SEARCH

SEARCH BY CITATION

Keywords:

  • allergic bronchopulmonary aspergillosis;
  • cross-reactivity;
  • molecular mimicry;
  • self-antigens;
  • thioredoxin

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Background:  Thioredoxins are cross-reactive allergens involved in the pathogenesis of atopic eczema and asthma. Cross-reactivity to human thioredoxin can contribute to the exacerbation of severe atopic diseases.

Methods:  Human thioredoxin, Asp f28 and Asp f29, two thioredoxins of Aspergillus fumigatus, and thioredoxin of Malassezia sympodialis were cloned and produced as recombinant proteins. Allergenicity and cross-reactivity to thioredoxins in allergic bronchopulmonary aspergillosis patients were assessed by enzyme-linked immunosorbent assay (ELISA), inhibition ELISA, immunoblot analysis, proliferation assays and skin tests. Molecular homology modelling was used to identify conserved, surface-exposed amino acids potentially involved in immunoglobulin E (IgE)-binding.

Results:  All thioredoxins, including the human enzyme, bind IgE from patients with allergic bronchopulmonary aspergillosis and induce allergen-specific proliferation in peripheral blood mononuclear cells and positive skin reactions in thioredoxin-sensitized patients. Inhibition experiments showed that the thioredoxins are cross-reactive indicating humoral immune responses based on molecular mimicry. To identify structural surface elements involved in cross-reactivity, the three-dimensional structures were modelled based on solved thioredoxin structures. Analysis of the molecular surfaces combined with sequence alignments allowed identification of conserved solvent exposed amino acids distantly located in the linear sequences which cluster to patches of continuous surface areas. The size of the surface areas conserved between human and fungal thioredoxins correlates well with the inhibitory potential of the molecules in inhibition ELISA indicating that the shared amino acids are involved in IgE-binding.

Conclusions:  Conserved, solvent exposed residues shared between different thioredoxins cluster to continuous surface regions potentially forming cross-reactive conformational B-cell epitopes responsible for IgE-mediated cross-reactivity and autoreactivity.

Skin test surveys suggest that 3–10% of the worldwide population is affected by fungal allergy (1, 2), and sensitization to Aspergillus species is quite common in patients with asthma or cystic fibrosis (CF) (3, 4). Among the moulds involved in pulmonary complications Aspergillus fumigatus plays a predominant role (5).

The different A. fumigatus-related pulmonary diseases can be broadly classified according to their clinical conditions into systemic mycoses, saprophytic colonization and allergic diseases (6). Aspergillus fumigatus-related allergy include rhinitis, sinusitis, allergic asthma and allergic bronchopulmonary aspergillosis (ABPA) (7). ABPA is the most severe complication related to A. fumigatus sensitization, affects almost exclusively patients with severe asthma or CF, and is difficult to diagnose (8). Although ABPA as a syndrome can only be diagnosed based on clinical signs, serologic tests are used to confirm a suspected ABPA and recombinant A. fumigatus allergens have been proven useful to discriminate between A. fumigatus sensitization and ABPA in both, asthmatic and CF patients (4, 9, 10). Aspergillus fumigatus allergic patients mount immunoglobulin E (IgE)-immune responses to secreted fungal allergens present as aeroallergens like Asp f1 (11), or secreted proteases (12). ABPA patients, however, have or had the fungus growing in the lung (13) and, as a result of fungal damage due cellular defence mechanisms, become exposed also to nonsecreted proteins. Classical examples of intracellular allergens are manganese-dependant superoxide dismutase (MnSOD, rAsp f6) (14, 15) differentially expressed only during germination (16) and ribosomal P2 protein (rAsp f8) (17). IgE-responses to these allergens are detected almost exclusively in sera of patients with ABPA. In concordance with the presence of rAsp f6 or rAsp f8-specific serum IgE, sensitized ABPA patients strongly react in skin test challenges to these allergens (14, 17) as well as to homologous proteins (18, 19). Interestingly, phylogenetically highly conserved proteins like MnSOD (18), thioredoxin (Trx) (20, 21) and cyclophilin (22, 23) are members of large pan-allergen families exhibiting a high degree of cross-reactivity also with their homologous human self-antigens (14, 15, 17, 20–24). Here we show that two cross-reactive thioredoxins are recognized by sera of patients with ABPA. As already shown for patients suffering from atopic eczema (20) and for baker’s asthma (21), also ABPA patients sensitized to fungal thioredoxin recognize human thioredoxin as an IgE-binding self-antigen further corroborating the involvement of IgE-mediated autoreactivity in the pathogenesis of chronic atopic disorders.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Cloning, protein expression and purification of thioredoxins

