Permanent Link: Research Collection · 12 SUMMARY cessfulbythemultiwavelengthanomalousdispersion(MAD)method.Thestructurewas re ned to an R and Rfree value of 18.9% and 21.4%, respectively

Research Collection

Doctoral Thesis

Structural characterisation of cross-reactive allergensCrystal structures of M. sympodialis cyclophilin and thioredoxinallow identification of putative cross-reactive B-cell epidopes,crystal structure of A. fumigatus cyclophilin reveals, 3D domainswapping of a central element

Author(s): Limacher, Andreas

Publication Date: 2005

Permanent Link: https://doi.org/10.3929/ethz-a-005063556

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

https://doi.org/10.3929/ethz-a-005063556

http://rightsstatements.org/page/InC-NC/1.0/

https://www.research-collection.ethz.ch

https://www.research-collection.ethz.ch/terms-of-use

Dissertation ETH No. 16075

Structural Characterisation of Cross-Reactive AllergensCrystal Structures of M. sympodialis Cyclophilin and Thioredoxinallow Identi�cation of Putative Cross-Reactive B-Cell Epitopes

Crystal Structure of A. fumigatus Cyclophilin reveals3D Domain Swapping of a Central Element

A dissertation submitted to theSwiss Federal Institute of Technology Zurich

for the degree ofDoctor of Sciences

presented byAndreas Limacher

Biochemist, University of Berneborn April 13th, 1973citizen of Inwil, LU

accepted on the recommendation ofProf. Dr. L. Scapozza, examinerProf. Dr. R. Crameri, co-examinerProf. Dr. G. Folkers, co-examiner

2005

2

Contents

Summary 11

Zusammenfassung 14

1 Introduction 171.1 Allergy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1.1.1 IgE-mediated hypersensitivity reaction . . . . . . . . . . . . . . . 171.1.2 Nature of allergens . . . . . . . . . . . . . . . . . . . . . . . . . . 181.1.3 Cross-reactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

1.2 Characterisation of antibody-binding epitopes . . . . . . . . . . . . . . . 201.2.1 Antigen-antibody interactions . . . . . . . . . . . . . . . . . . . . 201.2.2 Conformational changes upon antibody binding . . . . . . . . . . 201.2.3 Mutational e�ects on antigen-antibody binding . . . . . . . . . . 211.2.4 Crystal structure of an allergen-antibody complex . . . . . . . . . 22

1.3 Cyclophilin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241.3.1 Structure of cyclophilin . . . . . . . . . . . . . . . . . . . . . . . . 241.3.2 Cis-trans isomerisation activity . . . . . . . . . . . . . . . . . . . 241.3.3 Suppression of T-cell activation . . . . . . . . . . . . . . . . . . . 261.3.4 Binding of HIV-1 capsid protein . . . . . . . . . . . . . . . . . . . 26

1.4 Thioredoxin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271.4.1 Redox biochemistry of thioredoxin . . . . . . . . . . . . . . . . . . 271.4.2 Structure of thioredoxin . . . . . . . . . . . . . . . . . . . . . . . 281.4.3 Biological activities . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1.5 3D domain swapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301.5.1 De�nition and terminology . . . . . . . . . . . . . . . . . . . . . . 301.5.2 Types of 3D domain swapping . . . . . . . . . . . . . . . . . . . . 311.5.3 Recent examples of 3D domain swapping . . . . . . . . . . . . . . 321.5.4 Internal domain swapping . . . . . . . . . . . . . . . . . . . . . . 34

3

4 CONTENTS

1.5.5 Biological relevance . . . . . . . . . . . . . . . . . . . . . . . . . . 341.6 Aims of the presented work . . . . . . . . . . . . . . . . . . . . . . . . . 35Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

2 Analysis of the Cross-Reactivity and of the 1.5 Å Crystal Structureof the Malassezia sympodialis Mala s 6 Allergen, a Member of theCyclophilin Pan-Allergen Family 452.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462.3 Experimental procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

2.3.1 Cloning and production of cyclophilins . . . . . . . . . . . . . . . 482.3.2 Subjects, routine assessments and skin tests . . . . . . . . . . . . 492.3.3 Immunoassays for IgE antibodies binding to cyclophilins . . . . . 492.3.4 IgE immunoblots . . . . . . . . . . . . . . . . . . . . . . . . . . . 502.3.5 Crystallisation and data collection of Mala s 6 . . . . . . . . . . . 502.3.6 Structure determination and re�nement . . . . . . . . . . . . . . . 502.3.7 Calculation of the solvent-accessible area . . . . . . . . . . . . . . 51

2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522.4.1 Molecular cloning of Asp f 27 and production of recombinant cy-

clophilins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522.4.2 Demonstration of IgE antibody responses to recombinant cyclophilins 532.4.3 Overall structure of Mala s 6 . . . . . . . . . . . . . . . . . . . . . 562.4.4 Active site of Mala s 6 . . . . . . . . . . . . . . . . . . . . . . . . 572.4.5 Superposition of Mala s 6 on human CyP B reveals putative IgE-

binding residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

3 Structure Solution of Cross-Reactive Thioredoxins: Success and At-tempts 693.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703.3 Experimental procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

3.3.1 Malassezia sympodialis thioredoxin . . . . . . . . . . . . . . . . . 72Cloning, protein expression and puri�cation . . . . . . . . . . . . 72Crystallisation and data collection . . . . . . . . . . . . . . . . . . 73

CONTENTS 5

Structure determination and re�nement . . . . . . . . . . . . . . . 73Calculation of the solvent-accessible area . . . . . . . . . . . . . . 75

3.3.2 Malassezia sympodialis thioredoxin without His tag . . . . . . . . 75Cloning, protein expression and puri�cation . . . . . . . . . . . . 75Crystallisation and data collection . . . . . . . . . . . . . . . . . . 76

3.3.3 Triticum aestivum thioredoxin . . . . . . . . . . . . . . . . . . . . 78Cloning, protein expression and puri�cation . . . . . . . . . . . . 78Crystallisation and data collection . . . . . . . . . . . . . . . . . . 79

3.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 813.4.1 Structure of Malassezia sympodialis thioredoxin . . . . . . . . . . 813.4.2 Superposition of Malassezia sympodialis Trx on human Trx reveals

putative IgE-binding residues . . . . . . . . . . . . . . . . . . . . 843.4.3 Crystallisation of Malassezia sympodialis thioredoxin without His

tag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 883.4.4 Crystallisation of Triticum aestivum thioredoxin . . . . . . . . . . 89

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

4 The Crystal Structure of Aspergillus fumigatus Cyclophilin reveals 3DDomain Swapping of a Central Element 934.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 944.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 944.3 Experimental procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

4.3.1 Cloning, protein expression and puri�cation . . . . . . . . . . . . 954.3.2 Crystallisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 974.3.3 Data collection, phasing and re�nement . . . . . . . . . . . . . . . 974.3.4 Characterisation of quaternary structure and activity assay . . . . 99

4.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1004.4.1 Structure determination . . . . . . . . . . . . . . . . . . . . . . . 1004.4.2 Overall structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 1014.4.3 3D domain swapping . . . . . . . . . . . . . . . . . . . . . . . . . 1054.4.4 Dimer interface and binding of W-loop into active site . . . . . . 1084.4.5 Biological relevance . . . . . . . . . . . . . . . . . . . . . . . . . . 1104.4.6 Misfolding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

6 CONTENTS

5 Structural Aspects of Cross-Reactive Allergens 1195.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1215.3 Clinical aspects of cross-reactivity . . . . . . . . . . . . . . . . . . . . . . 1225.4 Cross-reactive structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 1235.5 Cross-reactive fungal allergens . . . . . . . . . . . . . . . . . . . . . . . . 1245.6 Autoreactivity to human homologues of fungal allergens . . . . . . . . . . 1255.7 Three-dimensional structures of allergens . . . . . . . . . . . . . . . . . . 1265.8 Identi�cation of cross-reactive epitopes . . . . . . . . . . . . . . . . . . . 1305.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

6 Final Discussion 143Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

Acknowledgements 151

List of publications 152

Curriculum vitae 153

List of Figures

1.1 The type I hypersensitivity reaction . . . . . . . . . . . . . . . . . . . . . 181.2 Patch of conserved residues in the Fagales order and the epitope of Bet v 1 231.3 Cartoon representation of human CyPA in complex with AAPF . . . . . 251.4 A putative mechanism of CyP-catalysed cis-trans isomerisation . . . . . 261.5 Proposed mechanism of thioredoxin-catalysed protein disul�de reduction 281.6 Structure of the thioredoxin-1 dimer . . . . . . . . . . . . . . . . . . . . 291.7 Schematic diagram illustrating terms related to 3D domain swapping . . 31

2.1 Sequence alignment of Mala s 6 with homologous CyPs . . . . . . . . . . 532.2 Protein puri�cation, Western blot analysis and inhibition ELISA . . . . . 542.3 Serum IgE antibodies to recombinant cyclophilins . . . . . . . . . . . . . 552.4 Overall structure and active site of Mala s 6 . . . . . . . . . . . . . . . . 582.5 Putative IgE-binding residues . . . . . . . . . . . . . . . . . . . . . . . . 60

3.1 Crystals of Mala s 13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.2 Crystals of cleaved Mala s 13 . . . . . . . . . . . . . . . . . . . . . . . . 773.3 Crystals of wheat Trx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803.4 Cartoon representation of Mala s 13 . . . . . . . . . . . . . . . . . . . . . 823.5 Cartoon representation of the crystallographic dimer . . . . . . . . . . . 833.6 Putative IgE-binding residues . . . . . . . . . . . . . . . . . . . . . . . . 853.7 Sequence alignment of Mala s 13 with homologous Trxs . . . . . . . . . . 86

4.1 Cartoon representation of Asp f 11 . . . . . . . . . . . . . . . . . . . . . 1014.2 Topology plot of Asp f 11 . . . . . . . . . . . . . . . . . . . . . . . . . . 1024.3 Structural comparison of Asp f 11 to human CyPA . . . . . . . . . . . . 1034.4 Sequence alignment of Asp f 11 with homologous CyPs . . . . . . . . . . 1044.5 W-loop bound to active site and 2nd hinge loop . . . . . . . . . . . . . . 1064.6 Cartoons of putative forms of 3D domain swapping . . . . . . . . . . . . 1084.7 SDS-PAGE of crystallised and misfolded Asp f 11 protein . . . . . . . . . 109

7

8 LIST OF FIGURES

4.8 Gel �ltration and peptidyl-prolyl cis-trans isomerase activity . . . . . . . 112

5.1 Schematic diagram of sensitisation mechanisms . . . . . . . . . . . . . . 1225.2 The major structural families of allergens . . . . . . . . . . . . . . . . . . 129

List of Tables

1.1 Structures of 3D domain swapping proteins not reviewed previously . . . 33

2.1 Data collection and re�nement statistics . . . . . . . . . . . . . . . . . . 512.2 Induction of immediate skin reactions with recombinant cyclophilins . . . 56

3.1 Data collection and re�nement statistics . . . . . . . . . . . . . . . . . . 74

4.1 Structures of cyclophilins that have been solved to date . . . . . . . . . . 954.2 Data collection, phasing and re�nement statistics . . . . . . . . . . . . . 98

5.1 Allergen crystal structures . . . . . . . . . . . . . . . . . . . . . . . . . . 128

9

10 LIST OF TABLES

SUMMARY 11

SummaryAllergy is a speci�c deviation of the immune system caused by IgE-mediated hypersen-sitivity reactions to otherwise innocuous substances. Allergen exposure leads to cross-linking of IgE molecules bound to mast cells and basophils, which triggers the release ofin�ammatory mediators like histamine and leukotrienes. Characteristic manifestations ofallergic reactions are hay fever, eczema, urticaria and asthma. Allergic individuals oftenreact to various allergen sources, either due to polysensitisation or due to cross-reactivitybetween common, homologous allergens from di�erent sources. Cross-reactivity resultsfrom sharing antigenic determinants by antigens. Antibodies raised against a speci�c epi-tope of an allergen can also recognise common epitopes present on structurally relatedallergens. Cross-reactivity between allergens and closely related homologous human anti-gens is often observed in allergic individuals su�ering from chronic in�ammatory atopicdiseases, which leads to humoral and cell-mediated IgE autoreactivity.The aim of this work was to better understand the structural basis of cross-reactivity andto identify putative common B-cell epitopes that could be responsible for IgE-mediatedcross- and autoreactivity by comparing the three-dimensional structures of fungal al-lergens with homologous human proteins. Cyclophilins and thioredoxins from the fungiAspergillus fumigatus and Malassezia sympodialis, which are both involved in allergiesand long lasting atopic diseases, were chosen for this comparison. These proteins act as al-lergens in vivo as shown by positive skin reactions of sensitised individuals and were iden-ti�ed as IgE-binding proteins in vitro. Cross-reactivities among cyclophilins and amongthioredoxins of di�erent sources, including Homo sapiens, were shown by Western blotanalysis and inhibition ELISA. Cyclophilins as well as thioredoxins are highly conservedamong each other and are termed pan-allergens. They show cross-reactivity due to sharedstructural features, the B-cell epitopes.Since there is no structural information available on fungal cyclophilins and thioredox-ins, the primary goal was to solve the crystal structure of these proteins. Cyclophilinfrom A. fumigatus and M. sympodialis as well as thioredoxin from M. sympodialis andwheat (Triticum aestivum) were cloned, expressed and crystallised. Good crystals wereobtained from M. sympodialis and A. fumigatus cyclophilin. The former di�racted to1.5 Å at the synchrotron belonging to space group P41212. The structure was solved bymolecular replacement and re�ned to an R and Rfree value of 14.3% and 14.9%, respec-tively. Crystals from A. fumigatus cyclophilin di�racted to 1.85 Å belonging to spacegroup P3121. Structure solving was not feasible by molecular replacement, but was suc-

12 SUMMARY

cessful by the multiwavelength anomalous dispersion (MAD) method. The structure wasre�ned to an R and Rfree value of 18.9% and 21.4%, respectively. Promising crystalswere also obtained from the thioredoxins of M. sympodialis and T. aestivum. Unfortu-nately, two M. sympodialis as well as two wheat crystal forms showed various types oftwinning and additional non-crystallographic translations, which made structure solvinginfeasible. Finally, a change in the M. sympodialis thioredoxin construct - the His tag wasnot removed - resulted in a non-twinned crystal form belonging to space group P21. Thestructure was solved by molecular replacement and re�ned to a resolution of 1.41 Å withR and Rfree values of 14.0% and 16.8%, respectively.The crystal structure of M. sympodialis cyclophilin reveals the typical fold of cyclophilinsand shows the co-crystallised dipeptide Ala-Pro in the active site, whose binding modeis identical to other cyclophilin-dipeptide-complexes. The solvent-accessible surfaces ofM. sympodialis cyclophilin and cross-reactive human cyclophilin B were calculated andcompared in order to identify conserved, putative IgE-binding amino acids. The analysisrevealed three conserved, contiguous surface patches, which might represent conforma-tional, cross-reactive IgE-binding epitopes. Mutational studies might answer the question,whether the de�ned conserved patches really represent immunodominant cross-reactiveepitopes.The solved crystal structure of M. sympodialis thioredoxin was also used for the identi�-cation of putative IgE-binding amino acids. The solvent-accessible surface was comparedto the surface of human thioredoxin revealing two conserved patches. One of them de�nesthe area around the active site, which is responsible for the binding to target proteins.These results suggest that shared structural features of homologous proteins can be foundby comparing the three-dimensional structures, leading to a better understanding of IgE-mediated cross- and autoreactivity found in chronic atopic diseases. Moreover, a thoroughknowledge of the IgE-binding epitopes might allow the development of allergen variantswith reduced IgE-binding capacity that can be used as vaccines for a safer immunother-apy with reduced anaphylactic side e�ects.

Surprisingly, the structure of A. fumigatus cyclophilin revealed dimerisation by 3D do-main swapping representing one of the �rst proteins with a swapped central domain.This behaviour has not been observed among homologous cyclophilins, which all fea-ture a monomeric conformation. The domain swapped element consists of two β-strandsand a subsequent loop carrying a conserved tryptophan. The tryptophan binds into theactive site, inactivating cis-trans isomerisation. This might be a means of biological reg-ulation. Alternative 3D domain swapping could lead to misfolded forms such as N- or

SUMMARY 13

C-terminally swapped dimers, oligomers and aggregates, which indeed were observed inrelevant experiments. As for now, it is not known, if this domain swapped dimer has anybiological or pharmaceutical meaning. Further work might focus on the characterisationof the di�erent misfolded multimers and on homologous proteins in order to investigateif they show the same multimerisation behavior as A. fumigatus cyclophilin.

14 ZUSAMMENFASSUNG

ZusammenfassungEine Allergie ist eine Störung des Immunsystems, die durch eine IgE-vermittelte Re-aktion auf ansonst harmlose Substanzen verursacht wird. Allergene führen zur Vernet-zung von IgE-Molekülen, die an Mastzellen und Basophilen gebunden sind, und lösendie Freisetzung von entzündungsfördernden Mediatoren wie Histamin und Leukotrienenaus. Typische Zeichen allergischer Reaktionen sind Heuschnupfen, Ekzeme, Urtikaria undAsthma. Allergiker reagieren oft auf unterschiedliche Allergenquellen, entweder wegen ei-ner mehrfachen Sensibilisierung oder wegen einer Kreuzreaktivität zwischen ähnlichen,homologen Allergenen verschiedener Herkunft. Kreuzreaktivität erfolgt, wenn verschie-dene Allergene die gleichen antigenen Merkmale tragen. Antikörper, die gegen ein be-stimmtes Epitop eines Allergens gerichtet sind, können ähnliche Epitope auf strukturver-wandten Allergenen erkennen. Bei Allergikern, die unter chronisch atopischen Entzün-dungskrankheiten leiden, wird oft Kreuzreaktivität zwischen humanen Antigenen undverwandten, homologen Allergenen beobachtet. Dies führt zu humoraler und zellvermit-telter IgE-Autoreaktivität.Ziel dieser Arbeit war, Kreuzreaktivität strukturell besser zu verstehen und mögliche ge-meinsame B-Zell Epitope zu identi�zieren, die für die IgE-vermittelte Kreuz- und Auto-reaktivität verantwortlich sein könnten, indem man die dreidimensionale Struktur vonPilzallergenen und homologen humanen Proteinen miteinander vergleicht. Dafür wurdenCyclophiline und Thioredoxine der beiden Pilze Aspergillus fumigatus und Malasseziasympodialis ausgewählt, die bei Allergien und langanhaltenden atopischen Erkrankungeneine Rolle spielen. Diese Proteine wirken in vivo als Allergene, wie durch positive Haut-reaktionen bei sensibilisierten Personen gezeigt werden konnte, und wurden in vitro alsIgE-bindend charakterisiert. Kreuzreaktivität unter Cyclophilinen und unter Thioredoxi-nen verschiedener Arten, u.a. auch Homo sapiens, wurde mittels Western Blot Analyseund Inhibitions-ELISA gezeigt. Cyclophiline wie auch Thioredoxine sind untereinanderstark konserviert und werden deshalb als Pan-Allergene bezeichnet. Sie zeigen Kreuzre-aktivität, die durch strukturelle Gemeinsamkeiten, den B-Zell Epitopen, bedingt ist.Da keine Strukturinformation über Pilzcyclophiline und -thioredoxine vorhanden ist, wares primäres Ziel, die Struktur dieser Proteine zu lösen. Cyclophiline aus A. fumigatus undM. sympodialis, als auch Thioredoxine aus M. sympodialis und Weizen (Triticum aesti-vum) wurden kloniert, exprimiert und kristallisiert. Die beiden Cyclophiline führten zuqualitativ guten Kristallen. Die M. sympodialis Kristalle streuten am Synchrotron bis zueiner Au�ösung von 1.5 Å und gehören zur Raumgruppe P41212. Die Struktur wurde mit

ZUSAMMENFASSUNG 15

der Methode des molekularen Ersatzes gelöst und bis zu einem R-Faktor von 14.3% undeinem freien R-Faktor von 14.9% verfeinert. Die A. fumigatus Kristalle streuten bis 1.85 Åund gehören zur Raumgruppe P3121. Das Lösen der Struktur war mittels molekularenErsatzes nicht möglich, jedoch erfolgreich mit der MAD-Methode. Die Struktur wurde biszu einem R-Faktor von 18.9% und einem freien R-Faktor von 21.4% verfeinert. Vielver-sprechende Kristalle wurden auch mit den Thioredoxinen von M. sympodialis und T. aes-tivum erhalten. Leider zeigten zwei M. sympodialis wie auch zwei Weizen-Kristallformenverschiedene Arten von Verzwilligung und nicht-kristallographische Translation, die ei-ne Strukturlösung unmöglich machten. Schliesslich führte eine Änderung am Konstruktdes M. sympodialis Thioredoxins - der His-Tag wurde nicht abgeschnitten - zu einernicht-verzwillingten Kristallform, die der Raumgruppe P21 angehört. Die Struktur wur-de mittels molekularen Ersatzes gelöst und bis zu 1.41 Å mit einem R-Faktor von 14.0%und einem freien R-Faktor von 16.8% verfeinert.Die Kristallstruktur des M. sympodialis Cyclophilins zeigt die typische Faltung von Cy-clophilinen sowie das Dipeptid Ala-Pro im aktiven Zentrum, das denselben Bindemodusaufweist wie andere Cyclophilin-Dipeptid-Komplexe. Von diesem Protein und dem kreuz-reaktiven, humanen Cyclophilin B wurde die Lösungsmittel-zugängliche Ober�äche be-rechnet, um konservierte, mutmasslich IgE-bindende Aminosäuren zu identi�zieren. Diesergab drei konservierte, zusammenhängende Ober�ächenbereiche, die konformationelle,kreuzreaktive Epitope darstellen dürften. Mutationsstudien könnten Aufschluss geben, obdie identi�zierten, konservierten Bereiche auch wirklich immunodominante, kreuzreaktiveEpitope de�nieren.Die Struktur des M. sympodialis Thioredoxins wurde ebenfalls zur Charakterisierung vonIgE-bindenden Aminosäuren verwendet. Die Lösungsmittel-zugängliche Ober�äche wur-de mit jener von humanem Thioredoxin verglichen und führte zur Identi�zierung vonzwei konservierten Bereichen. Ein Bereich wird durch die Region um das aktive Zentrumde�niert, der für die Bindung von Zielproteinen verantwortlich ist. Diese Resultate zei-gen, dass gemeinsame Strukturmerkmale homologer Proteine durch Vergleichen der 3DStrukturen gefunden werden können, die zum vertieften Verständnis der IgE-vermitteltenKreuz- und Autoreaktivität bei chronisch atopischen Erkrankungen beitragen. Zudemkönnte das Wissen über IgE-bindende Epitope die Entwicklung von modi�zierten Aller-genen mit reduzierter IgE-A�nität ermöglichen, die als Impfsto�e in Immuntheraphienmit verminderten anaphylaktischen Nebenwirkungen eingesetzt werden könnten.

Überraschenderweise zeigt die Struktur des A. fumigatus Cyclophilins Dimerisierungdurch 3D Domänen-Austausch und stellt somit eines der ersten Proteine mit einem ausge-

16 ZUSAMMENFASSUNG

tauschten zentralen Element dar. Andere, homologe Cyclophiline zeigen diese Eigenschaftnicht; sie weisen alle eine monomere Konformation auf. Das ausgetauschte Element be-steht aus zwei β-Strängen und der folgenden Schlaufe, die ein konserviertes Tryptophanträgt. Das Tryptophan bindet am aktiven Zentrum und inaktiviert somit die cis-trans Iso-merisierung, was einer biologischen Regulation dienen könnte. Alternative Formen von3D Domänen-Austausch könnten zu falsch gefalteten Multimeren führen, z.B. N- oderC-terminal ausgetauschte Dimere, Oligomere oder Aggregate. Dies wurde auch tatsäch-lich in entsprechenden Experimenten beobachtet. Eine biologische oder pharmazeutischeBedeutung ist für das Dimer zum jetzigen Zeitpunkt unbekannt. Weitere Arbeiten soll-ten auf die Charakterisierung der verschiedenen falsch gefalteten Multimere eingehen unduntersuchen, ob homologe Proteine dieselben Multimerisierungs-Eigenschaften aufzeigen.

Chapter 1

Introduction

1.1 AllergyAllergy is a speci�c deviation of the immune system caused by IgE-mediated hyper-sensitivity reactions to otherwise innocuous substances [1]. These substances, termedallergens, are usually small proteins or glycoproteins, but can also be metalloproteins orlow molecular weight substances such as antibiotics, food additives, or perfumes [2].

1.1.1 IgE-mediated hypersensitivity reactionDuring the sensitisation phase upon the �rst contact with the immune system, environ-mental allergens are taken up by antigen-presenting cells (APCs), processed and presentedby class II major histocompatibility complex (MHC) molecules to naive T helper (Th)cells favouring development of Th2 cell subsets [3, 4]. Allergen-speci�c Th2 cells thencross-talk with B cells to promote their growth and di�erentiation. This results in thesynthesis of speci�c IgE molecules, and thus priming of sensitised individuals for sub-sequent encounters with the same allergen (Figure 1.1) [4]. During the provocation orchallenge phase, mast cells, basophils and eosinophils displaying IgE molecules, whichare bound on their surface via the high-a�nity receptor FcεRI, become involved in theprocess of the disease. Allergen exposure during this phase leads to cross-linking ofFcεRI-bound IgE molecules which results in activation of signal transduction from theIgE receptor and release of in�ammatory mediators like histamine and leukotrienes [4,5].Thus, allergens must have at least two di�erent or identical IgE-binding epitopes to beable to cross-link IgE antibodies.The released mediators cause increased vascular permeability, vasodilation, bronchial andvisceral smooth muscle contraction and local in�ammation [6]. This reaction is called

17

18 CHAPTER 1. INTRODUCTION

B cell

IgEAntigen

Fc RI

Mast cell

Basophil

Antigen

Antigen-

presenting cell

IL-4

IL-13

Th2 cell

IL-5

Eosinophil

Atopic

disease

Inflammatory

mediators

MHCII

TCR

Figure 1.1: The type I hypersensitivity reaction [7].

immediate hypersensitivity. In its most extreme systemic form, called anaphylaxis, mastcell-derived or basophil-derived mediators can restrict airways to the point of asphyxiationand produce cardiovascular collapse leading to death. Individuals prone to develop strongimmediate hypersensitivity responses are called atopic and are said to su�er from allergies.In di�erent individuals, atopy may take di�erent forms presenting as hay fever, asthma,urticaria, or chronic eczema [6].

1.1.2 Nature of allergensAntigens that elicit strong immediate hypersensitivity reactions are called allergens. Al-lergens may be either proteins or chemicals bound to proteins. It is not known why someantigens cause strong allergic responses whereas other antigens, which may be encoun-tered by the same route of administration, are simply not allergenic and instead result inprotective humoral or cell-mediated immune responses. The property of being allergenicmay reside in the antigen itself, perhaps in epitopes seen by certain T cells [6].Respiratory allergies can be initiated and provoked by inhaled allergens that occur in-doors and outdoors. Common sources of indoor allergens are house dust mites andanimals like cats, dogs and birds. Typical examples of outdoor allergens are pollens fromtrees and grasses. Fungi like Aspergillus fumigatus, Malassezia sympodialis, Alternaria al-ternata and Cladosporium herbarum can be present as outdoor as well as indoor allergenicsources [8].

1.1. ALLERGY 19

Fungi are saprophytic, eukaryotic organisms that grow in �lamentous (moulds) or uni-cellular (yeasts) form. They occur ubiquitously in the environment growing on plants,animals, foods and decaying matter [9]. Fungal life cycles have been divided into perfect(sexual) and imperfect (asexual) states, but in either case spores or conidia are producedand released into the environment [10]. As universal atmospheric components, fungalspores are now recognised as an important cause of respiratory allergy. There are usu-ally high numbers of airborne spores in outdoor air throughout the year that frequentlyexceed pollen concentration by 100 - 1000 fold [11,12].Anaphylactic responses to foods typically involve highly glycosylated small proteins.These structural features probably protect these allergens from denaturation and degra-dation in the gastrointestinal tract, allowing them to be absorbed intact. The intactproteins serve as a source of foreign peptides for activation of T-cells and can directlyinteract with IgE [6]. Relevant allergens for intestinal food allergy are milk proteins,vegetables, hazelnut, wheat, apple, pork and egg. The clinical symptoms include nau-sea, vomiting, abdominal pain, cramping and distension, �atulence, constipation anddiarrhoea [13].

1.1.3 Cross-reactivityAllergic individuals often react to various allergen sources. The symptoms can resulteither from polysensitisation or from cross-reactivity between common homologous aller-gens from di�erent sources. Cross-reactivity results from sharing antigenic determinantsby two antigens. Antibodies raised against a speci�c epitope of an allergen can recog-nise common epitopes present on structurally related allergens. Due to cross-reactivitybetween homologous allergens from di�erent sources it is possible that an individual ex-periences an allergic reaction at the �rst contact with a new source of exposure withoutbeing aware of a sensitisation. All the cross-reactive proteins described so far have beenfound to re�ect common features on the level of both, primary and tertiary structure. Incontrast, not all proteins that share a similar fold are necessarily cross-reactive. Similarprotein folds can be found with a sequence identity as little as 25% while cross-reactivityis rare below 50% sequence identity [14]. Cross-reactive proteins found in various organ-isms are also termed pan-allergens [15, 16].

The article �Structural Aspects of Cross-reactive Allergens� published in �Recent ResearchDevelopments in Allergy & Clinical Immunology� gives a good overview on the topic(see Chapter 5). It focuses on cross-reactive fungal allergens and allergens in general.


Moreover, it summarises the 3D structures of allergens known to date and discusses anapproach to the identi�cation of cross-reactive epitopes.

1.2 Characterisation of antibody-binding epitopesA simple approach to identify putative cross-reactive B-cell epitopes is to determineconserved, solvent-exposed amino acids shared among cross-reactive allergens. Contigu-ous surface patches that are made up by these conserved residues might be responsiblefor cross-reactive IgE-binding. In order to accurately interpret this comparable stud-ies, knowledge about antibody-antigen interactions in general is necessary. The atomicdetails of interaction are known for more than 30 antigen-antibody combinations [14].

1.2.1 Antigen-antibody interactionsLaver et al. [17] studied the crystal structures of �ve antibody Fab-antigen complexes.They conclude that epitopes occupy large areas composed of 15-22 amino acids, which lieon two to �ve surface loops. Energetic calculations suggest that a smaller subset of 5-6of these residues contribute most of the binding energy, with the surrounding residuesmerely involved in complementarity [18]. All �ve of the structurally de�ned epitopes areof the conformational type; they are discontinuous or nonlinear and are assembled fromresidues from several di�erent portions of the polypeptide chain [17]. Each epitope hasa buried surface area on the antigen of 650-900 Å2. There are several hydrogen bondsbetween the antibody and antigen as well as salt links and hydrophobic interactions.Davies and Cohen [19] studied the complexes of lysozyme with four di�erent antibodyFab-fragments as well as two Fab complexes with in�uenza virus neuraminidase and threeFab complexes with their anti-idiotype Fabs. The antibody-antigen interfaces bury asurface area of 560-855 Å2 on the antigens. 10 to 20 amino acids of the antigens are makinginteractions (H-bonds, salt bridges or Van-der-Waals contacts) with the antibodies. Thefour epitopes formed on lysozyme cover 45% of the molecular surface area, supportingthe hypothesis that the entire surface of an antigen is potentially antigenic. Two of theepitopes are partially overlapping, while the other ones are spatially separated.

1.2.2 Conformational changes upon antibody bindingThe interaction of antibody with antigen involves conformational changes in both theantibody and the antigen that can range from insigni�cant to considerable. In general,the formation of a complex will adopt many of the characteristics of induced �t, similar

1.2. CHARACTERISATION OF ANTIBODY-BINDING EPITOPES 21

to those seen in other macromolecular interactions [19]. The four crystal complexeswith lysozyme that were studied by Davies and Cohen [19] clearly demonstrate that theinteraction of antibody with antigen can produce signi�cant conformational changes inthe antigen, mainly in regions that are demonstrably �exible. The largest Cα separationin the superimposed lysozymes was 8.17 Å between Gly102 of lysozyme in the Fab-D1.3 complex [20] and the same residue in the Fab-D11.5 complex [21]. There is alsosome �exibility in this region in the uncomplexed lysozyme; when the two independentmolecules in the monoclinic form [22] are superimposed, the Cα atoms of Gly102 aredisplaced by 3.40 Å. Therefore, �exibility in this region permits the enhancement of thecomplementarity with antigens.

1.2.3 Mutational e�ects on antigen-antibody bindingMutations in either antibody or antigen can be used to analyse the contributions ofindividual residues to the formation of the complex. There have been several studieswhere the binding e�ects have been correlated with the known three-dimensional struc-ture. A mutational analysis of the epitope of human growth hormone showed that onlyone-quarter of the side chains buried by the antibody could account for most of thebinding energy [23,24]. A similar analysis of the receptor epitope revealed a complemen-tary hot spot on the receptor surface, where there is a hydrophobic region dominatedby two tryptophan residues, which account for more than three-quarters of the bindingenergy [24,25].Chacko et al. [26] have examined the changes in the complex of lysozyme with the an-tibody HyHEL-5 that accompany the �conservative� substitution of lysine for arginineat position 68 of lysozyme. The mutation produces a decrease by a factor of 1000 inbinding, which can be explained by the net loss of a hydrogen bond. Tulip et al. [27]examined two complexes of mutant neuraminidase with antibody NC41. The mutationsAsn329-Asp and Ile368-Arg produced only a slight reduction in a�nity. Both structuresdemonstrate that local structural rearrangements can be used to accommodate theseamino acid substitutions.What is the implication of these �ndings on the comparative identi�cation of cross-reactive B-cell epitopes? Similar amino acids in the same position of a putative cross-reactive B-cell epitope on two homologous allergens can have a big impact on antibody-binding as shown above. On the other hand, non-similar amino acids can have no e�ectat all, which may be explained by conformational rearrangements in both, the allergenepitope and the antibody. Therefore, in a B-cell epitope analysis, the e�ect of non-


identical amino acids is di�cult to predict, and it is probably better to consider onlyidentical residues. Amino acids that are highly solvent-exposed will probably contributefor most of the binding energy. Therefore, about 4 to 8 amino acids, which are identicaland clearly solvent exposed in two or more cross-reactive allergens forming contiguoussurface patches and lying on a few di�erent surface loops, are the most likely candidatesfor forming cross-reactive IgE-binding epitopes. On the other hand, one should not forgetthat - in rare cases - even a low-conserved surface patch could account for a cross-reactiveB-cell epitope, due to e�ects like induced �t. Obviously, this kind of epitopes will hardlybe predictable.

1.2.4 Crystal structure of an allergen-antibody complexThe only method allowing a complete de�nition of a B cell epitope remains co-crystalli-sation of the allergen with a monoclonal antibody Fab fragment and solving the structureof the complex by X-ray crystallography. This method requires homogeneous reagents forgrowth of crystals and cannot be performed using polyclonal human serum IgE. Becausemonoclonal allergen-speci�c human IgE is di�cult to obtain, monoclonal antibodies fromother sources are used.Mirza et al. [28] presented the �rst crystal structure of an allergen-Fab complex betweenthe major birch pollen allergen Bet v 1 and the Fab fragment from a murine monoclonalIgG1 antibody (BV16), that has been solved to 2.9 Å resolution. The IgG antibody isshown to inhibit the binding of allergic patients' serum IgE to Bet v 1. Therefore, theepitope de�ned by BV16 is partly or completely overlapping Bet v 1 epitopes recognisedby human serum IgE. The allergen-IgG complex may therefore serve as a model for thestudy of allergen-IgE interactions relevant in allergy [28].No major conformational changes in the structure of Bet v 1 are seen upon binding to themonoclonal BV16 Fab. The buried area on Bet v 1 is calculated to be 931 Å2 (of a totalmolecular surface of 9119 Å2). This is comparable to what is found for other protein-Fabcomplexes (see above). Given the small size of the epitope (only around 10% of thetotal allergen surface), it is obvious that it is theoretically possible to bind more thanone IgE molecule to the same allergen molecule. The epitope is constituted of 16 aminoacids; half of them are involved in hydrogen bonds, the remaining ones in hydrophobicinteractions with the antibody. The epitope can be classi�ed as discontinuous. However,residues 42-52 constitute 80% of the contact surface. Although this continuous region isparticipating in the majority of complex interactions, the Bet v 1-synthetic peptide 39-53is not able to inhibit the binding of Bet v 1 to BV16 as judged by ELISA [28]. Thus, the

1.2. CHARACTERISATION OF ANTIBODY-BINDING EPITOPES 23

Figure 1.2: Patch of conserved residues in the Fagales order and the epitope of Bet v 1.The solvent-accessible surface of Bet v 1 is shown. (A) Patch of conserved residues among theallergens from the Fagales tree order (black area), comprising residues 41-52. (B) The residuescomprising the BV16 epitope (black area) with the same orientation of the molecule [28].

constraints imposed by the structure of Bet v 1 on the conformation of loop 42-52 seemto be of major importance for antibody binding.The crystal structure of uncomplexed Bet v 1 was solved earlier [29]. Data on surface-exposed amino acids of this allergen structure was combined with data on conserved aminoacids comparing 57 homologous sequences within the Fagales. By this methodology,three conserved patches large enough to accommodate antibody-binding epitopes wereidenti�ed and proposed to account for the clinically observed cross-reactivity of birchpollen allergic patients towards alder, hazel and hornbeam [29]. Figure 1.2 shows oneof these conserved patches and the epitope de�ned by the BV16 antibody. The surfacearea de�ned by the conserved residues is contained in the surface area de�ned by theinteractions with BV16 Fab. This �nding clearly suggests that cross-reactive, dominanthuman IgE epitopes are located in conserved surface patches with the size of averageantigen-antibody contact areas [28].

Therapeutic implication

The only curative treatment for allergy today is immunotherapy, based on repeated sub-cutaneous injections of allergen [30]. One major disadvantage of current immunotherapyis that it can cause severe side e�ects, such as asthma attacks and anaphylactic shock.To overcome these problems, new strategies have been developed to reduce the allergenicpotential of allergens [31]. The identi�cation of IgE epitopes on allergens allows the mod-i�cation of important allergens such that they display strongly reduced allergenic activityby disrupting the conformational IgE epitopes. Such hypoallergenic allergen derivatives


can be used as candidates for vaccines in allergen-speci�c immunotherapy aiming at thestimulation of IgG- and avoiding IgE-production, thus exhibiting a reduced risk of im-mediate side e�ects [31].

1.3 CyclophilinCyclophilins (CyPs) constitute a family of cytosolic proteins present in a wide num-ber of eukaryotic and prokaryotic species; over three hundred sequences of CyPs havebeen reported to date [32]. They are involved in many biological processes. Belongingto the family of immunophilins, CyP binds the immunosuppressive drug cyclosporin A(CsA) [33]. CyP is an enzyme that catalyses peptidyl-prolyl cis-trans isomerisation(PPIase) [34,35]. CyPs can be classi�ed into two categories: small and large CyPs. Thesmall CyPs contain a single domain called PPIase domain that has an in vitro catalyticactivity for peptidyl-prolyl cis-trans isomerisation. The large CyPs consist of a PPIasedomain and one or more additional domains such as a tetratricopeptide repeat (TPR)domain like in cyclophilin 40 [36].

1.3.1 Structure of cyclophilinThe three-dimensional structures of many CyPs have been solved, unligated and in com-plex with its partners like CsA, proline-containing peptides and HIV-1 capsid as well as ina ternary complex with CsA and calcineurin (for a review see [37]). Human cyclophilin A(CyPA), a cytoplasmic protein with a molecular weight of about 18 kDa, is a represen-tative of the small CyPs. The three-dimensional structure of CyPA was independentlysolved for the unligated CyPA [38] and CyPA in complex with a tetrapeptide N -acetyl-Ala-Ala-Pro-Ala-amidomethylcoumarin [39]. The structure of CyPA is a beta-barrel witheight antiparallel beta-strands and two alpha-helices covering the bottom and top of thebarrel (Figure 1.3). A shallow pocket on the barrel surface constitutes the active sitefor cis-trans isomerisation of a peptidyl-prolyl bond. This pocket is mainly hydrophobicand contains residues Arg55, Phe60, Met61, Gln63, Phe113, Trp121, Leu122 and His126(numbering referring to human CyPA).

1.3.2 Cis-trans isomerisation activityA peptidyl-prolyl cis-trans isomerase (PPIase) catalyses cis-trans isomerisation of apeptidyl-prolyl amide bond [41]. Four families of PPIases have been identi�ed: CyP,FKBP, parvulin and Pin1, which share no sequence and structure similarity. CyP and

1.3. CYCLOPHILIN 25

Figure 1.3: Cartoon representation of human CyPA in complex with AAPF (stick model inlight grey) showing the overall fold of CyPs (PDB code 1RMH [40]).

FKBP are known as immunophilins involved in the T cell activation [42]. The mechanismof cis-trans isomerisation of a peptidyl-prolyl bond by CyP has been extensively studiedby biochemical and structural approaches. �Catalysis by distortion� and �protonation onthe amide nitrogen� are the most likely mechanisms. �Catalysis by distortion� proposesthat the N-C=O peptide plane is distorted and stabilised upon binding to CyP [43].The mechanism of �protonation on the amide bond� (Figure 1.4) assumes that a CyPresidue protonates or forms a hydrogen bond with the amide nitrogen to deconjugate theN-C=O amide bond, on the basis of quantum chemistry calculations [44]. The crystalstructure of CyPA in complex with the substrate succinyl-Ala-Ala-Pro-Phe-nitroaniline(AAPF) shows that AAPF binds to the hydrophobic pocket on the surface of the CyPAbarrel (Figure 1.3) [40]. In the structure, the guanidine group of Arg55 forms a hydrogenbond with the carbonyl oxygen of the proline of AAPF and is also located about 4 Åaway from the prolyl nitrogen of AAPF. It has been hypothesised that the positivelycharged guanidine nitrogen of Arg55 may dynamically approach the amide nitrogen dur-ing catalysis to interact with the lone pair electrons of the peptide nitrogen atom [45]. Asresult, the conjugation system of the peptide plane may be substantially weakened andthe isomerisation initiates. This putative mechanism is consistent with the mechanismof �protonation on the amide nitrogen� (Figure 1.4).


Figure 1.4: Protonation on the amide nitrogen: Schematic presentation of a putative mecha-nism for the peptidyl-prolyl cis-trans isomerisation by CyP [37,45].

1.3.3 Suppression of T-cell activationCyP binds the immunosuppressive drug cyclosporin A (CsA) [33] in a 1:1 ratio at the ac-tive site, which is responsible for cis-trans isomerisation. On one hand, CsA is an inhibitorof the cis-trans isomerisation activity. On the other hand, the complex supresses T-cellactivation. CyP-CsA binds calcineurin (CN), thereby inhibiting its dephosphorylationactivity. CN is a Ca2+/calmodulin dependent serine/threonine protein phosphatase [46].The molecule of calcineurin is a heterodimer comprising a 59 kDa catalytic subunit cal-cineurin A and a 19 kDa regulatory subunit calcineurin B. CN is involved in severalbiological processes. The most extensively studied function of CN is its involvement inthe signal transduction pathway towards the T-cell activation. Calcineurin is a commonreceptor for binding of two immunophilin-immunosuppressant families: CyP-CsA andFKBP-FK506 [47, 48]. The binding of CyP-CsA or FKBP-FK506 inhibits CN dephos-phorylation activity on transcription factors such as nuclear factor of activated T-cell(NFAT), thus leading to suppression of T-cell activation [49,50].

