A New Immunodiagnostic Test for the Detection of Southern Hemisphere Fish Residues in Processed Foods

Home , Australian bonito, Eastern school whiting, Hyperoglyphe, Hyporhamphus australis, Salted squid, Scombriformes

Ji Liang

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy

School of Chemical Engineering Faculty of Engineering The University of New South Wales Sydney Australia

August 2018 THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet Surname or Family name: Liang First name: Ji Other name/s: Abbreviation for degree as given in the University calendar: PhD School: School of Chemical Engineering Faculty: Faculty of Engineering Title: A New Immunodiagnostic Test for the Detection of Southern Hemisphere Fish Residues in Processed Foods Abstract 350 words maximum: (PLEASE TYPE) Fish products are regulated for mandatory allergen labelling in many countries including Australia. However, the commercially available fish ELISA kits have been developed using parvalbumin (PAV) or fish protein extracts from northern hemisphere fish species. This study, therefore, aims to develop an immunodiagnostic test for the detection and quantification of (commercially important) southern hemisphere fish residues in processed foods, by developing antibodies with broad species specificity for southern hemisphere fish species.

In order to quantitatively estimate and rank the cross-reactivity between fish species, the cross-reactivity indices were developed using two model antibodies specific to fish parvalbumin (anti-cod PoAb and anti-carp MoAb). Compared with the crude extracts, improved correlation between the PAV content and the immunoreactivity of heated PAV was observed for both antibodies (i.e., R2=0.82 for the anti-cod PoAb and R2=0.55 for the anti-carp MoAb). There was a strong phylogenetic relationship, showing strong immunoreactivity amongst fish species from the order of Perciformes; and weak or no immunoreactivity amongst species from Salmoniformes, Ophiidiformes, Scombriformes, Scorpaeniformes, and Tetraodontiformes.

Polyclonal antibodies were developed using a heat-treated immunogen prepared from thirteen selected fish species. Two sensitive sandwich ELISAs, differing in their capture-detection antibody combinations, were developed. The (HM)S627#4 – (HM)RB#2 assay yielded an average detection limit of 0.08 μg/L; and the (HM)RB#2 – (HM)S627#4 assay yielded an average detection limit of 0.59 μg/L. The newly developed assays showed significantly improved detection of fish species from Salmoniformes, Ophidiiformes, and Scopaeniformes, which previously showed weak or no immunoreactivity.

The assay buffer was optimised with addition of various additives (e.g., glycerol, glycine and NaCl) to increase the protein recovery from < 10% to 85 – 122%. The spike and recovery study of SM PAV in four blank food matrices achieved good protein recovery of 87 – 117%, with satisfactory assay precision (<18%). A product survey of Thai-manufactured foods showed positive results in 50% of the products. Amongst the negative products were the fish sauce and the yellow curry paste, which did not show quantifiable fish residues, possibly due to the extensive protein hydrolysis.

Declaration relating to disposition of project thesis/dissertation I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only). ………………………...... ……….……………………...…….

Signature Witness Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research. FOR OFFICE USE ONLY Date of completion of requirements for Award: Originality Statement

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed......

Date...... COPYRIGHT STATEMENT

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed

Date ......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed

Date ......

Abstract

Fish products are regulated for mandatory allergen labelling in many countries including Australia. However, the four commercially available fish ELISA kits have been developed using parvalbumin (PAV) or fish protein extracts from northern hemisphere fish species. This study, therefore, aims to develop an immunodiagnostic test for the detection of (commercially important) southern hemisphere fish residues in processed foods, by developing antibodies with broad species specificity for southern hemisphere fish species.

(43%) showed a positive correlation (R2=0.74) between the PAV content in crude fish extracts and the immunoreactivity of PAV with the anti-cod PoAb; while no correlation between the same was established with the anti-carp MoAb.

The thermal treatment of the crude extracts, further improved the correlation between the PAV content and the immunoreactivity of heated PAV for both antibodies (i.e.,

R2=0.82 for the anti-cod PoAb and R2=0.55 for the anti-carp MoAb). There was a strong phylogenetic relationship, showing strong immunoreactivity amongst fish species from the order of Perciformes; and weak or no immunoreactivity amongst species from Salmoniformes, Ophiidiformes, Scombriformes, Scorpaeniformes, and

Tetraodontiformes.

The polyclonal antibodies were developed using a heat-treated immunogen prepared from thirteen selected fish species. Two sensitive sandwich ELISAs, differing in their capture-detection antibody combinations, were developed. The (HM)S627#4 –

(HM)RB#2 assay yielded an average detection limit (LLOD, determined as C15 (a i

concentration of fish protein produce 15% of colour relative to the maximum colour development) of 1.7 ± 0.9 μg/L; and the (HM)RB#2 – (HM)S627#4 assay yielded an average LLOD of 1.6 ± 0.4 μg/L. The detection limit calculated as the mean measure value of 10 blank sample replicates plus 3 times of the standard deviation (SD) of the mean value (LLOD’) for the (HM)S627#4 – (HM)RB#2 assay was 0.08 μg/L, and the

LLOD’ for the (HM)RB#2 – (HM)S627#4 assay was 0.59 μg/L. The (HM)S627#4 –

(HM)RB#2 assay showed a wider species specificity, ranging from 0% for swordfish to

686% for eastern school whiting. While the (HM)RB#2 – (HM)S627#4 assay exhibited a narrower species specificity ranging from 0.0% for swordfish to 122.2% for Spanish mackerel. The newly developed assays showed significantly improved detection of fish species from Salmoniformes, Ophidiiformes, and Scopaeniformes, which previously showed weak or no immunoreactivity.

The assay buffer was optimised with addition of various additives (e.g., glycerol, glycine and NaCl) to increase the protein recovery from less than 10% to 84.8 – 122.1%.

Food matrices (i.e., rice cake and corn flour) with high carbohydrate contents had detectable immunological responses with the (HM)RB#2 – (HM)S627#4 assay, due to the high affinity of the carbohydrate with the antibodies. This matrix effect could be eliminated by overnight storage when the carbohydrate contents become precipitates.

Three food products (i.e., salted pepper squid, chicken breast bites and mussel) among the thirty-four foods and ingredients showed cross-reactivity with the assay. The cross-reactivity could be eliminated by diluting the sample extracts to a non-detectable concentration, but in a range that does not compromise. The spike and recovery study of

SM PAV in four blank food matrices (i.e., almond coconut muesli, chicken corn soup, ham, and vegetable-lentil pie) achieved good protein recoveries: 88.1 – 112.2% for the intra-assay validation and 87.2 – 117.3% for the inter-assay validation, with satisfactory assay precisions (<18%).

A product survey of Thai-manufactured foods with possible fish residues showed positive results in 50% of the products tested. Amongst the negative products were the fish sauce and the yellow curry paste, which did not contain quantifiable fish residues possibly due to the extensive protein hydrolysis.

iii

Acknowledgments

I would like to express my most sincere gratitude to my dear supervisor Associate

Professor Nanju Alice Lee for her expertise, understanding, encouragement and tireless support throughout my Ph.D. study. She is always so thoughtful and patient. She has always been there whenever I needed her advices. I deeply thank her for all her contribution of time, ideas, and efforts to make my Ph.D. study such a joyful and memorable journey.

I would like to thank Dr. Sri Muralidharan for her great technical and mental support throughout my Ph.D. study.

I would like to thank my co-supervisor Professor Steve Taylor and Associate Professor

Joe Baumert for their support and valuable advices. I also would like to thank our collaborator Professor Andreas Lopata for providing the fish samples and constructive feedback on our publication.

I would like to thank Mr. Camillo Taraborelli for his kind technical assistance.

I would like to thank all my previous and current colleagues from the Food Science group: Yan Yee Poon, Johanna Rost, Jun Ma, Yiqing Zhao, RH Fitri Faradilla, Steffi

Angela, Kornelia Kaczmarska, Scott Markham, Xin Sun, Chatchaporn Uraipong, and

Huynh Tien Dat. I thank them for their support for my PhD study and their lovely friendship.

I would like to thank Australian Research Council (ARC) and the ARC Training Centre for Advanced Technologies in Food Manufacture (ARC ATFM) for the Ph.D. scholarship and the funding support of my project. I also would like to thank Food

Allergy Research and Resource Program (FARRP) for partially funding of the project.

Last but not least, I would like to especially thank my parents and my husband for their endless love and support throughout my study.

List of publication and presentations

Ji Liang, Chui Choo Tan, Steve L. Taylor, Joseph L. Baumert, Andreas L. Lopata & N. Alice Lee. 2017. Quantitative analysis of species specificity of two anti-parvalbumin antibodies for detecting southern hemisphere fish species demonstrating strong phylogenetic association. Food Chemistry, 237, 588-596.

Ji Liang, Sridevi Muralidharan, Steve L. Taylor, Joseph L. Baumert, Andreas L. Lopata & N. Alice Lee. Evaluation of the species-specificity of three IgG antibodies against the major fish allergen parvalbumin among southern hemisphere fish species. Food Allergen Management Symposium, 11-14 May 2015, Sydney, Australia. [Accepted for oral presentation]

Ji Liang, Sridevi Muralidharan, Steve L. Taylor, Joseph L. Baumert, Andreas L. Lopata & N. Alice Lee. Impacts of thermal processing on the immunoreactivity of parvalbumin from southern hemisphere fish species. Pacifichem, 15-20 December 2015, Hawaii, US. [Accepted for oral presentation]

Ji Liang, Shi Liu, Steve L. Taylor, Joseph L. Baumert, Andreas L. Lopata & N. Alice Lee. Selecting a suitable antibody for detection of southern hemisphere fish allergens. 1st International Conference on Food Analysis, 21-23 November 2016, Melbourne, Australia. [Accepted for oral presentation]

Ji Liang, Shi Liu, Steve L. Taylor, Joseph L. Baumert & N. Alice Lee. Developing a sensitive enzyme-linked immunosorbent assay (ELISA) for detection of southern hemisphere fish allergens. 2nd Food Allergen Management Symposium, 21-24 May 2017, Sydney, Australia. [Accepted for oral presentation]

List of Abbreviations

Abs Absorbance

ANOVA Analysis of variance

AU Absorbance Unit

BCA Bicinchoninicc acid

BSA Bovine serum albumin

CD Circular dichroism

CI Correlation index

(CM)R Rabbit anti-crude fish protein antibody

CV Coefficient of variation

DC Dendritic cell

DMSO Dimethyl sulfoxide

DTT Dithiothreitol

EDTA Ethylenediaminetetraacetic acid

ELISA Enzyme-linked immunosorbent assay

HCl Hydrochloric acid

(HM)G Goat anti-heat-treated mixed fish protein antibody

(HM)R Rabbit anti-heat-treated mixed fish protein antibody

(HM)S Sheep anti-heat-treated mixed fish protein antibody

HRP Horseradish peroxide

IgE Immunoglobulin E

IgG Immunoglobulin G

ISTD Isotoped-lablled internal standards kDa kilo Dalton

LC-MS-MS Liquid chromatography- tandem mass spectrometry

LLOD Lower limit of detection

vii

LOD Limit of detection

LOQ Limit of quantitation

LTQ-FT Linear ion trap-Fourier transform

MHC Major histocompatibility complex

MoAb Monoclonal antibody

MW Molecular weight

NaCl Sodium chloride

NCBI National Centre for Biotechnology Centre

OFC Oral food challenge

PAL Precautionary allergen labelling

PAV Parvalbumin

PBS Phosphate buffered saline

PCR polymerase chain reaction pI Isoelectric point

PoAb Polyclonal antibody

RO water Reverse osmosis water

SD Standard deviation

SDS-PAGE Sodium dodecyl sulphate-polyacrylamide gel electrophoresis

TCR T cell receptor

TMB 3,3'5,5'-tetramethylbenzidine

ULOD Upper limit of detection v/v Volume to volume ratio

VITAL Voluntary Incidental Trace Allergen Labelling w/v Weight to volume ratio

viii

Table of Contents

Abstract ...... i Acknowledgments ...... iv List of publication and presentations ...... vi List of Abbreviations ...... vii Table of Contents ...... ix List of Figures ...... xv List of Tables ...... xxi Chapter 1 Introduction ...... 1 1.1. Background ...... 1 1.2. Aim of study ...... 3 1.3. Objectives ...... 3 Chapter 2 Literature review ...... 5 2.1. Adverse reactions to food ...... 5 2.2. Food allergy ...... 6 2.2.1. General mechanism of IgE-mediated food allergic reactions ...... 6 2.2.2. Routes of exposure and sensitization ...... 8 2.2.3. Clinical manifestations ...... 9 2.2.4. Common food allergens in different geographic regions ...... 11 2.2.5. Management of food allergy ...... 13 2.2.5.1. Legislation about food allergen declaration ...... 13 2.2.5.2. Population-based food allergen thresholds and precautionary allergen labelling ...... 16 2.3. Fish allergy ...... 19 2.3.1. Overview of Fish ...... 19 2.3.1.1 Fish as an important animal protein source for human diet ...... 19 2.3.1.2. Fish proteins ...... 20 2.3.2. Fish allergy ...... 21 2.3.2.1. Introduction to fish allergy ...... 21 2.3.2.2. Prevalence of fish allergy ...... 23 2.3.2.3 Clinically important fish species ...... 27 2.3.2.4. Paralbumin (PAV) – the major fish allergen ...... 29 2.3.2.5. Protein structure of PAV ...... 31 ix

2.3.2.6. Molecular heterogeneity of PAV from different fish species ...... 33 2.3.2.7. PAV content relative to the allergenicity ...... 35 2.3.2.8. Minor fish allergens ...... 36 2.4. Current detection methodologies for fish allergen residues ...... 38 2.4.1. Overview ...... 38 2.4.2. ELISA for fish allergen detection ...... 41 2.4.2.1. Sandwich ELISA ...... 42 2.4.2.2. Competitive inhibition ELISA ...... 43 2.4.2.3. Commercial ELISA and ELISA reported in the literature for fish allergen detection ...... 45 2.4.3. Steps involved in the development of a sandwich ELISA for fish allergen detection ...... 56 2.4.3.1. Immunogen preparation-extraction and purification of PAV ...... 56 2.4.3.2. Production of antibodies ...... 57 2.4.3.3. Immunoassay development and optimisation ...... 58 2.4.3.4. Immunoassay validation ...... 62 2.5. The impacts of food processing on the immunoreactivity and allergenicity of fish allergens ...... 67 2.6. Conclusion ...... 69 Chapter 3 Quantitative analysis of species specificity of two anti-PAV antibodies for detecting southern hemisphere fish species demonstrating strong phylogenetic association ...... 71 3.1. Introduction ...... 71 3.2. Materials and methods ...... 72 3.2.1. Materials ...... 72 3.2.1.1. List of fish species ...... 72 3.2.1.2. Chemicals and reagents ...... 74 3.2.1.3. Buffers ...... 75 3.2.1.4. Instruments ...... 75 3.2.2. Methods ...... 76 3.2.2.1. Identification of fish species ...... 76 3.2.2.2. Soluble fish protein extraction ...... 76 3.2.2.3. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis x

(SDS-PAGE) ...... 76 3.2.2.4. Immunoblotting analysis ...... 77 3.2.2.5. PAV identification and relative quantification based on SDS-PAGE..... 77 3.2.2.6. Quantitative indirect ELISA ...... 79 3.2.2.7. Data analysis ...... 79 3.3. Results and discussion ...... 80 3.3.1. Protein profile of the crude fish protein extracts by SDS-PAGE and immunoblot analysis ...... 80 3.3.2. Indirect ELISA ...... 88 3.3.2.1. Qualitative indirect ELISA ...... 88 3.3.2.2. Quantitative indirect ELISA indicating correlation between PAV content and the immunoreactivity ...... 88 3.4. Conclusion ...... 93 Chapter 4 Impacts of thermal processing on the immunoreactivity of PAV with two anti-PAV antibodies ...... 95 4.1. Introduction ...... 95 4.2. Materials and methods ...... 96 4.2.1. Materials ...... 96 4.2.1.1. List of fish species ...... 96 4.2.1.2. Chemicals and reagents ...... 96 4.2.1.3. Instruments ...... 96 4.2.2. Methods ...... 97 4.2.2.1. Soluble fish protein extraction ...... 97 4.2.2.2. Heat treatment of crude fish protein extracts ...... 97 4.2.2.3. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis (SDS-PAGE) ...... 97 4.2.2.4. Immunoblotting analysis ...... 98 4.2.2.5. Thermally stable PAV identification of relative quantification based SDS-PAGE ...... 98 4.2.2.6. Quantitative indirect ELISA ...... 98 4.2.2.7. Data analysis ...... 98 4.2.2.8. Structure investigation of the fish protein extracts by fluorescence spectroscopy ...... 98

4.3. Results and discussion ...... 99 4.3.1. Protein profile of the heated fish protein extracts by SDS-PAGE and immunoblot analysis ...... 99 4.3.2. Quantitative indirect ELISA ...... 107 4.3.3. Heat-induced protein structural changes of selected fish species assessed by fluorescence analysis ...... 111 4.4. Conclusion ...... 115 Chapter 5 Development of sandwich ELISAs with broad specificity for southern hemisphere fish allergen detection ...... 116 5.1. Introduction ...... 116 5.2. Materials and methods ...... 116 5.2.1. Materials ...... 116 5.2.1.1. List of fish species ...... 116 5.2.1.2. Chemicals and Reagents ...... 118 5.2.1.3. Buffers ...... 119 5.2.1.4. Instruments ...... 119 5.2.2. Methods ...... 120 5.2.2.1. Identification of fish species ...... 120 5.2.2.2. Soluble fish proteins extraction ...... 120 5.2.2.3. Heat treatment of crude protein extracts ...... 120 5.2.2.4. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis (SDS-PAGE) ...... 120 5.2.2.5. Preparation of crude-mixed and heat-treated mixed immunogens ...... 121 5.2.2.6. Production of polyclonal anti-PAV antibodies ...... 121 5.2.2.7. Antisera titration ...... 121 5.2.2.8. Purification of polyclonal antibodies ...... 122 5.2.2.9. Immunoblotting analysis ...... 123 5.2.2.10. ELISA calibration curve – indirect sandwich ELISA ...... 123 5.2.2.11. Checkerboard ELISA ...... 124 5.2.2.12. Cross-reactivity by a checkerboard ELISA...... 126 5.2.2.13. Cross-reactivity study by calibration curves ...... 126 5.2.2.14. Data analysis ...... 127 5.2.2.15. LC-MS-MS analysis...... 127 xii

5.3. Results and discussion ...... 129 5.3.1. The selection of fish species for mixed immunogen preparation ...... 129 5.3.2. Antisera titration ...... 130 5.3.3. Checkerboard ELISA ...... 133 5.3.4. Indirect sandwich ELISA – calibration curve development ...... 142 5.3.5. Immunoblotting analysis of polyclonal antibodies ...... 145 5.3.6. Cross-reactivity by a checkerboard ELISA ...... 151 5.3.7. Cross-reactivity (species specificity) study using the full calibration curves ...... 153 5.3.8. Intra-assay precision of important analytical parameters...... 156 5.3.9. Identification of the 36 kDa proteins from the selected fish species by LC-MS-MS ...... 159 5.4. Conclusion ...... 162 Chapter 6 Validation of the indirect sandwich ELISA with optimised assay buffer for fish protein detection ...... 164 6.1. Introduction ...... 164 6.2. Materials and methods ...... 165 6.2.1 Materials ...... 165 6.2.1.1. Fish species and commercial food samples ...... 165 6.2.1.2. Chemicals and reagents ...... 165 6.2.1.3. Buffers ...... 166 6.2.1.4. Instruments ...... 167 6.2.2. Methods ...... 167 6.2.2.1. Soluble fish protein extraction ...... 167 6.2.2.2. Heat treatment of crude fish protein extracts ...... 167 6.2.2.3. Separation of PAV from heated fish protein extracts by ammonium sulphate precipitation ...... 167 6.2.2.4. Soluble protein extraction of blank food matrices and Thai-imported fish products ...... 168 6.2.2.5. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis (SDS-PAGE) ...... 169 6.2.2.6. Immunoblot analysis ...... 169 6.2.2.7. Indirect sandwich ELISA ...... 169

xiii

6.2.2.8. Spike and recovery studies ...... 170 6.2.2.9. Circular dichroism (CD) analysis and fluorescence analysis of partially purified Spanish mackerel PAV ...... 170 6.2.2.10. Data analysis ...... 171 6.3 Results and discussion ...... 171 6.3.1. Purification of PAV from the heated crude fish protein extracts by ammonium sulphate precipitation ...... 171 6.3.2. Cross-reactivity of the partially purified PAVs related to the heat-treated mixed immunogen determined by the (HM)RB#2 – (HM)S627#4 assay ...... 173 6.3.3. The preliminary spike and recovery study of SM PAV in the standard diluent buffer ...... 174 6.3.4. Protein structural investigation of SM PAV by fluorescence and CD analysis ...... 175 6.3.5. Effects of different additives on assay performances...... 177 6.3.6 Study of the effect of blank food matrices on the performances of the assay and the potential cross-reactivity of non-fish foods with the assay ...... 183 6.3.7 Spike and recovery of SM PAV in blank food matrices ...... 186 6.3.8. A pilot survey of Thai-imported products containing fish in the ingredient list or on PAL statement ...... 188 6.4. Conclusion ...... 190 Chapter 7 Conclusion and future work ...... 193 7.1. Conclusion ...... 193 7.2. Future work ...... 198 List of references ...... 200 Appendix ...... 218

xiv

List of Figures

Figure 2.1 Classification of adverse reactions to food modified from Dean, (2000)...... 5

Figure 2.2 The general mechanism of IgE-mediated allergic reactions...... 7

Figure 2.3 Phylogeny of the ray-fin fish species (Understanding Evolution, 2017)...... 20

Figure 2.4 Crystal structure of PAV from carp (Permyakov and Kretsinger, 2010)...... 31

Figure 2.5 Phylogenetic relations of thirty-eight PAVs (Saptarshi et al., 2014)...... 35

Figure 2.6 The workflows of a direct (a) and an indirect (b) sandwich ELISA...... 43

Figure 2.7 The workflows of a direct (a) and an indirect (b) competitive ELISA...... 45

Figure 3.1 Protein profiles of the crude fish protein extracts separated by SDS-PAGE with Coomassie brilliant blue staining. The red rectangles indicate the protein bands that were recognized by the two anti-PAV antibodies collectively in the immunoblot...... 81

Figure 3.2 Immunoblot analysis of the crude fish protein extracts with the anti-cod

PoAb...... 83

Figure 3.3 Immunoblot analysis of the crude fish protein extracts with the anti-carp

MoAb...... 84

Figure 3.4 The correlation between relative quantity of PAV estimated from SDS-PAGE and absorbance values of ELISA for 1 μg crude protein using (a) anti-cod PoAb and (b) anti-carp MoAb ...... 88

Figure 3.5 The correlation between relative quantity of PAV estimated by SDS-PAGE and cod-PAV equivalent concentrations by ELISA using the anti-cod PoAb for (a) quantifiable samples with the anti-cod PoAb (R2 = 0.30, y = 0.80x + 0.13) and (b) quantity-related samples with the anti-cod PoAb (R2 = 0.74, y = 1.13x – 0.08)...... 91

Figure 3.6 The correlation between relative quantity of PAV estimated by SDS-PAGE and cod-PAV equivalent concentrations by ELISA using the anti-carp MoAb for quantifiable samples with the anti-carp MoAb (R2 = 0)...... 92

Figure 4.1 Protein profiles of the heated fish protein extracts separated by SDS-PAGE xv

with Coomassie brilliant blue staining. The red rectangles indicate the PAV bands that were recognized by the two anti-PAV antibodies collectively in the immunoblot...... 100

Figure 4.2 Immunoblot analysis of the heated fish protein extracts with the anti-cod

PoAb...... 102

Figure 4.3 Immunoblot analysis of the heated fish protein extracts with the anti-carp

MoAb...... 103

Figure 4.4 The correlation between relative quantity of thermally stable PAV estimated by SDS-PAGE and cod-PAV equivalent concentrations by ELISA using the anti-cod

PoAb for (a) quantifiable samples with the anti-cod PoAb (R2 = 0.60, y = 1.02x – 0.62) and (b) quantity-related samples with the anti-cod polyclonal antibody (R2 = 0.82, y =

1.02x – 0.22)...... 110

Figure 4.5 The correlation between relative quantity of thermally stable PAV estimated by SDS-PAGE and cod-PAV equivalent concentrations by ELISA using the anti-carp

MoAb for quantifiable samples with the anti-carp MoAb (R2 = 0.55, y = 1.4x + 0.17).

...... 111

Figure 4.6 Fluorescence emission spectra of crude protein extracts (blue line), heated protein extracts with EDTA (red line) and heated protein extracts without EDTA (green line) for (a) Australian pilchard, (b) Atlantic salmon, (c) rainbow trout, (d) sea mullet, (e) pink ling, (f) barramundi (g) coral trout, (h) pink snapper, (i) sand whiting, (j) Spanish mackerel, (k) yellowfin tuna and (l) tiger flathead...... 113

Figure 5.1 Protein profiles of (a) crude fish protein extracts and (b) heated protein extracts for the immunogen preparation. The 36 kDa bands in red rectangles were chosen for LC-MS-MS analysis...... 130

Figure 5.2 Titration curves of (a) (HM)RA & (b) (HM)RB antisera to heated-mixed fish protein, and (c) (CM)RC & (d) (CM)RD to crude-mixed fish protein for (●) first bleed,

(■) second bleed, (▲) third bleed, (▼) fourth bleed, (♦) fifth bleed (terminal bleed). 132

Figure 5.3 Titration curves of (a) (HM)G (b) (HM)S606 and (c) (HM)S627 antisera to

xvi

heated-mixed fish protein for (●) first bleed, (■) second bleed, (▲) third bleed, (▼) fourth bleed (terminal bleed)...... 133

Figure 5.4 Calibration curves development using (●) (HM)RB#2 – (HM)G#1 1:200, (■)

(HM)RB#2 – (HM)G#2 1:200, (▲) (HM)RB#3 – (HM)G#3 1:200, (▼) (HM)S627#2 –

(HM)RB#2 1:500, (♦) (HM)S627#3 – (HM)RB#2 1:500, (○) (HM)S627#4 –

(HM)RB#2 1:500 and (◊) (HM)RB#2 – (HM)S627#4 1:500...... 143

The results were expressed as mean ± standard deviation (n = 2)...... 144

Figure 5.5 Immunoblotting analysis of heated fish extracts with (HM)RB#2...... 147

Figure 5.6 Immunoblotting analysis of heated fish extracts with (HM)G#3...... 148

Figure 5.7 Immunoblotting analysis of heated fish extracts with (HM)S627#4...... 149

Figure 5.8 The cross-reactivity studies screening of (a) (HM)S627#2 – (HM)RB#2, (b)

(HM)S627#3 – (HM)RB#2, (c) (HM)S627#4 – (HM)RB#2, (d) (HM)RB#2 –

(HM)S627#4 with 37 fish species. The middle solid line represents 100 %R; the (∙∙∙∙) dotted lines represent 120 or 80 %R; and the (----) dotted lines represent 150 or

50 %R. %Abs(100) was used for (a) (HM)S627#2 – (HM)RB#2, (b) (HM)S627#3 –

(HM)RB#2, and %Abs(50) was used for (c) (HM)S627#4 – (HM)RB#2, (d) (HM)RB#2

– (HM)S627#4...... 152

Figure 5.9 The control chart plots of the (●) C80, (■) C50 and (▲) C15 values from 15 assay replicates for (a) (HM)S627#4 – (HM)RB#2 and (b) (HM)RB#2 – (HM)S627#4.

The middle solid lines represent the mean values and the dotted lines represent 95% confidence intervals...... 158

Figure 6.1 Protein profiles of the partially purified PAVs separated by SDS-PAGE with

Coomassie brilliant blue staining...... 172

Figure 6.2 Immunoblot analysis of the partially purified PAVs with the rabbit antibody

(HM)RB#2...... 173

Figure 6.3 Immunoblot analysis of the partially purified PAVs with the sheep antibody

(HM)S627#4...... 173

xvii

Figure 6.4 The far UV CD analysis of SM PAV (blue line) and native cod PAV (red line).

...... 176

Figure 6.5 Fluorescence analysis of SM PAV (blue line) and native cod PAV (red line).

...... 177

Figure 6.6 Calibration curves of the (HM)RB#2 – (HM)S627#4 assay prepared in (●) 1%

BSA-PBS-T; (■) 10% (v/v) glycerol/1% BSA-PBS-T; (▲) 20% (v/v) glycerol/1%

BSA-PBS-T; (▼) 40% (v/v)/1% BSA-PBS-T with SM PAV...... 178

Figure 6.7 Calibration curves of the (HM)RB#2 – (HM)S627#4 assay prepared in (●) 1%

BSA-PBS-T; (■) 0.1 M glycine/1% BSA-PBS-T; (▲) 0.5 M glycine/1% BSA-PBS-T;

(▼) 1.0 M glycine/1% BSA-PBS-T with SM PAV...... 180

Figure 6.8 Calibration curves of the (HM)RB#2 – (HM)S627#4 assay were prepared in

(●) 1% BSA-PBS-T; (■) 0.5 M NaCl/1% BSA-PBS-T; (▲) 1.0 M NaCl/1%

BSA-PBS-T; (▼) 1.5 M NaCl/1% BSA-PBS-T with SM PAV...... 181

Figure 6.9 Protein profiles of selected food samples separated by SDS-PAGE...... 186

Figure 6.10 Immunoblot analysis of selected food samples with (a) (HM)RB#2 and (b)

(HM)S627#4...... 186

Appendix Figure 1 Illustration of tissue position obtained from fish samples...... 218 Appendix Figure 2 Calibration curve of purified cod PAV for PAV relative quantification on SDS-PAGE...... 218

Appendix Figure 3 Calibration curve of quantitative ELISA using the anti-cod PoAb for

Chapter 3. The calibration curve was fitted to the sigmoidal dose-response (variable slope) model using GraphPad Prism® v.7.0...... 219

Appendix Figure 4 Calibration curve of quantitative ELISA using the anti-carp MoAb for Chapter 3. The calibration curve was fitted to the sigmoidal dose-response (variable slope) model using GraphPad Prism® v.7.0...... 219

Appendix Figure 5 Time course study of heat treatment of barramundi protein extract; protein profile of soluble proteins after heat treatment for 2 min (lane 1), 5 min (lane 2),

xviii

10 min (lane 3), 20 min (lane 4), 30 min (lane 5); protein profile of precipitates after heat treatment for 2 min (lane 6), 5 min (lane 7), 10 min (lane 8), 20 min (lane 9), 30 min (lane 10); land 11 – barramundi crude protein extract...... 220

Appendix Figure 6 Fluorescence spectra of the selected raw and heated fish protein extracts for different heating time intervals (20 min, 30 min, 45 min and 60 min)...... 220

Appendix Figure 7 Protein profiles of the heated fish protein extracts of the selected fish species: pink ling after heat treatment for 20 min (lane 1), 30 min (lane 2), 45 min (lane

3), 60 min (lane 4); barramundi after heat treatment for 20 min (lane 5), 30 min (lane 6),

45 min (lane 7), 60 min (lane 8); school whiting after heat treatment for 20 min (lane 9),

30 min (lane 10), 45 min (lane 11), 60 min (lane 12)...... 221

Appendix Figure 8 Calibration curve of quantitative ELISA using the anti-cod PoAb for

Chapter 4. The calibration curve was fitted to the sigmoidal dose-response (variable slope) model using GraphPad Prism® v.7.0...... 222

Appendix Figure 9 Calibration curve of quantitative ELISA using the anti-carp MoAb for Chapter 4. The calibration curve was fitted to the sigmoidal dose-response (variable slope) model using GraphPad Prism® v.7.0...... 222

Australian pilchard, Atlantic salmon, rainbow trout, sea mullet, sea garfish, pink ling, blue cod and blue-eye trevalla determined by the (HM)S627#4 – (HM)R#2 assay. .... 223

Appendix Figure 11 The cross-reactivity of the heat-treated mixed immunogen with fish species including silver gemfish, yellowfin tuna, red emperor, sand whiting, Spanish mackerel, silver perch, yellowfin bream, yellowtail kingfish, yellowtail scad, yellowbelly flounder, Australian bonito, swordfish, eastern red scorpionfish, dusky flathead, ocean perch, red gurnard, tiger flathead, ocean jacket determined by the

(HM)S627#4 – (HM)R#2 assay...... 224

xix

Appendix Figure 12 The cross-reactivity of the heat-treated mixed immunogen with fish species including Australian pilchard, Atlantic salmon, John dory, bight redfish, orange roughy, sea mullet, sea garfish, pink ling, blue cod, blue-eye trevalla, coral trout, crimson snapper, eastern school whiting, jewfish, pink snapper, red emperor, sand whiting, Spanish mackerel, silver perch, yellowfin bream, yellowtail kingfish, yellowtail scad, yellowbelly flounder and Australian bonito determined by the

(HM)R#2 – (HM)S627#4 assay...... 225

Appendix Figure 13 The cross-reactivity of the heat-treated mixed immunogen with fish species including swordfish, eastern red scorpion fish, dusky flathead, ocean perch, red gurnard, ocean jacket, mahi mahi, silver gemfish, yellowfin tuna, Murray cod, tiger flathead, pink snapper, rainbow trout and barramundi determined by the (HM)R#2 –

(HM)S627#4 assay...... 226

Appendix Figure 14 Cross-reactivity of the heat-treated mixed immunogen with partially purified PAVs of barramundi, Murray cod, Atlantic salmon, rainbow trout, sea mullet, pink snapper, sand whiting, Spanish mackerel, Australian pilchard, coral trout, pink ling and tiger flathead determined by the (HM)R#2 – (HM)S627#4 assay...... 227

List of Tables

Table 2.1 The main clinical symptoms associated with the IgE-mediated food allergies

...... 10

Table 2.2 Foods listed for mandatory declaration in different countries (Ebisawa et al.,

2017; Gendel, 2012; Taylor and Hefle, 2006; van Hengel, 2007) ...... 15

Table 2.3 Estimated population-based thresholds (reference doses) of the main allergenic foods (Allen et al., 2014a) ...... 18

Table 2.4 Occurrence of symptoms of fish allergy from the Philippines, Singapore, and

Thailand (Connett et al., 2012) ...... 23

Table 2.5 Prevalence of fish allergy among different graphic regions ...... 25

Table 2.6 List of most prevalent fish species from selected epidemiology studies of fish allergy ...... 28

Table 2.7 The nineteen peptide biomarkers for the detection of fish β-PAV ...... 40

Table 2.8 A summary of the performance of ELISAs reported in the literature for the fish allergen detection ...... 49

Table 2.9 The commercial sandwich ELISA test kits for the fish allergen detection ..... 55

Table 3.1 The list of fish species analyzed in this Chapter ...... 73

Table 3.2 The calculated MW of PAV bands detected by the anti-cod PoAb and the anti-carp MoAb in fish protein extracts ...... 87

Table 4.1 The calculated MW of thermally stable PAV bands detected by the anti-cod

PoAb and the anti-carp MoAb in the heated fish protein extracts ...... 104

Table 4.2 The relative quantity of thermally stable PAV estimated by SDS-PAGE and the quantitative ELISA analysis of heated fish protein extracts using anti-cod PoAb

(ELISA-A) and using anti-carp MoAb (ELISA-B) ...... 108

Table 4.3 The summary of maximum wavelength (λmax) and maximum intensity (Imax) of the fluorescence emission spectra for 12 selected fish species ...... 114 xxi

Table 5.1 The list of fish species analysed in this Chapter ...... 117

Table 5.2 Gradient of mobile phase B for liquid chromatography ...... 128

Table 5.3 The titre values of rabbit antisera against crude-mixed/heated-mixed fish protein ...... 132

Table 5.4 The titre values of antisera heated-mixed fish proteins, raised in goat and sheep ...... 133

Table 5.5 Performances of the capture-antibody antibody combinations with (HM)RA#2 and (HM)RB#2 as the capture antibodies ...... 136

Table 5.6 Performances of the capture-detection antibody combinations with

(CM)RC#3 and (CM)RD#3 as the capture antibodies ...... 137

Table 5.7 Performances of the capture-antibody antibody combinations with (HM)G#1 and (HM)G#2 as the capture antibodies...... 138

Table 5.8 Performances of the capture-antibody antibody combinations with

(HM)S606#2 and (HM)S606#3 as the capture antibodies ...... 139

Table 5.9 Performances of the capture-detection antibody combinations with

(HM)S627#2 and (HM)S627#3 as the capture antibodies ...... 140

Table 5.10 Performances of the capture-detection antibody combinations with

(HM)S627#4, (HM)RB#2 and (HM)G#4 as the capture antibodies ...... 141

Table 5.11 The summary table of analytical parameters for the calibration curves developed by the seven capture-detection antibody combinations ...... 144

Table 5.12 The calculated MW of thermally stable PAV bands in the heated fish protein detected by (HM)RB#2 and (HM)S627#4 ...... 150

Table 5.13 The cross-reactivity of 37 fish species related to the heat-treated immunogen determined by the assays (HM)S627#4 – (HM)R#2 and (HM)RB#2 – (HM)S627#4 . 155

Table 5.14 A summary of the analytical parameters of the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay based on data collected from fifteen analyses performed on different days ...... 158

xxii

Table 5.15 Identification of the 36 kDa protein band of crude fish extracts from five fish species by LC-MS-MS ...... 160

Table 5.16 Identification of the 36 kDa protein band of heated fish extracts from five fish species by LC-MS-MS ...... 161

Table 6.1 The cross-reactivity of the purified PAVs relative to the heat-treated immunogen determined by the (HM)RB#2 – (HM)S627#4 assay ...... 174

Table 6.2 Recovery of fish protein of Spanish mackerel in 1% BSA-PBS-T ...... 175

Table 6.3 Analytical parameters of the calibration curves prepared in three concentrations of glycerol, SM PAV as the calibration standard ...... 179

Table 6.4 One-way ANOVA analysis of the calibration curves prepared in three concentrations of glycerol, SM PAV as the calibration standard ...... 179

Table 6.5 Analytical parameters of the calibration curves prepared in three concentrations of glycine, SM PAV as the calibration standard ...... 180

Table 6.6 One-way ANOVA analysis of the calibration curves prepared in three concentrations of glycine, SM PAV as the calibration standard ...... 180

Table 6.7 Analytical parameters of the calibration curves prepared in three concentrations of NaCl, SM PAV as the calibration standard ...... 181

Table 6.8 One-way ANOVA analysis of the calibration curves prepared in three concentrations of NaCl, SM PAV as the calibration standard ...... 182

Table 6.9 Recovery of SM PAV in three buffer solutions ...... 182

Table 6.10 Overview of thirty-four different foods tested for matrix effect and cross-reactivity with the (HM)RB#2 – (HM)S627#4 assay ...... 185

Table 6.11 Intra-assay recovery of SM PAV in four different food matrices ...... 187

Table 6.12 Inter-assay recovery of SM PAV in four different food matrices ...... 187

Table 6.13 Determination of fish protein concentration in ten Thai-imported fish products using the (HM)RB#2 – (HM)S627#4 assay ...... 190

xxiii

Chapter 1 Introduction

1.1. Background

Fish, as a valuable meat source in human diet for high-quality proteins, unsaturated fatty acids, vitamins and minerals (Arino et al., 2005), is one of the eight food groups accounting for the majority (> 90%) of IgE-mediated food allergies (Bousquet et al.,

1998; Taylor and Hefle, 2001). Unlike milk and egg allergies with evidence that children showed high frequencies to develop tolerance (Gupta et al., 2013; Wood, 2003); fish allergy is known to be a life-long health issue (Sharp and Lopata, 2014). Fish allergy induces high proportion of severe reactions and is the primary cause of death related to food allergy in the emergency rooms in Australia (Schulkes et al., 2014; Sharp and Lopata, 2014). The medical advice usually given to fish allergic patients is strict avoidance of all fish species due to a high rate of clinical cross-reactivity between different fish species (Sicherer et al., 2004).

To ensure the safety and improve the diversity of food choice for fish allergic consumers, fish added as an ingredient is regulated by mandatory labelling on pre-packaged foods in many countries worldwide, including Australia. Analytics is a critical and integral part of an effective food allergen management. The current detection methods available for fish allergen include polymerase chain reaction (PCR), mass spectrometry (MS) and enzyme-based immunoassays methods (Lee, 2012).

Analytical challenges with PCR is that the detection of a specific DNA fragment does not necessarily guarantee the presence of allergenic proteins in the foods (Kirsch et al.,

2009); and for MS, is that the method development is still an ongoing effort.

Complicated method development, developed methods not directly transferable between instruments, sample preparation methods, methods of quantification, as well as requirement for the expensive and delicate equipment and highly trained personnel still

hinder the implementation of MS methods for routine food allergen analysis in food laboratories. Enzyme-Linked Immunosorbent Assay (ELISA) is still most commonly used method for food industry and regulatory bodies to detect and quantigy allergen residues in food products.

An ELISA offers rapid, sensitive and specific analysis of food allergens with high throughput capability (Lee and Sun, 2016). Most of the ELISAs reported in the literature (Cai et al., 2013; Chen and Hsieh, 2014; Fæste and Plassen, 2008; Gajewski et al., 2009; Shibahara et al., 2013; Weber et al., 2009) and the four commercially available fish ELISA test kits have been developed primarily for the detection of fish residues from northern hemisphere fish species. Additionally, the commercial test kits have not been validated for the detection of residues from southern hemisphere fish species, thus not particular useful for Australian food industry. Therefore, there is an urgent need for developing a sensitive ELISA that can target detection of most commercially important fish species from the southern hemisphere and Australia.

One significant technical challenge of fish allergen detection is the molecular diversity

(Saptarshi et al., 2014) and the quantity diversity (Griesmeier et al., 2010b; Kobayashi et al., 2016d; Kobayashi et al., 2016b; Liang et al., 2017; Van Do et al., 2005) of parvalbumin (PAV), the main allergen in fish. Most of the quantitative fish ELISAs reported in the literature (Cai et al., 2013; Fæste and Plassen, 2008; Shibahara et al.,

2013; Weber et al., 2009) demonstrated high assay sensitivity and good assay precision, but their antibodies, showed limitations in detecting all commercially important species, such as tuna (Thunnus spp.), swordfish (Xiphias gladius), rainbow trout (Oncorhynchus mykiss), flounder (Paralichthys spp.), Japanese eel (Anguilla japonica) and European eel (Anguilla anguilla). This limitation rooted from that the antibodies were raised against one PAV isoform of one selected fish species rather than many fish species. On the other hand, an anti-barramundi antibody developed by Sharp et al. (2015), which

was raised against two most distantly related barramundi PAV isoforms, showed a broader species specificity in qualitatively detecting thirty-five crude fish samples and thirty-four heat-treated fish samples.

Therefore, developing suitable antibodies with broad species specificity to commercially important southern hemisphere fish species for the development of more effective fish immunodiagnostic test still remains a crucial task for the fish allergen management in processed foods.

1.2. Aim of study

The project, therefore, aims to develop an immunodiagnostic test with improved quantification capability for (commercially important) southern hemisphere fish residues in processed foods, by developing antibodies with broad species specificity.

1.3. Objectives

To achieve the aim of this project, five specific objectives were established as outlined below.

1. To develop an analytical approach to statistically estimate the cross-reactivity index

for ranking different fish species, for the purpose of differentiating between fish

species showing high immunoreactivity and fish species showing low

immunoreactivity (Chapter 3).

2. To estimate the cross-reactivity of the two model anti-PAV antibodies in detecting

southern hemisphere fish species using the developed method and rank their

detectability (Chapter 3).

3. To investigate the effects of thermal treatment on the molecular modification and

immunoreactivity of PAV of southern hemisphere fish species, and their subsequent

cross-reactivity index (Chapter 4). 3

4. To raise polyclonal antibodies with broad species specificity by selecting fish

species with diverse immunoreactivity for the immunogen preparation (Chapter 5).

To develop and format a sensitive sandwich ELISA (Chapter 5) with optimised

buffer composition(s) specific for PAV extraction to achieve an optimal protein

recovery (i.e., 50 and 150%) (Chapter 6)

5. To conduct a pilot survey on precautionary allergen labels (PAL statements) of

Thai-imported products for fish residue. (Chapter 6).

Chapter 2 Literature review

2.1. Adverse reactions to food

Adverse reactions to food are any unpleasant reactions caused by ingestion of food or food additives (Sampson and Leung, 2011). Figure 2.1 illustrates the classification of adverse reactions to food modified from Dean (2000), which classified into toxic and non-toxic reactions. Food allergy belongs to the non-toxic reaction of food hypersensitivity.

Figure 2.1 Classification of adverse reactions to food modified from Dean, (2000).

Food hypersensitivities include IgE (Immunoglobulin E)-meditated and/or non-IgE-mediated immune system responses (Bruijnzeel-Koomen, 1995; Sampson and

Leung, 2011). Our immune system has evolved to protect multicellular organisms from foreign pathogens like virus, bacteria, parasites, antigens and other invaders. When a foreign invader is recognised, the immune system will trigger an effector response, aiming to eliminate or neutralise the invader. Normally when dealing with antigens, the effector molecules induce a localised inflammatory response without doing extensively tissue damage. However, unexpected events could happen when the inflammatory

responses have detrimental effects, resulting in significant tissue injury, serious disease, even death. Such an inappropriate and damaging immune response is called hypersensitivity (Kindt et al., 2007; Lee, 2012).

Food hypersensitivity can be classified into four types: IgE-mediated hypersensitivity

(Type I), IgG or IgM-mediated cytotoxic hypersensitivity (Type II), immune complex-mediated hypersensitivity (Type III) and cell-mediated hypersensitivity (Type

IV) (Kindt et al., 2007). Clinical and experimental evidence supports that Type I, III, IV hypersensitivities are associated with food, often referring to as true food allergies

(Taylor and Baumert, 2012). Type II hypersensitivity could theoretically associate with food allergy, as food antigens capable of binding to cell surfaces would result in Type I hypersensitivity (Lee, 2012; Sweeney and Klotz, 1987).

2.2. Food allergy

Among the different types of food allergies, the mechanisms of IgE-mediated type are most extensively studied (Dean, 2000).

2.2.1. General mechanism of IgE-mediated food allergic reactions

IgE is one of the five classes of antibodies presented in the human body to prevent disease. Normally IgE only presents in a trace amount in the body, and its main role is to defend parasitic infections; but atopic individuals could produce IgE when they are exposed to certain antigens (Taylor and Hefle, 2001). Statistical data shows that

IgE-mediated reactions account for the majority of food allergic reactions (Samartín et al., 2001). The general mechanism of IgE-mediated reactions is illustrated in Figure 2.2.

Figure 2.2 The general mechanism of IgE-mediated allergic reactions.

The mechanism of IgE-mediated food allergic reactions involved two phases: sensitisation phase and elicitation phase (National Academic of Sciences, 2016).

In the sensitization phase, allergenic proteins pass through the epithelium of small intestine or skin and are taken up by dendritic cells (DCs). The DCs present the processed allergen-derived peptides to the T cell receptors (TCRs) of naïve T cells. The allergen-derived peptides are physically associated with the major histocompatibility complex (MHC) class II molecules on the DC membrane. The naïve T cells are then polarised to T helper cell type 2 (TH2 cells) which bear the TCR that can recognise the specific allergen-derived peptides and interact with B cells. By this stage, the immune response is skewed towards a TH2 response. Because the allergen might have encountered the environment of B cells located in the lymphoid tissue when the allergen travels through lymph draining tissues of intestines or skin, the B cells are able to recognise the same allergen and internalise and process it into allergen-derived peptides.

The allergen-derived peptides are then presented to the TH2 cells also by way of MHC class II molecules. The TH2 cells secrete cytokines including IL-4 and IL-13 and express CD40 ligand to act on the B cells (Samartín et al., 2001). This TH2 cell-B cell interaction activates both cells. Plasma cells are then differentiated from B cells, 7

producing food allergen-specific IgEs. The produced IgE will quickly bind to Fc receptors on the membranes of tissue mast cells, which is known as the completion of the sensitisation process (Dean, 2000). Noticeably, sensitisation does not necessarily elicit clinical symptoms. Sensitisation can continue to exist without patients eliciting any clinical symptoms (National Academic of Sciences, 2016).

In the elicitation phase, when the same allergen is consumed, the allergen is recognised by two or more IgE molecules bound to the Fc receptors of the sensitised mast cells, causing the degranulation of the mast cells. Granules inside the cell are transferred to the plasma membrane and released into the extracellular environment. As a result, a variety of inflammatory mediators such as histamine and platelet-activating factor are released, causing a wide range of adverse effects. The adverse effects include increasing intestinal peristalsis and contractions and increasing the permeability of certain blood vessels, leading to the leakage of fluid into the tissues. The occurrence of the adverse effects favours the allergens to pass into bloodstream, causing the sensitisation of blood basophils and release of inflammatory mediators. The mediators then travel to different muscle, blood or nervous endpoints, resulting in allergic manifestations of various organs (National Academic of Sciences, 2016; Kindt et al., 2007). It is still unknown why some sensitised individuals elicit clinical symptoms while some others do not.

Moreover, the mechanism that certain allergens produce organ-specific reactions is still not well understood (National Academic of Sciences, 2016)

2.2.2. Routes of exposure and sensitization

The exposure routes of food allergy can include foetal exposure to allergenic proteins ingested by the mother during pregnancy, exposure to allergenic proteins in mothers’ breast milk, ingestion of food allergens, skin contact of food allergens by contact and exposure to airborne food allergens (National Academic of Sciences, 2016).

The sensitisation process of food allergy, as discussed above, can be through small intestine after ingestion of allergens, or through the respiratory route when the allergens are aerosolised, or through skin exposure. Most sensitisation, however, occurs in the gastrointestinal tract (Sellge and Bischoff, 2008). In the intestinal tract, the components including gastric acid, mucus, the intact epithelial layer, digestive enzymes and the normal intestinal peristalsis together form a non-immunological barrier to prevent the body from exposing to allergens. Food allergens are often resistant to high temperature, acids, and proteolytic enzymes, making them easier to survive the food processing and while going through the intestinal barrier. Moreover, the intestinal barrier is sometimes disrupted, allowing the penetration of immunoreactive peptides in larger quantities, which is thought to accelerate the development of food allergy. Intestinal permeability is reported to be higher in patients with adverse reactions to food (Ventura et al., 2006).

2.2.3. Clinical manifestations

The symptoms of IgE-mediated allergic reactions normally onset from a few minutes to several hours after ingestion of the offending food (Taylor and Baumert, 2012). The clinical manifestations (Table 2.1) range from mild symptoms like nausea and eczema to life-threatening symptoms like severe asthma and anaphylactic shock (Kindt et al.,

2007; Lemke and Taylor, 1994). Gastrointestinal and respiratory symptoms are more common than other symptoms (Taylor and Hefle, 2001). Apart from histamine and platelet activating factor, other mediators include leukotrienes and prostaglandins are also realised during the elicitation phase. Leukotrienes are responsible for delayed symptoms like late-phase asthmatic reactions (Taylor and Baumert, 2012).

Table 2.1 The main clinical symptoms associated with the IgE-mediated food allergies

Type Symptom Nausea Vomiting Gastrointestinal Diarrhoea Abdominal cramping Urticaria Dermatitis or eczema Cutaneous Angioedema Pruritus Rhinitis Respiratory Asthma Laryngeal edema Anaphylactic shock Other symptoms Hypotension

The most severe manifestation is the anaphylactic shock (also called systemic anaphylaxis), which is a systemic syndrome involving multiple organs. Symptoms might come from the gastrointestinal tract, respiratory tract, skin and cardiovascular system all at once; and death can occur due to severe hypotension coupled with respiratory and cardiovascular complications (Kindt et al., 2007).

Allergic reactions induced by food and anaphylaxis caused half a billion economic burden in the United States (US) in 2007 (Patel et al., 2011). Two random-digit-dial surveys (a public survey and a patient survey) investigating the prevalence of anaphylaxis in the US were conducted by Wood et al. (2014). From the public survey, the prevalence of anaphylaxis among the whole adult population was estimated 1.6 –

5.1% (Wood et al., 2014). From the patient survey, they found 344 out of 1059 adult patient respondents reported a history of anaphylaxis over a 10 – year period from 2004 to 2013, among which nearly one third (31%) was caused by food (Wood et al., 2014).