Asp f28 and Asp f29, two A. fumigatus Trxs’, Malassezia sympodialis (Mala s13), a Trx of M. sympodialis, and human Trx were cloned and produced as described (20). In brief, fungal Trx cDNAs were amplified by PCR from clones isolated from phage surface displayed fungal libraries (25). Human Trx was amplified from a commercial human lymphoma U937 lung cDNA library (Stratagene, La Jolla, CA, USA). The BamHI/HindIII restricted amplicons were subcloned into pQE-30 (Qiagen, Hilden, Germany) and transformed into Escherichia coli M15 cells. Clones carrying correct sequenced inserts were used to produce recombinant N-terminal hexahistidine-Trx tagged proteins, purified under native conditions by nickel affinity chromatography (20). Molecular size and purity of the recombinant proteins were analysed by SDS–PAGE (NuPAGE, 12% Bis Tris; Invitrogen, Amsterdam, The Netherlands) under denaturing, reducing conditions. Enzymatic activity was assessed by determining the catalytic reduction of insulin (21) and lipopolysaccharide (LPS) was measured with the QCL-1000 endotoxin kit (Cambrex Bio Science, MD, Walkersville, USA).

Subjects selection, routine assessments and skin tests

Sera from 40 ABPA patients, selected according to clinical history and positive skin prick test (SPT) to the fungal extract (Allergopharma, Hamburg, Germany), were analysed together with sera from 10 healthy controls. Allergen-specific IgE to A. fumigatus extract was quantitatively determined using the CAP system (Phadia, Uppsala, Sweden), according to the package inserts. The diagnosis of ABPA was based on at least seven of the eight criteria used for the diagnosis of the disease in A. fumigatus-sensitized asthmatics (26). All patients were free of chest infiltrates, had stable bronchial asthma, and received no medication during the time of the study. SPT with recombinant Trx proteins were performed as described (15). The study was carried out according to a clinical protocol approved by the local ethical committee. All participants gave written informed consent after a full explanation of the procedure given individually before testing.

Immunoblots, IgE immunoassays and inhibition ELISA

For Western blots, 1 μg of recombinant protein was subjected to 12% SDS–PAGE (Invitrogen), electro-transferred onto Hybond ELC membranes (Amersham Biosciences, Buckinghamshire, UK) and processed as described (22). IgE-binding to recombinant Trx’s was determined by a direct solid phase enzyme-linked immunosorbent assay (ELISA) in Nunc-Immuno Plates (Maxisorp Surface; Nunc, Roskilde, Denmark) coated and processed as described (22). Results were expressed as EU/ml calibrated against the absorbency (OD405nm) of an in-house reference standard arbitrarily defined as 100 EU/ml for each allergen tested (11). For inhibition ELISA coating and blocking was performed as for standard ELISA. In a separate 96-well plate Trx were serially diluted 1 : 3 in blocking buffer starting with a concentration of 45 μM and incubated with 1 : 10 diluted patients’ sera for 2 h at 37°C. Bovine serum albumin (BSA) served as a negative control. The preincubated sera were transferred to the coated plates and incubated for 2 h at 37°C. Residual IgE-binding was determined as described above. Percentage of inhibition was calculated from the background-corrected absorbency values of samples where serum was incubated in blocking buffer without inhibiting antigen.