1.3.4 Binding of HIV-1 capsid proteinCapsid protein p24 (CA) is the coating protein of HIV-1 and oligomerises to form the innerconical viral particle. The assembly of HIV virion particles is essential for understandingthe HIV infectivity and for design of antivirus drugs, but is poorly understood [51, 52].CyP binds to the N-terminal domain of HIV-1 CA and is required for full infectiousactivity of HIV-1 [53�55]. The structure of CyPA in complex with the N-terminal domain

1.4. THIOREDOXIN 27

of CA showed that CyPA binds to loop 85-93 of HIV-1 CA [56]. The loop occupies thesame binding pocket as the substrate for cis-trans isomerisation sharing many commoninteractions. The role of CyPA in the HIV infectivity remains unclear. In considerationof the peptidyl-prolyl cis-trans isomerisation as a rate determining step of protein folding,one might expect that CyP helps with the folding and thus assembly of CA into the coreof the HIV-1 particle. However, the crystal structure of the complex suggests that CyPacts as a molecular chaperone, rather than a PPIase, in the HIV-1 infectious process [37].

1.4 ThioredoxinThioredoxin (Trx) was �rst described in 1964 by Laurent et al. [57] as a small redoxprotein from Escherichia coli. Trx family members have subsequently been shown to bepresent in a wide number of eukaryotic and prokaryotic species (for reviews see [58�60]).The mammalian thioredoxins undergo NADPH-dependent reduction by thioredoxin re-ductase and in turn reduce oxidised cysteine groups on proteins. They have been impli-cated in a number of mammalian cell functions (for a review see [61]). Activity has beenfound outside the cell (cell growth stimulation and chemotaxis), in the cytoplasm (as anantioxidant and a reductant cofactor), in the nucleus (regulation of transcription factoractivity) and in the mitochondria.

1.4.1 Redox biochemistry of thioredoxinTrxs constitute a family of proteins that serve as general protein disul�de oxido-reducta-ses. They contain a conserved catalytic site (-Trp-Cys-Gly-Pro-Cys-Lys-) that undergoesreversible oxidation to the cystine disul�de (Trx-S2) through the transfer of two electronsand two protons from the catalytic site cysteine residues (Trx-(SH)2) to, typically, adisul�de substrate (X-S2). Subsequently, the oxidised Trx is reduced back to the Trx-(SH)2-form by the NADPH-dependent �avoprotein thioredoxin reductase (TrxR). Thereactions that Trx undergoes are:

Trx-(SH)2 + X-S2 ⇀↽ Trx-S2 + X-(SH)2TrxR

Trx-S2 + NADPH + H+ ⇀↽ Trx-(SH)2 + NADP+

A mechanism for the thioredoxin-catalysed protein disul�de reduction has been pro-posed [58,62]. Trx has a hydrophobic surface area, which is conserved in the Trx-family.First, reduced Trx binds via this surface to a substrate protein X making a complex


Figure 1.5: Proposed mechanism of thioredoxin-catalysed protein disul�de reduction. Reducedthioredoxin [Trx-(SH)2] binds to a target protein via its hydrophobic surface area. Nucleophilicattack by the thiolate of Cys32 results in formation of a transient mixed disul�de, which isfollowed by a nucleophilic attack of the deprotonated Cys35 generating Trx-S2 and the reducedprotein [58,62,64].

(Figure 1.5). Second, in the hydrophobic environment of the complex, the thiolate ofCys32, acting as a nucleophile, attacks the target protein to form a covalently linkedmixed disul�de transition state. Last, attack of the now deprotonated thiolate of Cys35on the disul�de generates a dithiol in the target protein and a disul�de in thioredoxin.Trxs normally contain a buried Asp behind the less exposed cysteine (Cys35 in human).A water molecule is hydrogen bonded to one of the oxygens on the carboxylate. Thenucleophilicity of Cys35 is thought to be increased by the Asp, which has been suggestedto take up a proton from the thiol, and which most probably acts via the bound watermolecule [63]. Small conformational changes in thioredoxin and the target protein occurduring binding and the subsequent electron-transfer steps.

1.4.2 Structure of thioredoxinSolution and crystal structures have been solved of Trxs from several organism includinghuman thioredoxin-1 [65, 66]. Trx is a compact globular protein with a �ve-strandedbeta-sheet forming a hydrophobic core surrounded by four alpha-helices on the externalsurface (Figure 1.6). The conserved active site amino acids, -Trp-Cys-Gly-Pro-Cys-, linkthe second beta-strand to the second alpha-helix and form the �rst turn of the secondhelix. This stable tertiary structure is known as the Trx fold [67]. In addition to thiore-doxins, this structure has been found in another nine protein classes, including the redoxproteins glutaredoxin, Dsb protein from Escherichia coli, glutathione peroxidase, glu-tathione s-transferase, protein disul�de isomerase, cytochrome c oxidase (COX protein,conserved among prokaryotes and eukaryotes), peroxiredoxins, evolutionary conserved

1.4. THIOREDOXIN 29

Figure 1.6: Structure of the thioredoxin-1 dimer showing the Trx fold as well as the position ofthe Cys32 and Cys35 catalytic site residues and the cross-linked Cys73 residues [61](PDB code1ERT [66]).

Dim 1 protein and iodothyronine selenodeiodinases, together forming the thioredoxin-like-fold superfamily.Human thioredoxin-1 forms covalently linked homodimers in solution [66], especially inthe presence of a strong oxidant or when stored at high concentrations [68]. The crystalstructure revealed a dimeric structure in which monomers are disul�de-linked throughCys73 from each subunit, and active site residues are reduced and buried in the dimerinterface (Figure 1.6). It is not clear what role thioredoxin-1 dimers might play, sincethe active site becomes buried on dimer formation, leaving the protein inactive. One hy-pothesis is a possible regulatory mechansism [69]. Very similar non-covalent dimers havebeen described for Chlamydomonas reinhardtii Trx h [70] and Drosophila melanogasterTrx [71], whereas in all other known Trx structures, this dimerisation behaviour was notobserved.

1.4.3 Biological activitiesTrxs have many biological activities. Bacterial Trx has been reported to act as a cofactor.It is a source of reducing equivalents for ribonucleotide reductase [57], which catalysesthe conversion of nucleotides to deoxynucleotides.Human thioredoxin-1 acts as a growth factor and is secreted by a variety of cells. Itstimulates the growth of lymphocytes [72], normal �broblasts [73] and various cancer


cells [74, 75]. Second, Trx-1 acts as antioxidant. It has direct antioxidant propertiesremoving hydrogen peroxide [76]. It also acts as an e�cient electron donor to humanglutathione peroxidase, which reduces oxidant species such as H2O2 and alkyl preox-ides [77]. Third, Trx-1 selectively activates or regulates the DNA-binding of a number oftranscription factors. This includes the transcription factor NF-κB [78], the glucocorti-coid receptor [79] and the transcription factor AP-1 [80]. Furthermore, Trx-1 has beenshown to prevent apoptosis of lymphoid cells [81].Trx-1 may play a role in a number of human diseases but particularly cancer, in whichincreased levels of Trx-1 are found in many tumors and are associated with aggressivetumor growth [74, 82, 83]. Thus, Trx-1 represents a target for drugs aimed to inhibitgrowth of cancer cells.

1.5 3D domain swappingThree-dimensional domain swapping is the event by which a monomer exchanges partof its structure with identical monomers to form an oligomer where each subunit hasa similar structure to the monomer. The accumulating number of observations of thisphenomenon in crystal structures has prompted speculation as to its biological rele-vance. Domain swapping was originally proposed to be a mechanism for the emergenceof oligomeric proteins and as a means for functional regulation, but also to be a poten-tially harmful process leading to misfolding and aggregation (for reviews on 3D domainswapping see [84�86]).

1.5.1 De�nition and terminology3D domain swapping is a mechanism for two or more protein molecules to form a dimeror higher oligomer by exchanging an identical structural element (�domain�). The in-tertwined oligomer is composed of subunits having the same structure as the originalmonomer, with the exception of the hinge loop that connects the exchanging part withthe rest of the structure and which is usually folded back on itself in the monomer and ex-tended in the domain-swapped oligomer [87,88]. The term �domain� is not used in a strictsense, as proteins have been reported to swap entire tertiary globular domains [88, 89]but also often only a single element of secondary structure, such as an α-helix [90, 91]or a single strand of β-sheet [92]. The exchanging part of the structure makes the sameinteractions in the oligomer as it makes in the monomer, but its interactions are formedinter- rather than intramolecularly. These interactions that are identical in the monomer

1.5. 3D DOMAIN SWAPPING 31

Figure 1.7: Schematic diagram illustrating terms related to 3D domain swapping. The swappeddomain in an oligomer is a globular domain or a structural element of one subunit that extendsinto another subunit and interacts with the main domain of this subunit. This interaction isessentially identical to that of the same domain in the monomer [85].

and oligomer form the closed interface (Figure 1.7). Because subunits are often close toeach other in a domain swapped oligomer, a new interaction interface that is absent inthe monomer may be formed, and this is termed the open interface [88].If both the monomer and the dimer (or higher oligomer) of a molecule exist in stableforms, in which the dimer adopts a domain-swapped conformation and the monomeradopts a closed conformation, then this protein is considered to be a bona �de exampleof 3D domain swapping. Some proteins form intertwined, apparently domain-swappedoligomers without a known closed monomer. If these proteins have homologues knownto be closed monomers, these oligomers are considered to be �quasi-domain swapped�. Ifa protein forms an oligomer by exchanging domains, but there is no monomeric form orhomologue for the protein, this protein is considered a candidate for 3D domain swap-ping [93].

1.5.2 Types of 3D domain swappingThe swapped domains have diverse sizes and sequences. A swapped �domain� can beone structural element made of several residues. It can also be an entire tertiary domainconsisting of hundreds of residues. Sequence comparison shows the lack of sequence


similarity among these domains. No speci�c sequence motif can be found among thesedomains. Therefore, based on its sequence, a protein cannot be predicted to be domain-swapped or not. The diverse size and sequence of the swapped domains indicate thatthe closed interfaces of these domain-swapped proteins are di�erent from each other. Inaddition, various types of interactions are formed at di�erent closed interfaces, includinghydrophobic interactions, hydrogen-bonding, electrostatic interactions and even disul�debridge interactions [94�96].Until the beginning of 2002, about 40 crystal or solution structures of domain swappedproteins were known, which were reviewed and listed by Liu and Eisenberg [85]. Theyshow all N- or C-terminal domain swapping, with one exception. Blood coagulant factorIX/X-binding protein (IX/X-bp) shows internal domain swapping; it exchanges an inter-nal element instead of the N- or C-terminus [97]. IX/X-bp is an anticoagulant isolatedfrom the venom of the habu snake. It consists of two homologous subunits linked byan intermolecular disul�de bond. The two subunits form a heterodimer by exchanginga loop in the central part of the molecules. Structural comparison of the two subunitswith mannose binding protein (MBP) shows that they adopt the same fold, except thatthe exchanged loop in the IX/X-bp folds back to the same polypeptide chain in MBP.Thus, IX/X-bp is quasi-domain swapped. Because domain swapping takes place in themiddle of IX/X-bp, there are two hinge loops in each subunit. Until 2002, IX/X-bp wasthe only known example of a domain-swapped heterodimer; all other domain-swappedproteins were homooligomers. Some proteins are also known to domain swap more thanone domain. RNase A can swap both, the N-terminus and/or the C-terminus. Crystalstructures are known of RNase dimers, where the N-termini are swapped [90], or wherethe C-termini are swapped [98]. Moreover, biochemical studies support a model of anRNase trimer, where N- and C-terminal domain swapping occur simultaneously [98].

1.5.3 Recent examples of 3D domain swappingIn the review of Liu and Eisenberg [85], 14 domain swapped proteins were classi�edas bona �de 3D domain-swapping proteins, where both the monomer and the domainswapped dimer form of a speci�c protein are known. 13 proteins were classi�ed as quasi-domain swapping, where a homologue is known in the non-swapping monomer form, and14 proteins were classi�ed as putative candidates for 3D domain swapping, where nohomologous monomer form is known. In a PubMed and PDB data bank search, a further22 protein structures showing 3D domain swapping were found that have been publishedsince 2002 or that were missed in the previous review (Table 1.1).

1.5. 3D DOMAIN SWAPPING 33

Protein Swapping Classi�- Quaternary PDB Ref.part cation state code

N-terminal oligomerisation domain N-terminal candidate homodimer 1K1F [99]of human Bcr-Abl (Bcr1−72)Bacillus subtilis N-terminal bona �de homodimer 1MU4 [100]catabolite repression HPr (Crh)Escherichia coli C-terminal candidate homodimer 1K6W [101]cytosine deaminaseEscherichia coli 2-dehydro-3-deoxy- C-terminal candidate homodimer 1DXF [102]galactarate aldolase (DDGA)Human 8-kDa dynein light chain internal improper homodimer 1CMI [103](DLC8) bona �deEscherichia coli C-terminal quasi homo- 1L6W [104]fructose-6-phosphate aldolase (F6PA) pentamerFlammulina velutipes N-terminal candidate homodimer 1OSY [105]Fve (major fruiting body protein)Human glucocorticoid receptor internal improper homodimer 1NHZ [106]ligand-binding domain (GR-LBD) bona �deBacillus subtilis C-terminal quasi homodimer 1WKQ [107]guanine deaminaseMethanococcus jannaschii C-terminal candidate homo-24-mer 1SHS [108]heat shock protein HSP16.5 & internalWheat heat shock protein N-, C-term. candidate homo- 1GME [109]HSP16.9 & internal dodecamerEscherichia coli C-terminal candidate homodimer 1K75 [110]L-histidinol dehydrogenase (HisD) & internalAspergillus nidulans C-terminal candidate homodimer 1DQU [111]isocitrate lyase (ICL)Mycobacterium tuberculosis C- and candidate homodimer 1SR9 [112]α-IPMS (encoded by leuA) N-terminalM. tuberculosis ketopantoate C-terminal candidate homodimer 1OY0 [113]hydroxymethyltransferase (KPHMT)C-terminal noncollagenous (NC1) internal candidate heterotrimer 1M3D [114]domain of bovine type IV collagen [(α1)2α2]C-terminal PGN-binding domain C-terminal bona �de homodimer 1SK4 [115]of human PGRP-IαPyrococcus furiosus N-terminal quasi homodimer 1NNQ [116]rubrerythrinSulfolobus tokodaii N- or quasi homodimer 1J30 [117]sulerythrin C-terminalShewanella massilia C-terminal candidate homodimer 1N1C [118]TorD chaperoneHuman myeloid cell N-terminal quasi homodimer 1Q8M [119]activating receptor TREM-1Llama VHH domain C-terminal quasi homodimer 1SJV [120](VHH-R9)

Table 1.1: Structures of 3D domain swapping proteins not reviewed previously [85].


1.5.4 Internal domain swappingIn the data bank search, six more cases of internal 3D domain swapping have been found.Two of them were claimed to be bona �de domain swapping: human DLC8 [103] andhuman GR-LBD [106]. In fact, the domain swapping elements of these two proteins donot adopt the same conformation or position as the relevant elements in the monomerform. In the human 8-kDa dynein light chain (DLC8), the exchanged element of thedomain swapped dimer adopts a β-strand, whereas in the monomer state, it forms acoiled loop, which doesn't fold back on the monomer. In the human glucocorticoidreceptor ligand-binding domain (GR-LBD), the exchanged element adopts an α-helix inboth, the domain swapped dimer as well as the monomer, but the exchanged helix of aneighbouring subunit is oriented in a di�erent way than the respective non-swapped helixin the monomer state. Therefore, they do not follow the classical de�nition of proper 3Ddomain swapping. They are thus termed improper 3D domain swapping.Three candidates for 3D domain swapping show a combination of internal, N-terminal andC-terminal domain swapping: M. jannaschii heat shock protein HSP16.5 [108], wheatheat shock protein HSP16.9 [109] and E. coli HisD [110]. They form all homooligomers.The NC1 domain of bovine type IV collagen is also a candidate for internal 3D domainswapping, but forms a heterotrimer [114]. It consists of two α1 chains and one α2 chain,involving domain swapping of two central β-strands.

1.5.5 Biological relevanceDomain swapping was originally proposed to be a mechanism for the emergence ofoligomeric proteins and as a means for functional regulation, but also to be a poten-tially harmful process leading to misfolding and aggregation [84]. Clear evidence for afunctional role of domain swapping is still lacking and, since all these structures have beenobserved in crystal forms, they could easily be suspected to be an artefact of the non-physiological conditions and high protein concentrations used for crystallisation. Severalhypotheses together with coincident lines of evidence nevertheless suggest a signi�cancefor domain swapping in vivo.One potential biological role for domain swapping is the functional regulation of proteins.Glyoxalase I from Pseudomonas was shown to exist both in an active domain-swappeddimeric state and as a metastable and less active monomer. In vitro conversion fromdimeric to metastable monomeric glyoxalase I can be triggered upon addition of glu-tathione [121]. After removal of glutathione, glyoxalase I slowly reverts to a more stableand active domain-swapped form. Although no evidence exists for such a mechanism

1.6. AIMS OF THE PRESENTED WORK 35

in vivo, this demonstrates that functional regulation by domain swapping can, in prin-ciple, be achieved. Another interesting �nding is that formation of domain-swappedoligomers can introduce allosteric regulation of the protein activity already present inthe monomer, as shown in bovine seminal RNase [122]. Also, several receptor proteins,including G-coupled receptors, are thought to dimerise by domain swapping [123].A second proposed biological implication of domain swapping is a mechanism for proteinmisfolding, aggregation and amyloid formation, which might lead to amyloid and priondiseases [93, 124�126]. Direct evidence for domain swapping as a mechanism for aggre-gation and amyloid formation is lacking, but di�erent observations strongly suggest thatit might indeed be the case. First, transient aggregation observed during refolding ofseveral proteins has been suggested to occur by domain swapping [127�129]. Second, arole for domain swapping in the processes of prion and amyloid formation is suggestedby the crystallisation of both the human prion protein [96] and the amyloidogenic humancystatin C [130,131] in domain-swapped forms.Finally, domain swapping has been proposed to contribute to structural diversi�cationand the emergence of oligomers during evolution [93]. Some modern day proteins mayhave evolved by 3D domain swapping, even if their structures do not presently reveala domain exchange [86]. Chirgadze et al. [132] showed that a domain-swapped dimercan be converted to a side-by-side dimer by a single point mutation. If such an eventoccurred during the natural evolution of a protein, its domain-swapped history would becompletely lost and it would appear to be merely a side-by-side dimer. Some present dayside-by-side dimers might have evolved through a similar pathway, in which the domainswap of the ancient dimer was an essential step because it provided the cohesion energyfor initial dimerisation.

1.6 Aims of the presented workCyclophilins and thioredoxins of di�erent fungi were identi�ed as IgE-binding proteins.Among other allergens, they are responsible for allergies and long lasting atopic diseases.Serum IgE of individuals sensitised to A. fumigatus or M. sympodialis also recognisehomologous proteins from other organisms including human. These proteins are highlyconserved and termed pan-allergens. They show cross-reactivity due to shared structuralfeatures, so called B-cell epitopes.The overall aim of this work was to compare the three-dimensional structures of thefungal allergens and the homologous human proteins in order to identify common B-cellepitopes that could be responsible for IgE-mediated cross- and autoreactivity, which


is often observed in patients su�ering from chronic atopic diseases. Since there is nostructural information available on fungal cyclophilins and thioredoxins, the primarygoal was to solve the crystal structure of these proteins.Cyclophilin from A. fumigatus and M. sympodialis as well as thioredoxin from M. sympo-dialis and T. aestivum (wheat) were cloned, expressed and crystallised. The structures ofboth cyclophilins as well as the M. sympodialis thioredoxin were solved and compared ona structural level to the human homologue and on a sequential level to other fungal homo-logues. This study allowed the characterisation and identi�cation of common, conservedsurface patches, which might be responsible for IgE mediated cross- and autoreactivity.Mutational studies might answer the question whether the de�ned conserved patchesreally represent immunodominant cross-reactive epitopes. Knowledge of cross-reactivitiesbetween di�erent allergenic sources will allow to reduce the number of allergens requiredfor the diagnosis and therapy of allergic diseases. Elucidation of the crystal structure offungal allergens could lead to a better understanding of the observed cross-reactivities onmolecular level. In the future, the modi�cation of B-cell epitopes could result in a saferuse of the modi�ed allergen in immunotherapy with reduced side e�ects.The crystal structure of A. fumigatus cyclophilin revealed a big surprise. It showeddimerisation by 3D domain swapping of two central β-strands, which is completely un-known to occur in homologous cyclophilins. This very interesting �nding gave the op-portunity to study the rare phenomenon of 3D domain swapping and to address thebiological relevance of the dimerised A. fumigatus cyclophilin.

Bibliography[1] J. Ring. Angewandte Allergologie. MMV Medizin, Munich, 2nd edition, 1988.

[2] S. J. Ono. Molecular genetics of allergic diseases. Annu. Rev. Immunol., 18:347�66, 2000.

[3] A. Daser, N. Meissner, U. Herz, and H. Renz. Role and modulation of T-cell cytokines in allergy.Curr. Opin. Immunol., 7(6):762�70, 1995.

[4] D. B. Corry and F. Kheradmand. Induction and regulation of the IgE response. Nature, 402(6760Suppl):B18�23, 1999.

[5] B. J. Sutton and H. J. Gould. The human IgE network. Nature, 366(6454):421�8, 1993.

[6] A. K. Abbas, A. H. Lichtman, and J. S. Pober. Cellular and molecular immunology. W.B. SaundersCompany, Philadelphia, 3rd edition, 1997.

[7] S. Flückiger. Cloning, production, characterisation, and structural aspects of cross-reactive aller-gens. PhD thesis, No. 14274, ETH, Zürich, 2001.

[8] A. J. Neuman Taylor. ABC of allergies. Asthma and allergy. BMJ, 316(7136):997�9, 1998.

BIBLIOGRAPHY 37

[9] G. D'Amato and F. T. Spieksma. Aerobiologic and clinical aspects of mould allergy in Europe.Allergy, 50(11):870�7, 1995.

[10] W. E. Horner, A. Helbling, J. E. Salvaggio, and S. B. Lehrer. Fungal allergens. Clin. Microbiol.Rev., 8(2):161�79, 1995.

[11] J. Lacey. The aerobiology of conidial fungi. In G.T. Cole and B. Kendrick, editors, Biology ofconidial fungi, pages 373�416. Academic Press, Inc., New York, 1981.

[12] S. B. Lehrer, L. Aukrust, and J. E. Salvaggio. Respiratory allergy induced by fungi. Clin. ChestMed., 4(1):23�41, 1983.

[13] S. C. Bischo�, J. H. Mayer, and M. P. Manns. Allergy and the gut. Int. Arch. Allergy Immunol.,121(4):270�83, 2000.

[14] R. C. Aalberse. Structural biology of allergens. J. Allergy Clin. Immunol., 106(2):228�38, 2000.

[15] C. Mayer, S. Hemmann, A. Faith, K. Blaser, and R. Crameri. Cloning, production, characterizationand IgE cross-reactivity of di�erent manganese superoxide dismutases in individuals sensitized toAspergillus fumigatus. Int. Arch. Allergy Immunol., 113(1-3):213�5, 1997.

[16] R. Valenta, M. Duchene, C. Ebner, P. Valent, C. Sillaber, P. Deviller, F. Ferreira, M. Tejkl,H. Edelmann, D. Kraft, et al. Pro�lins constitute a novel family of functional plant pan-allergens.J. Exp. Med., 175(2):377�85, 1992.

[17] W. G. Laver, G. M. Air, R. G. Webster, and S. J. Smith-Gill. Epitopes on protein antigens:misconceptions and realities. Cell, 61(4):553�6, 1990.

[18] J. Novotny. Protein antigenicity: a thermodynamic approach. Mol. Immunol., 28(3):201�7, 1991.

[19] D. R. Davies and G. H. Cohen. Interactions of protein antigens with antibodies. Proc. Natl. Acad.Sci. USA, 93(1):7�12, 1996.

[20] T. N. Bhat, G. A. Bentley, G. Boulot, M. I. Greene, D. Tello, W. Dall'Acqua, H. Souchon, F. P.Schwarz, R. A. Mariuzza, and R. J. Poljak. Bound water molecules and conformational stabilizationhelp mediate an antigen-antibody association. Proc. Natl. Acad. Sci. USA, 91(3):1089�93, 1994.

[21] V. Chitarra, P. M. Alzari, G. A. Bentley, T. N. Bhat, J. L. Eisele, A. Houdusse, J. Lescar,H. Souchon, and R. J. Poljak. Three-dimensional structure of a heteroclitic antigen-antibodycross-reaction complex. Proc. Natl. Acad. Sci. USA, 90(16):7711�5, 1993.

[22] K. Harata. X-ray structure of a monoclinic form of hen egg-white lysozyme crystallized at 313 K.Comparison of two independent molecules. Acta Cryst., D50(Pt 3):250�7, 1994.

[23] B. C. Cunningham and J. A. Wells. Comparison of a structural and a functional epitope. J. Mol.Biol., 234(3):554�63, 1993.

[24] J. A. Wells. Binding in the growth hormone receptor complex. Proc. Natl. Acad. Sci. USA,93(1):1�6, 1996.

[25] T. Clackson and J. A. Wells. A hot spot of binding energy in a hormone-receptor interface. Science,267(5196):383�6, 1995.

[26] S. Chacko, E. Silverton, L. Kam-Morgan, S. Smith-Gill, G. Cohen, and D. Davies. Structureof an antibody-lysozyme complex: unexpected e�ect of a conservative mutation. J. Mol. Biol.,245(3):261�74, 1995.


[27] W. R. Tulip, J. N. Varghese, R. G. Webster, W. G. Laver, and P. M. Colman. Crystal structuresof two mutant neuraminidase-antibody complexes with amino acid substitutions in the interface.J. Mol. Biol., 227(1):149�59, 1992.

[28] O. Mirza, A. Henriksen, H. Ipsen, J. N. Larsen, M. Wissenbach, M. D. Spangfort, and M. Gajhede.Dominant epitopes and allergic cross-reactivity: complex formation between a Fab fragment of amonoclonal murine IgG antibody and the major allergen from birch pollen Bet v 1. J. Immunol.,165(1):331�8, 2000.

[29] M. Gajhede, P. Osmark, F. M. Poulsen, H. Ipsen, J. N. Larsen, R. J. Joost van Neerven, C. Schou,H. Lowenstein, and M. D. Spangfort. X-ray and NMR structure of Bet v 1, the origin of birchpollen allergy. Nat. Struct. Biol., 3(12):1040�5, 1996.

[30] J. Bousquet, R. Lockey, and H. J. Malling. Allergen immunotherapy: therapeutic vaccines forallergic diseases. A WHO position paper. J. Allergy Clin. Immunol., 102(4 Pt 1):558�62, 1998.

[31] R. Valenta. The future of antigen-speci�c immunotherapy of allergy. Nat. Rev. Immunol., 2(6):446�53, 2002.

[32] A. Galat. Peptidylprolyl cis/trans isomerases (immunophilins): biological diversity-targets-functions. Curr. Top. Med. Chem., 3(12):1315�47, 2003.

[33] R. E. Handschumacher, M. W. Harding, J. Rice, R. J. Drugge, and D. W. Speicher. Cyclophilin:a speci�c cytosolic binding protein for cyclosporin A. Science, 226(4674):544�7, 1984.

[34] G. Fischer, B. Wittmann-Liebold, K. Lang, T. Kiefhaber, and F. X. Schmid. Cyclophilin andpeptidyl-prolyl cis-trans isomerase are probably identical proteins. Nature, 337(6206):476�8, 1989.

[35] N. Takahashi, T. Hayano, and M. Suzuki. Peptidyl-prolyl cis-trans isomerase is the cyclosporinA-binding protein cyclophilin. Nature, 337(6206):473�5, 1989.

[36] P. Taylor, J. Dornan, A. Carrello, R. F. Minchin, T. Ratajczak, and M. D. Walkinshaw. Twostructures of cyclophilin 40: folding and �delity in the TPR domains. Structure, 9(5):431�8, 2001.

[37] H. Ke and Q. Huai. Crystal structures of cyclophilin and its partners. Front. Biosci., 9:2285�96,2004.

[38] H. M. Ke, L. D. Zydowsky, J. Liu, and C. T. Walsh. Crystal structure of recombinant humanT-cell cyclophilin A at 2.5 Å resolution. Proc. Natl. Acad. Sci. USA, 88(21):9483�7, 1991.

[39] J. Kallen, C. Spitzfaden, M. G. Zurini, G. Wider, H. Widmer, K. Wuthrich, and M. D. Walkin-shaw. Structure of human cyclophilin and its binding site for cyclosporin A determined by X-raycrystallography and NMR spectroscopy. Nature, 353(6341):276�9, 1991.

[40] Y. Zhao and H. Ke. Crystal structure implies that cyclophilin predominantly catalyzes the transto cis isomerization. Biochemistry, 35(23):7356�61, 1996.

[41] R. L. Stein. Mechanism of enzymatic and nonenzymatic prolyl cis-trans isomerization. Adv.Protein Chem., 44:1�24, 1993.

[42] S. L. Schreiber. Chemistry and biology of the immunophilins and their immunosuppressive ligands.Science, 251(4991):283�7, 1991.

[43] R. K. Harrison and R. L. Stein. Mechanistic studies of peptidyl prolyl cis-trans isomerase: evidencefor catalysis by distortion. Biochemistry, 29(7):1684�9, 1990.

BIBLIOGRAPHY 39

[44] J. L. Kofron, P. Kuzmic, V. Kishore, E. Colon-Bonilla, and D. H. Rich. Determination of ki-netic constants for peptidyl prolyl cis-trans isomerases by an improved spectrophotometric assay.Biochemistry, 30(25):6127�34, 1991.

[45] S. Hur and T. C. Bruice. The mechanism of cis-trans isomerization of prolyl peptides by cyclophilin.J. Am. Chem. Soc., 124(25):7303�13, 2002.

[46] C. B. Klee, G. F. Draetta, and M. J. Hubbard. Calcineurin. Adv. Enzymol. Relat. Areas Mol.Biol., 61:149�200, 1988.

[47] J. Liu, J. D. Farmer, Jr., W. S. Lane, J. Friedman, I. Weissman, and S. L. Schreiber. Calcineurinis a common target of cyclophilin-cyclosporin A and FKBP-FK506 complexes. Cell, 66(4):807�15,1991.

[48] J. Friedman and I. Weissman. Two cytoplasmic candidates for immunophilin action are revealedby a�nity for a new cyclophilin: one in the presence and one in the absence of CsA. Cell,66(4):799�806, 1991.

[49] J. Jain, P. G. McCa�rey, Z. Miner, T. K. Kerppola, J. N. Lambert, G. L. Verdine, T. Curran, andA. Rao. The T-cell transcription factor NFATp is a substrate for calcineurin and interacts withFos and Jun. Nature, 365(6444):352�5, 1993.

[50] N. A. Clipstone and G. R. Crabtree. Identi�cation of calcineurin as a key signalling enzyme inT-lymphocyte activation. Nature, 357(6380):695�7, 1992.

[51] H. R. Gelderblom. Assembly and morphology of HIV: potential e�ect of structure on viral function.Aids, 5(6):617�37, 1991.

[52] D. Baltimore. The enigma of HIV infection. Cell, 82(2):175�6, 1995.

[53] J. Luban, K. L. Bossolt, E. K. Franke, G. V. Kalpana, and S. P. Go�. Human immunode�ciencyvirus type 1 Gag protein binds to cyclophilins A and B. Cell, 73(6):1067�78, 1993.

[54] E. K. Franke, H. E. Yuan, and J. Luban. Speci�c incorporation of cyclophilin A into HIV-1 virions.Nature, 372(6504):359�62, 1994.

[55] M. Thali, A. Bukovsky, E. Kondo, B. Rosenwirth, C. T. Walsh, J. Sodroski, and H. G. Gottlinger.Functional association of cyclophilin A with HIV-1 virions. Nature, 372(6504):363�5, 1994.

[56] T. R. Gamble, F. F. Vajdos, S. Yoo, D. K. Worthylake, M. Houseweart, W. I. Sundquist, andC. P. Hill. Crystal structure of human cyclophilin A bound to the amino-terminal domain ofHIV-1 capsid. Cell, 87(7):1285�94, 1996.

[57] T. C. Laurent, E. C. Moore, and P. Reichard. Enzymatic synthesis of deoxyribonucleotides. IV.Isolation and characterization of thioredoxin, the hydrogen donor from Escherichia coli B. J. Biol.Chem., 239:3436�44, 1964.

[58] A. Holmgren. Thioredoxin. Annu. Rev. Biochem., 54:237�71, 1985.

[59] H. Nakamura, K. Nakamura, and J. Yodoi. Redox regulation of cellular activation. Annu. Rev.Immunol., 15:351�69, 1997.

[60] H. Follmann and I. Haberlein. Thioredoxins: universal, yet speci�c thiol-disul�de redox cofactors.Biofactors, 5(3):147�56, 1995.


[61] G. Powis and W. R. Montfort. Properties and biological activities of thioredoxins. Annu. Rev.Pharmacol. Toxicol., 41:261�95, 2001.

[62] G. B. Kallis and A. Holmgren. Di�erential reactivity of the functional sulfhydryl groups of cysteine-32 and cysteine-35 present in the reduced form of thioredoxin from Escherichia coli . J. Biol. Chem.,255(21):10261�5, 1980.

[63] D. M. LeMaster. Structural determinants of the catalytic reactivity of the buried cysteine ofEscherichia coli thioredoxin. Biochemistry, 35(47):14876�81, 1996.

[64] A. Holmgren. Thioredoxin structure and mechanism: conformational changes on oxidation of theactive-site sulfhydryls to a disul�de. Structure, 3(3):239�43, 1995.

[65] J. Qin, G. M. Clore, and A. M. Gronenborn. The high-resolution three-dimensional solutionstructures of the oxidized and reduced states of human thioredoxin. Structure, 2(6):503�22, 1994.

[66] A. Weichsel, J. R. Gasdaska, G. Powis, andW. R. Montfort. Crystal structures of reduced, oxidized,and mutated human thioredoxins: evidence for a regulatory homodimer. Structure, 4(6):735�51,1996.

[67] H. Eklund, F. K. Gleason, and A. Holmgren. Structural and functional relations among thiore-doxins of di�erent species. Proteins, 11(1):13�28, 1991.

[68] J. R. Gasdaska, D. L. Kirkpatrick, W. Montfort, M. Kuperus, S. R. Hill, M. Berggren, andG. Powis. Oxidative inactivation of thioredoxin as a cellular growth factor and protection bya Cys73->Ser mutation. Biochem. Pharmacol., 52(11):1741�7, 1996.

[69] J. F. Andersen, D. A. Sanders, J. R. Gasdaska, A. Weichsel, G. Powis, and W. R. Montfort.Human thioredoxin homodimers: regulation by pH, role of aspartate 60, and crystal structure ofthe aspartate 60 -> asparagine mutant. Biochemistry, 36(46):13979�88, 1997.

[70] V. Menchise, C. Corbier, C. Didierjean, M. Saviano, E. Benedetti, J. P. Jacquot, and A. Aubry.Crystal structure of the wild-type and D30A mutant thioredoxin H of Chlamydomonas reinhardtiiand implications for the catalytic mechanism. Biochem. J., 359(Pt 1):65�75, 2001.

[71] M. C. Wahl, A. Irmler, B. Hecker, R. H. Schirmer, and K. Becker. Comparative structural analysisof oxidized and reduced thioredoxin from Drosophila melanogaster . J. Mol. Biol., 345(5):1119�30,2005.

[72] N. Wakasugi, Y. Tagaya, H. Wakasugi, A. Mitsui, M. Maeda, J. Yodoi, and T. Tursz. Adult T-cellleukemia-derived factor/thioredoxin, produced by both human T-lymphotropic virus type I- andEpstein-Barr virus-transformed lymphocytes, acts as an autocrine growth factor and synergizeswith interleukin 1 and interleukin 2. Proc. Natl. Acad. Sci. USA, 87(21):8282�6, 1990.

[73] J. E. Oblong, M. Berggren, P. Y. Gasdaska, and G. Powis. Site-directed mutagenesis of active sitecysteines in human thioredoxin produces competitive inhibitors of human thioredoxin reductaseand elimination of mitogenic properties of thioredoxin. J. Biol. Chem., 269(16):11714�20, 1994.

[74] H. Nakamura, H. Masutani, Y. Tagaya, A. Yamauchi, T. Inamoto, Y. Nanbu, S. Fujii, K. Ozawa,and J. Yodoi. Expression and growth-promoting e�ect of adult T-cell leukemia-derived factor. Ahuman thioredoxin homologue in hepatocellular carcinoma. Cancer, 69(8):2091�7, 1992.

[75] J. R. Gasdaska, M. Berggren, and G. Powis. Cell growth stimulation by the redox protein thiore-doxin occurs by a novel helper mechanism. Cell Growth Di�er., 6(12):1643�50, 1995.

BIBLIOGRAPHY 41

[76] A. Spector, G. Z. Yan, R. R. Huang, M. J. McDermott, P. R. Gascoyne, and V. Pigiet. The e�ectof H2O2 upon thioredoxin-enriched lens epithelial cells. J. Biol. Chem., 263(10):4984�90, 1988.

[77] H. Z. Chae, S. J. Chung, and S. G. Rhee. Thioredoxin-dependent peroxide reductase from yeast.J. Biol. Chem., 269(44):27670�8, 1994.

[78] T. Hayashi, Y. Ueno, and T. Okamoto. Oxidoreductive regulation of nuclear factor kappa B.Involvement of a cellular reducing catalyst thioredoxin. J. Biol. Chem., 268(15):11380�8, 1993.

[79] K. A. Hutchison, G. Matic, S. Meshinchi, E. H. Bresnick, and W. B. Pratt. Redox manipulation ofDNA binding activity and BuGR epitope reactivity of the glucocorticoid receptor. J. Biol. Chem.,266(16):10505�9, 1991.

[80] H. Schenk, M. Klein, W. Erdbrugger, W. Droge, and K. Schulze-Ostho�. Distinct e�ects ofthioredoxin and antioxidants on the activation of transcription factors NF-kappa B and AP-1.Proc. Natl. Acad. Sci. USA, 91(5):1672�6, 1994.

[81] S. Iwata, T. Hori, N. Sato, K. Hirota, T. Sasada, A. Mitsui, T. Hirakawa, and J. Yodoi. Adult T cellleukemia (ATL)-derived factor/human thioredoxin prevents apoptosis of lymphoid cells inducedby L-cystine and glutathione depletion: possible involvement of thiol-mediated redox regulation inapoptosis caused by pro-oxidant state. J. Immunol., 158(7):3108�17, 1997.

[82] A. Gallegos, J. Ra�el, A.K. Bhattacharyya, and G. Powis. Increased expression of thioredoxin inhuman primary and metastatic colon cancer. Proc. Am. Assoc. Cancer Res., 41:189, 2000.

[83] T. M. Grogan, C. Fenoglio-Prieser, R. Zeheb, W. Bellamy, Y. Frutiger, E. Vela, G. Stemmerman,J. Macdonald, L. Richter, A. Gallegos, and G. Powis. Thioredoxin, a putative oncogene product,is overexpressed in gastric carcinoma and associated with increased proliferation and increased cellsurvival. Hum. Pathol., 31(4):475�81, 2000.

[84] F. Rousseau, J. W. Schymkowitz, and L. S. Itzhaki. The unfolding story of three-dimensionaldomain swapping. Structure, 11(3):243�51, 2003.

[85] Y. Liu and D. Eisenberg. 3D domain swapping: as domains continue to swap. Protein Sci.,11(6):1285�99, 2002.

[86] M. J. Bennett and D. Eisenberg. The evolving role of 3D domain swapping in proteins. Structure,12(8):1339�41, 2004.

[87] M. J. Bennett, S. Choe, and D. Eisenberg. Domain swapping: entangling alliances between pro-teins. Proc. Natl. Acad. Sci. USA, 91(8):3127�31, 1994.

[88] M. J. Bennett, M. P. Schlunegger, and D. Eisenberg. 3D domain swapping: a mechanism foroligomer assembly. Protein Sci., 4(12):2455�68, 1995.

[89] A. A. Kortt, R. L. Malby, J. B. Caldwell, L. C. Gruen, N. Ivancic, M. C. Lawrence, G. J. Howlett,R. G. Webster, P. J. Hudson, and P. M. Colman. Recombinant anti-sialidase single-chain variablefragment antibody. Characterization, formation of dimer and higher-molecular-mass multimersand the solution of the crystal structure of the single-chain variable fragment/sialidase complex.Eur. J. Biochem., 221(1):151�7, 1994.

[90] Y. Liu, P. J. Hart, M. P. Schlunegger, and D. Eisenberg. The crystal structure of a 3D domain-swapped dimer of RNase A at a 2.1-Å resolution. Proc. Natl. Acad. Sci. USA, 95(7):3437�42,1998.


[91] S. M. Green, A. G. Gittis, A. K. Meeker, and E. E. Lattman. One-step evolution of a dimer froma monomeric protein. Nat. Struct. Biol., 2(9):746�51, 1995.

[92] N. Khazanovich, K. Bateman, M. Chernaia, M. Michalak, and M. James. Crystal structure of theyeast cell-cycle control protein, p13suc1, in a strand-exchanged dimer. Structure, 4(3):299�309,1996.

[93] M. P. Schlunegger, M. J. Bennett, and D. Eisenberg. Oligomer formation by 3D domain swapping:a model for protein assembly and misassembly. Adv. Protein Chem., 50:61�122, 1997.

[94] K. Diederichs, S. Jacques, T. Boone, and P. A. Karplus. Low-resolution structure of recombinanthuman granulocyte-macrophage colony stimulating factor. J. Mol. Biol., 221(1):55�60, 1991.

[95] M. V. Milburn, A. M. Hassell, M. H. Lambert, S. R. Jordan, A. E. Proudfoot, P. Graber, andT. N. Wells. A novel dimer con�guration revealed by the crystal structure at 2.4 Å resolution ofhuman interleukin-5. Nature, 363(6425):172�6, 1993.

[96] K. J. Knaus, M. Morillas, W. Swietnicki, M. Malone, W. K. Surewicz, and V. C. Yee. Crystalstructure of the human prion protein reveals a mechanism for oligomerization. Nat. Struct. Biol.,8(9):770�4, 2001.

[97] H. Mizuno, Z. Fujimoto, M. Koizumi, H. Kano, H. Atoda, and T. Morita. Structure of coagulationfactors IX/X-binding protein, a heterodimer of C-type lectin domains. Nat. Struct. Biol., 4(6):438�41, 1997.

[98] Y. Liu, G. Gotte, M. Libonati, and D. Eisenberg. A domain-swapped RNase A dimer with impli-cations for amyloid formation. Nat. Struct. Biol., 8(3):211�4, 2001.

[99] X. Zhao, S. Gha�ari, H. Lodish, V. N. Malashkevich, and P. S. Kim. Structure of the Bcr-Abloncoprotein oligomerization domain. Nat. Struct. Biol., 9(2):117�20, 2002.

[100] M. Juy, F. Penin, A. Favier, A. Galinier, R. Montserret, R. Haser, J. Deutscher, and A. Bock-mann. Dimerization of Crh by reversible 3D domain swapping induces structural adjustments toits monomeric homologue Hpr. J. Mol. Biol., 332(4):767�76, 2003.

[101] G. C. Ireton, G. McDermott, M. E. Black, and B. L. Stoddard. The structure of Escherichia colicytosine deaminase. J. Mol. Biol., 315(4):687�97, 2002.

[102] T. Izard and N. C. Blackwell. Crystal structures of the metal-dependent 2-dehydro-3-deoxy-galactarate aldolase suggest a novel reaction mechanism. EMBO J., 19(15):3849�56, 2000.