These results indicate that anaphylaxis may be a common issue in the US. Mullins et al.

(2015) examined the hospital food-induced anaphylaxis admission rate in Australia between 2005 – 2006 and 2011 – 2012, and found that the admission rate among the 10

whole population increased from 0.0056% per year in 2005-2006 to 0.0082% per year in 2011 – 2012. Mullins et al. (2015) also revealed that the highest admission rate due to food-induced anaphylaxis occurred in children aged 0 – 4 years (0.0217% in 2005 –

2006 and 0.0303% in 2011 – 2012). There was a dramatic increase in admission rate for the same cause in youngsters aged 5 – 14 years (from 0.0058% in 2005 – 2006 to

0.0210% in 2011 – 2012) (Mullins et al., 2015).

2.2.4. Common food allergens in different geographic regions

Food allergens are the components within food that elicit allergic reactions. They are usually specific glycoproteins that interact with the body’s immune cells and trigger the development of food allergy (National Academic of Sciences, 2016). More than 170 foods have been evaluated to be potentially allergenic food groups (Burks et al., 2012).

Food and Agricultural Organisation (FAO) identified eight major food groups (denoted as the “big eight”) – milk, eggs, peanuts, tree nuts, fish, shellfish, soybean and wheat, which accounted for the majority of IgE-mediated food allergies (Bousquet et al., 1998;

Taylor and Hefle, 2001). Among the big eight food groups, fish, shellfish and tree nuts belong to the collective allergenic food groups comprising many species.

Globally, food allergy is likely to affect 5% of adults and 8% of children, and the total population suffering from food allergy is estimated to be around 240 – 550 million

(Sicherer and Sampson, 2014; Fiocchi et al., 2013). Infant and young children are more susceptible to cow milk, eggs, peanuts and tree nuts. Shellfish, fish, peanuts, fruits, and vegetable allergies are more common in adults (Liu et al., 2010a; McGowan and Keet,

2013; Nwaru et al., 2014; Soller et al., 2012; Taylor, 2006).

A National Health and Nutrition Examination Survey based on self-reported assessments from a total of 20,686 individuals in the US conducted between 2007 and

2010 revealed that the prevalence of food allergies was 8.96%, corresponding to the

prevalence of 6.53% and 9.72% in adults (McGowan and Keet, 2013). Among the major allergens, milk, peanut and shellfish seemed to be the top 3 food allergens in both children and adults (McGowan and Keet, 2013). Another similar national survey conducted in Canada based on self-reported assessments also reported a high prevalence of 6.69% among 9,667 individuals (Soller et al., 2012). The prevalence was 7.14% in children and 6.56% in adults. In Canada, the top 4 food allergens to children were peanuts, tree nuts, milk and eggs; and the top 4 food allergens to adults were shellfish, milk, fruits and vegetables (Soller et al., 2012).

The prevalence data available in Australia is mainly in infants. A number of 2,848 infants participated in the study involving skin prick test (SPT) and oral food challenge

(OFC) (Osborne et al., 2011). The results showed the prevalence of positive SPT to raw egg white was 16.5%, followed by 8.9% to peanuts, 5.6% to cow’s milk, 2.5% to sesame, and 0.9% to shellfish. The prevalence of oral challenge proven peanut allergy was 3.0%, raw egg allergy 8.9% and sesame allergy 0.8% (Osborne et al., 2011). This study came to an astonishing conclusion that more than 10% of the one-year-old infants showed an IgE-mediated allergy to more than one of the major food allergens (Osborne et al., 2011).

A systematic review was performed to estimate the prevalence of common food allergies in Europe (Nwaru et al., 2014). The estimated self-reported prevalence for the common allergenic foods across was 6.0% for cow’s milk, 2.5% for egg, 3.6% for wheat, 1.5% for soy, 0.4% for peanuts, 1.3% for tree nuts, 2.2% for fish and 1.3% for shellfish (Nwaru et al., 2014). In addition, the estimated prevalence proven by the food challenge was 0.6% for cow’s milk, 0.2% for egg, 0.1% for wheat, 0.3% for soy, 0.2% for peanut, 0.5% for tree nuts, 0.1% for fish and 0.1% for shellfish, which was much lower than the self-reported data (Nwaru et al., 2014).

Egg allergy (2.8 – 4.0%) and cow’s milk allergy (0.8 – 3.5%) seem to be the most 12

prevalent food allergies among children in Asia (Chen et al., 2012; Kim et al., 2011; Lee et al., 2013). Peanut (0.43 – 0.7%) and tree nuts (0.3 – 0.7%) seem to be less prevalent in Asian countries (Kim et al., 2011; Shek et al., 2010). Fish and shellfish allergies seem to be more common in Asia. In a questionnaire-based survey from Taiwan, clinician diagnosed fish allergy (19.0%, n = 2,086) was the second most prevalent food allergy after seafood allergy (85.6% to shrimp and crab, n = 2,086) (Wu et al., 2012). A group of preschool children (n = 441) aged 3 – 7 year from Northern Thailand were examined for allergy, and 22% reported allergic reactions to shrimp and 8% of them reported allergic reactions to fish (Lao‐araya and Trakultivakorn, 2012).

2.2.5. Management of food allergy

2.2.5.1. Legislation about food allergen declaration

There is currently no cure for food allergies, and the only proven management is strict avoidance of the offending food(s) (Sicherer and Sampson, 2014). From the prevalence data discussed in Section 2.2.4, food allergy is apparently a public health and food safety issue worldwide. From the food industry perspective, legislation on food allergen declaration became crucial to regulate food industry to take part in the management of food allergens in processed foods, to ensure the safety as well as food choice diversity for allergic consumers. Table 2.2 lists the allergenic foods currently to be declared in different countries (Ebisawa et al., 2017; Gendel, 2012; Taylor and Hefle, 2006; van

Hengel, 2007). The legislations vary in different countries and jurisdictions because each country has their own prioritised list of regulated food allergens. There are, however, no clear criteria of how the priorities have been established (Gendel, 2012).

The WTO codex recognise nine food groups and ingredients for food allergen labelling of pre-packaged foods. (Gendel, 2012). To date fourteen major allergens have been identified in the European Union (EU) that require mandatory labelling, and the regulation applies to all the member states within EU. Apart from pre-packaged foods, 13

foods in restaurants and street kitchen are also required allergen declarations. Canada has similar prioritised food allergens as the EU except for celery and lupin. The US,

China, and Hong Kong recognised mainly the “big eight” food allergens and does not regulate sulphite as recommended by the Codex. Only Singapore and Thailand follow the recommendation of the Codex (Table 2.2). Apart from the Codex recommendation and sesame seeds, Australia and New Zealand recently added lupin into the mandatory list (FSANZ, 2017). Japan and Korea have their unique food allergen lists. As buckwheat is an emerging issue in Japan and Korea, it is recognised as a prioritised food allergen. Peanut and tree nuts seemed to be not a problem in Japan and Korea.

Shrimp and crab are the only crustaceans requiring labelling in Japan and Korea

(Ebisawa et al., 2017; MFDS, 2016). Mackerel is the only fish species that required to be declared in Korea (MFDS, 2016), which might require the use of detection methods

(e.g., ELISA) specific to mackerel to enforce this regulation. The “other” food allergens listed in Table 2.2 for Japan are all recommended for labelling, but not mandatory

(Ebisawa et al., 2017).

Table 2.2 Foods listed for mandatory declaration in different countries (Ebisawa et al., 2017; Gendel, 2012; Taylor and Hefle, 2006; van Hengel, 2007)

Food Group Codex Alimentarius1 US EU Canada Australia/New Zealand Japan2 Korea3 China Hong Kong Singapore Thailand Wheat/Cereals ● ● ● ● ● ● ● ● ● ● ● Eggs ● ● ● ● ● ● ● ● ● ● ● Milk ● ● ● ● ● ● ● ● ● ● ● Peanut ● ● ● ● ● ● ● ● ● ● ● Tree nuts ● ● ● ● ● ● ● ● ● Soy ● ● ● ● ● ● ● ● ● ● Crustaceans/shellfish ● ● ● ● ● ● ● ● ● ● ● Fish ● ● ● ● ● ● ● ● ● ● Sesame ● ● ● Mustard ● ● Celery ● Lupin ● ● Sulphite ● ● ● ● ● ● Buckwheat ● ● Others ● ● 1 According to the Codex recommendation: cereals containing gluten are required to be declared; lactose in milk is required to be declared; and sulphite in concentration of 10 mg/kg or more is required to be declared. Barbados, Chile, Papua New Guinea, the Philippines, St. Vincent and the Grenadines use the wordings of the Codex in their regulatory framework. 2 “Others” in Japan include: Abalone, squid, salmon roe, orange, cashew nut, kiwifruit, beef, walnut, sesame, salmon, mackerel, soybean, chicken, banana, pork, matsutake mushroom, peach, yam, apple, and gelatine; these are all recommended allergen for labelling but not mandatory. 3 “Others” in Korea include: pork, peach, and tomato. 15

2.2.5.2. Population-based food allergen thresholds and precautionary allergen labelling

The legislations often only require food allergens that are part of the ingredient to be labelled. Whereas hidden allergen residues, which present through the unintended use of ingredients derived from an allergenic source or the undeclared use of rework and leftovers or the use of shared equipment and facilities, still pose high risks to food allergic consumers (Allen et al., 2014a). Precautionary allergen labelling (PAL), such as those of “may contain” statements or similar, is then encouraged by many countries

(except for Japan and US) (Allen et al., 2014b), to provide an additional information about the potential danger of undeclared residues to help allergic consumers make well-informed choices. However, the current PAL statements are not officially regulated, and manufacturers tend to misuse or overuse the PAL statements as an alternative for evidence-based allergen risk management (Allen et al., 2014b; Hefle et al., 2007;

Zurzolo et al., 2013). Zurzolo et al. (2013) investigated the prevalence of PAL statements among 1,355 pre-packaged food products in a large supermarket in

Melbourne, Australia and found that 882 products (65%) had a precautionary statement for one or more allergens. Another study by Ford et al. (2010) reported that detectable allergen residues were only found in 5.3% of the investigated products with PAL statements (including 57 products for egg, 59 for milk and 112 for peanut). The misuse or overuse of PAL statements has created huge confusion to allergic consumers, and many consumers found PAL statements untrustworthy (Barnett et al., 2011).

To solve the issue with PAL statements, an initiative was taken by the Allergen Bureau

Australia to establish a Voluntary Incidental Trace Allergen Labelling (VITAL®) program. The VITAL® program helps food companies to thoroughly review the allergen status of all the ingredients and the processing conditions that may contribute to the allergen status of the finished product through an evidence-based risk assessment 16

approach, and then makes recommendation on whether or not a PAL statement is needed on the pre-packaged foods (Allergen Bureau, 2016).

® The key component of the VITAL program is the VITAL Action Level Grid. The Action Levels are determined based on the reference doses of the main food allergens

(established by the VITAL scientific expert panel) in conjunction with appropriate their serving sizes (Allergen Bureau, 2012). The reference doses have been established based on the estimated allergen thresholds of the probability distribution models with global clinical data of individual no-observed adverse effect levels (NOAELs) and lowest-observed adverse effect levels (LOAELs) (Taylor et al., 2014). The reference dose, EDp, of a food allergen is defined as the eliciting dose that is predicted to produce a response in a certain percentage (p%) of the allergic population (Taylor et al., 2014).

Table 2.3 lists the reference doses of eleven food allergens (Allen et al., 2014a). The reference doses of peanut and cow’s milk (studied with a large number of subjects, n >

200) are based on the eliciting doses that are predicted to provoke an allergic reaction in

1% of the allergic population who are highly sensitive (=0.01). While the reference doses of soy, wheat, cashew nut, mustard, lupine, sesame and shrimp (studied with a smaller number of subjects, n < 80) are based on the eliciting doses that were predicted to provoke an allergic reaction in 5% of the highly sensitive allergic population

(=0.05). The reference doses of egg and hazelnut have been estimated using both ED01 and ED05.

Using the estimated reference doses, an action level transition point of each allergen is then calculated by multiplying the reference dose (mg) by 1000 per serving size (g)

(Allergen Bureau, 2012). The serving size might be the maximum amount of a product consumed by consumers or the reference amount of a whole product presented to consumers. There are currently two action levels in the VITAL® program guided by the action level transition point (Allergen Bureau, 2012). The Action Level 1 is adopted

when the contamination concentration of a food allergen is below the transition point and a PAL statement is not required for the relevant allergen under evaluation. The Action

Level 2 is adopted when the contamination concentration of a food allergen is above the transition point and a PAL statement is required for the relevant allergen using the standard VITAL® statement, “may be present”. The VITAL® is now a useful decision support tool to improve the proper use of PAL statements and has been widely promoted in Australia. Currently, no reference doses have been established for fish and celery, due to inadequate data to support the model fitting for these two allergens with reliability

(Taylor et al., 2014).

Table 2.3 Estimated population-based thresholds (reference doses) of the main allergenic foods (Allen et al., 2014a)

Allergen Proposed eliciting dose Protein level (mg)

Peanut ED01 0.2

Cow's milk ED01 0.1

Egg ED01 and lower 95% confidential interval (CI) of ED05 0.03

Hazelnut ED01 and lower 95% CI of ED05 0.1

Soy Lower 95% CI of ED05 1

Wheat Lower 95% CI of ED05 1

Cashew nut Lower 95% CI of ED05 2 (Provisional)

Mustard Lower 95% CI of ED05 0.05

Lupin Lower 95% CI of ED05 4

Sesame Lower 95% CI of ED05 0.2

Shrimp Lower 95% CI of ED05 10

A more recent study by Ballmer-Weber et al. (2015) investigated the threshold dose distributions of five main allergenic foods including celery and fish among the

European population. The threshold data were collected from the double-blind placebo-controlled food challenge studies from food allergic patients drawn from

EuroPrevall birth cohort, community surveys, and outpatient clinic studies. By the interval-censoring survival analysis, they obtained the reference doses ED10 of 1.6 mg protein for celery and the ED10 of 27.3 mg cod fish protein for fish (Ballmer-Weber et 18

al., 2015). However, the 95% confidence intervals obtained for fish was too wide, therefore, they concluded that the fish ED10 was not statistically valid.

2.3. Fish allergy

2.3.1. Overview of Fish

2.3.1.1 Fish as an important animal protein source for human diet

Fish is a valuable protein source for the human diet, because it provides high-quality protein, unsaturated fatty acids, vitamins and minerals such vitamin A, iodine, zinc, and iron (Arino et al., 2005). In 2013, fish contributed to 17% of the animal protein intake of the global population (FAO, 2016). The world fish utilisation as food supply continues to grow over the past six decades, rising from 9 kg/capita to around 22 kg/capita (FAO, 2016). A significant increase in fish consumption has been observed in

Asia and Africa: from 10.8 kg/capita in 1961 to 39.2 kg/capita in 2013 in East Asia; from 13.1 to 33.6 kg/capita in Southeast Asia, and from 2.8 to 16.4 kg/capita in North

Africa (FAO, 2016). Moreover, fish accounted for around 20% of the animal protein intake in developing countries and approximately 18% in low-income food-deficit countries.

There are 27,977 fish species being recorded, but only a few orders of fish are commonly consumed, mainly finfish (Lee, 2012). These include Salmoniformes

(salmon, trout, whitefish), Perciformes (perch, snapper, tuna, mackerel, tilapia),

Gadiformes (cod, pollock, hake), Pleuronectiformes (sole, whiff, flounder),

Clupeiformes (herring, sardine, anchovy), Scombriformes (tuna and swordfish) and

Cypriniformes (carp) (Lee, 2012). The main compositions of fish include water (60 –

80%), protein (15 – 21%), lipids (varying from < 1% to > 20%), and minerals (ash) (1 –

2%) (Arino et al., 2005; Lee, 2012). Figure 2.3 displays the phylogenetic relationship of the main ray-finned fish species (Understanding Evolution, 2017). The phylogenetic 19

tree demonstrates the evolutionary relationship between different fish species, and fish species sharing common ancestors are connected by the nodes (Understanding

Evolution, 2017).

Figure 2.3 Phylogeny of the ray-fin fish species (Understanding Evolution, 2017).

2.3.1.2. Fish proteins

Fish proteins account for around 15 – 21% of the total weight (wet basis). There are three major protein groups based on their water solubility characteristics: myofibrillar proteins, sarcoplasmic proteins, and stroma proteins (Tahergorabi and Jaczynski, 2011). 20

Myofibrillar proteins are the main group of fish proteins, accounting for 60 – 80% of the total proteins (Dyer et al., 1950; Tahergorabi and Jaczynski, 2011). They are composed of myosin, actin, tropomyosin, troponin and actinin. Myofibrillar proteins have been characterised as the salt-soluble proteins, which can either be extracted in saline solutions with ionic strength (IS) above 0.6 or extracted in very low IS

(Tahergorabi and Jaczynski, 2011).

Sarcoplasmic proteins are the main water-soluble proteins that can be easily extracted with low IS saline solution or just simply pressing the muscle tissue (Tahergorabi and

Jaczynski, 2011). This group of proteins includes myoglobin, haemoglobin, globin, albumins and some enzymes (Tahergorabi and Jaczynski, 2011). Pelagic fish that live neither close to the bottom nor near the shore of ocean or lake waters have more sarcoplasmic proteins than demersal fish that live on or near the bottom of waters

(Tahergorabi and Jaczynski, 2011). Parvalbumin (PAV), the primary fish allergen, is a sarcoplasmic protein (Lee et al., 2012).

Stroma proteins come from the connective tissue of fish muscle. They are water-insoluble proteins (Tahergorabi and Jaczynski, 2011). Neither acid nor alkaline solution nor physiological saline solution is feasible for isolation of these proteins

(Tahergorabi and Jaczynski, 2011). The main proteins are collagen and elastin. Fish gelatine, which is derived from collagen on the fish skin, has been suspected to be a minor fish allergen (Taylor et al., 2004). But the reactivity of patients from ingested fish gelatine has not been proven, and it is still unclear whether the patients are allergic to fish gelatine or the gelatine was contaminated with the fish allergen during processing.

2.3.2. Fish allergy

2.3.2.1. Introduction to fish allergy

Unlike milk and egg allergies with evidence supporting that children may develop 21

tolerance (Wood, 2003); once fish allergy is onset, it is known to persist as a lifetime issue (Sharp and Lopata, 2014). The routes of exposure to fish allergens include ingestion of raw, cooked or processed fish, skin contact with raw fish, or even inhalation of aerosolised fish vapours from fish heading, degutting or boiling (Sharp and Lopata,

2014). The clinical symptoms of fish allergy range from mild such as vomiting, diarrhea, and dermatitis to life-threatening symptoms like anaphylactic shock (Sharp and Lopata,

2014). Table 2.4 lists the occurrence of the main symptoms of fish allergy reported from the Philippine, Singapore, and Thailand. Apart from the hive, which is the most common fish allergy symptom, respiratory symptoms are the dominant symptoms in all of the three countries, followed by gastrointestinal symptoms and cardiovascular symptoms. The routes of exposure have shown that the sensitisation of fish allergy can occur in the intestinal tract as well as in the respiratory tract. Sensitisation via intestinal tract is considered the main route and has been confirmed by a clinical study

(Untersmayr et al., 2007). Compared to other food allergies, fish allergy tends to induce high proportion of severe reactions and has shown to be the primary cause of death related food allergy in emergency rooms in Australia (Schulkes et al., 2014; Sharp and

Lopata, 2014). In the questionnaire-based study by Schulkes et al. (2014), more than 50%

(twenty of thirty-eight patients) reported severe allergic reactions including oral symptoms, skin symptoms, gastrointestinal symptoms and respiratory symptoms; and six of twenty patients even reported cardiovascular symptoms. PAV has been identified as a cross-reacting allergen that patient IgE antibodies are able to recognise between fish species (Wild and Lehrer, 2005). The national telephone survey on fish allergy conducted in the US informs that 67% of the fish allergic consumers have reported reactions to multiple fish species (Sicherer et al., 2004). Therefore, it is unfortunate but compelling for fish allergic patients to avoid all the fish species to prevent cross-reactivity.

Table 2.4 Occurrence of symptoms of fish allergy from the Philippines, Singapore, and Thailand (Connett et al., 2012)

Type Symptom Philippines1 (%) Singapore2 (%) Thailand3 (%) Cutaneous Hives 217 (82.8) 8 (47.1) 3 (50.0) Throat itchiness 98 (37.4) 6 (35.3) 0 Lip swelling 73 (27.9) 6 (35.3) 2 (33.3) Coughing 43 (16.4) 2 (11.8) 0 Respiratory Throat tightness 43 (16.4) 1 (5.9) 0 Wheezing 28 (10.7) 1 (5.9) 0 Nasal congestion 20 (7.6) 1 (5.9) 2 (33.3) Eye swelling 19 (7.3) 2 (11.8) 2 (33.3) Abdominal pain 40 (15.3) 0 0 Gastrointestinal Vomiting 28 (10.7) 3 (17.6) 3 (50.0) Diarrhoea 19 (7.3) 0 0 Dizziness 22 (8.4) 3 (17.6) 0 Cardiovascular Loss of consciousness 21 (8.0) 1 (5.9) 0 1262 cases of reported among 11,434 Pilipino teenagers (14 – 16-year-olds); 217 cases reported among 6,498 Singaporean teenagers (14 – 16-year-olds); 36 cases reported among 2,034 Thai teenagers (14 – 16-year-olds).

2.3.2.2. Prevalence of fish allergy

Table 2.5 summarises the prevalence of fish allergy worldwide based on three assessing approaches of questionnaire-based, IgE testing, and oral food challenge. There are more prevalence data on the questionnaire-based than the IgE sensitisation and oral food challenge (OFC). In Europe, fish allergy seems to be highly prevalent in Finland, Spain,

Germany, and Bulgaria. The survey conducted in the US and Canada (Ben-Shoshan et al., 2010; Sicherer et al., 2004), the survey in Taiwan (Wu et al., 2012), and the oral food challenge study in Denmark (Osterballe et al., 2005) imply that fish allergy is more prevalent in adults than in children. Fish allergy seems to be more problematic in Asia than Europe according to both the questionnaire and the oral challenge data (Table 2.5).

In the study investigating the prevalence of fish allergy among Thailand, the Philippines and Singapore (Connett et al., 2012), the Philippines demonstrated the highest prevalence of 2.3%. The traditional processing of drying and salting in the Philippines

might have increased the immunological reactivity of fish (Sharp and Lopata, 2014).

The self-reported data with 2,999 Australian children was relatively high (~2.3%)

(Turner et al., 2011). Fish allergy is also an occupational allergy (Jeebhay and Lopata,

2012). Workers in seafood industry are at high risks of developing allergic symptoms during fish processing or harvesting activities. The IgE sensitisation among fish processing worker in South Africa is as high as 6% (Jeebhay et al., 2008). As the fish production and consumption are increasing steadily, the prevalence of fish allergy is expected to grow as well.

Table 2.5 Prevalence of fish allergy among different graphic regions

Continent Country Population size Age group Prevalence (%) Reference Questionnaire-based approach Sakellariou et al. Greece 2,003 20 – 54-year-olds 1.5 (2008) Osterballe et al. Denmark 843 22-year-olds 0.2 to codfish (2009) Pénard-Morand et al. France 6,672 9 – 11-year-olds 0.2 (2005) Turkey 2,739 6 – 9-year-olds 0.3 Orhan et al. (2009) Europe Kristjansson et al. Iceland 324 18-month-olds 2.2 (1999) UK 798 6-year-olds 0.3 Venter et al. (2006) Kralimarkova et al. Bulgaria 160 0 – 75-years-olds 10.0 (2008) Kristjansson et al. Sweden 328 18-month-olds 3.1 (1999) Greenhawt et al. US 287 > 18-year-olds 4.9 (2009) 3,607 0 – 17-year-olds 0.2 North US 8,816 18 – 67-year-olds 0.5 Sicherer et al. (2004) America 14,948 All age groups 0.4 Not specified Children 0.2 Ben-Shoshan et al. Canada Not specified Adults 0.6 (2010) 9,667 All age groups 0.5 813 < 3-year-olds 0.5 15,169 4 – 18-year-olds 1.5 Taiwan Wu et al. (2012) 14,036 > 19-year-olds 1.2 30,018 All age groups 1.3 Asia Thailand 2,034 14 – 16-year-olds 0.3 Philippines 11,434 14 – 16-year-olds 2.3 Connett et al. (2012) Singapore 6,498 14 – 16-year-olds 0.3 Hong Kong, 3,677 2 – 7-year-olds 0.3 Leung et al. (2009) China Oceania Australia 2,999 Children ~2.3 Turner et al. (2011) Africa Ghana 1,407 5 – 16-year-olds 0.3 Obeng et al. (2011)

Table 2.5 (continued)

Continent Country Population size Age group Prevalence (%) Reference IgE testing approach (mainly SPT and RAST) Median age 50 Germany 1,537 2.9 to mackerel Schäfer et al. (2001) years old Finland 708 15 – 17-year-olds ~9.0 Haahtela et al. (1980) UK 391 Young adults 0.2 Burney et al. (2010) Sweden 602 Young adults 0.3 Pénard-Morand et al. France 6,672 9 – 11-year-olds 0.7 Europe (2005) Spain 2,700 Not Specified 4.0 Crespo et al. (1995) Turkey 2,739 6 – 9-year-olds 0.2 Orhan et al. (2009) Kristjansson et al. Iceland 324 18-month-olds 0.6 (1999) Kristjansson et al. Sweden 328 18-month-olds 0.3 (1999) Asia China 382 0 – 24-month-olds 0.8 Hu et al. (2010) Children >6-year- South Africa 275 2.2 Mercer et al. (2002) Africa olds South Africa 575 25 – 46-year-olds1 6.0 Jeebhay et al. (2008) Oral food challenge approach 898 Children 0.0 to codfish Osterballe et al. Denmark 936 Adults 0.2 to codfish (2005) Europe Turkey 2,739 6 – 9-year-olds 0.0 Orhan et al. (2009) UK 798 6-year-olds 0.0 Venter et al. (2006) Japan 2,954 Not Specified 2.02 Ebisawa et al. (2017) China 1,604 0 – 2-year-olds 0.2 Chen et al. (2012) Asia Lao‐araya and Thailand 452 3 – 7-year-olds 0.2 Trakultivakorn (2012) 1The studied subjects were all workers in the fish processing industry. 2The prevalence was based on clinical records that doctors observed adverse reactions within 60 min after ingestion of fish.

2.3.2.3 Clinically important fish species

Table 2.6 lists some of the most prevalent fish species in the selected epidemiology studies of fish allergy. The fish species adopted in the OFC study by Sørensen et al.

(2017) were restricted to cod, salmon, and mackerel. It is evident that each country has its unique list of most clinically important fish species and the whole clinically relevant fish species spectrum globally covers a wide range of species. Among all the studies, cod, salmon, tuna, and mackerel seem to be the most studied as clinically important fish species. Tuna was considered to have low allergenicity determined by SPT (Van Do et al., 2005). Whereas tuna was repeatedly self-reported in many countries including the

Netherlands (Schulkes et al., 2014), Australia (Ng et al., 2011; Turner et al., 2011), the

US (Sicherer et al., 2004) and South Africa (Zinn et al., 1997). So, it may require further investigation of the role of tuna plays in fish allergy from the clinical perspective. From the fish allergy management point of view, the list of clinically important fish species will guide food authorities, regulatory bodies and researcher to focus more on the consequential species.

Table 2.6 List of most prevalent fish species from selected epidemiology studies of fish allergy Country Assessing approach Population group (size) Age group Most prevalent fish species (%) Reference Cod (20.0%) Salmon (5.7%) Mackerel (0%) Sørensen et al. Norway DBPCFC Fish allergic patients (35) 5 – 19-year-olds Cod and salmon (17.1%) Cod and mackerel (11.4%) (2017) Cod (71.1%) Tuna (36.8%) Tilapia (26.3%) Salmon (52.6%) Mackerel (28.9%) Sardine (23.7%) Schulkes et al. Netherlands Questionnaire Fish allergic patients (38) Not Specified Eel (44.7%) Plaice (28.9%) Swordfish (23.7%) (2014) Herring (39.5%) Pangasius (26.3%) Hake (10.5%) Pollock (36.8%) Unspecified white fish (10.0%) Salmon (9.0%) Tuna (8.0%) Fish allergic patients (94 Australia Questionnaire Children Cod (6.0%) Barramundi (4.0%) Bream (3.0%) Ng et al. (2011) families) Basa (2.5%) Australia Questionnaire General population (2999) Children Unspecified white fish (10.0%) Tuna (8.0%) Salmon (8.0%) Turner et al. (2011) Salmon (41.4%) Cod (29.3%) Trout (19.0%) Sicherer et al. US Questionnaire Fish allergic patients (58) All age group Tuna (36.2%) Flounder (25.9%) Bass (17.2%) (2004) Catfish (36.2%) Halibut (24.1%) Pilchard (2.6%) Anchovy (2.6%) Maasbanker1 (1.4%) Jeebhay et al. South Africa SPT General population (575) 20 – 45-year-olds Redeye (0.9%) Mackerel (0.3%) (2008) Hake (24.8%) Snoek (10.5%) Carp (0.95%) Yellowtail (21.9%) Tuna (2.8%) Trout (0.95%) South Africa Questionnaire Fish allergic patients (105) 7 – 74-year-olds Salmon (15.2%) Haddock (1.9%) Maasbanker (0.95%) Zinn et al. (1997) Mackerel (15.2%) Kob (1.9%) Pilchard (0.95%) Kingklip (13.3%) Sole (1.9%)

2.3.2.4. Paralbumin (PAV) – the major fish allergen

PAV, which is involved in muscle contraction and relaxation, can be found in the muscle tissue of all vertebrates. Rapidly contracting muscles tend to contain high quantities of parvalbumin (Kuehn et al., 2017). PAV is recognised as the pan-allergen existing in the muscle tissue of all fish species (Taylor et al., 2004), showing IgE recognition rate of 67

– 100% of fish allergic patients (Kobayashi et al., 2016c). They are a group of calcium-binding proteins belonging to the EF-hand protein family with a molecular weight (MW) of 10-15 kDa (Permyakov and Kretsinger, 2010). It is abundant in white muscle with a relative amount of up to 5 mg/kg fresh weight in some species

(Breiteneder, 2006); and it is less abundant in dark muscle showing lower allergenicity

(Kobayashi et al., 2006). It is reported that large migratory fish (longer than 1 metre in total length) such as salmon, tuna, and swordfish, which migrate on a periodical basis, have less PAV content then sedentary fish such as carp and cod, which live in the pelagic zone of ocean or fresh waters (Kobayashi et al., 2016d). There are probably two main reasons for this phenomenon: 1) the speed of the muscle relaxation, which is accelerated by PAV, decreases as animals get bigger (Berchtold et al., 2000), and 2) white muscles are in charge of short burst swimming and dark muscles are more responsible for low-speed swimming (Tsukamoto, 1984). Sedentary fish performs more short burst swimming for preying foods or escaping from predators than migratory fish that continuously swims at low speed (Kobayashi et al., 2016d). However, there are some exceptions such Atlantic cod, mackerel and Japanese jack that are migratory fish with high PAV contents (Kobayashi et al., 2016d). In addition, PAV contents are more abundant in the dorsal side than in ventral side of the fish body and decrease from anterior to posterior locations (Kobayashi et al., 2016d; Lee et al., 2012).

The intensive study of PAV dates back to the 1950s and 1960s when several myogen fractions were isolated from carp white muscle by ammonium sulphate fractionation 29

(Konosu et al., 1965; Pechère, 1968). They discovered that these fractions exhibited a strongly acidic behaviour with low MWs between 9 and 13 kDa. These myogen fractions have amino acid compositions of 12.5% ± 1% (w/w) of phenylalanine but no tryptophan and methionine, which differentiated them from other low MW proteins such as cytochrome c or lysozyme. Later the name “PAV” was proposed for this family of homologous proteins (Pechère and Capony, 1969). This name came from the concept of low MW “albumin”, extracted by low salt concentration (Heizmann, 1984). It has high solubility in water, with isoelectric points (pI) ranging from 3.9 to 5.5

(Bugajska-Schretter et al., 2000). PAV has high contents of aspartic acid (8 to 15 residues per molecule), glutamic acid (7 to 13), alanine (9 to 23), leucine (7 to 12), phenylalanine (8 to 10), and lysine (10 to 16) (Permyakov, 2006). PAV also has low content of arginine (1 to 3), methionine (0 to 3), cysteine (0 to 3), proline (0 to 3), tyrosine (0 to 2), tryptophan (0 to 1) and histidine (0 to 3) (Permyakov, 2006).

Furthermore, PAV can be subdivided into two lineages: α-PAVs with 109 amino acid residues and β-PAVs with 108 residues. The pI of α-PAVs is slightly higher (> 5) than the pI of β-PAVs is lower (< 4.5) (Goodman and Pechére, 1977). -PAVs are responsible for most allergenic reactions (Ma et al., 2008; Permyakov and Kretsinger, 2010). Most fish species express more than one β-PAV isoforms, of which the β-PAV isoforms show significant variations in amino acid sequence and degree of allergenicity (Kuehn et al.,

2014; Sharp and Lopata, 2014). For instance, the β1 and β2 isoforms of Atlantic salmon only show 64 % identity, yet β1 shows allergenicity, while β2 does not (Perez-Gordo et al., 2012). Aas and Elsayed (1969) used trypsin, pepsin, subtilisin, and pronase to extensively hydrolyse Gad c 1, the PAV from Baltic cod (Gadus callarias). The result showed that when Gad c 1 was with other proteins, the allergenic structure could not be hydrolysed entirely, suggesting that the other non-allergenic proteins may provide protective effects of the allergen and prevent it from being completely broken down in the intestinal tract. This phenomenon also has been observed for other allergens such

milk and nuts.

2.3.2.5. Protein structure of PAV

Figure 2.4 shows the crystal structure of carp PAV. It has six α-helices, from A to F, connected by loops, and they make up 52 of the total 108 residues (Permyakov and

Kretsinger, 2010). The core of this spherical molecule is hydrophobic, composed of side chains of phenylalanine, isoleucine, leucine, and valine; while the polar surface has charged groups exposed to the solvent or aqueous environment (Permyakov and

Kretsinger, 2010). The molecule has three primary domains: AB, CD, and EF. All three domains share a similar helix-loop-helix structure, which is called the EF-hand structure.

An EF-structure is like a right hand with its thumb and forefinger extending at the approximate right angle. Unlike the EF loop and CD loop that bind to calcium/magnesium ions, the AB loop does no bind to any ions (Kretsinger and

Nockolds, 1973; Permyakov and Kretsinger, 2010).

Figure 2.4 Crystal structure of PAV from carp (Permyakov and Kretsinger, 2010).

The calcium-binding capacity of PAV greatly affects the stability and allergenicity of

PAV. Permyakov et al. (1983) observed in their study that the pike PAV with the pI of

5.0 was more heat-stable and urea resistant than the pike PAV with a lower pI of 4.2. It was found that the PAV with the pI of 5.0 had higher affinity to Ca2+ in the EF domain than the PAV with the pI of 4.2 (Permyakov et al., 1983). Ca2+ depletion induced loss of

IgE binding had been reported previously (Bugajska-Schretter et al., 1998;

Bugajska-Schretter et al., 2000; Kobayashi et al., 2016a; Tomura et al., 2008). Tomura et al. (2008) investigated the effects of mutations on the calcium binding sites of CD and EF domains on the IgE recognition of Pacific mackerel (Scomber japonicas). The mutant D51/90A with the replacements of Asp 51 in the CD domain and Asp 90 in the

EF domain had significantly lower IgE reactivity of 0 – 7.7% compared to the natural mackerel PAV Sco j 1 (Tomura et al., 2008). Their explanation to this effect, though not fully confirmed experimentally, was the complete loss of calcium binding capability that had caused the tremendous reduction in the IgE binding.

An epitope is a portion of an allergen that is recognised by an antibody, where an antigen-antibody binding complex is formed. An epitope can be linear (sequential), which represents a primary peptide segment of a protein; or can be conformational, which is a conserved three-dimensional region of a protein (Sathe et al., 2005). Most of the studied animal IgG and IgE epitopes of PAV have been conformational.

IgG epitopes of PAV were first studied in Gad c 1, the first identified PAV from Baltic cod (Gadus callarias) (Permyakov and Kretsinger, 2010; Wild and Lehrer, 2005). Three immunoreactive regions were identified from the native peptide hydrolysis of Gad c 1

(Elsayed and Apold, 1983); 1) the residues 33 – 34 on the junction between AB and CD domains, 2) the residues 65 – 74 on the intersection between CD and EF domains, and 3) the residues 88 – 96 located in the calcium-binding loop of the EF domain. The three repeated tetrapeptides interspaced with six amino acids within the CD domain play an important role in the interaction with rabbit anti-Gad c 1 MoAb, and the change of the six interspaced amino acids did not change the immunoreactivity of the CD domain

(Elsayed and Apold, 1983). The residues 88 – 103 of the EF domain expressed immunoreactivity similar to that of the CD domain; but this segment lacks the same characteristic tetrapeptides as those observed in the CD domain (Elsayed and Apold,

1983).

Similarly, three epitopes have been identified from Sal s 1 β1: 1) residues 1 – 18 of the

AB domain, 2) residues 28 – 45 of the AB domains, and 3) residues 61 – 81 on the junction of the CD and EF domains (Perez-Gordo et al., 2012). The last two epitopes seem to correlate with the first two epitopes of Gad c 1. The epitope 3 of Sal s 1 β1 was considered to be a biomarker of the protein as it triggered the most severe symptoms among the subjected fish-allergic patients (Perez-Gordo et al., 2012). The different allergenicity of β1 and β2 is probably due to β2 having a flat confirmation, while the three epitopes of β1 form a three-dimensional structure, providing a better antibody-binding site.

A unique allergenic site of Gad m 1 from Atlantic cod (Gadus morhua) has later been identified by the same group (Perez-Gordo et al., 2013), using thirteen patient sera. This reactive site is composed of residues 95 – 109, located in the calcium-binding site of the

EF domain, correlating with the immunologically reactive residues 88 – 103 of Gad c 1.

The 3D modeling of the protein structure shows the reactive site lies in the periphery of the protein.

2.3.2.6. Molecular heterogeneity of PAV from different fish species

Databases including Allergome (http://www.allergome.org), WHO/IUIS Allergen

Nomenclature (http://www.allergen.org), AllergenOnline

(http://www.allergenonline.com) and Allfam (http://www.meduniwien.ac.at/allfam/) document amino acid sequences and molecular characteristics of current registered

PAVs from different fish species. For example, there are fifty-four PAV isoform entries from twenty-seven fish species in the database of AllergenOnline. Most of the fish species in AllergenOnline present more than one PAV isoforms with documented amino acid sequences in NCBI (National Centre for Biotechnology Centre). Though the

secondary and tertiary structures of PAVs are relatively conserved (Sharp and Lopata,

2014), their amino acid sequences show variations between different species. The oligomeric PAV (Gad m 1) from Atlantic cod (Gadus morhua) showed only 62.3% homology with Gad c 1; while it shared higher homology (75%) with the PAV (Sal s 1) from Atlantic salmon (Salmo salar) (Das Dores et al., 2002b). Saptarshi et al. (2014) summarised the phylogenetic relations between thirty-eight PAVs from fish, human, frog, and chicken (Figure 2.5). The number in the Figure represents the number of amino acid units that can be substituted between two species, so the smaller the number is, the less homology the two species share. The molecular heterogeneity of PAVs creates a huge challenge on consistent detection of PAVs among different fish species.

Figure 2.5 Phylogenetic relations of thirty-eight PAVs (Saptarshi et al., 2014).

2.3.2.7. PAV content relative to the allergenicity

PAV contents vary significantly between fish species, and they tend to correlate positively with their respective allergenicity (Griesmeier et al., 2010b; Kuehn et al.,

2010; Kobayashi et al., 2016d; Kobayashi et al., 2016b; Van Do et al., 2005). For example, the PAV content from Atlantic cod was 20 times higher than swordfish

(Griesmeier et al., 2010b); and the PAV content in yellowfin tuna was 100 times lower

than Atlantic herring (Kuehn et al., 2010). Both Atlantic cod and yellowfin tuna showed lower allergenicity and the low PAV contents of Atlantic cod and yellowfin tuna were perceived to account for this effect (Van Do et al., 2005; Griesmeier et al., 2010b).

To quantify the content of PAVs, Kobayashi et al. (2016d) developed a quantitative

SDS-PAGE method with the detection by a rabbit anti-Pacific mackerel (Scomber japonicus) PAV antibody in twenty-two commonly consumed fish species from Japan.

The detectable PAV contents varied between 11.2 mg/g in splendid alfonsino (Beryx splendens) and 0.234 mg/g in bigeye tuna (Thunnus obesus) (Kobayashi et al., 2016d).

Their results showed a positive correlation between the estimated PAV content and the

IgE reactivity determined by a direct IgE ELISA with the Spearman’s rank order correlation coefficients (r) ranging between 0.63 – 0.88. The same authors (Kobayashi et al., 2016b) defined an allergenicity index to further investigate the correlation between fish allergenicity and the PAV contents of nine fish species in another study.

The allergenicity index was calculated by multiplying the normalised IgE reactivity

(also obtained from a direct IgE ELISA) with the relative PAV content using the

Japanese eel PAV content as a control (Japanese eel PAV content = 1). Their results showed that the allergenicity index did not correlate with the IgE reactivity per fixed amount of immobilised purified PAV determined by the ELISA (Spearman’s rank order correlation coefficients r = −0.007). But, the allergenicity index significantly correlated with the PAV content (r = 0.880) (Kobayashi et al., 2016b). These studies raised an important question about the immunoreactivity of antibodies against different PAV contents, which could be an important factor contributing to the observed inconsistent immunochemical detection and quantification of PAV content.

2.3.2.8. Minor fish allergens

Apart from PAV, several minor fish allergens have been identified. A 41 kDa protein was isolated and purified from Baltic cod by Galland et al. (1998). Their immunoblots 36

showed that this protein was recognised by both cod-allergic patient sera and an anti-PAV MoAb, showing its allergenic potential. This protein was later confirmed to be homologous to an aldehyde phosphate dehydrogenase (APDH) (Das Dores et al.,

2002a).

From a fish processing plant, a 36 kDa was identified by the sera from 9 processing workers as a major allergen in pilchard, and later it was identified as glyceraldehyde-3-phosphate dehydrogenase (van der Ventel et al., 2011).

In the immunoblotting study by Rosmilah et al. (2005) using the patient sera, most of the tested patient sera recognised a 51 kDa protein in both fish species, a 46 kDa protein in golden snapper (Lutjanus johnii) and a 42 kDa in red snapper (Lutjanus argentimaculatus). These detected proteins in the raw fish extracts were found to be heat-sensitive.

The precursor of fish gelatine, collagen, has been suspected as an important fish allergen next to PAV (Hamada et al., 2003; Kanamori et al., 2011). The study by

Hamada et al. (2001) showed that 5 out of 8 patient sera reacted to bigeye tuna collagen.

Hamada et al. (2003) showed the patient IgE that reacted to bigeye tuna collagen also cross-reacted to collagens from Japanese eel (Anguilla japonica), alfonsino, Pacific mackerel and skipjack (Katsuwonus pelamis) by a competitive inhibition ELISA.

A 28 kDa protein was reported as a major allergen of chub mackerel (Scomber japonicus) (Wang et al., 2011). The immunoblots with five fish-allergic patient sera revealed strong binding to the 28 kDa protein, in the absence of IgE-binding to the PAV.

The liquid chromatography tandem-mass spectrometry (LC-MS-MS) analysis matched this protein to triosephosphate isomerase (TPI). Interestingly, TPI has shown to be an allergen in lychee, wheat, latex and crangon (Crangon crangon) (Wang et al., 2011).

The 50 kDa protein and 40 kDa proteins were detected in Atlantic cod, Atlantic salmon, 37

and yellowfin tuna by the immunoblots with the serum IgE (Kuehn et al., 2013). By the means of LC-MS-MS, the 50 kDa protein was identified as β-enolase, and the 40 kDa was fructose-bisphosphate aldolase A. The IgE ELISA showed that, in a cohort of 62,

62.9% of the patient sera showed IgE bindings to enolase, and 50% showed IgE bindings to aldolase. In addition, both enolase and aldolase showed inter-species cross-reactivity (Kuehn et al., 2013).

2.4. Current detection methodologies for fish allergen residues

2.4.1. Overview

The prevalence data of fish allergy (Table 2.5) indicates that fish allergy is a common issue worldwide, especially in Asia and South Africa. Apart from Japan, all of the other countries listed in Table 2.2 have implemented the mandatory labelling for fish containing ingredients in their food allergen labelling systems. As the detection of fish

PAV is challenging due to the molecular heterogeneity derived from the phylogenetic diversity of PAV; the regulation of unlabelled fish containing ingredients or residues in pre-packaged food is still a regulatory challenging, which has become the main concern for fish allergic consumers. Cross-reactivity and sensitivity (including the limit of detection) of the detection methodologies are the main research focus to ensure PAV residues from commercially important fish species are readily detected.

The detection methods available for fish allergen include polymerase chain reaction

(PCR), mass spectrometry (MS) and enzyme-based immunoassays methods (Lee, 2012;

Lee and Sun, 2016).

PCR is a DNA-based method for indirect detection of food allergen residues in processed food. It involves three main steps: extraction and purification of DNA from an allergenic protein of interest, amplification of a specific DNA segment by a thermostable DNA polymerase, and detection of the PCR products using methods such 38

as DNA sequencing, restriction fragment length polymorphism (RFLP) and single-stranded conformational polymorphism (SSCP) (Koppelman and Hefle, 2006;

Lee, 2012). PCR method is reported highly specific and sensitive for certain food allergen detection. A real-time PCR based on TaqMan-minor groove binder (MGB) probe technology was developed by Sun et al. (2008) to detect fish PAV. Using the forward primer 5’-CAGGACAAGAGTGGCTTCAT-3’ and the reverse primer

5’GAAGTTCTGCAGGAACAGCTT-3’, the assay was able to detect twenty-eight fish species except for golden threadfin bream (Nemipterus virgatus) and yellowfin tuna at a sensitivity of 5 picograms purified fish DNA (Sun et al., 2008). The assay achieved high intra-assay reproducibility (0.84 – 2.11%) and high inter-assay reproducibility (0.16 –

0.43%); and had no cross-reactivity with other species such as sheep, chicken, and crustaceans (Sun et al., 2008). However, the main concern of PCR method is that it targets a specific DNA fragment of an allergenic protein instead of the allergenic proteins themselves, which does not necessarily prove the presence of the allergens

(Kirsch et al., 2009).

MS in combination with high solution separation technique such as 2-dimensional electrophoresis (2-DE) has been a very popular detection method for food allergens. A rapid MS approach with selected MS/MS ion monitoring (SMIM) for fish allergen detection was developed by Carrera et al. (2012). The SMIM approach is a monitoring-scanning mode (within a limited mass-to-charge ratio) that obtains structural information by performing continuous MS/MS scans of one or more selected precursor ions (Jorge et al., 2007). In this study, PAV from sixteen fish species and six commercial products was qualitatively detected within 2 h by monitoring nineteen biomarkers of fish β-PAV from the most conserved region of PAV (amino acid residues 46 to 77)

(Table 2.7) (Carrera et al., 2012).

Table 2.7 The nineteen peptide biomarkers for the detection of fish β-PAV

Biomarker Code Peptide Sequence B1 SGFIEEDELK B2 SGFIEEEELK B3 SDFVEEDELK B4 LFLQNFSAGAR B5 LFLQTFSAGAR B6 LFLQNFSASAR B7 VIDQDASGFIEVEELK B8 SGYIEEEELK B9 FFAIIDQIHSGFIEEELK B10 LFLQNFAAGAR B11 LFLQNFCPK B12 AFAIIDQDNSGFIEEDELK B13 LFLQTFGAGAR B14 EGFIEEDELK B15 AFAIIDQDNSGFIEEEELK B16 AFAIIDQDISGFIEEEELK B17 AFHLLDADNSGFIEEEELK B18 AVGAFSAAESFNYK B19 IGVEEFQALVK

Quantitative-MS approaches of food allergen detection include comparative relative quantification and absolute quantification (Muralidharan et al., 2017). Both isotope labelling and label-free techniques can be applied to comparative relative quantification or absolute quantification (Faeste et al., 2011). Absolute quantification of food allergen by MS, which has gained an increased attention, requires the determination of the correlation between ion signal of a sample and a calibration curve established with a control (Faeste et al., 2011). Synthetic isotope-labelled signature peptides as the internal standards (ISTD) are ideal for absolute quantification (Faeste et al., 2011; Kirsch et al.,

2009). Synthetic ISTD may be spiked into samples, and differences between the ISTD intensities reflecting their abundances in the samples enable the measurement and comparison of ISTD to samples (Muralidharan et al., 2017).

In the targeted proteomics, a signature peptide can be a unique peptide of an allergen from a specific species or cultivar, or it can be a common peptide representing a group of allergenic proteins (e.g., PAV) (Ahsan et al., 2016; Koeberl et al., 2014). Although there have been many studies regarding absolute quantification of food allergens including peanut, soybean, milk and crustacean, no study has been conducted on fish

PAV so far (Ahsan et al., 2016). This is largely owing to the molecular diversity of PAV that no reliable common peptide(s) have been identified from PAVs from different fish species.

The selected reaction monitoring (SRM) approach, in particular the multiple reaction monitoring approach conducted using a triple quadrupole (QQQ) mass spectrometer, is best suited for food allergen quantification (Muralidharan et al., 2017). One advantage of MS approach for food allergen detection is that it allows simultaneous detection of multiple food allergens; however, the high cost of expensive equipment, high equipment maintenance fee and the requirements for well-trained personnel are some impediments for food laboratories to adopt MS as the routine fish allergen detection method.

Enzyme-Linked Immunosorbent Assay (ELISA) is still the most commonly used method in food industry laboratories and regulatory bodies to detect and quantify allergen residues in food products (Kirsch et al., 2009; Lee and Sun, 2016; Poms et al.,

2004). ELISA is a sensitive and specific assay for both qualitative or quantitative analysis of a target analyte without relying on sophisticated and costly equipment

(Konstantinou, 2017). The fundamental basis of ELISA is the inherent ability of an antibody to specifically bind to the corresponding antigen (Muralidharan et al., 2017).

2.4.2. ELISA for fish allergen detection

Two ELISA formats are often used for the quantification of allergens in food: a sandwich ELISA and a competitive inhibition ELISA (Muralidharan et al., 2017; Poms

et al., 2004).

2.4.2.1. Sandwich ELISA

As illustrated in Figure 2.6, a sandwich ELISA involves a pair of antibodies both specific to the same allergen (Lee and Sun, 2016; Wild, 2013). A primary antibody, which is highly specific to the allergen, is immobilised onto a solid phase to capture the allergen to form an antibody-antigen complex; then a second antibody also specific to the same allergen is added to detect the immobilised antibody-antigen complex. If the second antibody is conjugated to an enzyme for colour development, the ELISA is called a direct sandwich ELISA. If a third anti-species antibody conjugated to an enzyme is required to produce colour development; the ELISA is called an indirect sandwich ELISA. The generated colour development is positively correlated with the concentration of the allergen. Compared to the other formats, the sandwich ELISA is more sensitive, precise, robust and most commonly used in food allergen detection (He,

2013; Lee and Sun, 2016). This format depends on two antibodies binding to separate epitopes on the same allergen, so the allergen molecule has to be sufficiently large

(typically with a MW greater than 6,000 Da) (He, 2013; Lee and Sun, 2016).

(a)

(b)

Figure 2.6 The workflows of a direct (a) and an indirect (b) sandwich ELISA.

2.4.2.2. Competitive inhibition ELISA

A competitive inhibition ELISA typically has been used to detect of small molecules

(He, 2013), but it has been applied in the food allergen detection, particularly to detect small MW proteins. As shown in Figure 2.7, a pre-titration is conducted to determine the optimal antigen-antibody concentration to yield a maximum absorbance (Crowther,

2009). In a competitive ELISA, the allergen is immobilised onto the solid phase; the competition invoked by adding a free allergen (in a sample) and an antibody into the

microwell plates.

In the direct competitive ELISA (Figure 2.7a), if the (enzyme-labelled) antibody has a better binding affinity for free allergen than the immobilised allergen, the antibody will bind to the free allergen, leaving less or no antibody to bind to the immobilised allergen and less colour development. If the (enzyme-labelled) antibody has lower binding affinity for free allergen than the immobilised allergen, the antibody will bind to the immobilised allergen, resulting in more colour development (Crowther, 2009).