Proliferation assay of peripheral blood mononuclear cells

Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized peripheral venous blood samples by Ficoll density gradient centrifugation, washed three times in PBS and resuspended in RPMI 1640 medium supplemented with 1 mM sodium pyruvate, 2 mM l-gutamine, 1% MEM nonessential amino acids and vitamins, 100 μg/ml streptomycin, 100 U/ml penicillin, 100 μg/ml kanamycin (all from Invitrogen, Basel, Switzerland) and 10% heat-inactivated foetal calf serum (BioConcept, Allschwil, Switzerland). The optimal concentration for proliferation was determined for each single allergen in dose finding experiments using equimolar concentrations of 1, 10, 100 and 1000 nM. Thereafter, aliquots (106 cells/well) were stimulated in triplicate with optimal antigen concentrations for 6 days in humidified atmosphere with 5% CO2. Proliferation was measured as incorporation of tritiated thymidine (Hartmann Analytics, Braunschweig, Germany) during the final 8 h of incubation.

Structure modelling and calculation of the solvent-accessible area

The unsolved A. fumigatus Trx structures were modelled with the prediction program 3D-JIGSAW by automated comparative modelling (27). A model of the three-dimensional structure of Asp f28 was obtained using the Trx structure of E. coli (PDB code 2TRX) (28) as template. Asp f29 was modelled based on the M. sympodialis Trx structure (PDB code 2J23) (20). Structure models were superimposed by structural sequence alignment assessed with DS Visualizer (Accelrys, Cambridge, UK). Solvent-accessible surface areas were calculated for all described pan-allergens with DS Visualizer (Accelrys), using a probe radius of 1.4 Å with 240 grid points per atom.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Production and characterization of Trxs

The cDNAs encoding the homologous allergens Asp f28 (AJ937744), Asp f29 (AJ937745), Mala s13 (AJ937746) and human thioredoxin (X54539) were cloned and produced as described (20). The proteins purified by Ni2+-NTA chromatography migrated as a single band in agreement with the predicted size and showed the expected enzymatic activity (data not shown). The proteins were tested for LPS content and values were below 1 EU/μg protein for all recombinant proteins.

Allergenic properties of thioredoxins

The IgE-binding capacity of the purified recombinant Trxs was first assessed with sera of patients sensitized to A. fumigatus by Western blot analysis. All Trxs were able to bind serum IgE (Fig. 1) indicating that the Trxs of different origins are cross-reactive. IgE-binding was confirmed and surveyed by standard ELISA using sera of 40 patients with ABPA and 10 sera of healthy control individuals. Results were expressed as arbitrary EU/ml (11) and were considered positive when determinations exceeded three times the mean EU/ml value of the healthy control group (Fig. 2). Sensitization prevalence against Asp f28 and Asp f29 of around 30% and 50%, respectively, was found for the patients suffering from ABPA. The frequency of sensitization against Mala s13 was approximately 50% and comparable to those found for human Trx.

image

Figure 1.  Western blot analysis: specific IgE-binding to recombinant human Trx (lane 1), Asp f28 (lane 2), Asp f29 (lane 3) and Mala s13 (lane 4) analysed by Western blotting using serum from A. fumigatus-sensitized individuals. Molecular-mass sizes (kDa) are indicated on the left site.

Download figure to PowerPoint

image

Figure 2.  IgE-binding to thioredoxins in an antigen-specific ELISA calibrated against the binding to a reference serum arbitrarily assigned to 100 EU/ml (11). Forty patients with ABPA were analysed. Mean values are indicated by lines. The hatched areas represent the threefold value of the mean EU/ml value of 10 healthy individuals for the respective allergen used as cut-off values for the discrimination between positive and negative patients.

Download figure to PowerPoint

SPT were performed in five ABPA patients with detectable Trx-specific serum IgE, five ABPA patients with Trx-specific IgE levels below the cut-off values reported in Fig. 2, and five healthy controls. SPT were positive for all Trx proteins in all five ABPA patients who had Trx-specific serum IgE antibodies. This result shows that all Trx induce IgE cross-linking on effector cells in vivo, demonstrating the allergenicity of the recombinant proteins. The lack of skin reactivity in healthy individuals and in ABPA patients without Trx-specific serum IgE shows that the proteins do not induce unspecific reactions.

