Characterizationofantibodyspeciﬁcity …kth.diva-portal.org/smash/get/diva2:762237/FULLTEXT01.pdf · 2014. 11. 11. · Thesis defense This thesis will be defended November 28th

Characterization of antibody specificityusing peptide array technologies

Björn Forsström

Kungliga Tekniska Högskolan, KTHRoyal Institute of Technology

School of BiotechnologyStockholm 2014

c© Björn Forsström 2014

KTH, Royal Institute of TechnologySchool of BiotechnologyDivision of Proteomics and NanobiotechnologyScience for Life LaboratoryTomtebodavägen 23A171 65 Solna, Sweden

ISBN 978-91-7595-316-8TRITA-BIO Report 2014:16ISSN 1654-2312

Printed by US-AB 2014

Abstract

Antibodies play an important role in the natural immune response to invadingpathogens. The strong and specific binding to their antigens also make themindispensable tools for research, diagnostics and therapy.

This thesis describes the development of methods for characterization of an-tibody specificity and the use of these methods to investigate the polyclonalantibody response after immunization. Paper I describes the developmentof an epitope-specific serum fractionation technique based on epitope map-ping using overlapping peptides followed by chromatographic separation ofpolyclonal serum. This technique together with another epitope mappingtechnique based on bacterial display of protein fragments were then used togenerate antibody sandwich pairs (Paper I), investigate epitope variations ofrepeated immunizations (Paper II) and to determine the ratio of antibodiestargeting linear and conformational epitopes of polyclonal antibodies (PaperIII). Paper IV describes the optimization of in situ-synthesized high-densitypeptide arrays for epitope mapping and how different peptide lengths influ-ence epitope detection and resolution. In Paper V we show the developmentof planar peptide arrays covering the entire human proteome and how thesearrays can be used for epitope mapping and off-target binding analysis. InPaper VI we show how polyclonal antibodies targeting linear epitopes canbe used for peptide enrichment in a rapid, absolute protein quantificationprotocol based on mass spectrometry.

Altogether these investigations demonstrate the usefulness of peptide arraysfor fast and straightforward characterization of antibody specificity. Thework also contributes to a deeper understanding of the polyclonal anti-body response obtained after immunization with recombinant protein frag-ments.

Keywords: Antibody, Epitope mapping, Peptide array, Suspension beadarray, Antigen, Specificity, Cross-reactivity, Immunization, Immunogenic-ity

iii

Populärvetenskaplig sammanfattning

Antikroppar är molekyler i kroppens immunförsvar som skyddar oss genom attneutralisera till exempel virus, bakterier och gifter. Vid en infektion eller immu-nisering (vaccinering) bildas det antikroppar som specifikt binder till de antigen,d.v.s. främmande molekyler och celler, som orsakat antikroppssvaret. Den mycketspecifika interaktionen mellan en antikropp och dess antigen är en egenskap somgör antikroppar till utmärkta forskningsredskap för att detektera kroppens olikaproteiner, men också för att diagnostisera och behandla sjukdomar.

För att ta fram en antikropp som är specifik mot ett visst protein så kan manimmunisera ett djur med proteinet. Djuret producerar då antikroppar mot pro-teinet och dessa kan sedan renas fram ur djurets serum. För att kunna lita påresultat från antikroppsbaserade tester så måste man först validera antikroppensspecificitet mot antigenet så att man vet att den faktiskt binder det den ska ochatt den inte binder andra proteiner. En viktig del av valideringen är att ta redapå vilka delar av proteinet som antikroppen binder till. Dessa delar kallas epitoperoch det finns många olika tekniker för att kartlägga var dessa är lokaliserade påproteinet, s.k epitopmappning.

Den här avhandlingen är baserad på fem publicerade artiklar och ett manuskriptsom har epitopmappning som röd tråd. I Paper I, II och III används epitopmapp-ning tillsammans med en epitopspecifik fraktionering för att få en ökad förståelse fördet antikroppssvar som uppstår efter immunisering med proteinfragment. Vi visaratt bara ett fåtal av alla antikroppar med olika epitopspecificiteter bidrar till denönskade bindningen i flera vanliga antikroppsbaserade analyser och att antikrop-parna till största del binder till linjära epitoper. Vi visar också att antikroppar somkommer från upprepade immuniseringar med samma antigen har stora likheter iepitopspecificitet, men att de är långt ifrån identiska. Paper IV och V beskriverutvecklingen av peptidarrayer med hög densitet, d.v.s. ordnade ytor av miljontalsproteindelar och hur dessa kan användas för epitopmappning samt analys av an-tikroppars bindning till delar av alla människans proteiner. I Paper VI beskrivervi utvecklingen av en metod för att med hjälp av antikroppar anrika peptider frånbiologiska prover och hur det tillsammans med masspektrometri möjliggör snabbanalys av flera proteiner på samma gång.

iv

Thesis defense

This thesis will be defended November 28th 2014 at 10.15, in Gar-daulan, Folkhälsomyndigheten, Nobels väg 18, Solna, for the degreeof Doctor of Technology in Biotechnology.

Respondent:Björn Forsström graduated as master of science in engineering from KTHBiotechnology in 2008 and worked as research engineer within the HumanProtein Atlas-project before starting his PhD studies at the Division of Pro-teomics and Nanobiotechnology, KTH.

Faculty opponent:Markus Templin, Head of the Department of Assay Development at theNatura and Medical Sciences Institute (NMI), University of Tübingen, Ger-many

Evaluation committee:Marita Troye-Blomberg, Professor at Stockholm University, The Wenner-Gren Institute, Immunology

Mats A. A. Persson, Associate Professor at Karolinska Institutet, Depart-ment of Clinical Neuroscience, Center for Molecular Medicine

Ola Söderberg, Senior lecturer at Uppsala University, Department of Im-munology, Genetics and Pathology

Chairman:Stefan Ståhl, professor at KTH, School of Biotechnology, Division of ProteinTechnology

Main supervisor:Mathias Uhlén, professor at KTH, School of Biotechnology, Division of Pro-teomics and Nanobiotechnology

Co-supervisor:Henrik Johannesson, Group Leader, Research and Development, Atlas Anti-bodies AB.

v

List of publications

The presented thesis is based on the following six articles, referred to by theirRoman numerals (I-VI). All articles are included in the Appendix.

Paper I - Barbara Hjelm, Björn Forsström, Ulrika Igel, Henrik Johan-nesson, Charlotte Stadler, Emma Lundberg, Fredrik Pontén, Anna Sjöberg,Johan Rockberg, Jochen M. Schwenk, Peter Nilsson, Christine Johansson,Mathias Uhlén (2011). Generation of monospecific antibodies based on affin-ity capture of polyclonal antibodies. Protein Science 20(11): 1824-35 doi:10.1002/pro.716

Paper II - Barbara Hjelm, Björn Forsström, John Löfblom, Johan Rock-berg, Mathias Uhlén (2012). Parallel immunizations of rabbits using thesame antigen yield antibodies with similar, but not identical, epitopes. PLoSOne 7(12): e45817 doi: 10.1371/journal.pone.0045817

Paper III - Björn Forsström, Barbara Bisławska Axnäs, Johan Rockberg,Hanna Danielsson, Anna Bohlin, Mathias Uhlén. Dissecting antibodies withregards to linear and conformational epitopes. Manuscript submitted to PLoSOne

Paper IV - Søren Buus, Johan Rockberg, Björn Forsström, Peter Nilsson,Mathias Uhlén, Claus Schafer-Nielsen (2012) High-resolution mapping of lin-ear antibody epitopes using ultrahigh-density peptide microarrays. Molecular& Cellular Proteomics 11(12): 1790-800 doi: 10.1074/mcp.M112.020800

Paper V - Björn Forsström, Barbara Bisławska Axnäs, Klaus-Peter Sten-gele, Jochen Bühler, Thomas J. Albert, Todd A. Richmond, Francis JingxinHu, Peter Nilsson, Elton P. Hudson, Johan Rockberg, Mathias Uhlén (2014).Proteome-wide epitope mapping of antibodies using ultra-dense peptide ar-rays. Molecular & Cellular Proteomics 13(6): 1585-97 doi: 10.1074/mcp.M113.033308

vi

Paper VI - Fredrik Edfors*, Tove Boström*, Björn Forsström, MarlisZeiler, Henrik Johansson, Emma Lundberg, Sophia Hober, Janne Lehtiö,Matthias Mann, Mathias Uhlén (2014). Immunoproteomics using polyclonalantibodies and stable isotope-labeled affinity-purified recombinant proteins.Molecular & Cellular Proteomics 13(6): 1611-24 doi: 10.1074/mcp.M113.034140

*Both authors contributed equally to the work.

vii

Abbreviations

3D Three-dimensionalANLN Actin-binding protein anilinBCR B cell receptorC-terminus Carboxy-terminusCDR Complementarity determining regionCNDP1 Carnosine dipeptidase 1CTL Cytotoxic T lymphocyteDNA Deoxyribonucleic acidFab Fragment antigen bindingFc Fragment crystallizableIg ImmunoglobulinKD Equilibrium dissociation constantkoff Association rate constantkon Dissociation rate constantLC-MS/MS Liquid chromatography-tandem mass spectrometryMHC Major histocompatibility complexN-terminus Amino-terminusPCR Polymerase chain reactionPODXL Podocalyxin-like proteinPrEST Protein epitope signature tagRBM3 RNA-binding protein 3SATB2 Special AT-rich binding protein 2scFv Single-chain variable fragmentsiRNA Short interfering ribonucleic acidTCR T cell receptorTH cell Helper T cellTYMP Thymidine phosphorylase

ix

Contents

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiPopulärvetenskaplig sammanfattning . . . . . . . . . . . . . . . . . ivThesis defense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of publications . . . . . . . . . . . . . . . . . . . . . . . . . . . viAbbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

1 Introduction 1Antibodies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Antibody structure . . . . . . . . . . . . . . . . . . . . . . . . 1The immune system . . . . . . . . . . . . . . . . . . . . . . . 3Antibody diversity . . . . . . . . . . . . . . . . . . . . . . . . 3B cell activation, clonal expansion, affinity maturation and

class switching . . . . . . . . . . . . . . . . . . . . . . 6Antibodies for research, diagnosis and therapy . . . . . . . . . . . . 9

Epitopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Immunization and generation of polyclonal antibodies . . . . . 9Immunogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Monoclonal antibodies . . . . . . . . . . . . . . . . . . . . . . 12

Array technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Protein microarrays . . . . . . . . . . . . . . . . . . . . . . . . 14Peptide microarrays . . . . . . . . . . . . . . . . . . . . . . . . 15Suspension bead arrays . . . . . . . . . . . . . . . . . . . . . . 17Reverse phase protein arrays . . . . . . . . . . . . . . . . . . . 17Antibody arrays . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Characterization of antibody binding . . . . . . . . . . . . . . . . . 18

x

CONTENTS

Affinity determination . . . . . . . . . . . . . . . . . . . . . . 18Cross-reactivity (off-target binding) . . . . . . . . . . . . . . . 19Epitope mapping . . . . . . . . . . . . . . . . . . . . . . . . . 20

2 Aims of the thesis 24

3 Present investigations 25Generation of monospecific antibodies based on affinity cap-

ture of polyclonafl antibodies. (Paper I) . . . . . . . . 25Parallel immunizations of rabbits using the same antigen yield

antibodies with similar, but not identical, epitopes.(Paper II) . . . . . . . . . . . . . . . . . . . . . . . . . 28

Dissecting antibodies with regards to linear and conforma-tional epitopes. (Paper III) . . . . . . . . . . . . . . . 29

High-resolution mapping of linear antibody epitopes using ultrahigh-density peptide microarrays. (Paper IV) . . . . . . . . 30

Proteome-wide Epitope Mapping of Antibodies Using Ultra-dense Peptide Arrays. (Paper V) . . . . . . . . . . . . 30

Immunoproteomics using polyclonal antibodies and stable isotope-labeled affinity-purified recombinant proteins. (PaperVI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4 Concluding remarks and future perspectives 35

5 Acknowledgements 39

6 References 41

xi

Introduction

Antibodies

Secretion of antibodies is one of the tools used by the immune system ofhigher organisms to ward off foreign pathogens such as bacteria, viruses andparasites. When the immune system recognizes something as foreign, it willtry to neutralize this potential threat and clear it from the body. To achievethis, dedicated cells of the immune system kill infected cells, e.g. to preventfurther spread of viruses to uninfected cells, and produce antibodies capableof strong and specific binding to pathogens and toxins existing outside thecells. This specific binding makes antibodies attractive as tools for scientistsin their effort to study the molecules that make up life, especially proteins.Proteins carry out most of the functions in our bodies, but are too small tobe studied using conventional microscopes. Instead antibodies specific forthe protein of interest are used for detection and quantification in tissues,cells, bodily fluids etc. [1, 2].