In an indirect competitive ELISA where a third anti-species enzyme-conjugated antibody is used (Figure 2.7b), if the antibody binds strongly to the free allergens and does not bind to the immobilised allergen, no bindings of anti-species enzyme-conjugated antibody will occur, and no colour will be developed. If the antibody binds strongly to the immobilised allergen and does not bind to the free allergen, then the anti-species enzyme-conjugated antibodies added will bind to the immobilised allergen-antibody complex, leading to colour development. In this case, the absorbance is inversely proportional to the allergen concentration. The sensitivity of competitive ELISA is dependent upon the equilibrium constant of an antibody, precise measurement of the signal generated and the level of nonspecific binding (He, 2013;

Lee and Sun, 2016). Compared to the sandwich ELISA, the competitive ELISA is less sensitive and more easily affected by matrices for the food allergen detection (He, 2013;

Lee and Sun, 2016).

(a)

(b)

Figure 2.7 The workflows of a direct (a) and an indirect (b) competitive ELISA.

2.4.2.3. Commercial ELISA and ELISA reported in the literature for fish allergen detection

2.4.2.3.1. ELISAs specific to fish protein reported in the literature

Table 2.8 summaries six quantitative and five qualitative ELISAs dedicated to fish allergen detection reported in the literature. Most of the ELISAs targeted fish species 45

from northern hemisphere (Cai et al., 2013; Chen and Hsieh, 2014; Fæste and Plassen,

2008; Gajewski and Hsieh, 2009; Gajewski et al., 2009; Lee et al., 2011; Weber et al.,

2009; Shibahara et al., 2013). Only two of the ELISAs focused on southern hemisphere fish species (Lopata et al., 2005; Sharp et al., 2015).

Many of the quantitative immunoassays reported high sensitivity and good assay precision (Cai et al., 2013; Fæste and Plassen, 2008; Lopata et al., 2005; Shibahara et al.,

2013; Weber et al., 2009). Apart from one quantitative assay detecting aerosolised fish allergen in a seafood processing plant (Lopata et al., 2005), the other quantitative assays examined a wide range of fish-related foods in various food matrices. The solid and aqueous food samples for fish protein including bread, sauces and soup were examined by Fæste and Plassen (2008) in the spike and recovery study, and they achieved the spike recoveries of 68 – 138%. The only food matrix with low spike recovery (14 –

40%) was wine, which was known to use fish gelatine and isinglass as fining agents

(Fæste and Plassen, 2008). The polyphenols and tannins in wine are the suspected interferences in wine (Abbott et al., 2010). The study by Weber et al. (2009) targeting fish gelatine and isinglass reported a high amount of sturgeon PAV content of 414.7 ±

30.6 mg/kg) in the isinglass samples; but they did not investigate further to confirm the isinglass residues in the real wine samples. Another study by Koppelman et al. (2012) examined the risk that fish skin-derived gelatine may have for fish allergic consumers.

From the ninety-five commercial fish gelatine samples they tested, seventy-three samples showed negative results (i.e., below LOD) and only one sample was detected at

0.15 μg/g, suggesting that fish gelatines are probably safe for fish allergic consumers.

The sandwich ELISA developed by Shibahara et al. (2013) examined a wider range of food samples including five model foods and thirty-seven commercial products. From the twenty-one food products with fish allergen labels, the ELISA detected fish residues from eighteen products, including twelve heated foods and five retorted foods, with a PoAb against Pacific mackerel PAV (Shibahara et al., 2013). When the effects of thermal 46

processing on the detectability of fish proteins in the ELISA were examined, the reactivity to horse mackerel PAV was not affected by thermal treatment at 100 °C, but the reactivity was reduced dramatically when it was heated above 100 °C (i.e., 121 °C).

One challenge of fish allergen detection by ELISAs is species specificity (Chen and

Hsieh, 2014; Gajewski and Hsieh, 2009; Gajewski et al., 2009; Lee et al., 2011; Sharp et al., 2015). Many studies made efforts to evaluate the cross-reactivity of anti-PAV antibodies between fish species. In the study of Lee et al. (2011), the species specificity of three anti-PAV antibodies: anti-cod PAV polyclonal antibody (FARRP PoAb), anti-carp PAV monoclonal antibody (PV235 MoAb) and anti-frog PAV monoclonal antibody (PARV-19) was evaluated among twenty-nine northern hemisphere fish species from the US and the Netherlands. The anti-cod PoAb was shown to be the better performer in detecting a wider range of fish species, though some species such as albacore tuna (Thunnus alalunga) and mahi-mahi (Coryphaena hippurus) still showed weak reactivity (Lee et al., 2011). Many other studies also tested the species specificity of the PARV-19 (Chen et al., 2006; Gajewski and Hsieh, 2009; Lee et al., 2011). Most of the northern hemisphere fish species tested with the immunoblots and the indirect

ELISAs showed good immunoreactivity to the PARV-19 antibody, except for swordfish

(Xiphias gladius), Atlantic salmon, and yellowfin tuna (Thunnus albacares).

The PARV-19 antibody was also evaluated for the detection of fish species from the southern hemisphere (Beale et al., 2009; Saptarshi et al., 2014; Sharp et al., 2015). It was shown to recognize both monomeric and oligomeric PAV of pilchard (Sardina pilchardus), anchovy (Engraulis encrasicolus), yellowtail (Seriola lalandi), snoek

(Thyrsites atun), and hake (Merluccius merluccius) from Southern Africa (Beale et al.,

2009); but it showed limited detection to species from Indo-Pacific region (Saptarshi et al., 2014; Sharp et al., 2015). Based on the phylogenetic analysis (Figure 2.5) and the inhibition ELISA by Saptarshi et al. (2014), it was concluded that immunoreactivity of

different fish species with PARV-19 (i.e., the anti-frog PAV antibody) was due to the molecular phylogenetic association of PAV.

Heated fish extracts were tested in four of the five qualitative ELISAs (Chen and Hsieh,

2014; Gajewski and Hsieh, 2009; Gajewski et al., 2009; Sharp et al., 2015). Among these, two ELISAs by Chen and Hsieh (2014) and Gajewski et al. (2009) targeted the detection of a heat-stable 36 kDa protein rather than PAV. The other two qualitative ELISAs targeted thermally processed PAV (Gajewski and Hsieh, 2009; Sharp et al., 2015). The monoclonal antibody 3E1, which was developed against heated fish extract (Gajewski and Hsieh, 2009), exhibited an identical detection pattern as the PARV-19. It recognised the majority of heated fish samples except for swordfish, yellowfin tuna, Pollock, cod,

Idaho rainbow trout (Oncorhynchus mykiss), wild salmon, whiting, and haddock (Gajewski and Hsieh, 2009). Sharp et al. (2015) developed four anti-PAV antibodies against four phylogenetically diverse fish species: Atlantic salmon, barramundi, basa, and pilchard. The anti-barramundi PoAb against two most distantly related barramundi

PAV isoforms showed broader species specificity to the tested fish species, but does not detect well swordfish, yellowfin tuna, and gurnard (Chelidonichthys cuculus).

Interestingly, the anti-barramundi antibody was shown to detect heated mahi-mahi extract, but not crude mahi-mahi extract (Sharp et al., 2015).

Table 2.8 A summary of the performance of ELISAs reported in the literature for the fish allergen detection

LOD/LOQ1 ELISA format Antigen Antibody Food Samples tested Cross-reactivity Performance of validation Reference (mg protein/kg) Bread, white sauce, soy The PoAb showed 100-0.2% sauce, fish sauce, Purified reactivity to 31 fish species Intra-assay precision: <12%; Fæste and Quantitative Rabbit mushroom soup and LOD: 0.01 Atlantic cod and no cross-reactivity to Inter-assay precision: <19%; Plassen sandwich ELISA PoAb wine spiked with LOQ: 0.02 PAV crustacean, meat, nuts and Spike recoveries: 68-138% (2008) different levels of fish legumes. protein

Canned High concentrations of No cross-reactivity with Rob pilchard pilchard were detected in the lobster (Jasus Ialandii); 24% Quantitative (Sardinops Twenty-seven ambient canning department; high reactivity of the anti-anchovy competitive sagax) & fresh Rabbit environmental samples concentrations of anchovy Lopata et Ab to fish meal (80% of LOD: 105 ng/m3 inhibition cape anchovy PoAb in a seafood processing were detected from fish meal al. (2005) anchovy and 20% of other fish ELISA (Engraulis factory bagging; no concentration species); 19% reactivity of the capensis) crude above LOD was detected in anti-pilchard Ab to fish meal. extracts the Lobster department.

1LOD: limit of detection; LOQ: limit of quantitation.

Table 2.8 (continued) LOD/LOQ ELISA format Antigen Antibody Food Samples tested Cross-reactivity Performance of validation Reference (mg protein/kg)

Cross-reactivity related to cod was 80 – 614% for 6 fish No PAV was detected in the Quantitative Five fish gelatines, 2 species: cod (Gadus ssp.), fish gelatines and the competitive Purified frog hydrolysed fish Pollock (Pollchius virens), hydrolysed fish gelatines; Weber et PARV-19 LOD: 0.1 – 0.5 inhibition PAV gelatines and 4 isinglass hake, haddock while isinglass was found to al. (2009) ELISA samples (Melanogrammus aeglefinus) have PAV content up to sturgeon (Acipenser ssp.) and 414.7 mg/kg. tilapia (Oreochromis ssp.).

Cross-reactivity of 22.6 - 99.0% to purified PAVs for 7 fish species: Japanese eel (Scomber japonicus), sardine (Anguilla Intra-assay precision: Thirty-seven Shibahara Quantitative Purified Pacific Rabbit japonica), horse mackerel LOD: 0.23 <9.5%; Inter-assay commercial processed et al. sandwich ELISA mackerel PAV PoAb (Trachurus japonicus), crimson LOQ: 0.70 precision: <10.5%; Spike foods (2013) sea bream (Evynnis japonica), recoveries: 69.4 - 84.8% Skipjack, bigeye tuna, and Japanese flounder (Paralichthys olivaceus).

Table 2.8 (continued) LOD/LOQ ELISA format Antigen Antibody Food Samples tested Cross-reactivity Performance of validation Reference (mg protein/kg)

The PAV content was 221- 2461 mg/kg in white muscle and 63 - 232 mg/kg in dark muscle from silver carp, crucian carp (Carassius auratus), grass carp (Ctenopharyngodon idellus), Mandarin fish (Siniperca chuatsi), common carp Intra-assay precision: 4.4%; Quantitative Purified silver Fish ball and waste (Cyprinus carpio), Tilapia inter-assay precision: 8.4%; competitive MoAb LOD: 0.04 Cai et al. carp (H. water derived from fish (Oreochromis mossambicus), 77 mg/kg PAV was detected inhibition B2-E1 LOQ: 0.3 (2013) molitrix) PAV ball preparation European eel (Anguilla in the fish ball after 10 ELISA anguilla), sea bream (Sparus washing cycles. latus), Japanese sea bass (Lateolabrax japonicus), blueround scad (Decapterus maruadsi), yellow catfish (Pelteobagrus fulvidraco), and rice field eel (Monopterus albus).

Table 2.8 (continued) LOD/LOQ ELISA format Antigen Antibody Food Samples tested Cross-reactivity Performance of validation Reference (mg protein/kg) Compared with the PARV-19, the 3E1 showed similar Heated catfish cross-reactivity to 43 species sarcoplasmic MoAb Gajewski Qualitative except for cod, haddock, Pollock, protein extract; 3E1; and Hsieh indirect ELISA Idaho rainbow trout, swordfish, purified frog PARV-19 (2009) whiting, wild salmon and PAV yellowfin tuna. No cross-reactivity was found with non-fish samples. The T7E10 and T1G11 only reacted to the fish family Pangasiidae of basa and tra, while MoAbs: the F7B8 and F1G11 cross reacted Heated tra T7E10, to all tested cooked fish extracts Gajewski Qualitative (Pangasius T1G11, including mahi mahi (Coryphaena et al. indirect ELISA hypophthalmus F7B8, hippurus), swordfish and (2009) ) extract F1G11 yellowfin tuna. But the F7B8 and F1G11 also reacted to non-fish samples such as shellfish and meat.

Table 2.8 (continued) LOD/LOQ ELISA format Antigen Antibody Food Samples tested Cross-reactivity Performance of validation Reference (mg protein/kg) The assay was able to detect both raw and cooked protein Partially Intra-assay precision: Chen and Qualitative Rabbit extracts of 63 common fish purified 36 kDa <8.9%; Inter-assay Hsieh sandwich ELISA PoAb species with no fish protein precision: <9.3%. (2014) cross-reactivity to non-fish samples.

The FARRP PoAb showed reactivity to the widest range Purified frog PARV-19; of fish species, while the PAV; purified Qualitative PV2352; detection was still limited to Lee et al. cod PAV; indirect ELISA FARRP some important fish species (2011) purified carp PoAb3 inducing mahi-mahi, albacore PAV tuna, chub mackerel, and swordfish.

2PV235: anti-carp PAV MoAb; 3FARRP PoAb: anti-cod PAV PoAb.

Table 2.8 (continued) LOD/LOQ ELISA format Antigen Antibody Food Samples tested Cross-reactivity Performance of validation Reference (mg protein/kg)

The anti-barramundi PoAb Purified PAVs showed the best from barramundi cross-reactivity for detecting (Lates calcarifer), 87.5% of the 45 tested fish Qualitative Rabbit Sharp et al. basa (Pangasius species, followed by the indirect ELISA PoAbs4 (2015) bocourti), pilchard anti-pilchard PoAb and the and Atlantic anti-basa PoAb. Mahi-mahi, salmon yellowfin tuna and swordfish were not detected.

The assay showed Seventy-three out of 95 cross-reactivity 37 – 100% commercial fish gelatine to Pacific cod, Pollock, samples were tested below Quantitative Purified PAV from Rabbit Koppelman Fish gelatines haddock, hake, Flounder dab 0.02 LOD, the highest fish sandwich ELISA Atlantic cod PoAbs et al. (2012) (Hippoglossoides gelatine sample showed a platessoides) and redfish positive result of 0.15 μg/g. (Sebastes fasciatus). .

4The 4 PoAbs include anti-barramundi PAV PoAb, anti-basa PAV PoAb, anti-pilchard PAV PoAb, and anti-salmon PAV PoAb. 54

2.4.2.3.2. Commercial ELISA kits for fish residue detection

Table 2.9 lists four currently available commercial test kits for the fish residue detection from Romer Labs, Elution Technologies, Bio-Check, and Immunolab. These tests were developed using the purified PAV from either cod or a crude cod protein extract from northern hemisphere species. Only the Romer Labs test kit considers species specificity by using conversion factors to calculate results from different species. A recent study compares the performance of these four commercial ELISA kits (Bugyi et al., 2016). The result showed poor quantitative detection of albacore tuna, swordfish, Atlantic salmon, and mahi-mahi, while cod (the antigen for these antibodies) was detected very well (Bugyi et al., 2016).

Table 2.9 The commercial sandwich ELISA test kits for the fish allergen detection LOD LOQ Supplier Assay name Comments (mg protein/kg) (mg protein/kg) Results are expressed as cod equivalent; conversion factors AgraQuant® Romer Labs 1.4 4 are used to convert results to Fish Assay different fish species including salmon, swordfish and tuna.

This assay targets fish PAV, Elution Fish Protein showing no cross-reactivity to 0.002 No information Technologies ELISA non-fish samples including scallop and shrimp.

The assay targets fish protein, especially the order of Gadiformes. The assay shows % reactivity <10 to monkfish, Fish-Check Bio-Check <1.0 5 tuna, skate, salmon, and trout. ELISA Non-fish samples including buckwheat, amaranth, red lentil and soya showed interferences to the assay. Results are expressed as cod Immunolab Fish ELISA 1.4 No information equivalent.

2.4.3. Steps involved in the development of a sandwich ELISA for fish allergen detection

Developing a sensitive and robust immunoassay for fish allergen detection involves the following critical steps (Lee and Sun, 2016):

1. Immunogen preparation (extraction and purification of targeted allergen) 2. Antibody production (immunisation, sera collection, and antibody purification) 3. Assay development, characterisation, and optimisation (cross-reactivity) 4. Sample preparation method development 5. Assay validation

2.4.3.1. Immunogen preparation-extraction and purification of PAV

The most commonly used buffers for PAV extraction include low ionic strength of PBS buffer and Tris-HCl buffer. PAV is a water-soluble protein and can be easily extracted with a low salt buffer (Brownridge et al., 2009; Chen et al., 2006; Lim et al., 2008; Koppelman et al., 2010; Lee et al., 2011; Saptarshi et al., 2014).

Ammonium sulphate precipitation was used in the study by Pechére et al. (1971) to purify PAV from hake muscles. The protein extracts were firstly precipitated by 70% saturation of ammonium sulphate, followed by ion exchange chromatography using a Sephadex G-75 column and size exclusion chromatography by a DEAE-cellulose column.

In addition, heat treatment that removes undesirable heat-susceptible protein from desirable heat-stable proteins including PAV is often used as the first step of PAV purification (Cai et al., 2013; Fæste and Plassen, 2008; van der Ventel et al., 2011). In the study by Fæste and Plassen (2008), the crude fish extract was firstly heated at 100 °C for 30 min, and the heat-stable proteins were sequentially precipitated with 55% and 90% ammonium sulphate. The precipitated proteins were then purified by ion exchange chromatography using Sephadex G-50 column.

The identification and characterisation of purified PAV were conducted using SDS-PAGE, amino acid composition analysis, metal ion analysis, Matrix-Assisted Laser Desorption/Ionization Time-of-Flight (MALDI-TOF) and Electrospray Ionisation (ESI) 56

mass spectrometry, ultraviolet and fluorescence spectroscopy, and Nuclear Magnetic Resonance (NMR) Spectroscopy (Permyakov, 2006).

2.4.3.2. Production of antibodies

Two types of antibodies are most commonly used in fish allergen ELISAs: polyclonal antibodies and monoclonal antibodies. High background colour generated by non-specific binding and the presence of cross-reactive antibodies in the host animal are the main concerns for polyclonal antibodies. It is advised to pre-titrate the blood sample of the host animal before immunisation to check the level of cross-reactive antibodies.

The production of polyclonal antibodies is relatively quick and inexpensive compared to the monoclonal antibody production. Host animals including rabbits, goats, and sheep have been commonly used to raise polyclonal antibodies against fish proteins. During immunisation, subcutaneous injection at multiple sites close to lymph nodes is preferred to reduce the discomfort of the host animals and elicit strong immune responses (Liddell, 2013). The production of specific IgG antibodies is monitored typically after three immunisation events. The antisera with high titre values should have a higher amount of specific antibodies to the immunogen. If the test bleed does not exhibit sufficiently high titre value, boosters injection are performed to assist with production and affinity maturation of specific antibodies (Liddell, 2013). More than one animal should be used for the immunisation as the immune responses even to the same immunogen vary significantly between individual animals. Adjuvants are typically used to stabilise the immunogen as well as to elicit the immune response so that the immune-response can be triggered persistently over a period of time (Hancock and O’Reilly, 2005). Goats and sheep are preferred sometimes for producing a larger volume of antisera are preferred, while the antisera produced by rabbits can be used more readily, which may not require further purification. Rabbit antibodies tend to exhibit a higher affinity to the corresponding antigen than other animal hosts in our experience (Lee and Sun, 2016).

Compared to the polyclonal antibodies which are produced from thousands of different types of B lymphocytes (i.e., B cells), monoclonal antibodies are produced from single hybridoma cells which are generated by the fusion of immunised B cells with immortal animal myeloma cells (Liddell, 2013). The B cells are usually isolated from the spleen

tissue or lymphocyte nodes of immunised mice or rats. After successfully producing the hybrid cells, antibody-screening tests are performed by immunoassays to select the relevant cells secreting the wanted monoclonal antibodies, and cloning by limiting dilution is performed to produce colonies that contain identical fused B cells (Liddell, 2013). The identical fused B cells then secret identical monoclonal antibodies that only recognise a single epitope of the immunogens. Monoclonal antibodies have the advantages of high specificity, purity, consistency and low non-specific bindings (He, 2013). The production of monoclonal antibodies can be long-lasting at industrial scales because of the immortality of the myeloma cells, providing unlimited quantities of the antibodies (Liddell, 2013; Immer and Lacorn, 2015). Compared to polyclonal antibodies, the production of monoclonal antibodies requires more sophisticated procedures (e.g., the maintenance of tissue culture work at a high standard) and a higher production cost. The detection by monoclonal antibodies can be easily affected when the single antibody-binding epitope is modified due to protein denaturation, hydrolysis or aggregation during processing (Muralidharan et al., 2017). On the other hand, polyclonal antibodies may be more suitable for the detection of food allergen residues in processed foods because they are more tolerant to small changes of allergenic proteins occurred during processing (Koppelman and Hefle, 2006).

2.4.3.3. Immunoassay development and optimisation

2.4.3.3.1. Coating and blocking

The immobilisation (i.e., coating) of an antibody or an antigen on the polystyrene surface is by the passive absorption (primarily consisting of hydrophobic/ionic interactions) between the biomolecules and the hydrophobic surface of ninety-six microwell plates (Gibbs and Kennebunk, 2001). Theoretically, the coating buffer should have a pH value 1 – 2 units higher than the isoelectric point of the biomolecules (Crowther, 2009). The most frequently used coating buffers include 1) 50 mM carbonate buffer, pH 9.6, 2) 20 mM Tris-HCl buffer, pH 8.5, and 3) 10 mM phosphate-buffered saline (PBS), pH 7.2 (Crowther, 2009; He, 2013). Factors such as incubation temperature, time and concentration of the biomolecules being attached affect the efficiency of immobilisation Typical condition for the coating step is at 37 °C for 1 – 3 h or 4 °C overnight, or a combination of the two (Crowther, 2009). The concentration of 58

the molecules being attached usually range from 1 to 10 μg/mL with a loading volume of 50 – 100 μL (He, 2013).

Blocking is a crucial step in conducting ELISAs to minimise non-specific bindings and increase the signal to noise ratio. After coating, reagents that are immunologically inert to the antibodies and antigens are used to block the unoccupied sites on the polystyrene surface (Crowther, 2009). Three types of blocking reagents are often used in ELISA systems: proteins, detergents, and miscellaneous non-protein polymers (Gibbs and Kennebunk, 2001). Proteins, such as bovine serum albumin (BSA), non-fat dry milk, casein, normal whole serum and fish gelatine, can stabilise the biomolecules attached on the surface and reduce the steric hindrance of the biomolecules. Detergents are seldom used alone, because they may interfere with the ionic interactions between the biomolecules and the surface, and they are easily washed away by water or aqueous buffers. Therefore, they are often used together with proteins. The commonly used detergents for blocking include Tween-20, Tween-80, Triton X-100 and sodium dodecyl sulphate (SDS) (Crowther, 2009). Under situations where proteins may react with the immuno-reagents in ELISAs, miscellaneous polymers become an alternative. Polymers such as polyethylene glycol (PEG), polyvinyl alcohol (PVA) and polyvinylpyrrolidone (PVP) are largely employed particularly in lateral flow systems due to their hydrophilicity-producing property (Gibbs and Kennebunk, 2001).

2.4.3.3.2. Washing

The washing step is also called the separation step in ELISAs. It involves several consecutive cycles of filling the microwells with washing buffer and emptying them by decanting or aspirating, to achieve the separation between unbound (free) and bound reagents. The washing buffer is usually a physiological buffer such as PBS with 0.5% Tween-20 (He, 2013). Assay sensitivity and precision can be affected by the number of washing cycles (Gibbs and Kennebunk, 2001). Inadequate washing circles could be a cause of non-specific bindings, thus reducing the assay sensitivity. Three to five washing cycles with 1-5 min soaking before the last washing is recommended to remove unbound reagents effectively.

2.4.3.3.3. Titration of reagents 59

The titration of reagents, which is achieved by checkerboard ELISAs, aims to optimise the working concentration of each component in an ELISA. The checkerboard titration, which is also called a two-dimensional serial dilution (He, 2013), involves serial dilutions of two reagents against each other for comparison of inherent compatibility of resulting combinations. An empirical process of the checkerboard titration for an indirect sandwich ELISA includes 1) titration of a fixed concentration of detection antibodies against antigens and negative control with different capture antibodies, and 2) if the detection antibodies are not conjugated to an enzyme, titration of anti-species antibody-enzyme conjugates against the optimised detection-capture antibody combination with antigens and negative control. Apart from selecting the optimal capture-detection antibody pair, background absorbance (non-specific binding), maximum absorbance, assay sensitivity and possible dynamic range (detection range) of a sandwich ELISA can also be estimated by checkerboard titrations.

2.4.3.3.4. Calibration standards

For quantitative ELISAs, a calibration curve of which the unknown concentration of an allergen can be photometrically related to photometrically is required. The calibration solutions in an ELISA should represent as closely as possible the target allergen in food matrices to be analysed (Lee and Sun, 2016). Quality control of a calibration curve with characterised sample matrices is usually performed by the test kit developers to make sure the results obtained from different lots of calibration standards are comparable. For situations where allergenic proteins from a group of species such as fish and crustaceans need to be measured, a mixture may be used as the calibration standard to represent as closely as the protein from diverse species (Immer and Lacorn, 2015). However, it is still challenging to compare results of the same food matrices from different commercial kits, due to lack of harmonised reference materials for allergens. Many efforts have made globally to calibrate the commercial test kits from different manufacturers (Immer and Lacorn, 2015). Several reference materials are suggested by the National Institute of Standards and Technology (NIST) (Abbott et al., 2010), which includes spray-dried whole egg powder (NIST RM-8445), non-fat milk powder (NIST RM-1549) and roaster peanut butter (NIST RM-2387) (Immer and Lacorn, 2015). Apart from the attempts to develop a certified reference material from various scientific communities including the 60

National Measurement Institute, Australian, selecting a preparation protocol as an international standard for the calibration standard may be one of the options for future harmonisation.

2.4.3.3.5. Incubation conditions and detection systems

Incubation allows the close proximity between antibodies and antigens for reactions to occur. The temperatures for incubation include 4 °C, room temperature (20 – 25 °C) and 37 °C. Compared to higher incubation temperature, lower incubation temperature may be used to reduce non-specific bindings, but it may require longer incubation time. A high-affinity antibody can bind to the corresponding antigen within 10 to 30 mins (Muralidharan et al., 2017). Commercial kits nowadays are usually optimised for incubation time within 10 to 30 mins at room temperature. It is an empirical process to optimise the incubation temperature and time to achieve the optimal signal-to-noise ratio and optimal assay sensitivity.

The three main detection systems used in enzyme-linked immunoassays are: 1) colorimetric system, 2) fluorescent system, 3) luminescent system (Gibbs and Kennebunk, 2001). In a colorimetric ELISA, coloured products are produced by an enzyme-substrate reaction followed by the spectrophotometric measurement. The absorbance of the coloured products is proportional to the amount of allergens being measured. Compared to the colorimetric system, fluorescence system offers a wider detection range with slightly higher sensitivity. The luminescence system is theoretically the most sensitive among the three detection systems (Gibbs and Kennebunk, 2001). Horseradish peroxidase (HRP), calf intestine alkaline phosphatase (ALP), and E. coli β-D-galactosidase are the commonly used enzymes, among which HRP is the most frequently adopted to label antibodies. In the colorimetric system, the substrates that produce water-soluble products for HRP include 3,3',5,5' tetramethylbenzidine (TMB), 2,2'-azino-di[3-ethylbenzthiazoline] sulfonate (ABTS), and o-phenylenediamine (OPD). TMB, being a highly sensitive substrate, is most commonly used. The blue products generated by TMB absorb light at 650 nm, and after adding an acid solution such sulphuric acid to stop the reaction, primarily yellow products are formed which allows the measurement at 450 nm.

2.4.3.3.6. Calibration curve fitting

Concentrations (dose) of an allergen are logarithmically related to the corresponding signals (response) in an ELISA. When plotting the dose in a logarithmic scale and the response in a linear scale, the dose-response ELISA curve is in sigmoidal in shape (Dunn and Wild, 2013). Non-linear regression curve fit models including four-parameter logistic (4PL) and five-parameter logistic (5PL) models are widely used in the calibration curve fitting of immunoassays including sandwich ELISA and competitive ELISA.

The equation for the 4PL model is shown in Equation 2.1, where A is the Y value at the bottom plateau, B is the Y value at the top plateau, C is an EC50 (effective concentration), the concentration of an allergen giving 50% of the maximum response, and D is termed hillslope describing the slope index of the curve (in the most linear part of the curve).

B-A Y = A + [ ] Equation 2.1 1+10(LogC-x)*D

The equation for the 5PL model is shown in Equation 2.2, where A, B, C, and D are the same as the 4PL model with an extra parameter of asymmetry factor S, controlling the degree of the asymmetry of the curve.

B-A Y = A + { } Equation 2.2 (LogC-x)*D S [1+10 ]

The 4PL model is a symmetrical fitting while the 5PL model is an asymmetrical fitting. As the dose-response curves of ELISAs are rarely symmetric, the 5PL model is considered a better curve fitting. On the other hand, computational fitting of the 4PL model is easier than that of the 5PL model, which can be accomplished by most mathematical software programs (Dunn and Wild, 2013).

2.4.3.4. Immunoassay validation

The purpose of immunoassays is to provide simple, specific, sensitive, consistent and reliable detection for food allergen analysis. Therefore, ELISAs for food allergen 62

detection must be validated for their sensitivity, specificity, accuracy, and precision, with an optimised sample preparation procedures, data, calculation and interpretation of the results and analysis for a given allergen in specific food matrices (Lipton et al., 2000). The analytical parameters involved in an in-house validation may include accuracy, recovery, precision, sensitivity (C50, limit of detection, limit of quantification, and detection range), specificity (cross-reactivity), matric interference, and extraction efficiency (Hirst and Miguel, 2015).

2.4.3.4.1. Sensitivity

The assessments for the assay sensitivity mainly include C50, limit of detection (LOD), limit of quantification (LOQ), and detection range (Lipton et al., 2000). C50, which is a concentration at 50% of maximum colour development, is a good indicator of assay sensitivity, as it is less fluctuating than LOD and LOQ (Lee and Sun, 2016). An LOD is defined as the lowest concentration of an allergen that can be differentiated from a true blank sample at a specified probability level; and an LOQ is defined as the lowest level of an allergen that can be quantified at a specified level of precision (Abbott et al., 2010). The detection range is the range between the lower and upper limits of concentration of an allergen in a calibration curve that yields quantitative results (Lipton et al., 2000). Despite the nonlinear relationship of the ELISA calibration curve, the LOD and LOQ for the allergen ELISA in the literature (Cai et al., 2013; Fæste and Plassen, 2008; Shibahara et al., 2013; Weber et al., 2009) are commonly calculated in a similar manner as the linear relationship of instrumental method as follows: 1) LOD is the mean measured value of a set of blank sample replicates (n = 8 – 34) plus 3 times of the standard deviation (SD) of the mean value, and 2) LOQ is the mean measured value of a set of blank sample replicates (n = 8 – 34) plus 10 times of the standard deviation (SD) of the mean value. Notably, this approach in the allergen detection is different from the typical quantification used by ELISAs of other targets (Lee and Kennedy, 2007). Typically, quantification is done within the linear part of the sigmoidal curve and the LODs are calculated as 15-20% of colour development or the lowest point within the linear range. The LOD and LOQ values of the reported fish ELISAs are listed in Table 2.7. A lower limit of detection (LLOD) and an upper limit of detection (ULOD) that represent the lower limit and upper limit of the linear range of the calibration curve are 63

also commonly used to describe assay performance (Lee and Sun, 2016).

2.4.3.4.2. Accuracy, recovery and precision

The accuracy of an immunoassay represents the closeness of an analytical result to the true value (Hirst and Miguel, 2015). It can be assessed by spiking a known amount of the target allergen into a model food matrix and measuring the percent recovery of the allergen within the detection range (Lipton et al., 2000). Using an incurred sample is believed to provide the best source of information about the assay performance of a food allergen detection method (Abbott et al., 2010). An incurred sample is prepared by mimicking the actual manufacturing conditions of a particular food matrix when a known amount of food allergen is incorporated in the sample during the food preparation (processing) (Abbott et al., 2010). It gives the most accurate response of a detection method to a particular food matrix. However, the preparation of incurred samples can sometimes be costly and challenging, especially when large quantities are needed.

Spiking samples after manufacturing become the alternative for validation studies. There are two ways to prepare spiked samples. 1) A large batch of a food sample containing a high level of a food allergen is prepared and gradually diluted to the test concentrations by mixing the food sample with another food sample that does not contain the food allergen. 2) A known amount of a food allergen is added to each sample of the test portion. The first preparation method is suitable for samples in liquid or fine powder form that are mixed readily; but the method has the limitation of obtaining well-homogeneous samples at low spiking levels. The second method is considered more precise and acceptable to overcome the issues mentioned above according to the AOAC and MoniQA food allergens communities (Abbott et al., 2010). For inter-laboratory validations, minimum two food matrices with four spiking levels per matrix are required in a spike and recovery study (Abbott et al., 2010). The ideal range for recovery levels should be within 80 to 120%. Recovery levels are affected by the extraction efficiency and the specificity of the antibody. As food matrices and matrix interferences are usually challenging to deal with in food allergen detection, the acceptable recovery range has been recommended by the allergen community to cover

between 50 and 150% (Abbott et al., 2010), which is wider than the typical recovery of 80 – 120% for immunoassays.

Most fish ELISAs reported in the literature showed acceptable recoveries against their standard protein extracts (Table 2.7). For example, Shibahara et al. (2013) reported the recoveries of 69.4 – 84.8% of Pacific mackerel PAV from five model foods (i.e., chicken meatball, creamy croquette, pork meatball dumpling, vegetable and chicken soup and rice gruel) by the sandwich ELISA. Cai et al. (2013) reported the %recovery 70.3 – 134.8% of silver carp PAV in PBS buffer, tofu soup and mushroom soup by the competitive ELISA.

The precision of an immunoassay reflects the closeness of individual results between a series of measurements of the same homogenous sample over a period of time (Hirst and Miguel, 2015). The intra-assay precision, which is also called repeatability, assesses variations of replicate analyses conducted within the assay developing laboratory; while the inter-assay precision, which is also called reproducibility, assesses variations of replicate analyses between different laboratories (Tiwari and Tiwari, 2010). Standard deviation (SD) and percent coefficient of variation (%CV) are often calculated to evaluate assay precisions. The acceptable range of %CV for intra-assay precision should be smaller than 20%, and the %CV for inter-assay precision should be smaller than 30% (recommended by the AOAC Food Allergens Analytical Community). The study by Shibahara et al. (2013) involved collaboration with three other laboratories for the evaluation of inter-assay precision. The intra-assay (n = 10) and inter-assay (n = 3) precisions they obtained for their fish ELISA using the same protein standard and reagents were < 9.5% and < 10.5%, respectively. In another sandwich ELISA by Fæste and Plassen (2008), the intra-assay precision from the replicate analyses collected from eight different days was < 19%. Cai et al. (2013) also achieved a good intra-assay precision (i.e., 8.4%) from replicate analyses from five different days.

2.4.3.4.2. Cross-reactivity

Assay cross-reactivity results in false positive responses of samples in the absence of the target analyte (Abbott et al., 2010). The assay cross-reactivity can occur due to the presence of common epitopes of proteins sharing similar protein structures or amino

acid sequences (Muralidharan et al., 2017), which is associated with antibody specificity.

In the assay validation, a wide range of food products and ingredients, especially those that are genetically similar to the target allergen and that might be present along with the allergen, should be tested to determine the ability (assay selectivity) of an ELISA to distinctly detect the target allergen without being disrupted by other components in a food matrix (Abbott et al., 2010). In the case of fish allergen detection, antibodies in fish ELISAs are required to show a broad species specificity to a wide range of PAVs from many commercially important fish species without showing undesired cross-reactivity to other proteins of similar structures from frog or mammals.

The method used by Fæste and Plassen (2008) to evaluate the cross-reactivity of their ELISA compared the PAV content detected by the sandwich assay to the total protein content of the crude protein extracts. This method, however, did not take the possible variation of PAV contents between different fish species into consideration. Another method employed by Shibahara et al. (2013) compared the C50 value of the calibration standard they used to the C50 of a fish sample. The detailed cross-reactivity of the fish ELISAs reported in the literature is listed in Table 2.7.

2.4.3.4.3. Extraction efficiency

Extraction efficiency measures the efficiency of an extraction method (the core of a sample preparation procedure) in isolating target allergen from a food matrix (Lipton et al., 2000). It greatly affects the evaluation of the assay accuracy which involves spike and recovery of an allergen from a food matrix as well as the reliability of analytical results. Most fish ELISAs reported in the literature use a simple Tris-HCl buffer or a PBS buffer to extract fish proteins from various food matrices (Cai et al., 2013; Fæste and Plassen, 2008; Lopata et al., 2005; Weber et al., 2009). For example, Weber et al.

(2009) used 50 mM Tris-HCl with 150 mM NaCl and 0.5 mM CaCl2 (pH 7.4) to extract PAV from several fish gelatine and isinglass samples; and they detected a high content of sturgeon PAV at 414.7 ± 30.6 mg/kg in the isinglass samples.

In addition, food processing induces protein structural modification (e.g., degradation, aggregation, hydrolysis, and glycosylation) and has great impacts on protein 66

extractability and antibody detectability (Muralidharan et al., 2017). The addition of SDS and 2-mercaptoethanol (2-ME) in extraction buffers has been proven to increase the extractability of denatured egg proteins from boiled eggs and fried noodles (Steinhoff et al., 2011; Watanabe et al., 2005). SDS and 2-ME can also increase the extractability of peanut, egg, milk proteins from a chocolate matrix which involves extensive processing procedures including heating, chilling, high pressure treatment and fermentation (Khuda et al., 2015).

In the study by Shibahara et al. (2013), the extraction buffer adopted was 0.12 M Tris-HCl buffer (pH 7.4) with 0.1% BSA, 0.05% Tween-20, 0.5% SDS and 2% 2-ME. They used this extraction buffer to examine thirty-seven commercial processed foods, which included some oily matrices such as pasta sauce, salad dressing and beef curry, and they were able to detect fish residues from eighteen out of twenty-one products that had fish allergen labels. Recently, sodium sulphite, which can also increase the protein extractability of highly processed foods such as retorted foods and fruit jams, was recommended to be a more human- and environment-friendly alternative of 2-ME (Ito et al., 2016). Apart from adding extraction additives, increasing extraction buffer pH and salt concentration also enhance the isolation of allergenic proteins from heat-processed food matrices (Faeste et al., 2007; Khuda et al., 2015).

2.5. The impacts of food processing on the immunoreactivity and allergenicity of fish allergens

Thermal processing is a conventional approach to ensure the microbial safety of food products as well as to improve the texture, flavour, taste. and digestibility of food products (Davis and Williams, 1998; de Jongh et al., 2015). Thermal processing modifies the protein structure, hence potentially altering the allergenicity of allergenic proteins. Proteins start to lose tertiary structures at temperatures above 55 °C (Davis and Williams, 1998). With the increase of heating, proteins continue to unfold, losing secondary structures and exposing hydrophobic cores to the outward environment at temperatures above 80 °C. Consequently, protein molecules may aggregate to overcome the unfavourable contact of the hydrophobic cores. The loss of tertiary and secondary structures might result in loss of some IgE or IgG-binding epitopes, while the

interaction between protein molecules and between protein and other components in food may generate new epitopes.

Formation of dimers and oligomers of PAV after heat treatment has been reported in fish species such as red stingray (Dasyatis akajei), silver carp (Hypophthalmichthy molitrix), orange roughy (Hoplostethus atlanticus) and whiff (Lepidorhombus whiffiagonis) (Cai et al., 2010; Griesmeier et al., 2010a; Liu et al., 2010b; Saptarshi et al., 2014). The IgG antibodies discussed in Section 2.4.2.3 were able to detect most of the heated fish samples tested. It is therefore speculated that the IgG epitopes of PAV might not be affected by thermal processing. Similar to the immunoreactivity of native PAV, the IgG responses to heated PAV varied among different fish species (Chen and Hsieh, 2014; Gajewski and Hsieh, 2009; Gajewski et al., 2009; Sharp et al., 2015).

The thermally induced changes in allergenicity also varied between different fish species. IgE reactive bands above 37 kDa after heat treatment were observed in several fish species including white pomfret (P. argentius), barramundi, Indian mackerel (Rastrelliger kanagurta) and whiff (Chatterjee et al., 2006; Griesmeier et al., 2010a); and IgG reactive bands at around 20, 40 and 60 kDa were reported in heated Atlantic cod protein extract (de Jongh et al., 2015). The complex of PAV with other proteins induced by heat treatment from allergenic and immunoreactive high MW proteins (de Jongh et al., 2015). Similarly, a 24 kDa protein in whiff cooked protein extract was detected by an anti-cod antibody, and the protein was suspected to be a PAV dimer (Griesmeier et al., 2010a).

Increased IgE reactivity after thermal treatment has been reported for many fish species including catfish, bass, tuna, flounder, barramundi, pilchard, anchovy, hake, snoek and yellowtail (Chatterjee et al., 2006; Beale et al., 2009; Bernhisel-Broadbent et al., 1992). The formation of new allergenic high MW proteins might generate new PAV IgE epitopes that contribute to the increased IgE reactivity. A murine model revealed that the mice sensitised with cooked pilchard extract produced more TH2 cytokine (i.e., a key cytokine in the sensitisation process (Section 2.2.1)) in the mediastinal lymph node cells and splenocytes (van der Ventel et al., 2011).

The allergenicity may only be sustained up to 100 °C because complete loss of IgE

bindings have been demonstrated in canned tuna, canned salmon, autoclaved cod extract and Pacific mackerel extract heated at 140 °C (Bernhisel-Broadbent et al., 1992; de Jongh et al., 2015; Kubota et al., 2016). The CD analysis of heated PAV demonstrated gradual reduction of the ellipticity at 222 nm (an indicator for the α-helical structure) with increased temperature (Griesmeier et al., 2010a; Kubota et al., 2016). Also, no Ca2+ was detected in the Pacific mackerel extract heated at 140 °C (Kubota et al., 2016). The link between the conformational changes and the loss of allergenicity of fish parvalbumin, however, still requires further investigation at the molecular level.

2.6. Conclusion

Food allergy is a significant heath issue, affecting 240 – 550 million people globally (Sicherer and Sampson, 2014; Fiocchi et al., 2013). For the safety of food allergic consumers, many countries have implemented mandatory food allergen labelling systems to manage food allergy (Ebisawa et al., 2017; Gendel, 2012; Taylor and Hefle, 2006; van Hengel, 2007). Fish is recognised as one of the big eight food allergens (Bousquet et al., 1998; Taylor and Hefle, 2001), and fish allergy is a serious health problem worldwide due to potential severity and life-long persistence (Table 2.5). The management of fish allergy is challenging and unique because fish PAV represents molecular heterogeneity derived from the phylogenetic diversity of PAV.

PAV is the pan-allergen existing in the muscle tissue of all fish species (Taylor et al., 2004), showing IgE recognition rate of 67 – 100% of fish allergic patients (Kobayashi et al., 2016c). Databases including Allergome (http://www.allergome.org), WHO/IUIS Allergen Nomenclature (http://www.allergen.org), AllergenOnline (http://www.allergenonline.com) and Allfam (http://www.meduniwien.ac.at/allfam/) document the amino acid sequences and molecular characteristics of PAV from a diverse range fish species globally. For example, there are 54 entries related to PAV in the database of AllergenOnline, covering 29 registered PAVs from 27 fish species. The molecular heterogeneity of PAV has led to the analytical challenges of inconsistent detection by anti-PAV antibodies derived from different fish species, resulting in limited detection of some commercially important fish species such as swordfish, Atlantic salmon, mahi-mahi, and yellowfin tuna. Compared to the intensive studies focusing on

evaluating the immunoreactivity of anti-PAV antibodies with northern hemisphere fish species, very few studies focus on southern hemisphere species. The PARV-19 antibody has been the main antibody evaluated for the detection of southern hemisphere species (Beale et al., 2009; Saptarshi et al., 2014; Sharp et al., 2015). It showed limited detection to species from Indo-Pacific region (Saptarshi et al., 2014; Sharp et al., 2015). Several studies have shown that PAV contents vary significantly between species, and that the PAV contents correlate positively with their respective allergenicity (Griesmeier et al., 2010b; Kuehn et al., 2010; Kobayashi et al., 2016d; Kobayashi et al., 2016b; Van Do et al., 2005).

ELISA is the most commonly used method in food industry laboratories and food regulatory bodies to detect and quantify food allergen residues including fish residues in food products (Kirsch et al., 2009; Poms et al., 2004). Most of the ELISAs reported in the literature and the four commercially available ELISA test kits were primarily developed for northern hemisphere fish species (Table 2.7). These test kits have not been validated to detect southern hemisphere fish species.

Therefore, the current situation has highlighted an urgent need to develop suitable antibodies with broad species specificity for the detection of southern hemisphere fish allergens, for more effective management of fish allergy in Australia.

Chapter 3 Quantitative analysis of species specificity of two anti-PAV antibodies for detecting southern hemisphere fish species demonstrating strong phylogenetic association

(Ji Liang, Chui Choo Tan, Steve L. Taylor, Joseph L. Baumert, Andreas L. Lopata & N. Alice Lee. 2017. Quantitative analysis of species specificity of two anti-parvalbumin antibodies for detecting southern hemisphere fish species demonstrating strong phylogenetic association. Food Chemistry, 237, 588-596)

3.1. Introduction

Evaluating the species specificity of anti-PAV antibodies to determine their capability of detecting fish residues from a wide range of southern hemisphere species used processed foods is the first step of developing a sensitive immunoassay targeting southern hemisphere fish allergens with broad cross-reactivity.

The attempts to study species specificity by various antibodies were discussed in Section 2.3.2.7 of Chapter 2. Mahi-mahi, yellowfin tuna, albacore tuna, swordfish and Atlantic salmon were some of the key species that exhibited limited reactivity to the antibodies. Regarding southern hemisphere fish species, PARV-19 was the frequently used antibody for species specificity evaluation, showing limited detection to species from Indo-Pacific region (Beale et al., 2009; Saptarshi et al., 2014; Sharp et al., 2015).

Species specificity is typically studied qualitatively by comparing binding intensity of crude protein extracts of an immunoblot, or by comparing assay absorbance generated from a set amount of crude protein extracts of an qualitative ELISA (Chen et al., 2006; Gajewski and Hsieh, 2009; Lee et al., 2011; Lee et al., 2012; Beale et al., 2009; Saptarshi et al., 2014; Sharp et al., 2015). The potential variations of PAV contents within the crude protein extracts of individual species as well as non-linear dose-response nature of the affinity interaction between an antibody and an antigen in an ELISA have not been considered. Strong evidence has shown that the varying contents of PAV between species positively correlate with their respective allergenicity (Griesmeier et al., 2010b; Kuehn et al., 2010; Van Do et al., 2005; Kobayashi et al., 2016d; Kobayashi et al., 2016b). These studies raised an important question about the 71

immunoreactivity of antibodies raised in host animals against PAV, in that PAV content might be an important factor contributing to the observed inconsistent immunochemical detection.

In order to evaluate analytical performance of two anti-PAV antibodies for detection of southern hemisphere fish residues in processed foods, we investigated a new analytical approach to evaluate species specificity of the antibodies as a function of PAV content. Therefore, this chapter presents the evaluation of immunoreactivity of PAV from thirty-seven southern hemisphere fish species by correlating results from a quantitative indirect ELISA and a quantitative SDS-PAGE method.

3.2. Materials and methods

3.2.1. Materials

3.2.1.1. List of fish species

Thirty-seven fish species from twelve orders were collected from the Sydney Fish Market and the local market in Townsville, Queensland (Table 3.1). The fish species were selected based on fish production, consumption and local availability (ABARES, 2014). Southern hemisphere fish species are defined as fish species that are harvested and commercially available in the southern hemisphere.

Table 3.1 The list of fish species analyzed in this Chapter

Common Name Scientific Name Family Order Distribution Monkfish Squatina australis[1] Squatinidae Squatiniformes S Australian Pilchard Sardinops neophilchardus Clupeidae Clupeiformes N & S /Australian Sardine /Sardinops sagax[2] Atlantic Salmon Salmo salar[1] Salmonidae Salmoniformes N & S Rainbow Trout Oncorhynchus mykiss[1] Salmonidae Salmoniformes N & S Cod Gadus morhua[1] Gadidae Gadiformes N & S Dory Cyttus sp.[1] Cyttidae Zeiformes N & S Bight Redfish Centroberyx gerrardi[3] Berycidae Beryciformes S Sea Mullet Mugil cephalus[3] Mugilidae Mugiliformes N & S Arrhamphus sclerolepis Garfish Hemiramphidae Beloniformes S /Hyporhamphus quoyi[3] Pink Ling Genypterus blacodes[3] Ophidiidae Ophidiiformes S Barramundi Lates calcarifer[4] Latidae Perciformes N & S Blue Threadfin Eleutheronema tetradactylum[3] Polynemidae Perciformes N & S Carangoides ferdau Blue Trevally Carangidae Perciformes N & S /Carangoides orthogrammus[4] Cobia Rachycentron canadum[5] Rachycentridae Perciformes N & S /Black Kingfish Plectropomus leopardus Coral Trout /Plectropomus maculatus Serranidae Perciformes N & S /Variola louti[3] Crimson Snapper Lutjanus erythropterus[3] Lutjanidae Perciformes N & S /Crimson Seaperch Eastern School Sillago flindersi[3] Sillaginidae Perciformes S Whiting Flame Snapper Etelis coruscans[4] Lutjanidae Perciformes N & S /Ruby Snapper Grass Emperor Lethrinus laticaudis[3] Lethrinidae Perciformes S /Grass Sweetlip Pomadasys kaakan Grunter Bream Haemulidae Perciformes N & S /Pomadays argenteus[4] Argyrosomus japonicus Jewfish Sciaenidae Perciformes N & S /Tandanus tandanus[4] Pagrus auratus Pink Snapper Sparidae Perciformes N & S /Chrysophrys auratus[2] Pumpkin Head Trevally Trachinotus blochii[4] Carangidae Perciformes N & S /Snub-nosed dart

Table 3.1 (continued)

Common Name Scientific Name Family Order Distribution Red Emperor Lethrinus miniatus[1] Lethrinidae Perciformes N & S /Redthroat Emperor Sweetlip Emperor Lethrinus sp.[1] Lethrinidae Perciformes N & S Saddletail Snapper Lutjanus malabaricus[3] Lutjanidae Perciformes N & S /Saddletail Seaperch Sand Whiting Sillago ciliata[3] Sillaginidae Perciformes S Scomberomorus commerson Spanish Mackerel Scombridae Perciformes N & S /Scomberomorus semifasciatus[3] Stripped Snapper Lutjanus kasmira[1] Lutjanidae Perciformes N & S Yellowfin Bream Acanthopagrus australis[3] Sparidae Perciformes S Yellowtail Scad Trachurus novaezelandiae[3] Carangidae Perciformes S Swordfish Xiaphias gladius[3] Xiphiidae Scombriformes N & S Yellowfin Tuna Thunnus albacares[3] Scombridae Scombriformes N & S Platycephalus endrachtensis Northern Sand Flathead Platycephalidae Scorpaeniformes N & S /Platycephalus arenarius[3] Chelidonichthys kumu Red Gurnard Triglidae Scorpaeniformes N & S / Chelidonichthys cuculus[1] Neoplatycephalus richardsoni Tiger Flathead Platycephalidae Scorpaeniformes S /Neoplatycephalus aurimaculatus[3] Mosaic Leatherjacket Eubalichthys mosaicus[4] Monacanthidae Tetraodontiformes S N = northern hemisphere; S = southern hemisphere; [1]Sharp et al. (2015); [2]DPI NSW (2014); [3]Yearsley, G. K., Last, P. R., & Ward, R. D. (1999); [4]Pauly, D., & Froese, R. (2016); [5]Fisheries, 2016

3.2.1.2. Chemicals and reagents

The two anti-PAV antibodies selected for this study were rabbit anti-cod PAV PoAb developed by Food Allergy Research and Resource Program (FARRP) of the University of Nebraska-Lincoln (Lee et al., 2011), and mouse anti-carp PAV MoAb (PV 235, Swant Inc, Marly, Switzerland). The anti-cod PoAb was a serum produced by immunizing a rabbit with the purified cod PAV (Lee et al., 2011), and the anti-carp MoAb was a reconstituted lyophilized ascites fluid produced by the mouse hybridoma cells generated from a mouse immunized with the purified carp PAV (Celio et al., 1988). Among the thirty-seven fish species, only four species (cod, swordfish, rainbow trout, and Atlantic salmon) had previously been tested with the anti-cod PoAb and the anti-carp MoAb (Lee et al., 2011).