Cross-reactivity between fungal and human thioredoxins

We preformed competition ELISA to investigate whether the IgE of patients sensitized to Asp f28 cross-reacts to different Trx. The result reported in Fig. 3A suggests an extended cross-reactivity between the homologous proteins. Cross-reactivity was detected between Asp f28 and Asp f29 (45% sequence identity), whereas cross-reactivity between Asp f28 and human Trx (41% sequence identity) was less pronounced (Fig. 3A). Mala s13, which shares 39% sequence identity to Asp f28 showed the weakest inhibition. BSA was used as a negative control and no inhibition could be observed. To investigate the cross-reactivity between human Trx and other allergenic Trx’s, we coated human Trx on solid phase and performed inhibition ELISA. The result (Fig. 3B) confirms an extended cross-reactivity between all Trx. Interestingly, Asp f28 showed the highest inhibition, but shares the lowest sequence identity to human Trx among the tested thioredoxins (41%). An explanation for the differences in cross-reactivity between the structures can be derived from detailed analysis of the modelled 3D structures of the Trxs (see below).

image

Figure 3.  Competitive inhibition of IgE-binding to recombinant human Trx coated on solid phase. Serum from A. fumigatus-sensitized patients was pre-incubated with increasing amounts of recombinant Asp f28 (□), Asp f29 (bsl00001), human Trx (Δ), Mala s13 (•), or BSA as a negative control (♦). Pre-incubated serum samples were transferred to wells coated with Asp f28 (A) or human Trx (B), and residual IgE-binding analysed by ELISA.

Download figure to PowerPoint

Primary proliferative responses of PBMC from Trx-sensitized individuals

As allergen-specific IgE production is dependent on T cell help, we measured the proliferative responses of PBMCs from five ABPA patients with serum IgE to Trx, five A. fumigatus-sensitized patients without anti-Trx IgE, and 10 healthy controls to the different Trx. Dose finding experiments using 1, 10, 100 and 1000 nM protein showed that the optimal concentration of Asp f28, Asp f29, or human Trx for PBMC stimulation was in the range of 100 nM (≈1.2 μg/ml) independently of the antigen used. The mean proliferation after stimulation for six days at optimal antigen concentrations (1.2 μg/ml) showed significant differences between Trx-sensitized ABPA patients, A. fumigatus-sensitized asthmatics lacking Trx-specific IgE antibodies, and healthy controls (Fig. 4).

image

Figure 4.  PBMC proliferation after 6 days stimulation with different Trxs: PBMCs of 5 ABPA patients, 5 sensitized to Aspergillus and 10 healthy individuals were exposed to optimal concentrations of the recombinant Asp28, Asp f29, human Trx and BSA for 6 days (see text). Proliferation was measured by incorporation of tritiated thymidine in counts per minute. Significant higher proliferation (*P > 0.005) was observed in Trx-sensitized ABPA patients compared healthy controls and A. fumigatus-sensitized individuals without Trx-specific serum IgE. The hatched area represents values of >3 for the stimulation indices considered below the level of relevance.

Download figure to PowerPoint

Superposition of Trx discloses putative IgE-binding surface areas

All thioredoxins show extended sequence identity at amino acid level ranging from 39% to 49% (Table 1). Asp f28 shares 41% identity with human Trx. Superposition of the backbones of the modelled Asp f28 structure with the solved human Trx structure revealed a close structural relationship with a root mean square deviation of 2.39 Å for all Cα atoms. This structural similarity is relatively low compared to those of the Asp f29 structure modelled using E. coli Trx as template with a RMSD of 1.30 Å, and those of 1.11 Å obtained by direct superposition of the solved structures of Mala s13 and human Trx (20). However, because Asp f28, Asp f29, Mala s13 and human Trx show cross-reactivity in vitro (Fig. 3) and in vivo as demonstrated by SPT, they must share at least two IgE-binding epitopes to allow cross-linking of effector cell-bound allergen-specific IgE. Obviously, only residues that are at least partly exposed to solvent can contribute to antigen–antibody interactions. Thus solvent-accessible amino acids, which are conserved between thioredoxin proteins, are potentially involved in the IgE-mediated cross-reactivity experimentally confirmed in vitro (Fig. 3A and B). A sequence alignment shows that a total of 45, 53 and 48 for Asp f28, Asp f29 and Mala s13, respectively, are identical to residues present in human Trx (Fig. 5). From the identical amino acids 14, 18 and 19 residues scattered over the linear sequences of Asp f28, Asp f29 and Mala s13, respectively, are identical to human Trx, and at least 30% solvent exposed (Fig. 5). The identical, strongly and weakly similar, and different amino acid residues compared to human Trx are shown in the linear sequence in Fig. 5 and topologically mapped on 3D-structure models of the different thioredoxins (Fig. 6).

