sign in sign up
<<
Home , Bacterial display, Protein purification

Generation and characterization of for research

Karin Larsson

Royal Institute of Technology School of Biotechnology Stockholm 2009 ii GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

© Karin Larsson Stockholm 2009

Royal Institute of Technology School of Biotechnology Division of Proteomics AlbaNova University Center SE-106 91 Stockholm Sweden

Printed by Universitetsservice US-AB Drottning Kristinas väg 53B SE-100 44 Stockholm Sweden

ISBN 978-91-7415-439-9 TRITA-BIO Report 2009:21 ISSN 1654-2312

Cover illustration: Confocal image of immunofluorescently stained A-431 cells, image and immunohistochemical of colon. KARIN LARSSON iii

Karin Larsson (2009): Generation and characterization of antibodies for proteomics research. School of Biotechnology, Royal Institute of Technology (KTH), Sweden. Abstract Specific antibodies are invaluable tools for proteomics research. The availability of thoroughly validated antibodies will help to improve our understanding of expression, localization and function; fundamental processes and features of all living organisms. The objectives of the studies in this thesis were to develop high-throughput methods to facilitate the generation and purification of monospecific antibodies, and to address problems associated with selection for difficult target and subsequent validation issues. In the first of the studies, it was demonstrated that antibodies specific to human proteins could be generated in a high-throughput manner using protein signature tags (PrESTs) as both and affinity ligands. A previously developed purification process was adapted to a high-throughput format and this, in combination with the development of a protein microarray , resulted in monospecific antibodies that were used for profiling protein expression in 48 human tissues. Data obtained in these analyses suggest that a complete Human Protein Atlas should be attainable within the next ten years. In order to reduce the number of animals needed for such a massive project, and improve the cost-efficiency of generation, a multiplex immunization strategy was developed in a further study. Antisera from rabbits immunized with mixtures of two, three, five and up to ten different PrESTs were successfully purified and analyzed for specificity using protein arrays. Almost 80% of the animals immunized with up to three PrESTs yielded antibodies towards all the PrESTs administered, and they yielded comparable immunohistochemical staining patterns (of consecutive human tissue sections) to those of antibodies obtained from traditional single PrEST immunizations. Proteins with highly similar sequences to other proteins present a major challenge for the proteome-wide generation of antibodies. In another study, Cytokeratin-17 which displays high sequence similarity to closely related members of the intermediate filament family, was used as a model and the specificity and cross-reactivity of antibodies generated against this target were investigated using in combination with comparative IHC analyses. Antibodies identified by epitope mapping as binding to the most unique parts of the Cytokeratin-17 PrESTs also showed the most Cytokeratin-17-like staining pattern, thus further supporting the strategy of using sequence identity scores as the main criteria for PrEST design. An alternative antigen design strategy was investigated for use in raising antibodies towards G-protein- coupled receptors (GPCRs). The extracellular loops and N-terminus of each of three selected GPCRs were assembled to form single antigens and the resulting antibodies were analyzed by flow cytometric and confocal microscopic analyses of cell lines over-expressing the respective receptors. The results from both flow cytometric and immunofluorescence analyses showed that the antibodies were able to bind to their targets. In addition, the antibodies were used successfully for the in situ analysis of human brain and pancreatic islet cells.

Keywords: antibody, antibody validation, , immunofluorescence, tissue microarray, Western blot, epitope mapping, antigen design © Karin Larsson iv GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH KARIN LARSSON v

List of publications

This thesis is based upon the following four papers, which are referred to in the text by the corresponding roman numerals (I-IV). The papers are included in the Appendix.

I Nilsson P., Paavilainen L.*, Larsson K.*, Ödling J.*, Sundberg M. Andersson A., Kampf C., Persson A., Al-Khalili Szigyarto C., Ottosson J., Björling E., Hober S., Wernérus H., Wester K., Pontén F. and Uhlén M. (2005) Towards a human proteome atlas: high-throughput generation of mono-specific antibodies for tissue profiling. Proteomics. 5, p. 4327-4337

II Larsson K., Wester K., Nilsson P., Uhlén M., Hober S., Wernérus H. (2006). Multiplexed PrEST-immunizations for high-throughput affinity proteomics. J. Immunol. Methods. 315, p. 110-120

III Larsson K., Eriksson C., Schwenk J.M., Berglund L., Wester K., Uhlén M., Hober S., Wernérus H. (2009). Characterization of PrEST-based antibodies towards human Cytokeratin-17. J. Immunol. Methods. 342 p. 20-32

IV Larsson K., Hofström C., Lindskog C., Hansson M., Angelidou P., Hökfelt T., Uhlén M., Wernérus H., Gräslund T., Hober S. (2009) Novel antigen design for the generation of G-protein-coupled antibodies. Manuscript

*These authors contributed equally to this work.

All papers are reproduced with the kind permission of the respective copyright holders.

vi GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Related publication

Uhlén M., Björling E., Agaton C., Szigyarto C. A., Amini B., Andersen E., Andersson A. C., Angelidou P., Asplund A., Asplund C., Berglund L., Bergström K., Brumer H., Cerjan D., Ekström M., Elobeid A., Eriksson C., Fagerberg L., Falk R., Fall J., Forsberg M., Björklund M. G., Gumbel K., Halimi A., Hallin I., Hamsten C., Hansson M., Hedhammar M., Hercules G., Kampf C., Larsson K., Lindskog M., Lodewyckx W., Lund J., Lundeberg J., Magnusson K., Malm E., Nilsson P., Ödling J., Oksvold P., Olsson I., Öster E., Ottosson J., Paavilainen L., Persson A., Rimini R., Rockberg J., Runeson M., Sivertsson A., Sköllermo A., Steen J., Stenvall M., Sterky F., Strömberg S., Sundberg M., Tegel H., Tourle S., Wahlund E., Waldén A., Wan J., Wernérus H., Westberg J., Wester K., Wrethagen U., Xu L. L., Hober S., Pontén F. (2005) A human protein atlas for normal and cancer tissues based on antibody proteomics. Mol Cell Proteomics. 4(12) p.1920-1932

KARIN LARSSON vii

Contents

Introduction...... 1 1. Proteomics...... 2 1.1 Mass spectrometry-based proteomics ...... 3 1.2 Affinity-based proteomics...... 4 1.2.1 Human Proteome Resource...... 5 1.2.2 Other affinity-based proteomics projects ...... 7 2. Affinity reagents...... 10 2.1 Antibodies...... 11 2.1.1 Polyclonal antibodies ...... 12 2.1.2 Monoclonal antibodies ...... 14 2.1.3 ...... 14 2.2 Alternative protein scaffolds...... 16 Experimental techniques...... 19 3. Methodology ...... 20 3.1 Antibody purification...... 20 3.1.1 Fc-binding proteins ...... 20 3.1.2 Antigen-based affinity purification ...... 21 3.2 Methods for antibody characterization and their use in proteomics research ...... 23 3.2.1 ELISA...... 23 3.2.2 Western blotting ...... 25 3.2.3 Immunohistochemistry and Tissue Microarrays ...... 25 3.2.4 Immunofluorescence microscopy...... 27 3.2.5 ...... 28 3.2.6 Epitope mapping ...... 28 Present Investigation ...... 31 4. Objective...... 32 5. High-throughput generation of monospecific antibodies (Paper I)...... 34 6. An alternative immunization strategy using multiplexed PrEST-antigens (Paper II)...... 38 7. Generation and validation strategies for antibodies raised towards targets with highly similar sequences (Paper III) ...... 43 8. An alternative strategy for designing antigens for membrane-bound receptors (Paper IV) ...... 48 viii GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

9. Concluding remarks ...... 53 References ...... 57 Acknowledgements...... 64 KARIN LARSSON ix

Abbreviations

2-DE Two-dimensional gel ABP Albumin binding protein AP Alkaline phosphatase CDR Complementarity-determining region ELISA -linked immunosorbent assay Fab Fragment, antigen binding Fc Fragment, crystallizable FACS Fluorescence-activated cell sorting HAMA Human anti-mouse antibody HIAR Heat-induced antigen retrieval HPR Human Proteome Resource HRP Horseradish peroxidase HUPO Human Proteome Organisation IEF Isoelectric focusing IF Immunofluorescence IHC Immunohistochemistry IMAC Immobilized metal ion affinity IPG Immobilized pH gradient mAb Monoclonal antibody MALDI Matrix-assisted laser desorption/ionization MS Mass spectrometry msAb Monospecific antibody pAb Polyclonal antibody pI Isoelectric point PrEST Protein Epitope Signature Tag PTM Post-translational modification RT-PCR Reverse transcriptase polymerase chain reaction ScFv Single-chain variable fragment SDS-PAGE polyacrylamide SpA Staphylococcal SpG Streptococcal Protein G TOF Time-of-flight TMA Tissue Microarray x GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

KARIN LARSSON 1

Introduction 2 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

1. Proteomics

The completion of the human genome sequence [4, 5] has allowed genome-wide studies, which have uncovered major features of our genetic code (e.g. less than 2% is protein- encoding and different structural variations such as single-nucleotide polymorphism) furthering our understanding of biology. The study of , transcriptomics, has added further to this knowledge through discoveries such as non-coding RNA and small RNAs with regulatory properties [6]. Even though this has generated important information about the causes of and/or susceptibility of individuals to many diseases [7, 8], further studies of protein function are required to fully understand cell organization and function. This is because the products of gene expression, the proteins, are the effectors of most tasks in the cell, in addition to constituting about 80% of all drug targets [9].

The term proteome was first used by Mark Wilkins, in 1996 [10], for the entire set of proteins expressed by a cell, tissue or organism at a given time point. Proteomics is the study of the proteome, and involves the large-scale study of proteins, particularly their structures, biological functions and interactions with other proteins. Currently, there are two distinct branches of proteomics; mass spectrometry-based proteomics and affinity-based proteomics.

Huge financial and technical resources have been expended in sequencing the human genome, leading to a number of breakthroughs (e.g. new sequencing technologies), but deciphering the function of all proteins is even more challenging. Many factors influence the proteome that cannot be identified or characterized by whole genome or transcriptome analysis, including mRNA degradation and translational efficiency, alternative splicing, post-translational modifications (PTMs) and protein degradation rates [11]. The number of protein-coding genes is estimated to be around 20500 [12, 13]. If alternative splicing and PTMs are included there are estimated to be over 100 000 proteins variants [14], greatly increasing the complexity.

Other problems facing proteomics researchers trying to analyze the whole proteome are posed by the huge range of abundances of proteins, spanning several orders (7-8) of magnitude in tissues and even more (>10) in plasma samples [11, 15]. Abundant proteins, such as albumin in plasma, tend to mask those that are less abundant in analyses by many techniques. Therefore, there are substantial challenges when using currently available proteome profiling and detection techniques. Indeed, prefractionation and removal of highly abundant proteins is a requirement when applying existing methods. In addition, unlike nucleic acids, proteins have KARIN LARSSON 3 high biochemical heterogeneity, making analysis of all proteins by any single method unfeasible.

1.1 Mass spectrometry-based proteomics

Two-dimensional gel electrophoresis (2-DE), in combination with mass spectrometry, is highly associated with proteomics research. It was in 1975 that separation of proteins in complex mixtures was made possible with the introduction of high-resolution two-dimensional gel electrophoresis (2-DE) [16-18]. Subsequent technical improvements, like the introduction of immobilized pH gradients (IPGs) [19, 20], have allowed the resolution of over 5000 proteins simultaneously with current technologies [21]. 2-DE provides scientists with knowledge of protein expression levels, post-translational modifications and protein isoforms. In the first dimension proteins are separated using isoelectric focusing (IEF) based on differences in their isoelectric points (pI) and then separated in the second dimension, according to their masses using SDS-PAGE (sodium dodecylsulfate polyacrylamide gel electrophoresis). Following visualization of protein spots with suitable staining methods (e.g. Coomassie blue or silver staining), spots are excised from the gel, enzymatically digested and extracted into a liquid phase. The protein digests obtained are then analyzed by mass spectrometry. A major advantage with 2-DE is its ability to simultaneously analyze the expression of proteins in complex protein mixtures, such as whole cell lysates and tissue lysates. Additionally there is the possibility of detecting changes in protein expression levels, protein isoforms and posttranslational modifications, such as phosphorylations and glycosylations [21]. Limitations with this technology include difficulties in separating very hydrophobic proteins like cell surface receptors, automation problems and the need for time-consuming optimization, depending on the type of proteins to be studied [22]. Moreover, it is necessary to reduce the complexity of samples using prefractionation methods and to deplete them of highly abundant proteins in order to detect the fractions containing less abundant proteins.

Mass spectrometry (MS) provides highly sensitive and accurate mass determinations of biomolecules. It is a versatile tool in proteomics research with a wide range of applications, including protein identification, quality control of recombinant proteins and studies of post- translational modifications [23]. The most prominent use of MS is in combination with 2-DE, in which masses of enzymatic digests of proteins are determined and peptide mass “fingerprints” are obtained [24-26] for comparison with entries in available protein databases.

4 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Various combinations of ionization techniques, mass analyzers and detectors may be used. The ionization sources used for protein research are electrospray [27] and matrix-assisted laser desorption/ionization (MALDI) systems [28], since their relatively gentle modes of action permit analysis of large polypeptides. These ionization systems are combined with different mass analyzers in order to separate the ions formed, depending on the application. This mass separation can be accomplished with time-of-flight (TOF), quadrupole, or ion trap mass spectrometers, as well as combinations of these instruments.

Mass spectrometry is also used without 2-DE in proteomics research sometimes, e.g. in shotgun proteomics [29], a method devoid of time-consuming gel handling, in which whole protein samples are enzymatically digested in solution with no prior separation on gels. Following digestion, the peptides are separated by liquid chromatography and then detected using tandem mass spectrometry. Other applications of MS include detection of protein- protein interactions, where tagged proteins are precipitated and their binding partners identified according to accurate mass determinations, taking advantage of affinity interactions but not to the target protein [23, 30].

1.2 Affinity-based proteomics

Although two-dimensional gel electrophoresis combined with mass spectrometry has been historically synonymous with proteomics research, other approaches, such as affinity-based proteomics are becoming increasingly important. In affinity-based proteomics the systematic exploration of the proteome is undertaken by high-throughput generation of various types of affinity reagents. These binders are valuable tools in a wide range of analyses and applications, such as: the detection of target proteins in various assays (e.g. Western blots, ELISA, protein array analyses and immunohistochemical assays) and the purification or pull-down of proteins or protein complexes from complex mixtures (e.g. serum and tissue extracts) for use in, for example, protein structure determinations [1].

Creating affinity-binders to the human proteome is considered an overwhelming task. Several hundred thousand binders would be needed in order to achieve complete coverage of the proteome, especially if protein splice variants and post-translational modifications and the possible need for several binders to each target (in order to perform ‘sandwich’ assays and for validation purposes) is taken into account. Therefore, ProteomeBinders, a European KARIN LARSSON 5 consortium was formed (amongst others) with the aim of coordinating and standardizing independent efforts to produce well-validated affinity binders for proteomics research [31].