The other chemicals used in this Chapter include: Bicinchoninic acid (BCA) protein assay kit and bovine serum albumin (BSA) standard (Sigma-Aldrich, St Louis, US); methanol, ethanol, acetic acid and sulfuric acid (Chemical store, UNSW, Australia); sodium dihydrogen phosphate (NaH2PO4) (Honeywell International Inc, Marris Plains,

US); sodium carbonate (Na2CO3), sodium bicarbonate (NaHCO3), di-sodium hydrogen orthophosphate (Na2HPO4), dodecahydrate, sodium chloride (NaCl), sodium acetate (Univar, Downers Grove, US); 40% acrylamide/bis solution; 2x Laemmli sample buffer, N,N,N’,N’-tetramethylethylenediamine (TEMED), dithiothreitol (DTT), sodium dodecyl sulphate (SDS), 10×Tris/glycine buffer/SDS (Bio-Rad, Hercules, US); β-cyclodextrin, urea hydrogen peroxide, Coomassie Brilliant Blue R-250 (Sigma-Aldrich, St Louis, US); Precision Plus ProteinTM Dual Colour Standards (Bio-Rad, Hercules, US); Tween-20, 3,3’,5,5’-tetramethylbenzidine (TMB) (Sigma-Aldrich, St Louis, US); Trans-Blot® Turbo™ 5x Transfer Buffer (Bio-Rad, Hercules, US), anti-rabbit and anti-mouse MoAbs conjugated with horseradish peroxidase (HRP) (Abcam, Cambridge, UK), BSA powder (Bovogen Biologicals Pty Ltd, East Keilor, Australia).

3.2.1.3. Buffers

The following buffer solutions were used and their abbreviation as follows will be used thereafter: coating buffer (50 mM carbonate buffer, pH 9.6); PBS buffer (0.01 M phosphate buffered saline, pH 7.4); blocking buffer 1 (1% BSA in 0.05 M Tris-HCl/0.9% NaCl, with 0.05% Tween-20 (1% BSA/TBS-T), pH 7.5); blocking buffer 2 (1% BSA in PBS buffer (1% BSA/PBS), pH 7.4); diluent buffer (1% BSA/PBS with 0.05% Tween-20 (1% BSA/PBS-T), pH 7.4); stop solution (1.25 M sulphuric acid); and washing buffer (0.05% Tween-20 in reverse osmosis water).

3.2.1.4. Instruments

Heraeus pico 17 centrifuge, Heraeus Multifuge X3R centrifuge (Thermo Fisher Scientific, Waltham, US); Mini high speed refrigerated centrifuge (LabCo); Julabo TW 20 water bath (John Morris Scientific Pty Ltd, Sydney, Australia); Water bath (Retak, Melbourne, Australia); Revolver (Labnet International Inc, Edison, US); Power Pac HV high-voltage power supply (Bio-Rad, Hercules, US); Heating and drying oven (Weiss technik, Königswinter, Germany); Trans-Blot® TurboTM transfer system (Bio-Rad, US);

Optima microplate reader (BMG LABTECH, Ortenberg, Germany); ChemiDocTM MP imaging system (Bio-Rad, Hercules, US); Magic bullet food processor (Homeland Housewares, US).

3.2.2. Methods

3.2.2.1. Identification of fish species

Identification of species was done by comparing the corresponding whole fish images, cross-section images of fish tissue, protein profiles and distribution information with previous study (Sharp et al., 2015), species handbook (Yearsley et al., 1999) and online databases (DPI, 2014; Department of Agriculture and Fisheries, 2016; Pauly and Froese, 2016).

3.2.2.2. Soluble fish protein extraction

The optimized fish protein extraction method was conducted as follows. Approximately 1.5g of fish tissue taken from the dorsal side of the anterior position of the fish body was ground with 20 mL of PBS using a food processor for 30 sec (Lee et al., 2011). Fish proteins were extracted overnight at 4 °C with gentle rolling on a rotating mixer. Supernatants were collected as crude protein extracts after centrifugation at 12,000 x g for 15 min at 4 °C. The protein content of the extract was determined by the BCA protein kit. Extraction was performed in duplicate samples, and the samples were stored at -20 °C until further use.

3.2.2.3. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis (SDS-PAGE)

Proteins were separated by SDS-PAGE using the Bio-Rad Mini PROTEAN Tetra cell (Bio-Rad, Hercules, US). 10 µg of crude protein extract with Bio-Rad sample buffer containing 6% dithiothreitol (DTT, w/v) was heated at 95 °C for 5 min and separated in 15% Tris-HCl hand-casting gels at a constant voltage of 150 V for 75 min. Proteins bands were stained with the Coomassie Brilliant Blue R250. The gel images were captured by the Bio-Rad ChemiDocTM MP imaging system equipped with Bio-Rad Image LabTM software (Bio-Rad, Hercules, US).

3.2.2.4. Immunoblotting analysis

Crude fish protein extracts (5 µg) were separated by SDS-PAGE using the conditions above. After the electrophoresis, the proteins were transferred onto a 0.2 µm nitrocellulose membrane (Bio-Rad, Hercules, US) using Bio-Rad Trans-blot system. After washing the membrane three times with the washing buffer, the membrane was blocked with the blocking buffer 1. After blocking and washing, the anti-cod PoAb or anti-carp MoAb diluted in blocking buffer 1 was incubated with the membrane. After incubation and washing the membrane again, the membrane was incubated with a secondary anti-rabbit or anti-mouse MoAb conjugated with HRP. TMB substrate solution was sued to visualize the bound antibodies and the same signal development time (1 min) was applied to all the blots. The images were captured using the Bio-Rad ChemiDocTM MP imaging system. All blocking and incubation steps were performed for 1 hr at room temperature.

3.2.2.5. PAV identification and relative quantification based on SDS-PAGE

PAV bands were identified based on the specific bindings with the two anti-PAV antibodies (anti-cod PAV PoAb and anti-carp PAV MoAb) on the immunoblot, and the corresponding MWs on SDS-PAGE were calculated using the Bio-Rad Image LabTM software. The Bio-Rad precision plus protein standards with MW distribution ranging from 2 to 250 kDa (2, 5, 10, 15, 20, 25, 37, 50, 100, 250 kDa) were used for the MW calibration.

A quantitative SDS-PAGE based on the method of Kobayashi et al. (2016d) with modification was conducted to estimate immune-recognized PAV content from SDS-PAGE. Five different amounts of the purified cod PAV (2, 2.5, 3, 3.5, and 4 μg/lane) were run on SDS-PAGE for five times. Based on the intensities obtained using the Bio-Rad Image LabTM software, a normalization factor F was calculated as follows for each run:

Xi̅̅̅ Fi = Equation 3.1 Xi

Where Xi̅ is average band intensity of i μg/lane of the purified cod PAV of the five SDS-PAGE runs, Xi is the band intensity of i μg/lane of the purified cod PAV of a 77

individual run. Based on the five different normalization factors F2, F2.5, F3, F3.5 and F4 obtained for each run, a correction coefficient W̅̅̅̅ was calculated as Equation 3.2 for each run:

F + F + F + F + F w̅ = 2 2.5 3 3.5 4 Equation 3.2 5

Then a normalized band intensity Yi of i μg/lane of the purified cod PAV was calculated as Equation 3.3 for each run:

Yi = Xi × W̅̅̅̅ Equation 3.3

By averaging the normalized intensities Yi of the five SDS-PAGE runs, an SDS-PAGE calibration curve (Appendix Figure 2) was established using the calculated Yi̅, which was used for the following relative quantification of PAVs.

3 µg of purified cod PAV (the middle point of the calibration curve) was chosen as a reference to run with 3 µg per lane of all the fish protein extracts on the SDS-PAGE. A correction factor C was applied to normalize all the band intensity unit of test PAV bands.

d C = 3μg Equation 3.4 D3μg

Where d3µg is the band intensity unit of the reference cod PAV obtained from the individual gel; and D3µg is the normalized band intensity obtained from the calibration curve.

Then estimated PAV content was expressed as a quantification index Q relative to the reference cod PAV:

P×C Q = Equation 3.5 D3μg

Where Q is the relative quantity index; P is the band intensity unit of test PAV bands estimated by the Image LabTM software.

3.2.2.6. Quantitative indirect ELISA

Microtiter plates (Thermo Fisher Scientific, Waltham, US) were immobilised with the purified cod PAV in a range of 0.001 - 3.3 μg/mL as the calibration standards. The crude fish protein extracts of each fish species prepared in three concentrations of 0.11, 0.33, 1 μg/mL in the coating buffer were also immobilized in the same manner. Thereafter, all incubation steps, except for the blocking and incubation with the substrate solution, were performed for 30 min at room temperature. After the immobilisation step, the microwell plates were washed 3 times with washing buffer and incubated with 200 µL of blocking buffer 2 for 1 h. After washing the plate, 100 μL per well of the anti-PAV antibodies diluted in the diluent buffer was added to respective wells for incubation. Next, the plates were washed again and incubated with 100 μL per well of HRP conjugated anti-species antibodies in the diluent buffer. In the final washing, the plates were washed at least five times and the TMB substrate solution (100 μL per well) was added to all the wells and incubated for 20 min for colour development. The stop solution (50 μL per well) was added to stop the reaction and the absorbance was measured at 450 nm using the Optima microplate reader. All assays were performed in duplicate at room temperature.

3.2.2.7. Data analysis

A standard calibration curve was fitted to a sigmoidal dose-response (with a variable slope) model with the equation shown as follows using GraphPad Prism® v.7.0; where A is the Y value at the bottom plateau, B is the Y value at the top plateau, C is an EC50 (effective concentration), the concentration of fish protein giving 50% of the maximum response, and D is hillslope describing the slope of the curve (in the most linear part of the curve):

B-A Y = A + [ ] Equation 3.6 1+10(LogC-x)*D

All of the ELISA results were calculated and expressed as cod-PAV equivalent (CPQ) concentrations. A correlation index (CI) was determined using the equation below to assess the correlation between immunoreactivity of PAV and the SDS-PAGE estimated PAV content. 79

E CI = Equation 3.7 Q

E is the fish protein concentration determined by ELISA and Q is the quantity index of PAV bands relative to the reference cod PAV. Cut-off points of PAV quantity were chosen to be ± 50% (50 – 150% or CI of 0.5 – 1.5) for the evaluation based on the recommended recovery of an ELISA for fish allergen detection (Shibahara et al., 2013). Species with correlation index within the range from 0.5 to 1.5 were considered having a positive correlation between immunoreactivity of PAV and PAV content.

3.3. Results and discussion

3.3.1. Protein profile of the crude fish protein extracts by SDS-PAGE and immunoblot analysis

PAV contents are more abundant in the dorsal side than in ventral side of the fish body and decrease from anterior to posterior locations (Lee et al., 2012; Kobayashi et al., 2016d). To minimize variation in the PAV contents in the extracts, tissues used for this study were taken only from the dorsal side of the anterior position of the fish body (Appendix Figure 1). Prior to conducting series of experiments, the consistency of our optimized extraction protocol was tested, which had confirmed no significant differences in the intensity of PAV bands run in replicates from each of 24 selected species (data not shown). As crude protein extracts were susceptible to degradation and cryo-precipitation, fresh aliquots were used in all of the experiments. The protein profiles (Figure 3.1) show the most abundant proteins in the soluble fish protein extracts between 37 and 50 kDa; and 20 fish species had more than one band between 10-15 kDa where (monomeric) PAV bands were (Saptarshi et al., 2014). Our protein profiles agreed well with the results of the same species reported by Saptarshi et al. (2014) and Sharp et al. (2015), except for Spanish mackerel and stripped snapper. Unlike Sharp et al. (2015) who did not observe the PAV band in the crude extract, but did observe a band in the heated extract, we identified a distinct band in the crude extract of Spanish mackerel. Also, their SDS-PAGE image of stripped snapper showed one extra distinct band of 10-15 kDa, compared to our results. The extra protein band may be prone to degradation or oligomerisation, therefore inconsistently showing in different extracts.

The immunoblots of 37 southern hemisphere fish species with anti-cod PAV PoAb and anti-carp PAV MoAb are illustrated in Figure 3.2 and 3.3, respectively. For ease of comparison, the immunoblot images had been arranged according to the phylogenetic distance of the 12 fish orders (Wiley and Johnson, 2010; Nelson, 2006). The yellowfin tuna extract showed only one distinct band at 14 kDa on the SDS-PAGE, which agreed with Chen et al. (Chen et al., 2006). However, the MWs of IgE-reactive proteins of tuna identified in the literature include 12, 32, 46, 40 kDa (aldolase A), and 50 kDa (β-enolase) proteins (Yamada et al., 1999; Bernhisel-Broadbent et al., 1992; Kuehn et al., 2013), and none of these IgE-reactive bands correlate to the 14 kDa band on our SDS-PAGE. Considering the 14 kDa protein was not recognized by the antibodies in our immunoblots and in other previous studies (Chen et al., 2006; Sharp et al., 2015), we concluded that it may not be relevant to PAV or that tuna PAV bound very poorly with anti-cod or anti-carp PAV antibodies. As shown in Figure 3.2 and 3.3, there were two weakly binding bands in swordfish extract by both antibodies in the immunoblots, but there were no corresponding bands on the SDS-PAGE (Figure 3.1), suggesting that these may be low abundant proteins. Interestingly, despite that the rainbow trout PAV has more than 95% homology with the Atlantic salmon PAVs (Peñas et al., 2014), the rainbow trout showed stronger immunoreactivity with both antibodies than Atlantic salmon.

Figure 3.2 Immunoblot analysis of the crude fish protein extracts with the anti-cod PoAb.

Figure 3.3 Immunoblot analysis of the crude fish protein extracts with the anti-carp MoAb.

The two antibodies showed a similar binding trend, but with varying degree of binding intensity corresponding to their immunoreactivity. Our immunoblot analysis agreed well with Saptarshi et al. (2014) in that immunoreactivity among the test fish species demonstrated correlation with the molecular phylogeny of PAV. For example, fish species from the order of Perciformes showed strong immunoreactivity with both antibodies; whereas species from Salmoniformes, Ophidiiformes, Scombriformes, Scorpaeniformes, Tetraodontiformes including salmon, pink ling, swordfish, yellowfin tuna, red gurnard, tiger flathead, and mosaic leatherjacket showed weak or no immunoreactivity. Lee et al. (2011) and Sharp et al. (2015) also observed similar order specificity in their immunoblot analysis. Lee et al. (2011) showed that all of the test northern hemisphere fish species from Perciformes were recognized strongly by the anti-cod PAV antibody and the anti-carp PAV antibody. Mahi-mahi, swordfish and albacore tuna and the species from Salmoniformes showed low antibody binding. Sharp et al. (2015) demonstrated that most southern hemisphere fish species from Perciformes showed bindings to the PARV-19 antibody; while species from Salmoniformes and Ophidiiformes showed no binding.

Our immunoblot results showed that the anti-cod PoAb showed broader specificity than that of the anti-carp MoAb for the southern hemisphere species. This was expected, as polyclonal antibodies are likely to bind more epitopes on the same protein while monoclonal antibodies can only bind one single epitope. Interestingly, the anti-carp MoAb showed binding with 36 kDa protein for most fish species in our study. The identification of PAV bands in the SDS-PAGE (Figure 3.1) was guided by the binding of PAV specific antibodies in the immunoblots (Chen et al., 2006; Lee et al., 2011; Lee et al., 2012; Kuehn et al., 2010; Celio et al., 1988; Hemmi et al., 1997). Table 3.2 summarizes the calculated molecular weight (MW) of monomeric PAV bands that were immunochemically detected by the two antibodies. Of the thirty-seven species in this study, only nine species (Australian pilchard, barramundi, cod, cobia, swordfish, yellowfin bream, yellowfin tuna, Atlantic salmon and rainbow trout) have PAV amino acid sequences and molecular characteristics documented in the Allergome database and five among them (cod, swordfish, yellowfin tuna, Atlantic salmon and rainbow trout) have records in the database of WHO/IUIS Allergen Nomenclature and AllergenOline.

Considering there are still many species in our list with uncharacterized PAVs, we decided to mainly focus on monomeric PAV immunologically recognized by the two antibodies as identified by the immunoblots. Noticeably, several proteins between 22 – 25 kDa in monkfish, blue trevally, grunter bream and yellowtail scad showed strong bindings to the antibodies (Figures 3.2 and 3.3). These immunoreactive bands are considered to be PAV oligomers (Saptarshi et al., 2014), which were also included in the estimation of PAV content. The formation of PAV oligomers can be attributed to the readily reducible cysteine residues in PAV.

Table 3.2 The calculated MW of PAV bands detected by the anti-cod PoAb and the anti-carp MoAb in fish protein extracts Molecular weight (kDa) of PAV Fish samples 1 2 3 4 5 Monkfish 25.2 15.5 13.5 12.4 11.4 Australian Pilchard 11.7 10.8 10.3 Atlantic Salmon 13.4 11.8 Rainbow Trout 11.8 11.0 Cod 11.9 Dory 12.5 11.1 Bight Redfish 11.8 Sea Mullet 10.8 Garfish 12.8 12.2 11.4 10.8 Pink Ling 14.8 10.5 Barramundi 11.8 11.1 Blue Threadfin 12.8 11.4 Blue Trevally 26.1 11.9 11.4 Cobia 11.9 Coral trout 12.2 Crimson Snapper 12.3 11.4 Eastern School Whiting 11.2 10.7 Flame Snapper 11.3 Grass Emperor 11.6 11.1 Grunter Bream 21.8 10.9 Jewfish 11.9 Pink Snapper 13.2 11.1 Pumpkin Head Trevally 11.5 Red Emperor 11.7 11.2 10.7 Sweetlip Emperor 12.2 11.6 Saddletail Snapper 11.4 10.8 Sand Whiting 11.7 10.9 Spanish Mackerel 10.6 Stripped Snapper 11.9 11.3 Yellowfin Bream 10.8 Yellowtail Scad 22.6 11.4 Swordfish Yellowfin Tuna Northern Sand Flathead 11.6 Red Gurnard 12.7 Tiger Flathead 12.2 Mosaic Leatherjacket 12 11.4 The MW of PAV detected only by the anti-cod PoAb was shaded as light grey; MW of PAV detected only by the anti-carp MoAb was shaded as medium grey; MW of PAV detected by both the two antibodies was shaded as dark grey.

3.3.2. Indirect ELISA

3.3.2.1. Qualitative indirect ELISA

Species specificity of the two antibodies for the thirty-seven test fish species was evaluated using two analytical approaches. The first was a conventional approach by performing a qualitative indirect ELISA based on immobilized fish proteins, and examining the correlation between the absorbance values and the estimated PAV content as described in the method section (Figure 3.4). As shown in Figure 3.4, the anti-cod PoAb showed some trend of non-linear correlation, and the anti-carp antibody showed no correlation between absorbance values and the estimated PAV content. Noticeably, many of the species had high absorbance values, indicating binding saturation, therefore not allowing clear differentiation between low immunoreactive and high immunoreactive species.

(a) (b)

Figure 3.4 The correlation between relative quantity of PAV estimated from SDS-PAGE and absorbance values of ELISA for 1 μg crude protein using (a) anti-cod PoAb and (b) anti-carp MoAb

3.3.2.2. Quantitative indirect ELISA indicating correlation between PAV content and the immunoreactivity

As the dose-response relationship of an ELISA is sigmoidal, we took a second approach of constructing a quantitative ELISA to better estimate the quantifiable immunological responses from a fixed amount of crude protein extracts of different species, and subsequently, correlate the immunoreactivity with the SDS-PAGE estimated PAV 88

content. As purified cod PAV was used as a reference to quantify PAV contents on SDS-PAGE, it was also used as a standard in our quantitative ELISA with both antibodies (Appendix Figures 2 and 3), in order to establish a correlation between ELISA (i.e., immunoreactivity) and SDS-PAGE (i.e., relative quantity). The quantification range of the quantitative ELISA with the anti-cod PoAb was established between 0.2 – 3.4 µg/mL and the quantification range with the anti-carp MoAb was between 0.8 – 2.8 µg/mL.

According to the anti-cod PoAb ELISA (Table 3.3 and Figure 3.5a and 3.5b), twenty-five out of thirty-seven fish samples showed quantifiable responses (i.e., within the linear quantification range), and sixteen of the twenty-five samples (43% of total samples), shading as grey in Table 3, had a correlation index within the cut-off points of 0.5 – 1.5, showing a positive correlation between the PAV content and the immunoreactivity (R2 = 0.74). The correlation index of yellowfin tuna and swordfish is 1.00 because the absence of PAV bands on SDS-PAGE positively correlates with the “undetected” in the ELISA. Interestingly, there were no observable bands on the yellowfin tuna immunoblot, while there were two weakly visible bands on the swordfish immunoblot. For species that had responses below the limit of detection of the ELISA (< 0.2 µg/mL), an arbitrary concentration of 0.1 µg/mL was used only to calculate the corresponding correlation index. Noticeably, Atlantic salmon and pink ling had low reactivity (< 0.2 µg/mL) with the anti-cod PoAb, which agreed with the weak bands in the immunoblot. Conversely, although strong binding was observed in the immunoblot of rainbow trout, cobia, crimson snapper, flame snapper, jewfish, pumpkin head trevally and red gurnard, these fish showed low ELISA responses (< 0.2 µg/mL). Examining the species from the order of Perciformes with low ELISA responses, it was noted that they had relatively low PAV, which contributed to the low ELISA responses. Australian pilchard, which showed three distinct PAV isoforms on the SDS-PAGE (Figure 3.1) and in the immunoblot (Figure 3.2), was reported to have the greatest IgE reactivity among five fish species by ten fish-allergic patients (Beale et al., 2009). However, it showed very low immunoreactivity (< 0.2 µg/mL) with the anti-cod PoAb by the quantitative ELISA (Table 3.3).

Table 3.3 The relative quantity of PAV estimated by SDS-PAGE and quantitative ELISA analysis of fish protein extracts using anti-cod PoAb (ELISA 1) and using anti-carp MoAb (ELISA 2) SDS-PAGE ELISA 1 ELISA 2 Fish species CI-1 CI-2 RPQ CPQ (µg/mL) CPQ (μg/mL) Monkfish 1.8 0.7 0.39 <0.8 0.39 Australian Pilchard 2.2 <0.2 0.05 0.9 0.41 Atlantic Salmon 0.4 <0.2 0.25 <0.8 1.75 Rainbow Trout 0.7 <0.2 0.14 <0.8 1.00 Cod 0.7 1.1 1.47 1.2 1.68 Dory 1.5 2.2 1.50 <0.8 0.47 Bight Redfish 0.4 0.2 0.57 3.5 9.97 Sea Mullet 0.5 0.3 0.64 3.5 7.33 Garfish 1.7 0.8 0.44 2.9 1.66 Pink Ling 1.0 <0.2 0.10 0.0 0.00 Barramundi 1.0 0.8 0.79 3.4 3.42 Blue Threadfin 1.3 0.5 0.37 0.8 0.64 Blue Trevally 0.9 0.2 0.22 0.9 1.00 Cobia 0.3 <0.2 0.33 <0.8 2.33 Coral Trout 0.6 0.5 0.93 <0.8 1.17 Crimson Snapper 0.2 <0.2 0.50 <0.8 3.50 Eastern School Whiting 1.1 1.1 1.00 2.1 2.01 Flame Snapper 0.2 <0.2 0.50 <0.8 3.50 Grass Emperor 0.8 1.5 1.83 5.3 6.59 Grunter Bream 0.5 0.3 0.57 2.4 4.63 Jewfish 0.5 <0.2 0.20 <0.8 1.40 Pink Snapper 0.9 0.5 0.48 1.0 1.12 Pumpkin Head Trevally 0.2 <0.2 0.50 <0.8 3.50 Red Emperor 1.2 1.2 0.95 3.9 3.10 Sweetlip Emperor 1.1 1.5 1.38 2.9 2.69 Saddletail Snapper 0.9 0.8 0.97 3.0 3.50 Sand Whiting 1.8 1.4 0.77 2.8 1.59 Spanish Mackerel 1.0 1.1 1.03 2.7 2.63 Stripped Snapper 1.5 2.4 1.66 6.3 4.31 Yellowfin Bream 1.0 1.3 1.26 5.7 5.60 Yellowtail scad 1.2 1.9 1.58 6.1 5.08 Swordfish 0.0 0.0 1.00 0.0 1.00 Yellowfin Tuna 0.0 0.0 1.00 <0.8 0.00 Northern Sand Flathead 0.5 0.4 0.86 0.0 0.00 Red Gurnard 0.9 <0.2 0.11 <0.8 0.77 Tiger Flathead 0.8 0.2 0.26 0.0 0.00 Mosaic Leatherjacket 0.3 0.3 1.07 0.0 0.00 RPQ: Relative Quantity of PAV; CPQ: Cod PAV Equivalent. 90

(a) (b)

With respect to the anti-carp MoAb (Table 3.3 and Figure 3.6), twenty fish samples showed quantifiable results. For species that had results below the limit of detection of the ELISA (<0.8 µg/mL), 0.7 µg/mL was used to calculate the corresponding correlation index. Interestingly, the quantitative ELISA responses did not show any correlation with the PAV content, even though the overall binding pattern of the anti-carp MoAb shared close similarity with that of the anti-cod PoAb. Except for cobia, crimson snapper, flame snapper, jewfish and, pumpkin head trevally, all of the other species from the order of Perciformes presented strong ELISA responses. The species from the order of Salmoniformes, Ophidiiformes, Scombriformes, Scorpaeniformes,

Tetraodontiformes including Atlantic salmon, rainbow trout, pink ling, swordfish, yellowfin tuna, sand flathead, red gurnard, tiger flathead and mosaic leatherjacket showed either low or no reactivity.

The low immunoreactivity of swordfish observed in our study and in other studies (Lee et al., 2011; Sharp et al., 2015) is likely due to low PAV content. The absence of 12 kDa tuna PAV, which have been reported previously (Kuehn et al., 2010; Kawase et al., 2001;

Shiomi et al., 1999), could be due to low abundance in our extract. The weak reactivity of Atlantic salmon observed in our study and other studies (Lee et al., 2011; Sharp et al.,

2015), suggests the detection of salmon PAV probably requires a separate highly specific antibody because the antigenic region of salmon PAV is found to be the least conservative region of the protein (Saptarshi et al., 2014; Sharp et al., 2015). Among the studied fish species with detected PAVs by both the antibodies, low IgE reactivity

(allergenicity) of Atlantic salmon, swordfish and yellowfin tuna reported in previous studies (Griesmeier et al., 2010b; Kobayashi et al., 2016d) is believed to correlate with the low PAV contents. Though clear bands were observed for Australian pilchard in the immunoblot and high absorbance was observed in the qualitative ELISA, pilchard had very low quantitative responses in the quantitative ELISA with both antibodies (i.e.,

<0.2 µg/mL with the anti-cod PoAb and <0.8 µg/mL with the anti-cod PoAb, Table 3.3).

The low quantitative immunological responses of pilchard might attribute to the low stability of the soluble pilchard proteins. These contradictory results have further proved 92

that the qualitative ELISA overestimated the immunological responses of some fish species due to binding saturations, whereas the quantiataive ELISA was able to assess the immunological response of different fish species statically.

The anti-carp MoAb is reported to be a Ca2+ dependent antibody that recognizes one of the two Ca2+ binding sites on PAV (Tinner et al., 1990). Even though Ca2+ binding sites are reported to be the most conservative regions of PAV (Chen et al., 2006), our study together with other studies (Lee et al., 2011; Chen et al., 2006) have demonstrated that conservative protein structure could not guarantee similar binding capacity of anti-carp

MoAb and PARV-19 with certain species. The possible binding to other less conservative sites might have led to the discrete detection in some cases. The binding of the anti-carp MoAb to the unknown 36 kDa protein band may also be contributing to the frequently observed inconsistent results. This, on the other hand, highlights the advantage of anti-cod PoAb that it is able to recognize multiple epitopes of PAV in different species, allowing for better quantification.

3.4. Conclusion

The analytical approach based on the quantitative ELISA and quantitative SDS-PAGE enables us to assess the correlation between the immunoreactivity of PAV and PAV content statically and consequently evaluate species specificity of the two PAV specific antibodies for southern hemisphere fish. By examining the correlation index, we could differentiate species showing positive correlation from those showing no correlation.

Sixteen of the thirty-seven species (43%) exhibited a positive correlation between the immunoreactivity of PAV and PAV content with the anti-cod PoAb, while there was no such a correlation could be established with the anti-carp MoAb using this approach.

The sixteen fish species showing positive correlation are cod, dory, bight redfish, sea mullet, barramundi, coral trout, eastern school whiting, grunter bream, red emperor, sweetlip emperor, saddletail snapper, sand whiting, Spanish mackerel, yellowtin bream, 93

northern sand flathead, and mosaic leatherjacket.

The anti-cod PAV PoAb outperformed the anti-carp PAV MoAb, showing better species specificity for the thirty-seven commercially important fish from the southern hemisphere, as well as quantitatively detecting parvalbumin. Our results along with other related studies based on northern hemisphere species have confirmed that fish species from the order of Perciformes showed strong binding with the anti-PAV antibodies; whereas species from Salmoniformes, Ophidiiformes, Scombriformes,

Scorpaeniformes, and Tetraodontiformes showed weak or no binding. Order specificity in relation to PAV detection and potential for developing a new immunoassay with broad species specificity should be explored further.

Chapter 4 Impacts of thermal processing on the immunoreactivity of PAV with two anti-PAV antibodies

4.1. Introduction

Thermal processing plays an important role in food preservation as well as in modifications of texture, tastes, flavour, and digestibility of food products. Heat-induced structural changes of proteins, including loss of secondary and tertiary structures and molecule aggregation altering the antibody binding epitopes, thus affecting the antibody recognition of allergenic proteins (Davis and Williams, 1998). Therefore, thermal processing complicates the detection of PAV in processed forms (Sharp et al., 2015).

Several studies suggested that the allergenicity of PAV could be sustained up to 100 °C

(Bernhisel-Broadbent et al., 1992; Chatterjee et al., 2006; Beale et al., 2009; Griesmeier et al., 2010a; de Jongh et al., 2015). Increased IgE reactivity after heat treatment has been reported for many fish species including catfish, bass, tuna, flounder, barramundi, pilchard, anchovy, hake, snoek, and yellowtail (Seriola lalandi) (Chatterjee et al., 2006; Beale et al., 2009; Bernhisel-Broadbent et al., 1992). In this regard, food processing such as heating and frying may pose higher risks of allergic reactions to fish allergic consumers. To prevent such adverse reactions from allergenic residues in foods, effective detection methods albe to detect thermally stable PAV especially from commercially important fish species is urgently needed.

Four qualitative ELISAs targeting heated fish extracts have been reported (Chen and

Hsieh, 2014; Gajewski and Hsieh, 2009; Gajewski et al., 2009; Sharp et al., 2015).

Apart from two ELISAs especially targeted a heat-stable 36 kDa protein (Gajewski et al., 2009; Chen and Hsieh, 2014), the other two ELISA targeted detection of heat-stable crude fish proteins including PAV from a wide range of species from both northern and southern hemisphere (Gajewski and Hsieh, 2009; Sharp et al., 2015). Similar to the

detection of crude fish proteins, the detection of heated fish proteins was also limited to certain fish species such swordfish, yellowfin tuna and some commonly consumed southern hemisphere fish species such as jewfish and gurnard (Gajewski and Hsieh,

2009; Sharp et al., 2015). It is hard to predict the immunoreactivity of proteins from widely different fish species, since IgG responses to heat treated fish allergens/proteins could vary widely among different fish species.

In Chapter 3, a positive correlation between the native PAV content and the immunoreactivity with the anti-cod PoAb for the native PAV of sixteen fish species was demonstrated, but no correlation was observed with the anti-carp MoAb for the same species (Liang et al., 2017). This Chapter aims to evaluate the relationship between the quantity of thermally stable PAV and their corresponding immunoreactivity, and to investigate the impacts of thermal processing on the detectability of PAV from southern hemisphere fish species.

4.2. Materials and methods

4.2.1. Materials

4.2.1.1. List of fish species

The detailed list of fish species can be referred to Section 3.2.1.1 of Chapter 3.

4.2.1.2. Chemicals and reagents

Apart from ethylenediaminetetraacetic acid (EDTA) (Univar, Downers Grove, US), all the other chemicals can be referred to Section 3.2.1.2 of Chapter 3.

4.2.1.3. Instruments

Heraeus pico 17 centrifuge, Heraeus Multifuge X3R centrifuge (Thermo Fisher

Scientific, Waltham, US); Mini high speed refrigerated centrifuge (LabCo); Julabo TW

20 water bath (John Morris Scientific Pty Ltd, Sydney, Australia); Water bath (Retak,

Melbourne, Australia); Revolver (Labnet International Inc, Edison, US); Power Pac HV high-voltage power supply (Bio-Rad, US); Heating and drying oven (Weiss technik,

Königswinter, Germany); Trans-Blot® TurboTM transfer system (Bio-Rad, US); Optima microplate reader (BMG LABTECH, Ortenberg, Germany); Cary Eclipse Fluorescence

Spectrophotometer (Agilent Techonologies, Santa Clara, US); ChemiDocTM MP imaging system (Bio-Rad, Hercules, US); Magic bullet food processor (Homeland

Housewares, US).

4.2.2. Methods

4.2.2.1. Soluble fish protein extraction

Extraction of crude fish proteins followed the method described in Section 3.2.2.1 of

Chapter 3.

4.2.2.2. Heat treatment of crude fish protein extracts

The optimisation of the heat treatment condition is discussed in the Appendix

(Appendix Figures 5 to 7). The heat treatment of crude fish protein extracts was conducted as follows. Aliquots (5 ml) of crude fish protein extracts in 15 mL glass tubes heated at 100 °C for 45 mins. After the extracts cooled down to room temperature, they were centrifuged at 13,000 × g at 4 °C for 15 min. Soluble supernatant was collected as the heated fish protein extracts. Protein concentrations were checked with the BCA protein assay kit. All heated extracts were aliquoted and stored at -20 °C until use. New aliquots stored at -20 °C were used in all experiments.

4.2.2.3. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis

(SDS-PAGE)

Heat-treated fish proteins (5 μg per each fish sample) were separated by SDS-PAGE

based on the method described in Section 3.2.2.3 of Chapter 3. The consistency of the protein profiles was confirmed by running the SDS-PAGE of all heated extracts in replicates (data not shown).

4.2.2.4. Immunoblotting analysis

Heated fish proteins (5 μg per each fish sample) were separated by SDS-PAGE using the conditions in section 3.2.2.3 of Chapter 3. The separated proteins were transferred onto a 0.2 µm nitrocellulose membrane using the Bio-Rad Trans-blot system and the immunoblotting analysis was performed based on the method presented in Section

3.2.2.4 of Chapter 3.

4.2.2.5. Thermally stable PAV identification of relative quantification based

SDS-PAGE

The identification and relative quantification of thermally stable PAV bands on

SDS-PAGE were according to the methods stated in Section 3.2.2.5 of Chapter 3.

4.2.2.6. Quantitative indirect ELISA

The immunoreactivity of the thermally stable PAV was assessed by a quantitative indirect ELISA based on the method described in Section 3.2.2.6 of Chapter 3.

4.2.2.7. Data analysis

A sigmoidal dose-response (with a variable slope) model was incorporated in the data analysis. Detailed analysis method can be referred to Section 3.2.2.7 of Chapter 3.

4.2.2.8. Structure investigation of the fish protein extracts by fluorescence spectroscopy

The fluorescence spectra of crude and heated fish protein extracts at a concentration of

100 μg/mL were recorded from 300 to 450 nm upon excitation at 280 nm using the Cary

Eclipse Fluorescence Spectrophotometer. The scanning speed was 600 nm/min, and the slit width was 5 nm.

4.3. Results and discussion

4.3.1. Protein profile of the heated fish protein extracts by SDS-PAGE and immunoblot analysis

Compared with the crude fish protein profiles (Figure 3.1) shown in Chapter 3, most of the high MW proteins were denatured and became insoluble; the predominant heat-stable proteins remained the solution after heat treatment were proteins at around

36 kDa and PAV bands at 10 – 15 kDa (Figure 4.1), which is in agreement with the previous studies (Gajewski and Hsieh, 2009; Saptarshi et al., 2014; Sharp et al., 2015).

Most of the protein profiles at 10 – 15 kDa agreed with the results of the same species studied by Saptarshi et al. (2014) and Sharp et al. (2015), except for pilchard, cod, rainbow trout, Spanish mackerel, stripped snapper, coral trout and monkfish. Compared with the results of Saptarshi et al. (2014) and Sharp et al. (2015), our SDS-PAGE image showed more heat-stable bands at 10 – 15 kDa in pilchard, Spanish mackerel, stripped snapper, coral trout and monkfish. Conversely, there were more bands in cod and rainbow trout reported by Saptarshi et al. (2014) and Sharp et al. (2015), compared to our results. We observed a faint band at around 11 kDa in the heated yellowfin tuna extract, whereas there was no band at 11 kDa reported by Sharp et al. (2015). Saptarshi et al. (2014) and Sharp et al. (2015) demonstrated more heat-stable bands at higher MW region above 37 kDa than what we obtained in our study. The differences in protein profiles observed might be because we heated the crude fish extracts at a higher temperature (100 °C) for a longer time (45 min), while the studies of Saptarshi et al.

(2014) and Sharp et al. (2015) heated fish tissues 95 °C for 15 min.

Figure 4.1 Protein profiles of the heated fish protein extracts separated by SDS-PAGE with Coomassie brilliant blue staining. The red rectangles indicate the PAV bands that were recognized by the two anti-PAV antibodies collectively in the immunoblot.

100

The immunoreactive monomeric PAV bands were identified by the immunoblot analysis with the two anti-PAV antibodies (Figure 4.2 and 4.3). The corresponding MWs of the immunoreactive bands are listed in Table 4.1. Compared with the MW list in crude fish extracts (Table 3.2) in Chapter 3, the PAV in each fish species responded to the thermal processing differently. Among the thirty-seven tested species, twenty-two fish species showed consistent immunoreactive PAV bands before and after heat treatment. They are pilchard, cod, dory, bight redfish, sea mullet, pink ling, barramundi, blue threadfin, blue trevally, cobia, crimson snapper, flame snapper, grunter bream, jewfish, pumpkin head trevally, red emperor, sweetlip emperor, saddletail snapper, stripped snapper, yellowfin bream, yellowtail scad, sand flathead, and red gurnard. Five fish species exhibited reduced immunoreacitivity of the PAV bands after heat treatment and they are monkfish,

Atlantic salmon, rainbow trout, blue trevally, red emperor, and mosaic leatherjacket.

The rest of the fish species, except for swordfish, demonstrated greater immunoreactivity of the PAV bands after the heat treatment. More specifically, several proteins between 22 – 25 kDa in garfish, eastern school whiting, grass emperor, sand whiting and yellowtail scad, which did no show immunoreactivity in the crude extracts, showed reactivity with both the antibodies after heat treatment, and they were also included in the estimation of PAV content. These newly occurring proteins at 22 – 25 kDa were suspected to correlate with the PAV dimers. The heat-induced PAV dimers have been reported by the previous studies (Cai et al., 2010; Liu et al., 2010b; Saptarshi et al., 2014; Griesmeier et al., 2010a). Coral trout, Spanish mackerel, and tiger flathead showed two immunoreactive PAV bands in the heated extracts, compared with the corresponding crude extracts with only one immunoreactive PAV band. This is likely due to the degradation of PAV upon the heat treatment. Degradation of cod PAV has been reported in the autoclave condition but not boiling at 100 °C (de Jongh et al.,

2013). The degradation observed in our study may be attributed to the longer heating time (45 min). 101

Figure 4.2 Immunoblot analysis of the heated fish protein extracts with the anti-cod PoAb.

102

Figure 4.3 Immunoblot analysis of the heated fish protein extracts with the anti-carp MoAb.

103

Table 4.1 The calculated MW of thermally stable PAV bands detected by the anti-cod PoAb and the anti-carp MoAb in the heated fish protein extracts

Molecular weight (kDa) of PAV Fish samples 1 2 3 4 Monkfish 25.2 13.5 12.4 11.4 Australian Pilchard 11.7 10.8 10.3 Atlantic Salmon 11.8 Rainbow Trout 11.8 Cod 11.9 Dory 12.5 11.1 Bight Redfish 11.8 Sea Mullet 10.8 Garfish 22.4 12.2 11.4 10.8 Pink Ling 14.8 10.5 Barramundi 11.8 11.1 Blue Threadfin 12.8 11.4 Blue Trevally 11.9 Cobia 11.9 Coral trout 12.2 11.7 Crimson Snapper 12.3 11.4 Eastern School Whiting 23.8 11.2 10.7 Flame Snapper 11.3 Grass Emperor 25.8 11.6 11.1 Grunter Bream 21.8 10.9 Jewfish 11.9 Pink Snapper 13.2 11.1 Pumpkin Head Trevally 11.5 Red Emperor 11.7 11.2 Sweetlip Emperor 12.2 11.6 Saddletail Snapper 11.4 10.8 Sand Whiting 22.4 11.7 10.9 Spanish Mackerel 10.6 10.2 Stripped Snapper 11.9 11.3 Yellowfin Bream 10.8 Yellowtail Scad 22.6 11.4 Swordfish Yellowfin Tuna 11.3 Northern Sand Flathead 11.6 Red Gurnard 12.7 Tiger Flathead 11.2 10.6 Mosaic Leatherjacket 11.4 The MW of PAV detected only by the anti-cod PoAb is shaded as light grey; MW of PAV detected only by the anti-carp MoAb is shaded as medium grey; MW of PAV detected by both antibodies is shaded as dark grey.

104

The thermal processing affected IgG recognition in many ways (Figures 4.2 and 4.3).

Compared with the immunoblots of the crude fish extracts (Figure 3.2) shown in

Chapter 3, the detectability of the anti-cod PoAb was not greatly affected by the heat treatment, and strong bindings were still visible by most of the species except for swordfish, yellowfin tuna, sand flathead and mosaic leatherjacket. For some species which had weak or no bindings in the crude fish extracts, their PAVs were better recognised by the anti-cod PoAb in the denatured forms. These fish species include

Atlantic salmon, bight redfish, pink ling, pink snapper and tiger flathead. Interestingly, the anti-cod PoAb showed more bindings with proteins at around 25 kDa for most heated fish samples in our study, which was not previously observed in the crude fish samples (Liang et al., 2017).

On the other hand, the detectability of the anti-carp MoAb was greatly affected by the heat treatment, compared with its performance against the crude fish extracts (Figure

3.3). The anti-carp MoAb lost detectability to monkfish, rainbow trout, jewfish, pumpkin head trevally, red gurnard, and swordfish. Atlantic salmon, pink ling, sand flathead, tiger flathead and mosaic leather, which did not react with the anti-carp MoAb in the crude fish extracts, continued to show no bindings to the MoAb. Surprisingly, the weak band shown in the heated yellowfin tuna extract was detected by the MoAb.

Swordfish and yellowfin tuna were the two species showing limited detection by several studies (Gajewski et al., 2009; Gajewski and Hsieh, 2009; Sharp et al., 2015). In our study, we observed that crude swordfish protein extract was highly heat labile. After heat treatment, only the 36 kDa protein was observed in the SDS-PAGE (Figure 4.1), which explained why no antibody binding to PAV was observed in the heated swordfish fish extracts. Our observation has also agreed with the previous studies (Gajewski and

Hsieh, 2009; Gajewski et al., 2009; Sharp et al., 2015). Two MoAbs F1G11 and F7B8 developed against heated tra extract were able to detect heated yellowfin tuna extract

105

(Gajewski et al., 2009); and the PARV-19 was reported to demonstrate strong bindings to both heated light and dark tuna muscles (Saptarshi et al., 2014). This evidence suggests that yellowfin tuna PAV might be detected more easily in the denatured form than the native counterpart. The anti-carp MoAb weakly reacted to the heated tuna extract while the anti-cod PoAb did not react at all. This immunoreactive PAV band observed in our heated tuna extract correlated with the tuna PAV reported in the previous studies (Kuehn et al., 2010; Kawase et al., 2001; Shiomi et al., 1999).

Similar order specificity was more prominent in the heated fish extracts than the crude fish extracts. Despite some species showed no reactivity to the anti-carp MoAb, species from the order of Perciformes remained strongly immunoreactive to both antibodies; whereas species from Scombriformes, Scorpaeniformes and Tetraodontiformes showed weak or no immunoreactivity. Compared to the crude extracts, increased binding intensity with the anti-cod PoAb in the heated extracts was observed in species from

Salmoniformes and Ophidiiformes including Atlantic salmon, rainbow trout and pink ling. The study by Sharp et al. (2015), focusing on heated fish extracts, also observed similar order specificity. The anti-basa PoAb, anti-pilchard PoAb and anti-salmon PoAb showed stronger reactivity to the heated protein of species from Perciformes, while they showed weak or no reactivity to the species from Scombriformes, Scorpaeniformes and

Tetraodontiformes (Sharp et al., 2015). The anti-barramundi PoAb, which was raised against two distantly related barramundi PAV isoforms (β1 and β2, Figure 2.5), showed the broadest cross-reactivity to the tested species (Sharp et al., 2015). It displaced detection to thirty-five crude extracts and thirty-four heated extracts out of the forty-five fish samples tested (Sharp et al., 2015). Though the anti-barramundi PoAb did not show binding to raw extracts of species from Salmoniformes and Ophidiiformes including

Atlantic salmon, rainbow trout and pink ling in the immunoblots, it was able to qualitatively detect both raw and heated extracts of the same species by the indirect

106

ELISA (Sharp et al., 2015).

4.3.2. Quantitative indirect ELISA

Following the same approach described in Chapter 3, the quantitative indirect ELISA was also conducted with the heated extracts, to evaluate a correlation between the immunoreactivity of thermally stable PAV (ELISA) and PAV content, using the purified cod PAV as the reference. The quantification range of the ELISA with the anti-cod

PoAb was established between 0.02 – 0.4 μg/mL (Appendix Figure 4) and the quantification range with the anti-carp MoAb was between 0.8 – 2.8 μg/mL (Appendix

Figure 5). Table 4.2 illustrates the relative quantity of PAV estimated by SDS-PAGE and the immunoreactivity quantified by the ELISAs with the anti-cod PoAb and anti-carp

MoAb.

107

Table 4.2 The relative quantity of thermally stable PAV estimated by SDS-PAGE and the quantitative ELISA analysis of heated fish protein extracts using anti-cod PoAb (ELISA-A) and using anti-carp MoAb (ELISA-B)

SDS-PAGE ELISA-A ELISA-B Fish species CI-A CI-B RPQ CPQ (µg/mL) CPQ (μg/mL) Monkfish 1.8 0.2 0.09 <0.8 0.39 Australian Pilchard 1.4 0.05 0.04 <0.8 0.50 Atlantic Salmon 0.2 0.05 0.21 <0.8 3.04 Rainbow Trout 0.6 0.06 0.10 <0.8 1.15 Cod 1.5 1.6 1.05 1.0 0.67 Dory 1.3 1.2 0.94 <0.8 0.55 Bight Redfish 1.3 0.3 0.24 <0.8 0.54 Sea Mullet 1.0 0.2 0.22 1.5 1.56 Garfish 2.5 2.4 0.96 4.5 1.77 Pink Ling 1.1 0.1 0.10 0.0 0.00 Barramundi 2.7 2.7 0.99 4.5 1.69 Blue Threadfin 1.3 0.3 0.22 <0.8 0.64 Blue Trevally 0.3 0.3 0.97 1.2 4.00 Cobia 0.4 0.3 0.81 1.1 2.99 Coral Trout 1.4 0.6 0.42 <0.8 0.50 Crimson Snapper 1.2 0.04 0.04 <0.8 0.60 Eastern School Whiting 2.1 1.7 0.81 1.4 0.67 Flame Snapper 1.0 0.06 0.06 <0.8 0.74 Grass Emperor 2.0 1.0 0.51 3.2 1.64 Grunter Bream 0.8 0.02 0.03 <0.8 0.84 Jewfish 0.4 <0.02 0.03 0.0 0.00 Pink Snapper 1.0 0.3 0.28 <0.8 0.73 Pumpkin Head Trevally 0.5 <0.02 0.02 0.0 0.00 Red Emperor 1.4 0.3 0.24 1.0 0.73 Sweetlip Emperor 1.2 0.7 0.57 1.1 0.92 Saddletail Snapper 1.1 0.8 0.71 1.8 1.56 Sand Whiting 1.1 0.9 0.83 1.1 0.99 Spanish Mackerel 1.1 1.0 0.89 3.5 3.14 Stripped Snapper 2.4 2.9 1.21 3.8 1.62 Yellowfin Bream 1.7 1.1 0.67 3.3 1.92 Yellowtail scad 0.9 0.2 0.21 1.1 1.24 Swordfish 0.0 <0.02 0.00 <0.8 0.00 Yellowfin Tuna 0.1 0.0 0.00 <0.8 7.00 Northern Sand Flathead 0.7 0.06 0.09 0.0 0.00 Red Gurnard 1.3 0.08 0.06 <0.8 0.52 Tiger Flathead 1.3 0.1 0.09 0.0 0.00 Mosaic Leatherjacket 0.2 0.07 0.32 0.0 0.00 RPQ: Relative Quantity of heated PAV; CPQ: Cod PAV Equivalent. Based on the positive correlation cut-off points (0.5 – 1.5) for the correlation index (CI) as described in Chapter 3, fish species showing a positive correlation between the PAV content and the immunoreactivity with only the anti-cod PoAb were shaded in light grey; fish species showing a positive correlation with only the anti-carp MoAb were shaded in medium grey; fish species showing positive correlations with both antibodies were shaded in dark grey. 108

Thirty-three fish species had quantifiable responses with the anti-cod PoAb, demonstrating an increased correlation between the PAV content and the immunoreactivity (R2 = 0.60) (Figure 4.4a), compared with the correlation shown in the crude fish samples (Figure 3.5a). Fourteen species (42% of the quantifiable fish samples) had a CI between 0.5 and 1.5, showing a strong positive correlation between the PAV content and the immunoreactivity (R2 = 0.82) (Figure 4.4b). For species that had responses below the limit of detection of the ELISA (< 0.02 µg/mL), an arbitrary concentration of 0.01 µg/mL was used for the purpose of calculating the respective CI.

The heated swordfish extract, showing no PAV bands in the SDS-PAGE, was detected by the anti-cod PoAb in ELISA; while the heated yellow tuna extract, showing a suspected PAV band in the SDS-PAGE, was not detected by the anti-cod PoAb.

Australian pilchard, Atlantic salmon, rainbow trout, pink ling, crimson snapper, flame snapper, jewfish, pumpkin head trevally and red gurnard showed consistently low

ELISA responses with the anti-cod PoAb both before and after heat treatment (Tables

3.3 and 4.2). The low PAV contents of Atlantic salmon, rainbow trout, jewfish and pumpkin head trevally may attribute to their low ELISA responses. Whereas the high

PAV contents of pink ling, crimson snapper, flame snapper and grunter bream were contradicted to their low ELISA responses, suggesting that the anti-cod PoAb showed limited detectability to these fish species. The results of sand flathead, tiger flathead and mosaic leatherjacket showing low ELISA responses corresponded with the immunoblot results.

109

(a) (b)

Figure 4.4 The correlation between relative quantity of thermally stable PAV estimated by SDS-PAGE and cod-PAV equivalent concentrations by ELISA using the anti-cod PoAb for (a) quantifiable samples with the anti-cod PoAb (R2 = 0.60, y = 1.02x – 0.62) and (b) quantity-related samples with the anti-cod polyclonal antibody (R2 = 0.82, y = 1.02x – 0.22).