Antibody structure

Antibodies are Y-shaped proteins consisting of four polypeptide chains, twoidentical heavy chains and two identical light chains linked by disulfide bridges[3]. If an antibody is cleaved using the protease papain, three fragments willbe formed. The arms of the antibody form two identical antigen-binding frag-ments, Fabs, and the stem forms the fragment crystallizable, Fc [4].

1

Introduction

Fc

Fab

Hinge

Lightchain

Heavychain

Variable domain

CDR

Figure 1.1: Schematic drawing of an Immunoglobulin G antibody.CDR, complementarity determining region. Fab, fragment antigen binding.Fc, fragment crystallizable.

The antigen binding sites, or complementarity determining regions (CDRs)are located at the tip of each arm of the antibody and consist of six hypervariable loops, three each from the variable domains of the light and heavychains [5]. In humans and mice, the constant region of the heavy chains areencoded by the non-variable genes of one of the five heavy chain isotypes(δ, µ, α, γ, ε), which define the five major antibody classes, IgD, IgM, IgA,IgG and IgE, respectively. The constant regions are responsible for effectorfunctions such as recruitment of complement proteins and ability to passmembranes, which varies between the different classes [6]. In mammals thereare two types of antibody light chains, κ or λ [5]. However, certain speciessuch as camelids also have antibodies entirely lacking light chains and in theseantibodies the antigen binding is mediated only by the three CDR-loops onthe heavy chains [7].

2

Introduction

The immune system

The immune system of vertebrates is generally divided in two branches, in-nate and adaptive immunity. These systems are however not completelyseparate and influence each other via chemokines and other cytokines. Theinnate immune system, which is also found in less complex animals like in-sects, can for instance detect viral DNA and RNA or lipopolysaccharides ofbacterial cell walls, and in response produce unspecific antimicrobial sub-stances [8].

The adaptive immune system on the other hand, mounts a targeted responsespecific for the invading pathogen and, once trained to recognize moleculesfrom that pathogen, will form a memory of this infection and respond muchquicker when challenged a second time by the same pathogen. The adap-tive immune system can in turn be divided into cell-mediated and humoralimmunity [5].

In cell-mediated immunity, cytotoxic T lymphocytes (CTLs) recognize anti-gens on infected cells and force these cells to undergo controlled death, apop-tosis. Other parts of cell-mediated immunity include activating helper Tcells (TH cells), non-specific natural killer cells, phagocytic macrophages andneutrophils. The other branch, humoral immunity, encompasses the B lym-phocytes (B cells) and the antibodies they secrete into the circulatory sys-tem [5,7].

Antibody diversity

Under the right circumstances the immune system can produce antibodieswith affinity for almost any molecular surface, even including entirely newsynthetic molecules never before seen in nature. To achieve this, the aminoacid sequence of the CDR must be highly variable and capable of forming acomplimentary surface for any possible antigen.

3

Introduction

Antibodies are produced by B cells and during development each B cell ran-domly rearranges antibody gene segments encoding the variable regions ofthe heavy and light chains. Heavy chain germ-line DNA contain approxi-mately 39 V (variable), 27 D (diversity) and J (joining) gene segments [6].Rearrangement (see Figure 1.2) starts with joining of one D- and one J-segment, which are then joined to a V-segment. The lymphocyte specificenzymes RAG1 and RAG2 introduce breaks in the double stranded DNAnext to each of the two gene segments. The cut ends are then joined to-gether in an imprecise manner by DNA-repair enzymes and in this processadditional nucleotides are incorporated. The added nucleotides can encodeextra amino acids, but can also cause frameshifts that alter the amino acidssequence encoded by the following gene segment [5].

After rearrangement of heavy-chain genes, a primary RNA transcript isformed that contains the combined VDJ-segment, unused J-segments andthe Cµ or Cδ constant region genes. Alternative splicing then joins a leadingsequence (L) and the VDJ-segment to either one of the two C-genes, creatingmRNA encoding the heavy-chains of membrane bound IgM or IgD.

Light chain rearrangement is similar to that of the heavy chain, but thegerm-line DNA does not contain any D-segments. Instead V- and J-segmentare joined directly, resulting in less diversity than the one found in heavychains. In all, the gene segment rearrangement, addition of extra nucleotidesand heavy/light chain pairing, enable a repertoire of 1016 possible uniqueantibodies [6].

A developing B cell first expresses the heavy-chain and displays it on thesurface together with a surrogate light-chain. After that the light-chain rear-rangement starts and the translated light-chain replaces the surrogate to forma complete membrane bound antibody. Together with two other membranebound proteins, α and β, which are responsible for intracellular signaling, thismembrane bound antibody makes up the B cell receptor (BCR) [5].

New B cells with BCRs of random specificity are continuously produced inthe bone marrow. B cells with the ability to bind a foreign molecule presentat that time will get the signals necessary for further development, but non-

4

Introduction

V HV HV 1H V nHL L L L D 1H D H D H D nH J H J H J H J H Cμ Cδ

J H Cμ CδD 1H D H DJV HV HV 1H V nHL L L L

V HV 1HL L L V D J J H Cμ Cδ

L V D J Cμ L V D J Cδ

L V D J J H Cμ Cδ

1

2

3

4

Germ-line DNA

Rearranged DNA

Primary RNA transcript

Heavy-chain mRNA AAA AAA

Figure 1.2: VDJ-rearrangement of antibody heavy-chain gene seg-ments. First one D- and one J-gene segment of the germ-line DNA arejoined together by excision of the separating sequence (1) and then a V-segment is joined to the DJ-intermediate (2). Transcription of the rear-ranged heavy-chain DNA then creates a primary RNA transcript containingan L-exon, the VDJ- and non-deleted J-segments as well as genes for twoconstant region genes, Cµ and Cδ (3). Differential RNA-splicing then re-moves introns and extra exons creating two different mRNA molecules con-taining either Cµ or Cδ (4), which are then translated and processed intomature IgM or IgD heavy-chains.

binding B cells undergo apoptosis [9]. By creating an enormous number ofrandom antibody specificities and selecting only those that bind, the immunesystem can mount highly targeted antibody responses.

5

Introduction

B cell activation, clonal expansion, affinity maturationand class switching

When the membrane bound antibody on a B cell binds an antigen, signalingof the BCR will cause the whole BCR-antigen complex to be to be inter-nalized [10]. Increasing acidic environment and presence of proteases withinthe endocytic compartments degrade the antigen to peptides. Some of thepeptides from the degraded proteins are bound by major histocompatibilitycomplex (MHC) class II-molecules and are presented on the surface of the Bcell where they are accessible for recognition by TH cells [11, 12].

TH cells carry a surface bound T cell receptor (TCR), which is formed ina process similar to the genetic recombination of BCRs and naïve T cellscarry TCRs with random specificities [13]. During maturation in the thy-mus, autoreactive T cells with TCRs that bind MHC molecules carryingself-peptides are negatively selected and receive signals to undergo apopto-sis. This negative selection is one of the mechanisms used by the immunesystem to minimize the presence of self-targeting antibodies or cytotoxic Tcells that otherwise could cause autoimmune disease [14].

TH cells that carry a TCR with affinity to an MHC class II with boundnon-self peptide can be activated by antigen presenting cells (APCs), e.g.dendritic cells, macrophages and B cells, displaying this MHC-peptide com-plex on the surface. Upon activation, the TH cell starts to proliferate andcan in turn activate B cells that present this peptide [15]. When a B cell isactivated by a TH cell it will start a clonal expansion within the germinalcenters of secondary lymphoid organs [9]. During this phase the B cells aredividing rapidly and a phenomenon called somatic hypermutation introducessmall alterations to the genes of the antigen binding parts of the antibody,which in turn will introduce a slight change of the amino acid sequences ofthe CDR-loops. The new clones with slightly varying BCRs then compete forbinding to the limited antigen present in the meshwork of follicular dendriticcells within the germinal center and only the ones with highest affinity willbe selected for further expansion and rounds of mutation. This process ofrepeated somatic hypermutation and selection is know as affinity maturation

6

Introduction

2

1

3

4

B cell

T cellH

Figure 1.3: T cell-dependent B cell activation. Membrane boundantibodies bind antigen (1) and the antibody-antigen complex is internalized.The acidic environment and proteases within the endocytic pathway digeststhe antigen to peptides, of which some are bound by MHC class II molecules(2). The MHC-peptide complexes are displayed on the surface of the B cellwhere they can be recognized by the T cell-receptor of an activated TH cell(3). This causes the TH cell to express activating ligands on its surface andto secrete cytokines that together stimulate B cell proliferation (4).

and it enables the immune system to improve the efficacy of the antibodyrepertoire during the course of an ongoing infection [5].

The membrane bound antibodies on the developing B cell are mostly of theIgD class, but also to some extent IgM, which is the first antibody class to besecreted during an infection. While the somatic hypermutation introduceschanges to the variable domains during affinity maturation a second processcalled class switching occurs in the genes of the constant region. Here theµ heavy chain genes are substituted for genes of the α, γ or ε isotype. Thisprocess preserves the antibody specificity, but alters the effector functions ofthe antibody [16].

7

Introduction

Dark zone - clonal expansion & somatic hypermutation

MemoryB cell

ApoptoticB cell

Follicular dentritic cell

Light zone - selection, class switching & di�erentiation

Plasmacell

ActivatedB cell

2

13

4

T cellH

Figure 1.4: Cellular events within a germinal center. Antigen-activated B cells migrate into germinal centers (1) where rapid proliferationand somatic hypermutation within the dark zone create B cells with slightlyvarying affinity (2). In the light zone, competition for binding to a lim-ited amount of antigen on follicular dendritic cells results in selection forhigh affinity B cells, while lower affinity clones die by apoptosis (3). Afteractivation by TH cells, high affinity B cells undergo class switching and dif-ferentiation into long-lived memory cells or antibody-producing plasma cells(4).

When B cells have completed affinity maturation and class switching, matureB cells differentiate to antibody-secreting plasma cells which lose their BCRsand are short lived in the circulation or differentiate into memory B cells thatreside in the bone marrow where they divide slowly, sustaining a presenceof a few circulating cells. If these memory B cells encounter their antigenagain, e.g. as a result of re-infection with the same pathogen, the alreadyaffinity matured and class switched BCR binds the antigen, causing rapidproliferation and differentiation into plasma cells and a fast high affinity andhigh titer antibody response. [17].

8

Introduction

Antibodies for research, diagnosis and therapy

The generally high specificity and affinity of antibodies to their target anti-gens make them very well suited for detection of proteins. This in turn hasmade them invaluable as in vitro research reagents as well as for diagnosticsand therapeutics. Antibodies and other affinity reagents have to be gener-ated either by immunization, selected from a library in vitro, or rationallydesigned and synthesized.