A key issue in affinity proteomics is to ensure the quality of the binders generated. This can be addressed in various ways. can be used to minimize possible cross-reactivity by selecting antigens that are not similar to other human proteins, by performing Blast searches (if working with peptides or parts or proteins as antigens) [32, 33] or by selecting antibodies via epitope mapping that bind to dissimilar regions (Paper III). Analytical techniques like Western blotting are also useful for determining whether developed affinity-reagents bind specifically to their full-length target proteins. In situ hybridization permits comparison of levels of RNA and protein expression. Although large data sets have already been published, RNA and protein levels do not necessary correlate [11, 34, 35]. Comparative protein expression analysis using tissue microarrays and several binders to the same target is perhaps the most favorable strategy available to validate novel affinity reagents, especially when combined with literature searches. In particular, when working with binders of proteins of unknown function and cellular distribution, a careful approach to ensure quality is to produce several binders to non- overlapping and compare results obtained from different binding assays [36, 37]. Access for the scientific community to well-characterized affinity binders to all human proteins would greatly improve and facilitate research on protein function and biomarker discovery.

1.2.1 Human Proteome Resource

The overall objective of the Human Proteome Resource (HPR), an antibody-based proteomics program, is to generate affinity-binders (monospecific antibodies) to all non-redundant proteins in the human proteome, i.e. one representative protein per gene locus, and to compile a publicly available database with data showing the expression and localization of these proteins in a wide range of human tissues, cell lines and cancer types [38]. Generation of monospecific antibodies (msAbs) is carried out using Protein Epitope Signature Tags (PrESTs) as antigens [39]. The PrESTs are about 100-150 long polypeptide sequences derived from the human genome sequence retrieved from the Ensembl database [13]. A suitable region of the open reading frame is selected, using a bioinformatics tool [32], based on a set of criteria such as: low identity to other human proteins and absence of regions of low complexity, transmembrane regions, signal peptides and restriction sites. Once a PrEST has been selected, primers are designed and the gene fragments are amplified using the reverse transcriptase polymerase chain reaction (RT-PCR) utilizing different mRNA pools. Following 6 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

)*!$ !%&'( $ +,-

$ !"#$%&'()*+, -./,*,+ !"/0(*,12"/'340*/,

!"#$

7883,*<60*/, =,0*5/':123"*9*460*/, !"/0(*,16""6:

;()0(",15./0 %*))3(18*4"/6""6: 7883,/9.3/"()4(,4(

!35.*41'60656)( Figure 1. An overview of the production flow in the Human Proteome Resource (HPR) project. Protein

Epitope Signature Tags (PrESTs) are designed using a bioinformatics tool, cloned into an expression vector, produced in E. coli and used to immunize New Zealand White rabbits. Monospecific antibodies are then generated in a two-step purification method followed by thorough validation. The quality- assured antibodies are subsequently used for expression and localization studies by tissue microarray analyses and immunofluorescence microscopy.

KARIN LARSSON 7 cloning of the amplicon into a destination vector and sequence verification, the PrESTs are expressed in as fusions with a hexa-histidine (His6) tag, for purification of the antigen, and an albumin binding protein (ABP) for enhanced solubility [40]. Purification of the

His6-tagged proteins is performed with a semi-automated approach [41] using Immobilized

Metal ion (IMAC) [42]. In addition, the fusion partner, His6-ABP, is expressed and purified for later use in purification of the antibody. Following antigen preparation, the His6ABP-PrEST fusions are used to immunize rabbits and the resultant antisera are affinity-purified using a two-step protocol (see Figure 1). First, tag-specific antibodies are depleted from the antisera using His6ABP-coupled affinity columns and the flow-through is subsequently purified on a second affinity column coupled with the respective

His6ABP-PrEST previously used for immunization [43]. The monospecific antibodies are initially analyzed on protein arrays consisting of 384 purified PrEST fusions in order to ensure that no detectable/low cross-reactivity towards other proteins or the fusion tag remains. Antibodies passing the PrEST array screening are further validated by Western blots utilizing two cell lines; RT-4 and U-251MG and two human tissues; tonsil and liver, as well as human plasma. Furthermore, the antibodies are used for protein expression profiling on tissue microarrays (TMAs), first on a smaller test array and then on arrays consisting of 48 normal tissues, 20 different cancer types and 47 cell lines. The tissue and cancer moieties of the arrays are manually annotated by pathologists [44] and the cell arrays are annotated automatically by TMAx software [45]. Finally, immunofluorescence (IF) analysis using confocal microscopy is performed utilizing the monospecific antibodies to localize expression of particular proteins to specific cell compartments. Three different cell lines are used: A-431, U-2OS and U-251MG [46]. Based on the results obtained from Western blot, IHC and IF analyses, a validation score is appointed to the antibody and published together with the rest of the results on the Protein Atlas web site (www.proteinatlas.org).

In addition to creating a map of the human proteome, the enormous volume of data and images on the Protein Atlas has great potential utility in biomarker research aimed at finding cell- and tissue-specific protein expression. This could be used for the diagnosis, prognosis and monitoring of diseases.

1.2.2 Other affinity-based proteomics projects

In addition to the Human Proteome Resource program, other initiatives have been started with the common goal of producing validated affinity reagents to be made available to the 8 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH proteomics community. In this section, I will give a brief summary of current affinity-based proteomics projects.

The Sanger “Atlas of protein expression” initiative published an article in 2007 in which they described the selection and validation of a vast number of recombinant antibodies using and IHC [47]. A phage-antibody library was constructed containing over 1010 clones. Proteins used as targets for subsequent selection were produced both in bacterial and mammalian expression systems [48, 49]. A total number of 38000 antibody clones, recognising 290 target proteins, were selected and screened. Following DNA sequencing to reduce redundancy, the number was reduced to 7200 unique antibody clones and 4400 were selected for further evaluation. A selection success rate of over 70% enabled generation of 25 antibodies for each target protein. When an antibody had been successfully selected and found to meet various selection criteria it could be transferred into different expression systems, allowing a renewable source of antibodies. The recombinant antibodies produced were finally used to perform immunohistochemical staining of tissue microarrays. Even though time- consuming and costly screening of the numerous clones was required, preparation of numerous affinity-binders to the same target protein represents a great advantage in the validation of these molecules, since protein expression patterns in IHC analyses can be compared and evaluated as discussed above (section 1.2).

A completely different strategy was employed in a high-throughput antibody generation project in China [50]. The general idea was to immunize mice against fractionated plasma or tissue samples in order to produce monoclonal antibodies to proteins of high- and medium- abundance, thus avoiding the need to produce recombinant proteins. Initially, plasma was separated by liquid chromatography into three fractions, which were then injected into mice. Spleen cells from the immunized animals were homogenized and fused to myeloma cells. Hybridoma cells were selected and screened, and then in subsequent rounds of immunization the antibodies generated in the first or previous immunizations were utilized in order to deplete the highly abundant proteins in the plasma fractions. Since the monoclonal antibodies were produced using unidentified protein mixtures the target proteins then needed to be identified. This was accomplished with the use of ELISA and Western blot analyses as well as combined with MS. In this project, four immunizations generated 200 strains of monoclonal antibodies and over 30% of these antibodies recognized antigens of different sizes.

Among other initiatives is the Antibody Factory, a German national collaborative project aimed at developing in vitro methods for the generation of recombinant antibody fragments KARIN LARSSON 9

(scFv), and the American Clinical Proteomic Technologies for Cancer Reagents and Resources, funded by the US National Cancer Institute (NCI) aimed at the high-throughput generation of monoclonal antibodies. Other new projects aim to coordinate and promote the generation of well-characterized antibodies, including the HUPO Antibody Initiative and ProteomeBinders (as mentioned earlier). Also, the National Institutes of Health recently commenced the Protein Capture Tools program, which strives for high-throughput generation of monoclonal antibodies [31, 51-53].

10 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

2. Affinity reagents

In order to study the functions of proteins and their cellular organization in a systematic manner, efficient tools are needed for their high-throughput capture, detection and analysis. Invaluable tools in proteomics research are affinity reagents (i.e. molecules that bind specifically to intended target proteins) of various types, such as antibodies, antibody fragments and other non-antibody binders (summarized in Table 1). Access to affinity-binders is important for many methods for protein analysis such as immunoprecipitation of proteins and protein complexes, protein localization studies, development of new diagnostic assays and as potential drugs (e.g. for the inhibition/activation of receptors and delivery of toxins).

A good affinity reagent should be specific for its target, and should not bind to other proteins. They should also have suitable (usually high) affinity for the target protein and be functional in different technical platforms. In addition, it is highly desirable for them to be sufficiently stable to couple to effector molecules such as fluorophores and compatible with available secondary reagents.

Size Affinity reagent Origin Numbers of epitopes (kDa)

Polyclonal antibody Animal serum Several 150

Affinity-purified animal Mono-specific antibody Several 150 serum

Monoclonal antibody Hybridoma cells Single 150

Recombinant full-length Combinatorial libraries Single 150 antibody Recombinant antibody Combinatorial libraries Single 15-55 fragments Alternative scaffolds Combinatorial libraries Single 5-20

Table 1. Summary of affinity reagents discussed. Modified from [1].

Antibodies with natural variability, stability and high affinity, have historically been the most widely used affinity reagents. Technical advances in combinatorial display technologies [54] and the compilation of large antibody libraries have made it possible to select recombinant KARIN LARSSON 11 antibodies to most proteins and these, therefore, raise the possibility of an alternative source of high-affinity binders. An advantage of these molecules over polyclonal and monoclonal antibodies is that no animals are needed for their production, since this has been a limiting factor when generating binders at the whole proteome scale. Moreover, engineered antibody fragments could be promising alternatives to entire antibodies, since they still have the same binding properties, but are much easier to produce in high yield and at low cost in, for example, bacterial hosts [2]. In addition, has provided tools to develop binders that completely lack the globulin domain, here denoted alternative scaffolds, enabling innovative design of the binders depending on intended uses and targets.

2.1 Antibodies

Animals are exposed to diverse threats, ranging from invading pathogenic microorganisms to cancer (i.e. altered self-cells.). As a means of defense the immune system has evolved a remarkable set of mechanisms. The innate branch of the immune system is the first line of defense, comprising many different components, such as phagocytic cells, the skin and antimicrobial substances. It recognizes classes of molecules present on common pathogens opposed to the more specific, adaptive part of the immune system, which is activated upon such antigenic challenge and can be divided into cell-mediated and humoral immunity.

" # # "

" % $ $ & " # # & %

)* ! ! % ! ! ! ! $ $ ! ! % )+, % ' # # ' %

)- % ( # # ( %

Figure 2. Schematic representation of an antibody (IgG). Two heavy chains and two light chains are connected by disulfide bridges forming a bivalent molecule. The constant fragment (Fc) is responsible for effector functions while the antigen-binding fragment (Fab), primarily the variable fragment (Fv), is responsible for antigen binding.

12 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Effector T cells execute cell-mediated responses by activating phagocytic cells, as well as killing altered self-cells, such as tumor cells and virally infected cells. Humoral responses rely on B- cell-antigen interactions leading to proliferation of the B-cells and their differentiation into antibody-secreting plasma cells, upon activation by T-helper cells. Antibodies or immunoglobulins exist both as membrane-bound molecules conferring specificity to B-cells, and as secreted molecules circulating in the blood stream that detect, neutralize or mark foreign antigens for elimination from the body [55].

An antibody consists of two identical light chains and two identical heavy chains forming a characteristic Y-shaped molecule held together by disulfide bonds (Figure 2). The ‘arms’ of the Y-shaped molecule are responsible for antigen binding whereas the ‘foot’ of the molecule functions as an effector of various significant aspects of the immune response, such as complement activation, and as a receptor for phagocytic cells. The N-terminal domains of the heavy and light chains form the most variable part of the antibody and, therefore, denote the variable (V) region. Most of the amino acid variability is concentrated in three loops of each chain called complementarity-determining regions (CDRs). The remaining domains of the heavy and light chains, forming the constant (C) region, show less sequence variability. Immunoglobulins are divided into different subclasses depending on their heavy chain sequence. In humans, five different types of heavy chains are present (designated α, δ, ε, γ and µ), giving rise to five different subclasses: IgA, IgD, IgE, IgG and IgM, respectively. IgG is the most abundant type of immunoglobulin and hereafter when antibody is mentioned it will refer to IgG. In addition adaptive immunity has the remarkable feature of immunological ‘memory’, leading to a fast onset of the defense mechanism after a second exposure to a pathogen. Also, incredible variability results in the ability to create binders to most immunogenic substances. This is due to various events which lead to diversification, such as the rearrangement of somatic gene segments, junctional flexibility and somatic hypermutation [56, 57].

2.1.1 Polyclonal antibodies

A mixture of antibodies with varying affinities and specificities for the immunized antigen is present in polyclonal antisera. These are derived from non-identical B-cells. Before antibodies had been discovered the ability of polyclonal antisera to transmit passive immunity from an immunized to a non-immunized individual was demonstrated by Behring and Kitasato in 1890. KARIN LARSSON 13

This major breakthrough led to Behring being awarded the first Nobel Prize for research in immunology in 1901.

Polyclonal antibodies (pAbs) are produced by injecting an antigen into an animal such as a mouse, rabbit, goat or hen. The choice of animal depends on how much of the antibody is needed, as well as the phylogenetic relationship between the immunized animal and the original source of antigen. The bigger an animal the bigger its blood volume and thus the greater volume of antiserum will be obtained. In addition, the more distant the relationship between host and antigen, the more immunogenic the antigen should be. The antigen is usually administered together with an adjuvant such as Freund’s adjuvant, in order to enhance the immune response and the adjuvant-antigen mixture can be injected through various routes, for example subcutaneously or intramuscularly. After primary injection and possibly further booster injections, the blood is collected and centrifuged to remove blood cells and clotting factors. For egg yolk, several fractionation processes are available such as filtration, and precipitation in order to remove lipids and proteins. In polyclonal antisera, antibodies specific to the antigen used for immunization comprise approximately 1% of the total antibody population [14]. To purify these antibodies, the 99% of antibodies that are non- specific and unwanted can be removed using one or more of the purification strategies discussed in section 3.1. If the antigen used for immunization is used as an affinity in purification processes, a mixture of antibodies with varying specificities will still be present in the purified antiserum, but they will only recognize the target protein and be monospecific, in terms of their binding to the antigen.

Polyclonal antibodies work well in many technical applications [14]. Since they are able to recognize several different epitopes on the same target protein, because they contain a mixture of antibodies, problems with masked or denatured epitopes can be avoided. This kind of problem occurs during immunohistochemical staining of tissues where cross-linking of proteins often leads to antibody binding sites being inaccessible [58, 59]. Also, in SDS-PAGE dependent applications, such as Western blotting, most proteins are denatured, thus destroying many epitopes.