The ELISA results with the anti-carp MoAb agreed well with the corresponding immunoblot analysis. Sixteen fish species (49% of all tested species) had quantifiable

ELISA responses with the anti-carp MoAb. Another fifteen fish species showed immunological responses below the LOD of 0.8 μg/mL (Table 4.2); so for these species,

0.7 µg/mL was assigned in order to calculate the corresponding CI. Among the fifteen species showing immunological responses below the LOD, the heated swordfish extract and the heated tuna extract were detected by the anti-carp MoAb. The ELISA result for the low detection of the heated tuna extract correlated with the weak binding shown in the immunoblot analysis. Whereas, there was no PAV binding observed in the heated swordfish extract with the anti-carp MoAb. Pink ling, jewfish, pumpkin head trevally, sand flathead, tiger flathead, and mosaic leatherjacket were not detected by the anti-carp

MoAb. Compared with the crude extracts that showed no correlation between the PAV content and the immunoreactivity with the anti-carp MoAb (Liang et al. (2017), Figure

3.6), the heated extracts that had quantifiable ELISA responses exhibited a much stronger correlation (R2 = 0.55) (Figure 4.5). The heat treatment, by removing

110

undesirable proteins from the soluble extracts (Gajewski et al., 2009), probably helped concentrating the thermally stable PAV, thus increasing the correlation between the PAV content and the immunoreactivity.

Figure 4.5 The correlation between relative quantity of thermally stable PAV estimated by SDS-PAGE and cod-PAV equivalent concentrations by ELISA using the anti-carp MoAb for quantifiable samples with the anti-carp MoAb (R2 = 0.55, y = 1.4x + 0.17).

Cod, eastern school whiting, sweetlip emperor and sand whiting showed a positive correlation between the PAV content and the immunoreactivity with both antibodies.

Dory, barramundi, sweetlip emperor, saddletail snapper, sand whiting, Spanish mackerel and yellowfin bream showed a positive correlation consistently before and after heat treatment with the anti-cod PoAb. While bight fish, sea mullet, coral trout, red emperor and yellowtail scad, which showed positive correlations in the crude samples, demonstrated decreased ELISA responses with the anti-cod PoAb, resulting in reduced correlation.

4.3.3. Heat-induced protein structural changes of selected fish species assessed by fluorescence analysis

Fluorescence analysis was conducted to investigate the heat-induced structural changes of the fish protein extracts from twelve selected fish species (Figure 4.6). The depletion

111

of Ca2+ greatly reduces the structural stability of PAV, thus altering PAV’s fluorescence spectrometric features (Permyakov, 2006; Sudhakar et al., 1995). Treatment of crude extracts with 10 mM of EDTA as a chelating agent to capture Ca2+ ions was compared with the control group (without EDTA) during heat treatment, to investigate the effects of Ca2+ on PAV denaturation.

As shown in Figure 4.6 and Table 4.3, though it was not completely purified PAV used in the analysis, reduction of fluorescence intensity and red shift in λmax of all the fish species after heat treatment were observed, indicating protein unfolding. Furthermore, except for Atlantic salmon of which the heated extract without EDTA showed slight higher intensity than the heated extract with EDTA; all of the other species showed either similar or even slightly lower fluorescence intensity in the heated extracts without

EDTA, compared to the heated extracts with EDTA. This suggested that EDTA did not have a significant impact on destabilising the protein structure by removing Ca2+ when

2+ heat treatment was applied. The loss of Ca was confirmed by the red shift of λmax of all the heated fish samples to 350 – 355 nm in our fluorescence analysis (Table 4.3,

Permyakov (2006); Sudhakar et al. (1995)). This is probably because heat-denatured proteins cannot longer bind to Ca2+ strongly due to the protein unfolding. A similar finding was reported by Kubota et al. (2016) that the Pacific mackerel extract lost Ca2+ ions after heated at 140 °C. This finding could explain why the anti-carp MoAb, of which the antibody recognition is Ca2+ dependent (Tinner et al., 1990), lost the detectability to some fish species in our study. On the other hand, the detectability of the anti-cod PoAb was less affected by heat treatment, suggesting PAV epitope of the anti-cod PoAb may be sequential and Ca2+ independent.

112

113

Table 4.3 The summary of maximum wavelength (λmax) and maximum intensity (Imax) of the fluorescence emission spectra for 12 selected fish species

Fish samples Heated extract Heated extract Heated extract Heated extract Crude extract Crude extract with EDTA without EDTA with EDTA without EDTA Australian Pilchard Atlantic Salmon λmax (nm) 337.1 352.0 (+ 14.9) 354.0 (+ 16.9) 335.1 350.0 (+ 14.9) 345.0 (+ 9.9) Imax (A.U.) 131.2 28.3 (- 78%) 27.6 (- 79%) 130.1 39.8 (- 69%) 44.7 (- 66%) Rainbow Trout Sea Mullet λmax (nm) 339.1 350.0 (+ 10.9) 349.1 (+ 10) 338.0 349.1 (+ 11.1) 352.5 (+ 14.5) I (A.U.) 102.6 29.9 (- 71%) 26.2 (- 75%) 99.7 46.0 (- 54%) 37.3 (- 63%) max Pink Ling Barramundi

λ (nm) 342.0 348.0 (+ 6) 351.0 (+ 9) 340.0 350.0 (+ 10) 351.1 (+ 11.1) max I (A.U.) 194.3 25.7 (- 87%) 26.6 (- 86%) 160.9 44.9 (- 72%) 36.9 (- 77%) max Coral Trout Pink Snapper

λmax (nm) 340.9 351.1 (+ 10.2) 352.0 (+ 11.1) 337.1 350.0 (+ 12.9) 351.1 (+ 14)

Imax (A.U.) 96.7 24.6 (- 75%) 20.1 (- 79%) 87.6 27.9 (- 68%) 27.2 (- 69%)

Sand Whiting Spanish Mackerel

λmax (nm) 340.0 349.1 (+ 9.1) 350.0 (+ 10) 339.1 351.1 (+ 12) 355.1 (+ 16)

Imax (A.U.) 113.9 24.1 (- 79%) 24.2 (- 79%) 116.8 31.4 (- 73%) 28.2 (- 76%)

Yellowfin Tuna Tiger Flathead

λmax (nm) 339.1 352.0 (+ 12.9) 354.0 (+14.9) 340.9 352.0 (+ 11.9) 350.0 (+ 9.1)

Imax (A.U.) 115.1 43.7 (- 62%) 36.5 (- 68%) 109.3 25.1 (- 77%) 26.6 (- 76%)

The red shift of λmax of the heated extracts is stated in the bracket next to the λmax values; the %signal loss of the heated extracts is stated in the bracket next to the Imax (A.U.) values. 114

4.4. Conclusion

The detectability of the anti-cod PoAb was less affected by the thermal processing than that of the anti-carp MoAb. We speculate that heat-induced Ca2+ loss in PAV (loss of the most conservative structure) may have led to the loss of immunoreactivity of some species to the anti-carp MoAb, whereas the less conservative binding sites in the other fish species may have retained their immunoreactivity to the MoAb. The anti-cod PoAb still outperformed the anti-carp MoAb in quantitatively detecting thermally processed

PAV. Compared with the crude fish extracts, increased correlations between the PAV content and their immunoreactivity were observed for both antibodies, suggesting that thermally processed PAV is a more stable target for developing quantitative fish ELISA.

Similar trend of order specificity was observed with the heated extracts to that of the crude fish extracts. Fish species from the order of Perciformes showed strong immunoreactivity to both antibodies; whereas species from Scombriformes,

Scorpaeniformes and Tetraodontiformes showed weak or no immunoreactivity. Though increased binding intensity in immunoblot was observed for species from

Salmoniformes and Ophidiiformes which include Atlantic salmon, rainbow trout and pink ling, compared to the crude fish extracts; the corresponding quantitative ELISA still revealed limited detection to these species with the anti-cod PoAb.

These findings have guided us on species selection for immunogen preparation (i.e.,

Australian fish species from fish order that showed strong immunoreactivity to the antibodies (i.e., Perciformes) as well as fish species from fish order that showed weak or no immunoreactivity to the antibodies (e.g., Salmoniformes, Ophiidiformes,

Scombriformes, and Scorpaeniformes)) in order to develop polyclonal antibodies with broad specificity for detection of southern hemisphere fish allergen.

115

Chapter 5 Development of sandwich ELISAs with broad specificity for southern hemisphere fish allergen detection

5.1. Introduction

We have demonstrated in Chapters 3 and 4 that 1) the cod-parvalbumin specific polyclonal antibody outperformed the carp-parvalbumin specific monoclonal antibody in quantitatively detecting both native PAVs and thermally processed PAVs; 2) thermally processed PAVs are more stable proteins and better targets for developing a quantitative ELISA; 3) both native PAVs and thermally processed PAVs demonstrate similar order specificity, which has guided us in the selection of Australian fish species for antibody production.

Based on these findings, this Chapter presents the development of two sandwich ELISAs using polyclonal antibodies raised against mixed immunogens from the selected fish species, for the detection of fish allergens from southern hemisphere fish. The main tasks involved in Chapter include antisera titration, selection of the optimal capture-detection antibody combination to achieve high assay sensitivity by checkerboard ELISAs, calibration curve development, evaluation of cross-reactivity of two selected ELISAs to thirty-seven southern hemisphere fish species, and assessment of intra-assay precision of the two ELISAs.

5.2. Materials and methods

5.2.1. Materials

5.2.1.1. List of fish species

Thirty-seven fish species from twelve orders were collected from the Sydney Fish Market (Table 5.1). The fish species were selected based on the local availability, consumption and fish production (ABARES, 2014). Among the thirty-seven fish species, twenty-three species were the same as in the fish list in Chapters 3 and 4 and the other fourteen species were new species added to the fish list for the study. The thirteen fish species selected for mixed immunogen preparation were highlighted in bold in Table 5.1. 116

Table 5.1 The list of fish species analysed in this Chapter Common Name Scientific Name Family Order Distribution Australian Pilchard Sardinops neophilchardus Clupeidae Clupeiformes N & S /Australian Sardine /Sardinops sagax[2] Atlantic Salmon Salmo salar[1] Salmonidae Salmoniformes N & S Rainbow Trout Oncorhynchus mykiss[1] Salmonidae Salmoniformes N & S /Idaho Rainbow Trout John Dory Zeus faber[3] Zeidae Zeiformes N & S /Kuparu Bight Redfish Centroberyx gerrardi[3] Berycidae Beryciformes S Orange Roughy Hoplostethus atlanticus[3][5] Trachichthyidae Beryciformes N & S /Deep Sea Perch Sea Mullet Mugil cephalus[3] Mugilidae Mugiliformes N & S Hyporhamphus australis Sea Garfish /Hyporhamphus quoyi Hemiramphidae Beloniformes N & S /Hyporhamphus melanochir[4] Pink Ling Genypterus blacodes[3] Ophidiidae Ophidiiformes S Barramundi Lates calcarifer[4] Latidae Perciformes N & S Blue Cod Parapercis colias[4] Pinguipedidae Perciformes S Hyperoglyphe Antarctica Blue-eye Trevalla /Schedophilus velaini Centrolophidae Perciformes S /Schedophilus labyrinthica[3][4] Plectropomus leopardus Coral Trout /Plectropomus maculatus Serranidae Perciformes N & S /Variola louti[3] Crimson Snapper Lutjanus erythropterus[3] Lutjanidae Perciformes N & S /Crimson Sea Perch Eastern School Whiting Sillago flindersi[3] Sillaginidae Perciformes S Argyrosomus japonicus Jewfish Sciaenidae Perciformes N & S /Tandanus tandanus[4] Coryphaena hippurus Mahi Mahi Coryphaenidae Perciformes N & S /Coryphaena equiselis [1][4] Maccullochella peelii Murray Cod /Maccullochella Percichthyidae Perciformes S macquariensis[4] Pagrus auratus Pink Snapper Sparidae Perciformes N & S /Chrysophrys auratus[2] Red Emperor Lethrinus miniatus[1] Lethrinidae Perciformes N & S /Redthroat Emperor Sand Whiting Sillago ciliata[3] Sillaginidae Perciformes S

117

Table 5.1 (continued) Common Name Scientific Name Family Order Distribution Spanish Mackerel Scomberomorus commerson Scombridae Perciformes N & S /Scomberomorus semifasciatus[3] Silver Gemfish Rexea solandri[3] Gempylidae Perciformes S /Silver Kingfish Silver Perch Bidyanus bidyanus Terapontidae Perciformes N & S /Nibea soldado[3] Yellowfin Bream Acanthopagrus australis[3] Sparidae Perciformes S /Sciaenidae Yellowtail Kingfish Seriola lalandi[4] Carangidae Perciformes N & S /Yellowtail Amberjack Yellowtail Scad Trachurus novaezelandiae[3] Carangidae Perciformes S Yellowbelly Flounder Rhombosolea leporine[4] Pleuronectidae Pleuronectiformes S Australian Bonito Sarda spp.[3][4] Scombridae Scombriformes S Swordfish Xiaphias gladius[3] Xiphiidae Scombriformes N & S Yellowfin Tuna Thunnus albacares[3] Scombridae Scombriformes N & S Eastern Red Scorpaena cardinalis Scorpaenidae Scorpaeniformes S Scorpionfish /Scorpaena jacksoniensis[4] Dusky Flathead Platycephalus fuscus[3] Platycephalidae Scorpaeniformes S Ocean Perch Helicolenus spp. Sebastidae Scorpaeniformes N & S Pacific Ocean Perch /Sebastes alutus[4] Red Gurnard Chelidonichthys kumu Triglidae Scorpaeniformes N & S /Chelidonichthys cuculus[1] Tiger Flathead Neoplatycephalus richardsoni Platycephalidae Scorpaeniformes S /Neoplatycephalus aurimaculatus[3] Ocean Jacket Nelusetta ayraudi[3][4] Monacanthidae Tetraodontiformes S N = northern hemisphere; S = southern hemisphere; [1]Sharp et al. (2015); [2]DPI NSW (2014); [3]Yearsley, G. K., Last, P. R., & Ward, R. D. (1999); [4]Pauly, D., & Froese, R. (2016); [5]Saptarshi et al. (2014)

5.2.1.2. Chemicals and Reagents

Bicinchoninic acid (BCA) protein assay kit and bovine serum albumin (BSA) standard (Sigma-Aldrich, St Louis, US); acetonitrile, methanol, ethanol, acetic acid and sulfuric acid (Chemical store, UNSW, Australia); sodium dihydrogen phosphate (NaH2PO4)

(Honeywell International Inc, Marris Plains, US); sodium carbonate (Na2CO3), sodium bicarbonate (NaHCO3), di-sodium hydrogen orthophosphate (Na2HPO4), sodium chloride (NaCl), sodium acetate (Univar, Downers Grove, US); Tris

118

(Tris(hydroxymethyl)aminomethane), 40% Acrylamide/Bis solution; 2x Laemmli sample buffer, N,N,N’,N’-tetramethylethylenediamine (TEMED), dithiothreitol (DTT), iodoacetamide, sodium dodecyl sulphate (SDS), 10×Tris/glycine buffer/SDS, Precision Plus ProteinTM Dual Colour Standards (Bio-Rad, Hercules, US); β-cyclodextrin, urea hydrogen peroxide, Coomassie Brilliant Blue R-250, Tween-20, 3,3’,5,5’-tetramethylbenzidine (TMB), dimethyl sulfoxide (DMSO), Freund’s incomplete adjuvant, thiomersal (Sigma-Aldrich, St Louis, US); Trans-Blot® Turbo™ 5x Transfer Buffer (Bio-Rad, Hercules, US); anti-rabbit MoAb conjugated with horseradish peroxidase (HRP) (Abcam, Cambridge, UK); anti-sheep and anti-goat MoAbs conjugated with HRP (ThermoFisher Scientific, Waltham, US); trypsin (Promega Corporation, Madison, US); BSA powder (Bovogen Biologicals Pty Ltd, East Keilor, Australia).

5.2.1.3. Buffers

The following buffer solutions were used in the studies and their abbreviation will be used thereafter: PBS buffer 1 (0.01 M phosphate buffered saline (pH 7.4)); PBS buffer 2 (0.05 M phosphate buffered saline (pH 7.2)); blocking buffer 1 (1% BSA in 0.05 M Tris-HCl-0.9% NaCl (pH 7.5) with 0.05% Tween-20 (1% BSA-TBS-T)); coating buffer (50 mM carbonate buffer (pH 9.6); blocking buffer 2 (1% BSA in PBS buffer (1% BSA-PBS)); diluent buffer (1% BSA-PBS with 0.05% Tween-20 (1% BSA-PBS-T)); stop solution (1.25 M sulphuric acid); and washing buffer (0.05% Tween-20 in reverse osmosis water); buffer A (0.02 M sodium phosphate buffer (pH 7)); buffer B (0.1 M citric acid (pH 4)) for rabbit antibody purification; buffer B’ (0.1 M glycine-HCl (pH2.7)) for goat and sheep antibody purification; 1 M Tris-HCl buffer (pH 9.0).

5.2.1.4. Instruments

119

SpectraMax M2 plate reader (Molecular Devices, Sunnyvale, US); UV-1700 Spectrophotometer (Shimadzu, Kyoto, Japan); BioLogic LP low-pressure chromatography system (Bio-Rad, Hercules, US); Ultimate 3000 HPLC and auto-sampler system (Dionex, Amsterdam, Netherlands); LTQ-FT Ultra mass spectrometer (Thermo Electron Corporation, Bremen, Germany); ChemiDocTM MP imaging system (Bio-Rad, Hercules, US); Magic bullet food processor (Homeland Housewares, US).

5.2.2. Methods

5.2.2.1. Identification of fish species

Identification of species was done by comparing the corresponding images of whole fish, cross-section images of fish tissue, protein profiles and distribution information with the previous studies (Saptarshi et al., 2014; Sharp et al., 2015), species handbook (Yearsley et al., 1999), and online databases (Department of Agriculture and Fisheries, 2016; Pauly and Froese, 2016).

5.2.2.2. Soluble fish proteins extraction

Approximately 6g of fish tissue taken from the dorsal side of the anterior position of the body was ground with 40 mL of PBS for 30 sec using a food processor. Soluble fish proteins were extracted overnight at 4 °C with gentle rolling using a rotating mixer. Supernatants were collected as crude fish protein extracts after centrifugation at 12,000 × g for 15 min at 4 °C. The protein content of the extract was determined using the BCA protein kit. The protein extracts were stored at -20 °C until further use.

5.2.2.3. Heat treatment of crude protein extracts

The crude fish protein extracts were heat-treated based on the method described in Section 4.2.2.2 of Chapter 4.

5.2.2.4. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis (SDS-PAGE)

Heated fish proteins (5 μg per each fish sample) were separated by SDS-PAGE based on the method described in Section 3.2.2.3 of Chapter 3. 120

5.2.2.5. Preparation of crude-mixed and heat-treated mixed immunogens

Equal amounts of crude protein extracts from the thirteen selected fish species were mixed and referred to it as the crude-mixed immunogen, and equal amounts of heated protein extracts from the thirteen selected fish species were mixed and referred to it as the heat-treated mixed immunogen. The mixed immunogens were aliquoted and stored at -80 °C before use. New aliquots were used in every immunisation.

5.2.2.6. Production of polyclonal anti-PAV antibodies

Before immunisation, titration of the pre-treatment (control) blood samples from the host animals (i.e., rabbit, goat, and sheep) was performed to confirm no bindings to fish proteins.

Polyclonal antisera against the crude fish proteins were raised by injecting one pair of New Zealand white rabbits with 500 μg of crude-mixed proteins emulsified in Freund’s incomplete adjuvant (1:1 v/v) subcutaneously at multiple sites for the first injection; and the immunogen was reduced to 250 μg proteins for the subsequent injections. Polyclonal antisera against heated fish proteins were raised by injecting another pair of rabbits with 170 μg of heated-mixed proteins emulsified with Freund’s incomplete adjuvant (1:1 v/v) for every injection. There were seven injections in total over twenty-seven weeks.

Polyclonal antisera against heated proteins were also raised in one goat and two sheep by injecting the animals with 330 μg of heated-mixed proteins treated the same way as the rabbit immunisation. There were six injections in total over twenty-three weeks.

The first three injections were repeated in two-week intervals. From the fourth injection onwards, injection was repeated in four-weeks intervals. Blood samples were collected 7 – 9 days after each immunisation from the third injection. The blood samples were centrifuged at 1,509 × g for 15 min at 4 °C to separate the blood cells, and the antisera (supernatant) was collected. The antisera mixed with 0.005% (w/v) thiomersal (preservative) were stored at – 20 °C until further use.

5.2.2.7. Antisera titration

Titres of fish protein specific IgG were monitored by titrating the antisera against their 121

respective immunogens in an indirect ELISA. Briefly, 1 μg per well of the crude- or the heat-treated immunogen in coating buffer was immobilized on the microtiter plates (Thermo Fisher Scientific, Waltham, US) for overnight at room temperature. Thereafter, all incubation steps, except for the blocking and incubation with the substrate solution, were performed for 30 min at room temperature. After the immobilisation step, the plate was washed 3 times with washing buffer and blocked for 1 h with 200 μL per well of blocking buffer 2. The antisera starting with 1:100 dilution in diluent buffer were further serially diluted in 1/3, and 100 μL per well of each dilution was added in the respective wells for incubation. After another washing, bound fish protein specific IgG was detected by the HRP conjugated anti-species antibodies appropriately diluted in the diluent buffer. 100 μL per well of TMB substrate solution was incubated in all the wells for 20 min to develop colour. Absorbance was measured at 450 nm using the SpectraMax M2 plate reader. The assays were performed in duplicate wells. The titre value was defined as the dilution factor that gave the corresponding absorbance value at around 0.1 AU above the background (baseline).

5.2.2.8. Purification of polyclonal antibodies

The rabbit polyclonal antibodies were purified from the rabbit antisera using a HiTrap Protein A (protein derived from Staphylococcus aureus and immobilised on highly cross-linked agarose beads) HP column (5 ml; GE Healthcare, Little Chalfont, UK) and the goat and sheep polyclonal antibodies were purified using a HiTrap Protein G (protein derived from Streptococcus sp. and immobilised on highly cross-linked agarose beads) HP column (5 ml; GE Healthcare, Little Chalfont, UK). The columns were equilibrated with buffer A. The antisera were mixed with buffer A in 1:1 (v/v) and loaded onto the column. Then buffer A was loaded to wash the unbound proteins. After the UV conductivity returned to zero and remained a flat line for more than five minutes, buffer B/B’ was loaded to elute IgG antibodies. The collected IgG fragment was neutralised with 1 M Tris-HCl buffer (pH 9.0) and dialysed against PBS buffer 2 in 2 kDa cut-off dialysis membrane (Spectrum® Laboratory) for three days with several buffer changes. The purified antibody solutions mixed with 0.005% (w/v) thiomersal were stored at 4 °C.

The concentration of the purified antibodies was determined by the UV-1700 122

Spectrophotometer at 280 nm and calculated as follows using Beer-Lambert’s Law.

Abs (AU) C (mg/mL) = × DF Equation 5.1 ε

Where C is the concentration of the purified antibody, Abs is the UV/Vis absorbance at 280 nm, ε is the molecular extinction coefficient (ε = 1.35 for IgG), DF is the dilution factor of the purified antibodies.

5.2.2.9. Immunoblotting analysis

Heated fish proteins (5 μg per each fish sample) were separated by SDS-PAGE using the conditions in section 3.2.2.3 of Chapter 3. The separated proteins were transferred onto a 0.2 µm nitrocellulose membrane using the Bio-Rad Trans-blot system, and immunoblot analysis was performed as presented in Section 3.2.2.4 of Chapter 3.

5.2.2.10. ELISA calibration curve – indirect sandwich ELISA

For the indirect sandwich ELISA, microtiter plates were immobilised with 1 μg per well of a purified polyclonal antibody as the capture antibody (e.g., (HM)RB#2, refer the abbreviation to Section 5.3.2) in the coating buffer and incubated overnight at room temperature. Thereafter, all incubation steps, except for the blocking and incubation with the substrate solution, were performed for 30 min at room temperature. After the antibody immobilisation, the plates were washed 4 times with washing buffer, and incubated with 200 µL per well of blocking buffer 2 for 1 h. Calibration standard solutions of fish protein were prepared by titrating in 1/3 dilution with an initial concentration of 10,000 μg/L. After blocking and washing the plate for 5 times, 100 μL per well of the standard solutions was added to their respective wells. Next, the plates were washed 6 times and incubated with 100 μL per well of another purified polyclonal antibody as the detection antibody (e.g., (HM)S627#3) in the diluent buffer. The plates were washed 6 times again and incubated with the HRP conjugated anti-species antibodies diluted in the diluent buffer. Finally, 100 μL per well of TMB substrate solution was incubated for 20 min to develop colour. The stop solution (50 μL per well) was added to stop the reaction, and the absorbance was measured at 450 nm using the SpectraMax M2 plate reader. All the assays were performed in duplicate wells unless stated otherwise. 123

5.2.2.11. Checkerboard ELISA

A checkerboard ELISA in the indirect sandwich format was performed to select the best capture and detection antibody combination. in the indirect sandwich ELISA. The checkerboard ELISA workflow followed the ELISA procedure described in Section 5.2.2.11. Briefly, the microtitre plates were coated with 1μg per well of the purified capture antibody. After blocking and washing, the crude or the heat-treated immunogen was added to incubate with the (immobilised) capture antibody. After washing, the purified detection antibody diluted 1:200 or 1:500 or 1:10,000 in diluent buffer was added to detect the antibody-antigen complex and the unbound antibodies were removed by washing. The dilution factors of the detection antibody were chosen based on our previous experience. The anti-species antibody diluted in the diluent buffer was incubated in the respective well and unbound was removed by washing. Lastly, the colour was developed by adding the TMS solution.

The crude immunogen was used for the checkerboard ELISAs with the capture antibodies (CM)RC#3 and (CM)RD#3, which were raised against the crude immunogen. The heat-treated immunogen was used for the checkerboard ELISAs with the following capture antibodies: (HM)RA#2, (HM)RB#2, (HM)G#1, (HM)G#2, (HM)G#3, (HM)G#4, (HM)S606#2, (HM)S606#3, (HM)S606#4, (HM)S627#2, (HM)S627#3 and (HM)S627#4, which were raised against the heat-treated immunogen. Three concentrations of the crude / the heat-treated immunogens were selected to conduct the checkerboard ELISA: zero concentration as a negative control, an intermediate concentration, and a high concentration for maximum colour development. The intermediate concentration chosen were 500 μg/L or 100 μg/L. If 500 μg/L gave a %Abs (please see below for the description) higher than 50%, then 100 μg/L was used. The high concentration representing the maximum antibody binding were 10,000 μg/L or 100,000 μg/L. If 10,000 μg/L gave an absorbance less than 1.5 AU, then 100,000 μg/L was used.

Four analytical parameters were used to evaluate the performance of antibody pairs: %Abs(500/100), Blank Abs, max – blk and noise-to-signal ratio (N/S).

%Abs(500/100) is the percent absorbance either at 500 μg/L or 100 μg/L relative to the

124

maximum absorbance either at 10,000 μg/L or 100,000 μg/L. The calculation equation is shown below:

Abs (500/100) (AU) % Abs(500/100) = × 100 Equation 5.2 Absmax (AU)

Where Abs(500/100) is the absorbance at 500 μg/L or 100 μg/L, Absmax is the maximum absorbance. %Abs(500/100) is a key estimator for the assay sensitivity, and is rated as follows:

#### %Abs(500/100) > 75

### 50 < %Abs(500/100) < 75

## 25 < %Abs(500/100) < 50

# %Abs(500/100) < 25

Blank indicates the background colour which reflects the degree of non-specific bindings. To achieve high sensitivity, the background colour needs to be below a set level of 0.2 AU. This parameter is rated as follows: **** Blank < 0.2 AU *** 0.2 AU < Blank < 0.5 AU ** 0.5 AU < Blank < 1 AU * Blank > 1 AU

(Max – Blk) is the maximum absorbance minus the blank absorbance. This parameter indicates the absorbance unit of an ELISA. It is rated as follows: ¶¶¶¶ Max – Blk > 1.5 AU ¶¶¶ 1 AU < Max – Blk < 1.5 AU ¶¶ 0.5 AU < Max – Blk < 1 AU ¶ Max – Blk < 0.5 AU

Last parameter is the noise-to-signal ratio (N/S). This is an integrated parameter showing the percent non-specific binding relative to the maximum binding capacity. In practice, the lowest N/S is desirable to achieve the optimum sensitivity in an assay.

Blank Abs (AU) N/S = Equation 5.3 Maximum Abs (AU)

125

This parameter is rated as follows: ●●●● N/S < 0.25 ●●● 0.25 < N/S < 0.5 ●● 0.5 < N/S < 0.75 ● N/S > 0.75

5.2.2.12. Cross-reactivity by a checkerboard ELISA

Based on the results of the calibration curve development, heated fish protein extracts were used for the cross-reactivity study. Following the same approach of the previous approach described above, three points (negative control, an intermediate concentration, and a high concentration) were chosen to conduct initial screening of the cross-reactivity of four selected capture-detection antibody combinations to thirty-seven fish species listed in Table 5.1.

%Abs(100/50) was calculated according to Equation 5.2. A percent reactivity (%R) of selected fish species relative to the reactivity of the heat-treated mixed immunogen with each capture-detection antibody combination was calculated as follows:

%Abs s %R = (100/50) Equation 5.4 %Abs(100/50)std

%Abs(100/50)s is the percent absorbance at 100 μg/L or 50 μg/L of fish samples relative to the maximum absorbance determined at 10,000 μg/L of fish samples. %Abs(100/50)std is the percent absorbance at 100 μg/L or 50 μg/L of the heat-treated immunogen relative to the maximum absorbance at 10,000 μg/L of the immunogen. ± 20% (i.e., 80 – 120 %R) and ± 50% (i.e., 50 – 150 %R) were chosen as the cut-off points to assist us with the selection of the best capture-detection antibody combination, based on the recovery rates recommended by the ELISA validation guidance (Abbott et al., 2010). The desired antibody combinations are the ones exhibiting the least variation of the immunoreactivity with different fish species.

5.2.2.13. Cross-reactivity study by calibration curves

The cross-reactivity study based on the full calibration curves was conducted following the indirect sandwich ELISA protocol in Section 5.2.2.10. Heat-treated fish protein

126

extracts at 10,000, 2,000, 400, 80, 16, 3.2, and 0.64 μg/L and heat-treated immunogen in 1000, 200, 40, 8, 1.6, 0.32, 0.064 μg/L were used. The reactivity of a certain fish species relative to the heat-treated immunogen was calculated as follows:

C of heat-treated mixed immunogen (μg/L) % Cross-reactivity = 50 × 100 Equation 5.5 C50 of heated fish protein extract (μg/L)

C50 is the concentration of fish protein corresponding to the 50% of the maximum colour development. C50 values was determined from the calibration curves generated using GraphPad Prism® v.7.0, as described in Section 3.2.2.7 of Chapter 3.

5.2.2.14. Data analysis

The ELISA data analysis was the same as described in Section 3.2.2.7 of Chapter 3.

The analytical parameters for comparing the performances of fish ELISAs in our study include the C15 values (defined as the lower limit of detection, LLOD), the C50 values

(the indicator for assay sensitivity), and the C80 values (the upper limit of detection,

ULOD). The C15 is a concentration of fish protein produce 15% of colour relative to the maximum colour development and is the lower limit of the linear range of the calibration curves in our study. C80 is a concentration of fish protein produce 80% of colour relative to the maximum colour development and is and upper limit of the linear range of the calibration curves in our study.

Our definition of LLOD is different from the definition of LLOD in other studies (Cai et al., 2013; Fæste and Plassen, 2008; Shibahara et al., 2013; Weber et al., 2009), which is commonly calculated as the mean measured value of a set of blank sample replicates (n = 8 – 34) plus 3 times of the standard deviation (SD) of that mean value. This calculation results in lower LLOD than the LLOD in our study. The LLOD defined in our study is sometimes referred as the LOQ in some of the commercial ELISA test kits.

5.2.2.15. LC-MS-MS analysis

In-gel digestion

Targeted bands were cut into pieces from the SDS-PAGE gels, and each band was processed for in-gel digestion based on the protocol described as follows. After adjusting 127

the pH of the band pieces with 100 mM ammonium bicarbonate, the band pieces were de-stained with acetonitrile/50 mM ammonium bicarbonate (1:1 v/v) with vortex repeatedly until they turned transparent. Then the transparent band pieces were de-hydrated with acetonitrile. After removing the acetonitrile residues and letting the band pieces dry until the colour of the band pieces turned white, the proteins in the band pieces were reduced with 10 mM DTT in 100 mM ammonium bicarbonate for 30 min at 37 °C. Sequentially, the reduced proteins were alkylated with 55 mM iodoacetamide in 100 mM ammonium bicarbonate for 30 min at room temperature in the dark. The washing/dehydration circle was repeated twice, and the dried bands were digested overnight with 6.25 ng/mL trypsin at 37 °C. After digestion, the digesta were collected, and peptides were extracted several times with 50% acetonitrile containing 2% formic acid. The supernatants were combined, vacuumed dried, and reconstituted with 0.1% formic acid for storage at -20 °C until further use.

Liquid chromatography (LC)

Nano-Liquid chromatography (nano-LC) was performed using the Ultimate 3000 HPLC and autosampler system Digested peptides were injected into a fritless nanoLC column (75µm x ~12cm) containing C18 media (1.9µm, 120 Å ReproSil-Pur 120 C18-AQ, Dr Maisch GmbH) manufactured according to Gatlin et al. (1998). Peptides were eluted using a linear gradient according to the conditions on Table 5.2 below, over 52 min, at a flow rate of 0.200 µL/min. Mobile phase A consisted of 0.1% formic acid in H2O, while mobile phase B consisted of acetonitrile: H2O (8:2) with 0.1% formic acid.

Table 5.2 Gradient of mobile phase B for liquid chromatography Time (min) Percentage of mobile phase B (%) 0.0 2.0 4.0 2.0 36.0 45.0 37.0 80.0 37.5 80.0 39.0 2.0 52.0 2.0

128

Tandem mass spectrometry (MS/MS)

High voltage (2000 V) was applied to a low volume tee (Upchurch Scientific) and the column tip positioned ~ 0.5 cm from the heated capillary (T=275°C) of the LTQ Orbitrap Velos mass spectrometer. Positive ions were generated by electrospray and the mass spectrometer operated in a data-dependent acquisition mode (DDA). Full scan MS spectra were acquired (m/z 350-1750) by the Orbitrap at a resolution of 30,000. The 15 most abundant ions (>5,000 counts) with charge states ≥ +2 were sequentially isolated and fragmented within the linear ion trap using collisionally induced dissociation with an activation q = 0.25 and activation time of 10 ms at a target value of 30,000 ions. M/z ratios selected for MS/MS were dynamically excluded for 35 seconds.

Data analysis

The MS/MS spectra were searched by Mascot (Matrix Science, London, UK; version 2.3.02) against the NCBInr_24_10_15 databased with trypsin as the proteolytic enzyme. The search was restricted to the fish taxon Actinopterygii (ray-finned fishes) as only fish protein extracts were analysed in our study. The search engine was set to have a fragment ion mass tolerance of 0.4 Da and a peptide ion mass tolerance of 4.0 Da. The number of maximum missed cleavages was set to be 1 and the ion score cut-off point was 20. Carbamidomethylation and propionamide modification of cysteine and oxidation of methionine were set to be the variable modifications. The significant sequences threshold was set to be 0.05.

5.3. Results and discussion

5.3.1. The selection of fish species for mixed immunogen preparation

Based on our findings in Chapters 3 and 4, we selected fish species from the order that showed strong immunoreactivity (i.e., Perciformes) such as barramundi and Spanish mackerel as well as fish species from fish order that showed weak or no immunoreactivity (e.g., Salmoniformes, Ophiidiformes, Scombriformes, and Scorpaeniformes) such as Atlantic salmon, rainbow trout, pink ling and tiger flathead to prepare our mixed immunogens. The selected fish are also commercially important in Australia. In total, thirteen fish species across seven orders were selected (Figure 5.1).

129

The protein profiles of the crude and the heated fish extracts used for the mixed immunogens were consistent our previous results (Figures 3.1 and 4.1), which ensures that the right protein compositions have been immunised for in vivo antibody production.

(a)

(b)

Figure 5.1 Protein profiles of (a) crude fish protein extracts and (b) heated protein extracts for the immunogen preparation. The 36 kDa bands in red rectangles were chosen for LC-MS-MS analysis.

5.3.2. Antisera titration

The rationale for choosing polyclonal antibodies was based on the studies discussed in Chapter 3 and 4, and because of the expected molecular diversity of parvalbumins in our immunogen were prepared from multiple fish species.

The antisera against the heat-treated immunogen were raised from a pair of rabbits (A

130

and B), which are abbreviated to (HM)RA and (HM)RB. The antisera against the crude-mixed immunogen were raised from another pair of rabbits (C and D), which are abbreviated to (CM)RC and (CM)RD. The number of bleeds (i.e., 1, 2, 3, 4 and 5) is coded as #1, #2, #3, #4 and #5. For instance, the antibody purified from the first bleed of rabbit A is named as (HM)RA#1. One goat and two sheep were immunised with the heat-treated immunogen, so the antisera raised from goat are named as (HM)G, and the antisera raised from sheep are named as (HM)S606 and (HM)S627. 606 and 627 were the codes given to the immunised animals.

The titre values of the first bleeds of rabbit antisera were greater than 10,000, indicating that the immunisation was effective and specific IgG antibodies were formed (Figure 5.2 and Table 5.3). The production of the specific IgG antibodies reached maximum titre after four injections (the second bleed); thereafter it became stable but decreased somewhat at the final injection. Notably, the antisera from rabbits A and B exhibited higher titres than rabbits C and D, indicating the heated proteins may be more immunogenic than the crude proteins. This may be that thermal treatment helped to concentrate the desirable heat stable proteins including parvalbumins by removing undesirable insoluble proteins (Gajewski et al., 2009).

To minimise non-specific bindings, the bleed with a high titre value was selected for the further characterisation and antibody purification. In this case, (HM)RA#2, (HM)RB#2, (CM)RC#3 and (CM)RD#3 were chosen.

131

(a) (b)

Figure 5.2 Titration curves of (a) (HM)RA & (b) (HM)RB antisera to heated-mixed fish protein, and (c) (CM)RC & (d) (CM)RD to crude-mixed fish protein for (●) first bleed, (■) second bleed, (▲) third bleed, (▼) fourth bleed, (♦) fifth bleed (terminal bleed).

Table 5.3 The titre values of rabbit antisera against crude-mixed/heated-mixed fish protein

Number of bleeds (HM)RA (HM)RB (CM)RC (CM)RD #1 656,100 656,100 72,900 218,700 #2 2187,000 2187,000 243,000 729,000 #3 729,000 729,000 243,000 729,000 #4 729,000 729,000 243,000 729,000 #5 243,000 243,000 81,000 243,000

The production of specific IgG antibodies in goat and sheep showed a steady increase during the immunisation (Figure 5.3 and Table 5.4). The titre values of sheep antisera increased dramatically in the terminal bleed. All the goat antisera were chosen for the antibody purification. The sheep antisera chosen for antibody purification were bleeds #2, #3 and #4. 132

(a)

(b) (c)

Figure 5.3 Titration curves of (a) (HM)G (b) (HM)S606 and (c) (HM)S627 antisera to heated-mixed fish protein for (●) first bleed, (■) second bleed, (▲) third bleed, (▼) fourth bleed (terminal bleed).

Table 5.4 The titre values of antisera heated-mixed fish proteins, raised in goat and sheep

Number of Bleeds (HM)G (HM)S606 (HM)S627 #1 218,700 2,700 2,700 #2 72,900 2,700 2,700 #3 243,000 9,000 81,000 #4 656,100 218,700 656,100

5.3.3. Checkerboard ELISA

Selection of the best performing capture/detection antibody pair largely predetermines the performance of a sandwich ELISA. For this purpose, checkerboard ELISAs, examining the capture antibody/detection antibody combination and the working

133

concentrations of the antibodies, were conducted. For the best sensitivity, the capture antibody (primary antibody) is required to have high affinity and specificity to the target allergen, while the detection antibody (secondary antibody) can be of lower affinity and less specific (He, 2013; Lee and Sun, 2016). The two antibodies must be from unrelated species to avoid self-cross-reactivity in a sandwich ELISA. If two antibodies from the same species must be used, then the second (detection) antibody must be conjugated with an enzyme. Therefore, in our indirect sandwich ELISA, if rabbit antibodies are used as capture antibodies, goat or sheep antibodies should be used as detection antibodies.

The selection of antibody combinations is based on four analytical parameters which have been explained in the method Section 5.2.2.11. The four parameters as the assay performance indicators provide estimates of relative sensitivity, possible non-specific bindings, and detection range. In total, one hundred and fifty antibody pairs were evaluated, and the screening results are summarised in Tables 5.5 to 5.10.

As shown in Tables 5.5 and 5.6, several combinations with the rabbit antibodies as the capture antibodies and the goat antibodies as the detection antibodies offered good sensitivity (%Abs(500)) with acceptable background (blank Abs) and good signal range (max – blk). These include (HM)RB#2 – (HM)G#1 1:200, (HM)RB#2 – (HM)G#2 1:200, (HM)RB#2 – (HM)G#3 1:200, (CM)RD#3 – (HM)G#1 1:200 and (CM)RD#3 – (HM)G#2 1:200. So these five pairs were chosen for further development. The combinations with the rabbit antibodies (i.e., (HM)RB#2, (CM)RD#3) as the capture antibodies and the sheep antibodies (i.e., (HM)S606#2, (HM)S627#2) as the detection antibodies also produced good sensitivity, but the generated absorbance was relatively low, and they were not considered for further development.

Tables 5.7 to 5.9 show the results with goat and sheep antibodies as the capture antibodies. Though the sheep antibodies exhibited lower titre values than the goat antibodies (Figure 5.3 and Table 5.4); the sheep antibodies as the capture antibodies offered better assay sensitivity (higher %Abs(500) values ) than the goat antibodies, suggesting the sheep antibodies had higher affinity to the immunogen than the goat antibodies. Between the two sheep antibodies (Tables 5.8 and 5.9), (HM)S627 outperformed (HM)S606 in terms of assay sensitivity; in particular, (HM)S627#2 – 134

(HM)RB#2 1:500 and (HM)S627#3 – (HM)RB#2 1:500 offered good sensitivity, low background colour and satisfactory signal range, and they were chosen for further development.

Table 5.10 presents the combination of (HM)RB#2 with goat and sheep antibodies from the final bleed (i.e., (HM)G#4 and (HM)S627#4). The sheep antibody (HM)S627#4 showed higher affinity to the immunogen than the goat antibody (HM)G#4 (Table 5.10). The combinations of (HM)S627#4 – (HM)RB#2 1:500 and (HM)RB#2 – (HM)S627#4 1:500, showed high sensitivity, low background colour, and satisfied signal range and they were chosen for further development.

135

Table 5.5 Performances of the capture-antibody antibody combinations with (HM)RA#2 and (HM)RB#2 as the capture antibodies

Detection Antibodies Capture Antibodies %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S (HM)G#1 1:200 + Anti-G (1:10,000) (HM)G#1 1:500 + Anti-G (1:10,000) (HM)G#1 1:1,000 + Anti-G (1:10,000) (HM)RA#2 #### * ¶¶ ● #### * ¶¶ ●● ### * ¶ ● (HM)G#2 1:200 + Anti-G (1:10,000) (HM)G#2 1:500 + Anti-G (1:10,000) (HM)G#2 1:1,000 + Anti-G (1:10,000) (HM)RA#2 N/A * N/A N/A # * ¶ ● ## * ¶ ● (HM)S606#2 1:200 + Anti-S (1:10,000) (HM)S606#2 1:500 + Anti-S (1:10,000) (HM)S606#2 1:1,000 + Anti-S (1:10,000) (HM)RA#2 ### **** ¶ ●● ### **** ¶ ●● ## **** ¶ ●● (HM)S627#2 1:200 + Anti-S (1:10,000) (HM)S627#2 1:500 + Anti-S (1:10,000) (HM)S627#2 1:1,000 + Anti-S (1:10,000) (HM)RA#2 ### **** ¶ ●●● ### **** ¶ ●●● ## **** ¶ ●●● (HM)G#3 1:200 + Anti-G (1:10,000) (HM)G#3 1:500 + Anti-G (1:10,000) (HM)G#3 1:1,000 + Anti-G (1:10,000) (HM)RA#2† ### *** ¶¶ ●●● ## **** ¶ ●●● ### **** ¶ ●● (HM)G#1 1:200 + Anti-G (1:10,000) (HM)G#1 1:500 + Anti-G (1:10,000) (HM)G#1 1:1,000 + Anti-G (1:10,000) (HM)RB#2 #### *** ¶¶¶¶ ●●● #### **** ¶¶ ●●●● #### *** ¶¶¶ ●●●● (HM)G#2 1:200 + Anti-G (1:10,000) (HM)G#2 1:500 + Anti-G (1:10,000) (HM)G#2 1:1,000 + Anti-G (1:10,000) (HM)RB#2 #### *** ¶¶¶ ●●● #### **** ¶¶ ●●●● #### *** ¶¶¶ ●●● (HM)S606#2 1:200 + Anti-S (1:10,000) (HM)S606#2 1:500 + Anti-S (1:10,000) (HM)S606#2 1:1,000 + Anti-S (1:10,000) (HM)RB#2 #### **** ¶ ●●● #### **** ¶ ●●● #### **** ¶ ●● (HM)S627#2 1:200 + Anti-S (1:10,000) (HM)S627#2 1:500 + Anti-S (1:10,000) (HM)S627#2 1:1,000 + Anti-S (1:10,000) (HM)RB#2 #### **** ¶ ●●●● ### **** ¶ ●●●● #### **** ¶ ●●● (HM)G#3 1:200 + Anti-G (1:10,000) (HM)G#3 1:500 + Anti-G (1:10,000) (HM)G#3 1:1,000 + Anti-G (1:10,000) (HM)RB#2† #### *** ¶¶ ●●●● #### **** ¶ ●●●● #### **** ¶ ●●●

%Abs(500) = Percent Absorbance at 500 µg/L relative to maximum Absorbance at 10,000 µg/L † %Abs(500)’= Percent Absorbance at 500 µg/L relative to maximum Absorbance at 100,000 µg/L N/S = noise-to-signal ratio; Anti-G represents anti-goat monoclonal antibody; Anti-S represents anti-sheep monoclonal antibody 1:200, 1:500, 1:1,000 and 1:10,000 are the dilution factors; the rating criteria are referred to method section 5.2.2.11.

136

Table 5.6 Performances of the capture-detection antibody combinations with (CM)RC#3 and (CM)RD#3 as the capture antibodies

Detection Antibodies Capture Antibodies %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S (HM)G#1 1:200 + Anti-G (1:10,000) (HM)G#1 1:500 + Anti-G (1:10,000) (HM)G#1 1:1,000 + Anti-G (1:10,000) (CM)RC#3 # ** ¶¶¶ ●●● ## ** ¶¶¶ ●●● ## ** ¶ ●● (HM)G#2 1:200 + Anti-G (1:10,000) (HM)G#2 1:500 + Anti-G (1:10,000) (HM)G#2 1:1,000 + Anti-G (1:10,000) (CM)RC#3 ## ** ¶¶ ●●● ### *** ¶¶ ●●● ## ** ¶ ●● (HM)S606#2 1:200 + Anti-S (1:10,000) (HM)S606#2 1:500 + Anti-S (1:10,000) (HM)S606#2 1:1,000 + Anti-S (1:10,000) (CM)RC#3 ## **** ¶ ●●● N/A **** ¶ ●● N/A **** ¶ ● (HM)S627#2 1:200 + Anti-S (1:10,000) (HM)S627#2 1:500 + Anti-S (1:10,000) (HM)S627#2 1:1,000 + Anti-S (1:10,000) (CM)RC#3 ## **** ¶ ●●● ## **** ¶ ●●● ## **** ¶ ●●● (HM)G#1 1:200 + Anti-G (1:10,000) (HM)G#1 1:500 + Anti-G (1:10,000) (HM)G#1 1:1,000 + Anti-G (1:10,000) (CM)RD#3 ### *** ¶¶¶¶ ●●● ### *** ¶¶¶ ●●● ### *** ¶¶ ●●● (HM)G#2 1:200 + Anti-G (1:10,000) (HM)G#2 1:500 + Anti-G (1:10,000) (HM)G#2 1:1,000 + Anti-G (1:10,000) (CM)RD#3 ### *** ¶¶¶ ●●● #### *** ¶¶ ●●● ### *** ¶ ●●● (HM)S606#2 1:200 + Anti-S (1:10,000) (HM)S606#2 1:500 + Anti-S (1:10,000) (HM)S606#2 1:1,000 + Anti-S (1:10,000) (CM)RD#3 #### **** ¶ ●●● ### **** ¶ ●● ### **** ¶ ●● (HM)S627#2 1:200 + Anti-S (1:10,000) (HM)S627#2 1:500 + Anti-S (1:10,000) (HM)S627#2 1:1,000 + Anti-S (1:10,000) (CM)RD#3 ### **** ¶ ●●●● ### **** ¶ ●●● ### **** ¶ ●●● (HM)G#3 1:200 + Anti-G (1:10,000) (HM)G#3 1:500 + Anti-G (1:10,000) (HM)G#3 1:1,000 + Anti-G (1:10,000) (CM)RD#3† ### **** ¶ ●●● ### **** ¶ ●●● ### **** ¶ ●●●

137

Table 5.7 Performances of the capture-antibody antibody combinations with (HM)G#1 and (HM)G#2 as the capture antibodies

Detection Antibodies Capture Antibodies %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S (HM)RA#2 1:200 + Anti-R (1:5,000) (HM)RA#2 1:500 + Anti-R (1:5,000) (HM)RA#2 1:1,000+ Anti-R (1:5,000) (HM)G#1 # *** ¶¶¶¶ ●●●● # *** ¶¶ ●●● # **** ¶¶ ●●● (HM)RB#2 1:200+ Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:1,000 + Anti-R (1:5,000) (HM)G#1 ## *** ¶¶¶¶ ●●●● # *** ¶¶¶¶ ●●●● # *** ¶¶¶ ●●●● (CM)RC#3 1:200 + Anti-R (1:5,000) (CM)RC#3 1:500 + Anti-R (1:5,000) (CM)RC#3 1:1,000 + Anti-R (1:5,000) (HM)G#1 # **** ¶¶ ●● N/A **** ¶ ● # **** ¶ ● (CM)RD#3 1:200 + Anti-R (1:5,000) (CM)RD#3 1:500 + Anti-R (1:5,000) (CM)RD#3 1:1,000 + Anti-R (1:5,000) (HM)G#1 # *** ¶¶¶ ●●●● # *** ¶¶ ●●● # **** ¶ ●●● (HM)RA#2 1:200 + Anti-R (1:5,000) (HM)RA#2 1:500 + Anti-R (1:5,000) (HM)RA#2 1:1,000 + Anti-R (1:5,000) (HM)G#2 # **** ¶¶¶¶ ●●●● # **** ¶¶¶ ●●●● # **** ¶¶ ●●●● (HM)RB#2 1:200 + Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:1,000 + Anti-R (1:5,000) (HM)G#2 ## **** ¶¶¶¶ ●●●● ## **** ¶¶¶¶ ●●●● ## **** ¶¶¶ ●●●● (CM)RC#3 1:200 + Anti-R (1:5,000) (CM)RC#3 1:500 + Anti-R (1:5,000) (CM)RC#3 1:1,000 + Anti-R (1:5,000) (HM)G#2 # **** ¶¶ ●●●● # **** ¶ ●●● # **** ¶ ●●● (CM)RD#3 1:200 + Anti-R (1:5,000) (CM)RD#3 1:500 + Anti-R (1:5,000) (CM)RD#3 1:1,000 + Anti-R (1:5,000) (HM)G#2 # **** ¶¶¶ ●●●● # **** ¶¶ ●●●● # **** ¶ ●●●●

%Abs(500) = Percent Absorbance at 500 µg/L relative to maximum Absorbance at 10,000 µg/L N/S = noise-to-signal ratio; Anti-R represents anti-rabbit monoclonal antibody 1:200, 1:500, 1:1,000 and 1:5,000 are the dilution factors; the rating criteria are referred to method section 5.2.2.11. N/A: the ratings were not available due to the results of the parameter were zero.