Antibodies also have to be validated to ensure that they do indeed selectivelybind to the intended target and that they do not show binding to otherproteins that could interfere with the analysis or therapeutic effect. The lackof well-validated antibodies has limited affinity-based protein research [18],but huge undertakings like the Human Protein Atlas-project [19] and theProteomeBinders consortium [20] have made affinity reagents to most humanproteins readily available.

Epitopes

B cell epitopes are structures on the antigen complimentary to the paratopeor antigen-binding surface of the antibody. Protein epitopes are can be di-vided in two classes, linear (continuous) epitopes where a stretch of con-secutive amino acid residues of the primary sequence are bound and con-formational (discontinuous) epitopes, which are formed when distant aminoacid residues are brought in close proximity upon protein folding [21]. T cellepitopes on the other hand are confined to linear peptides formed upon prote-olytic cleavage and presented to T cells by MCH class I or II molecules.

Immunization and generation of polyclonal antibodies

The classical way of generating affinity reagents is by immunization of ani-mals with the antigen, for which antibodies are desired. Some antigens, e.g.haptens or peptides, are not always able to elicit an antibody response on

9

Introduction

a b

Figure 1.5: Schematic drawing of antibodies binding (a) linear epitopesof consecutive amino acid residues and (b) conformational epitopes formedby amino acids brought together by protein folding.

their own but rather have to be coupled to a carrier protein to make animmunogenic complex, or immunogen. The immunogen is injected into theanimal together with an adjuvant, which helps to activate the immune sys-tem and create a stronger response [22]. This procedure is repeated severaltimes over a few months to boost the response, which allow the B cells toundergo affinity maturation and class switching as well as to ensure highantibody titers in the blood.

After immunization, the blood from the animal is collected and since theantibodies are present in the serum faction, blood cells and clotting factorsare removed. This crude antiserum now includes antibodies to the antigenand can be used for detection or neutralization of the antigen in question.However, the antiserum also contains antibodies naturally present in theanimal and a purification step using the antigen as bait can be used to obtainan antibody reagent with defined specificity [23]. Since many different B cellclones contribute to the secretion of antibodies into the circulation, the finalreagent is referred to as a polyclonal antibody.

10

Introduction

Immunogens

Antibodies intended for research are made to bind target proteins and themost straightforward way is to use the protein itself as the immunogen. Im-munizations done with full-length proteins of natively folded structure oftengenerate a polyclonal antibody response targeting predominantly conforma-tional epitopes, but to some extent also to linear epitopes [24]. However,this requires production of full-length proteins under native conditions insufficient amounts for immunization [25] and there is no way to direct theantibody response to a certain part of the protein. Since many proteins sharecommon domains and have high sequence similarity the generated antibodiescan be cross-reactive to other proteins than the intended target.

One alternative way is to use protein domains or other protein fragmentsas immunogens, which can be selected to cover parts that are unique to theprotein of interest as compared to all other proteins of the same species [23].Choosing an entire protein domain improves the chances of obtaining a nativefolding of the immunogen, while a smaller protein fragment of approximately100 amino acid residues is more likely to adapt a non-native fold giving lessantibodies to conformational epitopes on the native protein. This may onthe other hand increase the amount of antibodies targeting linear epitopes,which can be beneficial if the antibody is intended for use in assays wherethe proteins have been subjected to partial or complete denaturation e.g.Western blot [26]. Recombinant protein fragments can be expressed togetherwith a fusion tag to prolong the half-life during immunization and increasethe immunogenicity.

Short peptides of 15-40 amino acid residues can also be used for immuniza-tion, even if they do not contain any T cell epitopes. Coupling the peptides tocarrier proteins provide them with amino acid sequences suitable for presen-tation on MHC class II molecules that also are recognized as foreign by theTH cells [27]. The rationale of this immunization strategy is that antibodiesraised to peptides should cross-react with the same amino acid sequence onthe full-length protein. However, most antibodies raised to short peptidesfail to bind their intended protein targets unless the peptides are chosen to

11

Introduction

cover inherently unstructured parts like the generally more flexible N- and C-termini [28]. Peptides are also used to generate antibodies intended to bindthe very peptide used for immunization, e.g. for specific peptide enrichmentin trypsin digested protein samples [29].

a b c

Figure 1.6: Schematic figure of different immunogens. a) Full-length protein (green). b) Protein fragment (blue) covering a unique partof the amino acid sequence expressed with a fusion tag. c) Peptides (red)conjugated to a carrier protein.

Since polyclonal antibodies are purified from the serum of an immunized ani-mal, the resulting affinity reagent is a limited resource, which will eventuallybe exhausted. New immunizations using the same immunogen can be made,but differences in immune response will yield batch-to-batch variations thatinfluence the performance of the antibody [30–32]. This problem can be ad-dressed to some degree by pooling serum from multiple immunizations tocreate large batches that last longer. Another problem with polyclonal anti-bodies is that it is hard to in detail define their binding, making the approvalof them for therapeutic purposes problematic.

Monoclonal antibodies

During the 1970’s a technology to produce monoclonal antibodies, i.e. anti-bodies all originating from the same B cell clone, was invented [33]. In thistechnique B cells from the spleen are fused with myeloma cells, cancerousplasma cells, to form hybridoma cells. These contain the genes coding for

12

Introduction

the antibody of the B cell and the ability to divide indefinitely, which is char-acteristic for myeloma cells. Cell cultivation using a selective medium allowsfused cells carrying both B cell and myeloma genes to survive while unfusedmyeloma cells do not. The different hybridoma clones that are formed arecultivated in separate compartments and secrete their antibodies to the cul-ture medium [5]. The supernatant of each compartment can then be testedfor functionality in different assays and clones fulfilling the desired criteriaare chosen for larger cultivation and antibody purification. The genes of asuitable hybridoma cell can also be sequenced and cloned into other cells bet-ter suited for large-scale antibody production [30]. This makes monoclonalantibodies renewable affinity reagents, a huge advantage over the exhaustiblenature of polyclonal antibodies.

Since all the antibody molecules of in a monoclonal culture originate from thesame B cell they are identical and they target one single epitope on the pro-tein. This is in some cases a drawback, since an epitope might be intact andaccessible in one assay but not in another [34]. On the other hand a mono-clonal antibody can be formulated as a well-defined and reproducible product,which is reflected in their use as therapeutic agents. Currently there are 34monoclonal antibodies for human therapy approved by the American Foodand Drug Administration, FDA, with many more in clinical trials [35].

In vitro selection of antibody fragments

Although immunization can yield functional monoclonal antibodies of highaffinity, not all immunizations are successful in producing a reagent withthe desired properties. In combination with high costs and the long time ofimmunization schemes this has prompted researchers to look for alternativeways to create monoclonal antibodies [36]. In vitro selection techniques canbe used to isolate high affinity antibody fragments from libraries constructedby random combination of antibody variable gene segments. Single-chainvariable fragments (scFvs), which are fusion proteins of only the variabledomains of the heavy and light chains of antibodies, or Fab fragments arecommonly used in selection systems. Both retain the antigen binding capac-

13

Introduction

ity of antibodies, but lack the effector functions [37].

The small size of antibody fragments make them compatible with e.g. phagedisplay-based selection, which utilizes the ability of some bacteriophages totake up genetic material and display the corresponding protein products onthe surface of the phage particle. This creates a link between the proteinon the surface and its corresponding DNA inside the phage particle. Thisenables deduction of the amino acid sequence of the selected protein by DNAsequencing [38]. A great advantage of phage display and some other displaysystems is that huge libraries of more than 1010 unique members can beencreated [39], which increases the chance of finding a suitable binder to theprotein target of interest.

During the selection, phages in the library are allowed to interact with theantigen of interest and phages displaying binders with affinity for the antigenare retained while non-binders are washed away. If needed, the retainedbinders can then be used to infect bacteria and by using different mutationstrategies slight variations can be introduced to the selected binders. Afterstringent washing, antibody fragments with improved affinity for the antigencan be isolated, thus mimicking the affinity maturation of an in vivo immuneresponse [40].

Array technologies

Protein microarrays

Biological processes often involve interactions between proteins and othermolecules such as nucleic acids, peptides and other proteins. Many of theseinteractions are still unknown and finding new interaction partners to a pro-tein can aid in revealing its biological function or to discover new drugs.However, testing each interaction one by one becomes impractical with agrowing number of proteins to be screened. To shorten the time needed,molecules can be arranged in ordered arrays where multiple interactions canbe probed simultaneously [41].

14

Introduction

Protein microarrays are usually made by spotting small volumes of recom-binant protein solutions onto an activated microscope slide surface and byusing this technique, arrays with many thousands of proteins can be pro-duced [42]. Another way of making protein microarrays is to first synthesizea DNA microarray and then use an in vitro translation system to producethe proteins encoded by the DNA [43]. This strategy omits the need for sep-arate recombinant protein expression and purification and each array can betranslated just before use, where the stability of the DNA molecules enableslonger storage time compared to traditional protein arrays.

Protein microarrays can be used to determine the specificity and potential off-target binding of affinity reagents e.g. polyclonal and monoclonal antibodies[44], as well as for discovery of autoantibodies targets in plasma samples frompatients with autoimmune disorders or cancer [45,46].

Peptide microarrays

Peptide microarrays share many of the features of protein arrays, but in-stead of studying binding to full-length proteins or longer protein fragmentsthey generally contain peptides with lengths in the range of 6-25 amino acidresidues. Peptide microarrays with overlapping peptides together coveringthe sequence of a protein antigen can be used to map linear B cell epi-topes [47,48] or to find potential T cell epitope peptides that stimulate pro-liferation of either TH cells or CTLs [49].

Traditionally peptide microarrays have been produced using SPOT synthesiswhere droplets, each containing one of 20 amino acids, are deposited to spec-ified positions on a solid support followed by washing of un-reacted aminoacids and removal of protecting groups to allow for incorporation of the nextamino acid to the growing peptide. In this way, repeated coupling cyclesgenerate an array of short peptides with known sequence in defined posi-tions [47]. However, arrays produced directly using SPOT-synthesis havelow spot density, which limits the number of unique peptide sequences onthe arrays and has prompted development of improved techniques. Spottingof peptide arrays using pre-synthesized peptides can achieve higher peptide

15

Introduction

density, but this technique is much more expensive per peptide and bestsuited for production of many identical arrays [50].

Lately, different photolithographic techniques for in situ synthesis of high-density peptide arrays have enabled millions of unique peptides to be syn-thesized in parallel on a standard microscope slide surface. The vast amountof unique peptides has made it possible to produce arrays of short overlap-ping peptides covering all human proteins, enabling both detailed epitopemapping and off-target binding analysis [51,52].

As with protein arrays, peptide arrays can be used for identification of pro-tein targets of antibodies in patient plasma samples. These results howeverpinpoint the specificity to a short amino acid sequences, which enables theseparation of signals generated by antibodies with different epitope specifici-ties. This distinction would not be possible to make if a full-length proteinwas used. At the same time a major drawback of the peptide arrays is thatantibodies towards conformational epitopes formed upon protein folding arenot detected using short peptides [53]. Many known autoantibodies targetconformational epitopes [54, 55] and peptide and protein arrays should notbe seen as competing techniques, but rather as complements.

An alternative to creating peptide arrays based on the amino acid sequencesof proteins is making arrays with random sequences. These arrays with thou-sands of random peptides contain peptides with sequence motifs mimickingepitopes on both pathogens like viruses and bacteria as well as human pro-teins. If a simple sample is analyzed, e.g. a purified monoclonal antibody,alignment of the bound peptide sequences can be used to deduce the actualepitope. With more complex samples, like human plasma, these arrays canbe used to obtain individual binding signatures. By comparing samples formdifferent disease groups, peptides patterns able to distinguish between thedifferent groups can be identified, although the actual epitopes and targetproteins remain elusive. [56].

