Allergenic lipid transfer proteins from plant-derived foods do not immunologically and clinically behave homogeneously: the kiwifruit LTP as a model

Abstract

Background: Food allergy is increasingly common worldwide. Tools for allergy diagnosis measuring IgE improved much since allergenic molecules and microarrays started to be used. IgE response toward allergens belonging to the same group of molecules has not been comprehensively explored using such approach yet.

Objective: Using the model of lipid transfer proteins (LTPs) from plants as allergens, including two new structures, we sought to define how heterogeneous is the behavior of homologous proteins.

Methods: Two new allergenic LTPs, Act d 10 and Act c 10, have been identified in green (Actinidia deliciosa) and gold (Actinidia chinensis) kiwifruit (KF), respectively, using clinically characterized allergic patients, and their biochemical features comparatively evaluated by means of amino acid sequence alignments. Along with other five LTPs from peach, mulberry, hazelnut, peanut, mugwort, KF LTPs, preliminary tested positive for IgE, have been immobilized on a microarray, used for IgE testing 1,003 allergic subjects. Comparative analysis has been carried out.

Results: Alignment of Act d 10 primary structure with the other allergenic LTPs shows amino acid identities to be in a narrow range between 40 and 55%, with a number of substitutions making the sequences quite different from each other. Although peach LTP dominates the IgE immune response in terms of prevalence, epitope recognition driven by sequence heterogeneity has been recorded to be distributed in a wide range of behaviors. KF LTPs IgE positive results were obtained in a patient subset IgE positive for the peach LTP. Anyhow, the negative results on homologous molecules allowed us to reintroduce KF in patients' diet.

Conclusion: The biochemical nature of allergenic molecule belonging to a group of homologous ones should not be taken as proof of immunological recognition as well. The availability of panels of homologous molecules to be tested using microarrays is valuable to address the therapeutic intervention.

Figure 1. Act d 10 complete primary structure.

Arrows indicate fragments obtained by enzymatic digestion with trypsin (T), Asp-N (D) and chymotrypsin (C). The amino acid sequence of the N-terminal region obtained by direct sequencing of the entire molecule is indicated by N-term. Peptides are numbered according to their order in the sequence.

Figure 2. Amino acid identity table and sequence alignments of Act d 10, Ara h 9, Art v 3, Cor a 8, Mor n 3, and Pru p 3.

Panel A: Table of amino acid identity reported as paired percent values; Panel B: LTP sequence alignments; Act d 10 and Pru p 3 sequence numbering are indicated at the top and at the bottom of the alignments, respectively. Background colors identify identical amino acids as follows: Yellow: identical amino acids in all six sequences; Blue: identical amino acids in Act d 10 and Pru p 3 and some other LTP sequences; Green: amino acids of Act d 10 substituted in Pru p 3, but conserved in at least one of the other sequences; Red: amino acids of Pru p 3 substituted in Act d 10, but conserved in at least one of the other sequences; Brown: identical amino acids only in Act d 10 and Pru p 3 sequences; White: no amino acid identities with Act d 10 and Pru p 3 sequences.

Figure 3. Comparative skin testing and IgE detection using nAct c 10, nAct d 10, and nPru p 3 in 44 peach clinically allergic subjects tested positive for nPru p 3 on ISAC 103.

Panels A to I report correlation and concordance results for each paired allergen preparation and test as in the graphs. The Spearman r correlation coefficient and the Fisher's exact test have been used for statistical evaluations. Statistics are reported below each graph. IgE results have been obtained using ISAC microarray Exp96 and reported as kU/l; skin test have been obtained by measuring wheal areas and reported as mm² _. For graphical visualization needs on log scales, zero values for the skin test have been set at 0.1 mm², and at 0.01 kU/l for ISAC values.

Figure 4. IgE prevalence for nAct c 10, nAct d 10, rAra h 9, nArt v 3, rCor a 8, nMor n 3, nPru p 3 on 431 sera.

Prevalence has been calculated on 431 patients having at least one positive IgE test for one of the LTP under study. Asterisk marks LTPs tested on ISAC 103; remaining have been tested on ISAC Exp96. Statistical comparative evaluation has been performed using the χ² test to evaluate the overall differences among all seven values. Pair-wise comparison has been performed using the Tukey post hoc test. “p” values, for paired letters in italics on top of the two corresponding bars, were as follows: a: p = 0.00015; b, e, g, h, j, k, l, m: p<0.0001; c: p = 0.0076; d: p = 0.00067; f: p = 0.0021; i: p = 0.045.