138

Table 5.8 Performances of the capture-antibody antibody combinations with (HM)S606#2 and (HM)S606#3 as the capture antibodies

Detection Antibodies Capture Antibodies %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S (HM)RA#2 1:200 + Anti-R (1:5,000) (HM)RA#2 1:500 + Anti-R (1:5,000) (HM)RA#2 1:1,000 + Anti-R (1:5,000) (HM)S606#2 ### *** ¶¶¶ ●●●● ### *** ¶¶ ●●● ### *** ¶ ●●● (HM)RB#2 1:200 + Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:1,000 + Anti-R (1:5,000) (HM)S606#2 ### *** ¶¶¶ ●●●● ## *** ¶¶ ●●●● ## *** ¶¶ ●●●● (CM)RC#3 1:200 + Anti-R (1:5,000) (CM)RC#3 1:500 + Anti-R (1:5,000) (CM)RC#3 1:1,000 + Anti-R (1:5,000) (HM)S606#2 ## *** ¶ ●●● ### *** ¶ ●● ### *** ¶ ● (CM)RD#3 1:200 + Anti-R (1:5,000) (CM)RD#3 1:500 + Anti-R (1:5,000) (CM)RD#3 1:1,000 + Anti-R (1:5,000) (HM)S606#2 ## *** ¶¶ ●●●● ## *** ¶ ●●● ## **** ¶ ●●● (HM)RA#2 1:200 + Anti-R (1:5,000) (HM)RA#2 1:500 + Anti-R (1:5,000) (HM)RA#2 1:1,000 + Anti-R (1:5,000) (HM)S606#3† ## **** ¶¶¶¶ ●●●● # **** ¶¶¶¶ ●●●● # **** ¶¶¶ ●●●● (HM)RB#2 1:200 + Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:1,000 + Anti-R (1:5,000) (HM)S606#3† ## **** ¶¶¶¶ ●●●● ## **** ¶¶¶¶ ●●●● # **** ¶¶¶ ●●●●

%Abs(500) = Percent Absorbance at 500 µg/L relative to maximum Absorbance at 10,000 µg/L † %Abs(500)’= Percent Absorbance at 500 µg/L relative to maximum Absorbance at 100,000 µg/L N/S = noise-to-signal ratio; Anti-R represents anti-rabbit monoclonal antibody 1:200, 1:500, 1:1,000 and 1:5,000 are the dilution factors; the rating criteria are referred to method section 5.2.2.11.

139

Table 5.9 Performances of the capture-detection antibody combinations with (HM)S627#2 and (HM)S627#3 as the capture antibodies

Detection Antibodies Capture Antibodies %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S %Abs(500) Blank Abs Max-Blk N/S (HM)RA#2 1:200 + Anti-R (1:5,000) (HM)RA#2 1:500 + Anti-R (1:5,000) (HM)RA#2 1:1,000 + Anti-R (1:5,000) (HM)S627#2 ### **** ¶¶¶¶ ●●●● ### **** ¶¶¶ ●●●● ### **** ¶¶¶ ●●●● (HM)RB#2 1:200 + Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:1,000 + Anti-R (1:5,000) (HM)S627#2 #### **** ¶¶¶¶ ●●●● ### **** ¶¶¶¶ ●●●● ### **** ¶¶¶ ●●●● (CM)RC#3 1:200 + Anti-R (1:5,000) (CM)RC#3 1:500 + Anti-R (1:5,000) (CM)RC#3 1:1,000 + Anti-R (1:5,000) (HM)S627#2 ### **** ¶¶ ●●●● ### **** ¶ ●●● ### **** ¶ ●●● (CM)RD#3 1:200 + Anti-R (1:5,000) (CM)RD#3 1:500 + Anti-R (1:5,000) (CM)RD#3 1:1,000 + Anti-R (1:5,000) (HM)S627#2 ### **** ¶¶¶ ●●●● ### **** ¶¶ ●●●● ### **** ¶¶ ●●●● (HM)RA#2 1:200 + Anti-R (1:5,000) (HM)RA#2 1:500 + Anti-R (1:5,000) (HM)RA#2 1:1,000 + Anti-R (1:5,000) (HM)S627#3† ### *** ¶¶¶¶ ●●●● ### *** ¶¶¶¶ ●●●● ### *** ¶¶¶ ●●●● (HM)RB#2 1:200 + Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:1,000 + Anti-R (1:5,000) (HM)S627#3† #### *** ¶¶¶¶ ●●●● #### *** ¶¶¶¶ ●●●● ### *** ¶¶¶¶ ●●●●

%Abs(500) = Percent Absorbance at 500 µg/L relative to maximum Absorbance at 10,000 µg/L † %Abs(500)’= Percent Absorbance at 500 µg/L relative to maximum Absorbance at 100,000 µg/L N/S = noise-to-signal ratio; Anti-R represents anti-rabbit monoclonal antibody 1:200, 1:500, 1:1,000 and 1:5,000 are the dilution factors; the rating criteria are referred to method section 5.2.2.11.

140

Table 5.10 Performances of the capture-detection antibody combinations with (HM)S627#4, (HM)RB#2 and (HM)G#4 as the capture antibodies

Detection Antibodies Capture antibodies %Abs(100) Blank Abs Max-Blk N/S %Abs(100) Blank Abs Max-Blk N/S %Abs(100) Blank Abs Max-Blk N/S (HM)RB#2 1:200 + Anti-R (1:5,000) (HM)RB#2 1:200 + Anti-R (1:7,500) (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)S627#4 #### ** ¶¶¶¶ ●●●● ## * ¶¶¶ ●● ### *** ¶¶¶¶ ●●●● (HM)RB#2 1:500 + Anti-R (1:7,500) (HM)RB#2 1:1000 + Anti-R (1:5,000) (HM)RB#2 1:1000 + Anti-R (1:7,500) (HM)S627#4 ## **** ¶¶¶¶ ●●●● ## **** ¶¶¶¶ ●●●● ## **** ¶¶¶¶ ●●●● (HM)G#4 1:200 + Anti-G (1:10,000) (HM)G#4 1:200 + Anti-G (1:12,500) (HM)G#4 1:200 + Anti-G (1:15,000) (HM)RB#2 #### * ¶¶¶¶ ●●● #### * ¶¶¶¶ ●●● #### * ¶¶¶¶ ●●● (HM)S627#4 1:200 + Anti-S (1:10,000) (HM)S627#4 1:200 + Anti-S (1:12,500) (HM)S627#4 1:200 + Anti-S (1:15,000) (HM)RB#2 #### * ¶¶¶¶ ●●● #### ** ¶¶¶¶ ●●●● #### ** ¶¶¶¶ ●●● (HM)G#4 1:500 + Anti-G (1:15,000) (HM)G#4 1:500 + Anti-G (1:20,000) (HM)G#4 1:1000 + Anti-G (1:15,000) (HM)RB#2 ## *** ¶¶¶¶ ●●●● ## *** ¶¶¶¶ ●●●● ## *** ¶¶¶¶ ●●●● (HM)G#4 1:1000 + Anti-G (1:20,000) (HM)S627#4 1:500 + Anti-S (1:15,000) (HM)S627#4 1:500 + Anti-S (1:20,000) (HM)RB#2 ## *** ¶¶¶¶ ●●●● ### **** ¶¶¶¶ ●●●● ### **** ¶¶¶¶ ●●●● (HM)S627#4 1:1000 + Anti-S (1:15,000) (HM)S627#4 1:1000 + Anti-S (1:20,000) (HM)RB#2 ## **** ¶¶¶¶ ●●●● ### **** ¶¶¶¶ ●●●● (HM)RB#2 1:500 + Anti-R (1:5,000) (HM)RB#2 1:500 + Anti-R (1:7,500) (HM)RB#2 1:1000 + Anti-R (1:5,000) (HM)G#4 # **** ¶¶¶¶ ●●●● # **** ¶¶¶¶ ●●●● # **** ¶¶¶¶ ●●●●

%Abs(100) = Percent Absorbance at 100 µg/L relative to maximum Absorbance at 100,000 µg/L N/S = noise-to-signal ratio 1:200, 1:500 and 1:1,000 are the dilution factors; the rating criteria are referred to method section 5.2.2.11. Anti-R represents anti-rabbit monoclonal antibody; Anti-S represents anti-sheep monoclonal antibody; Anti-G represents anti-goat monoclonal antibody.

141

5.3.4. Indirect sandwich ELISA – calibration curve development

The (CM)RD#3 was the only antibody raised against the crude immunogen selected from the checkerboard. However, the attempts of establishing calibration curves with (CM)RD#3 as the capture antibody showed low assay sensitivity (data not shown), and they were excluded from discussion in this section.

Seven indirect sandwich ELISAs were established with the antibodies raised against the heat-treated immunogen (Figure 5.4 and Table 5.11). The heat-treated mixed immunogen was used as the standard to establish the calibration standard curves. Among the assays with (HM)RB#2 as the capture antibody and (HM)G as the detection antibodies, (HM)RB#2 – (HM)G#3 1:200 showed the highest sensitivity (C50 = 52.2 ±

2.4 μg/L), followed by (HM)RB#2 – (HM)G#2 1:200 (C50 = 57.3 ± 0.3 μg/L) and

(HM)RB#2 – (HM)G#1 1:200 (C50 = 61.0 ± 0.8 μg/L). These three combinations exhibited a similar detection range (C15 – C80) of 10.3 – 310.6 μg/L.

The combinations of (HM)S627#2 – (HM)RB#2 1:500 and (HM)S627#3 – (HM)RB#2

1:500 yielded a similar assay sensitivity (C50) of 69.9 ± 0.0 μg/L and 68.7 ± 0.4 μg/L, respectively. While the (HM)S627#3 – (HM)RB#2 1:500 assay displayed a wider detection range (2.7 ± 0.0 – 911.2 ± 297.7 μg/L) than the (HM)S627#2 – (HM)RB#2 1:500 assay (14.5 ± 2.6 – 421.4 ± 106.0 μg/L).

The last two combinations that interchanged the capture antibody and detection antibody between the sheep antibody (HM)S627#4 and the rabbit antibody (HM)RB#2, exhibited much higher assay sensitivity than the combinations mentioned above. The (HM)S627#4 – (HM)RB#2 1:500 assay showed a sensitivity of 4.4 ± 0.3 μg/L; and the (HM)RB#2 – (HM)S627#4 1:500 assay showed a sensitivity of 9.9 ± 0.4 μg/L. The detection range of the (HM)S627#4 – (HM)RB#2 1:500 assay was 1.0 ± 0.1 – 29.0 ± 1.4 μg/L. The detection range of the (HM)RB#2 – (HM)S627#4 1:500 assay was 1.5 ± 0.0 – 35.2 ± 0.7 μg/L.

142

(a) (b) (c)

Figure 5.4 Calibration curves development using (●) (HM)RB#2 – (HM)G#1 1:200, (■) (HM)RB#2 – (HM)G#2 1:200, (▲) (HM)RB#3 – (HM)G#3 1:200, (▼) (HM)S627#2 – (HM)RB#2 1:500, (♦) (HM)S627#3 – (HM)RB#2 1:500, (○) (HM)S627#4 – (HM)RB#2 1:500 and (◊) (HM)RB#2 – (HM)S627#4 1:500.

143

Table 5.11 The summary table of analytical parameters for the calibration curves developed by the seven capture-detection antibody combinations

2 Capture-detection antibody combination Blank Abs (AU) Max-Blk (AU) C80 (μg/L) C50 (μg/L) C15 (μg/L) R (HM)RB#2 – (HM)G#1 1:200 0.3 ± 0.0 1.3 ± 0.0 310.6 ± 63.8 61.0 ± 0.8 10.3 ± 0.5 0.996 (HM)RB#2 – (HM)G#2 1:200 0.2 ± 0.0 1.1 ± 0.0 307.7 ± 35.6 57.3 ± 0.3 11.1 ± 0.2 0.995 (HM)RB#2 – (HM)G#3 1:200 0.3 ± 0.0 0.9 ± 0.0 308.4 ± 21.7 52.2 ± 2.4 11.9 ± 0.9 0.996 (HM)S627#2 – (HM)RB#2 1:500 0.2 ± 0.0 2.6 ± 0.1 421.4 ± 106.0 69.9 ± 0.0 14.5 ± 2.6 0.994 (HM)S627#3 – (HM)RB#2 1:500 0.3 ± 0.0 1.9 ± 0.1 911.2 ± 297.7 68.7 ± 0.4 2.7 ± 0.0 0.998 (HM)S627#4 – (HM)RB#2 1:500 0.2 ± 0.0 2.7 ± 0.1 29.0 ± 1.4 4.4 ± 0.3 1.0 ± 0.1 0.999 (HM)RB#2 – (HM)S627#4 1:500 0.1 ± 0.0 2.0 ± 0.0 35.2 ± 0.7 9.9 ± 0.4 1.5 ± 0.0 0.999

The results were expressed as mean ± standard deviation (n = 2).

144

5.3.5. Immunoblotting analysis of polyclonal antibodies

The specific bindings of the generated polyclonal antibodies with the 37 heated protein extracts were assessed by immunoblots (Figures 5.5 to 5.7). Three antibodies were selected based on the sensitivity (Figure 5.4) for the immunoblot analysis.

With regard to the rabbit antibody (HM)RB#2 and the goat antibodies ((HM)G#1, (HM)G#2 and (HM)G#3), because the (HM)RB#2 – (HM)G#3 assay showed the highest sensitivity (Table 5.11), due to the affinity maturation with booster injections, (HM)RB#2 and (HM)G#3 were selected for the immunoblots. For the sheep antibodies (HM)S627#2, (HM)S627#3 and (HM)S627#4, the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay exhibited the highest assay sensitivities (Table 5.11), so (HM)S627#4 was selected for the immunoblotting analysis.

The immunoblot images were arranged according to the phylogenetic distance of the 12 fish orders (Wiley and Johnson, 2010; Nelson, 2006). Compared to the goat antibody (HM)G#3 that showed limited bindings at 10 – 15 kDa, the rabbit antibody (HM)RB#2 and the sheep antibody (HM)S627#4 displayed broader cross-reactivity with the tested fish species. The calculated MWs of the monomeric PAV bands detected by (HM)RB#2 and (HM)S627#4 were listed in Table 5.12. Compared to (HM)RB#2, (HM)S627#4 showed more bindings at 10 – 15 kDa in some fish species such as Australian pilchard, coral trout, and sand whiting. These immunoreactive bands showed MWs different from the MWs of PAV demonstrated in Chapters 3 and 4 (Tables 3.2 and 4.1). They were not confirmed to be related to PAV, and they were not included in Table 5.12.

(HM)RB#2 showed bindings to PAVs in 36 out of 37 fish species except for swordfish and (HM)S627#4 showed bindings to PAVs in 35 out of 37 fish species except for swordfish and tiger flathead. A clear binding with the 36 kDa protein in the heated swordfish extract was observed for both (HM)RB#2 and (HM)S627#4. The absence of PAV binding and the presence of the 36 kDa protein binding of the heated swordfish extract were in-line with the SDS-PAGE (Figure 4.1); that is only a 36 kDa protein band was observed in the heated swordfish extract.

Compared to the anti-cod PoAb evaluated in Chapter 4 (Figure 4.2), our (HM)RB#2 and (HM)S627#4 detected more PAVs in species from Scombriformes, Scorpaeniformes, 145

and Tetraodonitiformes which previously showed weak or no reactivity with the anti-cod PoAb in the immunoblot. These detected fish species include Australian bonito, eastern red scorpionfish, dusky flathead, ocean perch, red gurnard and ocean jacket. Noticeably, both (HM)RB#2 and (HM)S627#4 showed weak bindings to yellowfin tuna and mahi-mahi. Yellowfin tuna and mahi-mahi also showed weak or no immunoreactivity in Chapters 3 (Liang et al., 2017) and 4 and in other studies (Lee et al., 2011; Gajewski and Hsieh, 2009; Sharp et al., 2015). As the heated-mixed protein extracts contained not only PAV but also other heat-stable proteins, our antibodies reacted to proteins other than PAV, including the 15 kDa, 25 kDa and 36 kDa proteins, which seemed to be detected in most of the fish species.

146

Figure 5.5 Immunoblotting analysis of heated fish extracts with (HM)RB#2.

147

Figure 5.6 Immunoblotting analysis of heated fish extracts with (HM)G#3.

148

Figure 5.7 Immunoblotting analysis of heated fish extracts with (HM)S627#4.

149

Table 5.12 The calculated MW of thermally stable PAV bands in the heated fish protein detected by (HM)RB#2 and (HM)S627#4

Molecular weight (kDa) of PAV Fish samples 1 2 3 Australian Pilchard 11.7 10.8 10.3 Atlantic Salmon 13.4 11.8 Rainbow Trout 11.8 John Dory 12.7 10.6 Bight Redfish 11.8 Orange Roughy 12.7 12.0 10.6 Sea Mullet 10.8 Sea Garfish 11.3 10.7 Pink Ling 14.8 10.5 Barramundi 11.8 11.1 Blue Cod 12.5 10.8 Blue Eye Trevalla 11.9 Coral trout 12.2 11.7 Crimson Snapper 12.3 11.4 10.5 Eastern School Whiting 11.2 10.7 Jewfish 11.9 Mahi Mahi 11.4 Murray Cod 10.7 10.0 Pink Snapper 13.2 11.1 Red Emperor 11.7 11.2 Sand Whiting 11.7 10.9 Spanish Mackerel 10.6 10.2 Silver Gemfish 11.3 Silver Perch 11.3 10.4 Yellowfin Bream 10.8 Yellowtail Kingfish 11.5 Yellowtail Scad 11.4 Yellowbelly Flounder 10.5 10.0 Australian Bonito 10.5 Swordfish Yellowfin Tuna 11.3 Eastern Red Scorpionfish 10.5 Dusky Flathead 11.5 10.7 Ocean Perch 11.9 11.3 10.8 Red Gurnard 12.7 11.5 10.5 Tiger Flathead 11.2 10.6 Ocean Jacket 11.9 11.4 The MW of PAV detected only by (HM)RB#2 was shaded as light grey; MW of PAV detected only by (HM)S627#4 was shaded as medium grey; MW of PAV detected by both the two antibodies were shaded as dark grey.

150

5.3.6. Cross-reactivity by a checkerboard ELISA

Cross-reactivity study is a critical step in the assessment the applicability of the developed assays. We first conducted a brief screening of cross-reactivity by a checkerboard ELISA.

The assays that include (HM)RB#2 – (HM)G#1, (HM)RB#2 – (HM)G#2 and (HM)RB#2 – (HM)G#3 were not selected for the cross-reactivity study because they did not offer competitive sensitivities and the goat antibody (HM)G#3 show limited bindings to the 37 fish species (Figure 5.6) and was not included in the cross-reactivity.

Figure 5.8 displays the plots of %R values of the tested species for the four selected pairs. If a species had a %R value close to 100, we could assume that the species is very likely to have similar immune-response as the heat-treated immunogen with the antibodies, which is what we were aiming for. Both capture and detection antibodies will influence the assay cross-reactivity.

The more scattered the graph is, the more diverse immune response the antibodies exhibited with the tested fish species. The (HM)S627#2 – (HM)RB#2 assay displayed the most scattered plot. For the (HM)S627#3 – (HM)RB#2 assay, twenty-nine fish species (78% of all tested species) showed %R values within the cut-off points 50 – 150, and among them, fourteen fish species (38%) had %R within the cut-off points 80 – 120. Noticeably, some low %R values were obtained in the (HM)S627#4 – (HM)RB#2 assay (Table 5.11). Nevertheless, the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay exhibited broader cross-reactivity than the (HM)S627#3 – (HM)RB#2 assay. For the (HM)S627#4 – (HM)RB#2 assay, twenty-seven species (72%) showed %R values within 50 – 150 and 16 species (43%) showed %R within 80 – 120. For the (HM)RB#2 – (HM)S627#4 assay, thirty species (84%) had %R values within 50 – 150 and 18 species (49%) showed %R within 20 – 120.

From these results, the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay, which also yielded the two highest assay sensitivities, were selected for further cross-reactivity study with the full calibration curves.

151

(a) (b)

Figure 5.8 The cross-reactivity studies screening of (a) (HM)S627#2 – (HM)RB#2, (b) (HM)S627#3 – (HM)RB#2, (c) (HM)S627#4 – (HM)RB#2, (d) (HM)RB#2 – (HM)S627#4 with 37 fish species. The middle solid line represents 100 %R; the (∙∙∙∙) dotted lines represent 120 or 80 %R; and the (----) dotted lines represent 150 or 50 %R. %Abs(100) was used for (a) (HM)S627#2 – (HM)RB#2, (b) (HM)S627#3 – (HM)RB#2, and %Abs(50) was used for (c) (HM)S627#4 – (HM)RB#2, (d) (HM)RB#2 – (HM)S627#4.

152

5.3.7. Cross-reactivity (species specificity) study using the full calibration curves

The two assays (HM)S627#4 – (HM)R#2 and (HM)RB#2 – (HM)S627#4 showing a broader cross-reactivity with the thirty-seven fish species were selected for more comprehensive investigation of the cross-reactivity quantitatively with the tested fish species. All the calibration curves are demonstrated in Appendix Figures 6 to 9.

The cross-reactivity of the thirty-seven species related to the heat-treated immunogen determined by the (HM)S627#4 – (HM)R#2 assay ranged widely from 0.0% for swordfish to 685% for eastern school whiting (Table 5.13). The (HM)RB#2 – (HM)S627#4 assay exhibited a narrower cross-reactivity ranging from 0.0% for swordfish to 122% for Spanish mackerel. Though a 36 protein kDa was detected by both (HM)RB#2 and (HM)S627#4 in the immunoblots of the swordfish extract, the two assays did not cross-react with swordfish in the sandwich assay.

Similar to our previous findings in Chapter 3 (Liang et al., 2017) and Chapter 4 that fish species from the order of Perciformes showed strong bindings with the anti-PAV antibodies, most of the studied species from Perciformes in this Chapter strongly cross reacted with both assays, except for blue cod and mahi-mahi which showed less than 10% cross-reactivity in both assays. The low cross-reactivity of mahi-mahi correlated with the weak bindings with our antibody (HM)RB#2 and (HM)S627#4 in the immunoblots (Figures 5.5 and 5.7). Other species that exhibited less than 10% cross-reactivity in both assays include yellowbelly flounder, eastern red scorpionfish, and yellowfin tuna. These species are from Pleuronectiformes, Scombrifomres, Scopaeniformes which were known to have weak or no immunoreactivity to anti-PAV antibodies (Lee et al., 2011; Liang et al., 2017; Sharp et al., 2015; Shibahara et al., 2013). However, apart from eastern red scorpion, all the other species from Scopaeniformes, including dusky flathead, ocean perch, red gurnard and tiger flathead did show cross-reactivity but relatively low (12.7 – 66.3%) to the (HM)RB#2 – (HM)S627#4 assay. These results correlate with the immunoblots in Figures 5.5 and 5.7. Compared to the (HM)S627#4 – (HM)R#2 assay that showed strong cross-reactivity to Australian pilchard, the (HM)RB#2 – (HM)S627#4 assay exhibited much lower cross-reactivity to Australian pilchard. The sandwich ELISA by Fæste and Plassen (2008) reported low cross-reactivity (< 1%) with

153

anchovy, which is from the same fish order of Clupeiformes as pilchard. Among other fish species that were used for the preparation of our immunogen, the (HM)S627#4 – (HM)R#2 assay showed stronger cross-reactivity (> 100%) to Atlantic salmon, rainbow trout, pink ling barramundi, coral trout, and Murray cod than the (HM)RB#2 – (HM)S627#4 assay (20 – 50%). This is expected because the rabbit antibody (HM)RB#2, showing higher affinity and specificity to the immunogen, would react strongly to the species in the immunogen.

The two antibodies produced in our study have enabled the development of two sensitive sandwich ELISAs with broad species specificity. In particular, compared to the specificity evaluation in Chapters 3 and 4 and the previous studies (Fæste and Plassen, 2008; Gajewski and Hsieh, 2009; Lee et al., 2011; Sharp et al., 2015), our assays showed a higher cross-reactivity (and broader species specificity) to fish species from Salmoniformes, Ophidiiformes and Scopaeniformes. The related species in our study include Atlantic salmon, rainbow trout, pink ling, dusky flathead, ocean perch, red gurnard and tiger flathead. Compared to the study of Sharp et al. (2015) that showed limited detection to Spanish mackerel, both our assays showed strong cross-reactivity to Spanish mackerel. The limited detection to mahi-mahi, swordfish, and yellowfin tuna observed in our assays and other studies (Chen et al., 2006; Gajewski and Hsieh, 2009; Lee et al., 2011; Sharp et al., 2015) may correlate with the low levels of PAV in these fish, as they are all large migratory fish with dominant dark muscle tissues (Kobayashi et al., 2016d; Tsukamoto, 1984).

154

Table 5.13 The cross-reactivity of 37 fish species related to the heat-treated immunogen determined by the assays (HM)S627#4 – (HM)R#2 and (HM)RB#2 – (HM)S627#4

Cross-reactivity (%) Fish order Fish species (HM)S627#4 – (HM)RB#2 (HM)RB#2 – (HM)S627#4 Clupeiformes Australian Pilchard 126.2 11.0 Rainbow Trout 213.8 45.5 Salmoniformes Atlantic Salmon 108.8 52.7 Zeiformes John Dory 19.2 3.1 Orange Roughy 309.8 4.3 Beryciformes Bight Redfish 18.2 4.9 Mugiliformes Sea Mullet 50.3 34.9 Beloniformes Sea Garfish 262.5 31.2 Ophidiiformes Pink Ling 307.3 50.3 Eastern School Whiting 685.2 96.3 Murray Cod 343.9 34.5 Pink Snapper 327.9 65.0 Barramundi 316.0 20.0 Coral Trout 166.7 20.9 Spanish Mackerel 158.3 122.2 Silver Gemfish 152.0 35.4 Jewfish 130.6 99.1 Sand Whiting 100.0 80.3 Perciformes Yellowtail Scad 82.6 112.1 Silver Perch 64.4 51.6 Yellowfin Bream 64.0 80.8 Crimson Snapper 32.3 11.0 Blue-eye Trevalla 29.6 44.6 Yellowtail Kingfish 18.8 47.8 Red Emperor 18.0 31.5 Blue Cod 1.4 2.5 Mahi Mahi 1.0 0.1 Pleuronectiformes Yellowbelly Flounder 3.0 1.5 Australian Bonito 19.8 12.9 Scombrifomres Yellowfin Tuna 0.2 0.6 Swordfish 0.0 0.0 Ocean Perch 7.1 12.7 Dusky Flathead 4.0 28.9 Scopaeniformes Eastern Red Scorpionfish 3.2 5.7 Red Gurnard 3.0 33.5 Tiger Flathead 0.6 66.3 Tetraodontiformes Ocean Jacket 10.7 13.9 C of heat-treated immunogen (μg/L) % Cross-reactivity = 50 × 100 C50 of heated fish protein (μg/L)

155

5.3.8. Intra-assay precision of important analytical parameters

Based on the recommendations by Thompson et al. (2002) and Abbott et al. (2010), the main validation parameters to be assessed for the developed fish ELISAs include lower limit of detection (C15), assay sensitivity (C50), assay cross-reactivity, assay precision and assay accuracy (including spike and recovery study). This section focuses on the evaluation of the intra-assay precision of the key analytical parameters including C15, C50,

C80, blank Abs (AU), and Max-Blk (AU). A set of data collected from fifteen individual assay runs of the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay performed on different days is plotted in Figure 5.9, to compare the intra-assay precision of the two assays.

Compared to the (HM)RB#2 – (HM)S627#4 assay, more significant variations were observed in the (HM)S627#4 – (HM)RB#2 assay (Figure 5.9a). The %CV values for

C15, C50, C80 and Blank Abs were significantly higher than the acceptable range (<20%) for the intra-assay precision (Table 5.14). The average sensitivity of the (HM)S627#4 –

(HM)RB#2 assay (C50 = 11.9 ± 7.7 μg/L) was almost three-fold lower than the initial sensitivity (C50 = 4.4 ± 0.3 μg/L); while the average sensitivity of the (HM)RB#2 –

(HM)S627#4 assay (C50 = 14.0 ± 5.4 μg/L) was 1.5-fold lower than the initial sensitivity

(C50 = 9.9 ± 0.4 μg/L); suggesting that the (HM)RB#2 – (HM)S627#4 assay showed better intra-assay precision than the (HM)S627#4 – (HM)RB#2 assay. This indicates that when the rabbit antibody (HM)RB#2 (raised against the heat-treated immunogen) were used the detection antibody, though it exhibited higher sensitivity (i.e., low LOD) than the sheep antibody (HM)S627#4 (raised against the same immunogen), it was more susceptible to changes in the antigens, resulting in greater variability of the (HM)S627#4 – (HM)RB#2 assay.

Although 67% of the C80, C50 and C15 values (10 out of 15 assay replicates) obtained from the (HM)RB#2 – (HM)S627#4 assay were within the 95% confidence intervals, only the parameters of max – blk and C15 showed %CV within or close to the acceptable range (<20%) (Abbott et al., 2010); the other 3 parameters including C80, C50, and blank Abs showed 30 – 38 %CV, which are higher than the acceptable range. The reasons which may contribute to this variability include 1) protein loss caused by the absorption

156

of the proteins to the container surface (Simpson, 2010) or 2) molecular modification via PAV polymerisation or PAV aggregation of the heated immunogens, affecting the antibody-antigen binding in the ELISAs.

Notably, both assays achieved very low lower limit of detection (C15): 1.7 ± 0.9 μg/L for the (HM)S627#4 – (HM)RB#2 assay and 1.6 ± 0.4 μg/L for the (HM)RB#2 – (HM)S627#4 assay. The detection limit calculated as the mean measure value of 10 blank sample replicates plus 3 times of the standard deviation (SD) of the mean value (LLOD’) for the (HM)S627#4 – (HM)RB#2 assay was 0.08 μg/L, and the LLOD’ for the (HM)RB#2 – (HM)S627#4 assay was 0.59 μg/L. As there is currently no threshold (reference dose) established for fish allergen, 0.1 mg of protein level (one of the low thresholds estimated for other food allergens, Table 2.3) is used as a putative threshold to estimate the limit of detection required for the detection of fish allergen. Assumingly, if the serving size for a particular fish-containing product is 100 g, the action level transition point based on the VITAL® calculator (Section 2.2.5.2) would be 1,000 μg per L food. This means the lower limit of detection of fish ELISAs is required to be at least 1,000 μg per L food to ensure the reliable detection of VITAL®. The lower limit of detections of both of our assays are much lower than 1,000 μg per L food, suggesting the assays can potentially be used in real world practice to detect trace amount of fish residue in processed foods.

157

Table 5.14 A summary of the analytical parameters of the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay based on data collected from fifteen analyses performed on different days

(HM)S627#4 – (HM)RB#2 Analytical parameters Value SD %CV

C15 (μg/L) 1.7 0.9 49

C50 (μg/L) 11.9 7.7 64

C80 (μg/L) 61.6 29.9 49 LLOD’(μg/L) 0.08 Blank Abs (AU) 0.1 0.0 40 Max - Blk (AU) 2.6 0.3 10 (HM)RB#2 – (HM)S627#4 Analytical parameters Value SD %CV

C15 (μg/L) 1.6 0.4 24

C50 (μg/L) 14.0 5.4 38

C80 (μg/L) 63.5 19.3 30 LLOD’(μg/L) 0.59 Blank Abs (AU) 0.1 0.0 30 Max - Blk (AU) 1.5 0.2 16 SD: standard deviation; %CV: percent coefficient of variation μg/L means μg fish protein per L food

(a) (b)

Figure 5.9 The control chart plots of the (●) C80, (■) C50 and (▲) C15 values from 15 assay replicates for (a) (HM)S627#4 – (HM)RB#2 and (b) (HM)RB#2 – (HM)S627#4. The middle solid lines represent the mean values and the dotted lines represent 95% confidence intervals.

158

5.3.9. Identification of the 36 kDa proteins from the selected fish species by LC-MS-MS

Among the proteins other than PAV that were also detected by our antibodies in most of the fish species (i.e., 15 kDa, 25 kDa and 36 kDa proteins), the 36 kDa proteins have drawn our special attention because they were also detected by the anti-PAV MoAb in Chapter 3, and the ELISAs by Chen and Hsieh (2014) and Gajewski et al. (2009). The ELISAs by Chen and Hsieh (2014) and Gajewski et al. (2009), which specifically targeted the detection of 36 kDa proteins, claimed to show cross-reactivity with a wide range of fish species in the heated form.

To identify and confirm the 36 kDa proteins, the LC-MS-MS analysis was conducted. The Mascot searches with the highest significant protein matches (i.e., number of MS-MS queries matched to the database) and the highest significant peptide sequence matches (n > 10) from the first two protein families are reported in Tables 5.15 and 5.16. Exempt for Atlantic salmon that matched the 36 kDa protein to glyceraldehyde-3-phosphate dehydrogenase only, other analysed fish species showed match to glyceraldehyde-3-phosphate dehydrogenase and tropomyosin α-1 chain. The 36 kDa protein in pilchard has been previously identified as glyceraldehyde-3-phosphate dehydrogenase (van der Ventel et al., 2011).

Considering that glyceraldehyde-3-phosphate dehydrogenase is heat-labile (Wrba et al., 1990) and likely to have been removed along with other precipitated protein after heat treatment, tropomyosin is probably the protein presenting 36 kDa in our heated extracts. Notably, tropomyosin has a very similar pI to that of PAV, which explains the technical challenges we had in separating PAV and the 36 kDa protein in our preliminary work of PAV purification by ion exchange chromatography using a Bio-Rad high Q column (data no shown). This had led us to seek for an alternative approach to purify PAV, which will be discussed in Chapter 6. Our antibodies did show cross-reactivity with tropomyosin.

159

Table 5.15 Identification of the 36 kDa protein band of crude fish extracts from five fish species by LC-MS-MS Protein Accession Prot_matches Prot_sequences Calculated Protein Band Protein identified Fish species coverage numbers _sig _sig pI (%) Glyceraldehyde-3-phosphate gi|185133678 Salmo salar 25 10 51.2 7.81 36 kDa band in crude Australian dehydrogenase pilchard extract Stegastes gi|657551885 Tropomyosin α-1 chain isoform X1 26 17 58.5 4.69 partitus Glyceraldehyde-3-phosphate gi|929063688 Salmo salar 344 27 80.5 8.63 36 kDa band in raw Atlantic dehydrogenase isoform X1 salmon extract gi|185132114 Glycerol-3-phosphate dehydrogenase Salmo salar 69 25 76 6.52 Glyceraldehyde-3-phosphate Stegastes gi|657600123 54 13 44.7 8.58 36 kDa band in raw sea mullet dehydrogenase partitus extract Stegastes gi|657551902 Tropomyosin α-1 chain isoform X3 41 23 74.3 4.69 partitus Glyceraldehyde-3-phosphate Stegastes gi|657600123 75 14 57.7 8.58 36 kDa band in raw pink ling dehydrogenase partitus extract Stegastes gi|657551902 Tropomyosin α-1 chain isoform X3 70 27 75 4.69 partitus Glyceraldehyde-3-phosphate Astyanax gi|597772191 60 13 55.3 8.24 36 kDa band in raw coral trout dehydrogenase mexicanus extract gi|60390740 Tropomyosin α-1 chain Liza aurata 32 19 68.7 4.69 Pro_matches_sig: number of significant protein matches; pro_sequences_sig: number of significant peptide sequence matches.

160

Table 5.16 Identification of the 36 kDa protein band of heated fish extracts from five fish species by LC-MS-MS Accession Prot_matches_ Prot_sequences Protein Caculated Protein Band Protein identified Fish species numbers sig _sig coverage (%) pI Tropomyosin alpha-1 chain Notothenia gi|736211277 95 26 65.5 4.69 36 kDa band in heated Australian isoform X1 coriiceps pilchard extract Myosin heavy chain, fast gi|736211277 Danio rerio 39 27 15.5 5.54 skeletal muscle Tropomyosin alpha-1 gi|185132405 Salmo salar 130 33 78.9 4.63 36 kDa band in heated Atlantic chain-like isoform X1 salmon extract gi|185132405 Myosin heavy chain Salmo salar 26 22 13 5.53 36 kDa band in heated sea mullet gi|60390740 Tropomyosin alpha-1 chain Liza aurata 156 36 81.3 4.69 extract 36 kDa band in heated pink ling Paralichthys gi|598074490 Tropomyosin alpha-1 143 38 75.7 4.67 extract olivaceus 36 kDa band in heated coral trout Epinephelus gi|295792268 Tropomyosin 139 37 80.6 4.69 extract coioides Pro_matches_sig: number of significant protein matches; pro_sequences_sig: number of significant peptide sequence matches.

161

5.4. Conclusion

Two sensitive ELISAs, which were denoted as (HM)S627#4 – (HM)RB#2 assay and

(HM)RB#2 – (HM)S627#4 assay, were developed for the detection of southern hemisphere fish allergens. The average assay sensitivity (C50) of (HM)S627#4 –

(HM)RB#2 assay was 11.9 ± 7.7 μg/L with a detection range (C15 – C80) of 1.7 ± 0.9 –

61.6 ± 29.9 μg/L. The average sensitivity (C50) of the (HM)RB#2 – (HM)S627#4 assay was 14.0 ± 5.4 μg/L with a detection range of 1.6 ± 0.4 – 63.5 ± 19.3 μg/L. The assay sensitivities achieved in our study were only slightly lower than the assay sensitivity

(C50 = 6.6 μg/L) reported by Shibahara et al. (2013), but much higher than the assay sensitivity (C50 = 97.4 μg/L) reported by Cai et al. (2013). The LLODs reported by Cai et al. (2013) and Shibahara et al. (2013) were 0.8 μg/L and 0.58 μg/L, respectively. As mentioned in Section 5.2.2.14, the definition of LLOD in these two studies is different from our study; for direct comparison of their LLODs with our study, we calculated the

LLOD’ using their definition. The LLOD’ of the (HM)S627#4 – (HM)RB#2 assay was

0.08 μg/L, which was much lower than the LLODs reported in the two studies; and the

LLOD’ of the (HM)RB#2 – (HM)S627#4 assay was 0.59 μg/L, which was similar to the

LLOD reported by Shibahara et al. (2013).

The (HM)S627#4 – (HM)RB#2 assay showed a wider cross-reactivity (0.0 – 685.2%) with the thirty-seven southern hemisphere fish species than the (HM)RB#2 –

(HM)S627#4 assay (0.0 – 122.2%). Twenty-seven fish species (73%) and twenty-eight fish species (76%) had cross-reactivity greater than 10% with the (HM)S627#4 –

(HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay, respectively. In particular, our assays showed an improved detection to fish species from Salmoniformes,

Ophidiiformes and Scopaeniformes which showed weak or no immunoreactivity with the anti-parvalbumin antibodies evaluated in Chapters 3 and 4 and in the previous studies (Fæste and Plassen, 2008; Gajewski and Hsieh, 2009; Lee et al., 2011; Sharp et 162

al., 2015). This suggests that our polyclonal antibodies raised against the heat-treated immunogen showed broader species specificity with the commercially important southern hemisphere fish species in Australia.

The comparison of the analytical parameters obtained from 15 individual assay runs of the two assays indicated that the (HM)RB#2 – (HM)S627#4 assay showed better intra-assay precision than the (HM)S627#4 – (HM)RB#2 assay. Therefore, the

(HM)S627#4 – (HM)RB#2 assay was selected for further assay validation of the important analytical parameters including assay cross-reactivity to non-fish samples and assay accuracy evaluation by spike and recovery study in Chapter 6.

163

Chapter 6 Validation of the indirect sandwich ELISA with optimised assay buffer for fish protein detection

6.1. Introduction

The (HM)RB#2 – (HM)S627#4 assay showed better intra-assay precision than the

(HM)S627#4 – (HM)RB#2 assay, but it still displayed higher variability of the C80, C50, and blank Abs values. The variability of the analytical parameters could be due to molecular modification of PAV (and other detectable proteins) via polymerisation or aggregation, therefore affecting antibody-antigen binding. To minimise the assay fluctuation due molecular changes, the purified PAV from one single fish species is proposed to be used as the standard reagent has been proposed for the further validation of the (HM)RB#2 – (HM)S627#4 assay.

Our preliminary work of PAV purification by ion exchange chromatography using a

Bio-Rad high Q column did not completely separate the 36 kDa heat-stable protein from

PAV (data not shown). The 36 kDa protein, which matched to tropomyosin (Section 5.3.9 in Chapter 5) in the LC-MS-MS analysis, shares a similar pI with PAV. Based on the studies by Chen and Hsieh (2014) and (Permyakov, 2006), who separated the 36 kDa protein by 50 – 60% of ammonium sulphate ((NH4)2SO4), and PAV by 70 – 90%

(NH4)2SO4, purification by a gradient ammonium sulphate precipitation was developed.

This Chapter presents the optimisation of the assay reagents and validation of the assay performance using spike and recovery studies. The assay buffer was optimised for the recovery of heated fish protein in food products. With the optimised buffer, the matric effects and cross-reactivity with non-fish proteins were evaluated; spike and recovery of

PAV in four different food matrices was conducted; and a pilot survey of Thai-imported products containing fish in the ingredient list or on PAL statement was also conducted to investigate the applicability of our assay.

164

6.2. Materials and methods

6.2.1 Materials

6.2.1.1. Fish species and commercial food samples

The fish species studied in this Chapter include: Australian pilchard (Sardinops neophilchardus), Atlantic salmon (Salmo salar), rainbow trout (Oncorhynchus mykiss), sea mullet (Mugil cephalus), pink ling (Genypterus blacodes), barramundi (Lates calcarifer), coral trout (Plectropomus leopardus), Murray cod (Maccullochella peelii), pink snapper (Pagrus auratus), sand whiting (Sillago ciliata), Spanish mackerel

(Scomberomorus commerson) and tiger flathead (Neoplatycephalus richardsoni).

Thirty-four food products were purchased from local supermarkets, Sydney, Australia to investigate the effect of blank food matrices on the performances of the (HM)RB#2 –

(HM)S627#4 assay as well as to investigate potential cross-reactivity of the food matrices.

To examine whether fish residues in processed foods can be detected using our immunoassay, ten selected Thai-imported fish products were purchased in Thai groceries, Sydney, Australia, based on the following criteria: 1) products showing the key word “fish” on the product name; 2) listing fish as the ingredient; 3) showing PAL statements of fish residues. The product list can be referred to Table 6.10 in the results and discussion section.

6.2.1.2. Chemicals and reagents

Bicinchoninic acid (BCA) protein assay kit and bovine serum albumin (BSA) standard

(Sigma-Aldrich, St Louis, US); methanol, ethanol, acetic acid and sulfuric acid

(Chemical store, UNSW, Australia); sodium dihydrogen phosphate (NaH2PO4)

(Honeywell International Inc, Marris Plains, US); sodium carbonate (Na2CO3), sodium bicarbonate (NaHCO3), di-sodium hydrogen orthophosphate (Na2HPO4), dodecahydrate,

165

sodium chloride (NaCl), ethylenediaminetetraacetic acid (EDTA), sodium acetate, ammonium sulphate ((NH4)2SO4) (Univar, Downers Grove, US); Tris (Tris(hydroxymethyl)aminomethane), 40% acrylamide/bis solution; 2x Laemmli sample buffer, N,N,N’,N’-tetramethylethylenediamine (TEMED), dithiothreitol (DTT), glycine, sodium dodecyl sulfate (SDS), 10×Tris/glycine buffer/SDS, Precision Plus ProteinTM

Dual Colour Standards (Bio-Rad, Hercules, US); β-cyclodextrin, urea hydrogen peroxide, Coomassie Brilliant Blue R-250, Tween-20, 3,3’,5,5’-tetramethylbenzidine

(TMB), dimethyl sulfoxide (DMSO), thiomersal (Sigma-Aldrich, St Louis, US);

Trans-Blot® Turbo™ 5x Transfer Buffer (Bio-Rad, Hercules, US); anti-rabbit MoAb conjugated with horseradish peroxidase (HRP) (Abcam, Cambridge, UK); anti-sheep and anti-goat MoAbs conjugated with HRP (ThermoFisher Scientific, Waltham, US);

BSA powder (Bovogen Biologicals Pty Ltd, East Keilor, Australia); glycerol (Chemi

Supply Ltd., Corby, UK); SuperSignal™ West Pico Chemiluminescent Substrate

(ThermoFisher Scientific, Waltham, US).

6.2.1.3. Buffers

The following buffer solutions were used in the studies and their abbreviation will be used thereafter: PBS buffer 1 (0.01 M phosphate buffered saline (pH 7.4)); PBS buffer 2

(0.05 M phosphate buffered saline (pH 7.2)); blocking buffer 1 (1% BSA in 0.05 M

Tris-HCl-0.9% NaCl (pH 7.5) with 0.05% Tween-20 (1% BSA/TBS-T)); coating buffer

(50 mM carbonate buffer (pH 9.6)); blocking buffer 2 (1% BSA in PBS buffer (1%

BSA-PBS)); diluent buffer (1% BSA/PBS with 0.05% Tween-20 (1% BSA-PBS-T)); stop solution (1.25 M sulphuric acid); washing buffer (0.05% Tween-20 in reverse osmosis water); extraction buffer for the commercial products (0.5% SDS in PBS-T); glycerol buffer including 10% (v/v) glycerol/1% BSA-PBS-T, 20% (v/v) glycerol/1%

BSA-PBS-T, and 40% (v/v)/1% BSA-PBS-T; glycine buffer including 0.1 M glycine/1%

BSA-PBS-T, 0.5 M glycine/1% BSA-PBS-T, and 1.0 M glycine/1% BSA-PBS-T; high salt buffer including 0.5 M NaCl/1% BSA-PBS-T, 1.0 M NaCl/1% BSA-PBS-T, and 1.5

166

M NaCl/1% BSA-PBS-T.

6.2.1.4. Instruments

Heraeus pico 17 centrifuge, Heraeus Multifuge X3R centrifuge (Thermo Fisher

Scientific, Waltham, US); Mini high speed refrigerated centrifuge (LabCo); Julabo TW

20 water bath (John Morris Scientific Pty Ltd, Sydney, Australia); Water bath (Retak,

Melbourne, Australia); Revolver (Labnet International Inc, Edison, US); Power Pac HV high-voltage power supply (Bio-Rad, Hercules, US); Heating and drying oven (Weiss technik, Königswinter, Germany); Trans-Blot® TurboTM transfer system (Bio-Rad, US);

Optima microplate reader (BMG LABTECH, Ortenberg, Germany); Cary Eclipse

Fluorescence Spectrophotometer (Agilent Techonologies, Santa Clara, US),

ChirascanTM-plus CD Spectrometer (Applied Photophysics Ltd., Leatherhead, UK),

IFM-700G food processor (Iwatani, Japan); ChemiDocTM MP imaging system (Bio-Rad,

Hercules, US); Magic bullet food processor (Homeland Housewares, US).

6.2.2. Methods

6.2.2.1. Soluble fish protein extraction

The protocol for soluble fish protein extraction can be referred to Section 5.2.2.2 in

Chapter 5.

6.2.2.2. Heat treatment of crude fish protein extracts

The crude fish protein extracts were heat-treated based on the method described in

Section 4.2.2.2 of Chapter 4.

6.2.2.3. Separation of PAV from heated fish protein extracts by ammonium sulphate precipitation

Heated fish protein extracts were first brought to 20% saturation of (NH4)2SO4, and the solution was mixed for 30 min. Then (NH4)2SO4 was increased to 40% saturation, and the solution was mixed for another 30 min. Precipitated proteins were separated by 167

centrifugation at 13,000 × g at 4 °C for 15 min. The collected pellet was denoted as the

20 – 40% fraction. The resulting supernatant was subjected to a second precipitation by increasing the (NH4)2SO4 to 60% saturation. After mixing for 1 h, precipitated proteins were separated by centrifugation as previously. The collected pellet was denoted as the

40 – 60% fraction. The resulting supernatant was subjected to a third precipitation by increasing the (NH4)2SO4 to 70% saturation. After mixing for 30 min, (NH4)2SO4 was finally brought to 99% saturation, and the solution was mixed for 1 h. After centrifugation, the resulting pellet, which is denoted as the 70 – 99% fraction, was suspended in PBS buffer 2 and dialysed against PBS buffer 2 for three days with several buffer changes. The final products (i.e., the 70 – 99% fraction) were the partially purified PAVs. Thiomersal (0.005%, w/v) was added to the partially purified PAVs for storage at 4 °C until use.

6.2.2.4. Soluble protein extraction of blank food matrices and Thai-imported fish products

The sample extraction protocol was modified from Watanabe et al. (2005). Around 10 g of each commercial product was homogenised with the Iwatani food processor. A portion of 1.0 g of each of the homogenised samples was mixed with 19 mL of extraction buffer and extracted overnight at room temperature with gentle rolling on a rotating mixer. After centrifugation at 4,193 × g at room temperature for 20 min, the supernatant was collected and filtered through 185 mm diameter Whatman filter papers

No. 1 (Sigma-Aldrich, St Louis, US). The filtered solutions were the original sample extracts, which was stored at 4 °C for up to 7 days.

For studying the effect of food matrices on the assay performances as well as the potential effects on cross-reactivity with the food matrices, three concentrations of the thirty-four food samples (in original sample extracts, 10 times dilution of the sample extracts in 1 M glycine/1% BSA-PBS-T and 20 times dilution of the samples extracts in

1 M glycine/1% BSA-PBS-T) were tested with the indirect sandwich ELISA. 168

For the Thai-product survey on PAL statements, the original sample extracts were diluted 20 times with 1 M glycine/1% BSA-PBS-T, then tested the indirect sandwich

ELISA.

Spanish mackerel PAV, staring from 1,000 μg/L, was diluted in serial dilutions in 1 M glycine/1% BSA-PBS-T as the calibration standard for both studies.

6.2.2.5. Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis

(SDS-PAGE)

The partially purified PAVs (3 μg) were separated by SDS-PAGE based on the method described in Section 3.2.2.3 in Chapter 3.

6.2.2.6. Immunoblot analysis

Immunoblots of the partially purified PAV and five selected food matrices with

(HM)RB#2 and (HM)S627#4 antibodies were performed based on the method described in Section 3.2.2.4 in Chapter 3.

6.2.2.7. Indirect sandwich ELISA

The indirect sandwich ELISA with the diluted partially purified PAVs or heat-treated immunogen or Thai fish product extracts was conducted according to the method described in Section 5.2.2.10 in Chapter 5. The partially purified PAVs and the heat-treated immunogen, staring from 1,000 μg/L, were diluted in serial dilutions in diluent buffer as the standard solutions for the cross-reactivity study by the full calibration curves. The calibration curve prepared with Spanish mackerel PAV was the control, for comparison of the key analytical parameters (i.e., C15, C50, C80, blank abs, and maximum abs – blank abs) calculated from the calibration curves prepared with

Spanish mackerel PAV with different additives (i.e., glycerol buffers, glycine buffers, or high salt buffers), to study the effects of different additives on the assay performances.

169

6.2.2.8. Spike and recovery studies

The spike and recovery studies were conducted with two approaches, one with a simple matrix and the other with food matrix effects.

For the first approach, the purified Spanish mackerel PAV spiked in diluent buffer with different additives to the final concentration levels of 100,000 μg/L, 20,000 μg/L, and

2,000 μg/L) were tested with the indirect sandwich ELISA, to evaluate the protein recoveries in optimised buffer solutions.

For the second approach, the partially purified Spanish mackerel PAV were spiked in a homogenised blank food samples (1 g of food sample homogenised with 19 mL extraction buffer) to the final concentration levels of 40,000 μg/L, 8,000 μg/L, 2,000

μg/L, and 400 μg/L. The spiked sample mixtures were further extracted overnight following the protein extraction protocol described in Section 6.2.2.4. The spiked sample extracts were further diluted 20 times with 1 M glycine/1% BSA-PBS-T and assayed with the indirect sandwich ELISA.

Recovery of the purified PAV was calculated as the quotient of detected PAV concentration and spiked PAV concentration (Equation 6.1).

detected PAV concentration (μg/L) % Recovery = ×100 Equation 6.1 spiked PAV concentration (μg/L)

6.2.2.9. Circular dichroism (CD) analysis and fluorescence analysis of partially purified Spanish mackerel PAV

The purified Spanish mackerel PAV or the native cod PAV (purified) was diluted in PBS buffer 2 to 100,000 μg/L. The solution was transferred to a quartz cuvette with a 0.5 mm optical path. The far ultraviolet (UV) CD spectra were recorded from 195 nm to 260 nm at room temperature by the ChirascanTM-plus CD Spectrometer. The resolution was 0.2 nm, with a bandwidth of 1.0 nm. The spectra were recorded at 0.5 s per point.