However, the fact that pAbs are mixtures of antibodies is a major drawback and possibly hindrance for potential therapeutic applications. The heterogeneity of any polyclonal antiserum means it is difficult to reproduce it with exactly the same composition. In therapeutics, therefore, polyclonal antisera are mainly used in replacement therapies where patients are suffering from immunodeficiency diseases, or for the neutralization of toxins and viruses [60, 61]. 14 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

2.1.2 Monoclonal antibodies

A major milestone in immunological research was the development of monoclonal antibodies by Köhler and Milstein in 1975; work for which they were later awarded the Nobel Prize in Physiology or Medicine [62]. They fused a mouse myeloma cell line with mouse spleen cells from an immunized donor. The resultant hybridoma cell line possessed the immortal growth properties of the cancerous plasma cells and the antibody secreting properties of the normal B- cells, thus creating an indefinitely growing cell line that produced large quantities of antibodies with the same specificity, i.e. recognizing a single epitope.

Monoclonal antibodies (mAbs) are produced by initially immunizing mice with the target protein. The subsequent fusion of spleen B-cells and myeloma cells is conducted in the presence of polyethylene glycol then the resultant cells are grown in a selective medium in which only hybridoma cells can proliferate [63]. Clones are subsequently screened for specificity using appropriate analytical methods (e.g. ELISA) and good antibody-producers are selected. For large-scale production of mAbs the hybridomas can be grown as cell cultures or injected into mice, where they produce tumors secreting antibody-rich fluid (ascites fluid). The benefits of monoclonal antibodies include the fact that they can be infinitely available and their single-epitope binding characteristics make them appropriate for therapeutic uses. Indeed, they are currently the most commonly used immunoreagents for diagnostic applications [64]. Limitations with mAb use include the time-consuming screening of the hybrid cells during their production and later problems with immune reactions when they are used in humans because of their murine origin. [63]. In addition, their single epitope recognition site means that they may have limited use in some applications in which antibody binding-sites may be disrupted or hidden, such as Western blotting.

2.1.3 Recombinant antibodies

As discussed above, the monospecific binding properties of monoclonal antibodies make them good reagents for therapeutic use. However, the murine origin of the antibodies has been shown to generate problems associated with poor effector function and immunogenicity, i.e. upon repeated administration human anti-monoclonal antibodies (HAMA) frequently develop [63, 65, 66]. Many attempts have been made to produce human hybridoma cells, but technical difficulties and problems with their stability and the efficiency of subsequent antibody production have hampered their development [67]. However, the introduction of recombinant KARIN LARSSON 15 full-length antibodies (e.g. chimeric antibodies, humanized antibodies and fully human antibodies) has reduced these problems [2]. Initially, efforts were made to exchange the constant part of the mouse mAb with constant regions of human origin and thereby create a chimeric mouse-human mAb [68, 69]. Antibodies with the constant part originating from humans would lead to improved effector functions as well as a decrease in immunogenicity compared to the murine equivalents. However, problems with anti-idiotypic reactions were reported for some chimeric monoclonal antibodies [70], leading to the development of humanized antibodies, in which as well as the constant region parts of the variable region are replaced by their human counterparts [71]. Humanization of the variable regions has been conducted by CDR grafting [72, 73]. In CDR grafting, the hypervariable loops of the mouse antibody are implanted onto the human framework and constant regions, thus reducing the murine content of the antibody considerably, leading to less potential immunogenicity than with the chimeric variants.

Intact antibodies have both advantages and disadvantages. They are bivalent binders, increasing their apparent affinity or avidity due to longer retention times on cell-surface receptors and on antigens with repeated motifs. They also maintain effector functions, as discussed above. These features are not always required for diagnostic and therapeutic purposes and they may lead to, for example, inappropriate complement activation [2]. The engineering of antibody fragments is an emerging field, offering potentially interesting alternatives to full-length antibodies. They may be more easily produced and fused successfully to different compounds (e.g. toxins and ). A vast range of antibody formats is available, as summarized in Figure 3.

Antibody Fab and single-chain Fv fragments consist of both VH and VL domains, joined by a flexible peptide linker in scFvs. These monovalent binders maintain the antigen-binding affinity of the parent IgG, but show improvements in, for example, tissue penetration. Single variable domains, especially from camels (VhH) and sharks (V-NAR), possess long surface loops which are able to penetrate cavities on antigens (e.g. viruses and enzyme active sites) that have been difficult to raise antibodies against by other means [74]. Unlike those produced by humans and mice, these single variable domains are soluble and stable, but their potential immunogenicity must be considered if used in vivo. Multivalent binders can be formed from

Fab and scFv fragments, such as Fab2 [75] and diabodies [76], triabodies [77] and tetrabodies, in order to obtain desired avidities [78]. Combining scFvs or Fabs with different specificities allows the formation of bispecific and even trispecific binders [79, 80].

16 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

'5*,."%*+),-&.)/+01 '()("%*+),-&.)/+01 9),%$,-97:

4.%&(/3 6%0.%&*+-0+7.()

!%&

8()1,%),-0+7.() #$!" !" 97: 20.%&(/3 >.1<+$.=.$-&.)/+01 #.)7*+-/(B%.)1 6?@-A8%B+*C

6;DEF-A#?%0GC 4.%&(/3;&.1<+$.=.$ 2+,0%&(/3 !%&--;&.1<+$.=.$2

Figure 3. Schematic representation of intact IgG monovalent binders (Fab, Fv and ScFv), multivalent binders (diabody, triabody and tetrabody), bispecific molecules (Fab2 and diabody) and single domain binders (VhH and V-NAR). Colored and uncolored ovals symbolize variable domains and constant domains, respectively, while circles represent antigen-binding sites. Black lines in single chain fragments and Fab2 fragments correspond to linkers. Modified from [2] and [3].

2.2 Alternative protein scaffolds

Recent advances in combinatorial engineering and library display technologies have allowed novel classes of binding molecules to be developed based on alternative scaffolds [81]. These binding molecules, typically selected through phage [82, 83], bacterial [84, 85] or ribosomal [86] display technologies, often show both high specificity and affinity for their targets. Antibody molecules have complex structures with long half-lives in serum, and they are also difficult and expensive to produce. Alternative scaffolds, on the other hand, are usually small, single-chain proteins lacking disulfide bonds or free cysteines, enabling their production in environments such as the bacterial [87]. The absence of free cysteines also permits the introduction of cysteines at desired positions allowing, for example, site-specific incorporation KARIN LARSSON 17 of fluorophores [88]. The small size of many alternative scaffolds also enhances tissue penetration and body clearance, which are advantageous features in bioimaging, and reduce their half-life in serum [87, 89]. However, the non-human origin of most alternative scaffolds means that if the binders are to be used in therapeutic applications, immunogenicity issues must be considered.

The is a three-helix bundle derived from an engineered domain of Staphylococcus aureus Protein A [90]. The binding surface of this 58-amino acid peptide consists of 13 amino acids distributed on two helices forming a relatively flat binding surface suitable for binding to proteins. Through randomization of these 13 amino acids a combinatorial library of about 109 binders was created, permitting (for example) the selection of binders of HER2 [91], and the amyloid beta peptide [92]. This cysteine-free scaffold can be selected by phage or , it can also be produced by solid-phase synthesis [93]. It has been demonstrated to be a stable scaffold, well suited for the purification, detection and targeting of proteins [94, 95].

Designed ankyrin-repeat proteins (DARPins) have been developed from naturally occurring mediators of protein-protein interactions [96]. These proteins are composed of 33 amino acid repeats, a β-turn followed by two antiparallel α-helices and a loop linking to the next repeat. The diversity of the DARPins comes both from the randomization of the interaction surface, and from variation in the number of repeats. Hence, this modularity allows the scaffold to be adapted to different binding surfaces. High-affinity molecules that recognize proteins such as the maltose binding protein have been documented [97].

The fibronectin type III domain has an antibody-like structure and is also a natural protein- protein interaction mediator, occurring in about 2% of all proteins. The 94 amino acid β- sandwich is similar to the VH domain of antibodies, but is not dependent on disulfide bridges to fold. Its variability lies in the randomization of the CDR-like loops [98]. A high-affinity binder to TNF-α has been selected using mRNA display [99].

Anticalins [100, 101] are derived from lipocalins, a family of proteins involved in the storage and transport of biological compounds. Although diverse in nature, lipocalins have a consensus structure; a barrel of eight antiparallel β-strands connected by four loops that form a cavity responsible for ligand binding. In anticalins 16 residues in the binding pocket are randomized in order to bind a diverse set of small molecules. Examples of such binders, selected with phage display, are the hapten fluorescein [100] and the toxin digoxigenin [102]. 18 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

KARIN LARSSON 19

Experimental techniques 20 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

3. Methodology

The purpose of this section is to summarize the techniques generally used, in particular those of central importance for the projects in this thesis, for the purification, validation and subsequent use of antibodies in affinity-based proteomics research.

3.1 Antibody purification

Regardless of the choice of affinity-binding molecule, purification increases the success rate of subsequent analyses. The removal of contaminating serum proteins and nonspecific host antibodies from, for example, polyclonal antisera decreases cross-reactivity and nonspecific binding, thus enhancing the quality of data obtained.

Affinity chromatography offers an excellent technique for antibody purification. The principles of affinity chromatography depend on the reversible interaction between a binding protein (e.g. antibody) and a ligand (e.g. antigen or Protein A). One of the interacting molecules is coupled to a solid support. The binding between antibodies and respective antigens involves a number of non-covalent interactions, including van der Waals forces, ionic bonds, hydrogen bonds and/or hydrophobic interactions. This enables elution of antibodies in a concentrated form by changing the conditions in the buffer (e.g. pH, ionic strength or polarity) used or introduction of a molecule that competes with the ligand for specific binding sites [103]. Various affinity purification methods are available, of which Fc binding and antigen-specific purification are the most frequently used, and will therefore be discussed in more detail below.

3.1.1 Fc-binding proteins

Staphylococcal Protein A (SpA), a protein found in the cell wall of Staphylococcus aureus, has been used for many years to purify antibodies, due to its natural affinity for the Fc part of different subclasses of derived from a variety of species. Coupling of SpA to a solid support enables efficient purification of IgG from crude protein samples, such as serum, ascites fluid or cell culture supernatants using affinity chromatography with the elution of bound IgG at low pH [104]. The low affinity of SpA for human IgG3 and immunoglobulins KARIN LARSSON 21 of some other species (e.g. rat) has also led to the use of other Fc-binding proteins. Among these is the streptococcal Protein G (SpG) [105]. Compared to Protein A, Protein G shows a wider ability to bind to IgGs from different species and the subclasses of human IgG [106]. The complementary binding of these two proteins inspired the development of a chimeric antibody-binding protein containing the IgG-binding domains from both Protein A and Protein G. This fusion protein possesses the additive properties of both parental proteins, thus making it an excellent choice for antibody purification [107]. Since Fc-binding proteins are unable to separate antibodies with different specificities, their use in affinity-purification is principally for the purification of monoclonal antibodies with single clonality, as opposed to polyclonal antiserum consisting of a pool of antibodies with varying epitope recognition properties.

3.1.2 Antigen-based affinity purification

Even though Fc-binding proteins like SpA are versatile purification reagents that are widely used in research facilities, both in industry and academia, other purification strategies are often preferred. As previously mentioned, Fc-binding proteins like SpA or SpG do not discriminate between antibodies with different epitope specificities, so they are used to separate the complete IgG pool from other host proteins. For the purification of polyclonal antisera containing more than 99% host-derived antibodies that are not specific for the antigen used for immunization, this is often not satisfactory. In addition, if the target protein or peptide is coupled to a fusion tag, for purification or immunogenicity purposes, there may be antibodies directed towards the fusion tag, resulting in unexpected cross-reactivity in subsequent assays [43]. Therefore, antigen-specific purification, using the target protein as an affinity ligand, is often a preferred choice when working with pAbs.

Many different purification methods are available, most of which involve the target protein being covalently coupled to a solid support. The most commonly used matrix for affinity media preparation is beaded [108]. Coupling of the ligand is enabled by derivatization of the hydroxyl group on the agarose sugar residue to appropriate activated groups. These activated groups are then covalently attached to different groups on the ligand molecule (e.g. primary amines, thiol, carboxyl or hydroxyl groups).

If the antigen used for immunization is a fusion protein containing, for example, a hexahistidine tag (His6) used for antigen purification, antibodies will be produced that are 22 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

directed towards the tag and this must be taken into consideration when planning a purification procedure. Tag-specific antibodies can be removed from the serum prior to the purification of target-specific antibodies simply by coupling the tag protein to an initial column and purifying the serum in two steps: the first aimed at removing antibodies with unwanted specificity and the second designed to elute desired target specific antibodies, i.e. monospecific antibodies (see Figure 4) [43].

#$%!&'(&)*+%, -$%!&'(&)*+%. /$%!&'(&)*+%0

!"#

Figure 4. Three antibody purification strategies. (A) In strategy 1, the column is coupled to Protein A (SpA) and all antibodies regardless of specificity are separated from other serum proteins. (B) In strategy 2, the protein used for immunization (in this case a fusion protein) is immobilized on the column and antibodies specific for the antigen (both to target and tag) will be separated from host antibodies and serum proteins. (C) In strategy 3, the antiserum is purified in two subsequent steps, the first using a column coupled with solely the fusion tag, and the second using a column coupled to the target protein. The tag-specific antibodies are eluted from the first column and the eluant is then purified on the second column, enabling elution of only the target-specific, i.e. monospecific, antibodies.

KARIN LARSSON 23

3.2 Methods for antibody characterization and their use in proteomics research

Having access to carefully validated antibodies is of great importance when carrying out proteomics research. Otherwise, interpreting data obtained from immunohistochemical analyses and immunofluorescence microscopy is difficult and false conclusions may be drawn concerning protein expression and localization. In particular, when characterizing unknown proteins, annotating tissue microarrays becomes a complex task if antibodies used for analyses are nonspecific. Ideally, two or more antibodies to each target protein should be used, to help decide which results are most reliable. Methods used in applications of antibodies are described below, and their uses are summarized in Table 2.

Method Epitope stage Aim

Protein assays (ELISA, Native (assay dependent) Relative amounts protein arrays, etc.)

Partially denatured Size determination Western blot () Identification Expression Partially denatured/native Immunohistochemistry Localization (formaldehyde/frozen) Tissue profiling Immunofluorescence Partially denatured Subcellular localization (confocal microscopy) (formaldehyde, methanol) Cell surface expression Flow cytometry Native analysis Cell identification Table 2. Summary of applications of antibodies. Modified from [1].