16

Introduction

Suspension bead arrays

One of the limitations of planar arrays is the fairly low sample throughputand suspension bead arrays are good complements when many samples haveto be analyzed. The bead array assays are based on a system of codedmicrospheres where each analyte is coupled to one unique bead-ID and thencombined into an array of analytes [57]. The bead array is then incubatedwith the sample and a flow cytometer determines the bead-ID, often based onthe ratio of two or more fluorescent dyes. The signal intensity of a reporterfluorophore then reveals the binding to the analytes in the sample.

Suspension bead arrays systems are highly compatible with an automatedworkflow in microtiter plates and the analysis time on the flow cytometer isapproximately a minute per sample. However, compared to the thousandsof analytes present on planar arrays, the number of available unique bead-IDs limits the size of the suspension bead arrays. Currently a system of 500different bead-IDs is used routinely [58], but with addition of more fluorescentdyes this number can be increased [59].

Reverse phase protein arrays

Another way to increase the sample throughput is to spot the samples on amicroscope slide and then look for expression of a certain protein in up toseveral thousand samples at once using an antibody. These arrays are knownas reverse phase protein arrays [60].

Antibody arrays

The array format is also attractive for multiplex protein analysis using anti-bodies. As an example, antibody arrays consisting of immobilized antibodiesspecific for the proteins of interest can be incubated with biotin labeled com-plex protein samples, e.g. tissue lysates or body fluids, and captured targetproteins are detected using a streptavidin-conjugated fluorophore [61, 62].

17

Introduction

To create an antibody array capable of reliable multiplex protein analysisrequires access to well validated and reliable antibodies to many protein tar-gets, which can be difficult to obtain in particular for less studied proteins.Large-scale projects like the Human Protein Atlas [19], which to date haveproduced validated polyclonal antibodies to almost 17,000 human genes, pro-vide a resource for antibody arrays. In another approach, scFvs selected fromphage display libraries are used as capturing reagents in multiplex antibodyarrays [63].

Characterization of antibody binding

Affinity determination

The binding strength, or affinity, between an antibody’s paratope and theantigen epitope depends on the complementarity of the two surfaces. Theattracting forces are mainly weak non-covalent van der Waals, hydrophobicand electrostatic interactions, but to some extent also hydrogen bonds [64].Affinities are often reported as the dissociation constant at equilibrium, KD,measured in molar units (M). KD is defined as the ratio between the concen-tration of free antigen and free antibody and the concentration of antibody-antigen complex.

KD =[Ab][Ag]

[AbAg](1.1)

There are many ways to determine the affinity and kinetics, i.e. the rates atwhich the antibody-antigen complex associate and dissociate. This can beperformed in solution, e.g. by determining the ratio of bound and free antigenat equilibrium for varying antigen concentrations while keeping the antibodyconcentration constant, but in some assays either the antibody or antigen hasto be immobilized [65,66]. This is the case when a surface plasmon resonance(SPR) biosensor is used to monitor the binding in real time and measures haveto be taken to ensure that the immobilization itself doesn’t alter the structure

18

Introduction

of the tethered molecule, thus influencing the affinity of the interaction [67].Still SPR is a powerful and widely used method for determination of affinity,association rate, (kon) and dissociation rate (koff ).

KD =[koff ]

[kon](1.2)

Antibodies carry multiple identical antigen binding regions, two in the caseof IgG, IgD and IgE, two or four for IgA and ten for IgM [6]. The combinedbinding strength of the immunoglobulin molecule for antigens with multipleepitopes is called the avidity or functional affinity. In some cases, this makesit difficult to determine the actual affinity of a single paratope-epitope in-teraction using the intact antibody. One way to circumvent this problem isby cleaving the antibody with papain to create monovalent Fab-fragments,which then can be used to determine the affinity, but mathematical modelscan also be used to account for the effects of bivalent binding [68].

Cross-reactivity (off-target binding)

Many proteins share stretches of their primary sequences that are identicalor differing by only a few amino acid residues and some proteins with similarfunctions have domains with surface patches of high similarity. Sometimes anantibody targeting an epitope in one of these regions shows cross-reactivityto other proteins than the intended target, making the results from an assaywith this antibody unreliable and hard to interpret [69]. Often the off-targetinteraction is much weaker than the one seen for the cognate antigen makingthese effects negligible when both proteins exist at equimolar concentrations,but the results will be erroneous if the off target protein is much more abun-dant in the sample. Sandwich immuno-assays [70] where two antibodiestargeting different epitopes of the same protein are preferably used in com-parison to a direct immunoassay since the chance of both antibodies beingcross-reactive to the same protein is very low.

19

Introduction

Epitope mapping

Most antibodies used in research are targeting protein antigens and anti-bodies to microbial proteins are common in a natural immune response.In addition to knowledge concerning which protein the antibody binds, itis also important determine the actual epitope and many epitope mappingtechniques have been developed for this purpose. Epitope mapping can havemany different meanings, varying from identification of which domain of anantigen the antibody binds to precise mapping of amino acid residues thatare crucial for binding to occur [71]. In addition to being a tool for vali-dation of antibodies, epitope mapping techniques have been used to studythe antibody response during infection, immunization and in autoimmunedisease [32, 72,73].

X-ray crystallography of antibody-antigen complexes

Solving the 3D-structure of an antibody’s Fab fragment in complex with theantigen provides an almost complete picture of the epitope-paratope interac-tion, making this method the gold standard of epitope mapping [74]. Boththe epitope and paratope surfaces are clearly defined and the contribution ofeach atom to the surface buried upon binding can be calculated. This in turnreveals which amino acid residues are contributing to the binding. However,producing Fab-antigen complexes require large amounts of both highly pureFab and antigen to form the crystals needed [75]. This makes X-ray crystal-lography a time consuming method for epitope mapping and has so far onlybeen used for a limited number of epitope mapping efforts [76,77].

20

Introduction

Peptide scanning

Solid-phase peptide synthesis enables short peptides of any amino acid se-quence to be synthesized efficiently and this has been used extensively to maplinear epitopes of antibodies. The most common method is to use overlappingpeptides for scanning of the primary sequence to identify the epitope, butalso truncations are often used to find the smallest consecutive amino acidstretch required for binding [78]. Employing a chemical synthesis procedurealso makes it possible to incorporate modified amino acids at any position,which allows analysis of e.g. citrullinated epitopes targeted by autoantibodiesin Rheumatoid arthritis [79], or amino acid substitutions which can be usedto investigate each amino acid’s contribution to antibody binding [48].

Antigen display systems

Display of proteins, domains and peptides on phage particles or cell surfaceshas also been used extensively for epitope mapping. In a method developedby Rockberg and Löfblom el al. [80] a library of short antigen fragmentsis displayed on the surface of the Gram-positive bacterium Staphylococcuscarnosus and incubated with the antibody. Clones with antibody-bindingfragments on the surface are then isolated using fluorescence-activated cellsorting (FACS) and the epitopes are determined by aligning the fragmentsequences. Analogously, yeast display can also be used to express proteinfragments, but yeast display is also well suited for random mutagenesis ofantigens using error prone PCR [81, 82]. In this way, detection of aminoacid substitutions in the antigen that have a negative effect on antibodybinding reveals which amino acids are located within the epitope. However,substitution of structurally important amino acids can alter the conformationof the antigen that in turn can lead to loss of binding although the aminoacid is not a part of the epitope surface.

Display systems, especially phage display, can also be used for creating hugepeptide libraries suitable for epitope mapping [83]. Random peptide librariescan be used to identify both linear epitopes and peptide sequences mimicking

21

Introduction

conformational epitopes [53] and Larman et al. [84] presented a phage librarydisplaying 36-mer peptides that together cover the human proteome. Thislibrary was in turn used to identify novel autoantibody targets present inspinal fluid of individuals with paraneoplastic neurological syndrome.

Epitope mapping using mass spectrometry

Tandem mass spectrometry can very accurately determine the mass andamino acid sequence of peptides in enzymatically digested protein samplesand has been used for read out in epitope mapping experiments. Epitopeextraction is a method for mapping linear epitopes and is based on enrich-ment of epitope containing peptides generated by enzymatic digestion of thetarget protein. The antibody binds the epitope containing peptides, whichafter elution are identified in a mass spectrometer [85, 86]. By using mul-tiple enzymes with different cleavage sites, e.g. trypsin and chymotrypsin,peptides of varying lengths and with overlapping sequences can be generatedwhich increases the chance of having peptides where the epitope has not beendestroyed during the enzymatic protein digestion [87]. In epitope excision,enzymatic digestion is carried out while the antigen is bound to the antibodyand thereby the epitope is shielded by the paratope. This decreases the rateof enzymatic cleavage of the epitope region and the epitopes are identifiedby detection of non-cleaved peptides [88].

A method similar to epitope excision is based on deuterium exchange wherethe antigen and antibody are diluted in heavy water, D2O, upon which theamide hydrogens in the peptide bonds of solvent exposed parts are replacedby deuterium. The labeled antigen and antibody are then allowed to form acomplex before changing the back to H2O buffer where back-exchange to hy-drogen occurs everywhere except where the antigen is shielded by the boundantibody. Subsequent digestion of the antigen and mass analysis of the pep-tides reveal which peptides and peptide fragments still contain deuteriumafter back-exchange, i.e. were part of the shielded epitope [87].

22

Introduction

Competition epitope mapping and epitope masking

Competition assays where an antibody’s ability to compete with a naturalligand or other antibodies of known epitope specificity can, by lack of binding,indicate at least partly overlapping epitopes [71]. Epitope masking is similarto competition mapping, but instead of using a competing molecule, differentsurfaces of the antigen are blocked, e.g. by selectively tethering one surfaceat a time to a solid support [89].

Epitope prediction algorithms

An epitope is not an intrinsic characteristic of an antigen and only exists inrelation to the paratope of an antibody, although some parts of a protein canbe more immunogenic than others [90]. Many in silico epitope predictionmethods have been developed which tries to identify these regions that aremost likely to be targeted in an antibody response.

The first attempts started with propensity scales of the amino acids’ phys-iochemical properties, e.g. hydrophilicity and hydrophobicity, [91, 92], aswell as structural properties such as surface exposure [93] and flexibility [94].Accurate prediction of immunogenic parts of an antigen would be of greathelp in designing new vaccines and for choosing immunogens for generationof affinity reagents, but B cell epitopes have proven difficult to predict [95].This could partly be explained by the high complexity of conformational epi-topes, but also that good prediction algorithms have to be trained using largesets of well-defined epitopes, which have thus far been lacking. In contrast tothe aforementioned methods based on qualitative occurrence of amino acidsand structural properties in epitopes, Rockberg and Uhlén devised a propen-sity scale based on the antibody titer of the polyclonal response of more than12,000 standardized immunizations with recombinant protein fragments. [96].Increasing availability of structural protein data stored in standardized andaccessible formats [97] has led to more sophisticated methods for predictionof conformational epitopes of proteins with solved 3D-structure [98,99].

23

Aims of the thesis

This thesis describes the development of peptide array assays for character-ization of antibody specificity and the use of these techniques to study thepolyclonal antibody response after immunization.

Paper I - The aim of this study was to establish a method for separationof antibodies with respect to the different epitope specificities present inpolyclonal antisera and determine their respective functionalities in commonantibody-based assays.

Paper II - The aim of this study was to analyze the epitope variation ofpolyclonal antibodies from parallel rabbit immunizations using the same re-combinant protein fragment as immunogen.

Paper III - The aim of this study was to determine the ratio between anti-bodies targeting linear and conformational epitopes of polyclonal antibodiesobtained after immunizations with recombinant protein fragments.

Paper IV - The aim of this study was to develop a method for epitopemapping using high-density planar peptide arrays.

