of Food Quality Design and Development, Graduate School of Agriculture, Kyoto University, Uji, Kyoto 611-0011, Japan
b
Department of Respiratory Medicine and Allergology, Sahlgrenska Academy, Göteborg University, Gothenburg, Sweden
c
Research & Development, Phadia AB, Uppsala, Sweden
d
Phadia KK, Tokyo, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Plant profilins are known pan-allergens involved in the cross-reactions between pollen and plant foods.
Received 19 February 2010 Peanut profilin, Ara h 5, is one of the important peanut allergens. Presently, most immunological, bio-
and in revised form 11 March 2010 chemical and structural studies on peanut allergens have focused on the three major allergens (Ara h
Available online 15 March 2010
1, 2 and 3). Here Ara h 5 was cloned, expressed in Escherichia coli, Rosetta2(DE3) (Novagen), purified using
a combination of ammonium sulfate fractionation and size-exclusion chromatography and yielded a total
Keywords: of 29 mg/l of culture. IgE reactivity was assayed using multiplexed microarray with other peanut aller-
Peanut allergen
gens (Ara h 1, 2, 3, and 8) and birch (Bet v 2) and timothy (Phl p 2) profilin using sera from peanut allergic
Profilin
Ara h 5
Swedish patients. Using homology modeling, Ara h 5 structure was also generated, compared against
Expression other profilins and utilized to predict surface-exposed residues potentially forming epitopes. The allergen
Cross-reactivity was recognized by 3 out of 33 sera (9.1%). IgE reactivity to Ara h 5 also coincided with that of two other
Homology modeling profilins, Phl p 12 and Bet v 2, confirming cross-reactivity. Interestingly, IgE reactivity to Ara h 5 was
Surface-exposed epitopes higher than above-mentioned profilins which may be indicating specificity of sera towards peanut pro-
filin. Eight surface-exposed epitopes were predicted and verified against experimentally validated
sequential epitopes. Three epitopes (#1, 5 and 7) mostly located at the accessible loops and neutral to
relatively electropositive sites were found common among profilins, which should be involved in
cross-reactivity. A specific putative epitope (#4) was also found which may explain the relative high
IgE reactivity to Ara h 5 as compared to the other profilins. Due to its close relation to other allergenic
profilins, Ara h 5 could be used as a model and allergen of choice for profilin allergy diagnosis.
Ó 2010 Elsevier Inc. All rights reserved.

Introduction To date, 11 peanut (Arachis hypogea) allergens have been offi-

cially listed by the Allergen Nomenclature Sub-committee of the
Peanut allergy is one of the ‘‘big eight” food allergies worldwide International Union of Immunological Societies (IUIS) that are
accounting for 0.2–1.5% prevalence in children and still rising [1– termed as Ara h 1 through Ara h 11 (http://www.allergen.org). Of
5]. The rise in peanut allergy is not limited to children. A National these, the three major allergenic peanut proteins are Ara h 1
Health and Nutritional Examination Survey in the United States (63.5 kD), Ara h 2 (17–19 kD) and Ara h 3 (64 kD). All three are
(1988–1994) showed that about 8.6% of Americans are sensitized storage proteins, Ara h 1 belongs to the vicilin-type 7S globulin
to peanuts [6]. Similarly, national surveys suggest that 1.1% of while Ara h 2 is a 2S albumin belonging to the conglutin storage
Americans or three million individuals are allergic to peanuts, tree proteins. Ara h 3 is an 11S globulin belonging to the glycinin fam-
nuts, or both [3,4]. Symptoms range from mild oral allergy syn- ily. Another identified peanut allergen is profilin, which is also
drome (OAS)1 to anaphylactic reactions and even death [7–9]. Thus known as Ara h 5. Although peanut profilin is considered a minor
far, there is no cure for peanut allergy. Therapy focuses mainly on allergen, it is an important cross-reactive allergen to patients suf-
peanut avoidance, early recognition of symptoms due to inadvertent fering from multivalent type I allergy leading to its designation
ingestions, and pharmacologic treatment of adverse reactions. as a pan-allergen [10]. Profilins are 14–17 kDa proteins which are
found in all eukaryotic cells. They were first described by Carlsson
* Corresponding author. Fax: +81 774 38 3761. and colleagues [11] as actin-binding proteins in calf spleen that are
E-mail address: marunobu@kais.kyoto-u.ac.jp (N. Maruyama). responsible for regulating the normal distribution of filamentous
1
Abbreviations used: PDB, protein data bank; OAS, oral allergy syndrome; IPTG, acting networks. Since the discovery of its allergenicity in 1991
Isopropyl b-D-1-thiogalactopyranoside; IUIS, International Union of Immunological
[12], profilins were identified as allergen of many fruits, vegeta-
Societies; APBS, adaptive Poisson-Boltzmann solver; RMSD, root mean square
deviation; HSE, half sphere exposure; PLP, poly-L-proline.
bles, foods, as well as pollen. Profilins are highly conserved with