3.2.1 ELISA

Enzyme-linked immunosorbent assay, or ELISA, is an immunoassay for the detection and quantification of antigens or antibodies in samples, which has numerous analytical and clinical applications. The technique was first developed by a group at Stockholm University in 1971 [109] and has since been widely applied for many clinical purposes, including the development 24 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH of a tremendous number of diagnostic assays, such as HIV and allergy tests, in which antibodies in serum are detected. In ELISA antibodies are covalently linked to an enzyme to detect the analyte. After addition of a substrate the amount of analyte can be visualized by a color change or fluorescence. Several different assay designs can be used, including indirect and sandwich ELISA. In indirect ELISA (Figure 5A), wells are coated with antigen and the antibody to be analyzed is added. After subsequent washing steps and addition of a secondary enzyme-labeled antibody, the substrate to be assayed is introduced. Formerly, chromophores were used to induce a measurable color change, but now fluorophores are more commonly used. Many methods and techniques employ the principles of indirect ELISA, including Western blotting, IHC, Gyrolab and Luminex technology. Sandwich [110] have a high degree of specificity, since two antibodies to the same antigen are used (Figure 5B). The wells are coated with one antibody and the second added after the analyte, hence an antigen- antibody sandwich is formed. However, a limitation when using sandwich ELISAs in proteome-wide analyses is that it requires two antibodies to each target, while producing one suitable antibody to each target is in itself difficult to achieve.

!

-. ,012)&3)'! 4565&!(73#/'! <. ,012)&3)'! 4565&!(73#/'

! *#89:' ! ! *#89:' -#)%/'#

!

! !

!

-#)%/'#!

"#$%&'()!*+",- ,3#$;%(7!*+",-

Figure 5. (A) In indirect ELISA, a primary antigen-specific antibody is added to antigen-coated wells. An enzyme-conjugated secondary antibody specific for the primary antibody is then added, followed by a substrate producing a color change. (B) In sandwich ELISA, the wells are coated with one antibody and a second enzyme labeled antibody is added subsequent to antigen addition.

KARIN LARSSON 25

3.2.2 Western blotting

Western blotting can be used for the identification of proteins in complex mixtures, such as tissue samples and serum [111, 112]. It is also an established method for validating that antibodies bind to a target protein of the expected size, as determined from amino acid sequences. It can also be used to detect specific antibodies in serum which is useful when, for example, confirming HIV test results. In addition, posttranslational modifications like phosphorylations can be studied. In SDS-PAGE, which is the most common form of gel electrophoresis, samples are boiled in buffer containing sodium dodecyl sulfate (SDS) and β- mercaptoethanol in order to break any secondary or tertiary structures and reduce disulfide bonds respectively (although separation under native conditions can also be performed). The reduced samples (e.g. tissue homogenates or plasma) are then separated on a polyacrylamide gel according to the lengths of their polypeptide chains, i.e. molecular weights. The separated proteins are then typically transferred to a nitrocellulose or polyvinylidene fluoride (PVDF) membrane, either by capillary blotting or by applying an electric field across a gel/membrane sandwich. Nonspecific binding to the membrane is blocked, using a suitable blocking buffer, and the target protein is detected in a similar manner as in indirect ELISA. The primary antibody, either to be validated or used to detect a protein, is applied to the membrane and after sufficient washing a secondary antibody (conjugated to a chromophore or chemiluminiscent enzyme) is incubated with the membrane for detection purposes. Although it is a very versatile method for both antibody validation and protein research, there are limitations with this method. Since SDS-PAGE is usually run under denaturing conditions, the epitopes of the proteins are linear and this can be a problem if monoclonal antibodies are used for analysis. Moreover, estimations of band size are not clear-cut and shifts in apparent molecular weight are not unusual. In addition, analyses of membrane-bound proteins, such as cell-surface receptors, are difficult and often fail, due to poor solubility, (especially of integral membrane proteins) and low abundance [113].

3.2.3 Immunohistochemistry and Tissue Microarrays

Information on protein expression in tissues, at both cellular and subcellular levels, can be provided by immunohistochemical (IHC) analysis. Due to technical developments in tissue microarray preparation [44, 114], automated staining procedures and image acquisition, IHC is nowadays a high-throughput technique that is widely used in tissue protein expression profiling and proteomics research [36]. 26 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Before any immunohistochemical staining can be performed, the tissue sample must be frozen in liquid nitrogen or fixed in order to stop protein degradation and to preserve tissue morphology. Fixation can be achieved by immersion of the tissue in a cross-linking fixative, such as formaldehyde. Following fixation the tissues are dehydrated using ethanol or acetone in increasing concentrations. Finally, the tissues are washed in xylene in order to remove the alcohol and embedded in paraffin wax to create a tissue block. Thin sections of the formalin- fixed, paraffin-embedded tissue sample block may then be prepared using a microtome and IHC analysis can be applied to detect proteins that have been expressed in the sample. By combining tissue cores from different paraffin blocks a tissue microarray (TMA) can be formed, enabling high-throughput histological analysis of many samples (e.g. tissues, tumors or cells) simultaneously. Before TMAs can be used for staining with antibodies they must be conditioned. Initially the paraffin blocks must be sectioned to an appropriate size, then deparaffinized. After this, tissue sections are rehydrated using decreasing concentrations of ethanol. Endogenous peroxidase is then blocked in order to avoid interference with the enzymes used for . The proteins in the tissue sections are cross-linked by fixation, which might hide the epitopes, therefore a procedure called antigen retrieval is usually performed. This can be executed using enzymes, but is much more commonly carried out using heat (heat induced antigen retrieval; HIAR) typically in a microwave oven or pressure cooker. This procedure breaks the cross-linking, but also denatures the proteins. The sections are now ready for staining with antibodies in an indirect ELISA-like procedure using a specific primary antibody and a secondary enzyme-labeled antibody (e.g. enzyme-linked dextran polymers) to produce a color change in positively stained tissues. Commonly used enzymes include horseradish peroxidase (HRP) and alkaline phosphatase (AP). The numerous and often harsh steps involved in immunohistochemical staining procedures of tissue sections, affect proteins and their epitopes and this must be taken into consideration when choosing the antibody to use.

The TMA technique is dependent on specific, well-validated antibodies. When working with unknown proteins, this is of great importance. Unspecific antibodies can lead to positive staining even when there is no target protein present or give misleading subcellular localizations of proteins. False negative results are often caused by epitopes becoming inaccessible due to cross-linking or denaturation. This is more often a problem when using monoclonal antibodies rather than polyclonal antibodies, due to their intrinsic single epitope recognition site [36]. Benefits with TMA compared to traditional IHC analysis of tissue sections are firstly, that less material is needed, which is especially important when using rare tissue specimens, secondly, that the technique is less time-consuming and thirdly, that a semi quantitative analysis can be accomplished due to identical/similar treatment of all samples on KARIN LARSSON 27 the same slide. The major drawback is that since smaller amounts of samples are used a less representative picture of the tissue is obtained, so important information may be lost in the process.

3.2.4 Immunofluorescence microscopy

Even though immunohistochemical analysis reveals some information about subcellular localization of proteins, the output is limited. Often, discrimination between nuclear or cytoplasmic staining is achievable and in some cases membranous staining can be recognized. By using confocal microscopy, on the other hand, high-resolution images can be obtained and proteins can be localized to a wide range of sub-cellular compartments (e.g. the nucleolus, Golgi or ). The reason why the subcellular localization of proteins is of interest is to improve understanding of the functions of individual proteins within the cell. Confocal microscopy uses either live cells [115] or fixed cells [46] for the immunofluorescent detection of protein expression and localization. Live cells have the advantage that momentary protein expression patterns can be observed, whereas in fixed material proteins can only be studied at particular end-points [46]. When working with large molecules like antibodies, cells must be permeabilized in order for the molecules to penetrate the cell and find their targets, this requires cell fixation.

The most commonly used fixative is paraformaldehyde and cells are regularly permeabilized using a detergent such as Triton-X 100. Cell samples can be mounted on glass slides or prepared in glass-bottomed micro-titer plates. Following fixation and permeabilization, cells are stained with a target-specific primary antibody and fluorophore-labeled secondary antibody. Cells are usually counterstained with a nuclear probe (e.g. DAPI) and several other cellular probes (e.g. phalloidin for actin filament staining and MitoTracker for staining of mitochondria) can be used in order to help in the annotation process.

Since strong fixative agents are also used in immunofluorescent analysis, problems with denatured epitopes and poor antibody binding may arise, as previously discussed for IHC analysis. However, cross-linking of proteins occurs to a lesser extent since much shorter incubation times are required in fixatives. In addition, epitopes on membrane-bound proteins may be disturbed during permeabilization of the plasma membrane, posing a potential problem in antibody-based immunofluorescence microscopy, since this could lead to reduced 28 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH antibody recognition and false negative results. In such cases live cells might be a better alternative.

3.2.5 Flow cytometry

Flow cytometry is a valuable technique that is used to obtain information about the physical characteristics of cells such as their size, shape and, more importantly (in molecular biology), proteins expressed on their surface. The first fluorescence-based flow cytometer was constructed by Wolfgang Göhde in 1968, and since then the technique has been extensively used for a wide range of diagnostic and clinical applications. An example of the routine use of flow cytometry is in the immunophenotyping of leukemia. The subtyping, i.e. analysis of surface marker expression of lymphocytes, helps to determine the prognosis of the disease and the appropriate course of treatment [116]. Another application is the monitoring of CD4+ lymphocyte levels in HIV patients, used as an indicator of disease progression, responses to therapy and susceptibility to infections [117].

The flow cytometer analyzes cells in suspension. The cells are streamed through the flow cytometer one by one, passing through a laser light beam focused on the stream. Detectors measure scattered light and emitted light, giving information about the size and inner complexity of the cells, and cell-surface antigens, respectively. An additional feature of some flow cytometers is their ability to sort cells based on their fluorescence and light scattering properties. This is called fluorescence activated cell sorting (FACS) and it is a versatile method for the enrichment of subpopulations in complex samples that exhibit certain characteristics. By fluorescently labeling antibodies, the cell surface expression of proteins can be monitored. By using a cell line with known surface expression, flow cytometry can also be used to evaluate the ability of antibodies to bind to native proteins.

3.2.6 Epitope mapping

An epitope, also called an antigenic determinant, is the part or parts of an antigen recognized by the immune system. The corresponding part of the antibody responsible for binding to the antigen is termed a paratope. There are two types of epitopes, termed linear or structural epitopes [118]. Linear or sequential epitopes correspond to the amino acid sequence of the protein (primary protein structure) while structural or conformational epitopes are KARIN LARSSON 29 discontinuous in terms of the amino acid sequence, but are brought together in space by the natural folding of the protein (tertiary protein structure). There is, however, no clear-cut distinction between these two types of antigenic determinants. Conformational epitopes may have sequential elements in them and linear stretches of amino acid sequences may both have structural features (e.g. loops in proteins) and be discontinuous in nature (e.g. not all amino acids in the stretch may be involved in the antigen-antibody interaction) [119].

Epitope mapping is a method used to determine the minimum part of an antigen needed for antibody binding. Characterization of the antibody-binding site is desirable for research and a necessity in applications such as development. Several methods are available for mapping structural epitopes, including X-ray crystallography [120], mutagenesis of the antigen [121] and site-directed masking of epitopes [122]. Although these methods have all been shown to work well, high-throughput determination of structural epitopes is not feasible. Both mutagenesis and site-directed masking of antigens depend on structural data for making appropriate decisions about which amino acids to modify. Due to these issues, epitope- mapping efforts have mainly focused on linear epitopes, [123] even though most epitopes are thought to be structural [124]. A wide range of methods is available for mapping sequential epitopes. The most common approaches involve measuring binding of the antibody to peptide fragments. These can be generated either from the protein’s primary sequence using an overlap covering the whole sequence at single amino acid resolution [125, 126], or using peptide libraries (random or antigen-derived) expressed by phage or bacterial display, among other methods [127-130]. Although mapping of linear epitopes can be considered a high-throughput method, peptide synthesis remains expensive and efforts have been made to predict linear epitopes in silico since the beginning of the 1980s [131]. Amino acid sequences of peptides has been used to create propensity profiles, based on various characteristics such as hydrophobicity, but a recent study found that the correlation between known linear epitopes and propensity profile peaks was only slightly better than random [123]. 30 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

KARIN LARSSON 31

Present Investigation 32 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

4. Objective

The main objective of the studies this thesis is based upon was to develop methods to facilitate the high-throughput generation and subsequent purification of monospecific antibodies (Papers I and II). In addition, some of the challenges associated with generating antibodies directed towards targets with highly similar sequences and their validation were investigated (Paper III), and finally an alternative antigen design strategy was investigated for generating antibodies towards G-protein-coupled receptors (Paper IV). All of these studies were performed as part of the Human Protein Atlas program.

Before the Human Protein Atlas program was officially started, while its details were still being refined, monospecific antibodies were purified by incubating antisera with Western blot membranes containing the target protein and cutting out the areas where antibodies apparently binding to target proteins were detected, these antibodies were then extracted from the membrane by eluting them at low pH [132]. Following subsequent developments, a system with peristaltic pumps and PrESTs immobilized on affinity columns was used, enabling two antibodies to be purified per pump in a single day [43]. Soon after the project was launched, the peristaltic pumps were upgraded to ÄKTA explorer systems, resulting in a doubling in the output of purified antibodies. Even though less hands-on time was then needed per antibody, the purification rate was still not sufficient for the requirements of the HPR project. However, the introduction of the ÄKTAxpress system greatly boosted the throughput (as described in Paper I), leading to the current output of about 600 purified antibodies per month. Methods for analyzing the specificity of these antibodies have been similarly enhanced, by using protein arrays with 384 spotted PrESTs rather than low-throughput manually manufactured dot blots with four PrESTs.

An alternative immunization strategy was also developed (Paper II), in which multiplexed PrEST antigens (consisting of two or more PrESTs) were used, providing both economic and ethical benefits. The use of PrESTs as both antigens and affinity-purification ligands allows both high-throughput antigen production [41] and antibody purification, in a format that can be readily adapted to previously described methods (Paper I).

The validation of antibodies is important but not always straightforward, especially for antibodies raised against difficult targets, such as proteins with highly similar sequences to other proteins. Paper III, describing the generation of ten antibodies to such a target, Cytokeratin-17, illustrates these issues. The resulting antibodies were characterized using a KARIN LARSSON 33 combination of epitope mapping and comparative IHC analysis, in addition to the analyses generally used in HPR procedures (e.g. protein array and Western blot analyses).

In Paper IV, a novel antigen design strategy for the generation of GPCR-specific antibodies is described. The structure of GPCRs (which have an extracellular N-terminal domain, seven transmembrane helices connected by short loops and an intracellular C-terminal tail) makes them difficult to produce in the expression systems normally used for antigen production. Here, an alternative approach was investigated, involving fusion of the extracellular loops and N-terminal to create an antigen that could be used for immunization and antibody purification.