Paper V - The aim of this study was to develop a method for characteriza-tion of antibody binding using high-density planar peptide arrays coveringthe entire human proteome.

Paper VI - The aim of this study was to investigate if polyclonal antibodiesraised against recombinant protein fragments can be used for specific peptideenrichment and in combination with heavy isotope labeled standards allow forrapid absolute protein quantification using tandem mass spectrometry.

24

Present investigations

Generation of monospecific antibodies based on affinitycapture of polyclonafl antibodies. (Paper I)

The potential cross-reactivity of affinity reagents makes results based on asingle binder less sensitive and reliable than results from e.g. a sandwichassay where two antibodies targeting different epitopes have to bind simulta-neously to produce a signal [100]. However, finding suitable antibody pairscan be difficult for many proteins and an alternative can be to separatepolyclonal antibodies into fractions based on their epitope specificities andthereby make paired antibodies from a single starting reagent. In this study,we developed such a method to generate epitope-specific, or monospecific,antibody fractions and show how the individual fractions and different an-tibody pairs perform in commonly used immunoassays. First, the linearepitopes of the polyclonal antibody were mapped using a suspension beadarray with synthetic overlapping 15-mer peptides covering the amino acidsequence of the antigen. The peptides corresponding to identified epitopeswere then immobilized on separate liquid chromatography columns and con-nected serially and followed by an antigen column to bind antibodies thatwere not captured by the peptide columns. Polyclonal antiserum was thenpassed over the connected columns before antibodies in the columns wereeluted in parallel to produce epitope-specific fractions.

25


Epitope 1

Epitope 2

Epitope 3

Epitope 4

Antigen

1

2

3

Polyclonal serum

Figure 3.1: Overview of generation of epitope-specific antibodyfractions. Polyclonal serum is epitope mapped using overlapping peptides(1). Peptides corresponding to identified epitopes and the antigen are thenused as ligands in a serial antibody capture (2). Parallel elution into sepa-rate epitope-specific fractions followed by epitope confirmation (3)

In total, epitope-specific fractions were isolated from polyclonal antisera tar-geting four potential cancer biomarkers, RBM3, SATB2, ANLN and CNDP1.All epitope-specific fractions, the antigen fractions and the polyclonal anti-bodies were tested for functionality in Western blot and immunohistochem-istry assays and for three, the subcellular localization was determined usingimmunofluorescence microscopy. For each, RBM3 and SATB2, two fractionswith non-overlapping epitope specificities were identified as functional in the

26


immunohistochemistry assay and these were used as paired antibodies toverify the expression of endogenous protein [19] in both normal and cancertissues.

A subcellular localization study of the polyclonal antibody targeting ANLNshowed the predicted nuclear staining, but also a cytoplasmic staining notsupported by literature, indicating potential binding to other proteins. Forthe four epitope-specific fractions, two were negative in the immunofluo-rescence microscopy assay, one showed only nuclear and one only vesicularstaining. To determine which of these staining patterns actually correspondto ANLN-binding, siRNA knock-down of ANLN expression in U-2 OS cellswas performed and this showed significant decrease of nuclear staining inten-sity, but unchanged intensity for the vesicular staining. The cross-reactivitycould thus be attributed to one epitope-specific fraction and the increasedspecificity of one other fraction compared to the polyclonal antibody.

The epitope-specific fractions targeting CNDP1 were screened to find a suit-able pair for a sandwich assay. Recombinant CNDP1 was spiked into humanplasma and different combinations of capture and detection antibodies weretested in a multiplex bead array assay. One combination of two epitope-specific fractions showed low limit of detection and low coefficient of varia-tion, which proved that good sandwich assay pairs can be produced from onepolyclonal serum.

In conclusion, we established a method to generate separate epitope-specificfractions from polyclonal antisera, which could be used as paired antibod-ies. The epitope mapping revealed 4-5 linear epitopes on each of the 100-130amino acids long protein fragments used as immunogens. The epitope-specificfractions from one polyclonal serum showed varying performance in commonimmunoassays like Western blot, immunohistochemistry and immunofluores-cence microscopy. However, some epitope-specific fractions were shown torecognize the target protein in all tested assays.

27


Parallel immunizations of rabbits using the same antigenyield antibodies with similar, but not identical, epitopes.(Paper II)

Polyclonal antibodies purified from sera of immunized animals are the mostcommon group of affinity reagents available to the research community, butin contrast to monoclonal antibodies they are a non-renewable resource. Newimmunizations using the same antigen can be made, but the properties ofresulting antibodies vary between the different immunizations [30, 31]. Inthis study we used two different epitope mapping techniques to compare theepitope profiles of the polyclonal antibody responses from rabbits immunizedwith the same antigen. Then Western blot or immunohistochemistry wereused to compare the functionality of the antibodies in common immuno-assays. In total, 30 polyclonal antibodies targeting ten different antigenswere epitope mapped using bacterial display of antigen fragments. Antibod-ies towards six of the antigens were also mapped using overlapping syntheticpeptides and the three antibodies towards TYMP where fractionated accord-ing to epitope specificity, as described in Paper I.

Western blot validation highlighted the batch-to-batch variation of polyclonalantibodies, since cross-reactivity to non-target proteins could be seen forsome, but not all three antibodies from immunizations with the same anti-gen. Both epitope mapping techniques showed that many epitope regionswere similar between the antibodies from all three immunizations, yet all tenantigens included at least one epitope not observable for all three antibod-ies. The epitope-specific fractions of the three polyclonal antibodies towardsTYMP also showed that the relative abundance of antibodies towards thedifferent epitopes varies greatly between the three immunizations and thatonly some of these fractions are functional in Western blot.

In conclusion, this work shows that re-immunization with the same antigencan create a similar, but not identical, epitope pattern. It also emphasizes theneed for thorough validation of each new polyclonal antibody batch.

28


Dissecting antibodies with regards to linear and confor-mational epitopes. (Paper III)

Immunization strategies using full-length proteins, protein fragments andshort peptides are extensively used to generate both polyclonal and mono-clonal antibodies. Immunization with native full-length protein have been re-ported to generate antibodies mainly targeting conformational epitopes [24],while short peptides, due to the their lack of tertiary structure, in generalelicit antibodies targeting linear epitopes. In this study we used a slightlymodified variant of the epitope fractionation technique used in Paper I todetermine the ratio of linear and conformational epitopes obtained from im-munizations using eight recombinant protein fragments with lengths rangingfrom 95 to 144 amino acids. Antibodies captured on the peptide columnswere considered to target linear epitopes, while antibodies not binding toany of the peptides, but captured by the subsequent antigen column, weredefined as targeting conformational epitopes. All epitope fractions and theoriginal polyclonal antibodies were then tested for ability to recognize theirtarget proteins in Western blot assays. For all eight proteins, at least threefractions targeting linear epitopes and one fraction targeting conformationalepitopes were obtained. For six out of the eight fractionated sera, less than30 % of the antibodies were targeting conformational epitopes. For only oneantigen, SYNJ2BP, the majority (83 %) of the antibodies targeted conforma-tional epitopes. For each antigen, at least one of the fractions towards linearepitopes was able to recognize a band of correct molecular weight in theWestern blot analysis while only two of the fractions towards conformationalepitopes were functional in this assay.

In conclusion, these results show that immunizations using recombinant pro-tein fragments yield mostly antibodies targeting linear epitopes and thatthese perform well in assays where the proteins have been denatured.

29


High-resolution mapping of linear antibody epitopes us-ing ultrahigh-density peptide microarrays. (Paper IV)

Advances in photolithographic in situ peptide synthesis have made it possibleto synthesize peptide arrays with hundreds of thousands of peptides. In thispaper, digital micromirrors were used for light-directed peptide synthesis,which enable unique amino acid sequences to be synthesized in parallel andarrays with overlapping peptides and single amino acid substitutions wereused for detailed epitope mapping of 22 polyclonal antibodies. First the an-tibodies were epitope mapped by scanning through the amino acid sequenceof the corresponding antigens with peptides of lengths varying between 5 and20 amino acids. At least one linear epitope was detected for 20 of the 22 an-tibodies and 15 antibodies targeted multiple epitopes. Some of the epitopeswere short and could be detected using 6-mer peptides, but many epitopesrequired at least 8-mer or longer peptides, however too long peptides showedreduced resolution. In addition to peptide scanning, new arrays containingsingle substitutions of the amino acid sequence of 79 epitopes were synthe-sized and used for very detailed epitope mapping that showed which aminoacids were crucial for antibody binding.

In conclusion, this article shows how high-density planar peptide arrays canbe used for efficient mapping of antibodies targeting many different antigens.It also showed that overlapping peptides with lengths between 12 and 15amino acids were able to detect most linear epitopes with high resolution andthat the detected linear epitopes in general consisted of 7-9 amino acids.

Proteome-wide Epitope Mapping of Antibodies UsingUltra-dense Peptide Arrays. (Paper V)

In this study we showed how planar in situ synthesized peptide arrays cover-ing the entire human proteome can be used for epitope mapping and off-targetbinding analysis for both monoclonal and polyclonal antibodies. For two ofthese antibodies, cross-reactivity to peptides corresponding to proteins otherthan the intended targets were tested for reactivity to longer recombinant

30


SSQRQVQNGPSPDEMDIQRRQVMEQHQQQRQESLERRTSATGPILPPGHPSSAASAPVSCSGPPPPPPPPVPPPPTGATPPP

32 ESLERRTSATGP33 SLERRTSATGPI34 LERRTSATGPIL35 ERRTSATGPILP36 RRTSATGPILPP37 RTSATGPILPPG38 TSATGPILPPGH39 SATGPILPPGHP

5 10 15 20 25 30 35 40 45 50 55 60 65 70

Fluo

resc

ence

inte

nsity

a

b

Figure 3.2: Example of epitope mapping of a polyclonal antibodyusing overlapping peptides. (a) Barplot showing the binding to eachpeptide reveals three linear epitopes. (b) Sequence alignment of reactiveoverlapping peptides where the consensus sequence show the minimal epitoperequired for antibody binding.

protein fragments on a protein array and in Western blot using lysates fromcells overexpressing some of these proteins. Reactivity to non-target pep-tides could be observed frequently for both antibodies and most of thesepeptides shared sequence elements with high similarity to the bound targetpeptides. However, little reactivity could be seen when these sequences werepart of longer protein fragments and only the monoclonal antibody towardsRBM3 bound to one of the overexpressed non-target proteins in Westernblot. These findings suggest that the high flexibility of short peptides, beingonly C-terminally tethered to the microarray surface, make them adapt tothe paratopes of the antibodies, but when these amino acids are present ina protein chain this flexibility is lost along with the ability to be bound bythe antibody.

31


In conclusion, this work showed that high-density peptide arrays covering thehuman proteome could be used to identify both on- and off-target epitopepeptides. However, in most cases we were not able to see any cross-reactivityto proteins containing the amino acids sequences corresponding to the boundpeptides. This suggests that antibody binding of linear epitopes is dependenton the context these amino acid sequences are presented in.

Immunoproteomics using polyclonal antibodies and sta-ble isotope-labeled affinity-purified recombinant proteins.(Paper VI)

Shotgun proteomics using tandem mass spectrometry can very accurately de-tect and sequence peptides from enzymatically digested protein samples anddetermine which proteins they originate from [101]. By using stable isotope-labeled peptide or protein standards, absolute quantification can also beachieved for the corresponding proteins. However, this approach has a biastowards detecting more abundant proteins and targeted approaches are of-ten needed to detect low abundant proteins in complex samples, e.g. plasma.One such method is immunoproteomics where the strength of antibodies tobind a target molecule is combined with the exact read-out of mass spec-trometry based proteomics.