1046-5928/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.pep.2010.03.005
 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45 37

sequence similarity ranging from 70 to 80% among known plant The recombinant protein was then expressed in LB medium con-
profilins [13]. taining 0.5 M NaCl and 50 lg/ml carbenicillin. Initially, three dif-
Most of the immunological, biochemical and structural studies ferent incubation temperatures were tested, namely, 20, 30 and
on peanut allergens have focused on the three major allergens. 37 °C. Although the level of recombinant protein expression was
Natural peanut profilin is also difficult to obtain in considerable comparable among the three conditions, cells were incubated at
amount since it is present in very low amounts in crude allergenic 20 °C due the concomitant high expression level of host cell pro-
extract [28,33]. It has also been shown that several plant profilins teins found when incubated at 30 and 37 °C, which may not pro-
can be efficiently produced as stable native-like recombinant pro- vide the relative ease in a tagless purification. Cultures were
teins in Escherichia coli [34] and have been reported to be purified induced by addition of Isopropyl b-D-1-thiogalactopyranoside
by affinity chromatography either as a histidine-tagged protein (IPTG) to a final concentration of 1 mM at OD600nm = 0.60 and incu-
(using metal affinity) [35] or as an untagged protein employing bation continued for 40 h. Cells were harvested by centrifugation
poly-L-proline interaction [28,36]. Here we describe the cloning, at 6700g and pellets were resuspended and sonicated in 35 mM so-
expression in E. coli, and purification of Ara h 5 using an alternative dium phosphate, pH 7.6, 0.4 M NaCl, 1 mM EDTA, 0.1 mM (p-ami-
(in the absence of affinity columns) two-step process involving dinophenyl) methanesulfonyl fluoride (p-APMSF), 1.2 lM
only ammonium sulphate precipitation and size-exclusion chro- leupeptin, 0.2 lM pepstatin A, 0.02% (w/v) NaN3. Insoluble material
matography. We also report the IgE reactivity to recombinant Ara was removed by centrifugation at 9800g and supernatant was sub-
h 5 by protein microarray assay of sera from peanut allergic Swed- jected to ammonium sulfate precipitation. Solid ammonium sulfate
ish patients in comparison with other peanut allergens (Ara h 1, 2, [(NH4)2SO4] was slowly added to the crude extract up to 35% satu-
3 and h 8) and known allergenic profilins (Phl p 12 and Bet v 2). We ration with stirring on an ice bath. After the sample was kept on ice
also describe the generated homology structure of Ara h 5, pre- with stirring for 30 min, it was centrifuged at 4 °C for 30 min
dicted surface-exposed epitopes and compared it against the (9800g). The supernatant was then further added with ammonium
above-mentioned profilins and two other – Cuc m 2 and Hel a 2 sulfate up to 60% saturation and centrifuged at 9800g. The precip-
profilin. We also attempted to give insight into the observed rela- itate was resuspended in the same buffer.
tive high reactivity of sera to Ara h 5 against that of other profilins The protein was then applied on HiLoad 16/60 Superdex 75
and their cross-reactivity. Complementary to the in silico predic- prep-grade column (GE Healthcare Life Sciences, NJ) equilibrated
tion, we further compared the identified potential epitopes of Ara with the same buffer at a flow rate of 0.5 mL/min. To check for pur-
h 5 to the experimentally validated linear or sequential epitopes ity, fractions were analyzed by SDS–PAGE using 11% polyacryl-
of other profilins. amide gels according to the procedure of Laemmli [37], and
those containing mainly the bands belonging to Ara h 5
(14.1 kDa) were pooled together as one fraction.
Methods

Reverse transcription-polymerase chain reaction (RT-PCR) and cloning Protein measurement

of Ara h 5 cDNA
Protein concentration was determined using a Protein Assay Ra-
By using RT-PCR, we cloned the cDNA encoding Ara h 5 based pid Kit (Wako, Osaka, Japan) with bovine serum albumin as the
from Genbank accession No. GU354312. Total RNA was isolated from standard.
peanut seeds using TRIzol reagent (Invitrogen). The profilin cDNA
was amplified using the RNA LA PCR Kit (AMV), Ver. 1.1 (Takara
Recombinant Ara h 5 biological activity and protein microarray assay
Bio, Japan). The first-strand cDNA was synthesized with an oli-
go(dT)-adapter primer containing an M13 primer M4 sequence
Peanut allergic Swedish patients (n = 33) sera (Phadia AB) se-
(contained in the RNA LA PCR Kit). The product was then used for
lected based on positive case history, skin prick test and Immuno-
PCR amplification using the forward primer 50 -TCGTGGCAAACC-
CAP IgE detection were analysed using multiplexed immunoassay
TACGTCGATAACC-30 corresponding to the N-terminus and reverse
[14]. All allergen reagents were standardized at a concentration
primer 50 -ATTCAATAAGCTTTTAAAGACCCGTATCAATGAGA-30 corre-
of 1 mg/ml and spotted at a volume of 300 pl. Purified Ara h 5
sponding to the C-terminal end of the gene containing stop codon
was covalently immobilized to polystyrene particles (0.11 lm
(bold font, shown in reverse manner) and HindIII restriction site
diameter, Bangs Laboratories Inc., Carmel, IN, USA) and arrayed
(underlined). This pair of primers was specifically designed to in-
onto nitrocellulose membrane together with other peanut aller-
clude only the mature region of the Ara h 5 protein and without
gens (from Phadia AB) – Ara h 1 and 3 (storage proteins), Ara h 2
any polypeptide or protein tag. The reaction was performed using
(2S albumin) and Ara h 8 (pathogenesis-related protein 10, PR-
LA Taq DNA Polymerase (Takara Bio) with 30 cycles at 94 °C for
10), and reference profilin allergens Bet v 2 (birch tree) and Phl p
2 min, 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min. The PCR prod-
12 (timothy grass). Particles were spotted in an array format using
uct was further amplified using KOD-Plus DNA Polymerase (TOY-
a nano-plotter (Type 2.0 system by Gsellschaft für Silizium-Mikro-
OBO, Japan) with 30 cycles at 94 °C for 2 min, 94 °C for 30 s, 55 °C
systeme mbH, GeSim, Grosserkmannsdorf, Germany). Diameter of
for 30 s, and 68 °C for 1.5 min. The amplified fragment with the ex-
arrayed spots was approximately 200 lm and the density of the ar-
pected size was then phosphorylated, and treated with HindIII. The
ray was around 400 spots/cm2. The assay procedure comprised
resultant fragment was eventually ligated with pET21d (Novagen,
sequential additions of 30 ll washing buffer (phosphate buffer at
CA) that was previously treated with NcoI, blunted, treated further
neutral pH with blocking agents and detergent), serum (30 ll),
with HindIII and finally dephosphorylated. Sequencing and insert
washing buffer (30 ll), fluorophore-conjugated monoclonal anti-
confirmation was carried out according to the dideoxy method using
human IgE antibody (20 ll) and washing buffer (2  30 ll), before
an ABI Prism 3100 DNA analyzer (Applied Biosystems).
reading the fluorescence intensity of each spot using a GenePix
4000B laser scanner (Molecular Devices, Sunnyvale, CA, USA). IgE
High-level expression and protein purification reactivity was reported semi-quantitatively and compared against
background binding adjacent to each spot and binding of sera to
The generated pET21d plasmid containing Ara h 5 coding region control spots containing only phosphate-buffered saline solution
was transformed into E. coli, Rosetta2(DE3) (Novagen, CA) strain. (PBS).
38 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45