Figure 5. Venn diagram representation of positive IgE results for selected LTPs.

LTP specific IgE have been determined using ISAC Exp96, excepting for nArt v 3 and rCor a 8where ISAC 103 has been used. Due to the highly similar behavior of the two kiwifruit LTPs, panel from B to E show how the other four LTPs behave compared to nAct d 10 and nPru p 3. Absolute and relative IgE prevalence are given for each combination on graphs as follows: Panel A: nAct c 10, nAct d 10, and nPru p 3; Panel B: nAct d 10, nPru p 3, rAra h 9; Panel C: nAct d 10, nPru p 3, nArt v 3; Panel D: nAct d 10, nPru p 3, rCor a 8; Panel E: nAct d 10, nPru p 3, nMor n 3.

Figure 6. Unsupervised two-way hierarchical clustering analysis of 431 subjects tested for IgE on the seven LTPs.

Logarithmic IgE value distribution has been used to generate the heat map. Subjects had at least one IgE-positive result to one LTP under study. LTP are reported on the y-axis, subjects on the x-axis with their respective relative distribution scales. Black to dark red scale corresponds to IgE values from negative to strongly positive. Color key legend gives an approx idea of visualized IgE values.

Figure 7. IgE values box-plot representation considering median values and 5–95 percentile distributions.

IgE value distribution is plotted for positive values. Kruskal-Wallis One-Way ANOVA by Ranks test gave a statistically significant p value (<0.0001). The pair-wise comparison with a Mann-Whitney test with Bonferroni correction applied to LTP IgE results gave p values, considering paired letters in italics on top of the two corresponding bars, as follows,: a: p = 0.00218; b, d, h: p<0.0001; c: p = 0.00014; e: p = 0.00093; f: p = 0.00079; g: p = 0.03402.

Figure 8. IgE value correlations for paired LTPs. All 1,003 subjects have been plotted in each graph.

Flags A, B, C, D, E, F, G, H, I in figure 8 indicate them as parts of the results shown also in figures 9, and 10. Consecutive letters have been used on purpose through the three figures. The * marks IgE detection performed on ISAC 103, remaining IgE determinations have been obtained by ISAC Exp96. For graphical visualization needs on log scales, zero values for ISAC testing have been set to 0.01 kU/l. The Spearman r correlation coefficient has been calculated and the χ² test has been used for statistical purposes. Statistics are reported below each graph.

Figure 9. IgE value correlations for paired LTPs. All 1,003 subjects have been plotted in each graph.

Flags J, K, L, M, N, O, P, Q, R in figure 9 indicate them as parts of the results shown also in figures 8, and 10. Consecutive letters have been used on purpose through the three figures. The * marks IgE detection performed on ISAC 103, remaining IgE determinations have been obtained by ISAC Exp96. For graphical visualization needs on log scales, zero values for ISAC testing have been set to 0.01 kU/l. The Spearman r correlation coefficient has been calculated and the χ² test has been used for statistical purposes. Statistics are reported below each graph.

Figure 10. IgE value correlations for paired LTPs. All 1,003 subjects have been plotted in each graph.

Flags S, T, U, V in figure 10 indicate them as parts of the results shown also in figures 8, and 9. Consecutive letters have been used on purpose through the three figures. The * marks IgE detection performed on ISAC 103, remaining IgE determinations have been obtained by ISAC Exp96. For graphical visualization needs on log scales, zero values for ISAC testing have been set to 0.01 kU/l. The Spearman r correlation coefficient has been calculated and the χ² test has been used for statistical purposes. Statistics are reported below each graph.

Figure 11. Single Point Highest Inhibition Achievable assay (SPHIAa) for LTP IgE inhibition.