[103] J. Liang, S. R. Ja�rey, W. Guo, S. H. Snyder, and J. Clardy. Structure of the PIN/LC8 dimerwith a bound peptide. Nat. Struct. Biol., 6(8):735�40, 1999.

[104] S. Thorell, M. Schurmann, G. A. Sprenger, and G. Schneider. Crystal structure of decamericfructose-6-phosphate aldolase from Escherichia coli reveals inter-subunit helix swapping as a struc-tural basis for assembly di�erences in the transaldolase family. J. Mol. Biol., 319(1):161�71, 2002.

[105] P. Paaventhan, J. S. Joseph, S. V. Seow, S. Vaday, H. Robinson, K. Y. Chua, and P. R. Kolatkar.A 1.7Å structure of Fve, a member of the new fungal immunomodulatory protein family. J. Mol.Biol., 332(2):461�70, 2003.

BIBLIOGRAPHY 43

[106] B. Kauppi, C. Jakob, M. Farnegardh, J. Yang, H. Ahola, M. Alarcon, K. Calles, O. Engstrom,J. Harlan, S. Muchmore, A. K. Ramqvist, S. Thorell, L. Ohman, J. Greer, J. A. Gustafsson,J. Carlstedt-Duke, and M. Carlquist. The three-dimensional structures of antagonistic and ago-nistic forms of the glucocorticoid receptor ligand-binding domain: RU-486 induces a transconfor-mation that leads to active antagonism. J. Biol. Chem., 278(25):22748�54, 2003.

[107] S. H. Liaw, Y. J. Chang, C. T. Lai, H. C. Chang, and G. G. Chang. Crystal structure of Bacillussubtilis guanine deaminase: the �rst domain-swapped structure in the cytidine deaminase super-family. J. Biol. Chem., 279(34):35479�85, 2004.

[108] K. K. Kim, R. Kim, and S. H. Kim. Crystal structure of a small heat-shock protein. Nature,394(6693):595�9, 1998.

[109] R. L. van Montfort, E. Basha, K. L. Friedrich, C. Slingsby, and E. Vierling. Crystal structure andassembly of a eukaryotic small heat shock protein. Nat. Struct. Biol., 8(12):1025�30, 2001.

[110] J. A. Barbosa, J. Sivaraman, Y. Li, R. Larocque, A. Matte, J. D. Schrag, and M. Cygler. Mecha-nism of action and NAD+-binding mode revealed by the crystal structure of L-histidinol dehydro-genase. Proc. Natl. Acad. Sci. USA, 99(4):1859�64, 2002.

[111] K. Britton, S. Langridge, P. J. Baker, K. Weeradechapon, S. E. Sedelnikova, J. R. De Lucas, D. W.Rice, and G. Turner. The crystal structure and active site location of isocitrate lyase from thefungus Aspergillus nidulans. Structure Fold. Des., 8(4):349�62, 2000.

[112] N. Koon, C. J. Squire, and E. N. Baker. Crystal structure of LeuA fromMycobacterium tuberculosis,a key enzyme in leucine biosynthesis. Proc. Natl. Acad. Sci. USA, 101(22):8295�300, 2004.

[113] B. N. Chaudhuri, M. R. Sawaya, C. Y. Kim, G. S. Waldo, M. S. Park, T. C. Terwilliger, andT. O. Yeates. The crystal structure of the �rst enzyme in the pantothenate biosynthetic pathway,ketopantoate hydroxymethyltransferase, from M. tuberculosis. Structure, 11(7):753�64, 2003.

[114] M. Sundaramoorthy, M. Meiyappan, P. Todd, and B. G. Hudson. Crystal structure of NC1domains. Structural basis for type IV collagen assembly in basement membranes. J. Biol. Chem.,277(34):31142�53, 2002.

[115] R. Guan, E. L. Malchiodi, Q. Wang, P. Schuck, and R. A. Mariuzza. Crystal structure of theC-terminal peptidoglycan-binding domain of human peptidoglycan recognition protein I alpha.J. Biol. Chem., 279(30):31873�82, 2004.

[116] W. Tempel, Z. J. Liu, F. D. Schubot, A. Shah, M. V. Weinberg, F. E. Jenney, Jr., W. B. Arendall,M. W. Adams, J. S. Richardson, D. C. Richardson, J. P. Rose, and B. C. Wang. Structural ge-nomics of Pyrococcus furiosus: X-ray crystallography reveals 3D domain swapping in rubrerythrin.Proteins, 57(4):878�82, 2004.

[117] S. Fushinobu, H. Shoun, and T. Wakagi. Crystal structure of sulerythrin, a rubrerythrin-likeprotein from a strictly aerobic archaeon, Sulfolobus tokodaii strain 7, shows unexpected domainswapping. Biochemistry, 42(40):11707�15, 2003.

[118] S. Tranier, C. Iobbi-Nivol, C. Birck, M. Ilbert, I. Mortier-Barriere, V. Mejean, and J. P. Samama.A novel protein fold and extreme domain swapping in the dimeric TorD chaperone from Shewanellamassilia. Structure, 11(2):165�74, 2003.

[119] S. Radaev, M. Kattah, B. Rostro, M. Colonna, and P. D. Sun. Crystal structure of the humanmyeloid cell activating receptor TREM-1. Structure, 11(12):1527�35, 2003.


[120] S. Spinelli, A. Desmyter, L. Frenken, T. Verrips, M. Tegoni, and C. Cambillau. Domain swappingof a llama VHH domain builds a crystal-wide beta-sheet structure. FEBS Lett., 564(1-2):35�40,2004.

[121] A. P. Saint-Jean, K. R. Phillips, D. J. Creighton, and M. J. Stone. Active monomeric and dimericforms of Pseudomonas putida glyoxalase I: evidence for 3D domain swapping. Biochemistry,37(29):10345�53, 1998.

[122] L. Vitagliano, S. Adinol�, F. Sica, A. Merlino, A. Zagari, and L. Mazzarella. A potential allostericsubsite generated by domain swapping in bovine seminal ribonuclease. J. Mol. Biol., 293(3):569�77,1999.

[123] P. R. Gouldson, C. Higgs, R. E. Smith, M. K. Dean, G. V. Gkoutos, and C. A. Reynolds. Dimer-ization and domain swapping in G-protein-coupled receptors: a computational study. Neuropsy-chopharmacology, 23(4 Suppl):S60�77, 2000.

[124] A. L. Fink. Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold. Des.,3(1):R9�23, 1998.

[125] F. E. Cohen and S. B. Prusiner. Pathologic conformations of prion proteins. Annu. Rev. Biochem.,67:793�819, 1998.

[126] N. Sinha, C. J. Tsai, and R. Nussinov. A proposed structural model for amyloid �bril elongation:domain swapping forms an interdigitating beta-structure polymer. Protein Eng., 14(2):93�103,2001.

[127] M. Silow, Y. J. Tan, A. R. Fersht, and M. Oliveberg. Formation of short-lived protein aggregatesdirectly from the coil in two-state folding. Biochemistry, 38(40):13006�12, 1999.

[128] C. Ganesh, F. N. Zaidi, J. B. Udgaonkar, and R. Varadarajan. Reversible formation of on-pathwaymacroscopic aggregates during the folding of maltose binding protein. Protein Sci., 10(8):1635�44,2001.

[129] B. Mannervik, A. D. Cameron, E. Fernandez, A. Gustafsson, L. O. Hansson, P. Jemth, F. Jiang,T. A. Jones, A. K. Larsson, L. O. Nilsson, B. Olin, P. L. Pettersson, M. Ridderstrom, G. Stenberg,and M. Widersten. An evolutionary approach to the design of glutathione-linked enzymes. Chem.Biol. Interact., 111-112:15�21, 1998.

[130] R. A. Staniforth, S. Giannini, L. D. Higgins, M. J. Conroy, A. M. Hounslow, R. Jerala, C. J.Craven, and J. P. Waltho. Three-dimensional domain swapping in the folded and molten-globulestates of cystatins, an amyloid-forming structural superfamily. EMBO J., 20(17):4774�81, 2001.

[131] R. Janowski, M. Kozak, E. Jankowska, Z. Grzonka, A. Grubb, M. Abrahamson, and M. Jaskol-ski. Human cystatin C, an amyloidogenic protein, dimerizes through three-dimensional domainswapping. Nat. Struct. Biol., 8(4):316�20, 2001.

[132] D. Y. Chirgadze, M. Demydchuk, M. Becker, S. Moran, and M. Paoli. Snapshot of protein structureevolution reveals conservation of functional dimerization through intertwined folding. Structure,12(8):1489�94, 2004.

Chapter 2

Analysis of the Cross-Reactivity and of the1.5 Å Crystal Structure of the Malasseziasympodialis Mala s 6 Allergen, a Member ofthe Cyclophilin Pan-Allergen Family

Andreas G. Glaser1,#, Andreas Limacher1,2,#, Sabine Flückiger1, AnnikaScheynius3, Leonardo Scapozza4, and Reto Crameri1,∗

1Swiss Institute of Allergy and Asthma Research (SIAF), 7270 Davos, Switzerland2Institute of Pharmaceutical Sciences, Department of Chemistry and Applied Biosciences,Swiss Federal Institute of Technology (ETH), 8093 Zurich, Switzerland3Dept. of Medicine, Clinical Allergy Research Unit, Karolinska Institutet, Stockholm,Sweden4Laboratoire de Chimie Thérapeutique, Section des Sciences Pharmaceutiques, Universitéde Genève, 1211 Genève 4, Switzerland#Equally contributing∗Correspondence: Reto Crameri, Swiss Institute of Allergy and Asthma Research (SIAF),Obere Strasse 22, CH-7270 Davos, Switzerland. Phone: +41 81 420 07 31, Fax: +41 81410 08 40, E-mail: [email protected]

Running title: Cross-reactivity and structure of Mala s 6

Key words: Allergy, IgE, Malassezia sympodialis, Aspergillus fumigatus, cyclophilin, crys-tal structure, autoreactivity

Submitted

45

46 CHAPTER 2. CROSS-REACTIVITY AND STRUCTURE OF MALA S 6

2.1 AbstractCyclophilins constitute a family of proteins involved in many essential cellular functions.They have been identi�ed also as a pan-allergen family able to elicit IgE-mediated hyper-sensitivity reactions. Moreover, it has been shown that human cyclophilins are recognisedby IgE from sera of patients sensitised to environmental cyclophilins. In this work, a newmember of the cyclophilin pan-allergen family, Asp f 27 from Aspergillus fumigatus, hasbeen cloned, expressed and showed to represent a highly cross-reactive structure. IgE-mediated autoreactivity to self antigens sharing sequence homology with allergens is aphenomenon often observed in chronic in�ammatory atopic disorders. Therefore, com-parison of the crystal structure of human proteins sharing homology to environmentalallergens should allow the identi�cation of structural similarities to explain IgE-mediatedautoreactivity. The three-dimensional structure of cyclophilin from Malassezia sympodi-alis (Mala s 6), a yeast involved in the pathogenesis of atopic eczema, has been determinedat 1.5 Å resolution by X-ray di�raction analysis. Crystals belong to space group P41212with unit cell dimensions of a = b = 71.99 Å and c = 106.18 Å. The structure wassolved by molecular replacement using the structure of human cyclophilin A as a searchmodel. The �nal re�ned model includes all 162 amino acids of Mala s 6, an active site-bound Ala-Pro dipeptide and 173 water molecules, with a crystallographic R-factor of14.3% and a free R-factor of 14.9%. Like other cyclophilins, M. sympodialis cyclophilinshows an overall structure consisting of an eight-stranded antiparallel β-barrel and twoα-helices covering the top and bottom of the barrel. We identi�ed conserved solvent-exposed amino acids in the fungal and the human crystal structure potentially involvedin the IgE-mediated cross-reactivity demonstrated in vitro and in vivo.

2.2 IntroductionCyclophilins (CyP) constitute a family of cytosolic proteins, which play a pivotal rolein protein folding through enzymatic catalysis of the peptidyl-prolyl cis-trans isomerisa-tion reaction (PPIase) [1, 2]. The primary structure of this important enzyme is highlyconserved among phylogenetically distant species indicating involvement of CyP in basiccellular functions [3, 4]. Belonging to the family of immunophilin, CyP binds the im-munosuppressive drug cyclosporin A (CsA) [5]. The complex of CyP with CsA binds andinhibits the protein phosphatase calcineurin [6], thus suppressing the signal transductionin T cells [7]. Therefore, CsA is one of the most important immunosuppressant drugsused for prevention of graft rejection after transplant surgery [8].

2.2. INTRODUCTION 47

CyP from Aspergillus fumigatus (Asp f 11 [9]), the etiologic agent identi�ed in the ma-jority of Aspergillus-related human diseases [10] and from the yeast Malassezia sympo-dialis (Mala s 6 [11]), a skin colonising yeast involved in the pathophysiology of atopiceczema [12], have been isolated as IgE-binding proteins from phage surface-displayedcDNA libraries [13]. Later, CyP was isolated as allergen also from birch pollen [14] andcarrots [15] demonstrating that this molecular structure can elicit IgE-mediated responsesin di�erent types of allergic disorders. Inhalation of environmental allergens from pol-lens, dust mites, animal dander and fungi is the most common cause of IgE-mediatedantibody responses in humans. The symptoms related to allergen exposure range fromrhinoconjunctivitis, atopic eczema, asthma and fatal anaphylaxis and the incidence ofallergic disease is dramatically increasing [16].Although the mechanisms leading to allergy are quite well understood [17, 18] and inspite of relevant progress in allergen cloning [13], our knowledge about the repertoire ofmolecular structures able to induce a switch to IgE production is still incomplete. Acommon hallmark of allergenic structures is their ability to induce cross-linking of high-a�nity FcεRI receptor-bound IgE on e�ector cells of sensitised individuals, thus causingimmediate release of anaphylactogenic mediators [18]. Although the 3D structures ofsome allergens are known [19], the question whether a protein exhibits special structuralcharacteristics responsible for its allergenicity is poorly understood.In contrast, elucidation of the 3D structures of allergens has essentially contributed to ourunderstanding of cross-reactivity among homologous structures derived from phylogenet-ically distant allergenic sources [20, 21]. Among these, manganese superoxide dismutase(MnSOD) of A. fumigatus [22] has been shown to cross-react with a wide variety ofMnSODs [23�26], including human MnSOD [27]. Cross-reactivity between allergens andclosely related homologous human antigens is often observed in allergic individuals suf-fering from chronic atopic diseases [27�31] and the availability of the 3D structure of agiven allergen and its human homologues allows a detailed study of the residues involvedin cross-reactivity [22]. These studies provide strong evidence for in vitro and in vivo hu-moral and cell-mediated IgE autoreactivity in patients su�ering from long lasting atopicdiseases [27, 30] potentially contributing to exacerbation and/or perpetuation of chronicallergic reactions [32]. Recently, up regulation of human MnSOD in eczematous skin ofpatients su�ering from atopic dermatitis has been shown, providing a direct link betweenIgE-sensitisation, cross-reactive structures and pathogenesis of the disease [33].Here we present the cloning, production and characterisation of Asp f 27, a new CyPfrom A. fumigatus, and the 3D structure of CyP from M. sympodialis (Mala s 6), alipophilic yeast involved in the pathogenesis of atopic eczema [12,33], the comparison of


the structure with the solved structure of human cyclophilin B [34] and the de�nition ofthe solvent-accessible residues shared by the two crystal structures. These amino acidresidues potentially involved in IgE-mediated cross-reactivity among cyclophilins o�er anexplanation for the IgE-mediated autoreactivity found in clinically distinct chronic atopicdiseases.

2.3 Experimental procedures

2.3.1 Cloning and production of cyclophilinsAsp f 11, the cyclophilin of A. fumigatus and human cyclophilin B were cloned and pro-duced as described [9]. A new cyclophilin of A. fumigatus formally termed Asp f 27 ac-cording to the recommendations of the international allergen nomenclature committee [35]was detected during a high-throughput screening program of cDNA libraries displayed onphage surface [36]. The gene encoding Asp f 27 was ampli�ed by PCR using the primers5' BamHI 5'-CGGCGGATCCATGGTTGTCAAGACTTTCTTC-3' and 3' HindIII 5'-GCCCAAGCTTACAGCTGACCACAGTCGG-3', subcloned as BamHI/HindIII frag-ment into pQE30 (Qiagen, Hilden, Germany) and transformed into E. coli M15 cellsfollowed by DNA sequence veri�cation. To produce recombinant proteins, E. coli M15cells were grown at 37◦C in LB medium to an OD600 of 0.6, induced with 1 mM IPTG, in-cubated at 30◦C for 16 h, harvested by centrifugation (6000 x g, 10 min, 4◦C) and storedat -20◦C. The cell pellet was resuspended in lysis bu�er (50 mM NaH2PO4, 300 mM NaCl,10 mM imidazole, pH 8.0) and lysed by French Press. Insoluble material was removedby centrifugation at 20,000 x g (20 min, 4◦C).His-tagged recombinant proteins were puri�ed by nickel a�nity chromatography using a5 ml HiTrap Chelating HP column (Amersham Pharmacia Biotech, Uppsala, Sweden).Proteins were eluted in a linear bu�er gradient (10 - 250 mM imidazole, 50 mM NaH2PO4,300 mM NaCl, pH 8.0) and dialysed against distilled water. Molecular size, purity andenzymatic activity of the recombinant CyP proteins were assessed as described [9].For crystallisation of Mala s 6, a thrombin-cleavable His-tagged protein was designed.The original Mala s 6 cDNA [11] was ampli�ed from the original clone by PCR using PfuTurbo DNA Polymerase (2.5 U/µl) (Stratagene, La Jolla, CA) with the primers 5' BamHI5'-CGGCGGATCCATGTCTAACGTTTTCTTCG-3' and 3' HindIII 5'-GCCCAAGCT-TTTAGCACACACCGCACTTG-3'. The PCR product was digested with BamHI andHindIII restriction endonucleases (NEB, Beverly, MA, USA), cleaned with QIAquick PCRpuri�cation kit (Qiagen) and ligated into a modi�ed pQE32 vector (Qiagen) containing

2.3. EXPERIMENTAL PROCEDURES 49

an N-terminal His6 tag followed by a thrombin cleavage site and a BamHI restrictionsite (HHHHHHLVPRGS). The ligation mixture was transformed into E. coli strain M15and the sequence of picked clones containing inserts of the correct size veri�ed by DNAsequencing. A correct clone was used to produce and purify His-tagged protein processedas before. The N-terminal His tag of Mala s 6 was cleaved o� by thrombin (20 Unitsper mg protein) in 300 mM NaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5 by incubation for30 h at 22◦C. The cleaved protein was further puri�ed by gel �ltration on a Superdex 75column (FPLC, Pharmacia, Uppsala, Sweden) equilibrated with 100 mM NaCl, 50 mMTris, pH 7.0. The eluted protein was diluted 1:10 with H2O, 2 mM DTT and concentratedto 10 mg/ml. Correct cleavage and purity were assessed by SDS-PAGE.

2.3.2 Subjects, routine assessments and skin testsSera from 40 patients su�ering from atopic eczema sensitised to M. sympodialis [12, 33]and from 40 patients sensitised to A. fumigatus [10, 27] selected according to clinicalhistory and immediate skin test reactivity to extracts were analysed together with serafrom 15 healthy controls. Allergen-speci�c IgE to extracts was quantitatively determinedby ImmunoCAP m3 (A. fumigatus) and m70 (M. sympodialis) using the Pharmacia CAPsystem (Pharmacia Diagnostics, Uppsala, Sweden) according to the package inserts. Skin-prick tests were performed with A. fumigatus extract purchased from ALK (Horsholm,Denmark) as commercially available skin prick test (SPT) solution and with an in houseM. sympodialis extract prepared as described [33]. Tests were considered positive if awheal surface > 9 mm2 surrounded by an erythema was induced in absence of reactionto the negative 0.9% saline control [37]. Intradermal skin tests with human recombinantCyP A, CyP B and Asp f 11 were performed as described [27, 30]. The study protocolwas carried out according to a clinical protocol approved by the ethical committee andall participants gave written informed consent after a full explanation of the proceduregiven individually before testing.

2.3.3 Immunoassays for IgE antibodies binding to cyclophilinsIgE-binding to recombinant CyP was determined by a standard direct solid phase ELISAin polystyrene microtiter plates (MaxisorpTM , Nunc, Roskilde, Denmark) coated andprocessed as described [9, 37]. Results were expressed as ELISA Units/ml (EU/ml) cal-ibrated against the absorbency of an in-house reference standard arbitrarily de�ned as100 EU/ml for each allergen tested [37]. Inhibition ELISA was performed using 1:10 di-luted patients' sera and BSA as a negative control as described [9]. Percentage inhibition


was calculated from the absorbency of the serial dilutions containing BSA in �uid phaseset as 0% inhibition.

2.3.4 IgE immunoblotsRecombinant CyP were separated by SDS-PAGE on gels (12%, Novex, San Diego, CA,USA) under denaturing, reducing conditions, transferred to HybondTM ECLTM nitro-cellulose membranes (Amersham, Little Chalfont, GB) and processed as described [9].Membranes were incubated with sera of patients su�ering either from A. fumigatus orM. sympodialis sensitisation and positive in IgE ELISA with Asp f 11 or Mala s 6 coatedto the solid phase. After incubation with mouse anti-human IgE mAb TN-142, spe-ci�c IgE-binding was detected with peroxidase-conjugated goat anti-mouse IgG (Dako,Copenhagen, Denmark) and bands visualised with SuperSignal R© West Pico Chemilumi-nescent Substrate (Pierce, Rockford, IL, USA) on Hyper�lmTM ECLTM (Amersham) asdescribed [9, 27].

2.3.5 Crystallisation and data collection of Mala s 6Crystallisation was performed using the hanging drop vapour di�usion method at 16◦C.The protein solution (6 mg/ml, 80 mM Ala-Pro, 2 mM DTT) was mixed in a 2:1 ratio withreservoir solution (1.5 M ammonium sulfate, 12% v/v glycerin, 0.1 M Tris pH 8.5). Thedrop was equilibrated against 500 µl of reservoir solution. After several months, a crystalgrew to a size of 300 x 300 x 250 µm and the drop solution had turned cryoprotectant.Therefore, the crystal was directly �ash-cooled in a stream of gaseous nitrogen of 100 K.A dataset was collected to 1.5 Å resolution on the synchrotron beamline X06SA at SwissLight Source (Villigen/CH) at 100 K. Data were processed and scaled with DENZO andSCALEPACK of the HKL program package [38]. The crystal belongs to the tetragonalspace group P41212 with cell parameters a=b=71.99 Å and c=106.18 Å (Table 2.1).

2.3.6 Structure determination and re�nementThe structure was solved by molecular replacement using MOLREP [39]. A polyalaninemodel of human CyP A (PDB code 2CPL) served as search model. One molecule waslocated in the asymmetric unit, which yields a Matthews coe�cient (VM) of 4.0 Å3/Daand a corresponding solvent content of 69.0%. Rigid body re�nement using REFMAC [40]as implemented in the CCP4 program suite [41], resulted in an R and Rfree of 46.2% and45.5%, respectively. Further re�nement was also performed with REFMAC, manual


Data CollectionX-ray source X06SA (SLS Villigen/CH)Wavelength (Å) 1.0001Space group P41212Unit cell axes a = b, c (Å) 71.99, 106.18Resolution (Å) 42.7-1.5 (1.55-1.50)Unique re�ections 45,175 (4438)Redundancy 14.2Completeness (%) 99.4 (100.0)Rsym (%) 6.1 (30.7)Average I/σ 26.9 (8.5)Re�nement StatisticsResolution (Å) 36.7-1.5Number of re�ections (working/test) 42,825 / 2272Number of atoms 1409Rcryst (%) 14.3Rfree (%) 14.9Mean B factor (Å2): All atoms 22.7Main chain atoms 19.6Side chain atoms 21.9Dipeptide atoms 23.2Solvent atoms 37.0

Rmsd bond lengths (Å) 0.015Rmsd bond angles (◦) 1.63

Table 2.1: Data collection and re�nement statistics. Numbers in parentheses are for the highestresolution shell. Rfree was calculated using a test set of 5%.

rebuilding and correction with XtalView [42]. 173 water molecules were introduced usingARP [43]. Final rounds of re�nement were carried out with individual anisotropic Bfactors. The side chain of Gln87 was modelled as double conformation with occupanciesof 0.5 each. Statistics from data collection and re�nement are provided in Table 2.1.The stereochemical quality of the �nal model was assessed with PROCHECK [44] andWHATCHECK [45]. Anisotropic validation was done with PARVATI [46]. Secondarystructure elements were assigned automatically with DSSP [47].

2.3.7 Calculation of the solvent-accessible areaSolvent-accessible surface areas were calculated with the program NACCESS [48], animplementation of the Lee and Richards solvent accessibility algorithm [49], using a probe


radius of 1.4 Å and a slice width of 0.01 Å. Ligands and water molecules were omittedin the calculation. The relative residue accessibility is the ratio of the accessible area ofa residue in the model to the accessible area of that residue in an extended Ala-X-Alatripeptide.

Accession numberStructural data is accessible from the PDB (ID code will be inserted after acceptance forpublication).

2.4 Results

2.4.1 Molecular cloning of Asp f 27 and production of recombi-nant cyclophilins

Screening of a phage surface displayed cDNA library of A. fumigatus yielded a vastvariety of clones carrying inserts encoding putative IgE-binding proteins [37]. Amongthese, 173 clones encoded Asp f 11, an already known allergen of the mould identi�ed asa member of the cyclophilin family [9]. Clone E6n01 coded for a variant of cyclophilinsharing 61% sequence identity with Asp f 11 at primary structure level. The cDNAcontained a 492 bp open reading frame, coding for a 163-amino acid protein with theconserved active site residues of cyclophilin (Figure 2.1) and a calculated molecular massof 17.74 kDa. The new allergen cannot be considered as an isoform of Asp f 11 accordingto the recommendations of the allergen nomenclature committee (www.allergen.org) andwas therefore termed Asp f 27. The sequence of Asp f 27 is available from the NBCIdatabase, accession number AJ937743. PCR ampli�ed DNA encoding the full lengthAsp f 27 was ligated into pQE30, expressed as N-terminal His-tagged protein in E. coliand puri�ed by Ni2+-NTA a�nity chromatography. The puri�ed protein migrated as asingle band of the expected size on 4-12% gradient SDS-PAGE as it was the case for Malas 6, Asp f 11 and human cyclophilin B produced and puri�ed in the same way (Figure2.2 A). All recombinant cyclophilins were able to catalyse the cis-trans isomerisation ofN -succinyl-Ala-Ala-Pro-Phe p-nitroanilide (data not shown) and enzymatic activity wasconsidered as a proof for native like folding [9].

2.4. RESULTS 53

== 1=== ==== 2==== ===== 1===== 3= = 4=

10 20 30 40 50 60 70 80

Mala s 6 ----MSNVFFDITK-----NGAPLGTIKFKLFDDVVPKTAANFRALCTGEKGFGYAGSHFHRVIPDFMLQGGDFTAGNGTGGKSIYGAKFADEN

Human CyPB GPKVTVKVYFDLRI-----GDEDVGRVIFGLFGKTVPKTVDNFVALATGEKGFGYKNSKFHRVIKDFMIQGGDFTRGDGTGGKSIYGERFPDEN

Asp f 11 ----MSQVFFDVEYAPVGTAETKVGRIVFNLFDKDVPKTAKNFRELCKRPAGEGYRESTFHRIIPNFMIQGGDFTRGNGTGGRSIYGDKFADEN

Asp f 27 ---MVVKTFFDITI-----DGQPAGRITFKLFDEVVPKTVENFRALCTGEKGFGYKGSSFHRIIPQFMLQGGDFTKGNGTGGKSIYGDRFPDEN

:.:**: * : * **.. ****. ** *.. * ** * ***:* :**:****** *:****:**** :*.***

= 5= = 6= 3 = == 7=== === 2=== === 8===10

90 100 110 120 130 140 150 160

Mala s 6 FQLKHNKPGLLSMANAGPNTNGSQFFITTVVTSWLDGKHVVFGEVID--GMNVVKAIEAEGS-GSGKPRS--RIEIAKCGVC-----------

Human CyPB FKLKHYGPGWVSMANAGKDTNGSQFFITTVKTAWLDGKHVVFGKVLE--GMEVVRKVESTKTDSRDKPLK--DVIIADCGKIEVEKPFAIAKE

Asp f 11 FSRKHDKKGILSMANAGPNTNGSQFFITTAVTSWLDGKHVVFGEVADEKSYSVVKEIEALGS-SSGSVRSNTRPKIVNCGEL-----------

Asp f 27 FQLKHDKPGLLSMANAGKNTNGSQFFITTVVTSWLDGAHVVFGEVED--GMDLVKKIESYGS-ASGTPKK--KITIADCGQL-----------

*. ** * :****** :**********. *:**** *****:* : . .:*: :*: : . .. . *..**

Figure 2.1: Structure alignment of Mala s 6 and human CyP B combined with a sequencealignment of Asp f 11 and Asp f 27. The sequence of human CyP B corresponds to the matureform, but missing 6 N-terminal amino acids. The top line shows the secondary structure assignedby DSSP [47], the second line shows the sequence numbering of Mala s 6. Asterisks indicateidentical, colons strongly similar and periods weakly similar amino acids in all sequences. Activesite residues are in bold. The two loops, which adopt a di�erent conformation in Mala s 6compared to human CyP B, are underlined. Residues that are identical and at least 50% orbetween 30% and 50% solvent-exposed in Mala s 6 and human CyP B are shown on a pink andblue background, respectively. Residues, which are also conserved in Asp f 11 or Asp f 27, arecoloured correspondingly.

2.4.2 Demonstration of IgE antibody responses to recombinantcyclophilins

The IgE-binding capacity of the recombinant cyclophilins was investigated by Westernblotting using serum pools of individuals sensitised to M. sympodialis and A. fumigatus(Figure 2.2 B). These results show that the cyclophilins of di�erent origin are cross-reactive and the speci�city of the antigen-antibody interaction was demonstrated byinhibition ELISAs (Figure 2.2 C). The prevalence of IgE-binding to the di�erent cy-clophilins was assessed by a standard solid phase ELISA in groups of patients with atopiceczema sensitised to M. sympodialis or patients su�ering from allergic bronchopulmonaryaspergillosis (ABPA) and therefore sensitised to A. fumigatus. Both conditions representlong lasting, chronic relapsing atopic diseases with a strong in�ammatory background.Results were expressed as arbitrary ELISA Units/ml (EU/ml) and considered positivewhen determinations exceeded three times the mean EU/ml value of a control group in-volving 15 healthy individuals regarded as background value for the ELISA determinationof allergen-speci�c IgE.The results reported in Figure 2.3 suggest a prevalence of about 50% and 75% to Asp


A C

0.01 0.1 1 10

0

25

50

75

100

Antigen in fluid phase [ M]

Inh

ibit

on

[%

]

B

Figure 2.2: Protein puri�cation, Western blot analysis and inhibition ELISA.(A) Recombinant proteins (5 µg) were separated on an SDS-PAGE and stained with Coomassieblue. Molecular mass standards (kDa) are indicated on Lane 1. Lane 2: Mala s 6, Lane 3:Asp f 27, Lane 4: Asp f 11, Lane 5: human CyP B.(B) Speci�c IgE-binding of recombinant cyclophilins analysed by Western blotting using serumpools of individuals sensitised to M. sympodialis (left) and A. fumigatus (right). Lane 1 and 5:Mala s 6, Lane 2 and 6: Asp f 27, Lane 3 and 7: Asp f 11, Lane 4 and 8: human CyP B.(C) Competitive inhibition of IgE-binding to recombinant human CyP B coated on a solid phase.Pooled serum from A. fumigatus-sensitised patients was preincubated with increasing amountsof recombinant human CyP B (blue), Mala s 6 (pink), Asp f 27 (light green), Asp f 11 (darkgreen) and BSA as negative control (black). Preincubated serum samples were transferred towells coated with recombinant human CyP B and IgE bound was analysed by ELISA.

2.4. RESULTS 55

A B

Figure 2.3: Serum IgE antibodies to recombinant cyclophilins. IgE binding to the allergenswas determined with an antigen-speci�c ELISA and compared to the binding of a referenceserum arbitrarily assigned to 100 EU/ml [36]. (A) Sensitisation to Mala s 6, Asp f 11, Asp f 27and human CyP B in a group of 40 atopic dermatitis patients sensitised to M. sympodialis.(B) Sensitisation to the same allergens in a group of 40 A. fumigatus-sensitised patients su�eringfrom ABPA. The lower level of assay sensitivity (hatched area) is indicated and corresponds tothree times the mean EU/ml values of a healthy control group of 15 individuals. Mean valuesare indicated by lines.

f 11 and Asp f 27 among patients su�ering from ABPA and atopic dermatitis (AD),respectively. The incidence of sensitisation against Mala s 6 was about 30% among both,patients su�ering from AD and ABPA. These results suggest, together with the resultsof the cross-inhibition ELISAs reported in Figure 2.2 C, an extended cross-reactivitybetween these homologous molecular structures. In accordance with results reportedearlier [9], patients sensitised to either of the environmental allergens also show cross-reactivity both in Western blot and in ELISA analyses with the human self antigen CyP B(Figure 2.2 B) now traceable back to structural features shown by comparison of the 3Dstructure of Mala s 6 solved during this work and the known 3D structure of humanCyP B (see below).The biological activity of human CyP A and CyP B was assessed by quantitative intra-dermal skin testing of 4 ABPA patients showing IgE antibodies against all cyclophilins,3 healthy controls and 3 ABPA patients with cyclophilin-speci�c IgE antibodies belowbackground (Table 2.2). These results show that positive skin tests were obtained using


Individual Histamine NaCl Asp f 11 hCyP A hCyP B IgE (EU/ml)a)

Asp f 11 Mala s 6 hCyP BBUb) 361 0 625 1215 1089 319 146 137MLb) 324 0 400 169 225 216 76 129Müb) 484 0 572 246 225 262 95 111LOb) 245 0 256 151 121 156 111 98BHc) 484 0 0 0 0 19 23 28KMc) 289 0 0 0 0 16 6 8BEc) 529 0 0 0 0 18 26 9GMd) 436 0 0 0 0 10 15 21ICd) 387 0 0 0 0 14 12 16SHd) 414 0 0 0 0 11 13 15

Table 2.2: Induction of immediate skin reactions with recombinant cyclophilins: Values are inmm2 calculated according to the formula [(D1+D2)/2]2 [36].a) Relative level of allergen speci�c IgE, cut-o� value 45 EU/ml (see Figure 2.3).b) ABPA patients sensitised to cyclophilin.c) ABPA patients not sensitised to cyclophilin.d) Healthy controls.

10−4 to 10−1 µg/ml recombinant Asp f 11, human CyP A or CyP B, and that all patientsshowing skin test reactivity also had detectable CyP-speci�c IgE levels in serum. In con-trast, non-allergic controls and ABPA patients without detectable serum IgE antibodiesto CyP were negative in skin tests using up to 10 µg/ml protein demonstrating a highspeci�city of the hypersensitivity responses in CyP-sensitised individuals (Table 2.2).

2.4.3 Overall structure of Mala s 6The X-ray structure of Mala s 6 was solved by molecular replacement and was re�ned to aresolution of 1.5 Å with R and Rfree values of 14.3% and 14.9%, respectively. 87.0% of thenon-glycine and non-proline residues have main chain dihedral angels in the most favouredregions of the Ramachandran plot, with the remaining ones located in the additionalallowed regions. Data collection and re�nement statistics are shown in Table 2.1. Thereis one monomer per asymmetric unit. This results in a Matthews coe�cient (VM) of4.0 Å3/Da and a corresponding solvent content of 69.0%. For such a high solvent content,the excellent resolution is remarkable. The monomeric state is also observed in solution,as shown in a gel �ltration experiment (data not shown).The �nal model consists of all the amino acids of Mala s 6 (aa 1-162), a dipeptide (Ala-Pro) bound to the active site, a glycerin molecule bound on the surface and 173 watermolecules. Like in all known cyclophilin structures, the fold consists of an eight-stranded

2.4. RESULTS 57

antiparallel β-barrel and two α-helices covering the top and the bottom of the barrel withan additional small 310-helix formed by Ser118, Trp119 and Leu120 (Figure 2.4 A).The two conserved cysteines in Mala s 6 could theoretically form a disul�de bond (Cys38and Cys159), however, in the structure they are present in the reduced form. The sulfuratoms are well de�ned in the electron density, are separated by 5.22 Å and are solventinaccessible. Simple rotation of the sulfur atoms about the Cα-Cβ cysteine side chainbonds towards each other would allow disul�de bond formation with a bond length ofabout 2 Å and chi torsion angles around 180◦. Potential roles of this disul�de bond couldbe the stabilisation of the protein in an oxidising environment or a signalling mechanismin response to oxidative stress, as hypothesised for C. elegans CyP 3 [50].

2.4.4 Active site of Mala s 6The dipeptide Ala-Pro binds into the active site, which is responsible for the cis-transisomerisation reaction. It is well de�ned in the electron density (Figure 2.4 B). The activesite lies in a hydrophobic pocket, which is formed by residues of the β-strands 3 to 6 andadjacent loops. The peptide bond of the ligand Ala-Pro is in the cis conformation, withan omega torsion angle of -0.2◦. The binding of the dipeptide involves hydrogen bondsand Van-der-Waals interactions with the active site residues Arg53, Phe58, Met59, Gln61,Ala99, Asn100, Phe111, Leu120 and His124 (Figure 2.4 B). The proline side chain sits inthe hydrophobic pocket, which is made up by the side chains of Phe58, Met59, Phe111,Leu120 and His124 of Mala s 6. The alanine of the dipeptide forms two hydrogen bondswith Asn100 and one with the water molecule Wat119. The C-terminal carboxyl groupof the proline forms two H-bonds with Arg53, one with Gln61 and one with Wat119.Wat119 forms an additional H-bond with Gln61.The active site residues are highly conserved among all cyclophilins. The binding modeof Ala-Pro to Mala s 6 is identical as in the human CyP A-Ala-Pro complex (PDB code1CYH). It involves the same residues and the same hydrogen bonding. For human CyP A,it has been questioned, whether Ala-Pro acts as a competitive inhibitor of the cis-transisomerisation function or as a substrate, which is subject to cis-trans isomerisation [51].With Caenorhabditis elegans CyP 3, Ala-Pro was shown to act as a weak inhibitor of thePPIase activity, with a measured Kd value of 23.3 mM [52], supporting the inhibitionhypothesis.


A

B

Figure 2.4: Overall structure and active site of Mala s 6.(A) Cartoon representation of Mala s 6. The fold consists of an eight-stranded antiparallelβ-barrel and two α-helices covering the top and the bottom of the barrel. There is an additionalsmall 310-helix formed by Ser118, Trp119 and Leu120. The ligand Ala-Pro (shown as cappedstick model) is situated in a hydrophobic pocket made up of β-sheets 3 to 6 and adjacent loops.(B) Stereo view of the active site with the bound dipeptide Ala-Pro. The cis form of thedipeptide is well de�ned in the electron density as shown by a Fo-Fc omit map contoured at4.5 σ. The side chain of the proline makes hydrophobic interactions with the side chains ofPhe58, Met59, Phe111, Leu120 and His124 of Mala s 6, while the N- and C-termini of Ala-Proform hydrogen bonds with Arg53, Gln61, Asn100 and a water molecule (W119).

2.4. RESULTS 59

2.4.5 Superposition of Mala s 6 on human CyP B reveals putativeIgE-binding residues

Mala s 6 shares 60% identity with human CyP B. The structure of the mature form -lacking the signal sequence and the following 6 N-terminal amino acids - was solved byX-ray crystallography (PDB code 1CYN, [34]). On the structural level, Mala s 6 andhuman CyP B superpose well with an rmsd of 1.05 Å for all Cα atoms (Figure 2.5 A).There are structural di�erences in two loops and in the extensions of the N- and C-terminiof human CyP B. The loop connecting the �rst two β-strands in Mala s 6 (Lys10-Ala13)and the 10-amino acid loop connecting the second α-helix to the next β-strand in Mala s 6(Glu143-Ser152) adopt a di�erent conformation than the corresponding loops in humanCyP B. There is very little sequence homology in these loops (Figure 2.1). The extensionof the C-terminus (11 residues) of human CyP B prolongs the last β-strand, makes a turnand forms - together with the extension of the N-terminus (4 residues) - an additionalβ-sheet.Since Mala s 6 and human CyP B show cross-reactivity, they must share common IgE-binding epitopes. Only those residues that are at least partly exposed to solvent cancontribute to antigen-antibody interactions in native proteins. Thus, solvent-accessibleresidues conserved in both proteins are potentially involved in the IgE-mediated cross-reactivity. A sequence alignment of Mala s 6 and human CyP B shows that a total of88 of the 162 aligned residues are identical (Figure 2.1). Based on the solved structures,the solvent-accessible surface area of Mala s 6 and human CyP B was calculated withthe program NACCESS, using a probe radius of 1.4 Å. 30 of the 88 identical aminoacids are at least 30%, and 10 thereof at least 50% solvent exposed in both structuresand are likely to form cross-reactive IgE-binding regions. These residues, as well as theremaining conserved active site residues were mapped on the solvent-accessible surface ofMala s 6 (Figure 2.5 B-D). The �gures reveal conserved, contiguous patches, which mightrepresent conformational, cross-reactive IgE-binding epitopes. Interestingly, there are noconserved patches on the top, the bottom and the backside of the molecule (Figure 2.5 B).Lys149 on the top lies in the conformationally di�erent loop after the second α-helix andcan thus not act as a cross-reactive IgE-binding residue despite its conservation in bothmolecules (Figure 2.1). Furthermore, this amino acid is not conserved in Asp f 11 andAsp f 27 and thus could not account for cross-reactivity. The residues at the back arepoorly conserved and the residues on the bottom are masked by the additional N- and C-terminal β-strand of human CyP B. Figure 2.5 C shows a conserved patch, which is madeup by the active site and �anking residues (Arg53, Asp57-Met59, Gln61, Gly70-Thr71,


A B

C D

Figure 2.5: Putative IgE-binding residues. (A) Superposition of Mala s 6 (pink) on humanCyP B (blue) reveals large structural similarities with three minor deviations: the loop afterthe �rst β-strand (1) and the loop after the second α-helix (2) adopt di�erent conformations.The N- and C-terminal extensions of human CyP B form an additional β-sheet (3). (B-D)Solvent-accessible surfaces of Mala s 6 showing putative IgE-binding residues. Amino acids thatare identical and at least 50% solvent-exposed in Mala s 6 and human CyP B are shown in pink.Residues that are identical and at least 30% solvent-exposed, as well as the remaining conservedactive site residues, are coloured in blue. Amino acids that are only conserved between Malas 6 and human CyP B, but not in Asp f 11 and/or Asp f 27 are in italics. (B) The backside,the top and the bottom of the molecule are devoid of conserved surface patches, due to lack ofhomology, to conformational di�erences (on top), or to the additional β-sheet in human CyP B(on the bottom). The view and scale are the same as in (A). (C) This side of the molecule showsa conserved patch, which is made up by the active site residues (AS) and �anking residues. Thesolvent-accessible surface area is quite large at 1017 Å2, and might thus represent two or moreoverlapping B-cell epitopes. The view is as in (B), but rotated by 100◦ about the y-axis. (D) Theother side shows two conserved surface patches: a lower patch (Pro28-K29, Tyr77-G78, Glu84)and an upper patch (Thr39-Phe44, Thr66, Lys74). Both patches could account for cross-reactiveIgE-binding epitopes in Mala s 6 and human CyP B. The view is as in (B), but rotated by 220◦

about the y-axis.