Homology modeling (www.salilab.org/modweb). In Modweb, models are built for each

one of the sequence-structure matches by satisfaction of spatial re-
In this part of the study, Ara h 5 was compared to either gener- straints based on the widely used program Modeller [15]. Ara h 5
ated model or PDB structure of other profilins, namely; Phl p 12 (GU354312) and Phl p 12 (CAA54686) were modeled using crystal
(Timothy grass, Phleum pretense), Bet v 2 (Birch tree, Betula verru- structure of Arabidopsis thaliana, Ara t 8, 3NUL (78% and 76% iden-
cosa, 1CQA), Cuc m 2 (Melon, Cucumis melo), Hel a 2 (Sunflower, tity, respectively) as template; Cuc m 2 (AJ565931) using Hevea
Helianthus annuus). The unsolved structures were modeled brasiliensis, Heb v 8, 1G5U (83% identity); and Hel a 2 (Y15210)
employing the automated comparative modeling server ModWeb using Bet v 2 (AAA16522), 1CQA (72% identity). Stereochemical

Fig. 1. (A) Amino acid sequence alignment of Ara h 5 (GU354312), Phl p 12 (CAA54686), Bet v 2 (AAA16522), Cuc m 2 (AJ565931) and Hel a 2 (Y15210). Residues in red and
green font are the predicted surface-exposed epitopes based on Discotope and BEpro algorithm. Residues in green are the common epitopes of these profilins. On top of the
alignment is the consensus secondary structure. coil or loop, alpha-helix, and beta-sheet. (B) Time course of Ara h 5 expression after induction with 1 mM
Isopropyl-b-D-1-thiogalactopyranoside at 20 °C incubation. Lane 1, 0 h; lane 2, 4 h; lane 3; 12 h; lane 4, 24 h; lane 5, 36 h; lane 6, 50 h after induction; lane M, molecular
weight markers. (C) Purification of recombinant peanut profilin expressed in Rosetta2(DE3) E. coli. Lane 1, soluble lysate; lane 2, after ammonium sulfate fractionation; lane 3,
HiLoad Superdex purified recombinant peanut profilin (Ara h 5); lane M, molecular weight markers. (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)
 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45 39

Fig. 2. (A) Individual patient data and summary microarray analysis of Ara h 5 along with Ara h 1, 2, 3, 8, Bet v 2 and Phl p 12 allergens. The low, medium, and high IgE
reactivity are shown in different shades of blue. (B) Array plot of patient J6 serum sample. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)

quality of the polypeptide backbone and side chains was evaluated tical sites (77 identical residues out of 133) among the five se-
using Ramachandran plots generated from RAMPAGE [16]. Amino quences. Ara h 5 sequence is most similar to Cuc m 2 (84.7%),
acid environment of models was evaluated using Verify3D [17]. followed by Phl p 12 (78.6%), Bet v 2 (75.2%), and Hel a 2 (69.9%).

Prediction of surface-exposed B-cell epitopes Expression and purification of rAra h 5