Previously, anti-peptide antibodies have been used to enrich peptides fromthe proteins of interest to dramatically reduce the complexity of the sampleinjected to the mass spectrometer [29], but this requires new anti-peptide an-tibodies to be produced for each studied peptide. To circumvent this limita-tion, other researchers have developed antibodies recognizing motifs commonfor many peptides, which allow enrichment of groups of peptides [102,103]. Inthis study we investigate the ability of polyclonal antibodies raised againstrecombinant protein fragments to capture peptides from trypsin digestedsamples and if these in combination with heavy isotope-labeled variants ofthe same recombinant protein fragments can be used for efficient absoluteprotein quantification.

32


1 2 3 4

Other proteins

Target protein

Heavy isotope-labeled standard

Figure 3.3: Schematic workflow of immunoproteomics with heavy-isotope labeled protein standards. Proteins from normal cell cultures(red and green) are mixed with known amount of heavy isotope-labeled anti-gen (blue) (1). Trypsin digestion specifically cleaves the proteins after argi-nine and lysine, which creates a mix of both light and heavy peptides (2).Antibodies on magnetic beads subsequently enrich target peptides, whichdrastically reduces the sample complexity (3). Mass spectrometric analy-sis detects the small mass shift between heavy-labeled and non-labeled pep-tides (4) and the original protein concentration is calculated from the ratiobetween the amount of heavy and light peptides.

Epitope mapping using high-density peptide arrays, as described in PaperIV and V, of more than 900 polyclonal antibodies produced within theHuman Protein Atlas-project showed that these on average recognize ap-proximately three linear epitopes. However, around 40 % of these epitopescontain a trypsin cleavage site, which makes these non-functional for pep-tide enrichment. Most antibodies however should still be able to enrich atleast one peptide. A semi-automated protocol for peptide enrichment onantibodies coupled to protein A-coated magnetic beads was established andcombined with a rapid 15-minute detection procedure on an LC-MS/MS in-strument. This experimental setup was then used for absolute quantificationof proteins in human HeLa cells and the results were similar to a 24-hourquantification without peptide enrichment.

In conclusion, peptide enrichment using antibodies coupled to mass spec-trometric readout is a well-established method, but it has so far generallyrequired generation of special anti-peptide antibodies to set up a new assay.

33


In this work however, we show that an already available antibody resourcetogether with heavy isotope-labeled antigens can be used for rapid multiplexprotein quantification, which greatly reduces the time and cost to set up newpeptide-enrichment assays.

34

Concluding remarks and future per-spectives

The immune system produces antibodies to neutralize foreign pathogens andto achieve this, without harming the body’s own cells, this binding has to bevery strong and selective. These binding properties make antibodies excellenttools for the study of proteins, since they can selectively detect even lowabundant proteins. The work presented in this thesis has been focused ondevelopment of techniques for characterization of antibody specificity andthe use of these methods for epitope analysis of the polyclonal response afterimmunization.

In Paper I II, III and to some extent Paper VI, answers were sought fordifferent questions regarding the typical polyclonal antibody response afterimmunizations with recombinant proteins fragments as immunogens. In Pa-per I epitope-specific fractionation of polyclonal antisera was used to createantibody pairs, but characterization of the different fractions also showedthat only a couple of these were actually functional in assays like Westernblot, immunohistochemistry and immunofluorescence microscopy.

In Paper II, using two different epitope mapping techniques and the frac-tionation strategy devised in Paper I, we showed that multiple immuniza-tions using the same antigen resulted in antibody responses with similarepitope patterns, but that there were still significant differences, which hadan impact on the performance of the antibodies in Western blot. These re-sults are consistent with the findings of Geysen et al. [104] based on epitope

35

Concluding remarks and future perspectives

mapping of polyclonal antisera from seven rabbits immunized with the samefull-length antigen. Their results showed that no common epitopes were rec-ognized by all seven sera, but there were regions that were more likely toevoke an antibody response.

Using a slightly modified fractionation strategy, Paper III showed howchoosing recombinant protein fragments as immunogens predominantly yieldantibodies towards linear epitopes. This might in fact be more desirable thanantibodies targeting conformational epitopes if they are intended for use inassays where the proteins are wholly or partially denatured e.g. Western blotand immunohistochemistry.

The poor performance of antibodies to conformational epitopes could beexplained by the denaturation of the target proteins during gel electrophoresisin Western blot. An alternate explanation is that antibodies targeting theconformational epitopes on the protein fragment probably have a very lowchance of being reactive also to the native protein, which most likely adaptsa different fold.

Defining an epitope as either linear or conformational is not as straightfor-ward as it might seem and some researchers have questioned the concept oflinear epitopes entirely. In an article from 1986, Barley et al. [105] statethat a vast majority, if not all, antibodies are targeting conformational epi-topes and that peptides bound by antibodies only mimic parts of the antigensurface. In 1990 Laver [106] goes as far as to describe linear epitopes as amisconception and argues that the word epitope should be used exclusivelyfor conformational epitopes on native protein. In this work we have decidedto use the word linear epitope if the antibodies bind short peptides, but wecannot rule out that other amino acids that those present in the peptide arealso contributing to the binding.

In Paper VI we showed how polyclonal antibodies could be used for enrich-ment of peptides form trypsin digested protein samples as a way of reducingthe complexity before mass spectrometric analysis. We also showed that thisenrichment strategy in combination with heavy isotope-labeled protein stan-dards could be used for fast absolute protein quantification. This approach

36


is very similar to the well-established capture of peptides by anti-peptide an-tibodies with addition of heavy-isotope labeled peptides developed by LeighAnderson and his group [29]. The major differences being that by using analready existing antibody resource we can bypass the time-consuming step ofgenerating new antibodies to set up each new assay. Another difference is theuse of heavy-isotope labeled protein fragments instead of labeled peptides,which was shown by Zeiler et al. [107] to compensate for possible miscle-avages during peptide digestion. However, heavy-labeled protein fragmentscould of course be combined with any type of peptide enrichment system, e.g.antibodies targeting common peptide motifs described by Poetz et al. [102],or Selected Reaction Monitoring where only peptides of interest are selectedand analyzed in the mass spectrometer [108].

In the described protocol, we have used a 15-minute gradient for peptideseparation in the high performance liquid chromatography step coupled tothe mass spectrometer. The very low complexity of enriched peptide sam-ples should make it possible to optimize the LC-MS/MS protocol further,which could decrease the analysis time on the mass spectrometer even more.This is of particular importance for use in clinical diagnostics, where largenumbers of samples have to be analyzed each day. Ideally, a multiplex assaywith absolute quantification of most of the clinically relevant protein markersshould require no more than a couple of minutes for analysis on the massspectrometer.

In Paper IV, V and VI we used high-density peptide arrays covering eitherantigen sequences or the entire human proteome to study both on- and off-target binding of polyclonal and monoclonal antibodies. Epitope mappingof over 900 polyclonal antibodies showed that each antibody on average rec-ognizes three linear epitopes on antigens of approximately 100 amino acidsand the amino acid substitution scanning revealed that most linear epitopeswere 7-9 amino acid residues long. Already in 1975 M. Z. Atassi [72] showedthat polyclonal antibodies to sperm-whale myoglobin were binding epitopesof 6-7 amino acid residues. These results were based on minimal consensussequences of overlapping peptides, which also in our mapping efforts haveshown slightly shorter epitopes than the ones identified using amino acid

37


substitutions. An illustrative example in Paper V is the PODXL mini-mal epitope YPKTP, where a full amino acid substitution scan showed thata third proline also contributes to the binding, thus extending the epitopeto YPKTPSP. One should therefore be careful when stating linear epitopelengths and clarify how the epitope was defined.

Apart from studying the cross-reactivity of antibodies, the great screeningpotential of the proteome-covering peptide arrays could also be used to dis-cover the targets of autoantibodies present in plasma or serum from patientssuffering from autoimmune diseases. This strategy has been used in ongoingprojects where interesting peptides have been found to show increased anti-body binding in samples from patients with narcolepsy and multiple sclerosis.A landmark paper by Larman et al. [84] describes the construction of a phagedisplay library covering the entire proteome with 36-mer peptides and howthis was used to identify a novel autoantibody target. The phage display ap-proach has the advantage that, once the library has been created, the cost peranalysis is relatively low and the longer peptides could also be an advantagecompared to planar peptide arrays. However, planar peptide array analysisis very fast without any sequencing steps and information from non-boundpeptides also contribute to the binding signature of a serum sample.

Autoantibodies may also arise to proteins leaked to the circulation in varioustypes of cancers. Here peptide arrays could be used to detect the autoan-tibodies in plasma at an early stage when the protein concentration itselfis too low to be detectable. Stafford et al. [56] demonstrated the feasibilityof this approach by using binding signatures of planar arrays with randompeptide sequences to accurately stratify patients with five different types ofcancer.

The vast number of peptides available on a high-density peptide array alsomakes it possible to construct arrays that can be used for epitope map-ping of antibody responses towards the entire proteomes of many commonpathogens. This would give researchers valuable information about com-monly targeted epitopes, which in turn could be of great importance fordesign of new drugs and vaccines.

38

Acknowledgements

First I would like to thank all the people who in any way have helped mewith this thesis. Since I know I would fail to mention everybody, I won’teven try.

However, I would like to give special thanks to:

Mathias, for always being a great source of inspiration and ideas and ofcourse for giving me the opportunity to work in your lab. I’ve had fantastictime during these four years.

Henrik, for all your help and support, especially during the first two years.

Per-Åke and all you other PI:s who have taken your time to help me duringthese 4 years.

Fellow group members, past and present, Basia, Fredrik, Johan and Masato.I’ve had so much fun working with you.

Peter, Jochen and all other PAPP-skallar for letting me be a part ofyour group too. In particular the autoimmunity bunch for a great timein Nice.

Alpha 6 for fantastic lunch and fika company. Let the fire extinguisherring!

39

Acknowledgements

The HPA-crew for producing all the antigens and antibodies that I haveused in my projects.

My old group Immunotech, for helping me with my ridiculously large an-tibody orders and for letting me use your equipment.

All the fantastic people at Plan 3 and Alpha 2 for providing a nice andcreative work environment. Especially the Beer foundation for obviousreasons.

Special thanks also to:

Arash for all your help with the planar peptide arrays and R.

Tove for the introduction to mass spectrometry world.

Paul for always having relevant comments about projects and manuscripts.

Cajsa and Hammou for convincing me to pursue a doctoral degree. I oweyou a lot.

Last but not least, thank you Paul, Tarek and Fredrik for reading mythesis and giving me invaluable feedback.

40

References

[1] M. Uhlen. Affinity as a tool in life science. Biotechniques, 44(5):649–54, 2008.

[2] J. M. Schwenk, U. Igel, B. S. Kato, G. Nicholson, F. Karpe, M. Uhlen, and P. Nilsson.Comparative protein profiling of serum and plasma using an antibody suspensionbead array approach. Proteomics, 10(3):532–40, 2010.

[3] G. M. Edelman and M. D. Poulik. Studies on structural units of the gamma-globulins. J Exp Med, 113:861–84, 1961.

[4] R. R. Porter. The hydrolysis of rabbit y-globulin and antibodies with crystallinepapain. Biochem J, 73:119–26, 1959.

[5] Thomas J. Kindt, Richard A. Goldsby, Barbara Anne Osborne, and Janis Kuby.Kuby immunology. W.H. Freeman, New York, 6th edition, 2007.

[6] Jr. Schroeder, H. W. and L. Cavacini. Structure and function of immunoglobulins.J Allergy Clin Immunol, 125(2 Suppl 2):S41–52, 2010.

[7] G. W. Litman, J. P. Rast, and S. D. Fugmann. The origins of vertebrate adaptiveimmunity. Nat Rev Immunol, 10(8):543–53, 2010.