2.5. DISCUSSION 61

Ala99-Gly102, Phe111, Val115, Trp119-Leu120, Gly122-Lys123). The solvent-accessiblesurface area is quite large at 1017 Å2, and might thus represent two or more overlappingB-cell epitopes.The other side of the molecule reveals two conserved patches, which might be responsiblefor two di�erent IgE-binding epitopes (Figure 2.5 D). The lower patch consists of aminoacids Pro28-Lys29, Tyr77-Gly78 and Glu84 covering a surface area of 386 Å2. The upperpatch is made up by the 6 amino acid loop (Thr39-Phe44) after the �rst α-helix and byresidues Thr66 and Lys74 and covers a solvent-accessible surface area of 597 Å2 ful�llingthe criteria for a putative antigen-antibody interaction quite well.Mala s 6 is also cross-reactive with Asp f 11 and Asp f 27. The structures of these twoproteins are not known but they can be compared on the primary level with Mala s 6and human CyP B (Figure 2.1). 18 identical residues that are at least 30% solvent-exposed in Mala s 6 are shared among all four proteins. Thereof, 6 residues are at least50% solvent exposed. In the conserved patch around the active site (Figure 2.5 C) all butthree (Asp57, Val115, Lys123) amino acids are conserved among the four proteins. Thesethree residues lie on the edge of the patch and, therefore, they probably do not have alarge e�ect on IgE-binding. The lower patch in Figure 2.5 D is completely conserved in allproteins and could thus account for their overall cross-reactivity. Interestingly, the upperpatch is not conserved in Asp f 11 and thus cannot be involved in the cross-reactivity ofAsp f 11 with Mala s 6.

2.5 DiscussionRecently it was demonstrated that sera from patients su�ering from long lasting, chronicallergic diseases with in�ammatory background contain IgE antibodies against a vastvariety of self antigens [27�33]. The structures were either derived from screening ofhuman cDNA expression libraries with patient's sera [29, 53] or from direct cloning ofhuman (phylogenetically highly conserved) proteins sharing sequence homology with en-vironmental allergens [9, 27, 30]. Homologous molecular structures have been implicatedin many clinical syndromes including food-pollen allergy [54], latex-mould syndrome [24]and exacerbation or perpetuation of severe atopic disorders [27,32,33].However, not many crystal structures of allergens in general [19] and of cross-reactiveallergens in particular [55] are available which would allow to study antibody mediatedcross-reactivity in detail. To date, the atomic details of the interaction between antibodyand antigen are known for more than 30 antibody-antigen complexes [19]. The B-cellepitope, which is buried upon antibody binding, is in all cases conformational. The


epitope is made up of residues, which lie on di�erent surface loops forming discontinuousB-cell epitopes. They occupy a buried surface in the range of 560-860 Å2 and consistof 10-20 amino acids that are in contact with antibody residues [56]. The whole surfaceof a protein is potentially antigenic. Therefore, a typical globular 20 kDa allergen canaccommodate at most 5 to 10 antibodies at the same time [19]. The only method todetermine the complete structure of a B-cell epitope is to co-crystallise the allergen with amonoclonal antibody Fab fragment and solve the X-ray structure of the complex. Becausemonoclonal human IgE is di�cult to obtain, antibodies from other sources are currentlyused. The �rst structure of an allergen-Fab complex that has been solved, was thestructure of a complex between the major birch pollen allergen Bet v 1 and the Fabfragment of the murine monoclonal IgG1 antibody BV16 [57].We used an alternative approach to identify B-cell epitopes of cyclophilins involved inIgE-mediated cross-reactivity by determining shared features on the level of primary andtertiary structure. Notably, cross-reactivity between human and fungal cyclophilins canbe demonstrated in vitro (Figure 2.2) as well as in vivo (Table 2.2). The crystal structureof M. sympodialis cyclophilin was determined at 1.50 Å resolution and compared with thestructure of human cyclophilin B, which had been determined at 1.85 Å resolution [34].As expected from the primary structure which shows 88 identical amino acids among 162aligned (Figure 2.1) the comparison revealed a high similarity between the two structures(Figure 2.5 A). Although the majority of the amino acids are identical, only 30 thereofare > 30% and only 10 are > 50% solvent exposed in both structures and therefore likelyto be accessible for interactions with cross-reactive antibodies (Figure 2.5 B-D). Thus alarge portion of the conserved amino acids is located in the core of the protein and notaccessible for antigen-antibody interactions. The conserved, surface exposed residues arescattered over the whole sequence (Figure 2.1) and thus likely to de�ne discontinuousstructures found in B cell epitopes [58]. In contrast, they are clustered over the surfaceforming three patches covering solvent-accessible surface areas of 386, 597 and 1017 Å2,respectively (Figure 2.5 C and D) able to accommodate three or more B-cell epitopeswhich involve 15-22 amino acid residues on di�erent surface loops [58] and occupy aburied surface in the range of 540-890 Å2 [59]. Although the crystal structures of Aspf 11 and Asp f 27 are not yet solved, modelling experiments based on the human CyP Bstructure as template show comparable results as those obtained by comparison of theCyP B and Mala s 6 crystal structures (data not shown) o�ering a plausible explanationfor the IgE-mediated cross-reactivity between cyclophilins demonstrated by Western blotanalysis and inhibition ELISA (Figure 2.2) and elsewhere [9].Based on these results the actual contribution of a conserved, solvent exposed residue to

BIBLIOGRAPHY 63

the binding of IgE and to the cross-reactivity among di�erent CyPs can be investigatedby site-directed mutagenesis and might lead to the production of an engineered moleculelacking IgE-binding capacity and thus be useful as a safe candidate vaccine [60]. However,the most exciting issue will be to use the sequence information and the recombinantproteins to investigate in more details the role played by these cross-reactive structuresin the pathogenesis of chronic in�ammatory allergic diseases. First promising results havebeen obtained with MnSOD, a pan-allergen involved in autoreactivity [33], which openthe way for a more detailed understanding of the mechanisms involved in eliciting andmaintaining clinically severe atopic disorders.

AcknowledgementsWe are grateful to G. Menz for performing intradermal skin tests. Data collection wasperformed at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland andwe thank T. Tomizaki for his great technical support. Supported by the Swiss NationalScience Foundation grant No. 3100-063381/2, by the OPO foundation, Zürich and bythe Swedish Research Council grants No. 74F-15193 and 74X-7924.

Bibliography[1] G. Fischer, B. Wittmann-Liebold, K. Lang, T. Kiefhaber, and F. X. Schmid. Cyclophilin and

peptidyl-prolyl cis-trans isomerase are probably identical proteins. Nature, 337(6206):476�8, 1989.


[3] C. C. Trandinh, G. M. Pao, and M. H. Saier, Jr. Structural and evolutionary relationships amongthe immunophilins: two ubiquitous families of peptidyl-prolyl cis-trans isomerases. FASEB J.,6(15):3410�20, 1992.

[4] J. Fanghanel and G. Fischer. Insights into the catalytic mechanism of peptidyl prolyl cis/transisomerases. Front. Biosci., 9:3453�78, 2004.

[5] R. E. Handschumacher, M. W. Harding, J. Rice, R. J. Drugge, and D. W. Speicher. Cyclophilin: aspeci�c cytosolic binding protein for cyclosporin A. Science, 226(4674):544�7, 1984.


[7] J. Jain, P. G. McCa�rey, Z. Miner, T. K. Kerppola, J. N. Lambert, G. L. Verdine, T. Curran, andA. Rao. The T-cell transcription factor NFATp is a substrate for calcineurin and interacts with Fosand Jun. Nature, 365(6444):352�5, 1993.


[8] J. F. Borel. Pharmacology of cyclosporine (sandimmune). IV. Pharmacological properties in vivo.Pharmacol. Rev., 41(3):259�371, 1990.

[9] S. Flückiger, H. Fijten, P. Whitley, K. Blaser, and R. Crameri. Cyclophilins, a new family ofcross-reactive allergens. Eur. J. Immunol., 32(1):10�7, 2002.

[10] R. Crameri. Molecular cloning of Aspergillus fumigatus allergens and their role in allergic bron-chopulmonary aspergillosis. Chem. Immunol., 81:73�93, 2002.

[11] M. Lindborg, C. G. Magnusson, A. Zargari, M. Schmidt, A. Scheynius, R. Crameri, and P. Whitley.Selective cloning of allergens from the skin colonizing yeast Malassezia furfur by phage surfacedisplay technology. J. Invest. Dermatol., 113(2):156�61, 1999.

[12] A. Zargari, G. Midgley, O. Back, S. G. Johansson, and A. Scheynius. IgE-reactivity to sevenMalassezia species. Allergy, 58(4):306�11, 2003.

[13] C. Rhyner, M. Weichel, S. Flückiger, S. Hemmann, T. Kleber-Janke, and R. Crameri. Cloningallergens via phage display. Methods, 32(3):212�8, 2004.

[14] P. Cadot, J. F. Diaz, P. Proost, J. Van Damme, Y. Engelborghs, E. A. Stevens, and J. L. Ce-uppens. Puri�cation and characterization of an 18-kd allergen of birch (Betula verrucosa) pollen:identi�cation as a cyclophilin. J. Allergy Clin. Immunol., 105(2 Pt 1):286�91, 2000.

[15] C. Fujita, T. Moriyama, and T. Ogawa. Identi�cation of cyclophilin as an IgE-binding protein fromcarrots. Int. Arch. Allergy Immunol., 125(1):44�50, 2001.

[16] J. M. Hopkin. Mechanisms of enhanced prevalence of asthma and atopy in developed countries.Curr. Opin. Immunol., 9(6):788�92, 1997.


[18] H. J. Gould, B. J. Sutton, A. J. Beavil, R. L. Beavil, N. McCloskey, H. A. Coker, D. Fear, andL. Smurthwaite. The biology of IgE and the basis of allergic disease. Annu. Rev. Immunol., 21:579�628, 2003.


[20] J. Holm, G. Baerentzen, M. Gajhede, H. Ipsen, J. N. Larsen, H. Lowenstein, M. Wissenbach, andM. D. Spangfort. Molecular basis of allergic cross-reactivity between group 1 major allergens frombirch and apple. J. Chromatogr. B. Biomed. Sci. Appl., 756(1-2):307�13, 2001.

[21] A. Henriksen, T. P. King, O. Mirza, R. I. Monsalve, K. Meno, H. Ipsen, J. N. Larsen, M. Gajhede,and M. D. Spangfort. Major venom allergen of yellow jackets, Ves v 5: structural characterizationof a pathogenesis-related protein superfamily. Proteins, 45(4):438�48, 2001.

[22] S. Flückiger, P. R. Mittl, L. Scapozza, H. Fijten, G. Folkers, M. G. Grutter, K. Blaser, andR. Crameri. Comparison of the crystal structures of the human manganese superoxide dismutaseand the homologous Aspergillus fumigatus allergen at 2-Å resolution. J. Immunol., 168(3):1267�72,2002.


BIBLIOGRAPHY 65

[24] S. Wagner, S. Sowka, C. Mayer, R. Crameri, M. Focke, V. P. Kurup, O. Scheiner, and H. Breiteneder.Identi�cation of a Hevea brasiliensis latex manganese superoxide dismutase (Hev b 10) as a cross-reactive allergen. Int. Arch. Allergy Immunol., 125(2):120�7, 2001.

[25] A. Andersson, O. Rasool, M. Schmidt, R. Kodzius, S. Flückiger, A. Zargari, R. Crameri, andA. Scheynius. Cloning, expression and characterization of two new IgE-binding proteins from theyeast Malassezia sympodialis with sequence similarities to heat shock proteins and manganese su-peroxide dismutase. Eur. J. Biochem., 271(10):1885�94, 2004.

[26] S. Flückiger, L. Scapozza, C. Mayer, K. Blaser, G. Folkers, and R. Crameri. Immunological andstructural analysis of IgE-mediated cross-reactivity between manganese superoxide dismutases. Int.Arch. Allergy Immunol., 128(4):292�303, 2002.

[27] R. Crameri, A. Faith, S. Hemmann, R. Jaussi, C. Ismail, G. Menz, and K. Blaser. Humoral andcell-mediated autoimmunity in allergy to Aspergillus fumigatus. J. Exp. Med., 184(1):265�70, 1996.

[28] U. Appenzeller, C. Meyer, G. Menz, K. Blaser, and R. Crameri. IgE-mediated reactions to au-toantigens in allergic diseases. Int. Arch. Allergy Immunol., 118(2-4):193�6, 1999.

[29] S. Natter, S. Seiberler, P. Hufnagl, B. R. Binder, A. M. Hirschl, J. Ring, D. Abeck, T. Schmidt,P. Valent, and R. Valenta. Isolation of cDNA clones coding for IgE autoantigens with serum IgEfrom atopic dermatitis patients. FASEB J., 12(14):1559�69, 1998.

[30] C. Mayer, U. Appenzeller, H. Seelbach, G. Achatz, H. Oberko�er, M. Breitenbach, K. Blaser, andR. Crameri. Humoral and cell-mediated autoimmune reactions to human acidic ribosomal P2 proteinin individuals sensitized to Aspergillus fumigatus P2 protein. J. Exp. Med., 189(9):1507�12, 1999.

[31] I. Mittermann, K. J. Aichberger, R. Bunder, N. Mothes, H. Renz, and R. Valenta. Autoimmunityand atopic dermatitis. Curr. Opin. Allergy Clin. Immunol., 4(5):367�71, 2004.

[32] R. Valenta, S. Seiberler, S. Natter, V. Mahler, R. Mossabeb, J. Ring, and G. Stingl. Autoallergy:a pathogenetic factor in atopic dermatitis? J. Allergy Clin. Immunol., 105(3):432�7, 2000.

[33] P. Schmid-Grendelmeier, S. Flückiger, R. Disch, A. Trautmann, B. Wüthrich, K. Blaser,A. Scheynius, and R. Crameri. IgE- and T cell mediated autoimmunity against manganese su-peroxide dismutase in atopic dermatitis. J. Allergy Clin. Immunol., in press.

[34] V. Mikol, J. Kallen, and M. D. Walkinshaw. X-ray structure of a cyclophilin B/cyclosporin complex:comparison with cyclophilin A and delineation of its calcineurin-binding domain. Proc. Natl. Acad.Sci. USA, 91(11):5183�6, 1994.

[35] T. P. King, D. Ho�man, H. Lowenstein, D. G. Marsh, T. A. Platts-Mills, and W. Thomas. Allergennomenclature. Allergy, 50(9):765�74, 1995.

[36] M. Moser, R. Crameri, E. Brust, M. Suter, and G. Menz. Diagnostic value of recombinant Aspergillusfumigatus allergen I/a for skin testing and serology. J. Allergy Clin. Immunol., 93(1 Pt 1):1�11,1994.

[37] R. Kodzius, C. Rhyner, Z. Konthur, D. Buczek, H. Lehrach, G. Walter, and R. Crameri. Rapididenti�cation of allergen-encoding cDNA clones by phage display and high-density arrays. Comb.Chem. High Throughput Screen., 6(2):147�54, 2003.

[38] Z. Otwinowski and W. Minor. Processing of X-ray di�raction data collected in oscillation mode.Methods Enzymol., 276:307�26, 1997.


[39] A. Vagin and A. Teplyakov. MOLREP: an automated program for molecular replacement. J. Appl.Crystallog., 30:1022�5, 1997.

[40] G. N. Murshudov, A. A. Vagin, and E. J. Dodson. Re�nement of macromolecular structures by themaximum-likelihood method. Acta Cryst., D53:240�55, 1997.

[41] CCP4. The CCP4 suite: programs for protein crystallography. Acta Cryst., D50:760�3, 1994.

[42] D. E. McRee. XtalView/X�t - a versatile program for manipulating atomic coordinates and electrondensity. J. Struct. Biol., 125(2-3):156�65, 1999.

[43] V. S. Lamzin and K. S. Wilson. Automated re�nement of protein models. Acta Cryst.,D49(Pt 1):129�47, 1993.

[44] R. A. Laskowski, M. W. MacArthur, D. S. Moss, and J. M. Thornton. PROCHECK: a program tocheck the stereochemical quality of protein structures. J. Appl. Cryst., 26:283�91, 1993.

[45] R. W. Hooft, G. Vriend, C. Sander, and E. E. Abola. Errors in protein structures. Nature,381(6580):272, 1996.

[46] E. A. Merritt. Expanding the model: anisotropic displacement parameters in protein structurere�nement. Acta Cryst., D55:1109�17, 1999.

[47] W. Kabsch and C. Sander. Dictionary of protein secondary structure: pattern recognition ofhydrogen-bonded and geometrical features. Biopolymers, 22(12):2577�637, 1983.

[48] S. J. Hubbard and J. M. Thornton. NACCESS, Computer Program, Department of Biochemistryand Molecular Biology, University College London, 1996.

[49] B. Lee and F. M. Richards. The interpretation of protein structures: estimation of static accessi-bility. J. Mol. Biol., 55(3):379�400, 1971.

[50] J. Dornan, A. P. Page, P. Taylor, S. Wu, A. D. Winter, H. Husi, and M. D. Walkinshaw. Biochemicaland structural characterization of a divergent loop cyclophilin from Caenorhabditis elegans. J. Biol.Chem., 274(49):34877�83, 1999.

[51] Y. Zhao and H. Ke. Mechanistic implication of crystal structures of the cyclophilin-dipeptidecomplexes. Biochemistry, 35(23):7362�8, 1996.

[52] S. Wu Sy, J. Dornan, G. Kontopidis, P. Taylor, and M. D. Walkinshaw. The �rst direct determinationof a ligand binding constant in protein crystals. Angew. Chem. Int. Ed. Engl., 40(3):582�6, 2001.

[53] R. Crameri, R. Kodzius, Z. Konthur, H. Lehrach, K. Blaser, and G. Walter. Tapping allergenrepertoires by advanced cloning technologies. Int. Arch. Allergy Immunol., 124(1-3):43�7, 2001.

[54] K. Ho�mann-Sommergruber, P. Demoly, R. Crameri, H. Breiteneder, C. Ebner, M. Laimer DaCamara Machado, K. Blaser, C. Ismail, O. Scheiner, J. Bousquet, and G. Menz. IgE reactivity toApi g 1, a major celery allergen, in a central european population is based on primary sensitizationby Bet v 1. J. Allergy Clin. Immunol., 104(2 Pt 1):478�84, 1999.

[55] S. Flückiger, A. Limacher, A. G. Glaser, L. Scapozza, and R. Crameri. Structural aspects of cross-reactive allergens. Devel. Allergy Clin. Immunol., 5:57�75, 2004.

[56] D. R. Davies and G. H. Cohen. Interactions of protein antigens with antibodies. Proc. Natl. Acad.Sci. USA, 93(1):7�12, 1996.

BIBLIOGRAPHY 67



[59] E. A. Padlan. X-ray crystallography of antibodies. Adv. Protein Chem., 49:57�133, 1996.

[60] P. Neudecker, K. Lehmann, J. Nerkamp, T. Haase, A. Wangorsch, K. Fotisch, S. Ho�mann, P. Rosch,S. Vieths, and S. Scheurer. Mutational epitope analysis of Pru av 1 and Api g 1, the majorallergens of cherry (Prunus avium) and celery (Apium graveolens): correlating IgE reactivity withthree-dimensional structure. Biochem. J., 376(Pt 1):97�107, 2003.


Chapter 3

Structure Solution of Cross-ReactiveThioredoxins: Success and Attempts

A manuscript is in preparation:

The Crystal Structure of Malassezia sympodialis Thioredoxin Mala s 13, aMember of a New Pan-Allergen Family

Andreas Limacher1,2,#, Andreas G. Glaser2,#, Christa Meier1, Reto Crameri2,∗, andLeonardo Scapozza3

1Institute of Pharmaceutical Sciences, Department of Chemistry and Applied Biosciences,Swiss Federal Institute of Technology (ETH), 8093 Zurich, Switzerland2Swiss Institute of Allergy and Asthma Research (SIAF), 7270 Davos, Switzerland3Laboratoire de Chimie Thérapeutique, Section des Sciences Pharmaceutiques, Universitéde Genève, 1211 Genève 4, Switzerland#Equally contributing∗Correspondence: Reto Crameri, Swiss Institute of Allergy and Asthma Research (SIAF),Obere Strasse 22, CH-7270 Davos, Switzerland. Phone: +41 81 420 07 31, Fax: +41 81410 08 40, E-mail: [email protected]

Running title: Crystal structure of the Malassezia sympodialis thioredoxin

Key words: allergy, allergen structure, cross-reactivity, IgE epitopes, thioredoxin, pan-allergens, Malassezia, fungi, auto-reactivity

69

70 CHAPTER 3. CROSS-REACTIVE THIOREDOXINS

3.1 AbstractThioredoxins are small redox proteins present in a wide number of species. They have alsobeen identi�ed as a new pan-allergen family able to elicit IgE-mediated hypersensitivityreactions. Moreover, it has been shown that human thioredoxin is recognised by IgE fromsera of patients sensitised to environmental thioredoxin. IgE-mediated autoreactivity toself antigens sharing sequence homology with allergens is a phenomenon often observedin chronic in�ammatory atopic disorders. Therefore, comparison of the crystal structureof human proteins sharing homology to environmental allergens should allow the iden-ti�cation of structural similarities to explain IgE-mediated autoreactivity. In order to�nd these common features, di�erent thioredoxins were crystallised. Promising crystalswere obtained from the thioredoxins of the yeast Malassezia sympodialis and the wheatTriticum aestivum. Unfortunately, two M. sympodialis as well as two wheat crystal formsshowed various types of twinning and additional non-crystallographic translations, whichmade structure solving infeasible. Finally, a change in the M. sympodialis thioredoxinconstruct - the His tag was not removed - resulted in a non-twinned crystal form belong-ing to space group P21. The structure was solved by molecular replacement and re�nedto a resolution of 1.41 Å with R and Rfree values of 14.0% and 16.8%, respectively. The�nal re�ned model consists of two independent molecules per asymmetric unit showingthe typical thioredoxin fold, consisting of a �ve-stranded β-sheet forming a hydrophobiccore surrounded by �ve α-helices. Moreover, both His tags are visible in the electron den-sity, which is quite rare. The solvent-accessible surface was compared to the surface ofhuman thioredoxin revealing two conserved patches potentially involved in IgE-mediatedcross-reactivity.

3.2 IntroductionThioredoxins (Trxs) are small redox proteins found in all living cells. They undergoNADPH-dependent reduction by thioredoxin reductase and in turn reduce oxidised cys-teine groups on target proteins. The catalytic activity of Trxs is based on the tworedox-active cysteines in the highly conserved sequence WCGPC. A nucleophilic attackby the thiolate of the �rst Cys breaks the disul�de bridge of the target protein form-ing a mixed disul�de intermediate. This intermediate is then broken by the second Cysleaving the target protein reduced and releasing thioredoxin in the oxidised form [1, 2].Trxs have been implicated in a number of mammalian cell functions. Activity has beenfound outside the cell (cell growth stimulation and chemotaxis), in the cytoplasm (as an


antioxidant and a reductant cofactor), in the nucleus (regulation of transcription factoractivity) and in the mitochondria [3].Thioredoxins have been identi�ed as a new pan-allergen family able to elicit IgE-mediatedhypersensitivity reactions from several species. First, Trx was identi�ed as allergenicprotein of Coprinus comatus [4] and human Trx was shown to induce type I skin reac-tions in individuals sensitised to C. comatus Trx (Crameri and Helbling, unpublished),therefore acting as autoreactive self-antigen. Trxs from the mould A. fumigatus [5], theetiologic agent identi�ed in the majority of Aspergillus-related human diseases [6] andfrom M. sympodialis (Flückiger et al., unpublished), a skin colonising yeast involved inthe pathophysiology of atopic eczema [7], were isolated by phage display from the fungalcDNA libraries. Screening a wheat germ and maize endosperm library with IgE fromwheat and maize allergic individuals, respectively, revealed that wheat and maize Trxsare also allergenic molecules (Weichel et al., unpublished). Furthermore, S. cerevisiaeTrx has been cloned and produced (Glaser et al., unpublished) and cross-reactivity stud-ies between Trxs of the organisms mentioned above, including human, are in process.Cross-reactivity between allergens and closely related homologous human antigens is of-ten observed in allergic individuals su�ering from chronic atopic diseases [8�11] and theavailability of the 3D structure of a given allergen and its human homologue allows adetailed study of the residues involved in cross-reactivity [12]. These studies providestrong evidence for in vitro and in vivo humoral and cell-mediated IgE-autoreactivityin patients su�ering from long lasting atopic diseases [8, 11] potentially contributing toexacerbation and/or perpetuation of chronic allergic reactions [13].The only method to determine the complete structure of a B-cell epitope is to co-crystallise the allergen with a monoclonal antibody Fab fragment and solve the X-raystructure of the complex. This approach is very time-consuming and not always success-ful. Since monoclonal human IgE is di�cult to obtain, antibodies from other sourcesare currently used. In this work, an alternative approach was used. Putative B-cellepitopes of homologous proteins involved in IgE-mediated cross-reactivity can be identi-�ed by determining shared features on the level of primary and tertiary structure. Thethree-dimensional structure of M. sympodialis thioredoxin (Mala s 13) was determinedat 1.41 Å resolution by X-ray di�raction analysis. The structure was solved by molecu-lar replacement and re�ned to an R-factor of 14.0% and a free R-factor of 16.8%. Thesolved 3D structure was compared to the structure of human thioredoxin [14], in order tode�ne solvent-accessible residues shared by the two crystal structures. These amino acidresidues potentially involved in IgE-mediated cross-reactivity among Trxs o�er an expla-nation for the IgE-mediated autoreactivity found in clinically distinct chronic diseases.



3.3.1 Malassezia sympodialis thioredoxinCloning, protein expression and puri�cation

The original Mala s 13 was cloned by phage surface display [15] as N-terminal non-cleavable His-tagged protein into a pQE30 vector containing an Xba-linker of 10 aminoacids (GSSRAAARAG) between the His tag and the second amino acid of the proteincoding sequence (Flückiger et al., unpublished). The start methionine was missing inthe construct. In order to introduce a thrombin-cleavable His tag, the original Malas 13 cDNA was ampli�ed from the original clone by PCR with the primers 5' BamHI 5'-CCGCGGATCCGTGCAAGTGATTTCTTC-3' and 3' HindIII 5'-GCCCAAGCTTTT-AGGCCGAGTGCTGG-3' using the Pfu Turbo DNA Polymerase (2.5 U/µl) (Stratagene,La Jolla, CA, USA). The PCR product was digested with BamHI and HindIII restrictionendonucleases (NEB, Beverly, MA, USA), cleaned with QIAquick PCR puri�cation kit(Qiagen, Hilden, Germany) and ligated into a modi�ed pQE32 vector containing anN-terminal His6 tag followed by a thrombin cleavage site (HHHHHHLVPRGS), whereGS corresponds to the BamHI site. The ligation mixture was transformed into E. colistrain M15 pREP4 and the sequence of picked clones containing inserts of the correctsize veri�ed by DNA sequencing.M15 cells containing the correct vector were grown at 37◦C in Terri�c Broth medium toan OD600 of 1.0, induced with 1 mM IPTG, incubated at 25◦C for 24 h, harvested bycentrifugation at 6000g for 10 min at 4◦C and stored at -20◦C. The cell pellet was resus-pended in lysis bu�er (50 mM NaH2PO4, 300 mM NaCl, 50 mM imidazole, pH 8.0) andlysed by French Press. The insoluble material was removed by centrifugation at 20,000g(20 min, 4◦C). The His-tagged recombinant protein was puri�ed by nickel a�nity chro-matography using a 5 ml HiTrap Chelating HP column (Amersham Pharmacia Biotech,Uppsala, Sweden). The protein was eluted in a linear bu�er gradient (50 - 250 mM imi-dazole, 50 mM NaH2PO4, 300 mM NaCl, pH 8.0). Thrombin digestion of the N-terminalHis tag was not successful using 10-50 Units thrombin per mg recombinant protein in300 mM NaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5 and incubation for up to 30 h at 22◦C.Therefore, further work was carried out with the uncleaved protein puri�ed by gel �ltra-tion on a Superdex 75 column (FPLC, Pharmacia, Uppsala, Sweden) equilibrated with100 mM NaCl, 50 mM Tris, pH 7.5. The eluted protein was diluted 1:10 with H2O andconcentrated to 10 mg/ml. Protein identity and purity were assessed using MALDI-TOFmass spectroscopy and SDS-PAGE.


Figure 3.1: Crystals of Mala s 13 grown at 23◦C by vapour di�usion against 1.8 M ammo-nium sulfate, 3% PEG400, 0.1 M imidazole pH 7.0 with a protein concentration of 10 mg/ml.Dimensions of the dent-like crystals are 250 x 60 x 60 µm.

Crystallisation and data collection

Crystallisation was performed using the hanging drop vapour di�usion method at 23◦C.4 µl of protein solution (10 mg/ml) were mixed with 2 µl of reservoir solution (1.8 Mammonium sulfate, 3% PEG400, 0.1 M imidazole pH 7.0). The drop was equilibratedagainst 500 µl of reservoir solution. After two weeks, small crystals grew, which reached asize of 250 x 60 x 60 µm after about two months (Figure 3.1). Crystals were cryoprotectedby soaking stepwise for 1 min in reservoir solution complemented with increasing amountsof ethylene glycol (5, 10 and 15%). The crystals were �ash-cooled in a stream of gaseousnitrogen and measured at 100 K. A dataset was collected to 1.41 Å resolution on thesynchrotron beamline X06SA at Swiss Light Source (Villigen/CH) at 100 K. Data wereprocessed and scaled with DENZO and SCALEPACK of the HKL program package [16].The crystals belong to the monoclinic space group P21 with cell parameters a=37.50 Å,b=51.99 Å, c=53.02 Å and β=99.49◦ (Table 3.1).

Structure determination and re�nement

The structure was solved by molecular replacement using MOLREP [17]. A polyalaninemodel of Chlamydomonas reinhardtii thioredoxin H (PDB code 1EP7 [18]) served assearch model. Two molecules were located in the asymmetric unit yielding a Matthewscoe�cient (VM) of 1.91 Å3/Da and a corresponding solvent content of 35.7%. An initialrigid body re�nement using REFMAC [19] as implemented in the CCP4 program suite[20] resulted in an R and Rfree of 52% and 53%, respectively. Further re�nement was


Data CollectionWavelength (Å) 1.00802Unit cell dimensions a, b, c (Å) 37.50, 51.99, 53.02α, β, γ (◦) 90.00, 99.49, 90.00

Resolution (Å) 37.0-1.41 (1.43-1.41)Unique re�ections 38,712Redundancy 4.4Completeness (%) 99.6 (93.2)Rsym (%) 4.5 (24.7)Average I/σ 18.9 (3.4)Re�nement StatisticsResolution (Å) 37.0-1.41Number of re�ections (working/test) 36,757 / 1938Number of atoms 2011Rcryst (%) 14.0Rfree (%) 16.8Mean B factor (Å2): All atoms 18.5Main chain atoms (chain A/B) 15.3 / 16.0Side chain atoms (chain A/B) 19.4 / 19.4Solvent atoms 29.8

Rmsd bond lengths (Å) 0.016Rmsd bond angles (◦) 1.45

Table 3.1: Data collection and re�nement statistics. Numbers in parentheses are for the highestresolution shell. Rfree was calculated using a test set of 5%.

also performed with REFMAC, manual rebuilding and correction with XtalView [21].Initially, NCS-restraints were used, which were stepwise released and �nally omitted.After a few re�nement cycles, positive peaks in the di�erence electron density indicatedwell ordered His tags in both molecules, which were modelled accordingly. 164 watermolecules were introduced using ARP [22]. Final rounds of re�nement were carried outwith individual anisotropic B factors. The side chain of Met74 in chain A as well as Gln3,Lys56 and Ser63 in chain B were modelled as double conformations with occupancies of0.6/0.4. During the �nal re�nement cycles, the Fo-Fc di�erence electron density clearlyshowed disordering in the residue ranges 30-37 and 73-75 of chain B with positive andnegative peaks indicating an alternate peptide conformation. Therefore, the main andside chains of these two ranges were modelled as double conformations with occupancies of0.6 and 0.4 for the original and alternate conformation, respectively. Statistics from datacollection and re�nement are provided in Table 3.1. The stereochemical quality of the�nal model was assessed with PROCHECK [23] and WHATCHECK [24]. Anisotropic


validation was done with PARVATI [25]. Secondary structure elements were assignedautomatically with DSSP [26]. Coordinates and structure factors will be deposited in theProtein Data Bank.

Calculation of the solvent-accessible area

Solvent-accessible surface areas were calculated from molecule A of the solved Mala s 13structure as well as from the only molecule of the oxidised human thioredoxin structure(PDB code 1ERU [14]). Calculations were performed with the program NACCESS [27],an implementation of the Lee and Richards solvent accessibility algorithm [28], using aprobe radius of 1.4 Å with a slice width of 0.01 Å and omitting all water molecules. Therelative residue accessibility is the ratio of the accessible area of a residue in the modelto the accessible area of that residue in an extended Ala-X-Ala tripeptide.

3.3.2 Malassezia sympodialis thioredoxin without His tagCloning, protein expression and puri�cation

Since the His tag was not removable from the Mala s 13-construct described before andsince a freely movable His tag can hinder crystallisation, it was attempted to improvethrombin-cleavage. In order to make the cleavage site more accessible to thrombin, twoadditional amino acids (His-Met) were introduced between the cleavage site and the �rstresidue of the protein coding sequence by site-directed mutagenesis. Mutagenesis wasperformed with the QuikChange site-directed mutagenesis kit (Stratagene) as described inthe instruction manual with the primers 5' HM-for 5'-GCGCGGATCCCACATGGTGC-AAGTGATTTCTTCG-3' and 3' HM-rev 5'-CGAAGAAATCACTTGCACCATGTGG-GATCCGCGC-3' using the Pfu Turbo DNA Polymerase (2.5 U/µl) and the pQE32/Malas 13-construct as template. The product was treated with DpnI (20 U/µl), cleaned withQIAquick PCR puri�cation kit (Qiagen) and transformed into E. coli strain M15 pREP4.The sequence of a correct clone was veri�ed by DNA sequencing.The M15 cells were grown at 37◦C in Terri�c Broth medium to an OD600 of 0.5, inducedwith 1 mM IPTG, incubated at 20◦C for 24 h, harvested by centrifugation at 6000gfor 10 min at 4◦C and stored at -20◦C. The cell pellet was lysed and the His-taggedprotein puri�ed as described before. Thrombin cleavage now was straightforward: TheN-terminal His tag was cleaved o� by thrombin (20 Units per mg protein) in 300 mMNaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5 by incubation for 16 h at 22◦C. The cleavedprotein was further puri�ed by gel �ltration as described before. The eluted protein was


diluted 1:10 with H2O and concentrated to 12 mg/ml. Protein identity and purity wereassessed using MALDI-TOF mass spectroscopy and SDS-PAGE.


Screening for crystallisation conditions was performed with Crystal Screen 1 and CrystalScreen Cryo (Hampton Research) using the sitting drop vapour di�usion method at 23◦C.2 µl of protein solution at 8 mg/ml were mixed with 1 µl of crystallisation solution andequilibrated against 400 µl of reservoir solution. Crystals grew readily out of severalconditions within 2 to 7 days, showing nice morphologies. Crystal Screen 1 yielded 8crystallisation conditions (N◦3, 9, 11, 17, 32, 34, 35, 40); two of them led to crystals bigenough for direct measurement (N◦11 and 34). Crystal Screen Cryo yielded ten di�erentcrystallisation conditions. Microcrystals grew in solutions N◦30, 31, 47 and 49. Biggercrystals were obtained with conditions N◦16, 17, 20, 32, 35 and 42.Crystals from three di�erent conditions were optimised, if necessary, and complete datasets collected. The �rst condition was obtained with Screen Cryo N◦20 and optimisedby the hanging drop vapour di�usion method at 23◦C as follows: 3 µl of protein at8 mg/ml were mixed with 1.5 µl of crystallisation solution (0.16 M ammonium sulfate,18% PEG4000, 22% glycerin, 0.08 M sodium acetate pH 4.5) and equilibrated against500 µl of reservoir solution. After 2-5 days, crystals grew to dimensions of 200 x 200x 100 µm (Figure 3.2 A). Since the solution was cryoprotecting, the crystals were di-rectly �ash-cooled in gaseous nitrogen and measured at 100 K on a home source. Acomplete dataset was collected to 2.1 Å resolution, which was processed and scaled withDENZO and SCALEPACK of the HKL program package [16]. The data was processablein the primitive tetragonal space group P422 with cell dimensions of a=b=66.28 Å andc=93.10 Å and scalable in P42212 with an Rsym of 8.0%. The intensity distribution in-dicated a non-crystallographic translation, which can complicate molecular replacementand mask possible merohedral twinning. Unfortunately, no structure solution was foundby molecular replacement, neither with AMORE [29] nor with MOLREP [17]. Humanthioredoxin (PDB code 1ERU), a polyalanine model of human Trx, a homology-modelof Mala s 13 based on human Trx and further structures of known Trxs served as searchmodels.The second condition, Screen 1 N◦11, yielded crystals that could be directly used formeasurement. They grew in 1.0 M ammonium dihydrogen phosphate, 0.1 M sodium cit-rate pH 5.6 reaching a diameter of 150 µm after 4-7 days. Crystals were cryoprotected bysoaking stepwise for 20 s in reservoir solution complemented with 10% and 20% ethylene


A B

Figure 3.2: Crystals of cleaved Mala s 13 obtained at 23◦C by vapour di�usion with a proteinconcentration of 8 mg/ml. (A) Primitive tetragonal crystals grew against 0.16 M ammoniumsulfate, 18% PEG4000, 22% glycerin, 0.08 M sodium acetate pH 4.5 to dimensions of 200 x200 x 100 µm. (B) Body centered tetragonal crystals grew against 0.6 M potassium dihydrogenphosphate, 0.6 M sodium dihydrogen phosphate, 23% glycerin, 0.075 M HEPES pH 7.5 to adiameter of 150 µm.

glycol, �ash-cooled in gaseous nitrogen at 100 K and measured to 1.57 Å resolution onthe synchrotron beamline X06SA at Swiss Light Source (Villigen/CH). Data was alsoprocessable in the primitive tetragonal space group P422 with isomorphous cell dimen-sions of a=b=66.44 Å and c=93.25 Å, but only scalable in P4222 with an Rsym of 4.1%.Again, the intensity distribution indicated a non-crystallographic translation and again,structure solution was unsuccessful using the before mentioned molecular replacementprograms.The third condition, Screen Cryo N◦35, was optimised by the hanging drop vapour dif-fusion method at 23◦C as follows: 3 µl of protein at 8 mg/ml were mixed with 1.5 µl ofcrystallisation solution (0.6 M potassium dihydrogen phosphate, 0.6 M sodium dihydro-gen phosphate, 23% glycerin, 0.075 M HEPES pH 7.5) and equilibrated against 500 µl ofreservoir solution. After 2-5 days, crystals grew to a diameter of 150 µm (Figure 3.2 B).Since the solution was cryoprotecting, the crystals were directly �ash-cooled in gaseousnitrogen and measured at 100 K on a home source. Many crystals showed splitting athigher resolution, which can be a sign for twinning. A complete dataset was collectedto 1.87 Å resolution, which was processable in the body centered tetragonal space groupI422 with cell dimensions of a=b=65.84 Å and c=93.07 Å and scalable in I422 with anRsym of 5.7%. The intensity distribution and the 2nd moment in Z indicated perfectmerohedral twinning, thus, the true space group is I4. A reasonable solution was found


by molecular replacement with MOLREP [17] using the homology-model of Mala s 13 asa search model and allocating two molecules per asymmetric unit, but this solution wasnot re�nable.Some test data was collected from other crystals, in order to �nd a new crystal form. Thiswas not successful. Crystals grown in Screen Cryo N◦16 showed again the body centeredtetragonal space group. Optimised crystals from Screen 1 N◦34 and N◦40 showed thealready known primitive tetragonal space group.

3.3.3 Triticum aestivum thioredoxinCloning, protein expression and puri�cation

The original full-length Trx from Triticum aestivum (wheat) was cloned by phage surfacedisplay [15] as N-terminal non-cleavable His-tagged protein, containing an alanine-richsignal peptide of 17 amino acids at the N-terminus (Weichel et al., unpublished). Dueto a cloning artefact, this construct contained a mutation at position 114 (Glu->Lys).In order to clone the wild-type protein without the signal peptide, but with a thrombin-cleavable His tag, the wheat trx gene was ampli�ed from a wheat cDNA library (pJuFo)by PCR with the primers 5' BamHI 5'-GCGGGATCCGTGATCTCCGTCCAC-3' and3' HindIII 5'-GGCCAAGCTTCTAGGCCGCGTGTAGC-3' using the Pfu Turbo DNAPolymerase (2.5 U/µl) (Stratagene). The PCR product was digested, cleaned and ligatedas described before. The ligation mixture was transformed into E. coli strain XL1-Blue.The sequence of an insert with the correct size was veri�ed by DNA sequencing showingthe correct Glu at position 114. The XL1-Blue cells were grown at 37◦C in LB mediumto an OD600 of 0.6, induced with 1 mM IPTG, harvested after another 15 h of incubationby centrifugation at 6000 x g for 10 min at 4◦C and stored at -20◦C. The cell pellet waslysed and the His-tagged protein puri�ed as described before. Unfortunately, thrombindigestion of the N-terminal His tag was not successful using 10-50 Units thrombin per mgrecombinant protein in 300 mM NaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5 and incubationfor 18-40 h at 22◦C.In order to make the cleavage site more accessible to thrombin, three additional aminoacids were introduced between the cleavage site and the �rst residue of the protein cod-ing sequence by site-directed mutagenesis. The inserted residues Ala-Gly-Glu correspondto the last three amino acids of the signal sequence. Mutagenesis was performed withthe QuikChange site-directed mutagenesis kit (Stratagene) as described before using theprimers 5' AGE-for 5'-GGTGCCGCGTGGATCTGCCGGTGAAGTGATCTCC-3' and3' AGE-rev 5'-GGAGATCACTTCACCGGCAGATCCACGCGGCCACC-3'. The prod-


uct was transformed into E. coli strain M15 pREP4 and the sequence of a correct cloneveri�ed by DNA sequencing. The M15 cells were grown at 37◦C in LB medium to anOD600 of 0.8, induced with 1 mM IPTG, incubated at 30◦C for 15 h, harvested by cen-trifugation at 6000g for 10 min at 4◦C and stored at -20◦C. The cell pellet was lysedand the His-tagged protein puri�ed as described before. Thrombin cleavage now wasstraightforward: The N-terminal His tag was cleaved o� by thrombin (30 Units per mgprotein) in 300 mM NaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5 by incubation for 18 h at22◦C. The cleaved protein was further puri�ed by gel �ltration as described before. Theeluted protein was diluted 1:10 with H2O, one part was concentrated to 15 mg/ml, theother to 42 mg/ml. Protein identity and purity were assessed by SDS-PAGE.