Structure-based prediction of surface-exposed B-cell epitopes The final construct was transformed into Rosetta2(DE3) E. coli.
was done using the DiscoTope 1.2 [18] and BEpro [19] (formerly This strain is a BL21 derivative designed to alleviate codon bias
PEPITO). Both algorithms combine the amino acid propensity when expressing heterologous proteins in E. coli. The cells carry a
scales and solvent accessibility scores. Discotope uses contact chloramphenicol-resistant plasmid with the tRNA genes that de-
numbers (i.e., the number of Ca atoms within a certain distance code seven rare codons (AGA, AGG, AUA, CUA, GGA, CCC, and
threshold) while BEpro employs the newly developed half sphere CGG) to improve the yield of full-length proteins. The time course
exposure values [20] at multiple distances from target residue in of the expression was examined in cell pellet after separation in
calculating accessibility. Half-sphere exposure (HSE) is a newly SDS–PAGE by protein staining with Coomassie blue (Fig. 1B). Major
developed two-dimensional solvent exposure measure. By concep- band with an apparent molecular mass of 14.1 kDa can be visual-
tually separating an amino acid’s sphere in a protein structure into ized, and the highest level of production was obtained after 36 h.
two half spheres which represent its distinct spatial neighborhoods A two-step procedure consisting of ammonium sulfate precipi-
in the upward and downward directions [20]. Both programs gen- tation and size-exclusion chromatography was used to purify re-
erate a pdb-like list file of residues with corresponding score. Here combinant Ara h 5. Cells were harvested by centrifugation, pellet
two threshold scores were used for both algorithms. The minimum was sonicated, and supernatant was ammonium sulfate fraction-
threshold signifies a sensitivity of 47% and specificity of 75% (7.7 ated. Fig. 1C shows the pooled fraction of the protein after ammo-
and 0.9 for Discotope and BEpro, respectively). The upper threshold nium sulfate precipitation. The precipitate was then resuspended
signifies sensitivity of 32% and a specificity of 85% (6.0 and 1.1 for in the same sonication buffer and loaded on HiLoad 16/60 Super-
Discotope and BEpro, respectively). Residues were then grouped as dex 75 pg column. The purified protein (Fig. 1C) migrated as a sin-
putative epitopes according to their proximity on surface of the gle band and in agreement with expected size as indicated by gel
profilins. Structures were superimposed by common geometrical filtration profile (data not shown) and SDS–PAGE. Total yield of
core using Multiprot [21]. Molecular surfaces and electrostatic the purification process was 29 mg/l of culture of cells.
potentials (APBS plug-in) were calculated and displayed using
PyMOL.
Allergenicity and cross-reactivity between Ara h 5 and other profilins

Results IgE reactivity was further classified into three semi-quantitative

categories: low reactivity, medium reactivity, and high reactivity
Cloning of peanut profilin (Fig. 2A). Of the thirty-three peanut allergic Swedish patients, only
three sera (J6, J19 and J30) showed significant positive response
The cDNA encoding the allergen Ara h 5 was obtained by using with medium reactivity to Ara h 5 (9.1%). Serum from patient J6
specific primers, containing linker and restriction site, correspond- (Fig. 2B) had high IgE reactivity to Ara 1, 2 and 3. J19 serum had
ing to the N and C terminus of the mature protein, and was directly low reactivity to Ara h 8 while J30 had medium reactivity to Ara
cloned into pET21d expression plasmid. The size of PCR product h 8 and low reactivity to Ara h 3. Ara h 5 also highly coincided with
was 406 bp (containing the 396 bp Ara h 5 coding region) at an the IgE reactivity of Bet v 2 and Phl p 12 against all sera confirming
optimal concentration 1 mM Mg2+ and primer annealing at 55 °C. the well-documented fact of profilin cross-reactivity. Interestingly,
Fig. 1A shows the amino acid sequence alignment of Ara h 5 with IgE reactivity to Ara h 5 appears to be higher than the two profilins
the other four profilins: Phl p 12, Bet v 2, Cuc m 2, and Hel a 2. The which may be indicating higher specificity (i.e. presence of specific
high sequence similarity is reflected by the presence of 57.9% iden- epitopes) of sera towards peanut profilin.
40 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45

Fig. 3. (A) Ribbon diagram of the three-dimensional model of Ara h 5. Strands of b-sheet and stretches of a-helix are in yellow and red, respectively. Coil structures or loops
are in green, N and C indicate the N- and C-terminus of the polypeptide, respectively. (B) Molecular surface of Ara h 5 calculated and displayed with PyMOL. (C) Alpha-carbon
superimposition of Ara h 5 (blue), Phl p 12 (black), Bet v 2 (orange), Cuc m 2 (green), and Hel a 2 (red). The top and bottom arrows indicate the N-terminal alpha helix and the
variable loop, respectively. (D–E) Ribbon representation of the superimposed structures of Ara h 5, Phl p12, Bet v 2, Cuc m 2 and Hel a 2 with their consensus epitopes shown
in yellow. (D) Coil view and (E) residues involved in the different epitopes are shown as sticks. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)