34 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

5. High-throughput generation of monospecific antibodies (Paper I)

As discussed above, high-throughput generation of affinity reagents is of great importance for proteomics research, but achieving it presents major challenges, hence specific antibodies are only currently available for a subset of the human proteome. Prior to the studies presented in this thesis, a strategy had been developed for the generation of monospecific antibodies based on conventional immunization of rabbits with recombinantly produced protein fragments, denoted Protein Epitope Signature Tags (PrESTs) followed by stringent affinity purification of the resulting polyclonal antisera [43]. In the study presented in Paper I this procedure was adapted for high-throughput applications, by applying a modular semi-automated chromatography system for the following operations: (i) depletion of unwanted tag-specific antibodies; (ii) capture of antibodies with the desired specificity and (iii) buffer exchange using gel filtration. Each module allows the automated, serial purification of up to four samples and

A. B. 1. 2. 3. (a) 216 116 87

] 48

mAU 32 [ 26

18 Absorbance

(b) (c)

Retention [ml]

Figure 6. Chromatogram and Western blot illustrating the outcome of the unit operations applied in the purification of an antibody specific for the estrogen receptor (ER-α) and subsequent characterization of isolated fractions using an MCF-7 protein lysate. (A) Chromatogram showing flow-through (a), the elution of anti-tag antibodies (b) and anti-ER-α antibodies (c). (B) Western blot membrane lanes showing detected proteins in MCF-7 cell extract: in a sample following primary incubation with unpurified antisera (panel 1), the eluate from the His6-ABP column (panel 2, corresponding to elution peak (b)) and the eluate from the PrEST-coupled column (panel 3, corresponding to elution peak (c)).

KARIN LARSSON 35 the modularity of the system allows up to 12 modules to be controlled by a single computer, thus enabling true high-throughput purification of polyclonal antisera. Monospecific antibodies (msAb) were validated with a newly developed PrEST array format as well as by Western blotting, and were then used to localize proteins in normal and diseased tissue.

In order to generate monospecific antibodies, a stringent affinity purification procedure was applied [43]. Initially, the antiserum was depleted of fusion tag-specific (anti-His6-ABP) antibodies by using an affinity column with immobilized tag protein. Eluted antibodies were discarded and the flow-through containing PrEST-specific antibodies was collected for use in subsequent purification steps. This depletion step also effectively captured most of the rabbit serum albumin [43]. In order to capture the target-specific antibodies in the depleted flow- through, affinity chromatography was applied using a matrix coupled with the same PrEST-tag fusion protein that had been used for immunization. For the second unit operation a chromatographic system, with the capacity to purify four samples per module in series was used (which also provided the possibility of increasing the number of modules operated in parallel), followed by buffer exchange. Since all antibodies towards the tag moiety of the immobilized fusion protein were depleted in the first step, only PrEST-specific antibodies were eluted with a low pH buffer and, therefore, captured by this procedure (Figure 6). Finally, the buffer was changed to a suitable storage buffer using a desalting column. In this study four modules of the chromatographic system were used, allowing 16 different antisera to be purified in less than 6 h. Antibodies in different fractions were analyzed by Western blotting against whole cell protein lysate (Figure 6). The stringency of the purification method and the specificity of antibodies obtained was assessed (for quality assurance) using a newly developed protein array method in which the specificity of the binding of acquired antibodies to the cognate PrEST was tested by examining its binding to a large set of unrelated PrESTs. The results indicated whether the PrEST-design had been valid and, most importantly, whether tag- specific antibodies had been removed (Figure 7). A dual-color dye labeling system was used for detecting PrEST-antibody interactions (and any possible cross-reactivity) and to ensure that all PrESTs were present on the array. This was accomplished by mixing tag-specific antibodies generated in hens together with the rabbit-derived PrEST-specific antibodies during the primary incubation step. A secondary incubation step then followed with anti-hen and anti- rabbit antibodies, which had been labeled with two different fluorophores. Using this procedure, 12 msAbs could be analyzed for their specificity and cross-reactivity towards 72 different PrESTs spotted on a single glass slide. This was accomplished by using an adhesive silicone mask with 12 wells, each containing 72 different PrESTs spotted in duplicate, including the PrEST corresponding to the antibody being analyzed. Ninety-six percent (446) of 36 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH the 464 affinity-purified antibodies included in this study showed satisfactory specificity in this protein array analysis.

A.

B.

C.

Figure 7. Protein array analysis of an antibody directed to the leukocyte common antigen (LCA). Red bars indicate binding to the target antigen (spotted in duplicate) and grey bars signify binding to nonrelated antigens. The green line shows the presence of spotted proteins on the glass slide. (A)

Analysis of unpurified polyclonal serum demonstrates binding of tag-specific antibodies to the His6ABP protein fused to all PrESTs. (B) Protein array analysis of depleted serum, illustrating the importance of the first purification step. (C) Analysis of PrEST-specific antibodies eluted from the last unit.

KARIN LARSSON 37

One hundred and ninety of the 446 antibodies that passed this specificity test were further assessed in terms of their ability to localize proteins in immunohistochemically stained tissue microarrays (TMAs). In total, 48 normal tissue samples were analyzed, in triplicate, with each antibody, and some of the antibodies were also tested using TMAs containing a set of samples representing some of the most common cancers. Pathologists manually annotated the TMAs, and 81% of the antibodies produced staining that appeared to be specific, as judged from staining intensity and distribution. In addition, further specificity checks were carried out for 25 of the antibodies using an adsorption assay in which they were pre-incubated with their respective PrEST-antigens, prior to (Figure 8). Immunostaining with all the tested antibodies disappeared after this procedure, further supporting the conclusion that the antibodies were specific for their respective targets. The tissue distribution patterns obtained, using a subset of the antibodies, were also compared to results obtained with clinically approved monoclonal antibodies, and there was a high degree of concordance between results.

A B

Figure 8. Results of adsorption test demonstrating that specific patterns of immunoreactivity were quenched after pre-incubation of the antibodies with the corresponding PrEST-antigen. Staining of glandular mucosa with a specific antibody to zinc finger protein 183 shows strong nuclear staining (brown) (A), which completely disappears after adsorption with PrEST (B).

In conclusion, the stringent, high-throughput purification scheme, in combination with an efficient protein array-based validation method described, enabled the systematic generation of antibodies with a high rate of success and these monospecific antibodies were shown to be highly specific in several immunoassays.

38 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

6. An alternative immunization strategy using multiplexed PrEST- antigens (Paper II)

Using traditional immunization methods the generation of both monoclonal and polyclonal antibodies on a whole proteome scale would require massive immunization programs. Therefore, from both ethical and economic perspectives, alternative immunization strategies are desirable, in order to reduce the number of animals needed for such large-scale production of antibodies, but to date there have been few reported attempts to develop suitable strategies, and most of those that have been published involve monoclonal antibody production [50, 133]. In addition to these ethical considerations, the overall potential to automate methods, as well as their complexity and cost-efficiency, are important considerations for all high- throughput projects.

Figure 9. Schematic diagram of the multiplexed immunization strategy for the generation of monospecific antibodies; from antigen preparation to affinity purification of individual antibody pools.

KARIN LARSSON 39

In this study, a multiplexed PrEST-based immunization strategy for generating polyclonal antibodies was investigated. Antigens composed of two, three, five or up to ten PrESTs in different combinations were used to immunize a total of 28 different rabbits and the resulting polyclonal antisera were purified using a modified two-step affinity purification method and the ÄKTA explorer system (Figure 9). After depleting tag-specific antibodies, PrEST-specific purification was performed by connecting affinity columns in series for loading and washing samples, then eluting individual columns at low pH, one-by-one, to obtain pure, monospecific antibodies. This method could easily be adapted for purification using the ÄKTAxpress chromatography system, by collecting the flow-through from one column and loading it onto the following columns, thus in effect individual PrEST-specific antibodies from a multiplexed immunization could be purified one at a time in an iterative procedure with increased throughput.

The concept of using multiplexed antigen was initially evaluated using a combination of two different PrESTs for immunization. The total amount of administered antigen remained constant, irrespective of the degree of multiplexing. Hence, in a double immunization the amount of each participating PrEST administered was half the amount administered in traditional, single PrEST immunizations. This strategy was necessary to meet national guidelines regarding the use of animals for scientific purposes, preventing the administration of such high doses for immunization that harmful reactions may result. In total, 13 rabbits were immunized with two different PrESTs in various combinations and in 10 of these (77%) the resultant antisera contained antibodies towards both PrESTs, while in two cases (15%) only one of the PrESTs gave rise to a detectable, specific immune response. Following one of the immunizations no specific immune response could be detected towards either of the participating PrESTs (Table 3). All of the affinity-purified antibodies were thoroughly tested using both protein arrays and Western blots against a larger set of both positive and negative control PrESTs. Any potential cross-reactivity resulting from the multiplexing was shown to have been effectively removed, resulting in pure monospecific antibodies that were comparable to corresponding antibodies obtained from traditional, single PrEST- immunizations. This was further corroborated by immunohistochemical staining of consecutive human tissue sections, which resulted in very similar immunoreactivity patterns for antibodies produced by single and multiplex immunizations.

Encouraged by these findings, the feasibility of increasing the degree of multiplexing was investigated in trials in which 28 rabbits, in total, were immunized with two, three, five or up to ten different PrESTs in various combinations. In order to accelerate the analysis of the resulting immune responses, PrEST array analysis was also used for pre-screening antisera that 40 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH had been depleted of tag-specific antibodies prior to affinity purification. The screening of depleted sera showed that 80% of the animals immunized with antigens composed of two or three PrESTs generated antibodies towards all of the administered PrESTs. Of the animals immunized with five PrESTs, 25% produced antisera recognizing all targets and 50% produced antibodies towards four out of the five PrESTs. None of the immunizations with ten PrESTs resulted in antibodies towards all the targets (Table 3).

Recognized Degree of Multiplexing PrESTs 2 PrESTs 3 PrESTs 5 PrESTs 10 PrESTs 0 1 - - - 1 2 - - - 2 10 2 - - 3 7 1 - 4 2 - 5 1 - 6 1 7 1 8 - 9 - 10 - No. immunized rabbits 13 9 4 2

Table 3. Immune responses determined by PrEST-array analysis of depleted antisera in rabbits immunized against multiplexed PrEST antigens.

Illustrative results from the analysis and purification of an antiserum resulting from one of the immunizations with five different PrESTs are shown in Figure 10. Initial screening of the immune response was performed by PrEST-array analysis of serum that had been depleted of tag-specific antibodies, which showed that antibodies towards all five PrESTs could be detected in the antisera (Figure 10A). Chromatograms obtained from samples following the subsequent purification showed that the total amounts of antibodies harvested corresponded well to intensities in PrEST-array analysis, suggesting that this method could be used as a screening tool to investigate the immune response and possibly to facilitate subsequent purification of individual antibodies (Figure 10B). Following purification, all antibodies were analyzed by Western blotting, which demonstrated that all detectable cross-reactivity resulting

KARIN LARSSON 41

A. α-HT432 α-HT1524

α-HT2323

Intensity α-HT2389

α-HT3

PrEST

α-HT432 B. α-HT1524

] α-HT2323 mAU

[ α-HT2389

Intensity α-HT3

Retention [ml]

α-HT432 α-HT2323 α-HT3 α-HT2389 α-HT1524 210 120 C. 84 60 39 28 18

M 1 2 3 4 5 6 Figure 10. Purification and analysis of an antiserum derived from a multiplex immunization with five unique PrESTs. The serum was pre-screened using a PrEST-array, followed by affinity purification and Western blot analysis. (A) Intensity plot from PrEST-array analysis of depleted sera. Bars illustrate interactions of antibodies with PrESTs, solid lines show the amounts of immobilized proteins. (B) Elution profile from an ÄKTA explorer system showing antibody purification. (C) Western blots of PrESTs in the multiplexed antigen, using the respective purified antibodies.

42 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH from the multiplexing had been effectively removed (Figure 10C). The amounts of antibodies obtained, from each PrEST, following these multiplexed immunizations were generally lower than amounts obtained from traditional single immunizations, and the antibody titers were PrEST (probably due to intermolecular antigen competition) as well as animal-dependent factors. The lower antibody yields from multiplexed immunizations were probably due to the lower amounts of each PrEST present in the multiplexed antigens, compared to immunizations with single PrEST antigens. As mentioned earlier, the total amount of protein was kept constant regardless of the number of PrESTs, hence the higher the degree of multiplexing, the lower the amount of each administered PrEST. This is probably the main reason why the success rate decreased with increases in the degree of multiplexing.

To summarize, a PrEST-based multiplex immunization strategy was developed in this study for the high-throughput generation of monospecific antibodies, providing both ethical and economic benefits due to a reduction in the numbers of animals needed. The method provides an efficient means of producing antibodies against up to three antigens simultaneously, with the possibility of further increasing multiplexing, although this would be at the expense of antibody yields. It would also result in a more complex purification process, lowering the throughput, although this could be partially overcome by recent technical improvements, e.g. use of the ÄKTAxpress purification system. Importantly, the antibodies obtained from these immunizations against multiplexed proteins showed excellent specificity in a number of assays (protein array analyses and Western blots), they also compared well with the corresponding antibodies obtained from traditional immunization procedures in immunohistochemical analyses of human tissue sections.

KARIN LARSSON 43

7. Generation and validation strategies for antibodies raised towards targets with highly similar sequences (Paper III)

In affinity-based proteomics the aim should be to generate at least one specific binder for each non-redundant protein of the human proteome, defined here as a protein representing each gene locus. Families of proteins exhibiting high degrees of similarity in their amino acid sequences can complicate this endeavor. Such similarities make it difficult to find unique regions that are appropriate for use as antigens and result in strict requirements for the validation of the generated antibodies. Here, Cytokeratin-17, a member of the intermediate filament family of proteins, which show a high degree of sequence similarity and play important roles in cell morphology, was chosen as a model for two main reasons. Firstly, to illustrate the challenges faced in raising antibodies to such proteins, and secondly to study batch-to-batch variations in antibodies produced from repeated immunizations to the same target. Two non-overlapping PrESTs were designed, expressed, and each used to immunize five rabbits. The resulting antibodies were purified and analyzed using protein arrays, Western blotting, immunofluorescence-based confocal microscopy and immunohistochemical staining of consecutive tissue sections. In addition, to characterize the antibodies obtained further, epitope mapping was performed using both the Gyrolab platform, which incorporates microfluidic structures in a CD-format for analyzing proteins at the nanoliter level, and a bead- based Luminex suspension array, which allows true multiplexing of the assays.

There are about 130 genes containing InterPro keratin domains (defining the intermediate filament family and the cytokeratins) in the Ensembl database [134]. Hence, the standard PrEST-design criteria; a threshold value of 60% identical amino acids over a 50 amino-acid window and no local sequence with eight out of ten amino acids identical to each other, were difficult to fulfill. This is illustrated by a bioinformatics analysis, which found >90% identity in domain and epitope-sized sequences between Cytokeratin-17 and other members of the family across its whole primary sequence (Figure 11A) [32]. However, if two genes predicted as paralogues to Cytokeratin-17 (ENSG00000128422 and ENSG00000131885) were excluded from the analysis the antigen design was considerably eased (Figure 11B). When all genes containing an InterPro keratin domain were excluded (Figure 11C) the sequence identity plot showed an identity well below 60% over the whole protein, for the 50-aa window. 44 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

A.

B.

C.

Figure 11. PrEST location and sequence identity plots including domain and epitope sized (50 and 10 amino acids respectively) sequence similarity comparisons of Cytokeratin-17 to all human genes (A), to all genes except two genes predicted as paralogs to Cytokeratin-17 (B) and to all non Inter Pro keratin domain containing proteins (C).