Screening for crystallisation conditions was performed with Crystal Screen 1, CrystalScreen 2, Crystal Screen Cryo, PEG 6K Screen (all from Hampton Research) and an inhouse ammonium sulfate screen (20-80% ammonium sulfate, pH 3.5-9.0) using the sittingdrop vapour di�usion method at 16 and 23◦C. 2 µl of protein solution (8-42 mg/ml withor without 5 mM DTT) were mixed with 1 µl of crystallisation solution and equilibratedagainst 400 µl of reservoir solution (supplemented with or without 5 mM DTT). Theprotein was highly soluble under many crystallisation conditions, therefore the proteinconcentration was increased up to 42 mg/ml.One condition was found with Crystal Screen 1, solution N◦39 (42 mg/ml protein at 16◦Cand 23◦C, no DTT), which was optimised by the hanging drop vapour di�usion methodat 16◦C as follows: 4 µl of protein at 20 mg/ml were mixed with 2 µl of crystallisationsolution (1.5% PEG550MME, 1.6-1.8 M ammonium sulfate, 0.1 M imidazole pH 7) andequilibrated against 500 µl of reservoir solution. Alternatively, 4 µl of protein at 20 mg/mlwere mixed with 3 µl of crystallisation solution (1.5% PEG1000, 1.6-1.8 M ammoniumsulfate, 0.1 M imidazole pH 6.5) and equilibrated against 500 µl of reservoir solution.In both cases, many small crystals grew to a size of 70 x 70 x 50 µm after about oneweek using 1.8 M ammonium sulfate. Bigger and fewer crystals (200 x 200 x 100 µm)were obtained after about two months, when using 1.6 M ammonium sulfate (Figure3.3 A). Crystals were cryoprotected by soaking stepwise for 30-60 s in reservoir solutioncomplemented with increasing glycerin concentration of 5 to 10 and to 16%, �ash-cooledin gaseous nitrogen and measured at 100 K on the synchrotron beamline X06SA at SwissLight Source (Villigen/CH).Handling of the crystals was not very easy. Some fell apart during cryoprotection, some


A B

Figure 3.3: Crystals of wheat Trx obtained by vapour di�usion. (A) Crystals grew at 16◦Cwith a protein concentration of 20 mg/ml against 1.5% PEG550MME, 1.6 M ammonium sulfate,0.1 M imidazole pH 7 to dimensions of 200 x 200 x 100 µm. (B) Primitive tetragonal crystalsgrew at 23◦C with a protein concentration of 42 mg/ml against 0.2 M ammonium sulfate, 30%PEG8000, 0.1 M sodium cacodylate pH 6.5 to dimensions of 180 x 100 x 50 µm.

showed heavy signs of splitting, non-merohedral twinning and/or high mosaicity aftercollection of test data. Finally, one crystal was found that showed a nice di�ractionpattern. A complete dataset was collected to 2.5 Å resolution, which was processed andscaled with DENZO and SCALEPACK of the HKL program package [16]. Processingand scaling was ambigous. On one hand, the data could be processed and scaled in thebody centered tetragonal space group I422 to a resolution of 2.8 Å with cell dimensionsof a=b=93.1 Å and c=141.7 Å and a rather high Rsym of 16.6%. The data showed anextreme intensity distribution with a 2nd moment in Z of 2.88, which is an indication fornon-crystallographic translation. On the other hand, the data could be processed andscaled in the primitive tetragonal space group P422 to a resolution of 2.5 Å with celldimensions of a=b=65.9 Å and c=70.9 Å and an Rsym of 13.7%. Visual inspection ofthe di�raction images clearly showed that there were extremely weak re�ections in everysecond position along one cell axis, which were left out from processing. Moreover, thescaled data now showed a �at intensity distribution with a 2nd moment in Z of 1.92,which is an indication for partial merohedral twinning. The data was also processed andscaled in P1 (Rsym=9.3%), but the strange intensity distribution remained. Nevertheless,molecular replacement was performed in all relevant space groups with MOLREP [17]using a polyalanine model of human Trx (PDB code 1ERU) as search model. Not verysurprisingly, no solution was found.A second crystallisation condition was found with Crystal Screen 1, solution N◦15 at 23◦Cyielding crystals that could be directly used for measurement. 2 µl of protein at 42 mg/ml

3.4. RESULTS AND DISCUSSION 81

were mixed with 1 µl of solution N◦15 (0.2 M ammonium sulfate, 30% PEG8000, 0.1 Msodium cacodylate pH 6.5) and equilibrated against 400 µl of the same solution. Crys-tals grew to a size of 180 x 100 x 50 µm after about one month (Figure 3.3 B). Theywere cryoprotected by soaking stepwise for 30 s in reservoir solution complemented withincreasing glycerin concentration of 5 to 10 and to 15%, �ash-cooled in gaseous nitro-gen and measured to 1.8 Å resolution on the synchrotron. The data was processable inthe primitive tetragonal space group P4 to a resolution of 1.9 Å with cell dimensionsof a=b=92.7 Å and c=71.5 Å and scalable only in P4 with a reasonable Rsym of 7.7%.Unfortunately, the scaled data showed rather high anisotropy and again an extreme in-tensity distribution, which is indicative for a non-crystallographic translation and canmask twinning. Moreover, the calculated unit cell volume of 614,400 Å3 indicates quitemany molecules (4-6) per asymmetric unit. All these e�ects can complicate structure so-lution by molecular replacement. Anisotropic correction was performed with Phaser [30],which is also a powerful program for molecular replacement, but cannot handle data withtranslational NCS. Therefore, MOLREP [17] was used, but did not yield any reasonablesolution.

3.4 Results and discussion

3.4.1 Structure of Malassezia sympodialis thioredoxinThe X-ray structure of Mala s 13 was determined by molecular replacement and wasre�ned to a resolution of 1.41 Å with R and Rfree values of 14.0% and 16.8%, respectively.94.2% of the non-glycine and non-proline residues have main chain dihedral angels in themost favoured regions of the Ramachandran plot with the remaining ones located inthe additional allowed regions. Data collection and re�nement statistics are shown inTable 3.1.There are two independent molecules per asymmetric unit, designated A and B. The�nal model comprises all amino acids of both independent monomers (aa 2-105) and164 water molecules. In molecule A, which was used for the calculation of the solvent-accessible surface, all the side chains of the amino acids on the surface are well de�ned,except of Lys51, Gln67 and Arg72. These side chains are solvent-exposed and thus freelymovable. Interestingly, the His tags of both molecules are visible in the electron densityfrom the third histidine onwards, which is quite rare. They are sandwiched between twoneighbouring, symmetry related molecules and are thereby stabilised. In the His tag ofmolecule B, residues Arg-Gly-Val were not modelled due to disorder. Both monomers


C-term

N-term

His tag

Cys30Active site

Cys33

4

5

1

2

3

1

4

2

5

3

Figure 3.4: Cartoon representation of Mala s 13, molecule A. The overall fold consists of a�ve-stranded β-sheet forming a hydrophobic core surrounded by �ve α-helices. The side chainsof the active site cysteines and the His tag are shown as capped stick model.

show the typical Trx fold, consisting of a �ve-stranded β-sheet forming a hydrophobiccore surrounded by four α-helices (Figure 3.4). There is an additional small α-helixformed by residues Ala48-Lys51.In chain A, the conserved active site amino acids Trp29-Cys-Gly-Pro-Cys are well de�nedin the electron density. They link the second β-strand to the second α-helix, with Pro-Cysforming the �rst turn of the helix. The cysteines are in the oxidised form, as expectedfrom the oxidising crystallisation condition used. The sulfur atoms are very well de�nedin the density map and were thus re�ned without a disul�de bond restraint resulting ina bond length of 2.29 Å and a di�erence electron density without any peaks around thesulfurs. In chain B, there is some disorder in the active site residues and the �rst halfof the following α-helix (aa Cys30-Gly37) as well as in the residue range Ala73-Pro75.These two ranges, which lie in the vicinity of each other, were thus modelled in a doubleconformation (Figure 3.5). The cysteines are also present in the oxidised form, but - dueto the worse density - were re�ned with a disul�de bond restraint.The two independent molecules form a crystallographic dimer related by a non-crystallo-graphic 2-fold axis. The �rst β-strand of each molecule is hydrogen-bonded to each other,


His tag A

His tag B

molecule A

molecule B

1

1

2

2

ss-bond

ss-bond

Figure 3.5: Cartoon representation of the crystallographic dimer. The �rst β-strand of eachmolecule is hydrogen-bonded to each other resulting in an extended β-sheet between the twomolecules. Molecule A and B are coloured in yellow and violet, respectively. The main chainranges of molecule B, which were modelled in a double conformation, are shown in pink. Theside chains of the active site cysteines and the His tag are shown as capped stick model.

which results in an extended β-sheet between the two molecules consisting of ten strands(Figure 3.5). The dimer interface buries a surface area of 1000 Å2 on each molecule.Without the His tags, the contact interface drops to 713 Å2, thus the His tags contributesubstantially to the monomer-monomer interactions. The arrangement of the dimer isdi�erent from the natural covalent dimer observed in human thioredoxin [14]. The humandimer results from an intermolecular disul�de bond via the non-conserved residue Cys73of each monomer. The dimer is supposed to have a regulatory mechansism [31], since theactive site becomes buried and thus inactive on dimer formation. In contrast, the Malas 13 dimer is not a�ecting the active site and is most probably only of crystallographicnature. Thioredoxins are known to be redox-active in their monomeric form. In solution,


Mala s 13 behaves as a monomer. The quaternary state was determined in a gel �ltrationexperiment with four marker proteins on a Superdex 75 column (FPLC, Pharmacia)(data not shown). This yielded an elution volume of 13.01 ml and a calibrated weight of12.8 kDa, which is well in agreement with the calculated monomer weight of 13.3 kDa.

3.4.2 Superposition of Malassezia sympodialis Trx on human Trxreveals putative IgE-binding residues

Mala s 13 shares 45% sequence identity with human Trx. Superposition of the solvedMala s 13 structure with the oxidised human Trx (PDB code 1ERU [14]) reveals a highstructural similarity with an rmsd of 1.11 Å for all Cα-atoms (Figure 3.6 A). There aretwo conformational di�erences. First, a deletion of two amino acids between Gly16 andGly17 of Mala s 13 shortens the end of the �rst α-helix. Second, an insertion of tworesidues (Gly49, Asp50) leads to an additional small α-helix after the second helix ofMala s 13.Mala s 13 and human Trx show cross-reactivity leading to IgE-autoreactivity in patientssu�ering from long lasting atopic diseases. Therefore, these proteins must share commonIgE-binding epitopes. Only those residues that are at least partly exposed to solvent cancontribute to antigen-antibody interactions in native proteins. Thus, solvent-accessibleresidues conserved in both proteins are potentially involved in the IgE-mediated cross-reactivity. A sequence alignment of Mala s 13 and human Trx shows that a total of 48 ofthe 105 aligned residues are identical (Figure 3.7). The solvent-accessible surface areasof Mala s 13 and human Trx were calculated with the program NACCESS. 19 of the 48identical amino acids are at least 30%, and 9 thereof at least 50% solvent-exposed in bothstructures and are likely to form cross-reactive IgE-binding regions.The residues, which are identical and at least 30% or 50% solvent-exposed in Mala s 13 andhuman Trx were mapped on the solvent-accessible surface of Mala s 13 (Figure 3.6 B-D).The �gures reveal conserved, contiguous patches, which might represent conformational,cross-reactive IgE-binding epitopes. Figure 3.6 B shows four patches, a large one on top(patch 1), one in the middle (patch 2) and one on the bottom (patch3). There is a fourthpatch of only 2 amino acids, which is most probably too small to form an IgE-bindingepitope. Patch 1 consists of amino acids, which form the active site and the beginningof the following α-helix (Thr28, Trp29, Gly31, Pro32, Lys34, Met35 and Pro38) lying allon the same peptide stretch (Figure 3.6 D). There are two more residues contributingto this patch, which are situated on sequentially more distant loops (Asp60 and Ala92).The whole patch covers a solvent-accessible surface area of 846 Å2 and ful�lls the criteria


A

C D

B patch 1

patch 2

patch 4

patch 3

patch 1

patch 3

E68

G83

Q84

K85

K96

A99

K19

D18

G17

T28

W29

G31

P32

K34

M35

P38

D60A92

I

D

N

C

Figure 3.6: Putative IgE-binding residues. (A) Superposition of Mala s 13 (pink) on humanTrx (blue, PDB code 1ERU) reveals large structural similarities with two minor deviations. The�rst α-helix is shortened due to a deletion [D]. There is an additional small α-helix due to aninsert [I]. The N- and C-termini are designated N and C, respectively. (B-D) Solvent-accessiblesurfaces of Mala s 13 showing putative IgE-binding residues. Amino acids that are identicaland at least 50% or between 30% and 50% solvent-exposed in Mala s 13 and human Trx areshown in pink and blue, respectively. (B) The front view reveals four conserved surface patches.Patch 2 comprises four amino acids (Glu68, Gly83-Lys85) covering a surface area of 350 Å2 andlikely accounts for a B-cell epitope. Patch 4 (Lys96, Ala99) is most probably too small to forman IgE-binding epitope. Scale and view are as in (A). (C) The backside is virtually devoid ofconserved residues. Patch 3 (Gly17-Lys19) on the bottom is rather small and lies in a regionwith structural deviation. Therefore, it probably does not account for an IgE-binding epitope.The view is as in (A), but rotated by 180◦ about the y-axis. (D) Patch 1 on top consists ofresidues forming and surrounding the active site. It ful�lls the criteria for a putative antigen-antibody interaction quite well and is also conserved among other IgE-binding Trxs. The viewis as in (A), but rotated by 110◦ about the x-axis.


2ndary structure 1= === 1=== == 2== ===== 2====== = 3=

Patch number 333 11 11 11 1

Sequence number 10 20 30 40 50

Mala s 13 -----VQVISSYDQFKQVTG--GDKVVVIDFWATWCGPCKMIGPVFEKISDTPAGDK

Human Trx ----MVKQIESKTAFQEALDAAGDKLVVVDFSATWCGPCKMIKPFFHSLSEKYS--N

Asp f 28 MSHGKVIAVDNPIIYKALT---SSGPVVVDFFATWCGPCRAVAPKVGELSEKYS--N

Asp f 29 MSH-NVEKITDAKVFKEKVQE-GSGPVIVDCSATWCGPCKAISPVFQRLSTSEEFKN

Cop c 2 ----MVQVISNLDEFNKLTN--SGKIIIIDFWATWCGPCRVISPIFEKFSEKYGANN

Sac c Trx ----MVTQFKTASEFDSAIA--QDKLVVVDFYATWCGPCKMIAPMIEKFSEQYP--Q

* . :. . :::* *******: : * . :* :

2ndary structure == 3== == 4=== == 4== == 5= == ==== 5=====

Patch number 1 2 222 1 4 4

Sequence number 60 70 80 90 100

Mala s 13 VGFYKVDVDEQSQIAQEVGIRAMPTFVFFKNGQKIDT-VVGADPSKLQAAITQHSA

Human Trx VIFLEVDVDDCQDVASECEVKCMPTFQFFKKGQKVGE-FSGANKEKLEATINELV-

Asp f 28 VRFIQVDVDKVRSVAHEMNIRAMPTFVLYKDGQPLEKRVVGGNVRELEEMIKSISA

Asp f 29 AKFYEIDVDELSEVAAELGVRAMPTFMFFKDGQKVNE-VVGANPPALEAAIKAHVA

Cop c 2 IVFAKVDVDTASDISEEAKIRAMPTFQVYKDGQKIDE-LVGANPTALESLVQKSLA

Sac c Trx ADFYKLDVDELGDVAQKNEVSAMPTLLLFKNGKEVAK-VVGANPAAIKQAIAANA-

* ::*** .:: : : .***: .:*.*: : . *:: :: :

Figure 3.7: Structure alignment of Mala s 13 and human Trx combined with a sequencealignment of two A. fumigatus Trxs (Asp f 28 and Asp f 29), C. comatus Trx (Cop c 2) andS. cerevisiae Trx (Sac c Trx). The top line denotes the secondary structure of Mala s 13 assignedby DSSP [26], the second line shows the number (1-4) of the conserved patch a residue belongs to,and the third line shows the sequence numbering of Mala s 13. Asterisks indicate identical, colonsstrongly similar and periods weakly similar amino acids in all sequences. Active site residuesare in bold. The two loops, which adopt a di�erent conformation in Mala s 13 compared tohuman Trx are underlined. Residues that are identical and at least 50% or between 30% and50% solvent-exposed in Mala s 13 and human Trx are shown on a pink or blue background,respectively. Residues, which are also conserved or strongly similar in Asp f Trx1, Asp f Trx2,Cop c 2, or Sac c Trx are coloured correspondingly.

for a putative antigen-antibody interaction quite well regarding distribution and numberof involved residues and total surface area of the patch.Patch 2 comprises four amino acids distributed on α-helix 4 and the beginning of β-sheet 5(Glu68, Gly83-Lys85), which cover a surface area of 350 Å2. This is smaller than theinterface of an average antibody-binding epitope, but still could be responsible for across-reactive IgE-binding epitope. Together with neighbouring amino acids, which areless solvent-exposed or which are not identical, but strongly similar in both proteins (e.g.I71, R72, I86), the conserved patch could form an area large enough to accommodate across-reactive IgE-antibody. Patch 3 is formed by amino acids of the loop after the �rstα-helix (Gly17-Lys19) and covers an area of 246 Å2 (Figure 3.6 C). On one side, this areais rather small. On the other side, this patch lies in a region with structural deviation


between Mala s 13 and human Trx, due to a deletion of two amino acids between Gly16and Gly17. Therefore, this patch most probably does not account for a cross-reactiveepitope.Two Trxs from A. fumigatus (Asp f 28 and Asp f 29) and a Trx from S. cerevisiae andC. comatus (Cop c 2) have also been isolated as IgE-binding proteins. Whether theseallergens are cross-reactive with Mala s 13 and human Trx and among each other iscurrently under investigation (Glaser et al., unpublished). The structures of these fourTrxs are not known, but they can be compared on a sequential level to Mala s 13 andhuman Trx, which should allow the prediction of overall cross-reactivity (Figure 3.7).The alignment shows that patch 1 is highly conserved among all proteins. All aminoacids are identical or replaced against a strongly similar residue (in the case of Lys34and Ala92), except Met35, which lies on the edge of the patch and which is only partlyconserved. Therefore, this patch might represent a dominant IgE-binding epitope, whichis responsible for overall cross-reactivity among these Trxs. The patch is situated aroundthe active site and is involved in the binding to target proteins, which are prone toreduction of their disul�de bonds. Therefore, the active site as well as the surroundinghydrophobic surface is conserved in the Trx-family.Patch 2 is also strongly conserved and might thus account for a second overall cross-reactive epitope. Allergens must have at least two IgE-binding epitopes in order tocross-link IgE antibodies and trigger hypersensitivity reaction in vivo. Therefore, thetwo suggested epitopes could indeed be responsible for cross-reactivty in vivo. Patches3 and 4 are hardly conserved among the six Trxs further supporting the argument thatthey are not accounting for cross-reactivity.These results clearly suggest that shared structural features of homologous proteins canbe found by comparing the three-dimensional structures, giving an explanation for theIgE-mediated autoreactivity found in clinically distinct chronic atopic diseases. The ac-tual contribution of a conserved, solvent-exposed residue to the binding of IgE and tothe cross-reactivity among di�erent Trxs can be investigated by site-directed mutagenesis.One major disadvantage of current immunotherapy is that it can cause severe side e�ects,such as asthma attacks and anaphylactic shock. The identi�cation of IgE epitopes onallergens allows the modi�cation of important allergens such that they display stronglyreduced allergenic activity by disrupting the conformational IgE epitopes. Such hypoal-lergenic allergen derivatives can be used as candidates for vaccines in allergen-speci�cimmunotherapy with a reduced risk of immediate side e�ects [32].


3.4.3 Crystallisation of Malassezia sympodialis thioredoxin with-out His tag

Often, freely movable parts of recombinant proteins like His tags, loops or even whole do-mains can hinder crystallisation. Therefore, it was attempted to improve the accessibilityof the thrombin-cleavage site by the insertion of two additional amino acids between thecleavage site and the protein coding sequence. This approach worked very nicely result-ing in thrombin-cleaved Mala s 13 protein. Once the structure of the uncleaved proteinwas solved, the model showed that the �rst protein coding amino acid is involved in aβ-sheet structure, which might have left the cleavage site unaccessible to thrombin. Thee�ect of the cleaved protein on crystallisation was impressive. Whereas the uncleavedprotein only crystallised in a few conditions resulting in hardly omptimisable needles, thecleaved protein crystillised readily in many di�erent conditions yielding nice and ratherlarge crystals.Unfortunately, none of them were solvable. Three complete datasets were collected andevaluated. Two crystal conditions (Screen Cryo N◦20, Screen 1 N◦11) led to primitivetetragonal crystals showing isomorphous cell dimensions. Interestingly, they have slightlydi�erent space groups P42212 and P4222, which can be distinguished by systematic ab-sences. Moreover, the intensity distribution suggests a non-crystallographic translation.This is not bad per se, but can complicate molecular replacement. Additionally, it canmask signs of merohedral twinning, since presence of twinning is normally also judged bythe intensity distribution. Therefore, twinning cannot be excluded. Molecular replace-ment was performed in all relevant space groups using di�erent search models and alsotaking twinning into consideration. Unfortunately, no solution was found, probably dueto the problems mentioned before.The third crystal form (Screen Cryo N◦35) was perfectly merohedrally twinned belongingto space group I4. After many attempts, a solution was found by molecular replacementusing a homology-model of Mala s 13 and allocating two molecules per asymmetric unit.This solution was hardly re�nable and work was abandoned, because the structure ofthe uncleaved protein could be solved in the meantime. There were also attempts togrow untwinned crystals by slightly changing the crystallisation condition. This was notsuccessful; test data showed splitted re�ections at higher resolution, which can be anindication for twinning. In order to �nd a non-twinned and solvable crystal form, testdata was collected from crystals grown in di�erent crystallisation conditions, but thissearch only yielded the already known crystal forms.In the end, ironically, only the His-tagged protein was solvable, despite the worse crystalli-

BIBLIOGRAPHY 89

sation behaviour. The di�raction data was of very high quality and showed no twinning.This protein actually crystallised only due to the His tag. It made additional crystalcontacts with two neighbouring, symmetry-related molecules resulting in a new crystalform (P21) and in an ordered, visible His tag, which is quite rare.

3.4.4 Crystallisation of Triticum aestivum thioredoxinIn order to solve at least one crystal structure of a cross-reactive, allergenic thiore-doxin, crystallisation attempts with M. sympodialis and wheat Trx were done in parallel.Whereas the structure of M. sympodialis Trx could eventually be solved, determinationof wheat Trx has not been successful to date. The wheat protein crystallised readily,but only at very high concentrations. It was highly soluble in most crystallisation con-ditions, which can vary strongly in component composition and pH range. This veryhigh stability could explain that the protein survives denaturation and degradation inthe gastrointestinal tract, allowing it to be absorbed intact, thereby acting as allergen.Unfortunately, the yielded crystals showed various signs of disorder. Processing andscaling of the �rst crystal form (I422 or P422) was very ambigous. The scaled datashowed unsatisfying statistics and was indicative for translational NCS or twinning or acombination thereof. Most probably, these crystals are not single crystals. They probablyconsist of two or more di�erent lattices, which are coinciding in one or two axes, havingdi�erent fractions. This would explain the very weak, interlacing re�ections along oneaxis. Slight changes of the crystallisation condition could favour one lattice over theother, which would yield single crystals.The second crystal form (P4) is well processable and scalable showing only one lattice.Unfortunately, the data is anisotropic. This problem could probably be solved by opti-misation of the crystal condition or a change in the cryo-protocol, which might lead toisotropically di�racting crystals. The data also shows a non-crystallographic translation,which complicates molecular replacement and can mask merohedral twinning. In order tosolve these crystals, better isotropic data and more attempts in molecular replacement,using di�erent programs and search models is a prerequisite.

Bibliography[1] A. Holmgren. Thioredoxin. Annu. Rev. Biochem., 54:237�71, 1985.

[2] G. B. Kallis and A. Holmgren. Di�erential reactivity of the functional sulfhydryl groups of cysteine-32 and cysteine-35 present in the reduced form of thioredoxin from Escherichia coli . J. Biol. Chem.,255(21):10261�5, 1980.


[3] G. Powis and W. R. Montfort. Properties and biological activities of thioredoxins. Annu. Rev.Pharmacol. Toxicol., 41:261�95, 2001.

[4] A. Helbling, K. A. Brander, W. E. Horner, and S. B. Lehrer. Allergy to basidiomycetes. Chem.Immunol., 81:28�47, 2002.


[6] R. Crameri. Molecular cloning of Aspergillus fumigatus allergens and their role in allergic bron-chopulmonary aspergillosis. Chem. Immunol., 81:73�93, 2002.

[7] A. Zargari, G. Midgley, O. Back, S. G. Johansson, and A. Scheynius. IgE-reactivity to sevenMalassezia species. Allergy, 58(4):306�11, 2003.


[9] U. Appenzeller, C. Meyer, G. Menz, K. Blaser, and R. Crameri. IgE-mediated reactions to au-toantigens in allergic diseases. Int. Arch. Allergy Immunol., 118(2-4):193�6, 1999.




[13] R. Valenta, S. Seiberler, S. Natter, V. Mahler, R. Mossabeb, J. Ring, and G. Stingl. Autoallergy:a pathogenetic factor in atopic dermatitis? J. Allergy Clin. Immunol., 105(3):432�7, 2000.

[14] A. Weichsel, J. R. Gasdaska, G. Powis, and W. R. Montfort. Crystal structures of reduced, oxidized,and mutated human thioredoxins: evidence for a regulatory homodimer. Structure, 4(6):735�51,1996.

[15] R. Crameri, R. Jaussi, G. Menz, and K. Blaser. Display of expression products of cDNA librarieson phage surfaces. A versatile screening system for selective isolation of genes by speci�c gene-product/ligand interaction. Eur. J. Biochem., 226(1):53�8, 1994.



[18] V. Menchise, C. Corbier, C. Didierjean, M. Saviano, E. Benedetti, J. P. Jacquot, and A. Aubry.Crystal structure of the wild-type and D30A mutant thioredoxin H of Chlamydomonas reinhardtiiand implications for the catalytic mechanism. Biochem. J., 359(Pt 1):65�75, 2001.

BIBLIOGRAPHY 91







[25] E. A. Merritt. Expanding the model: anisotropic displacement parameters in protein structurere�nement. Acta Cryst., D55:1109�17, 1999.




[29] J. Navaza. AMoRe - an automated package for molecular replacement. Acta Cryst., A50:157�63,1994.

[30] L. C. Storoni, A. J. McCoy, and R. J. Read. Likelihood-enhanced fast rotation functions. ActaCryst., D60(Pt 3):432�8, 2004.

[31] J. F. Andersen, D. A. Sanders, J. R. Gasdaska, A. Weichsel, G. Powis, and W. R. Montfort.Human thioredoxin homodimers: regulation by pH, role of aspartate 60, and crystal structure ofthe aspartate 60 -> asparagine mutant. Biochemistry, 36(46):13979�88, 1997.



Chapter 4

The Crystal Structure of Aspergillusfumigatus Cyclophilin reveals 3DDomain Swapping of a Central Element

Andreas Limacher1,2, Daniel P. Kloer3, Sabine Flückiger2, Gerd Folkers1, RetoCrameri2, and Leonardo Scapozza1,∗,#

1Institute of Pharmaceutical Sciences, Department of Chemistry and Applied Biosciences,Swiss Federal Institute of Technology (ETH), 8093 Zurich, Switzerland2Swiss Institute of Allergy and Asthma Research (SIAF), 7270 Davos, Switzerland3Institut für Organische Chemie und Biochemie, Albert-Ludwigs-Universität, 79104Freiburg, Germany∗Correspondence: [email protected], Phone: +41 22 379 33 63,Fax: +41 22 379 33 60#Current address: Laboratoire de Chimie Thérapeutique, Section des Sciences Phar-maceutiques, Université de Genève, EPGL, Quai Ernest-Ansermet 30, 1211 Genève 4,Switzerland

Running title: Crystal structure of A. fumigatus cyclophilin

Key words: 3D domain swapping, misfolding, immunophilin, fungi, allergy

Publication: Structure, in press, 2005

93

94 CHAPTER 4. STRUCTURE OF A. FUMIGATUS CYCLOPHILIN

4.1 AbstractThe crystal structure of Aspergillus fumigatus cyclophilin (Asp f 11) was solved by themultiwavelength anomalous dispersion method and was re�ned to a resolution of 1.85 Åwith R and Rfree values of 18.9% and 21.4%, respectively. Many cyclophilin structureshave been solved to date, all showing the same monomeric conformation. In contrast, thestructure of A. fumigatus cyclophilin reveals dimerisation by 3D domain swapping andrepresents one of the �rst proteins with a swapped central domain. The domain swappedelement consists of two β-strands and a subsequent loop carrying a conserved tryptophan.The tryptophan binds into the active site, inactivating cis-trans isomerisation. Thismight be a means of biological regulation. The two hinge loops leave the protein proneto misfolding. In this context, alternative forms of 3D domain swapping that can lead toN- or C-terminally swapped dimers, oligomers and aggregates are discussed.

4.2 IntroductionCyclophilins (CyPs) constitute a family of cytosolic proteins involved in many biologicalprocesses. CyP is an enzyme that catalyses the peptidyl-prolyl cis-trans isomerisation(PPIase) [1, 2]. Belonging to the family of immunophilins, CyP binds the immunosup-pressive drug cyclosporin A (CsA) [3]. The complex of CyP with CsA binds and inhibitsthe protein phosphatase calcineurin [4], thus suppressing signal transduction in T cells [5].Therefore, CsA is one of the most important immunosuppressant drugs used for preven-tion of graft rejection after transplant surgery [6]. CyPs are also assumed to participate inother biological functions such as cell surface recognition [7] and heat-shock response [8].CyPs exist abundantly and ubiquitously in a broad range of organisms [9]. They have alsobeen isolated as IgE-binding proteins from several fungi [10�12], including CyP from themould Aspergillus fumigatus (Asp f 11) [13]. A. fumigatus is the etiologic agent identi�edin 80% of Aspergillus-related human diseases. The ubiquitous mould is considered asan opportunistic pathogen associated with an impressive list of pulmonary and allergiccomplications in humans and animals [14]. The ability of the active Asp f 11 to actas an allergen in vivo was shown by positive skin reactions of A. fumigatus-sensitisedindividuals [10]. Cross-reactivity between Asp f 11 and other CyPs from Malasseziasympodialis, Candida albicans, Saccharomyces cerevisiae and Homo sapiens were shownby Western blot analysis and inhibition ELISA [10].Several structures of CyPs have been solved (Table 4.1). Despite the fact that crystalswere grown under di�erent conditions such as PEG and ammonium sulfate, resulting in


Protein (ligand) Resolu- R-factors, Space Quaternary PDB Referencetion [Å] R/Rfree[%] group structure code

Human CyPA 1.63 18.0/- P212121 monomer 2CPL [16]Human CyPB (CsA) 1.85 16.0/- P212121 monomer 1CYN [17]Human CyPH (U4/U6 SnRNP) 2.00 19.5/21.2 P1211 monomer 1MZW [18]Human Nucl CyP20 2.00 17.0/22.4 P212121 monomer 1QOI [19]Human CyPJ 2.60 17.5/23.6 P3121 monomer 1XYH [20]E. coli CyP 2.10 16.0/- C2221 monomer 2NUL [unpublished]E. coli CyPB 1.80 20.1/22.5 P3221 monomer 1J2A [21]Murine CyPC (CsA) 1.64 19.7/- P1211 monomer 2RMC [22]B. malayi CyP1 1.95 19.9/23.3 P412121 monomer 1A58 [23]P. falciparum CyP19 (CsA) 2.10 15.0/19.0 P1 monomer 1QNG [24]C. elegans CyP3 1.80 21.5/28.6 P412121 monomer 1DYW [25]C. elegans CyP5 1.75 18.2/25.0 P61 monomer 1H0P [26]Bovine CyP40 1.80 17.8/25.6 C121 monomer 1IHG [27]S. cerevisiae CyPA 1.90 19.9/21.6 P1 monomer 1IST [unpublished]M. tuberculosis CyPA 2.60 21.2/22.9 P31 monomer 1W74 [28]A. fumigatus CyP; Asp f 11 1.85 18.9/21.4 P3121 3D d.s. dimer 2C3B [present work]

Table 4.1: Structures of cyclophilins that have been solved to date. One representative of eachprotein is listed (unligated, if available). A. fumigatus CyP (Asp f 11) is the only one forminga 3D domain swapping (3D d.s.) dimer.

di�erent unit cell dimensions and space groups, all of them are monomeric (Table 4.1).They share the same fold consisting of a β-barrel with eight antiparallel β-strands andtwo α-helices covering the bottom and top of the barrel. In contrast, the crystal structureof Asp f 11, presented here, reveals a dimer formed by 3D domain swapping of two centralβ-strands and a subsequent loop. 3D domain swapping is the event by which a monomerexchanges part of its structure with identical monomers to form an oligomer, where eachsubunit has a similar structure to the monomer [15].In the present work, inhibition of cis-trans isomerisation activity as a consequence of do-main swapping and as a potential way of functional regulation is discussed. Furthermore,alternative forms of 3D domain swapping leading to misfolded dimers and oligomers areaddressed.


4.3.1 Cloning, protein expression and puri�cationThe original full-length CyP from A. fumigatus was cloned by phage surface display [29]as N-terminal non-cleavable His-tagged protein, containing a linker of 9 amino acids be-


tween the His tag and the start methionine [13]. In order to clone a thrombin-cleavableHis-tagged protein, the Asp f 11 gene was ampli�ed from the original clone by PCRwith the primers 5' BamHI-1 5'-CCGCGGATCCATGTCTCAGGTCTTCTTC-3' and 3'HindIII-1 5'-GCCCAAGCTTTTACAGCTCACCACAGTTGAC-3' using the Deep VentPolymerase (NEB, Beverly, MA, USA). The PCR product was digested with BamHIand HindIII restriction endonucleases (NEB), cleaned with QIAquick PCR puri�cationkit (Qiagen, Hilden, Germany) and ligated into a modi�ed pQE32 vector containing anN-terminal His6 tag followed by a thrombin cleavage site (HHHHHHLVPRGS), whereGS corresponds to the BamHI site. The recombinant vector was transformed into E. colistrain XL1-Blue and the sequence of the insert veri�ed by DNA sequencing.The XL1-Blue cells were grown at 37◦C in LB medium to an OD600 of 0.6, induced with1 mM IPTG, harvested after another 15 h of incubation by centrifugation at 6000g for10 min at 4◦C and stored at -20◦C. The cell pellet was resuspended in lysis bu�er (50 mMNaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0) and lysed by French Press. Theinsoluble material was removed by centrifugation at 20,000g (20 min, 4◦C).The His-tagged recombinant protein was puri�ed by nickel a�nity chromatography usinga 5 ml HiTrap Chelating HP column (Amersham Pharmacia Biotech, Uppsala, Sweden).The protein was eluted in a linear bu�er gradient (10 - 250 mM imidazole, 50 mMNaH2PO4, 300 mM NaCl, pH 8.0). The N-terminal His tag was cleaved o� by thrombin(20 Units per mg protein) in 300 mM NaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5 by in-cubation for 18 h at 22◦C. The cleaved protein was further puri�ed by gel �ltration ona Superdex 75 column (FPLC, Pharmacia, Uppsala, Sweden) equilibrated with 100 mMNaCl, 50 mM Tris, pH 7.5. The eluted protein was diluted 1:10 with H2O and concen-trated to 16 mg/ml. Protein identity and purity were assessed using MALDI-TOF massspectroscopy and SDS-PAGE.The selenomethionine-substituted protein was expressed under minimal media conditionsthat inhibit methionine synthesis [30]. The XL1-Blue cells were incubated over night inLB medium at 37◦C and harvested at 6000g for 10 min. The pellet was washed twicein H2O and then inoculated into 1 l M9 medium (supplemented with the appropriateantibiotics, 1 mM MgSO4, 0.4% glucose, 0.00005% thiamine, 100 mg Lys, Phe and Threach, and 50 mg Ile, Leu, Val and SeMet each) at 37◦C until an OD600 of 0.45 wasreached. After induction with 1 mM IPTG, the cells were grown at 37◦C for another15 h. The SeMet protein was puri�ed and thrombin-digested by the same procedure asthe native protein, except that all bu�ers were supplied with 10 mM β-mercaptoethanol.The digested SeMet protein was further puri�ed by gel �ltration on a Superdex 75 col-umn (FPLC, Pharmacia) equilibrated with 100 mM NaCl, 50 mM Tris, 10 mM DTT,


pH 7.5. The eluted protein was diluted 1:10 with H2O, 10 mM DTT and concentrated to9 mg/ml. The incorporation of SeMet was veri�ed by MALDI-TOF mass spectroscopy.

4.3.2 CrystallisationCrystallisation was performed using the hanging drop vapour di�usion method at 23◦C.The protein was crystallised by mixing 3 µl of protein solution (7 mg/ml, 80 mM Ala-Pro) with an equal volume of crystallisation solution (13% ammonium sulfate, 0.1 MMES pH 6), equilibrating it against 500 µl of reservoir solution (50% ammonium sulfate,0.1 M sodium citrate pH 5.5). Crystals grew to a size of 300 x 300 x 150 µm within 3-5days. They belong to space group P3121, with cell parameters a=b=64.8 Å, c=156.3 Å,and contain two monomers per asymmetric unit.The selenomethionylated protein crystals were obtained under the same conditions adding5 mM DTT to the reservoir solution. They grew to 200 x 140 x 140 µm and belongalso to space group P3121, with cell parameters a=b=64.9 Å, c=157.0 Å. Crystals werecryoprotected by soaking them stepwise for 20 s in reservoir solution complemented with80 mM Ala-Pro, increasing the ethylene glycol concentration from 6 to 12 and to 19%.The crystals were �ash-cooled in a stream of gaseous nitrogen and measured at 100 K.

4.3.3 Data collection, phasing and re�nementData were obtained on the synchrotron beamline X06SA at Swiss Light Source (Villi-gen/CH). A dataset of a native crystal was collected to 1.85 Å resolution (Table 4.2).MAD data was collected with a selenomethionylated crystal to 2.05 Å resolution, inorder to determine the structure by the multiwavelength anomalous dispersion (MAD)technique (Table 4.2).Data were processed and scaled with DENZO and SCALEPACK of the HKL programpackage [31]. The selenium atom positions were found with SHELXD [32] using MADFA values calculated with XPREP (Bruker Nonius). Experimental phases were calcu-lated and re�ned with SHARP [33]. The initial electron density was improved by sol-vent �attening with 2-fold NCS-averaging using RESOLVE [34]. A polyalanine modelof human CyPA (PDB code 2CPL [16]) was �t into the improved electron density byphased molecular replacement using MOLREP [35], followed by manual rebuilding usingXtalView [36]. The structure was re�ned against the native dataset with REFMAC [37]as implemented in the CCP4 program suite [38]. Initially, NCS-restraints were used,which were stepwise released and �nally omitted. Water molecules were introduced usingARP [39]. Final rounds of re�nement were carried out using TLS re�nement, de�ning the


Native Se (In�ection) Se (Peak) Se (Remote)Data CollectionWavelength (Å) 0.99990 0.97988 0.97894 0.97128Unit cell axes a=b, c (Å) 64.83, 156.29 64.85, 157.01 64.86, 156.83 64.99, 157.37Resolution (Å) 45.6-1.85 (1.92-1.85) 38.3-2.05 (2.12-2.05) 38.3-2.05 (2.12-2.05) 38.4-2.27 (2.35-2.27)Unique re�ections 33,079 46,229 46,153 34,063Redundancy 5.9 5.6 5.6 5.0Completeness (%) 99.1 (99.9) 99.7 (99.6) 99.5 (99.7) 99.0 (93.2)Rsym (%) 8.7 (48.8) 4.5 (23.5) 4.7 (24.2) 6.1 (28.8)Average I/σ 19.2 (2.6) 20.2 (4.8) 20.5 (5.0) 17.4 (3.3)MAD PhasingPhasing power (iso/ano) - / 0.66 0.96 / 0.82 0.14 / 0.48Rcullis (iso/ano) - / 0.94 0.75 / 0.78 1.00 / 0.90FOM (SHARP/RESOLVE) 0.32 / 0.60Re�nement StatisticsResolution (Å) 45.6-1.85Number of re�ections 33,035Number of atoms 2308Rcryst (%) 18.9Rfree (%) 21.4Mean B factor (Å2): All atoms 43.7Main chain atoms (chain A/B) 40.8 / 40.9Side chain atoms (chain A/B) 46.1 / 46.2Solvent atoms 46.0Rmsd bond lengths (Å) 0.014Rmsd bond angles (◦) 1.95

Table 4.2: Data collection, phasing and re�nement statistics. Numbers in parentheses are forthe highest resolution shell. Rfree was calculated using a test set of 5%.

domain-swapped central part and the N-terminus before together with the C-terminusthereafter of each chain as separate TLS groups. Statistics from phasing and re�nementare provided in Table 4.2. The stereochemical quality of the �nal model was assessedwith PROCHECK [40] and WHATCHECK [41]. Secondary structure elements were as-signed automatically with DSSP [42]. Surface areas were calculated with the programNACCESS [43], an implementation of the Lee and Richards solvent accessibility algo-rithm [44], using a probe radius of 1.4 Å and a slice width of 0.01 Å. In order to obtaina good experimental map of the hinge regions (aa 126-129) between the two subunits,solvent �attening was redone without NCS averaging. The resulting electron densitymap leaves tracing unambiguously. Building of the model in a non-swapped way wasnot feasible using this experimental map. To assess the correctness of the tracing of thehinge region, a real space map correlation was calculated by overlapmap [38] and plottedalong the sequence, showing reasonable correlation coe�cients higher than 0.7 in themain chain of the connecting loops as well as the rest of the protein. Coordinates andstructure factors have been deposited in the Protein Data Bank (accession code 2C3B).


4.3.4 Characterisation of quaternary structure and activity assayThe molecular weight of the Asp f 11 protein in solution was determined by gel �ltrationon a Superdex 75 column (Pharmacia), equilibrated in 100 mM NaCl, 50 mM Tris, pH 7.5.100 µl of protein at two di�erent concentrations (2.0 and 7.5 mg/ml) were eluted withthe same bu�er at a �ow rate of 0.8 ml/min at 23◦C. Aprotinin, cytochrome C, carbonicanhydrase and BSA (molecular weight marker kit, Sigma, St. Louis, CA, USA) servedas marker proteins. Dimerisation behaviour was studied in a modi�ed crystallisationbu�er (20% ammonium sulfate, 100 mM MES, pH 6) using the same column and thesame parameters. The protein (at two di�erent concentrations; 2 and 7.5 mg/ml) wasincubated in the modi�ed crystallisation bu�er for three hours at RT before application.Oligomerisation behaviour of misfolded protein was studied in 100 mM NaCl, 50 mMTris, pH 7.5 using the same column and parameters. Misfolding was induced by freeze-thawing. The protein was puri�ed under reducing conditions (as described above) andfrozen at -20◦C in 100 mM NaCl, 50 mM Tris, 10 mM DTT, pH 7.5. After thawing atRT, the precipitated part (about 30%) was removed by centrifugation at 15,000g, 5 min,4◦C and the bu�er exchanged against a non-reducing bu�er (100 mM NaCl, 50 mM Tris,pH 7.5). The protein was applied on the column at 2 mg/ml.The nature of intra- or intermolecular disul�de bonding of protein, which was used forgel �ltration and crystallisation, as well as crystallised protein, which was dissolved inH2O, was analysed by reducing and non-reducing SDS-PAGE using 15% polyacrylamidegels. For reducing SDS-PAGE, the samples were mixed with 3x SDS sample bu�er,supplemented with 10% (v/v) β-mercaptoethanol and 0.5 M DTT and boiled at 95◦C for20 min. For non-reducing SDS-PAGE, samples were mixed with 3x SDS sample bu�ercontaining no reducing agents.Peptidyl-prolyl cis-trans isomerase activity of the monomeric and dimeric fraction of thegel�ltration experiment with modi�ed crystallisation bu�er was measured using the assayby Kofron et al. [45] with minor modi�cations. 25 µl of 0.5 µM protein solution (�nalconcentration 25 nM) were diluted with 437.5 µl 50 mM HEPES, 100 mM NaCl, pH 8(all solutions cooled on ice). Then, 25 µl of 10 mg/ml α-chymotrypsin (Sigma) in 1 mMHCl were added and the reaction was started by adding this mixture to 12.5 µl of 4 mMN -succinyl-Ala-Ala-Pro-Phe p-nitroanilide (Sigma) in tri�uoroethanol, 470 mM lithiumchloride in a semi-micro cuvette. The increase in absorbency at 390 nm was measuredfor 3 min at 0◦C in a Cary 50 Conc spectrophotometer. The background, which is dueto thermal cis-trans isomerisation of the substrate, was measured with a blank samplewithout protein.