Structural analysis and epitope prediction of 0.8 Å while that between Ara h 5 and Bet v 2 showed 1.9 Å. The
similarity between Ara h 5 and Cuc m 2 showed an RMSD of 2.1 Å
As antibodies bind to epitopes formed by the spatial arrange- while that between Ara h 5 and Hel a 2 showed a value of 2.48 Å.
ment of amino acid residues on surface, model structure of Ara h The only major structural differences as shown in superimposition
5 was generated and compared to model structures of Phl p 12, of their peptide backbones were observed with respect to the
Cuc m 2 and Hel a 2, and PDB structure of Bet v 2 (1CQA). The direction of the N-terminal a-helix and orientation of the highly
structure of Ara h 5 is very identical to the other profilins accessible variable loop (Fig. 3C–E). Apparently peanut profilin is
(Fig. 3A and B). Peanut profilin is composed of three a-helices, se- closest to the backbone structure of timothy grass Phl p 12. Bet v
ven b-strands and ten turns. It also displays the same motif of se- 2 appears to lack the common variable loop conformation for high
ven-stranded but incomplete and anti-parallel up-and-down b- solvent accessibility. Both Cuc m 2 and Hel a 2 possess a slightly
strand barrels observed in other profilins. The similarity between bent loop oriented towards their N-terminal a-helix and may not
Ara h 5 and Phl p 12 revealed a root mean square deviation (RMSD) have the same accessibility as that of Ara h 5 and Phl p 12.
Putative surface-exposed epitopes of the five profilins. Their common epitopes are highlighted in gray. Residue number is shown as subscript. Residues in bold font are the amino acids with scores equal to or higher than the upper C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45 41
threshold value (discotope score P 6.0 and z-score P 1.1 for Discotope and BEpro program, respectively) generated by either one or both algorithms. The upper threshold value signifies a detection sensitivity of 32% and a specificity of
Variations in this loop conformation (loop 1) resulted in the ab-
sence of a putative epitope in Bet v 2 (Fig. 1A) and hence have a sig-
nificant effect on the antigenicity of this region. The difference in
antigenicity of this region can be expected as it is highly variable
in terms of amino acid sequence among profilins.
Using two algorithms, we found potential surface-exposed
sequential and discontinuous epitopes of Ara h 5 and other profi-
lins. Table 1 shows all predicted surface epitopes, which were
numbered identically based on Ara h 5, for the five profilins. There
were eight epitope sites found in Ara h 5, six sites for Phl p 12, Bet v
2 and Cuc m 2, and four sites for Hel a 2. Four out of eight (#1, 3, 4,
and 8) epitopes were sequential in Ara h 5, two out of six (#3 and
6) in Phl p and three out of six (#1, 3 and 8) in Bet v 2. Likewise,
three out of six epitopes (#1, 3 and 8) were sequential in Cuc m
2 and two out of four in Hel a 2 (#1 and 2). Most portions of these
epitopes, depicted as yellow and green (consensus epitopes) col-
ored surface patches, were located on the front surface of each pro-
filin (Fig. 4A–E). We also mapped the electrostatic potentials on the
surface of profilin based on the pdb structure of Bet v 2 and found
that most patches correspond to the highly-exposed and relatively
neutral to electropositive (Lys-rich) regions which were on the
front side (Fig. 4F).
There were also some epitopes found common to some but not
in all five profilins. Epitopes #2 was common to all except Bet v 2,
while epitope #3 was common only to Ara h 5, Phl p 12 and Bet v 2.
Epitopes #6 was found in all except Hel a 2 and epitope #8 was
found only in Ara h 5, Bet v 2 and Cuc m 2. Epitope #4 (S37N38)
was present only in Ara h 5, which remarkably has the most epi-
topes encompassing the entire protein. This could explain the rel-
atively high IgE reactivity in Fig. 2. As expected from structure
similarity, Ara h 5 shared more epitopes with Phl p 12 (#1–3, 5–
7), Bet v 2 (#1, 3, 5–8) and Cuc m 2 (#1, 2, 5–8) than with Hel a
2 (#1, 2, 5, 7). Three epitopes, also on the front side, were found
highly similar in sequence and conformation in all five proteins.
These epitopes should be the regions responsible for cross-reactiv-
ity. Based on Ara h 5 sequence, these were the M1S2W3Q4T5Y6 (#1),
P40Q41K43P44E45T48A55E56P57G58S59P62T63G64 (#5), and K87G88P89-
G90D107E108P109T111P112G113 (#7). Residues in bold typeface are
the ones which had scores equal to or higher than the upper
threshold score (6.0 and 1.1 for Discotope and BEpro, respec-
tively), which may also mean high antigenicity. Epitope #1 is com-
posed mostly of neutral to weakly polar residues (M1, S2, W3, T5,
Y6) and one neutral residue (Q3). Epitope #5 is composed of 10
neutral to weakly polar residues (P40P44T48A55P57G58S59P62T63G64)
and four polar residues (Q41K43E45E56). Epitope #7 is composed of
six neutral to weakly polar (G88P89G90P109T111P112) and three polar
residues (K87D107E108). Fig. 5 revealed in a close-up view the con-
formational similarity of epitopes #1, 5, and 7 among profilins.
Most pronounced resemblance was shown between Ara h 5 and
Phl p 12. We further looked at the common secondary structural
features of these consensus epitopes. Superimposition of ribbon
structures (Fig. 3D–E, Fig. 1A) showed that residues were located
mostly at the highly accessible loop regions of profilins. Epitope
#1 is an exception which was located at the N-terminal site of a-
helix 1. Most residues of epitope #5 were found at the loops flank-
ing a-helix 2 and a few residues within the helix itself. Portions of
epitope #7 were located at the loops in between b-sheets 5 and 6,
and in between b-sheet 7 and a-helix 3. Fig. 3E depicts the stick
view of their residues forming a protruding interface.