The two Cytokeratin-17 PrESTs, denoted here as PrEST-A and PrEST-B, were recombinantly produced in Escherichia coli as His6ABP-fusions, then each of them was used to immunize five New Zealand white rabbits. Following bleeding and serum collection, monospecific antibodies were obtained using the standard two-step immunoaffinity-based protocol described in Paper I. Initial specificity analysis was performed using planar protein arrays, which showed specific binding to the respective antigens, with no detectable background binding or other unwanted KARIN LARSSON 45 cross-reactivity. Further analysis by Western blotting showed that all of the antibodies were able to recognize a protein of the size corresponding to Cytokeratin-17 (48kDa), with little variation within or between the respective PrEST-groups.

PrEST-A PrEST-B AbA1 AbB1

AbA2 AbB2

AbA3 AbB3

AbA4 AbB4

AbA5 AbB5

Figure 12. Epitope mapping results from affinity-purified antibodies generated towards PrEST-A and PrEST-B of Cytokeratin-17. The binding intensity was measured as total integrated volume (TIV) towards each synthetic peptide. (A) Results obtained with antibodies AbA1-AbA5 generated using PrEST-A as antigen. (B) Results obtained using PrEST-B specific antibodies (AbB1-AbB5).

A thorough analysis of linear epitopes was performed for all antibodies using synthetic overlapping peptides (15 amino acids long with a 10 amino acid overlap) with two different systems, the Gyrolab platform and the Luminex. Mapping was carried out to assess differences between immune responses when different animals were immunized with the same PrEST, and to account for observed discrepancies in results of Western and immunohistochemical analyses. Using both systems, mapping showed that antibodies bound principally to the C- terminal peptides of PrEST-A and PrEST-B, possibly due to their flexibility and exposure to the immune system (in contrast to their N-termini, which are fused to the rather large His6- ABP fusion tag). This was especially evident for the five antibodies generated against PrEST-B, for which almost all the detectable binding was to the last six C-terminal peptides, with the 46 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH exception of a few outliers. A subset of the PrEST-A antibodies exhibited additional binding to the N-terminus (Figure 12).

To investigate how observed differences in epitope recognition between antibodies might influence the results when used to stain samples of human tissue, they were applied in immunohistochemical analyses of consecutive tissue sections cut from tissue microarray blocks of 48 tissue samples with normal morphology. The results, summarized in Figure 13, showed that all PrEST-A antibodies apart from one (AbA2) yielded similar staining patterns, with strong staining of a number of epithelial tissues, some of which was expected and some unexpected according to the literature [135, 136]. The PrEST-A antibody, AbA2, yielded a Cytokeratin-17-like staining pattern and four of the PrEST-B antibodies (AbB1-AbB4) exhibited this pattern of staining even more strongly. AbB5 stained most epithelial tissues intensely (Figure 13). When the PrEST sequences and their similarity to other cytokeratins were analyzed in more detail, it was found that epitopes recognized by AbA1 and AbA3-AbA5 were situated in a region of PrEST-A that shared 24 adjacent amino acids in common with Cytokeratin-19, for example. This is a cytokeratin that is widely distributed in simple

!"#$%&' !"#$%&( ')'* ')'+ ')', ')'- ')'. ')(* ')(+ ')(, ')(- ')(. ';"815=32=51; ("854/ ("01>?@4 A<$ #1;0:8/"6@: F5==0E6513/@)8 G54/"061/84/615=3/"5>/ H6;18C I6B8" I@12 !51>"854 !"04/5/8 $5=6B5"C32=51; $D@5:0@438E6/?8=65 %?C"06;32=51; J"615"C3)=5;;8"

$/"01234/561612 7859&:0;8"5/834/561612 <034/561612

Figure 13. Tissue or cell type distribution of immunohistochemical staining obtained with ten antibodies generated towards two Cytokeratin-17 PrESTs. Red spots represent strong staining, orange spots weak to moderate staining and white spots no detectable staining.

KARIN LARSSON 47 epithelium [135]. Antibody AbA2, which only bound to the last peptide (A23), only shared three out of six amino acids with Cytokeratin-19, hence generating a more Cytokeratin-17-like staining pattern in the TMA. Interestingly, the C-terminal part of PrEST-B, where most of the PrEST-B antibody binding was observed, does not share any identical sequence with other cytokeratins, thus explaining the more Cytokeratin-17-like staining pattern obtained for these antibodies. All antibodies were further analyzed using immunofluorescent confocal microscopy with a Cytokeratin-17 positive cell line, A-431. All PrEST-B antibodies yielded the expected sub-cellular localization pattern of immunoreactivity with distinctive intermediate filament staining, thus further validating the Western blot and IHC findings. For PrEST-A antibodies results were less supportive since more diffuse cytoskeletal staining was observed, and in one case clear nuclear staining was evident.

In conclusion, the possibility of generating antibodies towards targets which have highly similar sequences to other proteins was investigated, and a validation strategy was demonstrated, based on Western blotting, IF and comparative IHC of several binders to the same target combined with epitope mapping. The results showed that antibodies (including most of the PrEST-B antibodies) binding to the most sequence-unique part of Cytokeratin-17, the C-terminal region, also exhibited the most Cytokeratin-17-like staining pattern in IHC analyses. However, PrEST-A antibodies displayed additional binding, possibly to other cytokeratins in the analyzed tissues, in accordance with the higher degree of sequence similarity between PrEST-A and other members of the cytokeratin family observed in sequence analyses. Furthermore, the binding patterns (with relatively few distinct epitopes) of antibodies generated by the immunization of different animals with the same antigen were similar, although not identical. These data support the strategy of producing antigens with little sequence similarity to other proteins as an important criterion. They also suggest that for affinity-based proteomics projects use of at least two, and preferably more, binders raised to non-overlapping regions of the same target protein is desirable.

48 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

8. An alternative strategy for designing antigens for membrane-bound receptors (Paper IV)

G-protein-coupled receptors (GPCRs) constitute a large family of important cell-surface receptors. These are proteins with seven transmembrane domains that play key roles in regulating cellular functions via interactions with a wide range of ligands (e.g. neurotransmitters and hormones). These regulators of cell functions include a great number of important drug-binding sites, accounting for 60-70% of current pharmaceutical research and development targets. Well-validated antibodies towards GPCRs are potentially important tools which can be used for the discovery of novel drugs and for research aimed at increasing understanding of these endogenous receptors [137]. In addition to the seven transmembrane

!"

#"

,"-'./012

!"#$%&'()***)+

Figure 14. Schematic representation of a G-protein-coupled receptor and the antigen used for antibody generation. (A) GPCRs consist of a seven transmembrane domain core connected by loops, an extracellular N-terminus and an intracellular C-terminus. (B) The extracellular loops and N-terminus are assembled into a single antigen separated by a polar linker, and fused to His6-ABP.

KARIN LARSSON 49 domains, all GPCRs contain three extracellular loops and an N-terminal tail, which are responsible for binding their respective ligands, as well as three intracellular loops and a C- terminus responsible for intracellular signaling (Figure 14A). In this study the possibility of producing GPCR-specific antibodies by combining the N-terminus and extracellular loops to form a single antigen was investigated.

To test this approach, three GPCRs were selected: Neuropeptide Y type 5 receptor (NPY5R), beta 2 adrenergic receptor (B2AR) and the glucagon-like peptide 1 receptor (GLP1R). In each case, using solid-phase gene assembly [138, 139] the extracellular loops and the N-terminus were cloned to form a single antigen separated by a polar linker (GSSSG), which was then fused to a hexahistidine and an albumin binding protein (His6-ABP) tag, and expressed in E. coli (Figure 14B). The resulting antigens were each used to immunize two rabbits, and stringent affinity-purification of the resulting antibodies was performed using previously described methods (Paper I). Since GPCRs are generally expressed at very low levels [140], cell lines that over-expressed the receptors were generated with a lentiviral gene delivery system. Flow cytometry and immunofluorescence-based confocal microscopy were then applied to the over- expressing cell lines to characterize the binding of the antibodies obtained to intact receptors. Epitope mapping using synthetic peptides corresponding to the extracellular loops and the N- terminal sequence of the three GPCRs was used to characterize antibody-antigen interactions.

Once stable expression in the A-431 cells had been established, all the antibodies generated, as well as a number of commercial antibodies, were analyzed using a flow cytometer. All the antibodies towards NPY5R (Figure 15A) and GLP1R (Figure 15C) produced an increase in fluorescence, indicative of binding to their cognate full-length receptor, but only one of the B2AR-specific antibodies (B2AR:1) showed fluorescence suggestive of specific binding (Figure 15B). Only one of the antibodies (the commercial antibody, MAB28141) raised against GLP1R, exhibited nonspecific binding to control A-431 cells. Immunofluorescence analysis was carried out for further validation of binding to full-length proteins and to assure specificity; staining to cell compartments other than the membrane or membrane-associated compartments would suggest that antibodies are not sufficiently specific. In general, the analyzed antibodies produced the expected pattern of plasma membrane staining, which was more pronounced at the cell junctions. Additional staining in the Golgi could be observed with all the NPY5R antibodies, and the only B2AR antibody (B2AR:1) that generated any staining also stained this compartment. Since the Golgi apparatus modifies proteins before they are transported to the plasma membrane this was to be expected. The GLP1R antibodies also weakly stained the nuclear membrane. NPY5R:1 and the commercial anti-GLP1R antibody, GA1940, stained the nuclei of normal A-431 cells nonspecifically, but not the nuclei of cell 50 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH over-expressing the respective receptors. Epitope mapping did not reveal any general pattern of immunoreactivity. The results showed that antibody NPY5R:2 bound strongly to the N- terminal region of NPY5R, while antibody NPY5R:1 bound to most parts of the antigen with lower signal intensities. Both B2AR antibodies showed binding with high signal intensities:

!" &'()*+! &'()*+# ,'-"!$./" ,'*0#"/"!!%

!""

" ! # $ % ! # $ % ! # $ % ! # $ % !" !" !" !" !" !" !" !" !" !" !" !" !" !" !" !"

#" 1#-*+! 1#-*+# 23!$$""

!#"

" ! # $ % ! # $ % ! # $ % !" !" !" !" !" !" !" !" !" !" !" !" E=99A?;:@C

$" 45'!*+! 45'!*+# -1/%$$

!""

" ! # $ % ! # $ % ! # $ % !" !" !" !" !" !" !" !" !" !" !" !" 4-!/%" 6-1#7!%!

!"" F=?;@G2 '=?;@G2 '=?;@G2>B@OA-I%$!A?=99>

" ! # $ % ! # $ % !" !" !" !" !" !" !" !" 89:;<=>?=@?=AB@C=@>BCD

Figure 15. Flow cytometric analysis of antibodies specific for NPY5R (A), B2AR (B) and GLP1R (C). X-axes indicate cell counts and y-axes fluorescence intensity. Each histogram shows data obtained from three samples: wild A-431 cells incubated solely with secondary antibody (thin lines), A-431 cells incubated with both primary and secondary antibodies (dashed lines) and over-expressing cell lines immunostained with the respective primary antibody and a secondary antibody.

KARIN LARSSON 51

B2AR:1 to the N-terminus and second extracellular loop, and B2AR:2 to the N-terminus as well as to both the second and third extracellular loops. Interestingly, B2AR:2 bound with very strong signal intensity in the Luminex assay used for epitope mapping, but failed to generate any detectable staining in immunofluorescence assays or flow cytometry analyses. Due to the long (120 amino acids) N-terminal tail of GLP1R no peptides were synthesized to represent this part of the antigen, so no possible binding to this receptor region could be determined in these experiments. GLP1R:1 and GLP1R:2 showed binding to the first and third extracellular loops, respectively, but with weak intensity of detected signals in the former case.

Figure 16. Immunofluorescence micrograph taken from the dorsal part of the periaqueductal gray in the human mesencephalon. Unfixed cryostat sections incubated with the antibody NPY5N2 and stained with green fluorescein isothiocyanate (FITC). Bar indicates µm.

In addition, some of the antibodies were used for in situ staining of human tissue. The antibody NPY5R:2 were used for immunofluorescence staining of unfixed cryostat sections of human brain tissue. A large number of fluorescent processes were seen in the dorsal part of the periaqueductal gray in the human mesencephalon, staining that was lost after pre-incubation with the antigen used for immunization, thus indicating specific staining of the antibody to the target receptor. No cell bodies were visible (Figure 16). Moreover, the antibodies generated towards GLP1R were used for IHC staining of formalin-fixed paraffin embedded normal human pancreas. The GLP1R:1 antibody exhibited positive staining of the plasma membrane 52 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH in islets and no staining in exocrine tissue (Figure 17). GLP1R:2 also stained the plasma membrane of islet cells but with additional weak cytoplasmic staining in exocrine cells. Neither of the antibodies GA1940, AB9433 or MAB28141 gave rise to staining in the plasma membrane, but rather weak to strong cytoplasmic staining both in islet and exocrine cells. AB9433 and MAB28141 also exhibited nuclear staining in islet cells as well as exocrine cells.

Figure 17. Immunohistochemical staining of normal human pancreas using an antibody generated towards GPL1R (GLP1R:1). A moderate staining of the plasma membrane of islet cells was shown (depicted in brown), but no staining in exocrine tissue.

In summary, antibodies towards G-protein-coupled receptors are invaluable tools for studies of their sub-cellular localization, posttranslational modifications, interaction partners and molecular identity [141]. The paper describes a novel antigen design strategy for the straightforward generation of GPCR-specific antibodies. Antigen production is considerably eased when the N-terminus and extracellular loops are combined to form single antigens rather than using whole receptor molecules. In addition, use of an antigen with several receptor domains (rather than a single synthetic peptide) increases the likelihood of inducing an immune response. The results showed that the antibodies generated in this manner work well across several technical platforms, and that they recognize both denatured and native targets in over expressing cell lines and in human tissue samples. KARIN LARSSON 53

9. Concluding remarks

Antibodies are invaluable tools for use in proteomics research, since they exhibit both remarkable variability and specificity for their target proteins. The systematic generation of specific, well-validated antibodies to gain complete coverage of the proteome and hence facilitate increased understanding of protein expression, localization and function, is an important challenge for scientists. The main objectives of the studies presented in this thesis were to develop methods facilitating the high-throughput generation and subsequent purification of monospecific antibodies, and to address problems associated with validating antibodies.

Paper I describes a high-throughput approach for proteome-wide generation of monospecific antibodies (to non-redundant proteins) and their validation with a newly developed protein array assay. The antibodies generated were used for parallel profiling of protein expression and sub-cellular localization. The data obtained suggest that monospecific antibodies could be used to generate a comprehensive atlas of the human proteome. Today, around 600 antibodies are purified each month within the Protein Atlas program and a first draft of the human proteome is expected in 2014. More than 90% of human genes are currently in the HPR pipeline. The biggest current challenges are to generate antigens to gene products for which previous attempts have failed somewhere in the process, and to design antigens to targets that are highly similar to other proteins, and to targets with short sequence stretches that are suitable for PrEST generation.