4.4 Results and discussion

4.4.1 Structure determinationInitial crystals of the original protein, which contained a non-cleavable His tag and alinker region at the N-terminus, di�racted to about 3.0 Å resolution on a home source.Co-crystallisation with the dipeptide Ala-Pro improved the resolution by about 0.5 Å.Removal of the His tag and linker improved the resolution by a further 0.5 Å. The �nalprotein crystals, co-crystallised with Ala-Pro, di�racted to about 1.8 Å resolution at theSwiss Light Source (Villigen/CH), where all relevant datasets were collected.Considering the high homology to human CyPA (56% sequence identity), the struc-ture was expected to be solvable by molecular replacement. Surprisingly, no solutionwas found, neither with AMORE [46] nor with MOLREP [35]. Therefore, selenome-thionylated protein was expressed and crystallised in order to perform a multiwavelengthanomalous dispersion (MAD) experiment. The X-ray structure was solved by MAD-phasing and re�ned against native data to a resolution of 1.85 Å with R and Rfree valuesof 18.9% and 21.4%, respectively. 90.9% of the residues are in the most favored regionsof the Ramachandran plot, with the remaining ones located in the additional allowedregions. Data collection, phasing and re�nement statistics are shown in Table 4.2.There are two monomers per asymmetric unit. The �nal model reveals an intertwineddimer consisting of chain A and B. Both chains run from the start methionine to the lastresidue (Leu171) of the protein. The hinge region (aa 126-129) between the two subunitscan be traced unambiguously within a good quality electron density map resulting fromsolvent �attening done without NCS averaging. Correlation coe�cient values higherthan 0.7 resulting from the real space correlation analysis indicate the correctness of thetracing. Two loops are missing in the �nal model: Phe70 to Phe91 and Asn105 to Ser113in chain A and Asp69 to His95 and Asn105 to Gly112 in chain B. These loops are highly�exible resulting in a poor electron density. In chain A, the last residue of the thrombincleavage site (Ser0) is also visible. 123 water molecules and two sulfate ions have beenadded. They are all well de�ned in the electron density.In spite of the improvement in resolution, the ligand Ala-Pro is not visible in the electrondensity, neither in the active site nor on the surface. Due to dimerisation, the active siteis not accessible. Therefore, Ala-Pro cannot bind, despite its a�nity to CyP. The Kd ofAla-Pro for Asp f 11 is not known, but it is most probably comparable to the Kd of otherCyPs; Asp f 11 shows comparable activities to other CyPs and all the active site residuesare conserved. The enzymatic activity of Asp f 11 was demonstrated by its ability to


N-term

C-term

C-term

N-term

SS-bond

SS-bond

sulfate sulfate

1

1

2

2

1

1

2

6 5

4

3

2

3

45

6

7

7

8

8

W124

W124

2�

310

Figure 4.1: Cartoon representation of the 3D domain-swapped Asp f 11 dimer.Chain A and B are in red and blue, respectively. The conserved Trp124 is shown as cappedstick model.

catalyse the cis-trans isomerisation of N -succinyl-Ala-Ala-Pro-Phe p-nitroanilide [10].The human CyPA-Ala-Pro complex was obtained by soaking native crystals in 100 mMAla-Pro [47]. And with C. elegans CyP3, a Kd of 23.3 mM was measured for Ala-Pro [48]. Therefore, at a concentration of 80 mM Ala-Pro, one would expect binding ofthe dipeptide to Asp f 11. This is probably true for the monomeric form.

4.4.2 Overall structureThe intertwined chains A and B form two similar subunits. Subunit A consists of theN-terminal amino acids 1-69 and the C-terminal amino acids 128-171 of chain A plusthe domain swapped residues 96-127 of chain B. Subunit B consists of the N-terminalamino acids 1-68 and the C-terminal amino acids 128-171 of chain B plus the domainswapped residues 92-127 of chain A (Figure 4.1 and 4.2). Like in all other cyclophilins, thefold of each subunit consists of an eight-stranded antiparallel β-barrel and two α-helicescovering the top and the bottom of the barrel. In subunit A, there is an additionalsmall 310-helix formed by Thr122, Ser123 and Trp124 of chain B. In subunit B, Arg93 toLys97 of chain A adopt an α-helical conformation, whereas in subunit A and in all otherstructurally known cyclophilins, this stretch is coiled.


1 32 75648 1

1 32 75648 1

2

2

C

C

N

N

Figure 4.2: Topology plot of secondary structure elements of the Asp f 11 dimer, schematicallyshowing the 3D domain swapping.

Superposing human CyPA on either subunit of Asp f 11 reveals some structural deviation(Figure 4.3). The rmsd between the Cα positions of subunit A and CyPA is 1.18 Å, using acut-o� distance of 3.5 Å (120 out of 128 alignable pairs). For subunit B the correspondingrmsd is 1.26 Å (120 out of 130 alignable pairs). There is also some deviation betweensubunit A and B - mainly in the domain swapped part - as shown by the rmsd of 1.15 Å(no cut-o�). Using a 3.5 Å cut-o�, the rmsd between 130 out of 135 alignable pairs dropsto 0.65 Å.Compared to human CyPA, there are some structural di�erences due to residue inserts(see sequence alignment, Figure 4.4). One di�erence in sequence and structure is theprolongation of the �rst loop after the �rst β-strand by 5 residues. Another insert oftwo amino acids (Glu138/Lys139) leads to the elongation of helix 2 and a third insert(Asn160/Thr161) prolongs the loop after helix 2 (Figure 4.3).A further structural deviation is found in the loop after the �rst α-helix (R45-PAG-E49),a divergent loop, where some CyPs have an insert, e.g. C. elegans CyP3 [25]. Comparedto human CyPA and CyPB, which feature the sequence GEKGF in the correspondingregion and form a type I β-turn, the turn of Asp f 11 swings into the opposite directionforming a type II β-turn (Figure 4.3). The same conformation has been observed in thestructure of C. elegans CyP5 [26], featuring a similar sequence (KPKGE) as Asp f 11.The protein forms a disul�de bond between Cys43 of the N-terminus and Cys168 at thethird last position of the C-terminus. The cysteines oxidise very slowly over time. This


Figure 4.3: Structural comparison of Asp f 11 to human CyPA.Superposition of two independent monomers of human CyPA (2CPL) - one in green, one inyellow - on subunit A (core in red, swapped domain in blue) and on subunit B (core in blue,swapped domain in red) of Asp f 11, respectively. The view is the same as in Figure 4.1, butrotated by 90◦ perpendicular to the plane. The moieties of the protein diverging from humanCyPA have been numbered from 1 to 8.(1) The conformation of Trp124 of Asp f 11 is di�erent from Trp121 of human CyPA.(2) The end of the �rst hinge loop is α-helical in chain A of Asp f 11, whereas it is coiled inhuman CyPA.(3) The divergent loop of Asp f 11 forms a type II β-turn, whereas in CyPA, it forms a type Iβ-turn.(4) The �rst loop is prolongated in Asp f 11.(5) Helix 2 is prolongated in Asp f 11 by 2 residues.(6) The loop after helix 2 is elongated by another 2 residues.(7) The side chain carboxyl of Glu86 of human CyPA (homologous to Glu89 of Asp f 11)superposes on the sulfate ion in the Asp f 11 crystal structure.(8) Asp f 11 forms a disul�de bond, human CyPA doesn't.


== 1=== === 2==== ===== 1===== 3= = 4=

10 20 30 40 50 60 70 80

Asp f 11 ---MSQVFFDVEYAPVGTAETKVGRIVFNLFDKDVPKTAKNFRELCKRP-------AGEGYRESTFHRIIPNFMIQGGDFTRGNGTGGRSIYGDKFAD

human CyPA -MVNPTVFFDIAV-----DGEPLGRVSFELFADKVPKTAENFRALSTGEKG-------FGYKGSCFHRIIPGFMCQGGDFTRHNGTGGKSIYGEKFED

human CyPB +KVTVKVYFDLRIG-----DEDVGRVIFGLFGKTVPKTVDNFVALATGEKG-------FGYKNSKFHRVIKDFMIQGGDFTRGDGTGGKSIYGERFPD

C.elegans CyP3 -MSRSKVFFDITI-----GGKASGRIVMELYDDVVPKTAGNFRALCTGENGIGKSGKPLHFKGSKFHRIIPNFMIQGGDFTRGNGTGGESIYGEKFPD

C.elegans CyP5 +KVTDKVYFDMEIG-----GKPIGRIVIGLFGKTVPKTATNFIELAKKP-------KGEGYPGSKFHRVIADFMIQGGDFTRGDGTGGRSIYGEKFAD

S.cerevisiae CyPA ---MSQVYFDVEA-----DGQPIGRVVFKLYNDIVPKTAENFRALCTGEKG-------FGYAGSPFHRVIPDFMLQGGDFTAGNGTGGKSIYGGKFPD

P.falciparum CyP19 MSKRSKVFFDISID-----NSNAGRIIFELFSDITPRTCENFRALCTGEK-IGSRGKNLHYKNSIFHRIIPQFMCQGGDITNGNGSGGESIYGRSFTD

*:**: **: : *: . .*:* ** *.. : * ***:* ** ****:* :*:**.**** * *

= 5= = 6= == 7== ==== 2===== == 8===

90 100 110 120 130 140 150 160 170

Asp f 11 ENFSRKHDKKGILSMANAGPNTNGSQFFITTAVTSWLDGKHVVFGEVADEKSYSVVKEIEAL-GSSSGSVRSNTRPKIVNCGEL--------------

human CyPA ENFILKHTGPGILSMANAGPNTNGSQFFICTAKTEWLDGKHVVFGKVKE--GMNIVEAMERF-GSRNGKTSK--KITIADCGQLE-------------

human CyPB ENFKLKHYGPGWVSMANAGKDTNGSQFFITTVKTAWLDGKHVVFGKVLE--GMEVVRKVESTKTDSRDKPL--KDVIIADCGKIEVEKPFAIAKE---

C.elegans CyP3 ENFKEKHTGPGVLSMANAGPNTNGSQFFLCTVKTEWLDGKHVVFGRVVE--GLDVVKAVESN-GSQSGKPVK--DCMIADCGQLKA------------

C.elegans CyP5 ENFKLKHYGAGWLSMANAGADTNGSQFFITTVKTPWLDGRHVVFGKILE--GMDVVRKIEQTEKLPGDRPK--QDVIIAASGHIAVDTPFSVEREAVV

S.cerevisiae CyPA ENFKKHHDRPGLLSMANAGPNTNGSQFFITTVPCPWLDGKHVVFGEVVD--GYDIVKKVESL-GSPSGATKA--RIVVAKSGEL--------------

P.falciparum CyP19 ENFNMKHDQPGLLSMANAGPNTNSSQFFITLVPCPWLDGKHVVFGKVIE--GMNVVREMEKE-GAKSGYVKR--SVVITDCGEL--------------

*** :* * :****** :**.****: . ****:*****.: : . .:*. :* . :. .*.:

Figure 4.4: Structure based sequence alignment of Asp f 11 with homologous CyPs.The top line shows the secondary structure of Asp f 11 assigned by DSSP [42]. The 310-helixformed by chain B (Thr122-Trp124) and the additional α-helix 2' formed by chain A (Arg93-Lys97) are not indicated. The second line shows the sequence numbering of Asp f 11. A plussign at the N-terminus indicates truncation of the signal sequence. The 1st and 2nd hinge loopof Asp f 11 are underlined and the W-loop is in bold. The divergent loop (aa 45-49) is in italics.The aligned CyPs are human CyPA (PDB code 2CPL), human CyPB (1CYN), C. elegansCyP3 (1DYW), C. elegans CyP5 (1H0P), S. cerevisiae CyPA (1IST) and P. falciparum CyP19(1QNG) (Table 4.1).

results in a protein, which is half oxidised and half reduced after puri�cation and storageat 4◦C for several days. Treatment with oxidised glutathione has no e�ect, probably dueto the inaccessibility of the cysteines. The electron density re�ects the situation verywell. About 50% of the cysteines are visible in their reduced and 50% in their oxidisedform. The sulfur atoms of the reduced cysteines are separated by 4.58 Å and 4.49 Åin subunit A and B, respectively. Simple rotation about the Cα-Cβ cysteine side chainbonds towards each other allows disul�de bond formation with a bond length of 2.04 Åand chi torsion angles around 180◦ for all oxidised cysteines in both subunits. In theelectron density, no disorder or conformational change is observed in the proximity of thedisul�de bonds. Potential roles of this disul�de bond could be the stabilisation of theprotein in an oxidising environment or a signaling mechanism in response to oxidativestress, as hypothesised for C. elegans CyP3 [25]. Both cysteines are conserved in manycyclophilins (Figure 4.4).


4.4.3 3D domain swappingA striking feature is the dimerisation of the protein by 3D domain swapping. A centralsegment - the end of the �rst hinge loop (aa 92-99 in chain A, aa 96-99 in B), the 5th

and 6th β-strand (aa 100-118) and the subsequent loop (aa 119-127) - changes into theneighbouring subunit, thus displacing its corresponding element (Figure 4.1 and 4.2).3D domain swapping is quite rare. To date, about 60 structures of domain swappedproteins have been reported, most of them exchanging their N- or C-terminus [49]. Onlyfew proteins are known to swap a central part. For example, blood coagulant factor IX/X-binding protein, an anticoagulant isolated from the venom of the habu snake, consists oftwo homologous subunits linked by an intermolecular disul�de bond. The two subunitsform a heterodimer by exchanging a loop in the central part of the molecule [50]. The NC1Domain of type IV collagen forms a heterotrimer, consisting of two α1 chains and one α2chain, involving domain swapping of two central β-strands [51]. The structure of Asp f 11reported here also exchanges a central part, but forms a homodimer. Since monomericnon-swapping structures of homologous CyPs are known, and since the swapped domainin Asp f 11 adopts exactly the same conformation as the corresponding non-swappeddomain in homologous structures, this kind of domain swapping is termed quasi-domainswapping [49]. To our knowledge, Asp f 11 is the only quasi-domain swapped homodimerexchanging a central element.This kind of 3D domain swapping requires two hinge loops that adopt a di�erent con-formation in the domain swapped dimer and in the monomer. The second hinge loopis very short in Asp f 11. It runs from amino acid 126 to 129 (DGKH) and adopts adi�erent conformation in each chain. All backbone atoms of this hinge loop and the sidechains of the aspartate and the histidine are clearly de�ned in the experimental electrondensity, whereas the side chain of the lysine is not visible (Figure 4.5 A). All distances,angles and torsions in the hinge loop are chemically sound with torsion angles lyingin the most favoured or additional allowed regions of the Ramachandran plot. Thereare three H-bonds between the hinge loops of chain A and B (A:His129;N bonded withB:Leu125;O, A:Leu125;O with B:His129;NE2 and A:Gly127;N with B:Trp124;O). An ad-ditional H-bond is formed within the hinge loop of chain A (Trp124;O with His129;NE2)(Figure 4.5 A).The �rst hinge loop is very long, running from amino acid 68 to 99. In homologousstructures, this part is very well de�ned. It forms a long loop, which runs like a handlefrom one side of the molecule to the other side, connecting β-strand 4 and 5. In contrast,the larger part of this loop is disordered in Asp f 11 and thus not visible in the electron


A

B

Figure 4.5: W-loop bound to active site and 2nd hinge loop. (A) Electron density of the secondhinge loop and parts of the adjacent W-loop (aa 124-129). The experimental map is contouredat 1σ using solvent-�attened experimental phases calculated without NCS averaging. H-bondswithin and between the hinges of both chains are indicated by yellow, dashed lines. Residuesof chain A and B are coloured in violet and green, respectively. (B) Hydrophobic binding ofthe W-loop into its own active site. Trp124 and Leu125 of chain A sit into the hydrophobicpocket of subunit A making hydrophobic interactions with surrounding residues of both chains.In subunit B, Val121 and Trp124 of chain B sit into its active site. Position and conformation ofsome distinctive residues of human CyPA (2CPL) are shown for comparison. Two independentmonomers of CyPA were superimposed on either subunit of Asp f 11 (see Figure 4.3). Sidechains of CyPA superposed on subunit A and B are coloured in orange and yellow, respectively.


density. Therefore, its role as hinge loop is ambiguous. Nevertheless, the electron densityimplies that this loop must be arranged in a di�erent way than in homologous structures.First, a well-de�ned sulfate ion is positioned on top of α-helix 1 beside the pyrrolidinering of Pro33, where - in homologous structures - the side chain carboxyl of Glu89 issituated (Figure 4.3). This implies that Glu89 of the long loop of Asp f 11 is di�erentlypositioned. Both residues, Pro33 and Glu89, are highly conserved in all CyPs (Figure 4.4).Second, the visible end of the long loop of Asp f 11 adopts a di�erent conformationfrom the corresponding loops of homologous structures (Figure 4.3). Going backwardsfrom β-strand 5, this part of the loop folds back in homologous CyPs, running towardsβ-strand 4 within the same subunit. In Asp f 11, this part of the loop goes straight on,pointing towards β-strand 4 of the neighbouring subunit. It forms an α-helix in chain A(aa 93-97), whereas in chain B, it adopts a rather extended conformation (aa 96-99).Nevertheless, the missing electron density leads to two putative scenarios in respect todomain swapping. First, this long loop represents a hinge loop, thus swapping - togetherwith the second hinge loop - the central element, as suggested in this work (Figure 4.6 A).Second, this loop does not represent a hinge loop. Thus, the remaining second hinge loopwould domain-swap the whole C-terminus (aa 128-171) (Figure 4.6 B) instead of thecentral element. Taking distances and crystal packing into account, both cases would bepossible; β-strand 4 connected to β-strand 5 of the neighbouring subunit or to β-strand 5of the same subunit. In the �rst case, we end up with an intramolecular disul�de bond(Cys43-Cys168) between the N- and the C-terminus, as shown in Figure 4.6 A. Thesecond case would lead to an intermolecular disul�de bond (Figure 4.6 B).The second case can be ruled out easily by a denaturing, non-reducing SDS-PAGE.Figure 4.7 shows that the crystallised protein runs as two very close bands, both atthe height of the monomer and both with about the same intensity. The lower bandrepresents reduced protein, the upper band oxidised protein containing an intramoleculardisul�de bond. This �nding re�ects the situation in the electron density, where 50% ofthe cysteines are reduced and 50% oxidised. The disul�de bonding is intramolecular,connecting N- and C-terminus within the same chain. Therefore, the long loop mustrepresent a hinge loop, swapping - together with the second hinge loop - the centralelement.Interestingly, both regions depicted here as hinge regions are conserved in length andsequence between the homologous proteins (Figure 4.4) suggesting that domain swappingcould also occur in other CyPs. The structural analysis clearly showed that 3D domainswapping is the only solution, allowing �tting of the electron density map and structuralre�nement. Thus, a similar dimerisation arrangement would be expected for homologous


Figure 4.6: Cartoons of putative forms of 3D domain swapping.The �rst hinge loop, which is not visible in the crystal structure, is in dashes. The second hingeloop is fully drawn.(A) 3D domain swapping of the central element, as proposed in the crystal structure, leads tointramolecular disul�de bonds (shown in yellow).(B) 3D domain swapping of the C-terminus would lead to intermolecular disul�de bonds; thiscase can be ruled out for the crystal structure by a non-reducing SDS-PAGE (Figure 4.7).(B-E) C-terminal (B) and N-terminal domain swapping (C) can lead to misfolded dimers, whichare covalently linked by intermolecular disul�de bonds. A combination of central, C-terminaland N-terminal domain swapping can lead to covalently linked trimers (D) and tetramers (E).

proteins undergoing the same event, while for the structures showing the monomeric state(Table 4.1) di�erent crystal-packing arrangements have been observed.There is one other small loop not visible in the electron density, connecting β-strand 5and 6 (aa 105-112). But here - considering distances - the connectivity is unambiguous.

4.4.4 Dimer interface and binding of W-loop into active siteAsp f 11 dimerises by 3D domain swapping, which leads to a highly intertwined molecule.The calculated surface area of the interface between chain A and B is 2310 Å2 permonomer, which makes up about one fourth of the whole surface area of chain A (8880 Å2).The interface formed by purely non-swapped parts - excluding hinges and swapped do-mains - is very small at 180 Å2 (less than 8% of the whole interface). Thus, mostinteractions in the dimer interface are contributed by either of the swapped domains.The swapped β-strands 5 and 6 replace the corresponding elements in the neighbouringsubunit, adopting exactly the same conformation (Figure 4.3). Together with β-strand 3and 4, they form the hydrophobic pocket where the active site for the cis-trans isomeri-sation function is situated. The subsequent domain swapped loop (aa 119-125) contains


66.2

14.4

21.5

31.0

45.0

97.4116.3

1 2 3 4 5 6 7 8 9 10

reducing non-reducing

kDa

Figure 4.7: Reducing and non-reducing SDS-PAGE of crystallised and misfolded Asp f 11protein.Molecular weights of the marker proteins (lane 3) are indicated on the left side. Crystallisedprotein under reducing (lane 5) and non-reducing condition (lane 7). As a control, proteinbefore crystallisation under reducing (lane 4) and non-reducing condition (lane 6), each 3 µg.Misfolded protein under non-reducing conditions at 4 µg (lane 8), 8 µg (lane 9) and 16 µg (lane10). As a control, misfolded protein under reducing conditions, 8 µg (lane 1). Reduced proteinas a reference (lane 2). Under non-reducing conditions, the upper band around the 21.5 kDamarker represents the oxidised form of the protein whereas the lower band indicates the reducedone. The small di�erence in running behavior between reducing and non-reducing condition isdue to the presence or absence of DTT/ β-mercaptoethanol as shown by the analysis of partiallyreduced batches of protein (data not shown).

a highly conserved tryptophan (W124) and is thus called the W-loop. It also replacesthe corresponding W-loop in the neighbouring subunit, but - in this way - comes to liein its own active site. This additional intramolecular interaction is not observed in otherCyPs. The conformation of the two corresponding W-loops compared to each other andto the homologous W-loop in human CyPA is not identical, but similar (Figure 4.3).The binding mode of the W-loop within the active site is di�erent in each subunit of Aspf 11. In subunit A, the side chains of Trp124 and Leu125 of chain A make hydropho-bic interactions with the side chains of Ile60, Phe63, Met64 and His129 of chain A andwith the domain swapped side chains of Phe116, Thr118, Val121, Trp124 and Leu125 ofchain B (Figure 4.5 B). In subunit B, the side chains of Trp124 and Val121 of chain B


make hydrophobic interactions with the side chains of residues Phe63, Met64 and His129of chain B and Phe63, Ser102, Ala104, Phe116, Thr118, Thr122 and Leu125 of chain A(Figure 4.5 B). Interestingly, the position and orientation of the side chain of Trp124 isdi�erent in either subunit. In subunit B, the benzene ring of Trp124 makes π-π interac-tions with the imidazole ring of His129, while in subunit A it only makes Van-der-Waalsinteractions.The side chains of the W-loop, which interact with the active site (Trp and Leu in chain A,Trp and Val in chain B) are completely buried in Asp f 11. In homologous structures,these residues lie on the rim of the active site, and the side chains of the tryptophan andthe valine are solvent accessible.The binding of Trp and Leu/Val within the active site of Asp f 11 leads to a change inthe side chain conformation and/or in the position of the Cα atoms of all the residuesinvolved in the hydrophobic pocket, when compared to superposed human CyPA. Themost distinctive displacements include the Cα positions of A:Phe63 (2.0 Å), B:Phe63(2.2 Å), B:Val121 (2.6 Å), A:Trp124 (3.0 Å) and the side chain positions of A:Phe116;CZ(5.2 Å), B:Phe116;CZ (5.2 Å) and B:His129;CH2 (5.4 Å), to name but a few (Figure4.5 B). These �ndings explain the relatively high rmsd between the human CyPA andthe chain A and B of Asp f 11 (see above).

4.4.5 Biological relevanceIs the domain swapped Asp f 11 dimer physiologically relevant or is it just a crystal-lographic artefact? For domain swapped proteins in general, both cases have been re-ported [49]. Domain swapping was originally proposed to be a mechanism for the emer-gence of oligomeric proteins and as a means of functional regulation [52]. It can also bea potentially harmful process leading to misfolding, aggregation and amyloid formationof some proteins, e.g. cystatin C [53,54] and human prion protein [55]. Several artefactsof domain swapped proteins have been reported, where the biological function for thedimers are unknown [49]. They were obtained under nonphysiological low pH and of-ten, the domain swapping only occurred as a consequence of the truncation of the wholeprotein.The Asp f 11 crystals were grown under physiological pH, but under nonphysiological highsalt and high protein concentrations. Despite these facts, dimerisation cannot be excludedin vivo due to e�ects like local concentration, other binding partners or macromolecularcrowding in cells [56]. Binding of other molecules to Asp f 11 or an environmental changecould induce domain swapping and dimer formation. The e�ect of other macromolecules


on a speci�c macromolecule in cells has been studied and termed macromolecular crowd-ing [56]. Macromolecular crowding increases local protein concentration and facilitatesprotein oligomerisation [57,58].In vitro, the Asp f 11 protein is clearly monomeric under physiological conditions, evenat high protein concentration, as shown by gel �ltration (Figure 4.8 A). The proteinelutes at 12.55 ml forming one single peak. Calibration with marker proteins yields amolecular weight of 14.9 kDa, which is in agreement with the calculated monomer weightof 18.9 kDa. Under these conditions, the protein also shows cis-trans isomerisationactivity [10].However, dimerisation can be induced in solution, when using a modi�ed crystallisa-tion bu�er with an ammonium sulfate concentration of 20% (Figure 4.8 B). Besides themonomer peak, a second peak corresponding to a dimer is obtained. The height of thesecond peak is dependent on the concentration of the protein. The higher the proteinconcentration, the bigger is the fraction of the dimeric form. Gel �ltration of protein at7.5 mg/ml results in a dimer fraction of about 10%. Both peaks exhibit shoulders, whichmight represent re-equilibrating protein or intermediate states, such as partly unfoldedprotein.To assess the peptidyl-prolyl cis-trans isomerase activity of the monomeric and dimericfraction of the gel �ltration experiment performed with modi�ed crystallisation bu�er,enzymatic activity was measured using the assay by Kofron et al. [45]. The enzymaticassay clearly shows that the dimer fraction is inactive while the monomer fraction is active(Figure 4.8 D). This indicates that the dimer observed in the gel �ltration experiment isarranged in a similar way as the dimer seen in the crystal structure. Further, it impliesa putative regulation mechanism.In general, CyPs are active in their monomeric form. No dimerisation and thus, nopotential role for this hypothetical state have been reported to date. For Asp f 11, boththe monomer and the dimer form can be demonstrated (Figure 4.8 A/B). Dimerisation, asshown by the crystal structure, obviously inactivates the cis-trans isomerisation function,since both active sites are occupied by their own W-loops. Therefore, dimerisation of Aspf 11 in vivo could be a means of regulating its own biological function.

4.4.6 MisfoldingTwo �exible hinge loops could leave the protein prone to misfolding. 3D domain swappingcan be a harmful mechanism for misfolding, aggregation and amyloid formation [15].Indeed, misfolding of Asp f 11 can be induced by freeze-thawing the protein in its re-


A B

C

D

Figure 4.8: Gel �ltration of normal and misfolded Asp f 11 protein and peptidyl-prolyl cis-transisomerase activity of the monomeric and dimeric fraction.(A) Protein at 2 mg/ml (bold line) and 7.5 mg/ml (thin line) under physiological conditions(100 mM NaCl, 50 mM Tris, pH 7.5); the chromatogram shows only a monomer peak, whichelutes at 12.55 ml. Calibration with four marker proteins yields a molecular weight of 14.9 kDa(R2 = 0.98).(B) Protein at 2 mg/ml (bold line) and 7.5 mg/ml (thin line) in a modi�ed crystallisation bu�er(20% ammonium sulfate, 0.1 M MES, pH 6); the chromatogram shows an additional, smallerdimer peak, which is concentration dependent. The main fraction elutes at 13.77 ml whereasthe second peak elutes at 10.68 ml. The high salt conditions used cause the observed shift inthe elution times. Despite a low correlation factor (R2 = 0.94), calibration with the four markerproteins indicates a monomeric state for the major peak and a dimeric state for the minor peak.(C) Misfolded protein at 2 mg/ml under physiological conditions (100 mM NaCl, 50 mM Tris,pH 7.5). The major peak elutes at 12.5 ml corresponding to the monomeric state. The minorpeak, the shoulder to the right and the shoulder to the left elute at 10.05, 10.30 and 9.31 ml,which results in calibrated weights of 34.2, 31.5 and 43.7 kDa, respectively, indicating at leastthe formation of dimers and possibly a trimer.(D) Peptidyl-prolyl cis-trans isomerisation by monomeric (bold line) and dimeric (normal line)Asp f 11 and, as a negative control, in the absence of protein (thin line). The substrate N -succinyl-Ala-Ala-Pro-Phe p-nitroanilide was isomerised, the trans-isomer was cleaved by α-chymotrypsin and the increase in absorbency measured at 390 nm. The monomer fractionis active, whereas the dimer fraction is inactive.


duced form. Freeze-thawing of oxidised protein does not lead to unfolding and refold-ing/misfolding, probably due to the stabilisation of the protein by the intramoleculardisul�de bond. After freeze-thawing, about 30% of the reduced protein precipitated.The remaining soluble part was changed into a non-reducing bu�er, in order to allowoxidation. A subsequent gel �ltration under physiological conditions clearly showed ab-sorption peaks in the dimer- to tetramer-range, in addition to the monomer peak (Fig-ure 4.8 C). Analysis of the gel �ltration fractions by non-reducing SDS-PAGE showsmonomers as well as covalently linked dimers, trimers and tetramers (data not shown).The ratio between the peak heights was independent of the protein concentration; theoligomer peaks also appear at low concentration (data not shown). The protein usedfor gel �ltration was analysed by a non-reducing, denaturing SDS-PAGE (Figure 4.7).Again, the gel showed two close bands at the monomeric weight, which correspond tonon-oxidised and to intramolecularly oxidised protein. Moreover, there are bands visibleat the dimer, trimer, tetramer and at higher oligomer masses. These bands must repre-sent oligomers covalently linked by intermolecular disul�de bonds. They disappear underreducing conditions (Figure 4.7, lane 1), as expected for oligomers formed by intermolec-ular disul�de bonds. These results suggest that partial unfolding and refolding occursduring freeze-thawing under reducing conditions.The observed covalently linked multimers can be explained by alternative forms of 3D do-main swapping. The protein can refold/misfold into a dimer, where the whole C-terminusafter the second hinge loop or the whole N-terminus before the �rst hinge loop is domainswapped (Figure 4.6 B and 4.6 C). These two kinds of domain swapping only involve onehinge loop each. After oxidation, the two chains of the dimer will be covalently linked.A combination of N-terminal, C-terminal and/or central domain swapping can lead tolinear trimers and tetramers (Figure 4.6 D/E) and also to cyclic tetramers. Linear mul-timerisation could carry on to higher oligomers, aggregates and eventually precipitate,which indeed can be observed in vitro after freeze-thawing. Therefore, incorrect 3D do-main swapping could serve as a model for the mechanism of misfolding and precipitationof proteins in general.Another explanation for the covalently linked multimers could be unspeci�c disul�debonding between independent, non-swapping subunits, but this seems very unlikely.There are only two cysteines in Asp f 11, which can form the before mentioned disul�debond. Both cysteines are involved in secondary structure elements. Cys43 is situated inα-helix 1, Cys168 in β-sheet 8. This makes the whole region more rigid, despite the factthat Cys168 lies close to the C-terminus. The sulfur atoms are completely inaccessibleby solvent as calculated by NACCESS [43], both in the reduced and the oxidised form.


During freeze-thawing under reducing conditions, neither a speci�c, nor a non-speci�cdisul�de bond can form. Once the protein is refolded and after the bu�er is exchangedagainst a non-reducing bu�er, disul�de bonding is more likely to occur within the samesubunit, due to solvent inaccessibility of the cysteines.One wonders, why nature preserves features such as two �exible hinge loops and - asobserved in all other CyP structures - a solvent accessible tryptophan, which leave theprotein highly prone to misfolding. The two hinges, as well as the tryptophan and theleucine of the W-loop are highly conserved among all CyPs (Figure 4.4). This leadsto the assumption that the domain swapped protein, as observed in the present crystalstructure, could have a biological meaning, but its biological role still has to be found.Further work could focus on the characterisation of the di�erent misfolded multimersand on homologous CyPs, in order to investigate, if they show the same multimerisationbehaviour as Asp f 11.

AcknowledgementsWe thank T. Tomizaki from the Swiss Light Source for his great technical support andD. Kostrewa for the help in collecting the native dataset and for his very useful tips. Wethank M. Grütter and P. Mittl for their support and the opportunity to collect test data ontheir home source. We also thank M. Birringer for kindly providing the modi�ed pQE32vector and T. Schürpf for MALDI-TOF mass spectroscopy measurements. This workwas performed at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland.It was supported by the Swiss National Science Foundation grant No. 3100-063381/2and by the OPO foundation, Zürich.

Bibliography[1] G. Fischer, B. Wittmann-Liebold, K. Lang, T. Kiefhaber, and F. X. Schmid. Cyclophilin and

peptidyl-prolyl cis-trans isomerase are probably identical proteins. Nature, 337(6206):476�8, 1989.


[3] R. E. Handschumacher, M. W. Harding, J. Rice, R. J. Drugge, and D. W. Speicher. Cyclophilin: aspeci�c cytosolic binding protein for cyclosporin A. Science, 226(4674):544�7, 1984.


BIBLIOGRAPHY 115

[5] J. Jain, P. G. McCa�rey, Z. Miner, T. K. Kerppola, J. N. Lambert, G. L. Verdine, T. Curran, andA. Rao. The T-cell transcription factor NFATp is a substrate for calcineurin and interacts with Fosand Jun. Nature, 365(6444):352�5, 1993.

[6] J. F. Borel. Pharmacology of cyclosporine (sandimmune). IV. Pharmacological properties in vivo.Pharmacol. Rev., 41(3):259�371, 1990.

[7] S. K. Anderson, S. Gallinger, J. Roder, J. Frey, H. A. Young, and J. R. Ortaldo. A cyclophilin-relatedprotein involved in the function of natural killer cells. Proc. Natl. Acad. Sci. USA, 90(2):542�6,1993.

[8] K. Sykes, M. J. Gething, and J. Sambrook. Proline isomerases function during heat shock. Proc.Natl. Acad. Sci. USA, 90(12):5853�7, 1993.

[9] T. Hunter. Prolyl isomerases and nuclear function. Cell, 92(2):141�3, 1998.


[11] W. E. Horner, G. Reese, and S. B. Lehrer. Identi�cation of the allergen Psi c 2 from the basid-iomycete Psilocybe cubensis as a fungal cyclophilin. Int. Arch. Allergy Immunol., 107(1-3):298�300,1995.


[13] R. Crameri. Epidemiology and molecular basis of the involvement of Aspergillus fumigatus in allergicdiseases. Contrib. Microbiol., 2:44�56, 1999.

[14] E. J. Bardana, Jr. The clinical spectrum of aspergillosis - part 2: classi�cation and description ofsaprophytic, allergic, and invasive variants of human disease. Crit. Rev. Clin. Lab. Sci., 13(2):85�159, 1981.

[15] F. Rousseau, J. W. Schymkowitz, and L. S. Itzhaki. The unfolding story of three-dimensionaldomain swapping. Structure, 11(3):243�51, 2003.

[16] H. Ke. Similarities and di�erences between human cyclophilin A and other beta-barrel structures.Structural re�nement at 1.63 Å resolution. J. Mol. Biol., 228(2):539�50, 1992.

[17] V. Mikol, J. Kallen, and M. D. Walkinshaw. X-ray structure of a cyclophilin B/cyclosporin complex:comparison with cyclophilin A and delineation of its calcineurin-binding domain. Proc. Natl. Acad.Sci. USA, 91(11):5183�6, 1994.

[18] U. Reidt, M. C. Wahl, D. Fasshauer, D. S. Horowitz, R. Luhrmann, and R. Ficner. Crystal structureof a complex between human spliceosomal cyclophilin H and a U4/U6 snRNP-60K peptide. J. Mol.Biol., 331(1):45�56, 2003.

[19] U. Reidt, K. Reuter, T. Achsel, D. Ingel�nger, R. Luhrmann, and R. Ficner. Crystal structure of thehuman U4/U6 small nuclear ribonucleoprotein particle-speci�c SnuCyp-20, a nuclear cyclophilin.J. Biol. Chem., 275(11):7439�42, 2000.

[20] L. L. Huang, X. M. Zhao, C. Q. Huang, L. Yu, and Z. X. Xia. Structure of recombinant humancyclophilin J, a novel member of the cyclophilin family. Acta Cryst., D61(Pt 3):316�21, 2005.


[21] M. Konno, Y. Sano, K. Okudaira, Y. Kawaguchi, Y. Yamagishi-Ohmori, S. Fushinobu, and H. Mat-suzawa. Escherichia coli cyclophilin B binds a highly distorted form of trans-prolyl peptide isomer.Eur. J. Biochem., 271(18):3794�803, 2004.

[22] H. Ke, Y. Zhao, F. Luo, I. Weissman, and J. Friedman. Crystal structure of murine cyclophilin Ccomplexed with immunosuppressive drug cyclosporin A. Proc. Natl. Acad. Sci. USA, 90(24):11850�4, 1993.

[23] P. Taylor, A. P. Page, G. Kontopidis, H. Husi, and M. D. Walkinshaw. The X-ray structure of adivergent cyclophilin from the nematode parasite Brugia malayi . FEBS Lett., 425(2):361�6, 1998.

[24] M. R. Peterson, D. R. Hall, M. Berriman, J. A. Nunes, G. A. Leonard, A. H. Fairlamb, and W. N.Hunter. The three-dimensional structure of a Plasmodium falciparum cyclophilin in complex withthe potent anti-malarial cyclosporin A. J. Mol. Biol., 298(1):123�33, 2000.

[25] J. Dornan, A. P. Page, P. Taylor, S. Wu, A. D. Winter, H. Husi, and M. D. Walkinshaw. Biochemicaland structural characterization of a divergent loop cyclophilin from Caenorhabditis elegans. J. Biol.Chem., 274(49):34877�83, 1999.

[26] N. C. Picken, S. Eschenlauer, P. Taylor, A. P. Page, and M. D. Walkinshaw. Structural and biologicalcharacterisation of the gut-associated cyclophilin B isoforms from Caenorhabditis elegans. J. Mol.Biol., 322(1):15�25, 2002.

[27] P. Taylor, J. Dornan, A. Carrello, R. F. Minchin, T. Ratajczak, and M. D. Walkinshaw. Twostructures of cyclophilin 40: folding and �delity in the TPR domains. Structure, 9(5):431�8, 2001.

[28] L. M. Henriksson, P. Johansson, T. Unge, and S. L. Mowbray. X-ray structure of peptidyl-prolylcis-trans isomerase A from Mycobacterium tuberculosis. Eur. J. Biochem., 271(20):4107�13, 2004.

[29] R. Crameri, R. Jaussi, G. Menz, and K. Blaser. Display of expression products of cDNA librarieson phage surfaces. A versatile screening system for selective isolation of genes by speci�c gene-product/ligand interaction. Eur. J. Biochem., 226(1):53�8, 1994.

[30] G. D. Van Duyne, R. F. Standaert, P. A. Karplus, S. L. Schreiber, and J. Clardy. Atomic struc-tures of the human immunophilin FKBP-12 complexes with FK506 and rapamycin. J. Mol. Biol.,229(1):105�24, 1993.


[32] T. R. Schneider and G. M. Sheldrick. Substructure solution with SHELXD. Acta Cryst.,D58(Pt 10):1772�9, 2002.

[33] E. de La Fortelle and G. Bricogne. Maximum-likelihood heavy-atom parameter re�nement formultiple isomorphous replacement and multiwavelength anomalous di�raction methods. MethodsEnzymol., 276:472, 1997.

[34] T. C. Terwilliger. Reciprocal-space solvent �attening. Acta Cryst., D55(11):1863�71, 1999.



BIBLIOGRAPHY 117









[45] J. L. Kofron, P. Kuzmic, V. Kishore, E. Colon-Bonilla, and D. H. Rich. Determination of ki-netic constants for peptidyl prolyl cis-trans isomerases by an improved spectrophotometric assay.Biochemistry, 30(25):6127�34, 1991.

[46] J. Navaza. AMoRe - an automated package for molecular replacement. Acta Cryst., A50:157�63,1994.

[47] H. Ke, D. Mayrose, and W. Cao. Crystal structure of cyclophilin A complexed with substrateAla-Pro suggests a solvent-assisted mechanism of cis-trans isomerization. Proc. Natl. Acad. Sci.USA, 90(8):3324�8, 1993.

[48] S. Wu Sy, J. Dornan, G. Kontopidis, P. Taylor, and M. D. Walkinshaw. The �rst direct determinationof a ligand binding constant in protein crystals. Angew. Chem. Int. Ed. Engl., 40(3):582�6, 2001.

[49] Y. Liu and D. Eisenberg. 3D domain swapping: as domains continue to swap. Protein Sci.,11(6):1285�99, 2002.

[50] H. Mizuno, Z. Fujimoto, M. Koizumi, H. Kano, H. Atoda, and T. Morita. Structure of coagulationfactors IX/X-binding protein, a heterodimer of C-type lectin domains. Nat. Struct. Biol., 4(6):438�41, 1997.

[51] M. Sundaramoorthy, M. Meiyappan, P. Todd, and B. G. Hudson. Crystal structure of NC1 do-mains. Structural basis for type IV collagen assembly in basement membranes. J. Biol. Chem.,277(34):31142�53, 2002.

[52] M. J. Bennett, M. P. Schlunegger, and D. Eisenberg. 3D domain swapping: a mechanism foroligomer assembly. Protein Sci., 4(12):2455�68, 1995.

[53] R. Janowski, M. Kozak, E. Jankowska, Z. Grzonka, A. Grubb, M. Abrahamson, and M. Jaskol-ski. Human cystatin C, an amyloidogenic protein, dimerizes through three-dimensional domainswapping. Nat. Struct. Biol., 8(4):316�20, 2001.


[54] R. A. Staniforth, S. Giannini, L. D. Higgins, M. J. Conroy, A. M. Hounslow, R. Jerala, C. J. Craven,and J. P. Waltho. Three-dimensional domain swapping in the folded and molten-globule states ofcystatins, an amyloid-forming structural superfamily. EMBO J., 20(17):4774�81, 2001.

[55] K. J. Knaus, M. Morillas, W. Swietnicki, M. Malone, W. K. Surewicz, and V. C. Yee. Crystalstructure of the human prion protein reveals a mechanism for oligomerization. Nat. Struct. Biol.,8(9):770�4, 2001.

[56] A. P. Minton. The in�uence of macromolecular crowding and macromolecular con�nement onbiochemical reactions in physiological media. J. Biol. Chem., 276(14):10577�80, 2001.

[57] R. Lindner and G. Ralston. E�ects of dextran on the self-association of human spectrin. Biophys.Chem., 57(1):15�25, 1995.