Discussion

Profilin is an ubiquitous protein found in almost all eukaryotic

Table 1

cells. It was first identified as a minor allergen in birch pollen. So

85%.

far, profilins were described as pan-allergens in botanically closely

42 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45

Fig. 4. Putative exposed epitopes (in both yellow and green colored patches) of Ara h 5 (A), Phl p 12 (B), Cuc m 2 (C), Hel a 2 (D), and Bet v 2 (E) mapped on their surfaces
(front and back side). Epitope regions in green (#1, 5 and 7) are the shared or consensus epitopes of the three profilins. (F) Mapping of electrostatic potentials, calculated with
APBS of PyMOL, on the molecular surface of Bet v 2 (front and back side). Electropositive and electronegative regions are in blue and red, respectively. Neutral regions are
colored white. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45 43

Fig. 5. Three-dimensional conformation of the B-cell epitopes #1, 5 and 7 (green patches) rendered on the molecular surfaces of the five profilins. These images were
calculated and displayed with PyMOL. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

related and distantly related plants or unrelated plant [12,13]. This BL21 derivative designed to alleviate codon bias when expressing
study is primarily aimed at investigating the role of the pan-aller- heterologous proteins in E. coli. This superior strain carries a chlor-
gen profilin as a cross-reactive allergen in peanut and giving in- amphenicol-resistant plasmid with the tRNA genes that decode se-
sight into its cross-reactivity by structural homology modeling. ven rare codons (AGA, AGG, AUA, CUA, GGA, CCC, and CGG) to
In doing so, we cloned the cDNA encoding Ara h 5 and expressed improve the yield of full-length proteins. We were also able to pur-
it in Rosetta2(DE3) (Novagen) E. coli strain. Previous initial studies ify the protein using a two-step process, consisting of ammonium
have mentioned the use of BL21(DE3) and recently BL21-Codon- sulfate fractionation and size-exclusion chromatography, and
Plus(DE3)-RIL, containing extra copies of tRNA genes decoding yielded a total of 29 mg/l of cell culture, which is comparable with
for rare codons in E. coli, to compensate for any incompatibility be- the yield of affinity chromatography-based (either using Poly-L-
tween the codon preference in the mRNA of the recombinant gene proline-Sepharose or Ni-NTA resin) peanut profilin purifications
and in the E. coli host [38]. The strain used in this study is also a [28,35,38]. Our procedure can also be a good alternative option
44 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45