[8] L. A. O’Neill, D. Golenbock, and A. G. Bowie. The history of toll-like receptors -redefining innate immunity. Nat Rev Immunol, 13(6):453–60, 2013.

[9] C. C. Goodnow, C. G. Vinuesa, K. L. Randall, F. Mackay, and R. Brink. Controlsystems and decision making for antibody production. Nat Immunol, 11(8):681–8,2010.

[10] S. K. Pierce. Lipid rafts and b-cell activation. Nat Rev Immunol, 2(2):96–105, 2002.

[11] S. Cassard, J. Salamero, D. Hanau, D. Spehner, J. Davoust, W. H. Fridman, andC. Bonnerot. A tyrosine-based signal present in ig alpha mediates b cell receptorconstitutive internalization. J Immunol, 160(4):1767–73, 1998.

[12] D. Rodriguez-Pinto. B cells as antigen presenting cells. Cell Immunol, 238(2):67–75,2005.

41

References

[13] R. M. Cobb, K. J. Oestreich, O. A. Osipovich, and E. M. Oltz. Accessibility controlof v(d)j recombination. Adv Immunol, 91:45–109, 2006.

[14] K. A. Hogquist, T. A. Baldwin, and S. C. Jameson. Central tolerance: learningself-control in the thymus. Nat Rev Immunol, 5(10):772–82, 2005.

[15] M. L. Alegre, K. A. Frauwirth, and C. B. Thompson. T-cell regulation by cd28 andctla-4. Nat Rev Immunol, 1(3):220–8, 2001.

[16] Z. Xu, H. Zan, E. J. Pone, T. Mai, and P. Casali. Immunoglobulin class-switch dnarecombination: induction, targeting and beyond. Nat Rev Immunol, 12(7):517–31,2012.

[17] K. Igarashi, K. Ochiai, A. Itoh-Nakadai, and A. Muto. Orchestration of plasma celldifferentiation by bach2 and its gene regulatory network. Immunol Rev, 261(1):116–25, 2014.

[18] A. M. Edwards, R. Isserlin, G. D. Bader, S. V. Frye, T. M. Willson, and F. H. Yu.Too many roads not taken. Nature, 470(7333):163–5, 2011.

[19] M. Uhlen, P. Oksvold, L. Fagerberg, E. Lundberg, K. Jonasson, M. Forsberg,M. Zwahlen, C. Kampf, K. Wester, S. Hober, H. Wernerus, L. Bjorling, and F. Pon-ten. Towards a knowledge-based human protein atlas. Nat. Biotechnol., 28(12):1248–50, 2010.

[20] M. J. Taussig, O. Stoevesandt, C. A. Borrebaeck, A. R. Bradbury, D. Cahill, C. Cam-billau, A. de Daruvar, S. Dubel, J. Eichler, R. Frank, T. J. Gibson, D. Gloriam,L. Gold, F. W. Herberg, H. Hermjakob, J. D. Hoheisel, T. O. Joos, O. Kallion-iemi, M. Koegl, Z. Konthur, B. Korn, E. Kremmer, S. Krobitsch, U. Landegren,S. van der Maarel, J. McCafferty, S. Muyldermans, P. A. Nygren, S. Palcy, A. Pluck-thun, B. Polic, M. Przybylski, P. Saviranta, A. Sawyer, D. J. Sherman, A. Skerra,M. Templin, M. Ueffing, and M. Uhlen. Proteomebinders: planning a europeanresource of affinity reagents for analysis of the human proteome. Nat Methods,4(1):13–7, 2007.

[21] Israel Schechter Michael Sela, Bilha Schechter and Felix Borek. Antibodies to sequen-tial and conformational determinants. Cold Spring Harbor Symposia on QuantitativeBiology, 32:537–545, 1967.

[22] W. C. Hanly, J. E. Artwohl, and B. T. Bennett. Review of polyclonal antibodyproduction procedures in mammals and poultry. ILAR J, 37(3):93–118, 1995.

[23] P. Nilsson, L. Paavilainen, K. Larsson, J. Odling, M. Sundberg, A. C. Andersson,C. Kampf, A. Persson, C. Al-Khalili Szigyarto, J. Ottosson, E. Bjorling, S. Hober,H. Wernerus, K. Wester, F. Ponten, and M. Uhlen. Towards a human proteomeatlas: high-throughput generation of mono-specific antibodies for tissue profiling.Proteomics, 5(17):4327–37, 2005.

42

References

[24] 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, and et al. The antigenicstructure of proteins: a reappraisal. Annu Rev Immunol, 2:67–101, 1984.

[25] P. Braun, Y. Hu, B. Shen, A. Halleck, M. Koundinya, E. Harlow, and J. LaBaer.Proteome-scale purification of human proteins from bacteria. Proc Natl Acad Sci US A, 99(5):2654–9, 2002.

[26] J. Bordeaux, A. Welsh, S. Agarwal, E. Killiam, M. Baquero, J. Hanna, V. Anagnos-tou, and D. Rimm. Antibody validation. Biotechniques, 48(3):197–209, 2010.

[27] H. L. Niman, R. A. Houghten, L. E. Walker, R. A. Reisfeld, I. A. Wilson, J. M. Hogle,and R. A. Lerner. Generation of protein-reactive antibodies by short peptides is anevent of high frequency: implications for the structural basis of immune recognition.Proc Natl Acad Sci U S A, 80(16):4949–53, 1983.

[28] N. H. Trier, P. R. Hansen, and G. Houen. Production and characterization of peptideantibodies. Methods, 56(2):136–44, 2012.

[29] N. L. Anderson, N. G. Anderson, L. R. Haines, D. B. Hardie, R. W. Olafson, andT. W. Pearson. Mass spectrometric quantitation of peptides and proteins using sta-ble isotope standards and capture by anti-peptide antibodies (siscapa). J ProteomeRes, 3(2):235–44, 2004.

[30] P. N. Nelson, G. M. Reynolds, E. E. Waldron, E. Ward, K. Giannopoulos, and P. G.Murray. Monoclonal antibodies. Mol Pathol, 53(3):111–7, 2000.

[31] K. Larsson, C. Eriksson, J. M. Schwenk, L. Berglund, K. Wester, M. Uhlen,S. Hober, and H. Wernerus. Characterization of prest-based antibodies towardshuman cytokeratin-17. J Immunol Methods, 342(1-2):20–32, 2009.

[32] B. Hjelm, B. Forsstrom, J. Lofblom, J. Rockberg, and M. Uhlen. Parallel immu-nizations of rabbits using the same antigen yield antibodies with similar, but notidentical, epitopes. PLoS One, 7(12):e45817, 2012.

[33] G. Kohler and C. Milstein. Continuous cultures of fused cells secreting antibody ofpredefined specificity. Nature, 256(5517):495–7, 1975.

[34] N. S. Lipman, L. R. Jackson, L. J. Trudel, and F. Weis-Garcia. Monoclonal versuspolyclonal antibodies: distinguishing characteristics, applications, and informationresources. ILAR J, 46(3):258–68, 2005.

[35] M. S. Kinch. An overview of fda-approved biologics medicines. Drug Discov Today,2014.

[36] G. Winter and C. Milstein. Man-made antibodies. Nature, 349(6307):293–9, 1991.

[37] H. R. Hoogenboom. Selecting and screening recombinant antibody libraries. NatBiotechnol, 23(9):1105–16, 2005.

43

References

[38] C. Rader and 3rd Barbas, C. F. Phage display of combinatorial antibody libraries.Curr Opin Biotechnol, 8(4):503–8, 1997.

[39] M. Weber, E. Bujak, A. Putelli, A. Villa, M. Matasci, L. Gualandi, T. Hemmerle,S. Wulhfard, and D. Neri. A highly functional synthetic phage display library con-taining over 40 billion human antibody clones. PLoS One, 9(6):e100000, 2014.

[40] N. M. Low, P. H. Holliger, and G. Winter. Mimicking somatic hypermutation: affin-ity maturation of antibodies displayed on bacteriophage using a bacterial mutatorstrain. J Mol Biol, 260(3):359–68, 1996.

[41] F. X. Sutandy, J. Qian, C. S. Chen, and H. Zhu. Overview of protein microarrays.Curr Protoc Protein Sci, Chapter 27:Unit 27 1, 2013.

[42] H. Zhu, M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, N. Lan,R. Jansen, S. Bidlingmaier, T. Houfek, T. Mitchell, P. Miller, R. A. Dean, M. Ger-stein, and M. Snyder. Global analysis of protein activities using proteome chips.Science, 293(5537):2101–5, 2001.

[43] N. Ramachandran, E. Hainsworth, B. Bhullar, S. Eisenstein, B. Rosen, A. Y.Lau, J. C. Walter, and J. LaBaer. Self-assembling protein microarrays. Science,305(5680):86–90, 2004.

[44] P. F. Predki, D. Mattoon, R. Bangham, B. Schweitzer, and G. Michaud. Proteinmicroarrays: a new tool for profiling antibody cross-reactivity. Hum Antibodies,14(1-2):7–15, 2005.

[45] B. Ayoglu, A. Haggmark, M. Khademi, T. Olsson, M. Uhlen, J. M. Schwenk, andP. Nilsson. Autoantibody profiling in multiple sclerosis using arrays of human proteinfragments. Mol Cell Proteomics, 12(9):2657–72, 2013.

[46] A. Yeste and F. J. Quintana. Antigen microarrays for the study of autoimmunediseases. Clin Chem, 59(7):1036–44, 2013.

[47] Ronald Frank. Spot-synthesis: an easy technique for the positionally addressable,parallel chemical synthesis on a membrane support. Tetrahedron, 48(42):9217–9232,1992.

[48] S. Buus, J. Rockberg, B. Forsstrom, P. Nilsson, M. Uhlen, and C. Schafer-Nielsen.High-resolution mapping of linear antibody epitopes using ultrahigh-density peptidemicroarrays. Mol Cell Proteomics, 11(12):1790–800, 2012.

[49] S. Gaseitsiwe, D. Valentini, S. Mahdavifar, M. Reilly, A. Ehrnst, and M. Maeurer.Peptide microarray-based identification of mycobacterium tuberculosis epitope bind-ing to hla-drb1*0101, drb1*1501, and drb1*0401. Clin Vaccine Immunol, 17(1):168–75, 2010.

[50] F. Breitling, A. Nesterov, V. Stadler, T. Felgenhauer, and F. R. Bischoff. High-density peptide arrays. Mol Biosyst, 5(3):224–34, 2009.

44

References

[51] C. Schirwitz, F. F. Loeffler, T. Felgenhauer, V. Stadler, A. Nesterov-Mueller,R. Dahint, F. Breitling, and F. R. Bischoff. Purification of high-complexity pep-tide microarrays by spatially resolved array transfer to gold-coated membranes. AdvMater, 25(11):1598–602, 2013.

[52] B. Forsstrom, B. B. Axnas, K. P. Stengele, J. Buhler, T. J. Albert, T. A. Richmond,F. J. Hu, P. Nilsson, E. P. Hudson, J. Rockberg, and M. Uhlen. Proteome-wide epi-tope mapping of antibodies using ultra-dense peptide arrays. Mol Cell Proteomics,13(6):1585–97, 2014.

[53] R. F. Halperin, P. Stafford, and S. A. Johnston. Exploring antibody recognition ofsequence space through random-sequence peptide microarrays. Mol Cell Proteomics,10(3), 2011.

[54] G. Fenalti and M. J. Rowley. Gad65 as a prototypic autoantigen. J Autoimmun,31(3):228–32, 2008.

[55] J. M. Wenzlau, L. M. Frisch, J. C. Hutton, and H. W. Davidson. Mapping ofconformational autoantibody epitopes in znt8. Diabetes Metab Res Rev, 27(8):883–6, 2011.