The fluorescence spectra of the diluted partially purified PAV or the native cod PAV in 170

PBS buffer 2 (100,000 μg/L) were recorded based on the method described in Section

4.2.2.8 in Chapter 4.

6.2.2.10. Data analysis

The ELISA data analysis was the same as described in Section 3.2.2.7 in Chapter 3.

As described in Section 5.2.2.14 of Chapter 5, the C15 is defined as the lower limit of detection (LLOD) in our study. The LLOD (μg/g) of blank food matrices and Thai food products is calculated as LLOD of calibration curve (μg/L) × dilution factor /1000.

The cross-reactivity of the partially purified PAVs related to the heat-treated immunogen was calculated based on Equation 5.5 in Chapter 5.

The analytical parameters (C15, C50 and C80) of the indirect sandwich ELISA in response to the partially purified Spanish mackerel PAV diluted in the diluent buffer with additives were analysed using one-way analysis of variance (ANOVA) by Excel. The p-values > 0.05 for all the three parameters was considered statistically insignificant and a p-value < 0.05 for either one of the parameters was considered statistically insignificant.

6.3 Results and discussion

6.3.1. Purification of PAV from the heated crude fish protein extracts by ammonium sulphate precipitation

Figure 6.1 illustrates the protein profiles of the partially purified PAVs from twelve of the fish species that were selected for our immunogen. Apart from Atlantic salmon which had lost one of the PAV isoforms compared to the corresponding protein profile of the heated extract in Figure 5.1b, the other fish species showed an identical number of PAV isoforms to those of the heated extracts.

The purity of the partially purified PAV solutions was assessed by the immunoblots with

171

the rabbit antibody (HM)RB#2 (Figure 6.2) and the goat antibody (HM)S627#4 (Figure

6.3). Atlantic salmon, rainbow trout, and pink ling showed weak bindings at ~36 kDa with (HM)RB#2, which corresponds to the faint ~36 kDa bands in the SDS-PAGE

(reducing gel). Only Atlantic salmon and tiger flathead showed bindings at ~36 kDa with (HM)S627#4. Compared to the bindings at ~36 kDa, the bindings at ~25 kDa were more consistently shown in most of the fish species with both antibodies. Similar bindings with the anti-cod PoAb in the heated fish extracts were observed in Chapter 4

(Figure 4.2), of which the detected proteins are probably the PAV dimers. The predominant protein bands detected by both antibodies were PAVs at 10 – 15 kDa, which has confirmed the success of the purification of PAV from by the simple

(NH4)2SO4 precipitation.

Figure 6.1 Protein profiles of the partially purified PAVs separated by SDS-PAGE with Coomassie brilliant blue staining.

172

Figure 6.2 Immunoblot analysis of the partially purified PAVs with the rabbit antibody (HM)RB#2.

Figure 6.3 Immunoblot analysis of the partially purified PAVs with the sheep antibody (HM)S627#4.

6.3.2. Cross-reactivity of the partially purified PAVs related to the heat-treated mixed immunogen determined by the (HM)RB#2 – (HM)S627#4 assay

To select a purified PAV as the calibration standard for further assay validation, the immunological responses (to the polyclonal antibodies) of the partially purified PAVs

173

relative to the immunological response of the heat-treated immunogen were assessed by the (HM)RB#2 – (HM)S627#4 assay. As shown in Table 6.1, the % cross-reactivity ranged from 42.6% for Australian pilchard to 546.8% for coral trout. The calibration curves are shown in Appendix Figure 10. The partially purified PAV of Spanish mackerel, which will be denoted as SM PAV thereafter, showed a high cross-reactivity

(222.5%) with the (HM)RB#2 – (HM)S627#4 assay and was selected as the calibration standard for the further validation.

Table 6.1 The cross-reactivity of the purified PAVs relative to the heat-treated immunogen determined by the (HM)RB#2 – (HM)S627#4 assay

Fish species % Cross-reactivity Coral trout 546.8 Spanish mackerel 222.5 Rainbow trout 167.4 Sea mullet 161.1 Sand whiting 150.0 Barramundi 114.2 Pink snapper 100.2 Murray cod 83.2 Pink ling 73.1 Atlantic salmon 70.4 Tiger flathead 45.2 Australian pilchard 42.6

C of heat-treated immunogen (μg/L) % Cross-reactivity = 50 × 100; C50 of purified PAV (μg/L)

6.3.3. The preliminary spike and recovery study of SM PAV in the standard diluent buffer

The preliminary spike and recovery study of SM PAV in our standard diluent buffer (1%

BSA/PBS-T, pH 7.2) showed less than 10% recovery (Table 6.2). These unexpectedly low recoveries alered us that the diluent buffer was not a suitable assay buffer for conducting further assay validation.

The factors contributing to the low recoveries include: 1) pH, 2) ionic strength of the extraction and assay buffers, 3) solubility of and stability of PAV in PBS buffer 174

(L’Hocine and Pitre, 2016), and 4) molecular modification during the storage, sample preparation and assay (e.g., PAV polymerisation or PAV aggregation), affecting antibody-antigen binding. Since PAV is extremely water soluble and can be readily extracted with low ionic strength buffer or even just water (Permyakov, 2006), we proceeded to investigate the molecular structure of the SM PAV by the CD and fluorescence analyses.

Table 6.2 Recovery of fish protein of Spanish mackerel in 1% BSA-PBS-T

Spiking level (μg/L) % Recovery 100,000 8.3 20,000 BLD 2,000 BLD

BLD: below the limit of detection

6.3.4. Protein structural investigation of SM PAV by fluorescence and CD analysis

The CD analysis was conducted to investigate the secondary structure of SM PAV as a reference PAV. Since the secondary and tertiary structures of PAV are conserved among different fish species (Sharp and Lopata, 2014), including Atlantic cod and Spanish mackerel, the purified native cod PAV was selected as a reference (Figure 6.4). The CD spectrum of the native cod PAV showed two minima at 208 and 222 nm with a zero-crossing point at around 202 nm, which was in line with the study of

Bugajska-Schretter et al. (2000). Compared to the spectrum of the native cod PAV, the spectrum of SM PAV showed a shift of the zero-crossing point to a lower wavelength

(199 nm) and some losses of the ellipticity at 222 nm compared with that at 208 nm.

The shift of the zero-crossing to a lower wavelength indicates protein unfolding (de

Jongh et al., 2013) and the loss of the ellipticity at 222 nm indicates a decrease in the

α-helical content of the protein (Permyakov, 2006). This change correlates with a transition from a native protein state to a molten globule state (Bugajska-Schretter et al.,

2000). The molten globule state involves the exposure of some hydrophobic structures of a protein core into the solvent environment, which could initiate protein aggregation

175

(Somkuti et al., 2012). The CD spectrum of our heated SM PAV showed unfolding and become more hydrophobic, which leads to formation of protein aggregates in PBS buffer.

Figure 6.4 The far UV CD analysis of SM PAV (blue line) and native cod PAV (red line).

Fluorescence analysis was also conducted to investigate the tertiary structural difference between the native cod PAV and the SM PAV (Figure 6.5). The peak intensity (Imax) of

SM PAV was thirteen-fold lower than the Imax of the native cod PAV, revealing that tryptophan in the hydrophobic core was highly exposed to the solvent environment, which induced the sharp decrease of the fluorescence signal (Wrolstad et al., 2005).

176

Figure 6.5 Fluorescence analysis of SM PAV (blue line) and native cod PAV (red line).

The fluorescence analysis is in line with the CD analysis and provides further evidence of the speculated formation of protein aggregates in the heated SM PAV during storage, which may explain the low recoveries in the preliminary spike and recovery study. In addition, the observation of protein aggregation may further explain the binding at 25 kDa (potential PAV dimers) observed in the immunoblots with antibodies (HM)RB#2 and (HM)S627#4 (Figures 6.2 and 6.3) as well as the variability of assay performances

(Figure 5.9 and Table 5.14) observed in Chapter 5.

6.3.5. Effects of different additives on assay performances

To develop sample preparation conditions to prevent protein aggregation, we selected three additives, glycerol, glycine and NaCl, to increase the surface tension of the water, thus keeping the PAVs in their monomeric form. Glycerol (10 – 40%) is known to stabilise the hydration shells of the proteins and increase the molecular density of the solution, thus protecting the proteins from aggregation (Simpson, 2010). Glycine (20 –

500 mM) is known to increase the surface tension of the water without showing significant electrostatic interactions with the proteins (Simpson, 2010). High salt (1 M

NaCl) buffer was used to increase the solubility of proteins (Khuda et al., 2015).

Figure 6.6 illustrates the calibration curves of the (HM)RB#2 – (HM)S627#4 assay

177

prepared in three concentrations (i.e., 10%, 20%, 40%) of glycerol in diluent buffer. As shown in Table 6.4, with the increase in glycerol concentration, the assay sensitivity

(C50) decreased. The calibration curve prepared in 40% (v/v) glyceol/1% BSA-PBS-T that exhibited a 2.3-fold decrease in assay sensitivity (C50 = 57.6 ± 11.0 μg/L) with a wider detection range (11.4 ± 0.3 – 270.6 ± 47.7 μg/L), compared to the calibration curves prepared in the diluent buffer (i.e., control, C50 = 24.1 ± 2.2 μg/L, 3.9 ± 0.2 –

115.4 ± 1.1 μg/L). The one-way ANOVA analysis of the analytical parameters (C15, C50 and C80, Table 6.4) showed that there were no statistically significant differences observed between the calibration curves prepared with 10% and 20% glycine in diluent buffer and the control (the p-values were greater than 0.05), suggesting the assay performances were not affected by glycerol up to 20%.

Figure 6.6 Calibration curves of the (HM)RB#2 – (HM)S627#4 assay prepared in (●) 1% BSA-PBS-T; (■) 10% (v/v) glycerol/1% BSA-PBS-T; (▲) 20% (v/v) glycerol/1% BSA-PBS-T; (▼) 40% (v/v)/1% BSA-PBS-T with SM PAV.

178

Table 6.3 Analytical parameters of the calibration curves prepared in three concentrations of glycerol, SM PAV as the calibration standard

10% glycerol/ 20% glycerol/ 40% glycerol/ Parameter 1% BSA/PBS-T 1% BSA-PBS-T 1% BSA-PBS-T 1% BSA-PBS-T

C15 (μg/L) 3.9 ± 0.2 3.9 ± 0.1 3.9 ± 0.3 11.4 ± 0.3

C50 (μg/L) 24.2 ± 2.6 24.1 ± 2.2 30.3 ± 1.6 57.6 ± 11.0

C80 (μg/L) 115.4 ± 1.1 108.9 ± 7.8 116.0 ± 9.3 270.6 ± 47.7 Blank (AU) 0.03 ± 0.01 0.03 ± 0.00 0.03 ± 0.00 0.03 ± 0.00 Max – Blk (AU) 1.0 ± 0.0 0.9 ± 0.0 0.9 ± 0.0 0.9 ± 0.0 Results expressed as mean ± standard deviation (n = 3)

Table 6.4 One-way ANOVA analysis of the calibration curves prepared in three concentrations of glycerol, SM PAV as the calibration standard

1% BSA-PBS-T (Control)

C15 C50 C80 10% glycerol/1% BSA-PBS-T p = 0.8394 p = 0.9926 p = 0.3607 20% glycerol/1% BSA-PBS-T p = 0.8694 p = 0.1062 p = 0.9387 40% glycerol/1% BSA-PBS-T p = 0.0011 p = 0.0505 p = 0.0442

For determining the maximum concentration of glycine that did not affect the antibody-antigen binding, we screened a wider range of concentrations (i.e., 0.1 M, 0.5

M and 1.0 M) than what was recommended in the literature (Simpson, 2010). All the four calibration curves showed similar assay sensitivities (C50 = 22.8 ± 1.8 – 25.9 ± 3.3 μg/L) and similar detection ranges (~4.8 – 111.7 μg/L) (Figure 6.7, Table 6.5). The one-way ANOVA analysis of the three analytical parameters (Table 6.6) showed that there were no statistically significant differences observed between the calibration curves prepared in the three concentrations of glycine in diluent buffer and the control.

This indicates the assay performances were not affected by glycine up to 1.0 M.

179

Figure 6.7 Calibration curves of the (HM)RB#2 – (HM)S627#4 assay prepared in (●) 1% BSA-PBS-T; (■) 0.1 M glycine/1% BSA-PBS-T; (▲) 0.5 M glycine/1% BSA-PBS-T; (▼) 1.0 M glycine/1% BSA-PBS-T with SM PAV.

Table 6.5 Analytical parameters of the calibration curves prepared in three concentrations of glycine, SM PAV as the calibration standard

0.1 M glycine/ 0.5 M glycine/ 1.0 M glycine/ Parameter 1% BSA-PBS-T 1% BSA-PBS-T 1% BSA-PBS-T 1% BSA-PBS-T

C15 (μg/L) 3.9 ± 0.4 3.9 ± 0.3 4.8 ± 1.2 3.9 ± 0.05

C50 (μg/L) 23.1 ± 4.3 22.8 ± 1.8 25.4 ± 2.8 25.9 ± 3.3

C80 (μg/L) 109.3 ± 1.0 100.7 ± 26.6 111.7 ± 15.0 106.5 ± 18.6 Max abs (AU) 0.04 ± 0.00 0.04 ± 0.00 0.04 ± 0.00 0.04 ± 0.00 Blank (AU) 1.0 ± 0.0 1.0 ± 0.0 1.0 ± 0.0 1.0 ± 0.0 Results expressed as mean ± standard deviation (n = 3)

Table 6.6 One-way ANOVA analysis of the calibration curves prepared in three concentrations of glycine, SM PAV as the calibration standard

1% BSA-PBS-T (Control)

C15 C50 C80 0.1 M glycine/1% BSA-PBS-T p = 0.9250 p = 0.9624 p = 0.6921 0.5 M glycine/1% BSA-PBS-T p = 0.4621 p = 0.5593 p = 0.8441 1.0 M glycine/1% BSA-PBS-T p = 0.9665 p = 0.5155 p = 0.8522

As shown in Figure 6.8 and Table 6.7, increasing NaCl concentration from 0.5 M to 1.5

M gradually decreased the assay sensitivity (C50) from 29.8 ± 2.1 μg/L to 57.4 ± 23.0

180

μg/L, implying that high concentrations of NaCl affected the antibody-antigen binding in our assay. The one-way ANOVA analysis of the three analytical parameters (Table 6.8) showed that there were no statistically significant differences observed between the calibration curves prepared with 0.5 M glycine in diluent buffer and the control (the p-values were greater than 0.05), suggesting the assay performances were not affected by NaCl up to 0.65 M (0.5 M NaCl + 0.15 M NaCl in the PBS buffer).

Figure 6.8 Calibration curves of the (HM)RB#2 – (HM)S627#4 assay were prepared in (●) 1% BSA-PBS-T; (■) 0.5 M NaCl/1% BSA-PBS-T; (▲) 1.0 M NaCl/1% BSA-PBS-T; (▼) 1.5 M NaCl/1% BSA-PBS-T with SM PAV.

Table 6.7 Analytical parameters of the calibration curves prepared in three concentrations of NaCl, SM PAV as the calibration standard

0.5 M NaCl/ 1.0 M NaCl/ 1.5 M NaCl/ Parameter 1% BSA-PBS-T 1% BSA-PBS-T 1% BSA-PBS-T 1% BSA-PBS-T

C15 (μg/L) 3.9 ± 0.3 3.6 ± 0.3 3.7 ± 0.1 3.4 ± 0.6

C50 (μg/L) 22.6 ± 3.0 29.8 ± 2.1 37.3 ± 2.2 57.4 ± 23.0

C80 (μg/L) 111.5 ± 20.3 106.4 ± 17.5 115.4 ± 4.4 322.7 ± 33.8 Max abs (AU) 0.03 ± 0.01 0.03 ± 0.00 0.03 ± 0.03 0.03 ± 0.00 Blank (AU) 1.0 ± 0.0 1.0 ± 0.0 1.0 ± 0.0 0.8 ± 0.0 Results expressed as mean ± standard deviation (n = 3)

181

Table 6.8 One-way ANOVA analysis of the calibration curves prepared in three concentrations of NaCl, SM PAV as the calibration standard

1% BSA-PBS-T (Control)

C15 C50 C80 0.5 M NaCl/1% BSA-PBS-T p = 0.5311 p = 0.1115 p = 0.8125 1.0 M NaCl/1% BSA-PBS-T p = 0.6362 p = 0.0332 p = 0.8168 1.5 M NaCl/1% BSA-PBS-T p = 0.4768 p = 0.1427 p = 0.0169

After determining the optimum concentrations of the three additives that would not compromise the assay performances, three different buffers: buffer A (20% glycerol in 1%

BSA/PBS-T), buffer B (1.0 M glycine in 1% BSA/PBS-T) and buffer C (0.5 M NaCl in

1% BSA/PBS-T) were chosen to conduct the spike and recovery study of SM PAV

(Table 6.9). A drastic improvement of the protein recoveries was achieved with all three buffer compositions. The protein recovery with buffer A ranged from 91.6% to 102.4%; the protein recovery with buffer B ranged from 84.8% to 122.1%; and the protein recovery with buffer C ranged from 101.3% to 113.7%. Three individual assay runs were performed for all the spiking levels of the three buffers. All the recovery rates were within the ideal recovery range (80 – 120%) (Abbott et al., 2010) with good assay repeatability (%CV < 23). With the optimised assay buffer, we proceeded to surveying

Thai-imported fish products to investigate the applicability of the (HM)RB#2 –

(HM)S627#4 assay using one of the optimised buffer solutions (buffer B) on detecting fish residues in commercial products.

Table 6.9 Recovery of SM PAV in three buffer solutions

Spiking level (μg/L) 100,000 20,000 2,000 % Recovery SD (%) %CV % Recovery SD (%) %CV % Recovery SD (%) %CV Buffer A 102.4 6 6 91.6 18 20 99.3 6 6 Buffer B 122.1 14 12 113.4 17 15 84.8 20 23 Buffer C 101.3 8 8 113.7 9 8 107.2 12 11 Buffer A: 20% glycerol in 1% BSA/PBS-T; Buffer B: 1.0 M glycine in 1% BSA/PBS-T Buffer C: 0.5 M NaCl in 1% BSA/PBS-T SD: standard deviation; %CV: percent coefficient of variation. The SD and %CV were calculated from three individual assay runs. 182

6.3.6 Study of the effect of blank food matrices on the performances of the assay and the potential cross-reactivity of non-fish foods with the assay

Thirty-four blank foods and ingredients from various food categories (e.g., high carbohydrate foods such as rice cakes and corn flour, dairy products such as skim milk and yoghurt and high fat foods such as peanut and macadamia nut) were tested for matrices effects and cross-reactivity with the (HM)RB#2 – (HM)S627#4 assay (Table

6.10). The food and ingredients were tested in three dilutions: 1 in 20 dilution (the original samples extracts), 1 in 200 dilution (10 times dilution of the original sample extracts in 1 M glycine/1% BSA-PBS-T) and 1 in 400 dilution (20 times dilution of the original sample extracts in 1 M glycine/1% BSA-PBS-T). As illustrated in Table 6.10, the original sample extracts (1 in 20 dilution) of only three food products (i.e., salted pepper squid, chicken breast bites and mussel) showed immunological response above

LLOD with the assay. The cross-reactivity of the three food products was eliminated when further diluting the three food products to 1 in 200 dilution and 1 in 400 dilution.

Noticeably, the 1 in 200 dilutions of rice cake and corn flour showed immunological response above LLOD with the assay. The high carbohydrate contents of these two food products might have high affinity with our antibodies, resulting in the observed reactivity. However, the 1 in 20 dilutions of rice cake and corn flour, which were tested on the second day, did not react with the assay. This probably because the carbohydrates might have precipitated after overnight storage and the reactivity was eliminated. These results suggested that food products with high carbohydrate contents may interfere with the assay and overnight storage may be an option to eliminate the interference.

Immunoblot analysis was conducted to visualise the cross-reactivity of salted pepper squid, chicken breast bites and mussel, which showed detected immunological response, with our antibodies. Since our antibodies cross react with fish tropomyosin, prawn and crab have also been included in the immunoblot analysis to investigate if our antibodies cross-react with tropomysin in shellfish. As shown in Figure 6.10, the rabbit antibody

183

(HM)RB#2 showed more cross-reactivity with the selected food samples than the sheep antibody (HM)S627#4. The cross-reacting bands in salted pepper squid with

(HM)RB#2 were around 100 kDa, 50-37 kDa, and 15 kDa, while these cross-reacting bands were almost invisible with (HM)S627#4 in the immunoblot. The cross-reacting bands of chicken breast bites with both (HM)RB#2 and (HM)S627#4 were around

50-37 kDa. The cross-reacting bands in prawn and crab with (HM)RB#2 were around

100 kDa, 37 kDa, and 20 kDa, whereas similarly, these cross-reacting bands were almost invisible with (HM)S627#4. These results of prawn and crab correspond with the non-reactivity of the original sample extracts of prawn and crab in ELISA using

(HM)S627#4 as the capture antibody (Table 6.10). Contradictorily, the original sample extract of mussel showed the greatest cross-reactivity (0.1 μg/g) with the assay in

ELISA, while there were no obvious cross-reacting bands observed with both antibodies in the immunoblot. The immunoblot analysis was conducted three days after the sample extraction; the soluble proteins in the mussel sample extracts might have precipitated, resulting in the weak protein bands shown in the SDS-PAGE (Figure 6.9), and consequentially, no obvious bands observed in the immunoblot. Nevertheless, the 1 in

200 dilution and 1 in 400 dilution results have suggested that the cross-reactivity of mussel, as well as pepper and salted squid and chicken breast bites, can be eliminated by diluting the sample extracts to reduce the cross-reacting proteins to a non-detected concentration. Considering that all the food samples showed no matric effects and no cross-reactivity at 1 in 400 dilution, 1 in 400 dilution has been chosen as the dilution factor to conduct spike and recovery study in blank food matrices.

184

Table 6.10 Overview of thirty-four different foods tested for matrix effect and cross-reactivity with the (HM)RB#2 – (HM)S627#4 assay

1 in 20 dilution 1 in 200 dilution 1 in 400 dilution Product Name Abs SPQ Abs SPQ Abs SPQ (AU) (μg/g) (AU) (μg/g) (AU) (μg/g) Almond coconut muesli 0.012 BLD 0.000 BLD 0.005 BLD Rice cakes 0.106 BLD 0.286 0.2 0.004 BLD Corn cakes 0.000 BLD 0.252 BLD 0.004 BLD Wrap 0.000 BLD 0.127 BLD 0.000 BLD Bread 0.000 BLD 0.039 BLD 0.000 BLD Biscuit 0.000 BLD 0.041 BLD 0.000 BLD Corn flour 0.000 BLD 0.376 0.4 0.000 BLD Wheat flour 0.000 BLD 0.027 BLD 0.000 BLD Chocolate 0.000 BLD 0.014 BLD 0.000 BLD Chicken corn soup 0.000 BLD 0.015 BLD 0.000 BLD Salted pepper squid 0.393 0.05 0.081 BLD 0.025 BLD Ham 0.000 BLD 0.048 BLD 0.000 BLD Pumpkin soup 0.000 BLD 0.005 BLD 0.000 BLD Yoghurt 0.000 BLD 0.015 BLD 0.000 BLD Egg 0.000 BLD 0.002 BLD 0.000 BLD Vegetable Samosa 0.000 BLD 0.008 BLD 0.000 BLD Chicken breast bites 0.398 0.05 0.051 BLD 0.054 BLD Rocket salad 0.008 BLD 0.007 BLD 0.000 BLD Vegetable-lentil pie 0.000 BLD 0.004 BLD 0.000 BLD Walnut 0.000 BLD 0.032 BLD 0.000 BLD Hazelnut 0.000 BLD 0.007 BLD 0.000 BLD Almond 0.002 BLD 0.041 BLD 0.000 BLD Oats 0.000 BLD 0.048 BLD 0.000 BLD Sesame 0.000 BLD 0.019 BLD 0.000 BLD Green peas 0.000 BLD 0.027 BLD 0.005 BLD Beef Meat 0.119 BLD 0.008 BLD 0.007 BLD Prawn 0.000 BLD 0.003 BLD 0.000 BLD Mussel 0.687 0.1 0.095 BLD 0.138 BLD Crab 0.027 BLD 0.034 BLD 0.000 BLD Kiwi 0.000 BLD 0.035 BLD 0.000 BLD Soy milk 0.000 BLD 0.027 BLD 0.000 BLD Skim milk 0.072 BLD 0.020 BLD 0.000 BLD Macadamia nut 0.000 BLD 0.068 BLD 0.000 BLD Peanut 0.000 BLD 0.035 BLD 0.000 BLD LLOD (μg/g) 0.02 (Abs = 0.214) 0.2 (Abs = 0.284) 0.4 (Abs = 0.236) SPQ – SM PAV Equivalent; BLD: below lower limit of detection (LLOD) 0.02 = 1 × 20 / 1000 (μg/g); 0.2 = 1 × 200 / 1000 (μg/g); 0.4 = 1 × 400 / 1000 (μg/g). 185

Figure 6.9 Protein profiles of selected food samples separated by SDS-PAGE.

(a) (b)

)

Figure 6.10 Immunoblot analysis of selected food samples with (a) (HM)RB#2 and (b) (HM)S627#4.

6.3.7 Spike and recovery of SM PAV in blank food matrices

The spike and recovery study of SM PAV was performed in four food matrices of almond coconut muesli, chicken corn soup, ham, and vegetable-lentil pie. Regardless of the food type, the intra-assay protein recovery ranged from 88.1% to 112.2% (Table

6.11), which was within the recommended protein recovery range of 80 – 120%. The intra-assay precision (repeatability) was less than 18% which is also within the recommended range of less 20%. The inter-assay validation (performed by two analytes) also demonstrated good protein recovery of 87.2 – 117.3% with satisfactory inter-assay precision (reproducibility) of less 18%. 186

Table 6.11 Intra-assay recovery of SM PAV in four different food matrices

Spiking level (μg/L) 40,000 8,000 2,000 400 % Recovery SD (%) %CV % Recovery SD (%) %CV % Recovery SD (%) %CV % Recovery SD (%) %CV Almond coconut muesli 105.3 11 10 95.3 7 8 112.2 8 7 88.0 5 5 Chicken corn soup 110.2 20 18 101.3 10 10 103.4 14 14 89.0 10 10 Ham 101.9 9 9 90.9 3 3 98.6 5 5 97.3 14 14 Vegetable lentil pie 95.3 15 16 88.1 10 12 96.3 8 8 97.5 6 6

SD: standard deviation; %CV: percent coefficient of variation. The SD and %CV were calculated from four individual replicates.

Table 6.12 Inter-assay recovery of SM PAV in four different food matrices

Spiking level (μg/L) 40,000 8,000 2,000 400 % Recovery SD (%) %CV % Recovery SD (%) %CV % Recovery SD (%) %CV % Recovery SD (%) %CV Almond coconut muesli 106.9 1 1 94.0 2 2 117.3 7 6 101.0 18 18 Chicken corn soup 106.4 5 5 95.6 8 9 114.2 15 13 105.3 10 10 Ham 100.0 3 3 91.0 0 0 104.8 9 8 98.2 1 1 Vegetable lentil pie 98.3 4 4 87.2 1 2 105.6 13 12 101.6 6 6

SD: standard deviation; %CV: percent coefficient of variation. The SD and %CV were calculated from two individual assay runs performed by two people.

187

6.3.8. A pilot survey of Thai-imported products containing fish in the ingredient list or on PAL statement

Considering the fact that Thailand has played an important role in global food export market (Rimpeekool et al., 2015) and fish products are largely manufactured and consumed in Thailand, a product survey of Thai-manufactured food was conducted to evaluate the applicability of the (HM)RB#2 – (HM)S627#4 assay on detecting fish residues in commercial products (Table 6.12). The ten Thai-imported products were selected for their processing, food matrices and potential presence of fish residues.

The (HM)RB#2 – (HM)S627#4 assay was able to detect fish protein from five of the ten fish products (Table 6.13), including fermented fish seasoning powder, fermented fish extracts, chili paste with smoke salmon flavour, Mamya curry sauce for noodles and sour fish sausage. A sour fish sausage was selected as a positive control because more than 60% of the ingredients was basa fish according to the product information sheet; and the (HM)RB#2 – (HM)S627#4 assay detected 4.3 μg fish protein per gram food from the fish sausage. The highest fish protein concentration (6.0 μg/g) was obtained from the fermented fish seasoning powder, which had no fish allergen label.

The other two products with detected residues that were labelled showed positive results of 0.4 μg/g and 5 μg/g.

The negative results shown in the fish sauce and the yellow curry paste may be related to the processing conditions. The fermentation of fish sauce, involves extensive hydrolysis of fish proteins by endogenous and exogenous proteases of fish and bacteria

(Fukui et al., 2012; Lee et al., 2015), and may produce hydrolysates which may not be readily detectable by the developed ELISA. The difference between the detection of fish proteins in fermented fish seasoning powder and non-detection of fish proteins in fish sauce might be because of the different production processes of the two fish products. In

188

the production process of fish sauce, supernatant produced by fish fermentation for around 6 to 18 months is drained and collected as fish sauce; whereas, whole fermented fish after fermentation for around 1 month is dried and ground as the fish seasoning powder. The fermentation period of fish sauce is longer than that of fish seasoning powder, and only supernatant is collected in fish sauce. Therefore, the concentration of detectable fish proteins including PAV in fish sauce may be much lower than those in fish seasoning powder, leading to different results of the two products. The yellow curry paste with the PAL statement of “may contain fish derivatives in shrimp paste” may not contain detectable fish protein including PAV, because the fish derivatives may be fish gelatine and fish isinglass which can be largely free of PAV residues by preparation procedures such as washing of the skin and acid hydrolysis (Koppelman et al., 2012;

Taylor et al., 2004; Shibahara et al., 2013). Shibahara et al. (2013) also reported negative results in a fish sauce and a fish collagen powder tested in their study with their sandwich ELISA.

As for the other three products with fish allergen labels, but found no fish residues, the high oil content in the products could have interfered with protein extraction and the antibody-antigen binding in the ELISA, even though 0.5% SDS was added in the extraction buffer to improve the protein extractability (Khuda et al., 2015; Steinhoff et al., 2011). Though tween-20 was used in the extraction buffer to emulsify oil content in the sample, further clean-up procedure such as defatting the sample by hexane is still required to completely remove oil in the sample preparation to further improve protein recovery.

189

Table 6.13 Determination of fish protein concentration in ten Thai-imported fish products using the (HM)RB#2 – (HM)S627#4 assay

Product name SPQ (μg/g) Fish allergen labels Fermented fish seasoning powder 6.0 No fish ingredient indicated Holy basil seasoning paste BLD Contains soybean, wheat and fish Yellow curry paste BLD May contain fish derivatives in shrimp paste Fish sauce BLD Contains Anchovy fish extract 60% Spicy dipping sauce BLD Contains fish Fermented fish extract 0.4 Fish as ingredient Chili paste-smoked salmon flavour 1.6 Smoked fish and fish sauce as ingredient Mamya curry sauce for noodles 5.0 Contains fish Chilli paste in soybean oil BLD Contains fish Sour fish sausage 4.3 Basa fish as ingredient

SPQ – SM PAV Equivalent; BLD: below LLOD 0.4 μg/g (ppm). The C15 was 1 μg/L (ppb) for the calibration curve the Thai-product survey. Considering the dilution factor 400 and 1 g of food used in the assay, the LLOD was 1 × 400 / 1000 = 0.4 μg/g (ppm).

6.4. Conclusion

With a simple gradient ammonium sulphate precipitation, PAVs from twelve fish species were purified. The immunoblots with the antibodies (HM)RB#2 and

(HM)S627#4 confirmed that, apart from some medium or weak bindings at 36 kDa and

25 kDa, the predominant proteins detected in the solutions were PAVs.

Heat-induced PAV aggregation had caused extremely low protein recovery in the diluent buffer (1% BSA-PBS-T). With the addition of three different additives (i.e., glycerol, glycine and NaCl) primarily to increase the aqueous surface tension, the protein recovery showed a drastic improvement to 84.8 – 122.1% in all three optimised buffer solutions (i.e., 20% glycerol in 1% BSA-PBS-T, 1.0 M glycine in 1% BSA-PBS-T and

0.5 M NaCl in 1% BSA-PBS-T).

With the optimised assay buffer 1.0 M glycine in 1% BSA-PBS-T, thirty-four blank food and ingredients from various food categories (e.g., high carbohydrate foods such as rice cakes and corn flour, dairy products such as skim milk and yoghurt and high fat

190

foods such as peanut and macadamia nut) were tested for matrix effect and cross-reactivity with the (HM)RB#2 – (HM)S627#4 assay. The results showed food matrices with high carbohydrate contents showed detectable immunological responses with the assay, due to the high affinity of the carbohydrate with the antibodies. This matrix effect could be eliminated by overnight storage when the carbohydrate contents become precipitates. Three food products (i.e., salted pepper squid, chicken breast bites and mussel) showed cross-reactivity with the assay, of which the cross-reactivity could be eliminated by diluting the sample extracts to a non-detectable concentration, but in a range that does not compromise. The other two tested crustacean species prawn and crab did not show cross-reactivity with the assay, because the cross-reacting bands with the detection antibody (HM)RB#2 in the two species were almost invisible with the capture antibody (HM)S627#4 in the immunoblot analysis.

Based on the study of matrices effects and cross-reactivity of non-fish food and ingredients, spike and recovery study of SM PAV in four blank food matrices (i.e., almond coconut muesli, chicken corn soup, ham, and vegetable-lentil pie) was conducted to further validate the (HM)RB#2 – (HM)S627#4 assay. Regardless of the food types, both intra-assay and inter-assay validation showed good protein recoveries

(88.1 – 112.2% and 87.2 – 117.3%, respectively) with satisfactory assay precisions

(<18%).

After confirming good protein recoveries were obtained in blank food matrices using the optimised assay buffer, a product survey was conducted with ten Thai-imported fish products. The (HM)RB#2 – (HM)S627#4 assay showed positive results in five out of the ten products, including the sour fish sausage which was made from more than 60% of basa fish and the fermented fish seasoning powder without any fish allergen labels.

Among the products with no detectable residues, the fish sauce and the yellow curry paste may not have fish residues detectable by the ELISA, and this could be due to the

191

extensive processing such as extensive washing and acid hydrolysis of fish proteins.

The other three products with no detectable residues contained high oil content, which could interfere with the antibody-antigen binding. These products may require further clean-up procedures such as defatting the sample by hexane to completely remove interfering oil in the sample preparation to improve protein recovery.

192

Chapter 7 Conclusion and future work

7.1. Conclusion

This thesis presents a new approach to develop new polyclonal antibodies specific to heat stable fish protein/allergen and statistically assess their species specificity, and to develop a sensitive immunodiagnostic test (based on a sandwich ELISA) for the detection of commercially important southern hemisphere fish residues in processed food. This new immunodiagnostic test would be a valuable addition to the decision support tools for better management of fish allergen labelling in processed foods.

Two anti-PAV antibodies, each raised against different fish PAV (i.e., anti-cod PoAb and anti-carp MoAb have not validated for the detection of southern hemisphere fish species) were used as model antibodies to study cross-reactivity in relation to the phylogenetic relationship, and their detection capability for southern hemisphere fish species, especially those of commercial importance.

In Chapter 3, a new analytical approach by combining the quantitative indirect ELISA and the quantitative SDS-PAGE was developed to quantitatively assess the cross-reactivity of the two anti-PAV antibodies with thirty-seven fish species. A correlation index was established to statistically evaluate the correlation between the

PAV content (estimated from the quantitative SDS-PAGE) in crude fish protein extracts and the corresponding immunoreactivity of PAB (determined by the quantitative indirect ELISA). Sixteen of the thirty-seven species (43%) showed a positive correlation (R2 = 0.74) between the PAV content and the immunoreactivity of PAV with the anti-cod PoAb; while no correlation between the PAV content and the immunoreactivity of PAV was established with the anti-carp MoAb. The polyclonal antibody raised against cod PAV outperformed the monoclonal antibody raised against carp PAV, showing better species specificity for the 37 commercially important fish

193

from the southern hemisphere, as well as quantitatively detecting parvalbumins. Our approach of using the correlation index to assess the cross-reactivity confirmed the previous studies (Lee et al., 2011; Sharp et al., 2015), that the immunoreactivity of different fish species with the anti-PAV antibodies demonstrated order specificity. Fish species from the order of Perciformes showed strong binding with the anti-PAV antibodies; whereas species from Salmoniformes, Ophidiiformes, Scombriformes,

Scorpaeniformes, and Tetraodontiformes showed weak or no binding.

In Chapter 4, we adopted thermal processing to remove less desirable proteins from the crude fish extracts, thus further concentrating more desirable proteins containing heat stable PAV. With the analytical approach we established in Chapter 3, the impacts of thermal processing on the immunoreactivity of PAV in southern hemisphere fish species with the two model anti-PAV antibodies were investigated. The PoAb still outperformed the MoAb in quantitatively detecting thermally treated PAV. Heat treatment induces

Ca2+ loss in PAV, leading to the loss of detectability of the anti-carp MoAb in some species; while the detectability of the anti-cod PoAb was less affected by the thermal processing. A significantly improved correlation between the PAV content in the heated fish extracts and the immunoreactivity of PAV were observed for both antibodies (i.e.,

R2 = 0.82 for the anti-cod PoAb and R2 = 0.55 for the anti-carp MoAb). A similar trend of order specificity was observed with the heat-treated extracts. As the raw extracts, fish species from the order of Perciformes showed strong immunoreactivity with both antibodies; whereas species from Salmoniformes, Ophiidiformes, Scombriformes,

Scorpaeniformes, and Tetraodontiformes showed weak or no immunoreactivity. These results indicate that heat-treated PAV is molecularly more stable than untreated PAV and is a better immunogen for raising specific antibodies and a target for developing quantitative fish ELISA.

The knowledge we gained from Chapters 3 and 4 includes: 1) the polyclonal antibody

194

outperforms the monoclonal antibody in quantitatively detecting both native PAVs and thermally treated PAVs; 2) heat-treated PAV is more stable protein and a better target for developing a quantitative ELISA with broader species specificity; 3) both native PAVs and heat-treated PAVs demonstrate order specificity. This has guided us in the selection of specific Australian fish species/order to be included in the preparation of immunogen mix. Consequently, the fish species selected included the fish order that showed strong immunoreactivity to the antibodies (i.e., Perciformes) as well as the fish order that showed weak or no immunoreactivity to the antibodies (e.g., Salmoniformes,

Ophiidiformes, Scombriformes, and Scorpaeniformes) for antibody production.

Polyclonal antibodies were raised by immunising rabbits, goat, and sheep with the heat-treated mixed immunogen prepared from thirteen fish species (i.e., Australian pilchard, Atlantic salmon, rainbow trout, sea mullet, pink ling, barramundi, coral trout,

Murray cod, pink snapper, sand whiting, Spanish mackerel, yellowfin tuna, and tiger flathead) and developed two indirect sandwich ELISAs with broad specificity for southern hemisphere fish allergen detection as discussed in Chapter 5.

In Chapter 5, by screening a hundred and fifty antibody pairs, two sandwich ELISAs were developed, demonstrating high sensitivity. With the heat-treated immunogen as the calibration standard, the (HM)S627#4 – (HM)RB#2 assay yielded an average C50 value of 11.9 ± 7.7 μg/L with a detection range (C15 – C80) of 1.7 ± 0.9 – 63.5 ± 29.9 μg/L.

The (HM)RB#2 – (HM)S627#4 assay yielded an average C50 of 14 ± 5.3 μg/L with the detection range of 1.6 ± 0.4 – 63.5 ± 19.3 μg/L.

The immunoblot analysis of thirty-seven southern hemisphere fish species with our polyclonal antibodies showed that the rabbit antibody (HM)RB#2 showed bindings to

PAVs in thirty-six out of thirty-seven fish species (97.3%) except for swordfish and the sheep antibody (HM)S627#4 showed bindings to PAVs in thirty-five out of thirty-seven fish species (94.6%), except for swordfish and tiger flathead. Compared to the 195

immunoblot with the anti-cod PoAb in Chapter 4, the (HM)RB#2 and (HM)S627#4 showed bindings to PAV in a greater number of fish species from Scombriformes,

Scorpaeniformes, and Tetraodonitiformes, which showed week or no reactivity with the anti-cod PoAb or the anti-carp MoAb

Sequentially, the cross-reactivity study with the full calibration curves was conducted for more objective assessment of the cross-reactivity by the two quantitative assays with the tested fish species. Compared to the qualitative cross-reactivity evaluation by the immunoblots, the cross-reactivity evaluation by the full calibration curves in ELISAs revealed significantly greater diversity. The (HM)S627#4 – (HM)RB#2 assay showed a cross-reactivity ranging from 0.0% for swordfish to 685.2% for eastern school whiting; and the (HM)RB#2 – (HM)S627#4 assay exhibited a cross-reactivity ranging from 0.0% for swordfish to 122.2% for Spanish mackerel. Nevertheless, twenty-seven fish species

(73%) and twenty-eight fish species (76%) had cross-reactivity greater than 10% with the (HM)S627#4 – (HM)RB#2 assay and the (HM)RB#2 – (HM)S627#4 assay, respectively. In particular, our assays showed an improved detection of fish species from Salmoniformes, Ophidiiformes, and Scopaeniformes, including Atlantic salmon, rainbow trout, pink ling, dusky flathead, ocean perch, red gurnard, and tiger flathead, which showed weak or no immunoreactivity with the model antibodies evaluated in

Chapters 3 and 4 and in the previous studies (Fæste and Plassen, 2008; Gajewski and

Hsieh, 2009; Lee et al., 2011; Sharp et al., 2015). As (HM)S627#4 did not show detection to PAV (Figure 5.7 and Table 5.12), the detection of tiger flathead by the

ELISAs comes from other immunoreactive bands (e.g., 25 kDa and 15 kDa proteins) shown in Figure 5.7. Our assays showed broader cross-reactivity with the commercially important southern hemisphere fish species and are of great potential to be used as a decision support tool in detecting fish residues in processed foods.

Through the evaluation of the intra-assay precision of the important analytical

196

parameters of the two assays, we found that the (HM)RB#2 – (HM)S627#4 assay showed better assay precision than the (HM)S627#4 – (HM)RB#2 assay. However, the

(HM)RB#2 – (HM)S627#4 assay still displayed high variability in the analytical parameters C80, C50 and blank Abs. One of the possible reasons we speculated contributing to the assay variability was polymerisation of the heat-treated immunogen or protein aggregation, affecting antibody-antigen binding; and our speculation was supported by the evidence of protein aggregation by the CD analysis and fluorescence analysis in Chapter 6.

In Chapter 6, the PAVs from twelve selected fish species were purified from other heat-stable proteins, especially the 36 kDa protein (matched to tropomyosin by

LC/MS/MS), by a gradient ammonium sulphate precipitation in Chapter 5. Protein aggregation caused extremely low recoveries (< 8.3%) of the PAV from Spanish mackerel in the original assay buffer (1% BSA/PBS-T). The addition of additives (such as glycerol, glycine and NaCl) to our diluent buffer (i.e., 20% glycerol in 1% BSA/PBS-T, 1.0 M glycine in 1% BSA/PBS-T, and 0.5 M NaCl in 1% BSA/PBS-T) significantly increased the assay recovery rate to 84.8 – 122.1%.

Among the thirty-four non-fish foods and ingredients tested for matrices effects and cross-reactivity with the (HM)RB#2 – (HM)S627#4 assay, the high carbohydrate foods which may be sticky, showed non-specific binding. This matrix effect could be eliminated by resting it overnight to allow the carbohydrate to precipate. Three food products (i.e., salted pepper squid, chicken breast bites and mussel) showed cross-reactivity with the assay, and could be eliminated by diluting the sample extracts to a non-detectable concentration.

The spike and recovery study of SM PAV in four blank food matrices (i.e., almond coconut muesli, chicken corn soup, ham, and vegetable-lentil pie) was conducted to further validate the performance of (HM)RB#2 – (HM)S627#4 assay. Both intra-assay 197

and inter-assay validation showed good protein recoveries (88.1 – 112.2% and 87.2 –

117.3%, respectively) with satisfactory assay precisions (<18 %CV).

A pilot product survey was conducted with ten Thai-imported fish products differing in their processing, food matrices and the presence of fish residues. The (HM)RB#2 –

(HM)S627#4 assay showed positive results in five of the ten products (50%) The detected products include a sour fish sausage, which was made from more than 60% of basa fish, as a positive sample and the fermented fish seasoning powder without any fish allergen labels. Among the negative products, the fish sauce and the yellow curry paste may not contain detectable fish residues that can be detectable by the developed

ELISA possibly due to the fermentation, extensive processing such as extensive washing and acid hydrolysis of fish proteins.

This project has provided valuable insights into the possibility of the development of immunoassays for the detection/quantification of an allergen with molecular diversity such as fish PAV. We successfully developed an ELISA for the detection of fish residues in diverse fish species important to Australia. This will support the development of commercial ELISA test kits for the detection of southern hemisphere fish allergens in processed foods in the near future.

7.2. Future work

Order specificity relative to PAV detection may guide the development of more specific immunoassays targeting fish species from a wide range of fish orders to improve the quantitative detection of commercially important fish species.

More efforts should also be directed to improve the precision of the important analytical parameters (i.e., C15, C50, C80) of our assay. The molecular modification of the heat-treated immunogen, storage and assay conditions such as buffer composition, incubation time and temperature are attributed to the variability of our assay

198

performances. With the optimised assay buffers obtained in Chapter 6, stability tests of the partially purified PAVs combining with structural characterisation should be conducted to further optimise the calibration protein.

Sample extraction and preparation procedure should be optimised and validated for a wide range of food matrices especially products with high oil content such as fish paste or salad dressing.