in the absence of a suitable affinity column or resin. In contrast might explain the higher IgE reactivity of sera to Ara h 5 than that
with PLP-Sepharose-based affinity chromatography where high of Phl p 12 and Bet v 2.
concentration of urea is used to elute bound profilin [28], our pro- This study shows that the interplay of common and unique epi-
cedure uses non-denaturing reagents and eliminates the need for topes in influencing the IgE reactivity to Ara h 5 and, in general,
long period of dialysis (at least 24 h) to remove urea from the puri- profilin allergenicity and cross-reactivity. All these point to the
fied fractions. plausibility of using Ara h 5 as a model and allergen of choice for
Using the purified allergen, we were able to detect IgE reactivity profilin allergy diagnosis.
of peanut allergic Swedish patients’ sera to peanut profilin employ-
ing multiplexed protein microarray assay. This highly sensitive and
References
high-throughput, though semi-quantitative, immunoassay makes
it possible to determine specific IgE against multiple allergens [1] J. Grundy, S. Matthews, B. Bateman, T. Dean, S.H. Arshad, Rising prevalence of
simultaneously in one same patient, with a minimum amount of allergy to peanut in children: data from 2 sequential cohorts, J. Allergy Clin.
Immunol. 110 (2002) 784–789.
serum [22]. Here we presented that this technique can also aid in
[2] M. Osterballe, T.K. Hansen, C.G. Mortz, A. Host, C. Bindslev-Jensen, The
detecting cross-reactions. Three (J6, J19 and 30) of thirty-three sera prevalence of food hypersensitivity in an unselected population of children
(9.1%) had IgE reactivity to the recombinant profilin. We compared and adults, Pediatr. Allergy Immunol. 16 (2005) 567–573.
the reactivity profile of Ara h 5 with that of other peanut allergens [3] S.H. Sicherer, A. Munoz-Furlong, A.W. Burks, H.A. Sampson, Prevalence of
peanut and tree nut allergy in the US determined by a random digit dial
(Ara h 1, 2, 3 and 8) (data not shown) and noted that there seems to telephone survey, J. Allergy Clin. Immunol. 103 (1999) 559–562.
be two groups of individuals in terms of reactivity to peanut aller- [4] S.H. Sicherer, A. Munoz-Furlong, H.A. Sampson, Prevalence of peanut and tree
gens. As can also be inferred from previous studies [23–29], the nut allergy in the United States determined by means of a random digit dial
telephone survey: a 5-year follow-up study, J. Allergy Clin. Immunol. 112
first group is the high response group that appears to react strongly (2003) 1203–1207.
with the three major allergens (Ara h 1, 2 and 3) and probably their [5] S.E. Emmett, F.J. Angus, J.S. Fry, P.N. Lee, Perceived prevalence of peanut allergy
homologous allergens (Ara h 4 and 6). The fact that these major in Great Britain and its association with other atopic conditions and with
peanut allergy in other household members, Allergy 54 (1999) 380–385.
allergens are present in large amount in peanuts [24–26] could [6] S.J. Arbes Jr, P.J. Gergen, L. Elliott, D.C. Zeldin, Prevalences of positive skin test
mean that this group can be a high risk group. The second group responses to 10 common allergens in the US population: results from the third
is the low to medium response group comprised of individuals National Health and Nutrition Examination Survey, J. Allergy Clin. Immunol.
116 (2005) 377–383.
who may not necessarily react to major allergens but reactive to [7] S.A. Bock, A. Munoz-Furlong, H.A. Sampson, Fatalities due to anaphylactic
the minor allergens, such as, Ara h 5 and 8. We may consider this reactions to foods, J. Allergy Clin. Immunol. 107 (2001) 191–193.
group as a low risk group. It seems that patients J6, J19, and J30 be- [8] G. Roberts, Anaphylaxis to foods, Pediatr. Allergy Immunol. 18 (2007) 543–
548.
long to the latter group. Furthermore, Ara h 5 reactivity pattern
[9] H.S. Skolnick, M.K. Conover-Walker, C.B. Koerner, H.A. Sampson, A.W. Burks,
showed strong correspondence to other profilins (Phl p 12 and R.A. Wood, The natural history of peanut allergy, J. Allergy Clin. Immunol. 107
Bet v 2) indicating cross-reactivity. The two other profilins gave po- (2001) 367–374.
sitive reaction also only in the three patient sera. The presence of [10] R. Valenta, M. Duchene, C. Ebner, P. Valent, C. Sillaber, P. Deviller, F. Ferreira, M.
Tejkl, H. Edelmann, D. Kraft, Profilins constitute a novel family of functional
very high similarity in terms of RMSD (ranging from 0.8 to 2.4 Å) plant allergens, J. Exp. Med. 175 (1992) 377–385.
among model structures and known pdb structure of profilin signi- [11] L. Carlsson, L.E. Nystrom, I. Sundkvist, F. Markey, U. Lindberg, Actin
fies that the generated structures using homology modeling are polymerizability is influenced by profilin, a low molecular weight protein in
non-muscle cells, J. Mol. Biol. 115 (1977) 465–483.
reliable enough for structure-based epitope prediction. Eight [12] R. Valenta, M. Duchene, K. Pettenburger, C. Sillaber, P. Valent, P. Bettelheim, M.
surface-exposed epitopes were found in Ara h 5 namely; M1S2W3- Breitenbach, H. Rumpold, D. Kraft, O. Scheiner, Identification of profilin as a
Q4T5Y6 (#1), E14E16G17N18 (#2), D29G30S31V32 (#3), S37N38 (#4), novel pollen allergen; IgE autoreactivity in sensitized individuals, Science 253
(1991) 557–560.
P40Q41K43P44E45T48A55E56P57 G58S59P62T63G64 (#5), G68G69T70- [13] C. Radauer, K. Hoffmann-Sommergruber, Profilins, in: C. Mills, P. Shewry
K71M73I75Q76E78P79G80 (#6), K87G88P89G90D107E108P109T111P112- (Eds.), Plant Food Allergens, Blackwell, Oxford, UK, 2004, pp. 105–124.
G113 (#7) and D128T129G130 (#8). These putative epitopes, except [14] A. Adachi, T. Horikawa, H. Shimizu, Y. Sarayama, T. Ogawa, S. Sjolander, A.
Tanaka, T. Moriyama, Soybean b-conglycinin as the main allergen in a patient
for some portions of epitope #5 and 6, strongly coincided with
with food-dependent exercise-induced anaphylaxis by tofu: food processing
the regions in Bet v 2, Cuc m 2 and Hel a 2 containing experimen- alters pepsin resistance, Clin. Exp. Allergy 39 (2009) 167–173.
tally validated linear epitopes (shaded regions in Fig. 1A) [30–32]. [15] A. Sali, T.L. Blundell, Comparative modelling by satisfaction of spatial
restraints, J. Mol. Biol. 234 (1993) 779–815.
This strong correspondence between the predicted epitopes and
[16] S.C. Lovell, I.W. Davis, W.B. Arendall III, P.I.W. de Bakker, J.M. Word, M.G.
experimentally-determined linear epitopes further affirms our Prisant, J.S. Richardson, D.C. Richardson, Structure validation by C alpha
confidence in the two algorithms (Discotope and BEpro) used. geometry: phi, psi and C beta deviation, Protein Struct. Funct. Genet. 50 (2002)
Moreover, three epitopes (#1, 5, and 7) were shown to be very 437–450.
[17] R. Luthy, J.U. Bowie, D. Eisenberg, Assessment of protein models with three-
similar in all the five profilins in terms of sequence and three- dimensional profiles, Nature 356 (1992) 83–85.
dimensional conformation (Table 1, Fig. 5). Residues of these three [18] P.H. Andersen, M. Nielsen, O. Lund, Prediction of residues in discontinuous B
epitopes were also mostly located at the loop regions and neutral cell epitopes using protein 3D structures, Protein Sci. 15 (2006) 2558–2567.
[19] M.J. Sweredoski, P. Baldi, PEPITO: improved discontinuous B-cell epitope
to relatively electropositive sites on the protein surface. This seems prediction using multiple distance thresholds and half sphere exposure,
to be expected since the loop regions are also the most accessible Bioinformatics 24 (2008) 1459–1460.
site in terms of solvent exposure in profilins. We also noticed that [20] T. Hamelryck, An amino acid has two sides: a new 2D measure provides a
different view of solvent exposure, Proteins Struct. Func. Bioinf. 59 (2005) 38–
epitopes #5 and 7 also partially corresponded to Lys-rich actin- 48.
binding site of profilin, which explains the reported partial inhibi- [21] M. Shatsky, R. Nussinov, H.J. Wolfson, A method for simultaneous alignment
tion of IgE reactivity to the protein upon treatment with actin [30]. of multiple protein structures, Proteins Struct. Funct. Bioinf. 56 (2004) 143–
156.
Interestingly, the linear epitopes (shaded regions in Fig. 1A) over-
[22] M. Ferrer, M.L. Sanz, J. Sastre, J. Bartra, A. del Cuvillo, J. Montoro, I. Jáuregui, I.
lapping epitope #1, portions containing epitope # 5 and 7 are Dávila, J. Mullol, A. Valero, Molecular diagnosis in allergology: application of
the three highly IgE reactive epitopes in sunflower profilin, Hel a the microarray technique, J. Investig. Allergol. Clin. Immunol. 19 (Suppl. 1)
(2009) 19–24.
2 [31]. Among these three, the linear epitope containing portion
[23] F. Blanc, K. Adel-Patient, M.F. Drumare, E. Paty, J.M. Wal, H. Bernard, Capacity
of #5 is the most reactive epitope in Hel a 2 as well as in Cuc m of purified peanut allergens to induce degranulation in a functional in vitro
2 [32]. All or any of these three epitopes should therefore be in- assay: Ara h 2 and Ara h 6 are the most efficient elicitors, Clin. Exp. Allergy 39
volved in the cross-reactivity between different profilins. Remark- (2009) 1277–1285.
[24] A.W. Burks, Peanut allergy, Lancet 371 (2008) 1538–1546.
ably, Ara h 5 also possessed most epitopes found in the entire [25] A.E. Flinterman, E. van Hoffen, C.F. den Hartog Jager, S. Koppelman, S.G.
length of profilins and one specific epitope #4 (S37N38), which Pasmans, M.O. Hoekstra, C.A. Bruijnzeel-Koomen, A.C. Knulst, E.F. Knol,
 C. Cabanos et al. / Protein Expression and Purification 73 (2010) 36–45 45