[56] P. Stafford, Z. Cichacz, N. W. Woodbury, and S. A. Johnston. Immunosignaturesystem for diagnosis of cancer. Proc Natl Acad Sci U S A, 111(30):E3072–80, 2014.

[57] K. L. Kellar and M. A. Iannone. Multiplexed microsphere-based flow cytometricassays. Exp Hematol, 30(11):1227–37, 2002.

[58] K. Drobin, P. Nilsson, and J. M. Schwenk. Highly multiplexed antibody suspensionbead arrays for plasma protein profiling. Methods Mol Biol, 1023:137–45, 2013.

[59] W. Wu, H. Slastad, D. de la Rosa Carrillo, T. Frey, G. Tjonnfjord, E. Boretti, H. C.Aasheim, V. Horejsi, and F. Lund-Johansen. Antibody array analysis with label-based detection and resolution of protein size. Mol Cell Proteomics, 8(2):245–57,2009.

[60] C. P. Paweletz, L. Charboneau, V. E. Bichsel, N. L. Simone, T. Chen, J. W. Gille-spie, M. R. Emmert-Buck, M. J. Roth, Iii Ef Petricoin, and L. A. Liotta. Reversephase protein microarrays which capture disease progression show activation of pro-survival pathways at the cancer invasion front. Oncogene, 20(16):1981–9, 2001.

[61] C. Wingren, A. Sandstrom, R. Segersvard, A. Carlsson, R. Andersson, M. Lohr,and C. A. Borrebaeck. Identification of serum biomarker signatures associated withpancreatic cancer. Cancer Res, 72(10):2481–90, 2012.

[62] A. Haggmark, S. Bystrom, B. Ayoglu, U. Qundos, M. Uhlen, M. Khademi, T. Olsson,J. M. Schwenk, and P. Nilsson. Antibody-based profiling of cerebrospinal fluid withinmultiple sclerosis. Proteomics, 13(15):2256–67, 2013.

45

References

[63] C. A. Borrebaeck and C. Wingren. Recombinant antibodies for the generation ofantibody arrays. Methods Mol Biol, 785:247–62, 2011.

[64] C. J. Van Oss. Hydrophobic, hydrophilic and other interactions in epitope-paratopebinding. Mol Immunol, 32(3):199–211, 1995.

[65] B. Friguet, A. F. Chaffotte, L. Djavadi-Ohaniance, and M. E. Goldberg. Measure-ments of the true affinity constant in solution of antigen-antibody complexes byenzyme-linked immunosorbent assay. J Immunol Methods, 77(2):305–19, 1985.

[66] S. J. Seligman. Influence of solid-phase antigen in competition enzyme-linked im-munosorbent assays (elisas) on calculated antigen-antibody dissociation constants.J Immunol Methods, 168(1):101–10, 1994.

[67] A. E. Wammes, M. J. Fischer, N. J. de Mol, M. B. van Eldijk, F. P. Rutjes, J. C.van Hest, and F. L. van Delft. Site-specific peptide and protein immobilizationon surface plasmon resonance chips via strain-promoted cycloaddition. Lab Chip,13(10):1863–7, 2013.

[68] K. M. Muller, K. M. Arndt, and A. Pluckthun. Model and simulation of multivalentbinding to fixed ligands. Anal Biochem, 261(2):149–58, 1998.

[69] M. H. Van Regenmortel. From absolute to exquisite specificity. reflections on thefuzzy nature of species, specificity and antigenic sites. J Immunol Methods, 216(1-2):37–48, 1998.

[70] S. D. Gan and K. R. Patel. Enzyme immunoassay and enzyme-linked immunosorbentassay. J Invest Dermatol, 133(9):e12, 2013.

[71] R. C. Ladner. Mapping the epitopes of antibodies. Biotechnol Genet Eng Rev,24:1–30, 2007.

[72] M. Z. Atassi. Antigenic structure of myoglobin: the complete immunochemicalanatomy of a protein and conclusions relating to antigenic structures of proteins.Immunochemistry, 12(5):423–38, 1975.

[73] P. Stafford, R. Halperin, J. B. Legutki, D. M. Magee, J. Galgiani, and S. A. Johnston.Physical characterization of the "immunosignaturing effect". Mol Cell Proteomics,11(4):M111 011593, 2012.

[74] J. M. Gershoni, A. Roitburd-Berman, D. D. Siman-Tov, N. Tarnovitski Freund, andY. Weiss. Epitope mapping: the first step in developing epitope-based vaccines.BioDrugs, 21(3):145–56, 2007.

[75] E. Davidson and B. J. Doranz. A high-throughput shotgun mutagenesis approachto mapping b-cell antibody epitopes. Immunology, 143(1):13–20, 2014.

[76] G. N. Sivalingam and A. J. Shepherd. An analysis of b-cell epitope discontinuity.Mol Immunol, 51(3-4):304–9, 2012.

46

References

[77] J. W. Stave and K. Lindpaintner. Antibody and antigen contact residues defineepitope and paratope size and structure. J Immunol, 191(3):1428–35, 2013.

[78] U. Reineke and R. Sabat. Antibody epitope mapping using spot peptide arrays.Methods Mol Biol, 524:145–67, 2009.

[79] M. Brink, M. Hansson, L. Mathsson, P. J. Jakobsson, R. Holmdahl, G. Hallmans,H. Stenlund, J. Ronnelid, L. Klareskog, and S. Rantapaa-Dahlqvist. Multiplex anal-yses of antibodies against citrullinated peptides in individuals prior to developmentof rheumatoid arthritis. Arthritis Rheum, 65(4):899–910, 2013.

[80] J. Rockberg, J. Lofblom, B. Hjelm, M. Uhlen, and S. Stahl. Epitope mapping ofantibodies using bacterial surface display. Nat. Methods, 5(12):1039–45, 2008.

[81] G. Chao, J. R. Cochran, and K. D. Wittrup. Fine epitope mapping of anti-epidermalgrowth factor receptor antibodies through random mutagenesis and yeast surfacedisplay. J Mol Biol, 342(2):539–50, 2004.

[82] L. R. Pepper, Y. K. Cho, E. T. Boder, and E. V. Shusta. A decade of yeast surfacedisplay technology: where are we now? Comb Chem High Throughput Screen,11(2):127–34, 2008.

[83] V. Bottger and A. Bottger. Epitope mapping using phage display peptide libraries.Methods Mol Biol, 524:181–201, 2009.

[84] H. B. Larman, Z. Zhao, U. Laserson, M. Z. Li, A. Ciccia, M. A. Gakidis, G. M.Church, S. Kesari, E. M. Leproust, N. L. Solimini, and S. J. Elledge. Autoantigendiscovery with a synthetic human peptidome. Nat Biotechnol, 29(6):535–41, 2011.

[85] Y. Zhao and B. T. Chalt. Protein epitope mapping by mass spectrometry. AnalChem, 66(21):3723–6, 1994.

[86] H. Lund, E. Paus, P. Berger, U. H. Stenman, T. Torcellini, T. G. Halvorsen, andL. Reubsaet. Epitope analysis and detection of human chorionic gonadotropin (hcg)variants by monoclonal antibodies and mass spectrometry. Tumour Biol, 35(2):1013–22, 2014.

[87] C. Hager-Braun and K. B. Tomer. Determination of protein-derived epitopes bymass spectrometry. Expert Rev Proteomics, 2(5):745–56, 2005.

[88] V. H. Obungu, V. Gelfanova, and L. Huang. Epitope mapping of antibodies by massspectroscopy: a case study. Methods Mol Biol, 988:291–302, 2013.

[89] D. Paus and G. Winter. Mapping epitopes and antigenicity by site-directed masking.Proc Natl Acad Sci U S A, 103(24):9172–7, 2006.

[90] M. H. Van Regenmortel. What is a b-cell epitope? Methods Mol Biol, 524:3–20,2009.

47

References

[91] T. P. Hopp and K. R. Woods. Prediction of protein antigenic determinants fromamino acid sequences. Proc Natl Acad Sci U S A, 78(6):3824–8, 1981.

[92] J. Kyte and R. F. Doolittle. A simple method for displaying the hydropathic char-acter of a protein. J Mol Biol, 157(1):105–32, 1982.

[93] E. A. Emini, J. V. Hughes, D. S. Perlow, and J. Boger. Induction of hepatitis avirus-neutralizing antibody by a virus-specific synthetic peptide. J Virol, 55(3):836–9, 1985.

[94] P. A. Karplus and G. E. Schulz. Prediction of chain flexibility in proteins. Natur-wissenschaften, 72(4):212–213, 1985.

[95] M. J. Blythe and D. R. Flower. Benchmarking b cell epitope prediction: underper-formance of existing methods. Protein Sci, 14(1):246–8, 2005.

[96] J. Rockberg and M. Uhlen. Prediction of antibody response using recombinanthuman protein fragments as antigen. Protein Sci, 18(11):2346–55, 2009.

[97] J. A. Greenbaum, P. H. Andersen, M. Blythe, H. H. Bui, R. E. Cachau, J. Crowe,M. Davies, A. S. Kolaskar, O. Lund, S. Morrison, B. Mumey, Y. Ofran, J. L. Pelle-quer, C. Pinilla, J. V. Ponomarenko, G. P. Raghava, M. H. van Regenmortel, E. L.Roggen, A. Sette, A. Schlessinger, J. Sollner, M. Zand, and B. Peters. Towards aconsensus on datasets and evaluation metrics for developing b-cell epitope predictiontools. J Mol Recognit, 20(2):75–82, 2007.

[98] U. Kulkarni-Kale, S. Bhosle, and A. S. Kolaskar. Cep: a conformational epitopeprediction server. Nucleic Acids Res, 33(Web Server issue):W168–71, 2005.

[99] P. Haste Andersen, M. Nielsen, and O. Lund. Prediction of residues in discontinuousb-cell epitopes using protein 3d structures. Protein Sci, 15(11):2558–67, 2006.

[100] S. Fredriksson, M. Gullberg, J. Jarvius, C. Olsson, K. Pietras, S. M. Gustafsdottir,A. Ostman, and U. Landegren. Protein detection using proximity-dependent dnaligation assays. Nat Biotechnol, 20(5):473–7, 2002.

[101] R. Aebersold and M. Mann. Mass spectrometry-based proteomics. Nature,422(6928):198–207, 2003.

[102] O. Poetz, S. Hoeppe, M. F. Templin, D. Stoll, and T. O. Joos. Proteome widescreening using peptide affinity capture. Proteomics, 9(6):1518–23, 2009.

[103] N. Olsson, P. James, C. A. Borrebaeck, and C. Wingren. Quantitative proteomicstargeting classes of motif-containing peptides using immunoaffinity-based mass spec-trometry. Mol Cell Proteomics, 11(8):342–54, 2012.

[104] H. M. Geysen, J. A. Tainer, S. J. Rodda, T. J. Mason, H. Alexander, E. D. Get-zoff, and R. A. Lerner. Chemistry of antibody binding to a protein. Science,235(4793):1184–90, 1987.

48

References

[105] D. J. Barlow, M. S. Edwards, and J. M. Thornton. Continuous and discontinuousprotein antigenic determinants. Nature, 322(6081):747–8, 1986.

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

[107] M. Zeiler, W. L. Straube, E. Lundberg, M. Uhlen, and M. Mann. A protein epitopesignature tag (prest) library allows silac-based absolute quantification and multi-plexed determination of protein copy numbers in cell lines. Mol Cell Proteomics,11(3):O111 009613, 2012.

[108] P. Picotti and R. Aebersold. Selected reaction monitoring-based proteomics: work-flows, potential, pitfalls and future directions. Nat Methods, 9(6):555–66, 2012.

49

Characterizationofantibodyspeciﬁcity …kth.diva-portal.org/smash/get/diva2:762237/FULLTEXT01.pdf · 2014. 11. 11. · Thesis defense This thesis will be defended November 28th

Documents