[58] G. Rivas, J. A. Fernandez, and A. P. Minton. Direct observation of the self-association of diluteproteins in the presence of inert macromolecules at high concentration via tracer sedimentationequilibrium: theory, experiment, and biological signi�cance. Biochemistry, 38(29):9379�88, 1999.

Chapter 5

Structural Aspects of Cross-ReactiveAllergens

Sabine Flückiger1, Andreas Limacher1,2, Andreas G. Glaser1, LeonardoScapozza2, and Reto Crameri1

1Swiss Institute of Allergy and Asthma Research (SIAF), Davos, Switzerland2Institute of Pharmaceutical Sciences, Swiss Federal Institute of Technology (ETH),Zurich, Switzerland

Correspondence to: Prof. Reto Crameri, PhD, Head Molecular Allergology, Swiss Insti-tute of Allergy and Asthma Research (SIAF), Obere Strasse 22, CH-7270 Davos, Switzer-land. Phone: +41 81 410 08 48, Fax: +41 81 410 08 40, E-mail: [email protected].

Running title: Cross-reactive allergens

Publication: Recent Res. Devel. Allergy & Clin. Immunol., 5:57-75, 2004.

119

120 CHAPTER 5. STRUCTURAL ASPECTS OF ALLERGENS

5.1 AbstractIgE-mediated cross-reactivity is a frequently observed phenomenon in clinical practiceand results from sharing antigenic determinants by two antigens. Cross-reactive aller-gens are conserved proteins that show signi�cant similarity on the level of both, pri-mary and tertiary structure. These proteins, phylogenetically conserved between distantspecies, are also termed pan-allergens. Among many others, typical representatives in-clude manganese-dependent superoxide dismutases, acidic ribosomal P2 proteins, pro�linsand cyclophilins. Interestingly, some patients with severe long lasting atopic diseases dis-play IgE-mediated autoreactivity to the human homologues of these proteins, suggestingthat the human proteins may play a role in the exacerbation and/or perpetuation ofchronic allergic reactions. Molecular biology methods have allowed cloning and identi-�cation of an impressive list of di�erent allergenic proteins and the three-dimensionalstructures of an increasing number of allergens have been solved by X-ray crystallog-raphy or NMR spectroscopy. Crystal structures of allergens cannot (yet) answer thequestion why some proteins are allergenic while others are not as no common character-istic structural or functional features have become evident. Allergens belong to a widevariety of di�erent protein families including intracellular and secreted enzymes, struc-tural proteins and proteins without known biochemical function covering many di�erentfolds and combinations thereof. However, three-dimensional structures of allergens havesubstantially contributed to our understanding of cross-reactivity between homologousproteins. While the only method allowing a complete de�nition of a B-cell epitope isco-crystallisation of the allergen with a monoclonal antibody Fab fragment and solvingthe structure of the complex, a simple approach to identify cross-reactive B-cell epitopesis the determination of shared features of cross-reactive allergens on sequence and struc-ture level. Conserved amino acids that are solvent exposed in cross-reactive structuresare potentially involved in IgE-mediated cross-reactivity. In conclusion, the only knowncommon feature of allergens is their ability to cross-link IgE antibodies on the surfaceof e�ector cells leading to allergic reactions. Knowledge of cross-reactive structures hasdiagnostic and, perhaps, therapeutic implications contributing to reduce the number ofdi�erent molecules needed for a component resolved diagnosis of allergic conditions as aprerequisite for immunotherapeutic intervention based on patient tailored vaccines.


5.2 IntroductionAllergy results from inappropriate IgE immune responses to exposure to environmentalantigens occurring in predisposed individuals [1]. The symptoms associated with allergicreactions to complex sources of exposure like food, moulds, mites and pollens can betraced back to a limited number of allergenic components present in the extracts [2].Modern molecular biology allows rapid cloning and identi�cation of allergenic moleculesbased on phage surface display of cDNA libraries and high density arrays [3]. The elu-cidation of hundreds of sequences encoding allergens allows, combined with homologysearches, a �rst in silico identi�cation of molecular families of homologous allergens po-tentially involved in allergic reactions. Based on this information, old clinical observa-tions can gradually be explained at molecular level. It has been well known for manydecades that allergic individuals often react to various allergen sources. Monosensitisa-tion, with exception of insect sting allergy [4, 5], is rarely observed in clinical practiceas a consequence of homologous cross-reactive molecules present in di�erent allergenicsources. IgE-mediated cross-reactivity results from sharing antigenic determinants bytwo antigens, which might occur quite frequently among phylogenetically highly con-served proteins. However, to cross-link IgE receptors on the surface of e�ector cellsrequired for mediator release [6], a minimum of two di�erent IgE-binding epitopes permolecule pair is required. Antibodies originally raised against a speci�c epitope of anallergen can recognise common epitopes present on structurally related allergens (Fig-ure 5.1). In 1942, Tuft and Blumenstein reported for the �rst time an association be-tween pollen sensitisation and plant-derived food allergy [7]. In many subsequent clinicalstudies based on serology and skin tests, adverse reactions to food in patients with pollenallergy have been investigated [8]. The term `pollen-food allergy syndrome' describes al-lergy caused by IgE antibodies directed against cross-reacting allergens present in pollensand foods [9]. Several pollen-related food allergies have been described like the ragweedpollen-Cucurbitaceae vegetables-banana syndrome, the birch pollen-vegetable syndrome,the mugwort pollen-celery-spice syndrome, the grass pollen-vegetable syndrome and themould-latex syndrome (for a review see [8]). During the past few years, the primarystructure of an impressive number of di�erent allergens has been determined and thethree-dimensional structures of an increasing number of allergens have been determinedby X-ray crystallography or NMR spectroscopy. Although crystal structures have not yetcontributed to understand the reasons, why some proteins are allergenic and some oth-ers are not, they have substantially contributed to our understanding of cross-reactivitybetween homologous proteins, the major topic of this review.


allergen

B

Th2

APC

Th0

mast cells

allergic reaction

homologous

allergen

IgE

allergen

B

Th2

APC

Th0

mast cells

allergic reaction

homologous

allergen

IgE

Figure 5.1: Schematic diagram of the sensitisation mechanisms involved in cross-reactivitybetween homologous proteins. Molecular mimicry between conserved T- and B-cell epitopesis considered to be responsible for immunological cross-reactivity between structurally relatedproteins.

5.3 Clinical aspects of cross-reactivityThe phenomenon of allergologic cross-reactivity partly explains the polysensitisation fre-quently seen in clinical practice. To di�erentiate between parallel, independent sensitisa-tion and cross-reactivity due to shared structural features between homologous allergens,in vitro tests measuring IgE-inhibition can be applied [10]. Reactivity of a patient totwo or more extracts raises several questions regarding primary sensitisation and cross-reactivity. In the case of pollen-food allergy syndrome evidence suggests that a primarysensitisation to pollens is more likely or more common since allergy to fruits and vegeta-bles is much more frequent among patients with pollen allergy than those without [11,12].Further considerations regarding primary sensitisation may be based on the geographi-cal distribution of pollen allergens and associated food allergens [13]. Anaphylaxis hasbeen, for instance, observed in a patient with pre-existing pollenosis after his very �rstingestion of rambutan, a tropical fruit [14]. This shows that due to cross-reactivity be-tween allergens from di�erent sources it is possible that an individual experiences an

5.4. CROSS-REACTIVE STRUCTURES 123

allergic reaction at the �rst contact with a new source of exposure without being awareof a sensitisation. On the other hand, apples, carrots and other vegetables containinghomologues of pollen allergens are the earliest foreign proteins in infant nutrition. Sensi-tisation to these proteins in childhood is commonly believed to be due to the immaturityof the mucosal immune system and might account for the predisposition of children withfood hypersensitivity to develop pollen allergy later on [15]. Whether the observed cross-reactivity reactions are caused by molecular mimicry between conserved T- and B-cellepitopes present on the structurally related proteins or whether these proteins are aller-genic themselves remains to be elucidated. At least in some instances, however, thereis evidence for molecular mimicry. For example, it has been shown that IgE antibodiesof individuals sensitised to manganese-dependent superoxide dismutase (MnSOD) of As-pergillus fumigatus also recognise MnSOD of Drosophila melanogaster, however, exposureto D. melanogaster is unlikely to occur under normal circumstances [16,17].Uncovering of cross-reactive structures does not only help to understand polysensiti-sations observed in allergic individuals, but may also have clinical applications. Due toimmunological and structural similarities of allergens leading to cross-reactivity, the num-ber of epitopes needed for diagnosis and therapy of allergic diseases may be less diversethan initially expected [18]. If, for instance, the pollen-food allergy syndrome is due tocross-reacting allergens, desensitisation with pollen extracts could lead to a resolutionof both, allergic rhinitis and the associated food allergy. Thus, cross-reactive moleculesbear a high practical potential, because they might allow a reduction of the number ofproteins required for diagnostic and immunotherapeutic applications.

5.4 Cross-reactive structuresAs mentioned above, allergic individuals often react to various allergen sources. Thesymptoms can result either from polysensitisation or from cross-reactivity between homo-logous allergens from di�erent sources of exposure. All the cross-reactive proteins de-scribed so far have been found to re�ect common features on the level of both, primaryand tertiary structure. In contrast, not all proteins that share a similar fold are neces-sarily cross-reactive. Similar protein folds can be found with a sequence identity as littleas 25 % while cross-reactivity is rare below 50 % sequence identity [19]. Cross-reactiveproteins found in various organisms are also termed pan-allergens [16, 20].The �rst relevant cross-reactive allergens identi�ed were the major birch pollen allergensBet v 1 and Bet v 2, the latter belonging to the protein family of pro�lins. Allergenshomologous to Bet v 1 have been detected in other pollens [21,22] as well as in apple, pear,


celery, carrot, potato, hazel and cherry [23�30]. Pro�lins are highly conserved, ubiquitousproteins responsible for cross-reactivity between botanically unrelated species of fruits,vegetables and pollens [20, 23, 29, 31, 32]. Of course, cross-reactivity is not restricted toplant allergens. Cross-reactivity has, for example, also been demonstrated between themajor mite allergens Der p 1 and Der f 1, and Der p 2 and Der f 2, respectively [33�35],as well as between dog and cat allergen extract [36�38].

5.5 Cross-reactive fungal allergensCross-reactivity between di�erent fungal species is a well known phenomenon [39]. Wehave cloned and characterised a wide variety of fungal allergens, including enzymes,structural, intracellular and secreted proteins that allow to study cross-reactivity be-tween single components [40]. The allergens can be divided into species-speci�c andcross-reactive allergens. Interestingly, the major allergens of Aspergillus fumigatus, Al-ternaria alternata and Cladosporium herbarum, Asp f 1 [41, 42], Alt a 1 [43] and Clah 1 [44], respectively, represent species-speci�c proteins so far not found in any otherorganism. In contrast, cross-reactive allergens are conserved intracellular proteins thatshow signi�cant homology between phylogenetically distant species. Representatives ofthis group are enolases [45�47], peroxisomal proteins [48], manganese-dependent super-oxide dismutases (MnSOD) [16,17,49,50], acidic ribosomal P2 proteins [51,52], aldehydedehydrogenases [51], cyclophilins (CyP) [53�57], heat shock proteins (HSP) [58], nucleartransport factor 2 [59] and thioredoxins [60].A protein whose cross-reactivity between di�erent species has been extensively investi-gated is MnSOD. MnSOD of A. fumigatus was isolated as allergenic protein, formallytermed Asp f 6, from a cDNA library of the mould by phage display [61]. It was foundto represent a speci�c marker for allergic bronchopulmonary aspergillosis (ABPA), an in-tense in�ammatory response to Aspergillus in the lung [61,62]. In the following, MnSODfrom Drosophila melanogaster, Saccharomyces cerevisiae, Escherichia coli and humanMnSOD were cloned and demonstrated to cross-react with Asp f 6 on B- and T-celllevel [16, 17]. Furthermore, MnSOD from Hevea brasiliensis, Hev b 10, was isolated aslatex allergen that cross-reacts with Asp f 6 and human MnSOD [49]. Recently, alsoMalassezia sympodialis MnSOD, Mala s 10, was identi�ed as allergen of the opportunis-tic skin-colonising yeast [63]. The cloning and immunological characterisation of MnSODfrom di�erent sources demonstrated that MnSOD from phylogenetically distant speciesshare common IgE-binding epitopes.Another well-characterised pan-allergen family is constituted by cyclophilins. CyP was

5.6. AUTOREACTIVITY TO HUMAN HOMOLOGUES 125

isolated as IgE-binding protein from the basidiomycete Psilocybe cubensis [53]. In thefollowing, A. fumigatus CyP [54], Asp f 11, and M. sympodialis CyP [55], Mala s 6, wereisolated by phage display technology from the respective cDNA libraries and demon-strated to be cross-reactive allergens [56]. This led to the cloning of Candida albicansand S. cerevisiae CyP which were both found to cross-react with Asp f 11 [56]. Further-more cross-reactivity between human CyP A and B, respectively, with the fungal CyPscould be demonstrated [56]. Further allergens belonging to the CyP family were isolatedfrom birch (Bet v 7) [64] and carrot [65], however, no cross-reactivity between these twoallergens was observed [65]. By immunological screening of a cDNA library derived fromprotoscoleces of Echinococcus granulosus with IgE from patients with cystic echinococ-cosis and allergic manifestations, a protein identical to E. granulosus CyP was isolated(EA21). Despite the close homology, E. granulosus CyP did not cross-react with Malas 6 and human CyP in the performed assays [66].Thioredoxins constitute a further putative pan-allergen family that is currently underinvestigation. First, thioredoxin was identi�ed as allergenic protein of Coprinus coma-tus [67] and human thioredoxin was shown to induce type I skin reactions in individ-uals sensitised to C. comatus thioredoxin (Crameri and Helbling, unpublished results).A. fumigatus thioredoxin [60] and M. sympodialis thioredoxin (Flückiger et al., unpub-lished results) were isolated by phage display from the fungal cDNA libraries. Screeninga wheat germ and maize endosperm library with IgE from wheat and maize allergic in-dividuals, respectively, revealed that wheat and maize thioredoxins are also allergenicmolecules (Weichel et al., unpublished results). Furthermore, S. cerevisiae thioredoxinhas been cloned and produced (Glaser et al., unpublished results) and cross-reactivitystudies between thioredoxins of the organisms mentioned above are in process.

5.6 Autoreactivity to human homologues of fungal al-lergens

Serological investigations showed that some allergic patients also display IgE-mediatedautoreactivity to human homologues of environmental allergens. This was astonishing,because the human proteins would be expected to induce immune tolerance. In some situ-ations, however, tolerance to human proteins seems to be broken, resulting in autoreactiveIgE antibodies. Examples thereof are human MnSOD [17,61], human P2 protein [52] andhuman cyclophilin A and B [56], which were recognised by IgE antibodies of individualssensitised to the corresponding A. fumigatus allergens. Furthermore, human MnSOD


and human P2 protein are able to induce proliferation in peripheral blood mononuclearcells (PBMC) and to provoke strong type I skin reactions in patients su�ering from al-lergic bronchopulmonary aspergillosis or severe atopic dermatitis [17,52,61]. These dataprovide evidence for humoral and cell-mediated autoreactivity to human proteins in indi-viduals su�ering from long lasting chronic allergic disease, which might play a role in theexacerbation of the allergic reactions. The role of human MnSOD as autoallergen was alsoinvestigated in atopic dermatitis (AD). Speci�c IgE antibodies against human MnSODcould be detected in 36 % of the tested sera from patients with AD. These patientswere concomitantly sensitised to the skin colonising yeast M. sympodialis and stronglyreacted in skin tests to M. sympodialis extract, to recombinant human MnSOD and tostructurally related MnSODs. The human protein further induced proliferation of PBMCof MnSOD-sensitised AD patients and eczematous reactions in atopy patch test (APT).Immunohistochemical staining showed that human MnSOD, a stress-inducible enzyme,was upregulated in lesional and APT skin specimen compared to non-a�ected skin fromAD patients or healthy individuals. MnSOD from M. sympodialis is the likely sourceof primary sensitisation, which leads to the induction of cross-reactive antibodies ableto interact with the upregulated human MnSOD in in�amed skin. Thus, there is strongevidence for an important role of human MnSOD autoreactivity as a disease-exacerbatingfactor in a relevant subset of AD patients [68].It is expected that many other structures are involved in IgE-mediated autoreactivity.However, IgE-mediated reactions against self-antigens seem to be restricted to patientswith severe chronic allergic diseases and involve highly conserved intracellular proteins[17,52,61,69]. Obviously, cytoplasmic proteins are unlikely to be accessible for antibodiesunder normal circumstances. However, due to local in�ammatory processes in chronicdiseases such as ABPA or severe forms of AD, release of autoantigens could result as aconsequence of tissue damage. This might lead to exacerbation and/or perpetuation ofthe allergic disorder even in the absence of exogenous allergen exposure [16, 17, 52, 70].These mechanisms might explain some forms of severe atopic diseases.

5.7 Three-dimensional structures of allergensAlthough the primary structure of a protein contains all the clues for its tertiary struc-ture, prediction of the three-dimensional structure of a protein is not (yet) possible. Toelucidate the three-dimensional structures of proteins X-ray crystallography [71], nuclearmagnetic resonance (NMR) spectroscopy [72] and homology modelling [73] have beenapplied. Homology modelling allows to predict backbone folds of well aligned sequences

5.7. THREE-DIMENSIONAL STRUCTURES OF ALLERGENS 127

with high accuracy based on the coordinates of a solved homologous structure, however,con�gurations of solvent exposed side-chains are di�cult to predict [19]. With currenttechnologies, NMR analysis is limited to molecules with a maximal molecular weight ofaround 40 kDa [72], but can be applied for proteins that are not readily crystallisable.NMR structures are not as detailed and accurate as crystal structures, and the resultis an ensemble of alternative models, in contrast to the unique model obtained by crys-tallography. Thus, X-ray crystallography is the method of choice to exactly characterisesurface structures of allergens, which are relevant for antibody binding. In Table 5.1, theknown allergen crystal structures with coordinates deposited in the Protein Data Bank(PDB, http://www.rcsb.org/pdb/) [74] are listed. There are only a few additional aller-gen structures listed in the PDB that were determined by NMR spectroscopy, includingthe giant ragweed pollen allergen Amb t 5 [75], the birch pollen allergen Bet v 4 [76], Derf 2, the major mite allergen from Dermatophagoides farinae [77], a lipid transfer proteinfrom maize seedlings (Zea m 14) [78] and the cherry allergen Pru av 1 [30].Although the number of solved allergen structures is rapidly increasing, no characteristicstructural or functional features of allergens have been detected that would allow theprediction of the allergenicity of a protein. The allergens known belong to a wide varietyof di�erent families including intracellular and secreted enzymes, structural proteins andproteins without known biochemical function. It seems that in principle, any protein hasthe potential to become allergenic and therefore prediction of the allergenic potential ofa protein is a major challenge of molecular allergology. Prediction of allergenicity is ofspecial interest if a new protein is introduced in the environment, particularly in alteredfoods. Knowledge on the structure/allergenicity relationship could allow modi�cationor substitution of proteins by less allergenic proteins, resulting in a reduced food chainrisk for allergic patients [19]. Interestingly, the solved allergen structures cover a widevariety of di�erent folds and combinations thereof as schematically shown in Figure 5.2.Aalberse [19] describes the structure analysis of about 40 allergenic proteins or partsof these proteins, of which the folds are either known or can be predicted on the basisof homology. According to this study, most allergens can be classi�ed into four majorstructural families that potentially confer allergenicity: A) antiparallel β-strands: theimmunoglobulin-fold family (grass group 2, mite group 2), serine proteases (mite group3, 6 and 9) and soybean-type trypsin inhibitor (Ole e 1, grass group 11); B) antiparallelβ-sheets intimately associated with one or more α-helices: tree group 1, lipocalin, pro�linand aspartate protease (cockroach group 2); C) (α+β) structures, in which the α- andβ-structural elements are not intimately associated: mite group 1, lysozyme/lactalbuminand vespid group 5; D) α-helical: non-speci�c lipid transfer protein, seed S2 protein,


Source Allergen Function PDB ID Ref.Lamb's-quarters Che a 3 Polcalcin 1PMZ [79]Timothy grass pollen Phl p 1 1N10 [80]Timothy grass pollen Phl p 2 1WHO, 1WHP [81]Timothy grass pollen Phl p 5b Ag25 1L3P [82]Timothy grass pollen Phl p 6 1NLX [83]Timothy grass pollen Phl p 7 Calcium-bind. protein 1K9U [84]Thale cress Pro�lin I 1A0K, 3NUL [85]Birch pollen Bet v 1 1BV1, 1FM4 (Bet v 1l) [86,87]Birch pollen Bet v 2 Pro�lin 1CQA [88]Cedar pollen Jun a 1 1PXZ [89]House dust mite Der p 2 1KTJ [90]Domestic cattle (dander) Bos d 2 Lipocalin 1BJ7 [91]Domestic cattle (milk) Bos d 4 α-Lactalbumin 1HFZ, 1F6R, 1F6S [92,93]Domestic cattle (milk) Bos d 5 β-Lactoglobulin 1MFH [94]Domestic horse (dander) Equ c 1 Lipocalin 1EW3 [95]Cat (saliva) Fel d 1 1PUO [96]Mouse Mus m 1 α-2U-globulin 1MUP, 1I04 [95,97]Rat Rat n 1 α-2U-globulin 2A2G, 2A2U [98]Aspergillus restrictus Asp f 1 Ribotoxin 1AQZ [99]Aspergillus fumigatus Asp f 6 MnSOD 1KKC [50]Paecilomyces varioti Xylanase 1PVX [100]Acanthamoeba castellanii Pro�lin Ia 1PRQ [101]Acanthamoeba castellanii Pro�lin II 1F2K [102]Bee venom Api m 1 Phospholipase A2 1POC [103]Bee venom Api m 2 Hyaluronidase 1FCQ, 1FCU [104]Bee venom Api m 4 Melittin 2MLT [105]Wasp venom Ves v 5 1QNX [106]Midge Chi t 1-9 Haemoglobin 1ECA, 1ECD, 1ECN, 1ECO [107]Hen egg white Gal d 2 Ovalbumin 1OVA, 1UHG [108,109]Hen egg white Gal d 3 Conalbumin 1OVT, 1IEJ, 1IQ7, 1RYX [110�113]Hen egg white Gal d 4 Lysozyme 1HEL, 1BWH, 1BWI, 1BWJ [114,115]Carp Parvalbumin B 5CPV [116]Latex Hev b 6.02 Hevein 1Q9B, 1WKX [117,118]Latex Hev b 8 Pro�lin 1G5U [119]

Table 5.1: Allergen crystal structures deposited in the Protein Data Bank (PDB). Mutatedallergens and allergen complexes are unaccounted for.

insect haemoglobin, �sh parvalbumin, pollen calmodulin, mellitin from bee venom, Feld 1 chain 1 and serum albumin [19]. In conclusion, allergens have no characteristicstructural features other than that they need to display at least two B-cell epitopeson their surface required to crosslink receptor-bound IgE on e�ector cells leading todegranulation, mediator release and as a consequence thereof to allergic reactions.B-cell epitopes are generally of the discontinuous type and contain 15-22 amino acid

5.7. THREE-DIMENSIONAL STRUCTURES OF ALLERGENS 129

A B

C D

Figure 5.2: The major structural families of allergens as classi�ed in [19]. A) antiparallelβ-strands: timothy grass pollen allergen Phl p 2 (PDB ID 1WHP). B) antiparallel β-sheetsintimately associated with one or more α-helices: birch pollen allergen Bet v 1 (PDB ID 1BV1).C) (α+β) structures, in which the α- and β-structural elements are not intimately associated:hen egg white lysozyme Gal d 4 (PDB ID 1HEL). D) α-helices: timothy grass pollen allergenPhl p 7 (PDB ID 1K9U).

residues involving di�erent surface loops [120]. Energetic calculations suggest that asmaller subset of 5-6 of these residues may contribute most of the binding energy, withthe surrounding residues allowing structural complementarity [121]. The buried surfaceoccupied by B cell epitopes, corresponding to the surface area that becomes inaccessiblefor water molecules as a consequence of antibody binding, is in the range of 540 - 890 Å2

[122]. There is evidence that most, if not all, of the protein surface is covered by possibleIgE-binding epitopes [123], however, in spatial terms only a few IgE antibodies can beaccommodated at the same time [19]. Because the allergenicity of IgE-binding epitopes is


in most cases absolutely dependent upon conformation of the native protein [120], correctprotein folding is essential for B-cell epitope mapping. Proteins with biological activity,e.g. enzymatic activity, are only functional if they are folded correctly, therefore native-like conformation of these proteins can be corroborated by measuring their activity.

5.8 Identi�cation of cross-reactive epitopesA simple approach to identify cross-reactive B-cell epitopes is to determine shared featuresof cross-reactive allergens on the level of primary and tertiary structure. Combination ofdata on conserved amino acids in homologous sequences with data on the solvent exposureof these residues in the three-dimensional structures, and thus accessibility for antibodybinding, allows to identify amino acids that are potentially involved in IgE-mediatedcross-reactivity [86, 91]. The actual contribution of a residue to the binding of IgE canthereafter be investigated by site-directed mutagenesis and testing of the IgE-bindingcapacity of the mutants [95,124].We followed this strategy to elucidate, whether the mechanisms of the above men-tioned IgE-mediated cross-reactivity between A. fumigatus MnSOD (Asp f 6) and hu-man MnSOD is based on molecular mimicry between conserved B-cell epitopes of the twomolecules. The crystal structure of Asp f 6 was solved at 2 Å resolution [50] and comparedwith the structure of the human MnSOD, which had been determined at 2.2 Å resolutionand deposited in the PDB with the ID 1ABM [125]. On sequence level, A. fumigatusand human MnSOD share 45 % identity and 65 % similarity. The two enzymes werealso found to have a high similarity in the overall structure (rms deviation 1.48 Å). Therelative solvent-accessible area of the individual residues in the two crystal structures wascalculated. Only those residues that are at least partly exposed to solvent can contributeto IgE binding. Therefore, the amino acids that are identical or similar in the fungaland human sequence and solvent exposed in both structures are potentially involved inthe IgE-mediated cross-reactivity between the two proteins. Although the fungal and thehuman sequence share 101 identical amino acids, only 17 thereof are at least 30 % andonly 10 at least 50 % solvent exposed in both structures and therefore likely to be acces-sible for antigen-antibody interactions [50]. Thus, a large portion of the conserved aminoacid residues is located in the core of the protein and inaccessible for antibody binding,including the highly conserved residues de�ning the active center of the enzyme. Theputative IgE-binding residues that could be involved in several B-cell epitopes elicitedby a polyclonal immune response are scattered over the whole sequence, in agreementwith the �ndings that B-cell epitopes elicited by natural exposure are discontinuous.

5.8. IDENTIFICATION OF CROSS-REACTIVE EPITOPES 131

They are also clustered over the whole surface, indicating that the entire surface of theMnSOD is potentially antigenic, as postulated by Laver et al. [120]. Furthermore, ho-mology models of the three-dimensional structures of A. fumigatus, D. melanogaster andS. cerevisiae MnSOD were constructed using the crystal structure of the human MnSODas template and conserved amino acid residues exposed to the solvent were identi�edin these models [17]. D. melanogaster and S. cerevisiae MnSOD had previously beenshown to cross-react with Asp f 6 [16, 17]. The sequences of the four MnSODs sharedonly �ve identical amino acid residues that were more than 50 % solvent exposed andthus accessible for antibody binding. Altogether, homology modelling and X-ray crys-tallography yielded comparable results. Superimposition of the backbone atoms of thecrystal structure and the modelled structure of Asp f 6 resulted in an rms deviation of1.47 Å. Based on these results, the structures of other pan-allergens may be approxi-matively determined by homology modelling, if the structure of a related protein hasbeen solved already, allowing a fast identi�cation of putative cross-reactive epitopes. Bycomparing the three-dimensional structures of cross-reactive allergens, we intend to iden-tify further B-cell epitopes within other pan-allergen families, including cyclophilins andthioredoxins. M. sympodialis CyP, Mala s 6, and M. sympodialis thioredoxin have beencrystallised and their structures are currently being solved (Limacher et al., unpublishedresults).However, the only method allowing a complete de�nition of a B cell epitope remains co-crystallisation of the allergen with a monoclonal antibody Fab fragment and solving thestructure of the complex by X-ray crystallography [120]. For crystallisation and X-raycrystallography homogenous reagents are required, therefore polyclonal human serumIgE can not be used. Because monoclonal human IgE is di�cult to obtain, monoclonalantibodies from other sources are currently used. The atomic details of interaction areknown for more than 30 antigen-antibody combinations [19]. The �rst structure of anallergen-Fab complex that has been solved was the structure of a complex between themajor birch pollen allergen Bet v 1 and the Fab fragment of a murine monoclonal IgG1antibody, BV16 [126]. This structure may serve as a model for allergen-antibody interac-tions relevant in allergy. The epitope de�ned by the BV16 Fab overlaps, at least partly,with the Bet v 1 epitope recognised by human serum IgE because BV16 was shown topartially inhibit the binding of human IgE to Bet v 1. The crystal structure of theBet v 1-BV16 Fab complex revealed that the dominant epitope, covering about 10 % ofthe Bet v 1 surface, lies in the patch of conserved surface residues among di�erent aller-gens from Fagales, explaining the binding of BV16 to allergens from pollen of other treesand the cross-reactivity among them. Furthermore, the crystal structure of the complex


showed that Bet v 1 does not undergo any major conformational changes upon bindingof BV16 Fab.

5.9 ConclusionsThe only known common feature of allergens, which span a wide variety of structurallyand functionally heterogeneous molecules [19], is their ability to elicit IgE-mediated im-mune responses to exposure and, as a consequence thereof, hypersensitivity reactions bycross-linking IgE antibodies bound to e�ector cells [127]. The application of molecu-lar biology methods to the �eld of allergology during the last decade allowed to clone,produce and characterise many recombinant allergens and their number is rapidly in-creasing [60, 128�130]. The cloned allergens can be divided into conserved intracellularproteins that share high sequence homology with proteins from phylogenetically distantspecies, species-speci�c proteins and such without sequence homology to known proteins.Examples of cloned allergens that belong to the group of conserved proteins are pro�l-ins [20, 23, 29, 31, 32], Bet v 1 and its homologues [21�29], enolases [45�47], peroxisomalproteins [48], manganese-dependent superoxide dismutases [16, 17, 49, 50, 61], acidic ri-bosomal P2 proteins [51, 52], aldehyde dehydrogenases [51], cyclophilins [53�57, 64�66],heat shock proteins [58], nuclear transport factors 2 [59] and thioredoxins [60]. Many ofthese allergens were demonstrated to exhibit IgE-mediated immunological cross-reactivitywith homologous proteins from di�erent species. These proteins, termed pan-allergens,are considered to be responsible for complex allergic syndromes observed in clinical prac-tice such as pollen-related food allergy [9] or latex-mould syndrome [49]. Knowledge ofcross-reactive structures will facilitate the diagnosis of allergic diseases by reducing thenumber of molecular structures needed for a speci�c and sensitive identi�cation of allergicpatients. Moreover, immunotherapy with a cross-reactive allergen of one organism mayalso a�ect the immunologic reactivity of an allergic individual against structurally homo-logous proteins present in other allergenic sources. However, due to the limited structuralknowledge about B-cell epitopes, it is not possible at present time to suggest any struc-tural motif or sequence pattern common to all allergenic proteins, if existent at all, al-though the solved structures have substantially contributed to explain cross-reactivity atmolecular level. To elucidate, whether the mechanisms of IgE-mediated cross-reactivityare based on molecular mimicry between conserved B-cell epitopes of homologous pro-teins, exact comparisons of their three-dimensional structures are needed. Methods todetermine the three-dimensional structure of a protein are X-ray crystallography [71],nuclear magnetic resonance spectroscopy [72] and homology modelling [73]. X-ray crys-

BIBLIOGRAPHY 133

tallography results in the most detailed and accurate information about surface structuresof allergens, which are relevant for antibody binding. In contrast, homology modelling isa fast approach for an approximate comparison of 3D structures and it is not dependenton successful protein crystallisation. However, it can only be applied, if the 3D structureof a related protein is already solved. Once putative cross-reactive epitopes are identi�ed,their actual contribution to the total IgE binding of the allergen can be investigated bysite-directed mutagenesis, followed by analysis of the IgE-binding capacity of the mutants.If the mutations do not signi�cantly reduce the IgE-a�nity of the molecule, it may beconcluded that the identi�ed cross-reactive epitopes are not immunodominant. Know-ledge of the three-dimensional structure of an allergen and identi�cation of epitopes canexplain IgE-mediated cross-reactivity at molecular level and might help to understandthe molecular basis of allergy. Furthermore, there are therapeutic implications for theidenti�cation of IgE-binding epitopes as the information could be used for the design ofnew and safer vaccines for immunotherapy. This can be achieved by mutational changesof allergens, e.g. by site-directed mutagenesis, in order to reduce their interactions withIgE and thus reducing anaphylactic side e�ects without changing the epitopes responsiblefor the development of T-cell tolerance [131]. Another therapeutic approach is based onthe saturation of allergic e�ector cells with allergen derivatives unable to induce cross-linking of e�ector cell-bound IgE, thus preventing mediator release by subsequent contactwith the intact allergen [129].

AcknowledgementsThis work was supported by the Swiss National Science Foundation grants No. 3100-063381 and 3100-063381/2, by the OPO Foundation, Zürich and by the EMDO Founda-tion, Zürich.

Bibliography[1] V. Casolaro, S. N. Georas, Z. Song, and S. J. Ono. Biology and genetics of atopic disease. Curr.

Opin. Immunol., 8(6):796�803, 1996.

[2] R. Valenta, J. Lidholm, V. Niederberger, B. Hayek, D. Kraft, and H. Gronlund. The recombinantallergen-based concept of component-resolved diagnostics and immunotherapy (CRD and CRIT).Clin. Exp. Allergy, 29(7):896�904, 1999.

[3] R. Kodzius, C. Rhyner, Z. Konthur, D. Buczek, H. Lehrach, G. Walter, and R. Crameri. Rapididenti�cation of allergen-encoding cDNA clones by phage display and high-density arrays. Comb.Chem. High Throughput Screen., 6(2):147�54, 2003.


[4] I. Annila. Bee venom allergy. Clin. Exp. Allergy, 30(12):1682�7, 2000.

[5] T. P. King and M. D. Spangfort. Structure and biology of stinging insect venom allergens. Int.Arch. Allergy Immunol., 123(2):99�106, 2000.


[7] L. Tuft and G. I. Blumenstein. Studies in food allergy. J. Allergy, 13:574�81, 1942.

[8] T. Caballero and M. Martin-Esteban. Association between pollen hypersensitivity and ediblevegetable allergy: a review. J. Investig. Allergol. Clin. Immunol., 8(1):6�16, 1998.

[9] M. H. Lessof. Pollen-food allergy syndrome. J. Allergy Clin. Immunol., 98(1):239�40, 1996.

[10] G. Pauli. Evolution in the understanding of cross-reactivities of respiratory allergens: the role ofrecombinant allergens. Int. Arch. Allergy Immunol., 123(3):183�95, 2000.

[11] M. Hannuksela and A. Lahti. Immediate reactions to fruits and vegetables. Contact Dermatitis,3(2):79�84, 1977.

[12] N. E. Eriksson, H. Formgren, and E. Svenonius. Food hypersensitivity in patients with pollenallergy. Allergy, 37(6):437�43, 1982.

[13] K. Ho�mann-Sommergruber, P. Demoly, R. Crameri, H. Breiteneder, C. Ebner, M. Laimer DaCamara Machado, K. Blaser, C. Ismail, O. Scheiner, J. Bousquet, and G. Menz. IgE reactivity toApi g 1, a major celery allergen, in a central european population is based on primary sensitizationby Bet v 1. J. Allergy Clin. Immunol., 104(2 Pt 1):478�84, 1999.

[14] J. M. Kelso, R. T. Jones, and J. W. Yunginger. Anaphylaxis after initial ingestion of rambutan,a tropical fruit. J. Allergy Clin. Immunol., 102(1):145�6, 1998.

[15] R. Fritsch, B. Bohle, U. Vollmann, U. Wiedermann, B. Jahn-Schmid, M. Krebitz, H. Breiteneder,D. Kraft, and C. Ebner. Bet v 1, the major birch pollen allergen, and Mal d 1, the major appleallergen, cross-react at the level of allergen-speci�c T helper cells. J. Allergy Clin. Immunol.,102(4 Pt 1):679�86, 1998.


[17] S. Flückiger, L. Scapozza, C. Mayer, K. Blaser, G. Folkers, and R. Crameri. Immunological andstructural analysis of IgE-mediated cross-reactivity between manganese superoxide dismutases.Int. Arch. Allergy Immunol., 128(4):292�303, 2002.

[18] R. Valenta, P. Steinberger, M. Duchene, and D. Kraft. Immunological and structural similaritiesamong allergens: prerequisite for a speci�c and component-based therapy of allergy. Immunol.Cell Biol., 74(2):187�94, 1996.


[20] R. Valenta, M. Duchene, C. Ebner, P. Valent, C. Sillaber, P. Deviller, F. Ferreira, M. Tejkl,H. Edelmann, D. Kraft, et al. Pro�lins constitute a novel family of functional plant pan-allergens.J. Exp. Med., 175(2):377�85, 1992.

BIBLIOGRAPHY 135

[21] E. Florvaag, E. Holen, H. Vik, and S. Elsayed. Comparative studies on tree pollen allergens. XIV.Characterization of the birch (Betula verrucosa) and hazel (Corylus avellana) pollen extracts byhorizontal 2-D SDS-PAGE combined with electrophoretic transfer and IgE immunoautoradiogra-phy. Ann. Allergy, 61(5):392�400, 1988.

[22] H. Ipsen and O. C. Hansen. The NH2-terminal amino acid sequence of the immunochemicallypartial identical major allergens of Alder (Alnus glutinosa) Aln g I, birch (Betula verrucosa) Betv I, hornbeam (Carpinus betulus) Car b I and oak (Quercus alba) Que a I pollens. Mol. Immunol.,28(11):1279�88, 1991.

[23] C. Ebner, R. Hirschwehr, L. Bauer, H. Breiteneder, R. Valenta, H. Ebner, D. Kraft, andO. Scheiner. Identi�cation of allergens in fruits and vegetables: IgE cross-reactivities with theimportant birch pollen allergens Bet v 1 and Bet v 2 (birch pro�lin). J. Allergy Clin. Immunol.,95(5 Pt 1):962�9, 1995.

[24] P. G. Calkhoven, M. Aalbers, V. L. Koshte, O. Pos, H. D. Oei, and R. C. Aalberse. Cross-reactivityamong birch pollen, vegetables and fruits as detected by IgE antibodies is due to at least threedistinct cross-reactive structures. Allergy, 42(5):382�90, 1987.

[25] C. Ebner, T. Birkner, R. Valenta, H. Rumpold, M. Breitenbach, O. Scheiner, and D. Kraft.Common epitopes of birch pollen and apples - studies by western and northern blot. J. AllergyClin. Immunol., 88(4):588�94, 1991.

[26] B. Fahlbusch, O. Rudeschko, W. D. Muller, G. Schlenvoigt, S. Vettermann, and L. Jager. Puri�ca-tion and characterization of the major allergen from apple and its allergenic cross-reactivity withBet v 1. Int. Arch. Allergy Immunol., 108(2):119�26, 1995.

[27] S. Vieths, A. Jankiewicz, B. Schoning, and H. Aulepp. Apple allergy: the IgE-binding potency ofapple strains is related to the occurrence of the 18-kDa allergen. Allergy, 49(4):262�71, 1994.

[28] S. Vieths, A. Jankiewicz, B. Wuthrich, and W. Baltes. Immunoblot study of IgE binding allergensin celery roots. Ann. Allergy Asthma Immunol., 75(1):48�55, 1995.

[29] R. Hirschwehr, R. Valenta, C. Ebner, F. Ferreira, W. R. Sperr, P. Valent, M. Rohac, H. Rumpold,O. Scheiner, and D. Kraft. Identi�cation of common allergenic structures in hazel pollen andhazelnuts: a possible explanation for sensitivity to hazelnuts in patients allergic to tree pollen.J. Allergy Clin. Immunol., 90(6 Pt 1):927�36, 1992.

[30] P. Neudecker, K. Schweimer, J. Nerkamp, S. Scheurer, S. Vieths, H. Sticht, and P. Rosch. Allergiccross-reactivity made visible: solution structure of the major cherry allergen Pru av 1. J. Biol.Chem., 276(25):22756�63, 2001.

[31] R. van Ree, M. Fernandez-Rivas, M. Cuevas, M. van Wijngaarden, and R. C. Aalberse. Pollen-related allergy to peach and apple: an important role for pro�lin. J. Allergy Clin. Immunol.,95(3):726�34, 1995.

[32] R. van Ree, V. Voitenko, W. A. van Leeuwen, and R. C. Aalberse. Pro�lin is a cross-reactiveallergen in pollen and vegetable foods. Int. Arch. Allergy Immunol., 98(2):97�104, 1992.

[33] P. W. Heymann, M. D. Chapman, and T. A. Platts-Mills. Antigen Der f I from the dust mite Der-matophagoides farinae: structural comparison with Der p I from Dermatophagoides pteronyssinusand epitope speci�city of murine IgG and human IgE antibodies. J. Immunol., 137(9):2841�7,1986.


[34] P. Lind, O. C. Hansen, and N. Horn. The binding of mouse hybridoma and human IgE antibodiesto the major fecal allergen, Der p I, of Dermatophagoides pteronyssinus. Relative binding sitelocation and species speci�city studied by solid-phase inhibition assays with radiolabeled antigen.J. Immunol., 140(12):4256�62, 1988.

[35] H. Yasueda, H. Mita, Y. Yui, and T. Shida. Comparative analysis of physicochemical and im-munochemical properties of the two major allergens from Dermatophagoides pteronyssinus andthe corresponding allergens from Dermatophagoides farinae. Int. Arch. Allergy Appl. Immunol.,88(4):402�7, 1989.

[36] J. L. Ohman, Jr., K. J. Bloch, S. Kendall, and F. C. Lowell. Allergens of mammalian origin. IV.Evidence for common allergens in cat and dog serum. J. Allergy Clin. Immunol., 57(6):560�8,1976.

[37] M. Viander, E. Valovirta, T. Vanto, and A. Koivikko. Cross-reactivity of cat and dog allergenextracts. RAST inhibition studies with special reference to the allergenic activity in saliva andurine. Int. Arch. Allergy Appl. Immunol., 71(3):252�60, 1983.

[38] B. Wuthrich, B. Guerin, and B. E. Hewitt. Cross-allergenicity between extracts of hair fromdi�erent dog breeds and cat fur. Clin. Allergy, 15(2):87�93, 1985.

[39] W. E. Horner, A. Helbling, J. E. Salvaggio, and S. B. Lehrer. Fungal allergens. Clin. Microbiol.Rev., 8(2):161�79, 1995.

[40] M. Weichel, S. Flückiger, and R. Crameri. Molecular characterisation of mould allergens involvedin respiratory complications. In S. G. Pandalai, editor, Recent research developments in respiratoryand critical care medicine, volume 2, pages 29�45. Research Signpost, Trivandrum, 2002.

[41] M. Moser, R. Crameri, G. Menz, T. Schneider, T. Dudler, C. Virchow, M. Gmachl, K. Blaser, andM. Suter. Cloning and expression of recombinant Aspergillus fumigatus allergen I/a (rAsp f I/a)with IgE binding and type I skin test activity. J. Immunol., 149(2):454�60, 1992.