An assessment of an alternative immunization strategy using multiplexed PrESTs antigens (for ethical and economic reasons) is described in paper II. It was shown that it is feasible to generate antibodies towards two to three different protein targets in a single animal and, furthermore, that these antibodies could be separated in subsequent purification steps, without compromising quality or throughput. It would be interesting to investigate the PrESTs that failed to evoke an immune response further by using them in different combinations for immunizations, to see if this resulted in different patterns of immunoreactivity. Antibody titers could also be investigated and the possibility that there are immunodominant PrESTs (and, if so, their amino acid sequences). In addition, immunizations using the same amount of each PrEST as used in single immunizations would be of interest, to assess whether the low antibody titers observed in this study were, in fact, a consequence of lower antigen concentrations. 54 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Proteins with highly similar sequences to other proteins, related or unrelated, pose problems, both for the high-throughput generation of antibodies and their subsequent validation. Cytokeratin-17 belongs to the intermediate filament family, a group with closely related members, and was therefore chosen as a target in the studies described in Paper III. In these studies ten antibodies towards two non-overlapping PrESTs were produced and characterized. Epitope mapping revealed an immune preference for the C-terminal and the more flexible part of both PrESTs, especially for the PrEST designed from the C-terminus of Cytokeratin-17. These findings explain observed differences in the results of comparative immunohistochemical analyses with these antibodies. The C-terminus, corresponding to one of the PrESTs, is the most unique part of Cytokeratin-17 and the antibodies that bound principally to this region showed the most Cytokeratin-17-like staining patterns in IHC. Robust antigen design is, therefore, of utmost importance and the strategy of using regions with the least sequence similarity to other proteins as the main criterion is justified. For Cytokeratin-17, immunization with shorter peptides designed from the C-terminus would probably improve results. Alternatively, subjecting antibodies to a second round of purification with the C-terminal peptide as an affinity ligand could be attempted, to see if the results obtained in subsequent analyses change.

G-protein-coupled receptors are important targets for the development of new drugs since they regulate numerous cellular functions. In Paper IV, a novel antigen design strategy for generating specific antibodies to three different GPCRs is described and evaluated. The extracellular loops and the N-terminus were combined into a single antigen and the resulting antibodies showed specificity for their respective GPCR targets in cell lines over-expressing them. This was assessed by several techniques, including flow cytometric analysis and immunofluorescent staining. Two of the antibodies also produced positive staining of tissues expected to express the respective targets: human brain tissue (NPY5R) and pancreatic islets (GLP1R). This approach to the generation of antibodies towards GPCRs should also be applicable to other receptors (both nuclear and plasma membrane-associated) for which antigen design is difficult since they have short stretches of exposed amino acid residues. The protocol could also be adapted to purification processes that are already available by using fusion tag proteins other than His6-ABP.

Large numbers of antibodies and other binders are being generated in various proteomics initiatives, and validation of these affinity reagents is, as stated earlier, an important but difficult task. Widely used techniques, such as immunohistochemistry, have drawbacks in terms of standardization and reproducibility [142]. In order to ascertain that the results obtained in IHC are accurate, the use of several binders to non-overlapping regions of the KARIN LARSSON 55 protein under investigation is helpful. However, there are many proteins for which not even one specific binder is available, so obtaining two or more will consequently be difficult. Therefore, the use of other, complementary methods would be beneficial to the validation process. One alternative approach could be to use knockout mice, if they are available, and if the binder investigated is functional in mice. Other options, applicable to a more vide range of targets, include RNA interference (RNAi) of expression, following which a reduction of signal in immunofluorescence or Western blot analyses would provide strong evidence that the applied antibodies are specific for their respective targets. Using an arsenal of well- characterized antibodies, proteomics research will ultimately be able to increase our understanding of protein function and biological processes substantially.

56 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

KARIN LARSSON 57

References

1. Uhlen, M. and S. Hober, Generation and validation of affinity reagents on a proteome-wide level. J Mol Recognit, 2009. 22(2): p. 57-64. 2. Holliger, P. and P.J. Hudson, Engineered antibody fragments and the rise of single domains. Nat Biotechnol, 2005. 23(9): p. 1126-36. 3. Hudson, P.J. and C. Souriau, Engineered antibodies. Nat Med, 2003. 9(1): p. 129-34. 4. Lander, E.S., et al., Initial sequencing and analysis of the human genome. Nature, 2001. 409(6822): p. 860-921. 5. Venter, J.C., et al., The sequence of the human genome. Science, 2001. 291(5507): p. 1304- 51. 6. Lee, R.C., R.L. Feinbaum, and V. Ambros, The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell, 1993. 75(5): p. 843-54. 7. Easton, D.F., et al., Genome-wide association study identifies novel breast cancer susceptibility loci. Nature, 2007. 447(7148): p. 1087-93. 8. Scott, L.J., et al., A genome-wide association study of type 2 diabetes in Finns detects multiple susceptibility variants. Science, 2007. 316(5829): p. 1341-5. 9. Drews, J., Drug discovery: a historical perspective. Science, 2000. 287(5460): p. 1960-4. 10. Wilkins, M.R., et al., Progress with proteome projects: why all proteins expressed by a genome should be identified and how to do it. Biotechnol Genet Eng Rev, 1996. 13: p. 19-50. 11. Anderson, N.L. and N.G. Anderson, Proteome and proteomics: new technologies, new concepts, and new words. Electrophoresis, 1998. 19(11): p. 1853-61. 12. Clamp, M., et al., Distinguishing protein-coding and noncoding genes in the human genome. Proc Natl Acad Sci U S A, 2007. 104(49): p. 19428-33. 13. Hubbard, T.J., et al., Ensembl 2009. Nucleic Acids Res, 2009. 37(Database issue): p. D690-7. 14. Uhlen, M. and F. Ponten, Antibody-based proteomics for human tissue profiling. Mol Cell Proteomics, 2005. 4(4): p. 384-93. 15. Anderson, N.L. and N.G. Anderson, The human plasma proteome: history, character, and diagnostic prospects. Mol Cell Proteomics, 2002. 1(11): p. 845-67. 16. Klose, J., Protein mapping by combined isoelectric focusing and electrophoresis of mouse tissues. A novel approach to testing for induced point mutations in mammals. Humangenetik, 1975. 26(3): p. 231-43. 17. O'Farrell, P.H., High resolution two-dimensional electrophoresis of proteins. J Biol Chem, 1975. 250(10): p. 4007-21. 18. Scheele, G.A., Two-dimensional gel analysis of soluble proteins. Charaterization of guinea pig exocrine pancreatic proteins. J Biol Chem, 1975. 250(14): p. 5375-85. 19. Gorg, A., et al., The current state of two-dimensional electrophoresis with immobilized pH gradients. Electrophoresis, 2000. 21(6): p. 1037-53. 20. Gorg, A., W. Postel, and S. Gunther, The current state of two-dimensional electrophoresis with immobilized pH gradients. Electrophoresis, 1988. 9(9): p. 531-46. 21. Gorg, A., W. Weiss, and M.J. Dunn, Current two-dimensional electrophoresis technology for proteomics. Proteomics, 2004. 4(12): p. 3665-85. 58 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

22. Falk, R., et al., Approaches for systematic proteome exploration. Biomol Eng, 2007. 24(2): p. 155-68. 23. Mann, M., R.C. Hendrickson, and A. Pandey, Analysis of proteins and proteomes by mass spectrometry. Annu Rev Biochem, 2001. 70: p. 437-73. 24. Henzel, W.J., et al., Identifying proteins from two-dimensional gels by molecular mass searching of peptide fragments in protein sequence databases. Proc Natl Acad Sci U S A, 1993. 90(11): p. 5011-5. 25. Patterson, S.D. and R. Aebersold, Mass spectrometric approaches for the identification of gel- separated proteins. Electrophoresis, 1995. 16(10): p. 1791-814. 26. Yates, J.R., 3rd, et al., Peptide mass maps: a highly informative approach to protein identification. Anal Biochem, 1993. 214(2): p. 397-408. 27. Yamashita, M. and J. Fenn, Electrospray ion source. Another variation on the free-jet theme. The Journal of Physical Chemistry, 1984. 88(20): p. 4451-4459. 28. Karas, M. and F. Hillenkamp, Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons. Anal Chem, 1988. 60(20): p. 2299-301. 29. MacCoss, M.J., et al., Shotgun identification of protein modifications from protein complexes and lens tissue. Proc Natl Acad Sci U S A, 2002. 99(12): p. 7900-5. 30. Rigaut, G., et al., A generic protein purification method for protein complex characterization and proteome exploration. Nat Biotechnol, 1999. 17(10): p. 1030-2. 31. Taussig, M.J., et al., ProteomeBinders: planning a European resource of affinity reagents for analysis of the human proteome. Nat Methods, 2007. 4(1): p. 13-7. 32. Berglund, L., et al., A whole-genome bioinformatics approach to selection of antigens for systematic antibody generation. Proteomics, 2008. 8(14): p. 2832-9. 33. Lindskog, M., et al., Selection of protein epitopes for antibody production. Biotechniques, 2005. 38(5): p. 723-7. 34. Anderson, L. and J. Seilhamer, A comparison of selected mRNA and protein abundances in human liver. Electrophoresis, 1997. 18(3-4): p. 533-7. 35. Gry, M., et al., Correlations between RNA and protein expression profiles in 23 human cell lines. BMC Genomics, 2009. 10: p. 365. 36. Warford, A., W. Howat, and J. McCafferty, Expression profiling by high-throughput immunohistochemistry. J Immunol Methods, 2004. 290(1-2): p. 81-92. 37. Warford, A., et al., Assessing the potential of immunohistochemistry for systematic gene expression profiling. J Immunol Methods, 2007. 318(1-2): p. 125-37. 38. Uhlen, M., et al., A human protein atlas for normal and cancer tissues based on antibody proteomics. Mol Cell Proteomics, 2005. 4(12): p. 1920-32. 39. Agaton, C., et al., Affinity proteomics for systematic protein profiling of chromosome 21 gene products in human tissues. Mol Cell Proteomics, 2003. 2(6): p. 405-14. 40. Larsson, M., et al., High-throughput protein expression of cDNA products as a tool in functional genomics. J Biotechnol, 2000. 80(2): p. 143-57. 41. Steen, J., et al., High-throughput protein purification using an automated set-up for high-yeild affinity chromatography. Protein Expr Purif., 2006. 46(2): p. 173-8. 42. Porath, J., et al., Metal chelate affinity chromatography, a new approach to protein fractionation. Nature, 1975. 258(5536): p. 598-9. 43. Agaton, C., et al., Selective enrichment of monospecific polyclonal antibodies for antibody-based proteomics efforts. J Chromatogr A, 2004. 1043(1): p. 33-40. KARIN LARSSON 59

44. Kampf, C., et al., Antibody-based tissue profiling as a tool in clinical proteomics. Clin. Proteomics, 2004. 1(3-4): p. 285-300. 45. Strömberg, S., et al., A high-throughput strategy for protein profiling in cell microarrays using automated image analysis. Proteomics, 2007. 7(13): p. 2142-50. 46. Barbe, L., et al., Toward a confocal subcellular atlas of the human proteome. Mol Cell Proteomics, 2008. 7(3): p. 499-508. 47. Schofield, D.J., et al., Application of phage display to high throughput antibody generation and characterization. Genome Biol, 2007. 8(11): p. R254. 48. Dyson, M.R., et al., Production of soluble mammalian proteins in Escherichia coli: identification of protein features that correlate with successful expression. BMC Biotechnol, 2004. 4: p. 32. 49. Durocher, Y., S. Perret, and A. Kamen, High-level and high-throughput recombinant by transient transfection of suspension-growing human 293-EBNA1 cells. Nucleic Acids Res, 2002. 30(2): p. E9. 50. Ning, Y., et al., An alternative strategy for high throughput generation and characterization of monoclonal antibodies against human plasma proteins using fractionated native proteins as immunogens. Proteomics, 2006. 6(2): p. 438-48. 51. Haab, B.B., et al., A reagent resource to identify proteins and peptides of interest for the cancer community: a workshop report. Mol Cell Proteomics, 2006. 5(10): p. 1996-2007. 52. Uhlen, M., Mapping the human proteome using antibodies. Mol Cell Proteomics, 2007. 6(8): p. 1455-6. 53. Gloriam, D.E., et al., Report: A community standard format for the representation of protein affinity reagents. Mol Cell Proteomics, 2009. 54. McCafferty, J., et al., Phage antibodies: filamentous phage displaying antibody variable domains. Nature, 1990. 348(6301): p. 552-4. 55. Goldsby, R., et al., Immunology. 5th ed. 2003: W.H. Freeman and Company. 56. Li, Z., et al., The generation of antibody diversity through somatic hypermutation and class switch recombination. Genes Dev, 2004. 18(1): p. 1-11. 57. Tonegawa, S., Somatic generation of antibody diversity. Nature, 1983. 302(5909): p. 575-81. 58. Shi, S.R., R.J. Cote, and C.R. Taylor, Antigen retrieval immunohistochemistry: past, present, and future. J Histochem Cytochem, 1997. 45(3): p. 327-43. 59. Sompuram, S.R., et al., A molecular mechanism of formalin fixation and antigen retrieval. Am J Clin Pathol, 2004. 121(2): p. 190-9. 60. Keller, M.A. and E.R. Stiehm, Passive immunity in prevention and treatment of infectious diseases. Clin Microbiol Rev, 2000. 13(4): p. 602-14. 61. Newcombe, C. and A.R. Newcombe, Antibody production: polyclonal-derived biotherapeutics. J Chromatogr B Analyt Technol Biomed Life Sci, 2007. 848(1): p. 2-7. 62. Kohler, G. and C. Milstein, Continuous cultures of fused cells secreting antibody of predefined specificity. Nature, 1975. 256(5517): p. 495-7. 63. Little, M., et al., Of mice and men: hybridoma and recombinant antibodies. Immunol Today, 2000. 21(8): p. 364-70. 64. Borrebaeck, C.A., Antibodies in diagnostics - from immunoassays to protein chips. Immunol Today, 2000. 21(8): p. 379-82. 65. Kipriyanov, S.M. and M. Little, Generation of recombinant antibodies. Mol Biotechnol, 1999. 12(2): p. 173-201. 66. Khazaeli, M.B., R.M. Conry, and A.F. LoBuglio, Human immune response to monoclonal antibodies. J Immunother Emphasis Tumor Immunol, 1994. 15(1): p. 42-52. 60 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