Children with peanut allergy recognize predominantly Ara h2 and Ara h6, epitopes defined by monoclonal antibodies, Int. Immunol. 14 (2002) 993–
which remains stable over time, Clin. Exp. Allergy 37 (2007) 1221–1228. 1001.
[26] S.J. Koppelman, M. Wensing, M. Ertmann, A.C. Knulst, E.F. Knol, Relevance of [32] G. López-Torrejón, A. Díaz-Perales, J. Rodríguez, R. Sánchez-Monge, J.F. Crespo,
Ara h1, Ara h2 and Ara h3 in peanut-allergic patients, as determined by G. Salcedo, L. Pacios, An experimental and modeling-based approach to locate
immunoglobulin E Western blotting, basophil-histamine release and IgE epitopes of plant profilin allergens, J. Allergy Clin. Immunol. 119 (2007)
intracutaneous testing: Ara h2 is the most important peanut allergen, Clin. 1481–1488.
Exp. Allergy 34 (2004) 583–590. [33] A. Vallverdu, J.A. Asturias, M.C. Arilla, N. Gomez-Bayon, A. Martinez, J.
[27] G.W. Palmer, D.A. Dibbern Jr., A.W. Burks, G.A. Bannon, S.A. Bock, H.S. Martinez, R. Palacios, Characterization of recombinant Mercurialis annua
Porterfield, R.A. McDermott, S.C. Dreskin, Comparative potency of Ara h 1 and major allergen Mer a 1 (profilin), J. Allergy Clin. Immunol. 101 (1998) 363–370.
Ara h 2 in immunochemical and functional assays of allergenicity, Clin. [34] S. Laffer, S. Vrtala, M. Duchêne, R. van Ree, D. Kraft, O. Scheiner, R. Valenta, IgE-
Immunol. 115 (2005) 302–312. binding capacity of recombinant timothy grass (Phleum pratense) pollen
[28] T. Kleber-Janke, R. Crameri, S. Scheurer, S. Vieths, W.M. Becker, Patient-tailored allergens, J. Allergy Clin. Immunol. 94 (1994) 88–94.
cloning of allergens by phage display: peanut (Arachis hypogaea) profilin, a [35] M. Sankian, M. Yousefi, N. Pazouki, A. Varasteh, One-step purification of
food allergen derived from a rare mRNA, J. Chromatogr. B Biomed. Sci. Appl. histidine-tagged profilin with high purity and yield by using metal
756 (2001) 295–305. precipitation, Biotechnol. Appl. Biochem. 47 (2007) 185–189.
[29] A. Mari, E. Scala, P. Palazzo, S. Ridolfi, D. Zennaro, G. Carabella, Bioinformatics [36] G. Pauli, J.P. Oster, P. Deviller, S. Heiss, J.C. Bessot, M. Susani, F. Ferreira, D.
applied to allergy: allergen databases, from collecting sequence information to Kraft, R. Valenta, Skin testing with recombinant allergens rBet v 1 and birch
data integration. The allergome platform as a model, Cell. Immunol. 244 profilin, rBet v 2: diagnostic value for birch pollen and associated allergies, J.
(2006) 97–100. Allergy Clin. Immunol. 97 (1996) 1100–1109.
[30] A.A. Fedorov, T. Ball, N.M. Mahoney, R. Valenta, S.C. Almo, The molecular basis [37] U.K. Laemmli, Cleavage of structural proteins during the assembly of the head
for allergen cross-reactivity: crystal structure and IgE-epitope mapping of of bacteriophage T4, Nature 227 (1979) 680–685.
birch pollen profilin, Structure 5 (1997) 33–45. [38] T. Kleber-Janke, W.M. Becker, Use of modified BL21(DE3) Escherichia coli cells
[31] J.A. Asturias, N. Gómez-Bayón, M.C. Arilla, L. Sánchez-Pulido, A. Valencia, A. for high-level expression of recombinant peanut allergens affected by poor
Martínez, Molecular and structural analysis of the panallergen profilin B cell codon usage, Protein Expr. Purif. 19 (2000) 419–424.