[42] L. K. Arruda, T. A. Platts-Mills, J. W. Fox, and M. D. Chapman. Aspergillus fumigatus allergenI, a major IgE-binding protein, is a member of the mitogillin family of cytotoxins. J. Exp. Med.,172(5):1529�32, 1990.

[43] M. W. De Vouge, A. J. Thaker, I. H. Curran, L. Zhang, G. Muradia, H. Rode, and H. M. Vijay.Isolation and expression of a cDNA clone encoding an Alternaria alternata Alt a 1 subunit. Int.Arch. Allergy Immunol., 111(4):385�95, 1996.

[44] L. Aukrust and S. M. Borch. Partial puri�cation and characterization of two Cladosporiumherbarum allergens. Int. Arch. Allergy Appl. Immunol., 60(1):68�79, 1979.

[45] M. Breitenbach, B. Simon, G. Probst, H. Oberko�er, F. Ferreira, P. Briza, G. Achatz, A. Unger,C. Ebner, D. Kraft, and R. Hirschwehr. Enolases are highly conserved fungal allergens. Int. Arch.Allergy Immunol., 113(1-3):114�7, 1997.

[46] B. A. Baldo and R. S. Baker. Inhalant allergies to fungi: reactions to bakers' yeast (Saccharomycescerevisiae) and identi�cation of bakers' yeast enolase as an important allergen. Int. Arch. AllergyAppl. Immunol., 86(2):201�8, 1988.

[47] K. Ito, A. Ishiguro, T. Kanbe, K. Tanaka, and S. Torii. Detection of IgE antibody against Candidaalbicans enolase and its crossreactivity to Saccharomyces cerevisiae enolase. Clin. Exp. Allergy,25(6):522�8, 1995.

BIBLIOGRAPHY 137

[48] S. Hemmann, K. Blaser, and R. Crameri. Allergens of Aspergillus fumigatus and Candida boidiniishare IgE-binding epitopes. Am. J. Respir. Crit. Care Med., 156(6):1956�62, 1997.

[49] S. Wagner, S. Sowka, C. Mayer, R. Crameri, M. Focke, V. P. Kurup, O. Scheiner, and H. Breite-neder. Identi�cation of a Hevea brasiliensis latex manganese superoxide dismutase (Hev b 10) asa cross-reactive allergen. Int. Arch. Allergy Immunol., 125(2):120�7, 2001.


[51] G. Achatz, H. Oberko�er, E. Lechenauer, B. Simon, A. Unger, D. Kandler, C. Ebner, H. Prillinger,D. Kraft, and M. Breitenbach. Molecular cloning of major and minor allergens of Alternariaalternata and Cladosporium herbarum. Mol. Immunol., 32(3):213�27, 1995.

[52] C. Mayer, U. Appenzeller, H. Seelbach, G. Achatz, H. Oberko�er, M. Breitenbach, K. Blaser,and R. Crameri. Humoral and cell-mediated autoimmune reactions to human acidic ribosomal P2protein in individuals sensitized to Aspergillus fumigatus P2 protein. J. Exp. Med., 189(9):1507�12,1999.

[53] W. E. Horner, G. Reese, and S. B. Lehrer. Identi�cation of the allergen Psi c 2 from the basid-iomycete Psilocybe cubensis as a fungal cyclophilin. Int. Arch. Allergy Immunol., 107(1-3):298�300,1995.

[54] R. Crameri. Epidemiology and molecular basis of the involvement of Aspergillus fumigatus inallergic diseases. Contrib. Microbiol., 2:44�56, 1999.



[57] D. Roy, D. Ghosh, and S. Gupta-Bhattacharya. Homology modeling of allergenic cyclophilins: IgE-binding site and structural basis of cross-reactivity. Biochem. Biophys. Res. Commun., 307(2):422�9, 2003.

[58] A. Kumar, L. V. Reddy, A. Sochanik, and V. P. Kurup. Isolation and characterization of arecombinant heat shock protein of Aspergillus fumigatus. J. Allergy Clin. Immunol., 91(5):1024�30, 1993.

[59] M. Weichel, P. Schmid-Grendelmeier, S. Flückiger, M. Breitenbach, K. Blaser, and R. Crameri.Nuclear transport factor 2 represents a novel cross-reactive fungal allergen. Allergy, 58(3):198�206,2003.



[62] R. Crameri, S. Hemmann, C. Ismail, G. Menz, and K. Blaser. Disease-speci�c recombinant al-lergens for the diagnosis of allergic bronchopulmonary aspergillosis. Int. Immunol., 10(8):1211�6,1998.


[63] A. Andersson, O. Rasool, M. Schmidt, R. Kodzius, S. Flückiger, A. Zargari, R. Crameri, andA. Scheynius. Cloning, expression and characterization of two new IgE-binding proteins fromthe yeast Malassezia sympodialis with sequence similarities to heat shock proteins and manganesesuperoxide dismutase. Eur. J. Biochem., 271(10):1885�94, 2004.

[64] P. Cadot, J. F. Diaz, P. Proost, J. Van Damme, Y. Engelborghs, E. A. Stevens, and J. L. Ceup-pens. Puri�cation and characterization of an 18-kd allergen of birch (Betula verrucosa) pollen:identi�cation as a cyclophilin. J. Allergy Clin. Immunol., 105(2 Pt 1):286�91, 2000.

[65] C. Fujita, T. Moriyama, and T. Ogawa. Identi�cation of cyclophilin as an IgE-binding proteinfrom carrots. Int. Arch. Allergy Immunol., 125(1):44�50, 2001.

[66] E. Ortona, S. Vaccari, P. Margutti, F. Delunardo, R. Rigano, E. Profumo, B. Buttari, O. Ra-sool, A. Teggi, and A. Siracusano. Immunological characterization of Echinococcus granulosuscyclophilin, an allergen reactive with IgE and IgG4 from patients with cystic echinococcosis. Clin.Exp. Immunol., 128(1):124�30, 2002.

[67] A. Helbling, K. A. Brander, W. E. Horner, and S. B. Lehrer. Allergy to basidiomycetes. Chem.Immunol., 81:28�47, 2002.

[68] P. Schmid-Grendelmeier, S. Flückiger, R. Disch, A. Trautmann, T. Kundig, R. Dummer,B. Wuthrich, K. Blaser, and R. Crameri. Stress mediated up-regulation of human manganesesuperoxide dismutase elicits eczematous reactions in atopic dermatitis. Submitted.


[70] R. Valenta, S. Natter, S. Seiberler, M. Roschanak, N. Mothes, V. Mahler, and P. Eibensteiner.Autoallergy: a pathogenetic factor in atopic dermatitis? Curr. Probl. Dermatol., 28:45�50, 1999.

[71] M. S. Smyth and J. H. Martin. X-ray crystallography. Mol. Pathol., 53(1):8�14, 2000.

[72] A. E. Ferentz and G. Wagner. NMR spectroscopy: a multifaceted approach to macromolecularstructure. Q. Rev. Biophys., 33(1):29�65, 2000.

[73] R. Sanchez and A. Sali. Advances in comparative protein-structure modelling. Curr. Opin. Struct.Biol., 7(2):206�14, 1997.

[74] H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H. Weissig, I. N. Shindyalov, andP. E. Bourne. The Protein Data Bank. Nucleic Acids Res., 28(1):235�42, 2000.

[75] G. L. Warren, C. J. Tuner, G. A. Petsko, and A. T. Brunger. A highly precise solution 1H NMRstructure of ragweed allergen Amb t 5. http://www.rcsb.org/pdb, 1998.

[76] P. Neudecker, J. Nerkamp, A. Eisenmann, A. Nourse, T. Lauber, K. Schweimer, K. Lehmann,S. Schwarzinger, F. Ferreira, and P. Rosch. Solution structure, dynamics, and hydrodynamics ofthe calcium-bound cross-reactive birch pollen allergen Bet v 4 reveal a canonical monomeric twoEF-hand assembly with a regulatory function. J. Mol. Biol., 336(5):1141�57, 2004.

[77] S. Ichikawa, H. Hatanaka, T. Yuuki, N. Iwamoto, S. Kojima, C. Nishiyama, K. Ogura, Y. Okumura,and F. Inagaki. Solution structure of Der f 2, the major mite allergen for atopic diseases. J. Biol.Chem., 273(1):356�60, 1998.

BIBLIOGRAPHY 139

[78] J. Gomar, M. C. Petit, P. Sodano, D. Sy, D. Marion, J. C. Kader, F. Vovelle, and M. Ptak.Solution structure and lipid binding of a nonspeci�c lipid transfer protein extracted from maizeseeds. Protein Sci., 5(4):565�77, 1996.

[79] P. Verdino, R. Barderas, K. Westritschnig, R. Valenta, R. Rodriguez, and W. Keller. Structure ofthe calcium-binding pollen allergen Che a 3. http://www.rcsb.org/pdb, 2003.

[80] A. A. Fedorov, T. Ball, B. Leistler, R. Valenta, and S. C. Almo. X-ray crystal structure of Phlp 1, a major timothy grass pollen allergen. http://www.rcsb.org/pdb, 2002.

[81] A. A. Fedorov, E. V. Fedorov, C. Dolecek, M. Susani, R. Valenta, and S. C. Almo. Crystal structureof Phl p 2, a major timothy grass (Phleum pratense) pollen allergen. http://www.rcsb.org/pdb/,1997.

[82] K. Rajashankar, A. Bufe, W. Weber, S. Eschenburg, B. Lindner, and C. Betzel. Structure of thefunctional domain of the major grass-pollen allergen Phl p 5b. Acta Cryst., D58(Pt 7):1175�81,2002.

[83] A. A. Fedorov, T. Ball, E. V. Fedorov, S. Vrtala, R. Valenta, and S. C. Almo. Crystal structure ofPhl p 6, a major timothy grass pollen allergen co-crystallized with zinc. http://www.rcsb.org/pdb,2003.

[84] P. Verdino, K. Westritschnig, R. Valenta, and W. Keller. The cross-reactive calcium-binding pollenallergen, Phl p 7, reveals a novel dimer assembly. EMBO J., 21(19):5007�16, 2002.

[85] K. S. Thorn, H. E. Christensen, R. Shigeta, D. Huddler, L. Shalaby, U. Lindberg, N. H. Chua, andC. E. Schutt. The crystal structure of a major allergen from plants. Structure, 5(1):19�32, 1997.

[86] M. Gajhede, P. Osmark, F. M. Poulsen, H. Ipsen, J. N. Larsen, R. J. Joost van Neerven, C. Schou,H. Lowenstein, and M. D. Spangfort. X-ray and NMR structure of Bet v 1, the origin of birchpollen allergy. Nat. Struct. Biol., 3(12):1040�5, 1996.

[87] Z. Markovic-Housley, M. Degano, D. Lamba, E. von Roepenack-Lahaye, S. Clemens, M. Susani,F. Ferreira, O. Scheiner, and H. Breiteneder. Crystal structure of a hypoallergenic isoform of themajor birch pollen allergen Bet v 1 and its likely biological function as a plant steroid carrier.J. Mol. Biol., 325(1):123�33, 2003.

[88] A. A. Fedorov, T. Ball, N. M. Mahoney, R. Valenta, and S. C. Almo. The molecular basisfor allergen cross-reactivity: crystal structure and IgE-epitope mapping of birch pollen pro�lin.Structure, 5(1):33�45, 1997.

[89] E. W. Czerwinski, M. A. White, T. Midoro-Horiuti, E. G. Brooks, and R. M. Gol. 1.7 Angstromcrystal structure of Jun a 1, the major allergen from cedar pollen. http://www.rcsb.org/pdb/, onhold.

[90] U. Derewenda, J. Li, Z. Derewenda, Z. Dauter, G. A. Mueller, G. S. Rule, and D. C. Benjamin.The crystal structure of a major dust mite allergen Der p 2, and its biological implications. J. Mol.Biol., 318(1):189�97, 2002.

[91] J. Rouvinen, J. Rautiainen, T. Virtanen, T. Zeiler, J. Kauppinen, A. Taivainen, and R. Mantyjarvi.Probing the molecular basis of allergy. Three-dimensional structure of the bovine lipocalin allergenBos d 2. J. Biol. Chem., 274(4):2337�43, 1999.

[92] A. C. Pike, K. Brew, and K. R. Acharya. Crystal structures of guinea-pig, goat and bovine alpha-lactalbumin highlight the enhanced conformational �exibility of regions that are signi�cant for itsaction in lactose synthase. Structure, 4(6):691�703, 1996.


[93] E. D. Chrysina, K. Brew, and K. R. Acharya. Crystal structures of apo- and holo-bovine alpha-lactalbumin at 2. 2-Å resolution reveal an e�ect of calcium on inter-lobe interactions. J. Biol.Chem., 275(47):37021�9, 2000.

[94] J. J. Adams, B. F. Anderson, L. K. Creamer, and G. B. Jamson. The structure of bovine beta-lactoglobulin a in crystals grown at very low ionic strength. http://www.rcsb.org/pdb, 2002.

[95] M. B. Lascombe, C. Gregoire, P. Poncet, G. A. Tavares, I. Rosinski-Chupin, J. Rabillon,H. Goubran-Botros, J. C. Mazie, B. David, and P. M. Alzari. Crystal structure of the allergen Equc 1. A dimeric lipocalin with restricted IgE-reactive epitopes. J. Biol. Chem., 275(28):21572�7,2000.

[96] L. Kaiser, H. Gronlund, T. Sandalova, H. G. Ljunggren, M. van Hage-Hamsten, A. Achour, andG. Schneider. The crystal structure of the major cat allergen Fel d 1, a member of the secretoglobinfamily. J. Biol. Chem., 278(39):37730�5, 2003.

[97] D. E. Timm, L. J. Baker, H. Mueller, L. Zidek, and M. V. Novotny. Structural basis of pheromonebinding to mouse major urinary protein (Mup-I). http://www.rcsb.org/pdb/, 2001.

[98] B. N. Chaudhuri, G. J. Kleywegt, J. Bjorkman, L. D. Lehman-McKeeman, J. D. Oliver, and T. A.Jones. The structures of alpha 2u-globulin and its complex with a hyaline droplet inducer. ActaCryst., D55(Pt 4):753�62, 1999.

[99] X. Yang and K. Mo�at. Insights into speci�city of cleavage and mechanism of cell entry from thecrystal structure of the highly speci�c Aspergillus ribotoxin, restrictocin. Structure, 4(7):837�52,1996.

[100] S. Eswaramoorthy, P. J. Vithayathil, and M. A. Viswamitra. Crystallization and preliminary X-ray crystallographic studies of thermostable xylanase crystals isolated from Paecilomyces varioti .J. Mol. Biol., 243(4):806�8, 1994.

[101] S. Liu, A. A. Fedorov, T. D. Pollard, E. E. Lattman, S. C. Almo, and K. A. Magnus. Crystalpacking induces a conformational change in pro�lin-I from Acanthamoeba castellanii . J. Struct.Biol., 123(1):22�9, 1998.

[102] A. A. Fedorov, W. Shi, N. Mahoney, D. A. Kaiser, and S. C. Almo. A comparative structuralanalysis of pro�lins. http://www.rcsb.org/pdb, 2000.

[103] D. L. Scott, Z. Otwinowski, M. H. Gelb, and P. B. Sigler. Crystal structure of bee-venom phos-pholipase A2 in a complex with a transition-state analogue. Science, 250(4987):1563�6, 1990.

[104] Z. Markovic-Housley, G. Miglierini, L. Soldatova, P. J. Rizkallah, U. Muller, and T. Schirmer.Crystal structure of hyaluronidase, a major allergen of bee venom. Structure Fold. Des., 8(10):1025�35, 2000.

[105] M. Gribskov, L. Wesson, and D. Eisenberg. http://www.rcsb.org/pdb/, 1990.

[106] A. Henriksen, T. P. King, O. Mirza, R. I. Monsalve, K. Meno, H. Ipsen, J. N. Larsen, M. Gajhede,and M. D. Spangfort. Major venom allergen of yellow jackets, Ves v 5: structural characterizationof a pathogenesis-related protein superfamily. Proteins, 45(4):438�48, 2001.

[107] W. Steigemann and E. Weber. Structure of erythrocruorin in di�erent ligand states re�ned at1.4 Å resolution. J. Mol. Biol., 127(3):309�38, 1979.

[108] P. E. Stein, A. G. Leslie, J. T. Finch, and R. W. Carrell. Crystal structure of uncleaved ovalbuminat 1.95 Å resolution. J. Mol. Biol., 221(3):941�59, 1991.

BIBLIOGRAPHY 141

[109] M. Yamasaki, N. Takahashi, and M. Hirose. Crystal structure of S-ovalbumin as a non-loop-insertedthermostabilized serpin form. J. Biol. Chem., 278(37):35524�30, 2003.

[110] H. Kurokawa, B. Mikami, and M. Hirose. Crystal structure of diferric hen ovotransferrin at 2.4 Åresolution. J. Mol. Biol., 254(2):196�207, 1995.

[111] K. Mizutani, B. Mikami, and M. Hirose. Domain closure mechanism in transferrins: new view-points about the hinge structure and motion as deduced from high resolution crystal structures ofovotransferrin N-lobe. J. Mol. Biol., 309(4):937�47, 2001.

[112] K. Mizutani, B. K. Muralidhara, H. Yamashita, S. Tabata, B. Mikami, and M. Hirose. Anion-mediated Fe3+ release mechanism in ovotransferrin C-lobe: a structurally identi�ed SO2−

4 bindingsite and its implications for the kinetic pathway. J. Biol. Chem., 276(38):35940�6, 2001.

[113] P. G. Thakurta, D. Choudhury, R. Dasgupta, and J. K. Dattagupta. Tertiary structural changesassociated with iron binding and release in hen serum transferrin: a crystallographic and spectro-scopic study. Biochem. Biophys. Res. Commun., 316(4):1124�31, 2004.

[114] K. P. Wilson, B. A. Malcolm, and B. W. Matthews. Structural and thermodynamic analysis of com-pensating mutations within the core of chicken egg white lysozyme. J. Biol. Chem., 267(15):10842�9, 1992.

[115] J. Dong, T. J. Boggon, N. E. Chayen, J. Raftery, R. C. Bi, and J. R. Helliwell. Bound-solventstructures for microgravity-, ground control-, gel- and microbatch-grown hen egg-white lysozymecrystals at 1.8 Å resolution. Acta Cryst., D55(Pt 4):745�52, 1999.

[116] A. L. Swain, R. H. Kretsinger, and E. L. Amma. Restrained least squares re�nement of native(calcium) and cadmium-substituted carp parvalbumin using X-ray crystallographic data at 1.6-Åresolution. J. Biol. Chem., 264(28):16620�8, 1989.

[117] C. A. Reyes-Lopez, A. Hernandez-Santoyo, M. Pedraza-Escalona, G. Mendoza, A. Hernandez-Arana, and A. Rodriguez-Romero. Insights into a conformational epitope of Hev b 6.02 (hevein).Biochem. Biophys. Res. Commun., 314(1):123�30, 2004.

[118] C. A. Reyes-Lopez and A. Rodriguez-Romero. Crystal structure of a Hev b 6.02 isoallergen.http://www.rcsb.org/pdb, on hold.

[119] A. A. Fedorov, E. V. Fedorov, E. Ganglberger, H. Breiteneder, and S. C. Almo. A comparativestructural analysis of allergen pro�lins Hev b 8 and Bet v 2. http://www.rcsb.org/pdb, 2000.



[122] E. A. Padlan. X-ray crystallography of antibodies. Adv. Protein Chem., 49:57�133, 1996.

[123] D. C. Benjamin, J. A. Berzofsky, I. J. East, F. R. Gurd, C. Hannum, S. J. Leach, E. Margoliash,J. G. Michael, A. Miller, E. M. Prager, et al. The antigenic structure of proteins: a reappraisal.Annu. Rev. Immunol., 2:67�101, 1984.

[124] G. Schramm, H. Kahlert, R. Suck, B. Weber, H. T. Stuwe, W. D. Muller, A. Bufe, W. M. Becker,M. W. Schlaak, L. Jager, O. Cromwell, and H. Fiebig. Allergen engineering: variants of the timothygrass pollen allergen Phl p 5b with reduced IgE-binding capacity but conserved T cell reactivity.J. Immunol., 162(4):2406�14, 1999.


[125] G. E. Borgstahl, H. E. Parge, M. J. Hickey, W. F. Beyer, Jr., R. A. Hallewell, and J. A. Tainer.The structure of human mitochondrial manganese superoxide dismutase reveals a novel tetramericinterface of two 4-helix bundles. Cell, 71(1):107�18, 1992.


[127] H. Turner and J. P. Kinet. Signalling through the high-a�nity IgE receptor Fc epsilonRI. Nature,402(6760 Suppl):B24�30, 1999.

[128] R. Crameri and G. Walter. Selective enrichment and high-throughput screening of phage surface-displayed cDNA libraries from complex allergenic systems. Comb. Chem. High Throughput Screen.,2(2):63�72, 1999.

[129] R. Valenta and D. Kraft. Recombinant allergens for diagnosis and therapy of allergic diseases.Curr. Opin. Immunol., 7(6):751�6, 1995.

[130] R. Valenta, S. Vrtala, S. La�er, S. Spitzauer, and D. Kraft. Recombinant allergens. Allergy,53(6):552�61, 1998.

[131] O. Arquint, A. Helbling, R. Crameri, F. Ferreira, M. Breitenbach, and W. J. Pichler. Reducedin vivo allergenicity of Bet v 1d isoform, a natural component of birch pollen. J. Allergy Clin.Immunol., 104(6):1239�43, 1999.

Chapter 6

Final Discussion

An increasing incidence of allergies, a speci�c deviation of the immune system caused byIgE-mediated hypersensitivity reactions to environmental allergens, was observed in in-dustrial countries during the past decade. Immediate hypersensitivity either results fromsensitisation to unrelated substances or from cross-reactivity between homologous pro-teins of di�erent sources. Cross-reactivity also occurs between environmental allergensand homologous human proteins, which leads to humoral and cell-mediated autoreac-tivity to self antigens often observed in allergic patients su�ering from chronic atopicdiseases. This is surprising, since the presence of a human homologue to an environmen-tal allergen would be expected to induce immune tolerance. Intracellular proteins arenot likely to be accessible to the immune system under normal circumstances. However,due to tissue damage in in�ammatory processes, auto-antigens can be released favouringexacerbation and perpetuation of allergic diseases after primary sensitisation to envi-ronmental allergens [1, 2]. The reason, why certain substances lead to sensitisation, isunclear. To date, no structural features have been found, which might be responsible forthe onset of sensitisation and the switch to IgE production. In contrast, cross-reactivitycan be explained on a structural level. Cross-reactive proteins share common conservedIgE-binding residues, which cross-link IgE-molecules bound on e�ector cells. The IgE-binding epitopes can be characterised on a structural level by comparing the molecularsurface of cross-reactive allergens. At least one three-dimensional structure has to beknown, in order to de�ne the common solvent-exposed residues on the molecular surface;only these amino acids can interact with IgE molecules. Most accurate results will beobtained, when the second structure is also known. An alternative with highly identicalproteins is to generate the second structure by homology-modelling. If there are furthercross-reactive proteins known, their structures can be homology-modelled or - even moresimple - be compared on a sequential level, which will �ne-tune the comparative study.

143

144 CHAPTER 6. FINAL DISCUSSION

This is particularly true if the lengths of the amino acid sequences are similar.The feasibility of this approach was shown by Gajhede et al. [3] and Mirza et al. [4].Gajhede et al. solved the crystal structure of the major birch pollen allergen Bet v 1 andde�ned three conserved patches large enough to accommodate antibody-binding epitopesby comparing the structure with sequence data of homologous proteins. Later on, oneof these patches was con�rmed as an antibody binding epitope by the crystal structureof Bet v 1 in complex with the Fab fragment from a murine monoclonal IgG1 antibody(BV16) [4].The aim of this work was to obtain more structural information on allergens in generaland to characterise B-cell epitopes on cross-reactive fungal allergens. Crystal structuresof two fungal allergens from the yeast Malassezia sympodialis, a cyclophilin (Mala s 6)and a thioredoxin (Mala s 13), were solved. Crystallisation was in both cases not verystraightforward. With the C-terminal His-tag, Mala s 6 did not crystallise at all. Removalof non-structured, freely-movable parts like His-tags or loops and termini of proteinscan improve crystal contacts and crystallisation behaviour. Therefore, the His-tag wasremoved, which led to thin needles in a few crystallisation conditions, but which weredi�cult to optimise. Interestingly, after about one year, one beautiful crystal grew out ofa crystal screen condition. Crystallisation after such a long time happens rarely. Often,it is only salts, sometimes degradation of the protein leads to the crystallisation of onlya fragment. Sometimes, a drastic change in the crystallisation condition like drying outcan lead to crystals, which must have been the case with Mala s 6. Obviously, this kind ofcrystallisation is hardly reproducible. The crystal was of high quality di�racting to 1.5 Åresolution and structure determination was straightforward. The structure was solved bymolecular replacement using human CyP A as a search model. The excellent statisticson data processing and re�nement (Table 2.1) actually imply that the crystal would havedi�racted beyond 1.5 Å. Astonishingly, only one molecule is located in the asymmetricunit, which results in a solvent content of 69%. For such a high solvent content, theexcellent resolution is remarkable. Crystals of membrane proteins often feature solventcontents in this range, but typically only di�ract to around 4 Å.Besides Mala s 13, the cross-reactive thioredoxin of wheat (Triticum aestivum), whichacts as a food allergen, was crystallised. Both proteins posed similar problems. First,the N-terminal His-tags were not removable, despite the presence of a thrombin cleavagesite. It was assumed that the cleavage site is too close to the �rst structured elementin both proteins leaving the cleavage site inaccessible to thrombin. Therefore, Mala s 13was crystallised with the His-tag yielding one crystallisation condition with small andclustered crystals, which �nally could be improved to bigger single crystals. In a parallel

145

work, it was attempted to remove the His-tag by the insertion of two and three aminoacids between the His-tag and the protein coding sequence of Mala s 13 and wheat Trx,respectively. This approach was successful resulting in proteins with thrombin-cleavableHis-tags and supporting the hypothesis that the cleavage site was not accessible. Thesolved crystal structure of Mala s 13 indeed showed that the very �rst amino acid after thethrombin cleavage site is involved in a β-strand, thus obstructing the cleavage site. Thee�ect of the cleaved Mala s 13 protein on crystallisation behaviour was dramatic. Niceand big crystals grew readily out of several screening conditions. Crystallisation of wheatTrx was more di�cult, but was successful after increasing the protein concentration upto 42 mg/ml.Both cleaved proteins yielded a few di�erent crystal forms, all di�racting to high reso-lutions (1.6-2.5 Å). Unfortunately, none of them were solvable. One Mala s 13 crystalform was perfectly twinned, the other one showed non-crystallographic translation andwas not solvable by molecular replacement. One wheat Trx crystal form was not process-able indicating disorder, the other one also showed translational NCS. It is unclear, whyboth proteins show similar problems, despite the fact that they feature di�erent spacegroups. Both proteins are fairly globular and compact. Moreover, they vaguely featurea two-fold symmetry within the molecule; rotating the central β-sheet with the four sur-rounding α-helices by about 180◦ roughly results in the same overall fold. This mightlead to alternative incorporation of molecules into the crystal lattice yielding twinned ordisordered crystals. The vague intramolecular symmetry certainly complicates molecularreplacement. Finally, structure solution of the non-cleaved Mala s 13 was successful. Inthe beginning, unclustered and large crystals were di�cult to grow. They did not lookvery promising due to the lack of even planes and clear edges, but in the end, they turnedout to be of excellent di�racting quality. They belong to space group P21 and di�ractto 1.4 Å. Structure determination by molecular replacement was straightforward usingChlamydomonas reinhardtii thioredoxin H as a search model, leading to a well re�nedmodel (Table 3.1).In order to compare structural features on the surface of cross-reactive allergens, high-resolution crystal structures are a prerequisite, with all loops and surface residues wellde�ned. The solved structures of Mala s 6 and Mala s 13 both meet these requirements.They were compared to the relevant homologous human protein; Mala s 6 to humanCyP B and Mala s 13 to human Trx. Both human proteins act as self-antigens leading toautoimmune reactions often observed in allergic patients with severe atopic diseases. Onlythose residues that are solvent exposed and conserved in both structures can contributeto IgE-mediated cross- and autoreactivity. Three conserved surface patches were de�ned


on the cyclophilins (Figure 2.5) and two on the thioredoxins (Figure 3.6). They covera surface area between 350 Å2 and 1017 Å2 comprising 4 to 17 amino acids, whichare located on a few di�erent loops. Thus, all patches are discontinuous and three-dimensional. Often, about half or more of the residues are contributed by only a singleloop forming the core of the epitope, surrounded by a few residues from other loops. Theproperties of the described epitopes are in accordance with observations made in crystalstructures of antibody-antigen complexes. Laver et al. [5] conclude that epitopes occupylarge areas comprised of 15-22 amino acids, which lie on two to �ve surface loops. Eachepitope has a buried surface area on the antigen of 650-900 Å2. A smaller subset of 5-6residues contributes most of the binding energy [6].The de�ned conserved and solvent-exposed residues on Mala s 6 and Mala s 13 aremore or less randomly distributed over the whole sequence (Figures 2.1 and 3.7), but -interestingly - form only a few contiguous patches when mapped on the three-dimensionalsurface (Figures 2.5 and 3.6). Moreover, these patches are not randomly distributed overthe surface but are accumulated on one side of the molecules. In both protein families,one side is more conserved than the other. Therefore, the putative epitopes de�ned inthis study might be relevant and not just random noise. Why are certain surface areasconserved among the described homologous proteins? CyPs as well as Trxs have to bindto their target protein in order to exert their enzymatic activity. Both protein familiesinteract with their target proteins via a conserved area that is located around the activesite. It is thus not surprising that this contact area can act as a cross-reactive B-cellepitope. Additionally, both protein families are known to have many more biologicalfunctions in vivo. Some of them might also be dependent on conserved surface areas inorder to interact with putative target molecules.There are alternative methods in characterising cross-reactive B-cell epitopes. The onlymethod allowing a complete de�nition of a B cell epitope remains co-crystallisation of theallergen with a monoclonal antibody Fab fragment. This method requires homogeneousreagents for growth of crystals and cannot be performed using polyclonal human serumIgE. Since monoclonal allergen-speci�c human IgE is di�cult to obtain, monoclonal anti-bodies from other sources are used. To date, only one allergen-antibody complex hasbeen solved; the complex described above between Bet v 1 and a murine monoclonalIgG1 antibody [4]. This method is very time-consuming. A simpler, comparative studywas used in this work involving the solution of a crystal structure. Even more straightfor-ward is homology-modelling, when there is high homology among proteins and at leastone structure known. Cross-reactivity usually occurs only among proteins, which aremore than 40% identical. This degree of identity also allows the generation of homology-

147

models. One disadvantage of homology-models is that certain loops cannot be modelledaccurately, due to insertions or deletions, or due to conformational deviations. In thecase of Mala s 13, patch 3 (Figure 3.6 C) could be ruled out as an epitope due to theproximity of a deletion, which could clearly be allocated in the crystal structure. An-other possibility is to compare proteins on the sequential level with a structurally knownprotein. This might be helpful, when comparing several structures in order to de�neepitopes, which are responsible for overall cross-reactivities. This method was also usedin this work. The alignment of the two known CyP structures with two additional cross-reactive CyP sequences revealed that two of the three patches are conserved among allfour proteins (Figure 2.1). Similarly, the alignment of the two known Trx structures withfour additional Trx sequences showed that both patches are conserved among all six Trxs(Figure 3.7). In order to cross-link IgE receptors on the surface of e�ector cells, a mini-mum of two identical or di�erent IgE-binding epitopes per molecule pair is required. Inboth protein families, two di�erent putative B-cell epitopes can be de�ned, thus ful�llingthe required criteria.In conclusion, the taken approach - the structural comparison of cross-reactive allergens- is a useful and fast way to identify putative B-cell epitopes. Site-directed mutagenesis,followed by analysis of the IgE-binding capacity of the mutants will reveal the actualcontribution of individual residues to the total IgE-binding of the allergen. If mutationof these residues signi�cantly reduces the IgE a�nity of the molecule, it may be con-cluded that the identi�ed cross-reactive epitope is immunodominant. Disruption of theconformation of a putative epitope by an insertion or deletion might also be an e�ectiveway of reducing IgE a�nity. Besides new insights into the molecular basis of cross-reactivity, identi�cation of IgE-binding epitopes also has therapeutical implications. Theonly curative treatment for allergy today is immunotherapy, based on repeated subcuta-neous injections of allergen. One major disadvantage of current immunotherapy is thatit can cause severe side e�ects, such as asthma attacks and anaphylactic shock. Theidenti�cation of IgE epitopes on allergens allows the modi�cation of important allergenssuch that they display strongly reduced allergenic activity by disrupting the conforma-tional IgE epitopes. Such hypoallergenic allergen derivatives can be used as candidatesfor vaccines in allergen-speci�c immunotherapy, with a reduced risk of immediate sidee�ects [7]. Another therapeutic approach is based on the saturation of allergic e�ectorcells with allergen derivatives unable to induce cross-linking of e�ector cell-bound IgE,thus preventing mediator release by subsequent contact with the intact allergen [8].


On one hand, it is reasonable and straightforward to homology model proteins witha certain degree of identity to a structurally known protein. On the other hand, theconformation of a homologous protein is not always predictable due to mechanisms likeinduced �t, movement of loops or whole domains or domain swapping. Structure solutionof Aspergillus fumigatus cyclophilin (Asp f 11) revealed such an unpredictable event. Aspf 11 is a fungal allergen, which is homologous to and cross-reactive with Mala s 6 andhuman CyP B. Crystals that di�racted to about 3.0 Å resolution on a home source wereinitially obtained by S. Flückiger, but they were not solvable by molecular replacement.In this work, the resolution could be improved by about 1 Å by removing the N-terminalHis-tag and the following 9 amino acid linker and by co-crystallisation with the dipeptideAla-Pro. Despite the better data, molecular replacement still was not feasible. In orderto �nd another crystal form that might be solvable, more crystallisation screening andoptimisation was performed. This resulted in two new crystallisation conditions, butunfortunately, they showed the same crystal form as the original one. The structure was�nally solved by the multiwavelength anomalous dispersion method with data collectedat the synchrotron to 1.85 Å resolution.Once the structure was solved, the improvement in resolution and the failure in molec-ular replacement was partly explainable. The N-terminus is located in the vicinity of acrystal contact. This contact might have been disturbed by the His tag of the originalconstruct. Interestingly, the dipeptide is not visible in the electron density, neither inthe active site nor on the surface. Probably, it altered the physicochemical properties ofthe crystallisation condition leading to better packed crystals. Di�erent reasons mightbe responsible for the failure in molecular replacement. First, the highly intertwineddimer leads to a close assembly of the two subunits, which could not have been predictedassuming two independent monomers. Second, there are conformational di�erences inthe domain swapped element and the active site, as well as slight rearrangements in thevicinity of the active site. Third, 66 out of 372 residues per asymmetric unit are notvisible in the electron density due to disorder. Last, the average B-factor is rather highat 43.7 Å2.The Asp f 11 crystal structure was originally also attempted to be used for the char-acterisation of cross-reactive B-cell epitopes. Once the structure was solved, it becameobvious that this procedure was not feasible anymore. First, the monomeric rather thanthe dimeric form is most probably the allergenic form, naturally occurring in the environ-ment. Second, a large part of the surface is not de�ned due to the disorder of the abovementioned 66 residues. Nevertheless, the structure turned out to be very interesting froma structural point of view.

BIBLIOGRAPHY 149

The structure revealed dimerisation by 3D domain swapping representing one of the �rstproteins with a swapped central domain. This behaviour has not been observed amonghomologous cyclophilins, which all feature a monomeric conformation. The swappedregion, which is binding into its own active site, as well as the second hinge loop are wellde�ned in the experimental electron density. The side chains of all amino acids are visible,except of Lys128, which is solvent-exposed. The protein was shown to be monomericin solution, but dimerisation was inducible under high salt and protein concentrations.Therefore, the protein exists in two di�erent conformations. It is not clear yet, whetherthe dimeric form has any biological meaning or pharmaceutical implication. The crystalstructure implies that the dimer must be inactive; a conserved tryptophan located on thedomain swapped element binds into its own active site, which is responsible for the cis-trans isomerisation activity. Indeed, inactivity of the dimer - in contrast to the monomer -could be demonstrated by an activity assay. Therefore, switching between the monomericand dimeric form could be a means of biological regulation.Freeze-thawing of Asp f 11 led to misfolded dimers and oligomers, which formed inter-molecular disul�de bonds under oxidising conditions. These forms can be explained byalternative 3D domain swapping leading to N- or C-terminally swapped dimers, oligomersand aggregates (Figure 4.6). Besides biological regulation, domain swapping was pro-posed to be a mechanism for the emergence of oligomeric proteins, but also to be apotentially harmful process leading to misfolding and aggregation [9]. Further workmight focus on the characterisation of the di�erent misfolded oligomers and on homolo-gous proteins, in order to investigate, if they show the same oligomerisation behaviouras A. fumigatus cyclophilin.

Bibliography[1] R. Crameri, A. Faith, S. Hemmann, R. Jaussi, C. Ismail, G. Menz, and K. Blaser. Humoral and

cell-mediated autoimmunity in allergy to Aspergillus fumigatus. J. Exp. Med., 184(1):265�70, 1996.


[3] M. Gajhede, P. Osmark, F. M. Poulsen, H. Ipsen, J. N. Larsen, R. J. Joost van Neerven, C. Schou,H. Lowenstein, and M. D. Spangfort. X-ray and NMR structure of Bet v 1, the origin of birch pollenallergy. Nat. Struct. Biol., 3(12):1040�5, 1996.



[5] W. G. Laver, G. M. Air, R. G. Webster, and S. J. Smith-Gill. Epitopes on protein antigens: miscon-ceptions and realities. Cell, 61(4):553�6, 1990.



[8] R. Valenta and D. Kraft. Recombinant allergens for diagnosis and therapy of allergic diseases. Curr.Opin. Immunol., 7(6):751�6, 1995.

[9] F. Rousseau, J. W. Schymkowitz, and L. S. Itzhaki. The unfolding story of three-dimensional domainswapping. Structure, 11(3):243�51, 2003.

ACKNOWLEDGEMENTS 151

AcknowledgementsI would like to thank all the people who supported and helped me during my thesis andwho made everyday work more enjoyable:

Prof. Dr. Leonardo Scapozza for giving me the opportunity to work in his group in thefascinating �eld of X-ray crystallography, for guidance and support during the last threeyears and for his con�dence in my work. Prof. Dr. Reto Crameri who gave me the chanceto work on this exciting project. He always took the time for interesting discussions andgave me competent help and support throughout my thesis. Prof. Dr. Gerd Folkers forhis interest in my work, fruitful discussions and support as co-referee.

Markus Birringer for sharing the pain of crystallisation and the joy of structure solution,for sharing the o�ce, the lab and some working material and for brightening up life duringthe last few months, when people got fewer and fewer. Thomas Kuoni who was keen onlearning about X-ray crystallography and NMR and was looking up to the excellent workof crystallographers. He always had the time for a small talk in his o�ce or a coolbeer out and about. Christa Meier, my diploma student, and Corina Glanzmann, mysemester student, who worked both with great enthusiasm and success on my project.Remo Perozzo for tips in the lab, Marc Gasser for help in computation, Thomas Schürpffor MS measurements and Vivianne Otto for managing our stay at Hönggerberg.

Michael Weichel for assistance in the lab during my short stay in Davos, for his hospitalityand the excellent morning cacao. Sabine Flückiger for help and assistance and for passingon some beautiful but nasty crystals, which turned out to be gems after structure solution.Andreas Glaser who complemented my work by measuring IgE-binding and cross-reactivi-ties of cyclophilins and thioredoxins.

Daniel Kloer for great help with some of my tricky datasets. He is a crack in compu-tational crystallography. Dirk Kostrewa and Peer Mittl for assistance in data collectionand for giving good tips in X-ray crystallography. Dirk is excellent in teaching abstracttheories in an understandable and simple way.

All my working colleagues who made everyday work a bit more colourful.

All my family, friends and especially Karin for giving help, support, encouragement andbalance.

152 PUBLICATIONS

List of publicationsPublicationsS. Flückiger, A. Limacher, A. G. Glaser, L. Scapozza, and R. Crameri. Structural aspectsof cross-reactive allergens. Recent Res. Devel. Allergy & Clin. Immunol., 5:57-75, 2004.

A. Limacher, D. P. Kloer, S. Flückiger, G. Folkers, R. Crameri, and L. Scapozza. Thecrystal structure of Aspergillus fumigatus cyclophilin reveals 3D domain swapping of acentral element. Structure, in press, 2005.

A. G. Glaser1, A. Limacher1, S. Flückiger, A. Scheynius, L. Scapozza, and R. Crameri.Analysis of the cross-reactivity and of the 1.5 Å crystal structure of the Malassezia sym-podialis Mala s 6 allergen, a member of the cyclophilin pan-allergen family. Submitted,2005.

A. Limacher1, A. G. Glaser1, C. Meier, R. Crameri, and L. Scapozza. The crystal struc-ture of Malassezia sympodialis thioredoxin Mala s 13, a member of a new pan-allergenfamily. In preparation, 2005.

Oral presentationsCloning, expression and crystallisation of cross-reactive allergens. Swiss Institute of Al-lergy and Asthma Research (SIAF). Davos, Switzerland, December 2002.

Structural characterisation of cross-reactive allergens. EMBO course on automated ma-cromolecular structure solution (AMSS). NKI, Amsterdam, Netherlands, May 2004.

Structural characterisation of cross-reactive allergens. Doktorandentag of the Institute ofPharmaceutical Sciences. ETH, Zurich, Switzerland, October 2004.

Identi�cation of putative cross-reactive B-cell epitopes on cyclophilins and thioredoxins.Swiss Institute of Allergy and Asthma Research (SIAF). Davos, Switzerland, May 2005.

PostersA. Limacher, C. Meier, R. Crameri, and L. Scapozza. Structural characterisation of cross-reactive allergens. Pharmaday of the Center of Pharmaceutical Sciences Basel-Zurich.Biozentrum Basel, Switzerland, July 2003.

1equally contributing

CV 153

Curriculum vitaePersonal detailsName Andreas LimacherAddress Sternegg 6, 6005 LuzernBorn April 13th, 1973 in ChamCitizen of Inwil, Luzern, Switzerland

Education05/2005 Graduation as Doctor of Sciences, Swiss Federal Institute of

Technology (ETH), Zurich05/2004 EMBO course on `Automated Macromolecular Structure Solution',

NKI, Amsterdam2002 - 2005 Dissertation on `Structural Characterisation of Cross-reactive

Allergens' at the Institute of Pharmaceutical Sciences, ETH,Zurich, in collaboration with the Swiss Institute of Allergy andAsthma Research (SIAF), Davos

2002 - 2005 Training as high school chemistry teacher, ETH, Zurich10/2001 Graduation in Biochemistry at University of Berne2000 - 2001 Diploma work in protein crystallography with Prof. U. Baumann,

Dept. of Chemistry and Biochemistry, University of Berne1997 - 2001 Studies in Biology/Biochemistry at University of Berne1996 Cambridge Course in Advanced English in Sydney, Australia1989 - 1994 Teacher Training School in Luzern1986 - 1989 High School in Hochdorf

Professional experience2002 - 2005 Teaching assistant at ETH, Zurich1997 - 1999 Part-time German teacher for adults at Sfb, Luzern1997 Part-time teacher at di�erent primary schools1994 - 1996 Primary teacher 3rd/4th grade in Reussbühl

Documents

Permanent Link: Research Collection · 12 SUMMARY cessfulbythemultiwavelengthanomalousdispersion(MAD)method.Thestructurewas re ned to an R and Rfree value of 18.9% and 21.4%, respectively