IgE values on graphs are reported as percent inhibition. IgE values for each serum in each graph representation are given in brackets. The three inhibitors are indicated in the X axis, whereas on top of each graph the LTP immobilized on the microarray whose IgE binding has been inhibited is reported. Statistical evaluations have been applied to the three series using the Kruskal-Wallis One-Way ANOVA by Ranks , followed by the post hoc test of Nemenyi-Damico-Wolfe-Dunn comparing paired series. Statistics are as follows: Panel A - nAct c 10: nAct c 10 Median = 100%; Range 83.91–100%; nAct d 10 Median = 100%; Range 97.92–100%; nPru p 3 Median = 100%; Range 78.03–100%; Kruskal-Wallis = n.s.s.; Panel B - nAct d 10: nAct c 10 Median = 100%; Range 84.27–100%; nAct d 10 Median = 100%; Range 98.67–100%; nPru p 3 Median = 100%; Range 80.42–100%; Kruskal-Wallis = n.s.s.; Panel C - rAra h 9: nAct c 10 Median = 79.62%; Range 50.18– 97.85%; nAct d 10 Median = 85.39%; Range 51.00–100%; nPru p 3 Median = 100%; Range 88.05–100%; Kruskal-Wallis p = 0.00061; nAct c 10 vs nAct d 10 = n.n.s.; nAct c 10 vs nPru p 3 p = 0.00019; nAct c 10 vs nPru p 3 p = 0.0055; Panel D - nArt v 3: nAct c 10 Median = 75.45%; Range 0–100%; nAct d 10 Median = 79.04%; Range 0–100%; nPru p 3 Median = 100%; Range 90.80–100%; Kruskal-Wallis p = 0.0012; nAct c 10 vs nAct d 10 = n.n.s.; nAct c 10 vs nPru p 3 p = 0.00078; nAct c 10 vs nPru p 3 p = 0.0083; Panel E - rCor a 8: nAct c 10 Median = 95.83%; Range 32.44–100%; nAct d 10 Median = 85.79%; Range 0–100%; nPru p 3 Median = 100%; Range 80.01–100%; Kruskal-Wallis p = n.n.s.; Panel F - nMor n 3: nAct c 10 Median = 63.71%; Range 27.97–95.84%; nAct d 10 Median = 67.06%; Range 43.65–97.31%; nPru p 3 Median = 98.99%; Range 85.11–100%; Kruskal-Wallis p = 0.00031; nAct c 10 vs nAct d 10 = n.n.s.; nAct c 10 vs nPru p 3 p = 0.00035; nAct c 10 vs nPru p 3 p = 0.0011; Panel G - nPru p 3: nAct c 10 Median = 50.21%; Range 18.25–95.94%; nAct d 10 Median = 60.55%; Range 39.55–98.11%; nPru p 3 Median = 100%; Range 97.44–100%; Kruskal-Wallis p<0.0001; nAct c 10 vs nAct d 10 = n.n.s.; nAct c 10 vs nPru p 3 p<0.0001; nAct c 10 vs nPru p 3 p = 0.0003.

Figure 12. Biochemical, immunochemical, and clinical evaluation of LTP distribution in kiwifruit tissues.

Panel A, Left: RP-HPLC profiles of the pulp (upper part) and seed (lower part) protein extracts of gold kiwifruit tissues. The amount of loaded proteins was 1 mg. The arrow indicates the elution time of Act c 10. Panel A, Right: RP-HPLC profiles of the pulp (upper) and seed (lower) protein extracts of green kiwifruit tissues. The amount of loaded proteins was 0.3 mg. The arrow indicates the elution time of Act d 10. Panel B: Single Point Highest Inhibition Achievable assay (SPHIAa) using pulp and seed extract preparations from green and gold kiwifruits. Grey bars: total kiwifruit extracts; White bars: bars: kiwifruit pulp extracts; Black bars: kiwifruit seed extracts. Panel C: Skin test using nAct c 10, nAct d 10, and green kiwifruit seed and pulp preparations. Skin test wheal areas have been recorded and expressed in mm². The Mann-Whitney test applied to paired test result distributions gave p values as follows: Act c 10 vs Act d 10 p = n.n.s.; Act d 10 vs Seeds p = n.n.s.; Act d 10 vs Pulp p<0.004; Seeds vs Pulp p<0.017.

Figure 13. IgE and Skin test result evaluation comparing patients being tolerant or not to green kiwifruit ingestion.

NT = Green kiwifruit Non-Tolerant; T = Green kiwifruit Tolerant. Panel A: IgE values obtained testing serum samples for nAct c 10, nAct d 10, and nPru p 3 on ISAC Exp96. The Mann-Whitney test applied to tolerant versus non-tolerant patients for each test gave p values as follows: Act c 10-T vs Act c 10-NT p = n.n.s.; Act d 10-T vs Act d 10-NT p<0.05; Pru p 3-T vs Pru p 3-NT p = n.n.s. Panel B: Skin test wheal area values obtained for nAct c 10, nAct d 10, and nPru p 3. Skin test wheal areas have been recorded and expressed in mm². The Mann-Whitney test applied to tolerant versus non-tolerant patients for each test gave p values as follows: Act c 10-T vs Act c 10-NT p<0.02; Act d 10-T vs Act d 10-NT p<0.01; Pru p 3-T vs Pru p 3-NT p = n.n.s.