199

List of references

AAS, K. & ELSAYED, S. M. 1969. Characterization of a major allergen (cod): Effect of enzymic hydrolysis on the allergenic activity. Journal of Allergy, 44, 333-343. ABARES 2014. Australian fisheries and aquaculture statistics 2013. 12/2014 ed. Fisheries Research and Development Corporation: Department of Agriculture - Australian Government. ABBOTT, M., HAYWARD, S., ROSS, W., GODEFROY, S. B., ULBERTH, F., VAN HENGEL, A. J., ROBERTS, J., AKIYAMA, H., POPPING, B., YEUNG, J. M., WEHLING, P., TAYLOR, S. L., POMS, R. E. & DELAHAUT, P. 2010. Validation Procedures for Quantitative Food Allergen ELISA Methods: Community Guidance and Best Practices. Journal of AOAC International, 93, 442-450. AHSAN, N., RAO, R. S. P., GRUPPUSO, P. A., RAMRATNAM, B. & SALOMON, A. R. 2016. Targeted proteomics: Current status and future perspectives for quantification of food allergens. Journal of Proteomics, 143, 15-23. ALLEN, K. J., REMINGTON, B. C., BAUMERT, J. L., CREVEL, R. W. R., HOUBEN, G. F., BROOKE-TAYLOR, S., KRUIZINGA, A. G. & TAYLOR, S. L. 2014a. Allergen reference doses for precautionary labeling (VITAL 2.0): Clinical implications. Journal of Allergy and Clinical Immunology, 133, 156-164. ALLEN, K. J., TURNER, P. J., PAWANKAR, R., TAYLOR, S., SICHERER, S., LACK, G., ROSARIO, N., EBISAWA, M., WONG, G., MILLS, E. N. C., BEYER, K., FIOCCHI, A. & SAMPSON, H. A. 2014b. Precautionary labelling of foods for allergen content: are we ready for a global framework? The World Allergy Organization Journal, 7, 10-10. ALLERGEN BUREAU, A. 2012. Food Industry Guide to the Voluntary Incidental Trace Allergen Labelling (VITAL) Program Version 2.0 [Online]. Available: http://allergenbureau.net/wp-content/uploads/2013/11/VITAL-Guidance-docume nt-15-May-2012.pdf [Accessed 19/06/2017]. ALLERGEN BUREAU, A. 2016. The VITAL Program [Online]. Available: http://allergenbureau.net/vital/ [Accessed 19/06/2017]. ARINO, A., BELTRAN, J. A., HERRERA, A. & RONCALES, P. 2005. Fish. In: CABALLERO, B., ALLEN, L. H. & PRENTICE, A. (eds.) Encyclopedia of Human Nutrition. Second Edition ed. USA: Academic Press. BALLMER-WEBER, B. K., FERNANDEZ-RIVAS, M., BEYER, K., DEFERNEZ, M., SPERRIN, M., MACKIE, A. R., SALT, L. J., HOURIHANE, J. O. B., ASERO, R. & BELOHLAVKOVA, S. 2015. How much is too much? Threshold dose distributions for 5 food allergens. Journal of Allergy and Clinical Immunology, 135, 964-971. BARNETT, J., MUNCER, K., LEFTWICH, J., SHEPHERD, R., RAATS, M. M., 200

GOWLAND, M. H., GRIMSHAW, K. & LUCAS, J. S. 2011. Using 'may contain' labelling to inform food choice: a qualitative study of nut allergic consumers. BMC Public Health, 11, 734. BEALE, J. E., JEEBHAY, M. F. & LOPATA, A. L. 2009. Characterisation of purified parvalbumin from five fish species and nucleotide sequencing of this major allergen from Pacific pilchard, Sardinops sagax. Molecular Immunology, 46, 2985-2993. BEN-SHOSHAN, M., HARRINGTON, D. W., SOLLER, L., FRAGAPANE, J., JOSEPH, L., ST PIERRE, Y., GODEFROY, S. B., ELLIOT, S. J. & CLARKE, A. E. 2010. A population-based study on peanut, tree nut, fish, shellfish, and sesame allergy prevalence in Canada. Journal of Allergy and Clinical Immunology, 125, 1327-1335. BERCHTOLD, M. W., BRINKMEIER, H. & MÜNTENER, M. 2000. Calcium Ion in Skeletal Muscle: Its Crucial Role for Muscle Function, Plasticity, and Disease. Physiological Reviews, 80, 1215. BERNHISEL-BROADBENT, J., SCANLON, S. M. & SAMPSON, H. A. 1992. Fish hypersensitivity: I. In vitro and oral challenge results in fish-allergic patients. Journal of Allergy and Clinical Immunology, 89, 730-737. BOUSQUET, J., BJÖKSTEN, B., BRUIJNZEEL-KUOMEN, C. A. F. M., HUGGETT, A., ORTOLANI, C., WARNER, J. O. & SMITH, M. 1998. Scientific criteria and the selection of allergenic foods for product labelling. Allergy, 53, 3-21. BREITENEDER, H. 2006. Classifying food allergens. In: KOPPELMAN, S. J. & HEFLE, S. L. (eds.) Detecting Allergens in Food. Cambridge: Woodhead Publishing. BROWNRIDGE, P., DE MELLO, L. V., PETERS, M., MCLEAN, L., CLAYDON, A., COSSINS, A. R., WHITFIELD, P. D. & YOUNG, I. S. 2009. Regional variation in parvalbumin isoform expression correlates with muscle performance in common carp (Cyprinus carpio). Journal of Experimental Biology, 212, 184-193. BRUIJNZEEL-KOOMEN, C., ORTOLANI, C., AAS, K., BINDSLEV-JENSEN, C.,JORKSTEN, B., MONERET-VAUTRIN, D., WUTHRICH, B. 1995. Position paper: Adverse reaction to food. Allergy, 50, 623. BUGAJSKA-SCHRETTER, A., ELFMAN, L., FUCHS, T., KAPIOTIS, S., RUMPOLD, H., VALENTA, R. & SPITZAUER, S. 1998. Parvalbumin, a cross-reactive fish allergen, contains IgE-binding epitopes sensitive to periodate treatment and Ca2+ depletion. Journal of Allergy and Clinical Immunology, 101, 67-74. BUGAJSKA-SCHRETTER, A., GROTE, M., VANGELISTA, L., VALENT, P., SPERR, W. R., RUMPOLD, H., PASTORE, A., REICHELT, R., VALENTA, R. & SPITZAUER, S. 2000. Purification, biochemical, and immunological

201

characterisation of a major food allergen: different immunoglobulin E recognition of the apo- and calcium-bound forms of carp parvalbumin. Gut, 46, 661-669. BUGYI, Z., BAUMERT, J. L. & TAYLOR, S. L. 2016. Evaluation of commercial enzyme-linked immunosorbent assays for the detection of fish residues in food. Journal of Food Science, In Press. BURKS, A. W., TANG, M., SICHERER, S., MURARO, A., EIGENMANN, P. A., EBISAWA, M., FIOCCHI, A., CHIANG, W., BEYER, K., WOOD, R., HOURIHANE, J., JONES, S. M., LACK, G. & SAMPSON, H. A. 2012. ICON: Food allergy. Journal of Allergy and Clinical Immunology, 129, 906-920. BURNEY, P., SUMMERS, C., CHINN, S., HOOPER, R., VAN REE, R. & LIDHOLM, J. 2010. Prevalence and distribution of sensitization to foods in the European Community Respiratory Health Survey: a EuroPrevall analysis. Allergy, 65, 1182-1188. CAI, Q.-F., LIU, G.-M., LI, T., HARA, K., WANG, X.-C., SU, W.-J. & CAO, M.-J. 2010. Purification and Characterization of Parvalbumins, the Major Allergens in Red Stingray (Dasyatis akajei). Journal of Agricultural and Food Chemistry, 58, 12964-12969. CAI, Q. F., WANG, X. C., LIU, G. M., ZHANG, L., RUAN, M. M., LIU, Y. & CAO, M. J. 2013. Development of a monoclonal antibody-based competitive enzyme linked-immunosorbent assay (c-ELISA) for quantification of silver carp parvalbumin. Food Control, 29, 241-247. CARRERA, M., CAÑAS, B. & GALLARDO, J. M. 2012. Rapid direct detection of the major fish allergen, parvalbumin, by selected MS/MS ion monitoring mass spectrometry. Journal of Proteomics, 75, 3211-3220. CELIO, M. R., BAIER, W., SCHÄRER, L., DE VIRAGH, P. & GERDAY, C. 1988. Monoclonal antibodies directed against the calcium binding protein parvalbumin. Cell calcium, 9, 81-86. CHATTERJEE, U., MONDAL, G., CHAKRABORTI, P., PATRA, H. K. & CHATTERJEE, B. P. 2006. Changes in the Allergenicity during Different Preparations of Pomfret, Hilsa, Bhetki and Mackerel Fish as Illustrated by Enzyme-Linked Immunosorbent Assay and Immunoblotting. International Archives of Allergy and Immunology, 141, 1-10. CHEN, J., LIAO, Y., ZHANG, H., ZHAO, H. & LI, H. 2012. Prevalence of food allergy in children under 2 years of age in three cities in China. Zhonghua er ke za zhi= Chinese journal of pediatrics, 50, 5-9. CHEN, L., HEFLE, S. L., TAYLOR, S. L., SWOBODA, I. & GOODMAN, R. E. 2006. Detecting fish parvalbumin with commercial mouse monoclonal anti-frog parvalbumin IgG. Journal of agricultural and food chemistry, 54, 5577-5582. CHEN, Y.-T. & HSIEH, Y.-H. P. 2014. A sandwich ELISA for the detection of fish and

202

fish products. Food Control, 40, 265-273. CONNETT, G. J., GEREZ, I., CABRERA-MORALES, E. A., YUENYONGVIWAT, A., NGAMPHAIBOON, J., CHATCHATEE, P., SANGSUPAWANICH, P., SOH, S. E., YAP, G. C., SHEK, L. P. C. & LEE, B. W. 2012. A Population-Based Study of Fish Allergy in the Philippines, Singapore and Thailand. International Archives of Allergy and Immunology, 159, 384-390. CRESPO, J. F., PASCUAL, C., BURKS, A. W., HELM, R. M. & ESTEBAN, M. M. 1995. Frequency of food allergy in a pediatric population from Spain. Pediatric Allergy and Immunology, 6, 39-43. CROWTHER, J. R. 2009. The ELISA Guidebook, Totowa, NJ, Humana Press. DAS DORES, S., CHOPIN, C., ROMANO, A., GALLAND-IRMOULI, A. V., QUARATINO, D., PASCUAL, C., FLEURENCE, J. & GUÉANT, J. L. 2002a. IgE-binding and cross-reactivity of a new 41 kDa allergen of codfish. Allergy, 57, 84-87. DAS DORES, S., CHOPIN, C., VILLAUME, C., FLEURENCE, J. & GUÉANT, J. L. 2002b. A new oligomeric parvalbumin allergen of Atlantic cod (Gad mI) encoded by a gene distinct from that of Gad cI. Allergy, 57, 79-83. DAVIS, P. J. & WILLIAMS, S. C. 1998. Protein modification by thermal processing. Allergy, 53, 102-105. DE JONGH, H. H., DE LOS REYES JIMENEZ, M., BAUMERT, J. L., TAYLOR, S. L. & KOPPELMAN, S. J. 2015. Electrophoretic Behavior in Relation to the Structural Integrity of Codfish Parvalbumin upon Heat Treatment. Journal of agricultural and food chemistry, 63, 4683-4689. DE JONGH, H. H. J., ROBLES, C. L., TIMMERMAN, E., NORDLEE, J. A., LEE, P.-W., BAUMERT, J. L., HAMILTON, R. G., TAYLOR, S. L. & KOPPELMAN, S. J. 2013. Digestibility and IgE-Binding of Glycosylated Codfish Parvalbumin. BioMed Research International, 2013, 10. DEAN, T. 2000. Introduction. In: DEAN, T. (ed.) Food intolerance and the food industry. Cambridge: Woodhead publishing Ltd. DEPARTMENT OF AGRICULTURE AND FISHERIES, Q. G. 2016. Cobia [Online]. Available: https://www.daf.qld.gov.au/fisheries/species-identification/inshore-estuarine-spe cies/cobia [Accessed 22/02/2016]. DPI, N. 2014. Status of fisheries resources in NSW 2013-14 [Online]. Available: http://www.dpi.nsw.gov.au/__data/assets/pdf_file/0008/598436/INT16-61462-At tachment-C-Status-of-Fisheries-Resources-in-NSW-2013-14-Full-Report-406-pa ges-updated.pdf [Accessed 28/03/2017]. DUNN, J. & WILD, D. 2013. Chapter 3.6 - Calibration Curve Fitting. In: WILD, D. (ed.) The Immunoassay Handbook (Fourth Edition). Oxford: Elsevier. DYER, W. J., FRENCH, H. V. & SNOW, J. M. 1950. Proteins in Fish Muscle.: I.

203

Extraction of Protein Fractions in Fresh Fish. Journal of the Fisheries Research Board of Canada, 7d, 585-593. EBISAWA, M., ITO, K. & FUJISAWA, T. 2017. Japanese guidelines for food allergy 2017. Allergology International, 66, 248-264. ELSAYED, S. & APOLD, J. 1983. Immunochemical analysis of cod fish allergen M: locations of the immunoglobulin binding sites as demonstrated by the native and synthetic peptides. Allergy, 38, 449-459. FAESTE, C. K., LOVBERG, K. E., LINDVIK, H. & EGAAS, E. 2007. Extractability, stability, and allergenicity of egg white proteins in differently heat-processed foods. JOURNAL-AOAC INTERNATIONAL, 90, 427. FÆSTE, C. K. & PLASSEN, C. 2008. Quantitative sandwich ELISA for the determination of fish in foods. Journal of Immunological Methods, 329, 45-55. FAESTE, C. K., ROENNING, H. T., CHRISTIANS, U. & GRANUM, P. E. 2011. Liquid chromatography and mass spectrometry in food allergen detection. Journal of food protection, 74, 316-345. FAO 2016. The State of World Fisheries and Aquaculture 2016. In: US, F. A. A. D. O. T. (ed.). FIOCCHI, A., SAMPSON, H. A., BAHNA, S. L. & LACK, G. 2013. Food Allergy. In: PAWANKAR, R., HOLGATE, S. T., CANONICA, G. W., LOCKEY, R. F. & BLAISS, M. S. (eds.) WAO White Book on Allergy: Update 2013. Milwaukee, USA: World Allergy Organization. FORD, L. S., TAYLOR, S. L., PACENZA, R., NIEMANN, L. M., LAMBRECHT, D. M. & SICHERER, S. H. 2010. Food allergen advisory labeling and product contamination with egg, milk, and peanut. Journal of Allergy and Clinical Immunology, 126, 384-385. FSANZ 2017. Standard 1.2.3 - Mandatory warning and advisory statement and declarations. FUKUI, Y., YOSHIDA, M., SHOZEN, K.-I., FUNATSU, Y., TAKANO, T., OIKAWA, H., YANO, Y. & SATOMI, M. 2012. Bacterial communities in fish sauce mash using culture-dependent and-independent methods. The Journal of general and applied microbiology, 58, 273-281. GAJEWSKI, K. G., CHEN, Y. T. & HSIEH, Y. H. 2009. Production and characterization of monoclonal antibodies specific to pangasius catfish, basa, and tra. Journal of food science, 74, C241-C247. GAJEWSKI, K. G. & HSIEH, Y.-H. P. 2009. Monoclonal Antibody Specific to a Major Fish Allergen: Parvalbumin. Journal of Food Protection, 72, 818-825. GALLAND, A. V., DORY, D., PONS, L., CHOPIN, C., RABESONA, H., GUÉANT, J. L. & FLEURENCE, J. 1998. Purification of a 41 kDa cod-allergenic protein. Journal of Chromatography B: Biomedical Sciences and Applications, 706, 63-71.

204

GATLIN, C. L., KLEEMANN, G. R., HAYS, L. G., LINK, A. J. & YATES, J. R. 1998. Protein identification at the low femtomole level from silver-stained gels using a new fritless electrospray interface for liquid chromatography–microspray and nanospray mass spectrometry. Analytical biochemistry, 263, 93-101. GENDEL, S. M. 2012. Comparison of international food allergen labeling regulations. Regulatory Toxicology and Pharmacology, 63, 279-285. GIBBS, J. & KENNEBUNK, M. 2001. ELISA Technical Bulletin. Corning Incorporated Life Sciences, Kennebunk, ME. GOODMAN, M. & PECHÉRE, J. F. 1977. The evolution of muscular parvalbumins investigated by the maximun parisomony method. Journal of Molecular Evolution, 9, 131-158. GREENHAWT, M. J., SINGER, A. M. & BAPTIST, A. P. 2009. Food allergy and food allergy attitudes among college students. Journal of Allergy and Clinical Immunology, 124, 323-327. GRIESMEIER, U., BUBLIN, M., RADAUER, C., VÁZQUEZ-CORTÉS, S., MA, Y., FERNÁNDEZ-RIVAS, M. & BREITENEDER, H. 2010a. Physicochemical properties and thermal stability of Lep w 1, the major allergen of whiff. Molecular Nutrition & Food Research, 54, 861-869. GRIESMEIER, U., VÁZQUEZ-CORTÉS, S., BUBLIN, M., RADAUER, C., MA, Y., BRIZA, P., FERNÁNDEZ-RIVAS, M. & BREITENEDER, H. 2010b. Expression levels of parvalbumins determine allergenicity of fish species. Allergy, 65, 191-198. GUPTA, R. S., LAU, C. H., SITA, E. E., SMITH, B. & GREENHAWT, M. J. 2013. Factors associated with reported food allergy tolerance among US children. Annals of Allergy, Asthma & Immunology, 111, 194-198.e4. HAAHTELA, T., BJORKSTÉAN, F., HEISKALA, M. & SUONIEMI, I. 1980. Skin Prick Test Reactivity to Common Allergens in Finnish Adolescents. Allergy, 35, 425-431. HAMADA, Y., NAGASHIMA, Y. & SHIOMI, K. 2001. Identification of collagen as a new fish allergen. Bioscience, biotechnology, and biochemistry, 65, 285-291. HAMADA, Y., NAGASHIMA, Y., SHIOMI, K., SHIMOJO, N., KOHNO, Y., SHIBATA, R., NISHIMA, S., OHSUNA, H. & IKEZAWA, Z. 2003. Reactivity of IgE in fish‐allergic patients to fish muscle collagen. Allergology International, 52, 139-147. HANCOCK, D. C. & O’REILLY, N. J. 2005. Production of Polyclonal Antibodies in Rabbits. In: BURNS, R. (ed.) Immunochemical Protocols 3rd Edition ed. Totowa, NJ: Humana Press. HE, J. 2013. Chapter 5.1 - Practical Guide to ELISA Development In: WILD, D. (ed.) The Immunoassay Handbook (Fourth Edition). Oxford: Elsevier. HEFLE, S. L., FURLONG, T. J., NIEMANN, L., LEMON-MULE, H., SICHERER, S.

205

& TAYLOR, S. L. 2007. Consumer attitudes and risks associated with packaged foods having advisory labeling regarding the presence of peanuts. Journal of Allergy and Clinical Immunology, 120, 171-176. HEIZMANN, C. W. 1984. Parvalbumin, and intracellular calcium-binding protein; distribution, properties and possible roles in mammalian cells. Experientia, 40, 910-921. HEMMI, H., YAMASHITA, M. & NISHIOKA, F. 1997. Changes in parvalbumin content and its isoform pattern in the muscle of red sea bream along with muscle development. Fisheries science, 63, 446-448. HIRST, B. & MIGUEL, S. J. 2015. Validation, standardisation and harmonisation of analytical methods and test kits for detecting allergens in food. In: FLANAGAN, S. (ed.) Handbook of Food Allergen Detection and Control. Woodhead Publishing. HU, Y., CHEN, J. & LI, H. 2010. Comparison of food allergy prevalence among Chinese infants in Chongqing, 2009 versus 1999. Pediatrics International, 52, 820-824. IMMER, U. & LACORN, M. 2015. Enzyme-linked immunosorbent assays (ELISAs) for detecting allergens in food. In: FLANAGAN, S. (ed.) Handbook of Food Allergen Detection and Control. ITO, K., YAMAMOTO, T., OYAMA, Y., TSURUMA, R., SAITO, E., SAITO, Y., OZU, T., HONJOH, T., ADACHI, R. & SAKAI, S. 2016. Food allergen analysis for processed food using a novel extraction method to eliminate harmful reagents for both ELISA and lateral-flow tests. Analytical and bioanalytical chemistry, 408, 5973-5984. JEEBHAY, M. F. & LOPATA, A. L. 2012. Occupational Allergies in Seafood-Processing Workers. Advances in food and nutrition research, 66, 47. JEEBHAY, M. F., ROBINS, T. G., MILLER, M. E., BATEMAN, E., SMUTS, M., BAATJIES, R. & LOPATA, A. L. 2008. Occupational allergy and asthma among salt water fish processing workers. American Journal of Industrial Medicine, 51, 899-910. JORGE, I., CASAS, E. M., VILLAR, M., ORTEGA‐PÉREZ, I., LÓPEZ‐FERRER, D., MARTÍNEZ‐RUIZ, A., CARRERA, M., MARINA, A., MARTÍNEZ, P. & SERRANO, H. 2007. High‐sensitivity analysis of specific peptides in complex samples by selected MS/MS ion monitoring and linear ion trap mass spectrometry: Application to biological studies. Journal of mass spectrometry, 42, 1391-1403. KANAMORI, M., TANAKA, H., HAMADA, Y., NAGASHIMA, Y. & SHIOMI, K. 2011. New extraction method suitable for immunoblotting analysis of fish allergens. European Food Research and Technology, 233, 991-997. KAWASE, S., USHIO, H., OHSHIMA, T., YAMANAKA, H. & FUKUDA, H. 2001.

206

Preparation of monoclonal antibodies against tuna parvalbumin. Fisheries science, 67, 559-561. KHUDA, S. E., JACKSON, L. S., FU, T.-J. & WILLIAMS, K. M. 2015. Effects of processing on the recovery of food allergens from a model dark chocolate matrix. Food Chemistry, 168, 580-587. KIM, J., CHANG, E., HAN, Y., AHN, K. & LEE, S. I. 2011. The incidence and risk factors of immediate type food allergy during the first year of life in Korean infants: a birth cohort study. Pediatric Allergy and Immunology, 22, 715-719. KINDT, T. J., COLDSBY, R. A. & OSBORNE, B. A. 2007. Kuby Immunology, New York, W.H Freeman and Company. KIRSCH, S., FOURDRILIS, S., DOBSON, R., SCIPPO, M.-L., MAGHUIN-ROGISTER, G. & DE PAUW, E. 2009. Quantitative methods for food allergens: a review. Analytical and Bioanalytical Chemistry, 395, 57-67. KOBAYASHI, A., ICHIMURA, A., KOBAYASHI, Y. & SHIOMI, K. 2016a. IgE-binding epitopes of various ﬁsh parvalbumins exist in a stereoscopic conformation maintained by Ca2+-binding. Allergology International, 65, 345-348. KOBAYASHI, A., KOBAYASHI, Y. & SHIOMI, K. 2016b. Fish allergy in patients with parvalbumin-specific immunoglobulin E depends on parvalbumin content rather than molecular differences in the protein among fish species. Bioscience, biotechnology, and biochemistry, 1-4. KOBAYASHI, A., TANAKA, H., HAMADA, Y., ISHIZAKI, S., NAGASHIMA, Y. & SHIOMI, K. 2006. Comparison of allergenicity and allergens between fish white and dark muscles. Allergy, 61, 357-363. KOBAYASHI, Y., HUGE, J., IMAMURA, S. & HAMADA-SATO, N. 2016c. Study of the cross-reactivity of fish allergens based on a questionnaire and blood testing. Allergology International. KOBAYASHI, Y., YANG, T., YU, C.-T., UME, C., KUBOTA, H., SHIMAKURA, K., SHIOMI, K. & HAMADA-SATO, N. 2016d. Quantification of major allergen parvalbumin in 22 species of fish by SDS–PAGE. Food Chemistry, 194, 345-353. KOEBERL, M., CLARKE, D. & LOPATA, A. L. 2014. Next Generation of Food Allergen Quantification Using Mass Spectrometric Systems. Journal of Proteome Research, 13, 3499-3509. KONOSU, S., HAMOIR, G. & PECHERE, J.-F. 1965. Carp myogens of white and red muscles. Properties and amino acid composition of the main low-molecular-weight components of white muscle. Biochemical Journal, 96, 98-112. KONSTANTINOU, G. N. 2017. Enzyme-Linked Immunosorbent Assay (ELISA). In: LIN, J. & ALCOCER, M. (eds.) Food Allergens: Methods and Protocols. New

207

York, NY: Springer New York. KOPPELMAN, S., NORDLEE, J., LEE, P.-W., HAPPE, R., HESSING, M., NORLAND, R., MANNING, T., DESCHENE, R., DE JONG, G. & TAYLOR, S. 2012. Parvalbumin in fish skin-derived gelatin: is there a risk for fish allergic consumers? Food Additives & Contaminants: Part A, 29, 1347-1355. KOPPELMAN, S. J. & HEFLE, S. L. 2006. Detecting Allergens in Food. Cambridge: Woodhead Publishing. KOPPELMAN, S. J., ROMIJN, R. A., DE JONGH, H. H. J., NORDLEE, J. A., PIERSMA, S., HESSING, M. & TAYLOR, S. L. 2010. Purification of Parvalbumin from Carp: A Protocol that Avoids Heat-Treatment. Journal of Food Science, 75, T49-T56. KRALIMARKOVA, T., MUSTAKOV, T., LYUDSKANOVA, E. & POPOV, T. 2008. Characteristics of 160 consecutive cases of alleged food allergy referred to the National Centre of Allergology in Bulgaria. Allergy: European Journal of Allergy and Clinical Immunology, 63, 583. KRETSINGER, R. H. & NOCKOLDS, C. E. 1973. Carp Muscle Calcium-binding Protein: II. STRUCTURE DETERMINATION AND GENERAL DESCRIPTION. Journal of Biological Chemistry, 248, 3313-3326. KRISTJANSSON, I., ARDAL, B., JONSSON, J. S., SIGURDSSON, J. A., FOLDEVI, M. & BJÖRKSTÉN, B. 1999. Adverse reactions to food and food allergy in young children in Iceland and Sweden. Scandinavian journal of primary health care, 17, 30-34. KUBOTA, H., KOBAYASHI, A., KOBAYASHI, Y., SHIOMI, K. & HAMADA-SATO, N. 2016. Reduction in IgE reactivity of Pacific mackerel parvalbumin by heat treatment. Food Chemistry, 206, 78-84. KUEHN, A., HILGER, C., LEHNERS-WEBER, C., CODREANU-MOREL, F., MORISSET, M., METZ-FAVRE, C., PAULI, G., BLAY, F., REVETS, D., MULLER, C. P., VOGEL, L., VIETHS, S. & HENTGES, F. 2013. Identification of enolases and aldolases as important fish allergens in cod, salmon and tuna: component resolved diagnosis using parvalbumin and the new allergens. Clinical & Experimental Allergy, 43, 811-822. KUEHN, A., RADAUER, C., LOPATA, A. L., KLEINE-TEBBE, J. & SWOBODA, I. 2017. Extract-Based and Molecular Diagnostics in Fish Allergy. In: KLEINE-TEBBE, J. & JAKOB, T. (eds.) Molecular Allergy Diagnostics: Innovation for a Better Patient Management. Cham: Springer International Publishing. KUEHN, A., SCHEUERMANN, T., HILGER, C. & HENTGES, F. 2010. Important Variations in Parvalbumin Content in Common Fish Species: A Factor Possibly Contributing to Variable Allergenicity. International Archives of Allergy and Immunology, 153, 359-366.

208

KUEHN, A., SWOBODA, I., ARUMUGAM, K., HILGER, C. & HENTGES, F. 2014. Fish Allergens at a Glance: Variable Allergenicity of Parvalbumins, the Major Fish Allergens. Frontiers in Immunology, 5. L’HOCINE, L. & PITRE, M. 2016. Quantitative and qualitative optimization of allergen extraction from peanut and selected tree nuts. Part 2. Optimization of buffer and ionic strength using a full factorial experimental design. Food chemistry, 194, 820-827. LAO‐ARAYA, M. & TRAKULTIVAKORN, M. 2012. Prevalence of food allergy among preschool children in northern Thailand. Pediatrics International, 54, 238-243. LEE, A. & SUN, X. 2016. Chapter 10. Immunoanalytical Techniques for Peanut Allergen Detection. In: LEE, A., WRIGHT, G. C. & RACHAPUTI, R. C. N. (eds.) Peanuts: Bioactives and Allergens. Pennsylvania, US: DEStech Publications, Inc. LEE, A. J., THALAYASINGAM, M. & LEE, B. W. 2013. Food allergy in Asia: how does it compare? Asia Pac Allergy, 3, 3-14. LEE, N. A. & KENNEDY, I. R. 2007. Immunoassays. Food Toxicants Analysis. Elsevier. LEE, P.-W. 2012. Characterization of IgG and IgE binding to parvalbumin derived from commercially important fish species. Doctor of Philosophy, University of Nebraska. LEE, P.-W., NORDLEE, J. A., KOPPELMAN, S. J., BAUMERT, J. L. & TAYLOR, S. L. 2011. Evaluation and Comparison of the Species-Specificity of 3 Antiparvalbumin IgG Antibodies. Journal of Agricultural and Food Chemistry, 59, 12309-12316. LEE, P.-W., NORDLEE, J. A., KOPPELMAN, S. J., BAUMERT, J. L. & TAYLOR, S. L. 2012. Measuring parvalbumin levels in fish muscle tissue: Relevance of muscle locations and storage conditions. Food Chemistry, 135, 502-507. LEE, S. H., JUNG, J. Y. & JEON, C. O. 2015. Bacterial community dynamics and metabolite changes in myeolchi-aekjeot, a Korean traditional fermented fish sauce, during fermentation. International Journal of Food Microbiology, 203, 15-22. LEMKE, P. J. & TAYLOR, S. L. 1994. Allergic reactions and food intolerances. In: KOTSONIS, F. N., MACKEY, M. & HJELLE, J. (eds.) Nutritional Toxicology. New York: Raven Press. LEUNG, T. F., YUNG, E., WONG, Y. S., LAM, C. W. K. & WONG, G. W. K. 2009. Parent-reported adverse food reactions in Hong Kong Chinese pre-schoolers: epidemiology, clinical spectrum and risk factors. Pediatric Allergy and Immunology, 20, 339-346. LIANG, J., TAN, C. C., TAYLOR, S. L., BAUMERT, J. L., LOPATA, A. L. & LEE, N.

209

A. 2017. Quantitative analysis of species specificity of two anti-parvalbumin antibodies for detecting southern hemisphere fish species demonstrating strong phylogenetic association. Food Chemistry, 237, 588-596. LIDDELL, E. 2013. Chapter 3.1 - Antibodies. In: WILD, D. (ed.) The Immunoassay Handbook (Fourth Edition). Oxford: Elsevier. LIM, D. L. C., NEO, K. H., YI, F. C., CHUA, K. Y., GOH, D. L. M., SHEK, L. P. C., GIAM, Y. C., VAN BEVER, H. P. & LEE, B. W. 2008. Parvalbumin–the major tropical fish allergen. Pediatric Allergy and Immunology, 19, 399-407. LIPTON, C., DAUTLICK, J., GROTHAUS, G., HUNST, P., MAGIN, K., MIHALIAK, C., RUBIO, F. & STAVE, J. 2000. Guidelines for the validation and use of immunoassays for determination of introduced proteins in biotechnology enhanced crops and derived food ingredients. Food and Agricultural Immunology, 12, 153-164. LIU, A. H., JARAMILLO, R., SICHERER, S. H., WOOD, R. A., BOCK, S. A., BURKS, A. W., MASSING, M., COHN, R. D. & ZELDIN, D. C. 2010a. National prevalence and risk factors for food allergy and relationship to asthma: results from the National Health and Nutrition Examination Survey 2005-2006. Journal of Allergy and Clinical Immunology, 126, 798-806. e14. LIU, G.-M., WANG, N., CAI, Q.-F., LI, T., SUN, L.-C., SU, W.-J. & CAO, M.-J. 2010b. Purification and characterization of parvalbumins from silver carp (Hypophthalmichthy molitrix). Journal of the Science of Food and Agriculture, 90, 1034-1040. LOPATA, A. L., JEEBHAY, M. F., REESE, G., FERNANDES, J., SWOBODA, I., ROBINS, T. G. & LEHRER, S. B. 2005. Detection of fish antigens aerosolized during fish processing using newly developed immunoassays. International archives of allergy and immunology, 138, 21-28. MA, Y., GRIESMEIER, U., SUSANI, M., RADAUER, C., BRIZA, P., ERLER, A., BUBLIN, M., ALESSANDRI, S., HIMLY, M., VÀZQUEZ-CORTÉS, S., RINCON DE ARELLANO, I. R., VASSILOPOULOU, E., SAXONI-PAPAGEORGIOU, P., KNULST, A. C., FERNÁNDEZ-RIVAS, M., HOFFMANN-SOMMERGRUBER, K. & BREITENEDER, H. 2008. Comparison of natural and recombinant forms of the major fish allergen parvalbumin from cod and carp. Molecular Nutrition & Food Research, 52, S196-S207. MCGOWAN, E. C. & KEET, C. A. 2013. Prevalence of self-reported food allergy in the National Health and Nutrition Examination Survey (NHANES) 2007–2010. The Journal of allergy and clinical immunology, 132. MERCER, M. J., VAN DER LINDE, G. P. & JOUBERT, G. 2002. Rhinitis (allergic and nonallergic) in an atopic pediatric referral population in the grasslands of inland South Africa. Annals of Allergy, Asthma & Immunology, 89, 503-512.

210

MFDS, R. O. K. 2016. Food Labeling System [Online]. Available: http://www.mfds.go.kr/eng/index.do?nMenuCode=118 [Accessed 30/08/2017]. MULLINS, R. J., DEAR, K. B. & TANG, M. L. 2015. Time trends in Australian hospital anaphylaxis admissions in 1998-1999 to 2011-2012. Journal of Allergy and Clinical Immunology, 136, 367-375. MURALIDHARAN, S., ZHAO, Y., TAYLOR, S. L. & LEE, N. A. 2017. Detection of Food Allergen Residues by Immunoassays and Mass Spectrometry. In: LOPATA, A. L. (ed.) Food Allergy: Molecular and Clinical Practice. Boca Raton: CRC Press. NATIONAL ACADEMIC OF SCIENCES, E., AND MEDICINE 2016. Finding a Path to Safety in Food Allergy: Assessment of the Global Burden, Causes, Prevention, Management, and Public Policy, National Academies Press. NELSON, J. 2006. Fishes of the World, Hoboken, NJ, John Wiley & Sons. NG, I. E., TURNER, P. J., KEMP, A. S. & CAMPBELL, D. E. 2011. Parental perceptions and dietary adherence in children with seafood allergy. Pediatric Allergy and Immunology, 22, 720-728. NWARU, B., HICKSTEIN, L., PANESAR, S., ROBERTS, G., MURARO, A. & SHEIKH, A. 2014. Prevalence of common food allergies in Europe: a systematic review and meta‐analysis. Allergy, 69, 992-1007. OBENG, B. B., AMOAH, A. S., LARBI, I. A., YAZDANBAKHSH, M., VAN REE, R., BOAKYE, D. A. & HARTGERS, F. C. 2011. Food Allergy in Ghanaian Schoolchildren: Data on Sensitization and Reported Food Allergy. International Archives of Allergy and Immunology, 155, 63-73. ORHAN, F., KARAKAS, T., CAKIR, M., AKSOY, A., BAKI, A. & GEDIK, Y. 2009. Prevalence of immunoglobulin E-mediated food allergy in 6–9-year-old urban schoolchildren in the eastern Black Sea region of Turkey. Clinical & Experimental Allergy, 39, 1027-1035. OSBORNE, N. J., KOPLIN, J. J., MARTIN, P. E., GURRIN, L. C., LOWE, A. J., MATHESON, M. C., PONSONBY, A.-L., WAKE, M., TANG, M. L. & DHARMAGE, S. C. 2011. Prevalence of challenge-proven IgE-mediated food allergy using population-based sampling and predetermined challenge criteria in infants. Journal of Allergy and Clinical Immunology, 127, 668-676. e2. OSTERBALLE, M., HANSEN, T. K., MORTZ, C. G., HØST, A. & BINDSLEV-JENSEN, C. 2005. The prevalence of food hypersensitivity in an unselected population of children and adults. Pediatric Allergy and Immunology, 16, 567-573. OSTERBALLE, M., MORTZ, C. G., HANSEN, T. K., ANDERSEN, K. E. & BINDSLEV-JENSEN, C. 2009. The Prevalence of food hypersensitivity in young adults. Pediatric Allergy and Immunology, 20, 686-692. PATEL, D. A., HOLDFORD, D. A., EDWARDS, E. & CARROLL, N. V. 2011.

211

Estimating the economic burden of food-induced allergic reactions and anaphylaxis in the United States. Journal of Allergy and Clinical Immunology, 128, 110-115.e5. PAULY, D. & FROESE, R. 2016. FishBase. 10/2015 ed. PECHÉRE, J.-F., CAPONY, J.-P. & RYDEN, L. 1971. The Primary Structure of the Major Parvalbumin from Hake Muscle. European Journal of Biochemistry, 23, 421-428. PECHÈRE, J. F. 1968. Muscular parvalbumins as homologous proteins. Comparative Biochemistry and Physiology, 24, 289-295. PECHÈRE, J. F. & CAPONY, J. P. 1969. A comparison at the peptide level of muscular parvalbumins from several lower vertebrates. Comparative Biochemistry and Physiology, 28, 1089-1192. PÉNARD-MORAND, C., RAHERISON, C., KOPFERSCHMITT, C., CAILLAUD, D., LAVAUD, F., CHARPIN, D., BOUSQUET, J. & ANNESI-MAESANO, I. 2005. Prevalence of food allergy and its relationship to asthma and allergic rhinitis in schoolchildren. Allergy, 60, 1165-1171. PEÑAS, E., UBERTI, F., BAVIERA, G., DI LORENZO, C. & RESTANI, P. 2014. Clinical monosensitivity to salmon and rainbow trout: a case report. Pediatric Allergy and Immunology, 25, 98-100. PEREZ-GORDO, M., LIN, J., BARDINA, L., PASTOR-VARGAS, C., CASES, B., VIVANCO, F., CUESTA-HERRANZ, J. & SAMPSON, H. A. 2012. Epitope Mapping of Atlantic Salmon Major Allergen by Peptide Microarray Immunoassay. International Archives of Allergy and Immunology, 157, 31-40. PEREZ-GORDO, M., PASTOR-VARGAS, C., LIN, J., BARDINA, L., CASES, B., IBÁÑEZ, M. D., VIVANCO, F., CUESTA-HERRANZ, J. & SAMPSON, H. A. 2013. Epitope mapping of the major allergen from Atlantic cod in Spanish population reveals different IgE-binding patterns. Molecular Nutrition & Food Research, 57, 1283-1290. PERMYAKOV, E. A. 2006. Parvalbumin, New York, Nova Science Publishers, Inc. PERMYAKOV, E. A. & KRETSINGER, R. H. 2010. Parvalbumin. Calcium Binding Proteins. John Wiley & Sons, Inc. PERMYAKOV, E. A., MEDVEDKIN, V. N., KALINICHENKO, L. P. & BURSTEIN, E. A. 1983. Comparative study of physicochemical properties of two pike parvalbumins by means of their intrinsic tyrosyl and phenylalanyl fluorescence. Archives of Biochemistry and Biophysics, 227, 9-20. POMS, R. E., KLEIN, C. L. & ANKLAM, E. 2004. Methods for allergen analysis in food: a review. Food Additives & Contaminants, 21, 1-31. RIMPEEKOOL, W., SEUBSMAN, S.-A., BANWELL, C., KIRK, M., YIENGPRUGSAWAN, V. & SLEIGH, A. 2015. Food and nutrition labelling in Thailand: a long march from subsistence producers to international traders. Food

212

Policy, 56, 59-66. ROSMILAH, M., SHAHNAZ, M., MASITA, A., NOORMALIN, A. & JAMALUDIN, M. 2005. Identification of major allergens of two species of local snappers: Lutjanus argentimaculatus (merah/red snapper) and Lutjanus johnii (jenahak/golden snapper). Trop Biomed, 22, 171-77. SAKELLARIOU, A., ZANNIKOS, K., TZANNIS, K., MICHOPOULOU, C., EMMANOUIL, E., VASSILOPOULOU, E., SINANIOTIS, A., SAXONI-PAPAGEORGIOU, P. & PAPADOPOULOS, N. 2008. The prevalence of perceived food hypersensitivity in adults in the city of Athens. Allergy: European Journal of Allergy and Clinical Immunology, 63, 583. SAMARTÍN, S., MARCOS, A. & R.K., C. 2001. Food hypersensitivity. Nutrition Research, 21, 473-497. SAMPSON, H. A. & LEUNG, D. Y. M. 2011. Adverse reactions to foods. Allergy principles and practice. Amsterdam, United States of America: Elsevier Academic Press SAPTARSHI, S. R., SHARP, M. F., KAMATH, S. D. & LOPATA, A. L. 2014. Antibody reactivity to the major fish allergen parvalbumin is determined by isoforms and impact of thermal processing. Food Chemistry, 148, 321-328. SATHE, S. K., TEUBER, S. S. & ROUX, K. H. 2005. Effects of food processing on the stability of food allergens. Biotechnology advances, 23, 423-429. SCHÄFER, T., BÖHLER, E., RUHDORFER, S., WEIGL, L., WESSNER, D., HEINRICH, J., FILIPIAK, B., WICHMANN, H. E. & RING, J. 2001. Epidemiology of food allergy/food intolerance in adults: associations with other manifestations of atopy. Allergy, 56, 1172-1179. SCHULKES, K. J., KLEMANS, R. J., KNIGGE, L., DE BRUIN-WELLER, M., BRUIJNZEEL-KOOMEN, C. A., MARKNELL DEWITT, Å., LIDHOLM, J. & KNULST, A. C. 2014. Specific IgE to fish extracts does not predict allergy to specific species within an adult fish allergic population. Clinical and translational allergy, 4, 27. SELLGE, G. & BISCHOFF, S. C. 2008. The Immunological Basis of IgE-Mediated Reactions. In: METCALFE, D. D., SAMPSON, H. A. & SIMON, R. A. (eds.) Food Allergy: Adverse Reactions to Foods and Food Addtives. Fourth Edition ed. USA: Blackwell Pulishing. SHARP, M. F. & LOPATA, A. L. 2014. Fish allergy: In review. Clinical reviews in allergy & immunology, 1-14. SHARP, M. F., STEPHEN, J. N., KRAFT, L., WEISS, T., KAMATH, S. D. & LOPATA, A. L. 2015. Immunological cross-reactivity between four distant parvalbumins—Impact on allergen detection and diagnostics. Molecular immunology, 63, 437-448. SHEK, L. P.-C., CABRERA-MORALES, E. A., SOH, S. E., GEREZ, I., NG, P. Z., YI,

213

F. C., MA, S. & LEE, B. W. 2010. A population-based questionnaire survey on the prevalence of peanut, tree nut, and shellfish allergy in 2 Asian populations. Journal of Allergy and Clinical Immunology, 126, 324-331.e7. SHIBAHARA, Y., UESAKA, Y., WANG, J., YAMADA, S. & SHIOMI, K. 2013. A sensitive enzyme-linked immunosorbent assay for the determination of fish protein in processed foods. Food Chemistry, 136, 675-681. SHIOMI, K., HAMADA, Y., SEKIGUCHI, K., SHIMAKURA, K. & NAGASHIMA, Y. 1999. Two Classes of Allergens, Parvalbumins and Higher Molecular Weight Substances, in Japanese Eel and Bigeye Tuna. Fisheries science, 65, 943-948. SICHERER, S. H., MUÑOZ-FURLONG, A. & SAMPSON, H. A. 2004. Prevalence of seafood allergy in the United States determined by a random telephone survey. Journal of Allergy and Clinical Immunology, 114, 159-165. SICHERER, S. H. & SAMPSON, H. A. 2014. Food allergy: Epidemiology, pathogenesis, diagnosis, and treatment. Journal of Allergy and Clinical Immunology, 133, 291-307.e5. SIMPSON, R. J. 2010. Stabilization of Proteins for Storage. Cold Spring Harbor Protocols, 2010, pdb.top79. SOLLER, L., BEN-SHOSHAN, M., HARRINGTON, D. W., FRAGAPANE, J., JOSEPH, L., PIERRE, Y. S., GODEFROY, S. B., LA VIEILLE, S., ELLIOTT, S. J. & CLARKE, A. E. 2012. Overall prevalence of self-reported food allergy in Canada. Journal of Allergy and Clinical Immunology, 130, 986-988. SOMKUTI, J., BUBLIN, M., BREITENEDER, H. & SMELLER, L. 2012. Pressure– Temperature Stability, Ca2+ Binding, and Pressure–Temperature Phase Diagram of Cod Parvalbumin: Gad m 1. Biochemistry, 51, 5903-5911. SØRENSEN, M., KUEHN, A., MILLS, E. N. C., COSTELLO, C. A., OLLERT, M., SMÅBREKKE, L., PRIMICERIO, R., WICKMAN, M. & KLINGENBERG, C. 2017. Cross-reactivity in fish allergy: A double-blind, placebo-controlled food-challenge trial. Journal of Allergy and Clinical Immunology. STEINHOFF, M., FISCHER, M. & PASCHKE-KRATZIN, A. 2011. Comparison of extraction conditions for milk and hen's egg allergens. Food Additives & Contaminants: Part A, 28, 373-383. SUDHAKAR, K., PHILLIPS, C. M., OWEN, C. S. & VANDERKOOI, J. M. 1995. Dynamics of parvalbumin studied by fluorescence emission and triplet absorption spectroscopy of tryptophan. Biochemistry, 34, 1355-1363. SUN, M., LIANG, C., GAO, H., LIN, C. & DENG, M. 2008. Detection of parvalbumin, a common fish allergen gene in food, by real-time polymerase chain reaction. Journal of AOAC International, 92, 234-240. SWEENEY, M. J. & KLOTZ, S. D. 1987. Immunology of food antigens. Handbook of food allergies. New York and Basel: Marcel Dekker, Inc. TAHERGORABI, R. & JACZYNSKI, J. 2011. Seafood proteins. In: PHILLIPS, G. O.

214

& WILLIAMS, P. A. (eds.) Handbook of Food Proteins. Woodhead Publishing. TAYLOR, S. L. 2006. The nature of food allergy. In: KOPPELMAN, S. J. & HEFLE, S. L. (eds.) Detecting Allergens in Food. Cambridge: Woodhead Publishing Limited. TAYLOR, S. L. & BAUMERT, J. L. 2012. Food allergies and intolerances. In: MACDONALD, I. A., ERDMAN, J. W. & ZEISEL, S. H. (eds.) Present Knowledge in Nutrition. Hoboken: Wiley. TAYLOR, S. L., BAUMERT, J. L., KRUIZINGA, A. G., REMINGTON, B. C., CREVEL, R. W. R., BROOKE-TAYLOR, S., ALLEN, K. J. & HOUBEN, G. 2014. Establishment of Reference Doses for residues of allergenic foods: Report of the VITAL Expert Panel. Food and Chemical Toxicology, 63, 9-17. TAYLOR, S. L. & HEFLE, S. L. 2001. Food Allergies and other food sensitivities. Food Technology, 55, 68-83. TAYLOR, S. L. & HEFLE, S. L. 2006. Food allergen labeling in the USA and Europe. Current opinion in allergy and clinical immunology, 6, 186-190. TAYLOR, S. L., KABOUREK, J. L. & HEFLE, S. L. 2004. Fish allergy: Fish and Products Thereof. Journal of Food Science, 69, R175-R180. THOMPSON, M., ELLISON, S. L. & WOOD, R. 2002. Harmonized guidelines for single-laboratory validation of methods of analysis (IUPAC Technical Report). Pure and Applied Chemistry, 74, 835-855. TINNER, R., OERTLE, M., HEIZMANN, C. & BOSSHARD, H. 1990. Ca2+-binding site of carp parvalbumin recognized by monoclonal antibody. Cell Calcium, 11, 19-23. TIWARI, G. & TIWARI, R. 2010. Bioanalytical method validation: An updated review. Pharmaceutical Methods, 1, 25-38. TOMURA, S., ISHIZAKI, S., NAGASHIMA, Y. & SHIOMI, K. 2008. Reduction in the IgE reactivity of Pacific mackerel parvalbumin by mutations at Ca2+-binding sites. Fisheries Science, 74, 411-417. TSUKAMOTO, K. 1984. The Role of the Red and White Muscles during Swimming of the Yellowtail. NIPPON SUISAN GAKKAISHI, 50, 2025-2030. TURNER, P., NG, I., KEMP, A. & CAMPBELL, D. 2011. Seafood allergy in children. Clinical and Translational Allergy, 1, P36. UNDERSTANDING EVOLUTION, U. O. C. M. O. P. 2017. Phylogeny of the ray-finned fishes [Online]. Available: http://evolution.berkeley.edu/evolibrary/article/fullfishtree_01 [Accessed 20/06/2017]. UNTERSMAYR, E., VESTERGAARD, H., MALLING, H.-J., JENSEN, L. B., PLATZER, M. H., BOLTZ-NITULESCU, G., SCHEINER, O., SKOV, P. S., JENSEN-JAROLIM, E. & POULSEN, L. K. 2007. Incomplete digestion of codfish represents a risk factor for anaphylaxis in patients with allergy. Journal

215

of Allergy and Clinical Immunology, 119, 711-717. VAN DER VENTEL, M. L., NIEUWENHUIZEN, N. E., KIRSTEIN, F., HIKUAM, C., JEEBHAY, M. F., SWOBODA, I., BROMBACHER, F. & LOPATA, A. L. 2011. Differential responses to natural and recombinant allergens in a murine model of fish allergy. Molecular Immunology, 48, 637-646. VAN DO, T., ELSAYED, S., FLORVAAG, E., HORDVIK, I. & ENDRESEN, C. 2005. Allergy to fish parvalbumins: studies on the cross-reactivity of allergens from 9 commonly consumed fish. Journal of Allergy and Clinical Immunology, 116, 1314-1320. VAN HENGEL, A. 2007. Food allergen detection methods and the challenge to protect food-allergic consumers. Analytical and Bioanalytical Chemistry, 389, 111-118. VENTER, C., PEREIRA, B., GRUNDY, J., CLAYTON, C. B., ARSHAD, S. H. & DEAN, T. 2006. Prevalence of sensitization reported and objectively assessed food hypersensitivity amongst six-year-old children: A population-based study. Pediatric Allergy and Immunology, 17, 356-363. VENTURA, M. T., POLIMENO, L., AMORUSO, A. C., GATTI, F., ANNOSCIA, E., MARINARO, M., DI LEO, E., MATINO, M. G., BUQUICCHIO, R., BONINI, S., TURSI, A. & FRANCAVILLA, A. 2006. Intestinal permeability in patients with adverse reactions to food. Digestive and Liver Disease, 38, 732-736. WANG, B., LI, Z., ZHENG, L., LIU, Y. & LIN, H. 2011. Identification and characterization of a new IgE-binding protein in mackerel (Scomber japonicus) by MALDI-TOF-MS. Journal of Ocean University of China, 10, 93-98. WATANABE, Y., ABURATANI, K., MIZUMURA, T., SAKAI, M., MURAOKA, S., MAMEGOSI, S. & HONJOH, T. 2005. Novel ELISA for the detection of raw and processed egg using extraction buffer containing a surfactant and a reducing agent. Journal of Immunological Methods, 300, 115-123. WEBER, P., STEINHART, H. & PASCHKE, A. 2009. Competitive Indirect ELISA for the Determination of Parvalbumins from Various Fish Species in Food Grade Fish Gelatins and Isinglass with PARV-19 Anti-parvalbumin Antibodies. Journal of Agricultural and Food Chemistry, 57, 11328-11334. WILD, D. 2013. Chapter 1.2 - Immunoassay for Beginners. The Immunoassay Handbook (Fourth Edition). Oxford: Elsevier. WILD, L. G. M. D. & LEHRER, S. B. P. 2005. Fish and shellfish allergy. Current Allergy and Asthma Reports, 5, 74-9. WILEY, E. & JOHNSON, G. D. 2010. A teleost classification based on monophyletic groups. In: NELSON, J., SCHULTZEH, P. & WILSON, M. (eds.) Origin and phylogenetic interrelationships of teleosts. München, Germany: Verlag Dr. Friedrich Pfeil. WOOD, R. A. 2003. The Natural History of Food Allergy. Pediatrics, 111, 1631-1637. WOOD, R. A., CAMARGO, C. A., LIEBERMAN, P., SAMPSON, H. A., SCHWARTZ,

216

L. B., ZITT, M., COLLINS, C., TRINGALE, M., WILKINSON, M., BOYLE, J. & SIMONS, E. R. 2014. Anaphylaxis in America: The prevalence and charateristics of anaphylaxis in the United States. Journal of Allergy and Clinical Immunology, 133, 461-467. WRBA, A., SCHWEIGER, A., SCHULTES, V., JAENICKE, R. & ZAVODSZKY, P. 1990. Extremely thermostable D-glyceraldehyde-3-phosphate dehydrogenase from the eubacterium Thermotoga maritima. Biochemistry, 29, 7584-7592. WROLSTAD, R. E., ACREE, T. E., DECKER, E. A., PENNER, M. H., REID, D. S., SCHWARTZ, S. J., SHOEMAKER, C. F., SMITH, D. M. & SPORNS, P. 2005. Handbook of Food Analytical Chemistry, Volume 1: Water, Proteins, Enzymes, Lipids, and Carbohydrates, John Wiley & Sons. WU, T. C., TSAI, T. C., HUANG, C. F., CHANG, F. Y., LIN, C. C., HUANG, I. F., CHU, C. H., LAU, B. H., WU, L. & PENG, H. J. 2012. Prevalence of food allergy in Taiwan: a questionnaire‐based survey. Internal medicine journal, 42, 1310-1315. YAMADA, S., NOLTE, H. & ZYCHLINSKY, E. 1999. Identification and characterization of allergens in two species of tuna fish. Annals of Allergy, Asthma & Immunology, 82, 395-400. YEARSLEY, G. K., LAST, P. R. & WARD, R. D. 1999. Australian seafood handbook: an identification guide to domestic species, CSIRO Marine Research, Australia. ZINN, C., LOPATA, A., VISSER, M. & POTTER, P. 1997. The spectrum of allergy to South African bony fish (Teleosti): Evaluation by double-blind, placebo controlled challenge. South African Medical Journal, 87, 146-152. ZURZOLO, G. A., MATHAI, M. L., KOPLIN, J. J. & ALLEN, K. J. 2013. Precautionary allergen labelling following new labelling practice in Australia. Journal of Paediatrics and Child Health, 49, E306-E310.

217

Appendix

Appendix Figure 1 Illustration of tissue position obtained from fish samples.

8×106

y 6×10

t 2 i R = 0.999 s i t n e t 4×106 n I

d n a

B 2×106

0 2.0 2.5 3.0 3.5 4.0 Purified Cod PAV (µg/lane) Appendix Figure 2 Calibration curve of purified cod PAV for PAV relative quantification on SDS-PAGE.

218

Appendix Figure 3 Calibration curve of quantitative ELISA using the anti-cod PoAb for Chapter 3. The calibration curve was fitted to the sigmoidal dose-response (variable slope) model using GraphPad Prism® v.7.0.

219

Appendix Figure 5 Time course study of heat treatment of barramundi protein extract; protein profile of soluble proteins after heat treatment for 2 min (lane 1), 5 min (lane 2), 10 min (lane 3), 20 min (lane 4), 30 min (lane 5); protein profile of precipitates after heat treatment for 2 min (lane 6), 5 min (lane 7), 10 min (lane 8), 20 min (lane 9), 30 min (lane 10); land 11 – barramundi crude protein extract.

5 ml aliquot of the fish crude protein extract in a centrifuge tube was subjected to boiling water and heat-treated at 100 °C for different time intervals (2 min, 5 min, 10 min, 20 min, 30 min). Most of the crude fish proteins were found to become insoluble right after 2 min of heating and the 36 kDa protein and PAV were the predominant soluble proteins in the heat-treated extracts. The protein patterns on the SDS-PAGE remained unchanged with the increase of heating time (Appendix Figure 5).

Appendix Figure 6 Fluorescence spectra of the selected raw and heated fish protein extracts for different heating time intervals (20 min, 30 min, 45 min and 60 min).

220

Appendix Figure 7 Protein profiles of the heated fish protein extracts of the selected fish species: pink ling after heat treatment for 20 min (lane 1), 30 min (lane 2), 45 min (lane 3), 60 min (lane 4); barramundi after heat treatment for 20 min (lane 5), 30 min (lane 6), 45 min (lane 7), 60 min (lane 8); school whiting after heat treatment for 20 min (lane 9), 30 min (lane 10), 45 min (lane 11), 60 min (lane 12).

Then the heating time was extended to 45 min and 60 min, and fluorescence analysis was conducted to investigate the tertiary structural change of the heated fish protein extracts including PAV at different heating time intervals. The fluorescence spectra (Appendix Figure 6) showed the fluorescence signals of all the selected fish species did not demonstrate further decrease after heating for 30 min, suggesting the heat-treated fish protein extracts reached a relative stable unfolding stage after 30 min of heating time. No significant difference was found between the protein profiles of the heat-treated fish protein extracts for different heating time intervals (Appendix Figure 7). Based on these preliminary results, 45 min of heating time was chosen to induce unfolding of fish proteins including PAV to a stable stage.

221

Appendix Figure 8 Calibration curve of quantitative ELISA using the anti-cod PoAb for Chapter 4. The calibration curve was fitted to the sigmoidal dose-response (variable slope) model using GraphPad Prism® v.7.0.

222

Appendix Figure 10 The cross-reactivity of the heat-treated mixed immunogen with fish species including Johan dory, bight redfish, orange roughy, barramundi, coral trout, crimson snapper, eastern school whiting, jewfish, mahi mahi, Murray cod, pink snapper, Australian pilchard, Atlantic salmon, rainbow trout, sea mullet, sea garfish, pink ling, blue cod and blue-eye trevalla determined by the (HM)S627#4 – (HM)R#2 assay.

223

224

225

226

227