Table 1.   Biochemical characteristics and identity table
AllergenAA*MW†Asp f28Asp f29Human TrxMala s13
  1. *Amino acids.

  2. †Molecular weight (kDa) without HIS-tag.

Asp f2810811.9100454139
Asp f2911012.0 1004849
Human Trx10511.7  10044
Mala s1310611.6   100
image

Figure 5.  Structural and sequence alignment of Trxs. The top line shows the secondary structure assigned by DSSP (34), the second line shows the number and position of the conserved patch residue and the third line shows the sequence numbering of human Trx. Identical (*), strongly similar (:) and weakly similar (.) amino acids in all sequences are indicated. Active site residues are in bold. Loops, which adopt a different conformation compared with human Trx, are underlined Identical residues involved in the formation of patches which are at least 30% solvent exposed and involved in the formation of contiguous patches on the 3D structures are coloured in red (patch 1), yellow (patch 2), blue (patch 3) and green (patch 4), respectively.

Download figure to PowerPoint

image

Figure 6.  Shared solvent exposed surfaces of human and fungal Trx: (A) human Trx, (B) Asp f28, (C) Asp f29 and (D) Mala s13. The front of the molecules on the first row shows patch 1 (red) conserved throughout the structures and patch 2 (yellow) only conserved between human Trx and Asp f28. The back of the molecules (second row) depicts patch 3 (blue) conserved throughout the structures and the small patch 4 (green) only conserved between human Trx and Mala s13. Surface residues not conserved between the structures are shown in grey.

Download figure to PowerPoint

As shown in the structural models (Fig. 6), identical amino acids located distantly in the linear sequences (Fig. 5) cluster to form four different continuous surface areas. Patch 1 is formed by the residues constituting the active centre of the enzyme and flanking residues (Thr30, Trp31, Gly33, Pro34, Lys36, Met37, Pro40) plus two other amino acids located on sequentially more distance loops (Asp60 and Ala92). This patch matches completely between Mala s13 and human Trx as already described and covers a solvent-accessible area of 867 Å2 (20). In the structural model of Asp f29 the surface accessible area is reduced to 774 Å2 due to a point mutation of Met37 to Ala, whereas in the Asp f28 model a conserved mutation at position 36 (Lys36/Arg) and two exchanges at position 37 and 92 (Met37/Ala, and Ala92/Gly) results in patch surface size of 675 Å2. Patch 2 involves four residues located in the loop between the end of α helix 2 and the beginning of β sheet 3 (Glu47, Lys48, Ser50, Asn51; Fig. 5) forming a surface area of 414 Å2, only present in human Trx and Asp f28. Patch 3 consists of 6 aa on two different loops (Glu68, Gly83–Gln84–Lys85–Val86, Glu88) resulting in a solvent-accessible surface of 527 Å2 conserved between human Trx and Asp f29. Two mutations in Mala s13 (Val86/Ile and Glu88/Thr), and four amino acid exchanges in Asp f28 (Lys85/Pro, Val86/Leu and Glu88/Lys) contribute to a reduction of the conserved surface area to 433 and 344 Å2, respectively. Patch 4 is made up of three residues after the first α helix (Glu19–Lys21) and covers a surface region of 239 Å2 but is only conserved between human Trx and Mala s13. Due to the surface area which is smaller than those required for an energetic B-cell epitope (29) and the lack of conservation it is unlikely that this region of the molecule can considerably contribute to the cross-reactivity observed among the different Trx’s.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

IgE-binding proteins sharing homology belong to groups of phylogenetically conserved protein families called pan-allergens. The most interesting cross-reactive proteins are those showing sequence homology between environmental allergens and human self-antigens, which can lead to humoral and cell-mediated autoreactivity (14, 17, 19, 20, 23). Patients suffering from severe atopic diseases often show elevated IgE-levels against self-antigens indicating that the pathogenesis of allergic diseases can be influenced by cross-reactive structures. The most convincing evidence that self-reactivity is involved in the pathogenesis of atopic diseases derives from the demonstration that simple application of human MnSOD to healthy skin areas of patients with atopic eczema induces eczematous reactions (15). Here we performed a study on cross- and self-reactivity to thioredoxins in patients with ABPA. Asp f28, Asp f29, Mala s13, and human Trx were all recognized by serum IgE from the majority of the 40 ABPA patients tested (Fig. 2). The allergens induced proliferation exclusively in PBMC’s from ABPA patients showing detectable amounts of Trx-specific serum IgE, whereas PBMC’s of healthy controls and of A. fumigatus-sensitized individuals without Trx-specific serum IgE consistently failed to proliferate (Fig. 4). However, the demonstration that proteins act as allergen is their ability to elicit immediate skin reactions. Among 15 individuals tested, positive SPT to all Trxs were observed only in patients with detectable serum IgE antibodies to Trx. SPT reactions depend on rapid degranulation of mast cells and mediator release, particularly histamine, triggered through cross-linking of high-affinity IgE-receptors (30). The skin test results show that all Trxs elicit IgE cross-linking on mast cells in vivo, and reactivity to human Trx suggests humoral autoimmune responses.

Extended, although incomplete, cross-reactivity between all thioredoxins was demonstrated by inhibition ELISA (Fig. 3). This indicates that primary sensitization to fungal Trxs induces polyclonal responses eliciting additional IgE-responses to surface areas not conserved between the homologous structures.

A partial explanation for the incomplete cross-reactivity can be obtained by an extended structural analysis. Antibody–antigen interfaces bury a surface area of 560–900 Å2 involving 10–22 amino acids of the antigen which interact with the antibody (31, 32). Detailed structural study of Fv-lysozyme complexes revealed that the productive binding is mediated by only five to six energetically important residues constituting an ‘energetic’ epitope with a surface area of approximately 250–300 Å2 (29). We compared the modelled structures of Asp f28 and Asp f29 to the solved crystal structure of human Trx (33) by structural alignments. The identical solvent exposed residues, scattered over the linear sequence (Fig. 5), cluster to four independent patches on the surface of human Trx (Fig. 6). Patch 1 is well conserved in all Trxs and fulfils the criteria for a conformational B-cell epitope (20). Patch 2 covers a surface of 414 Å2 and lies in a region with sequential and structural deviation in Asp f29 and Mala s13 compared to human Trx and Asp f28 (Figs. 5 and 6). This potential B-cell epitope is only shared between human Trx and Asp f28. Patch 3 shows the largest conserved surface between human Trx and Asp f29 (527 Å2), whereas two and four amino acids mutations in Mala s13 and Asp f28, respectively, contribute to a reduction of the patch size to 433 and 344 Å2, still fulfilling the criteria required for energetic B-cell epitopes. Patch 4 is only conserved between human Trx and Mala s13 but considering the small surface (239 Å2) is unlikely to contribute to IgE-binding. Notably the gradual reduction of the total solvent-accessible surface areas shared between human Trx, Asp f28 (1433 Å2), Asp f29 (1301 Å2) and Mala s13 (1300 Å2), coincides with the decreasing inhibitory capacity of the proteins in competition ELISAs (Fig. 3B). These results suggest that the conserved patches are indeed involved in IgE-binding and cross-reactivity. This structural information could be exploited to engineer a Trx molecule with reduced IgE-binding capacity. Such molecules could be useful for the treatment of ABPA patients with an autoimmune background due to thioredoxin sensitization. However, because A. fumigatus produces a complex mixture of IgE-binding molecules (7) it is unlikely that a single molecule would be useful as a general vaccine against A. fumigatus-related allergic complications.

In summary, the data presented provide strong evidence for in vitro and in vivo humoral and cell-mediated autoreactivity to human Trx in ABPA clearly traceable back to conserved surface-exposed amino acids. Molecular mimicry is the most plausible explanation for the reactions observed (18, 23).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

This work was supported by the Swiss National Science Foundation (Grants 3100-63381/2 and 31000-114634/1), and by the OPO-Pharma Foundation, Zürich.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References