67. Carson, D.A. and B.D. Freimark, Human lymphocyte hybridomas and monoclonal antibodies. Adv Immunol, 1986. 38: p. 275-311. 68. Morrison, S.L., et al., Chimeric human antibody molecules: mouse antigen-binding domains with human constant region domains. Proc Natl Acad Sci U S A, 1984. 81(21): p. 6851-5. 69. Boulianne, G.L., N. Hozumi, and M.J. Shulman, Production of functional chimaeric mouse/human antibody. Nature, 1984. 312(5995): p. 643-6. 70. Jaffers, G.J., et al., Monoclonal antibody therapy. Anti-idiotypic and non-anti-idiotypic antibodies to OKT3 arising despite intense immunosuppression. Transplantation, 1986. 41(5): p. 572-8. 71. Riechmann, L., et al., Reshaping human antibodies for therapy. Nature, 1988. 332(6162): p. 323-7. 72. Jones, P.T., et al., Replacing the complementarity-determining regions in a human antibody with those from a mouse. Nature, 1986. 321(6069): p. 522-5. 73. Verhoeyen, M., C. Milstein, and G. Winter, Reshaping human antibodies: grafting an antilysozyme activity. Science, 1988. 239(4847): p. 1534-6. 74. Nuttall, S.D., et al., Selection and affinity maturation of IgNAR variable domains targeting Plasmodium falciparum AMA1. Proteins, 2004. 55(1): p. 187-97. 75. King, D.J., et al., Improved tumor targeting with chemically cross-linked recombinant antibody fragments. Cancer Res, 1994. 54(23): p. 6176-85. 76. Holliger, P., T. Prospero, and G. Winter, "Diabodies": small bivalent and bispecific antibody fragments. Proc Natl Acad Sci U S A, 1993. 90(14): p. 6444-8. 77. Iliades, P., A.A. Kortt, and P.J. Hudson, Triabodies: single chain Fv fragments without a linker form trivalent trimers. FEBS Lett, 1997. 409(3): p. 437-41. 78. Le Gall, F., et al., Di-, tri- and tetrameric single chain Fv antibody fragments against human CD19: effect of valency on cell binding. FEBS Lett, 1999. 453(1-2): p. 164-8. 79. Rodrigues, M.L., et al., Engineering a humanized bispecific F(ab')2 fragment for improved binding to T cells. Int J Cancer Suppl, 1992. 7: p. 45-50. 80. Tutt, A., G.T. Stevenson, and M.J. Glennie, Trispecific F(ab')3 derivatives that use cooperative signaling via the TCR/CD3 complex and CD2 to activate and redirect resting cytotoxic T cells. J Immunol, 1991. 147(1): p. 60-9. 81. Nygren, P.A. and A. Skerra, Binding proteins from alternative scaffolds. J Immunol Methods, 2004. 290(1-2): p. 3-28. 82. Smith, G.P. and V.A. Petrenko, Phage Display. Chem Rev, 1997. 97(2): p. 391-410. 83. Smith, G.P., Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science, 1985. 228(4705): p. 1315-7. 84. Daugherty, P.S., Protein engineering with bacterial display. Curr Opin Struct Biol, 2007. 17(4): p. 474-80. 85. Lee, S.Y., J.H. Choi, and Z. Xu, Microbial cell-surface display. Trends Biotechnol, 2003. 21(1): p. 45-52. 86. Hanes, J. and A. Pluckthun, In vitro selection and evolution of functional proteins by using . Proc Natl Acad Sci U S A, 1997. 94(10): p. 4937-42. 87. Binz, H.K., P. Amstutz, and A. Pluckthun, Engineering novel binding proteins from nonimmunoglobulin domains. Nat Biotechnol, 2005. 23(10): p. 1257-68. 88. Karlstrom, A. and P.A. Nygren, Dual labeling of a binding protein allows for specific fluorescence detection of native protein. Anal Biochem, 2001. 295(1): p. 22-30. 89. Orlova, A., et al., Update: affibody molecules for molecular imaging and therapy for cancer. Cancer Biother Radiopharm, 2007. 22(5): p. 573-84. KARIN LARSSON 61

90. Nord, K., et al., Binding proteins selected from combinatorial libraries of an alpha-helical bacterial receptor domain. Nat Biotechnol, 1997. 15(8): p. 772-7. 91. Wikman, M., et al., Selection and characterization of HER2/neu-binding affibody ligands. Protein Eng Des Sel, 2004. 17(5): p. 455-62. 92. Grönwall, C., et al., Selection and characterization of Affibody ligands binding to Alzheimer amyloid beta peptides. J Biotechnol, 2007. 128(1): p. 162-83. 93. Engfeldt, T., et al., Chemical synthesis of triple-labelled three-helix bundle binding proteins for specific fluorescent detection of unlabelled protein. Chembiochem, 2005. 6(6): p. 1043-50. 94. Hedhammar, M., et al., Negatively charged purification tags for selective anion-exchange recovery. Protein Eng Des Sel, 2004. 17(11): p. 779-86. 95. Engfeldt, T., et al., 99mTc-chelator engineering to improve tumour targeting properties of a HER2-specific Affibody molecule. Eur J Nucl Med Mol Imaging, 2007. 34(11): p. 1843-53. 96. Binz, H.K., et al., Designing repeat proteins: well-expressed, soluble and stable proteins from combinatorial libraries of consensus ankyrin repeat proteins. J Mol Biol, 2003. 332(2): p. 489- 503. 97. Binz, H.K., et al., High-affinity binders selected from designed ankyrin repeat protein libraries. Nat Biotechnol, 2004. 22(5): p. 575-82. 98. Koide, A., et al., The fibronectin type III domain as a scaffold for novel binding proteins. J Mol Biol, 1998. 284(4): p. 1141-51. 99. Xu, L., et al., of high-affinity antibody mimics using mRNA display. Chem Biol, 2002. 9(8): p. 933-42. 100. Beste, G., et al., Small antibody-like proteins with prescribed ligand specificities derived from the lipocalin fold. Proc Natl Acad Sci U S A, 1999. 96(5): p. 1898-903. 101. Skerra, A., Lipocalins as a scaffold. Biochim Biophys Acta, 2000. 1482(1-2): p. 337-50. 102. Schlehuber, S., G. Beste, and A. Skerra, A novel type of receptor protein, based on the lipocalin scaffold, with specificity for digoxigenin. J Mol Biol, 2000. 297(5): p. 1105-20. 103. biotech, a.p., Affinity Chromatography. Principles and Methods. 2001. 104. Hjelm, H., K. Hjelm, and J. Sjoquist, Protein A from Staphylococcus aureus. Its isolation by affinity chromatography and its use as an immunosorbent for isolation of immunoglobulins. FEBS Lett, 1972. 28(1): p. 73-6. 105. Langone, J.J., Protein A of Staphylococcus aureus and related immunoglobulin receptors produced by streptococci and pneumonococci. Adv Immunol, 1982. 32: p. 157-252. 106. Guss, B., et al., Structure of the IgG-binding regions of streptococcal protein G. EMBO J, 1986. 5(7): p. 1567-75. 107. Eliasson, M., et al., Chimeric IgG-binding receptors engineered from staphylococcal protein A and streptococcal protein G. J Biol Chem, 1988. 263(9): p. 4323-7. 108. Wilson, K. and J. Walker, Principles and Techniques of Practical Biochemistry. 5th ed. 2000: Cambridge University Press. 109. Engvall, E. and P. Perlmann, Enzyme-linked immunosorbent assay (ELISA). Quantitative assay of immunoglobulin G. Immunochemistry, 1971. 8(9): p. 871-4. 110. Maiolini, R. and R. Masseyeff, A sandwich method of enzymoimmunoassay. I. Application to rat and human alpha-fetoprotein. J Immunol Methods, 1975. 8(3): p. 223-34. 111. Burnette, W.N., "Western blotting": electrophoretic transfer of proteins from sodium dodecyl sulfate--polyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibody and radioiodinated protein A. Anal Biochem, 1981. 112(2): p. 195-203. 62 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

112. Renart, J., J. Reiser, and G.R. Stark, Transfer of proteins from gels to diazobenzyloxymethyl- paper and detection with antisera: a method for studying antibody specificity and antigen structure. Proc Natl Acad Sci U S A, 1979. 76(7): p. 3116-20. 113. Santoni, V., M. Molloy, and T. Rabilloud, Membrane proteins and proteomics: un amour impossible? Electrophoresis, 2000. 21(6): p. 1054-70. 114. Kononen, J., et al., Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med, 1998. 4(7): p. 844-7. 115. Sigal, A., et al., Variability and memory of protein levels in human cells. Nature, 2006. 444(7119): p. 643-6. 116. Jennings, C.D. and K.A. Foon, Recent advances in flow cytometry: application to the diagnosis of hematologic malignancy. Blood, 1997. 90(8): p. 2863-92. 117. Hengel, R.L. and J.K. Nicholson, An update on the use of flow cytometry in HIV infection and AIDS. Clin Lab Med, 2001. 21(4): p. 841-56. 118. Atassi, M.Z. and J.A. Smith, A proposal for the nomenclature of antigenic sites in peptides and proteins. Immunochemistry, 1978. 15(8): p. 609-10. 119. Van Regenmortel, M.H., Immunoinformatics may lead to a reappraisal of the nature of B cell epitopes and of the feasibility of synthetic peptide . J Mol Recognit, 2006. 19(3): p. 183- 7. 120. Teplyakov, A., et al., Epitope mapping of anti-interleukin-13 neutralizing antibody CNTO607. J Mol Biol, 2009. 389(1): p. 115-23. 121. Cunningham, B.C. and J.A. Wells, High-resolution epitope mapping of hGH-receptor interactions by alanine-scanning mutagenesis. Science, 1989. 244(4908): p. 1081-5. 122. Paus, D. and G. Winter, Mapping epitopes and antigenicity by site-directed masking. Proc Natl Acad Sci U S A, 2006. 103(24): p. 9172-7. 123. Blythe, M.J. and D.R. Flower, Benchmarking B cell epitope prediction: underperformance of existing methods. Protein Sci, 2005. 14(1): p. 246-8. 124. Barlow, D.J., M.S. Edwards, and J.M. Thornton, Continuous and discontinuous protein antigenic determinants. Nature, 1986. 322(6081): p. 747-8. 125. Geysen, H.M., R.H. Meloen, and S.J. Barteling, Use of peptide synthesis to probe viral antigens for epitopes to a resolution of a single amino acid. Proc Natl Acad Sci U S A, 1984. 81(13): p. 3998-4002. 126. Poetz, O., et al., Protein microarrays for antibody profiling: specificity and affinity determination on a chip. Proteomics, 2005. 5(9): p. 2402-11. 127. van Zonneveld, A.J., et al., Identification of functional interaction sites on proteins using bacteriophage-displayed random epitope libraries. Gene, 1995. 167(1-2): p. 49-52. 128. Fack, F., et al., Epitope mapping by phage display: random versus gene-fragment libraries. J Immunol Methods, 1997. 206(1-2): p. 43-52. 129. Smith, G.P. and J.K. Scott, Libraries of peptides and proteins displayed on filamentous phage. Methods Enzymol, 1993. 217: p. 228-57. 130. Rockberg, J., et al., Epitope mapping of antibodies using bacterial surface display. Nat Methods, 2008. 5(12): p. 1039-45. 131. Hopp, T.P. and K.R. Woods, Prediction of protein antigenic determinants from amino acid sequences. Proc Natl Acad Sci U S A, 1981. 78(6): p. 3824-8. 132. Starborg, M. and C. Hoog, The murine replication protein P1 is differentially expressed during spermatogenesis. Eur J Cell Biol, 1995. 68(2): p. 206-10. KARIN LARSSON 63

133. De Masi, F., et al., High throughput production of mouse monoclonal antibodies using antigen microarrays. Proteomics, 2005. 5(16): p. 4070-81. 134. Flicek, P., et al., Ensembl 2008. Nucleic Acids Res, 2008. 36(Database issue): p. D707- 14. 135. Moll, R., et al., The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Cell, 1982. 31(1): p. 11-24. 136. Troyanovsky, S.M., et al., Patterns of expression of keratin 17 in human epithelia: dependency on cell position. J Cell Sci, 1989. 93 ( Pt 3): p. 419-26. 137. Lundstrom, K., Structural genomics of GPCRs. Trends Biotechnol, 2005. 23(2): p. 103-8. 138. Nguyen, T.N., M. Uhlén, and S. Ståhl, De novo gene assembly using a paramagnetic solid support. In: M. Uhlén, E. Hornes, Ö. Olsvik (Eds.) Eaton Publishing Co., Natick, 1994. Advances in Biomagnetic Separation: p. 73-78. 139. Ståhl, S., et al., Solid-phase gene assembly of constructs derived from the Plasmodium falciparum malaria blood-stage antigen Ag332. Biotechniques, 1993. 14(3): p. 424-34. 140. Sarkar, C.A., et al., Directed evolution of a G protein-coupled receptor for expression, stability, and binding selectivity. Proc Natl Acad Sci U S A, 2008. 105(39): p. 14808-13. 141. Mackrill, J.J., Generation, use, and validation of receptor-selective antibodies. Methods Mol Biol, 2004. 259: p. 47-65. 142. Shi, S.R., et al., Protein-embedding technique: a potential approach to standardization of immunohistochemistry for formalin-fixed, paraffin-embedded tissue sections. J Histochem Cytochem, 2005. 53(9): p. 1167-70.

64 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Acknowledgements

Nu har jag faktiskt kommit till det allra sista! När jag började doktorera trodde jag nog aldrig att jag skulle komma hit, eller det kändes i alla fall oändligt långt borta. En massa år senare så sitter jag här och undrar hur det gick till. Utan all hjälp och allt stöd skulle jag inte ha kommit speciellt långt det är jag ganska säker på. Därför vill jag passa på att tacka alla (hoppas verkligen inte att jag glömmer någon) som har gjort sitt till för att den här avhandlingen skulle bli av.

Jag vill börja med att tacka mina handledare: Henrik, du har varit både en jättebra chef när jag jobbade i Immunotech gruppen och en superbra handledare sedan jag började doktorera. Jag skriver lite för kort, du lägger till så brukar jag ta bort lite och sen blir det ofta rätt bra på slutet. Att du dessutom är AIKare är bara ett plus i kanten. Särskilt tack för att du så imponerande snabbt kastade dig in i avhandlingsarbetet när du kom tillbaka från pappaledigheten. Sophia, efter varje möte vi haft känner jag mig alltid supertaggad och full med energi. Du är alltid så positiv och har en förmåga att sprida din entusiasm när jag har motivationssvackor. Sedan kommer du ofta med väldigt bra råd, både jobbmässiga och annat som rör livet i stort. Ditt inskolningstips räddade förra hösten!

Sedan vill jag såklart tacka övriga Professorer och PIs för att ni gör den här avdelningen till en rolig, minnesvär och intressant plats att jobba på. Jag misstänker att det är ganska ovanligt med en så stor avdelning (numera tre avdelningar) där alla på något sätt samarbetar, känner och har så roligt tillsammans.

Tack alla som jag har samarbetat med under den här tiden. Extra tack till Peter Nilsson och din grupp för hjälp med protein array spottning, Tobbe och Camilla för GPCR samarbetet och för att ni lärde mig allt om cell odling och FACSande, Cilla, Jochen och Lisa för hjälp, data och sunpunkter på Cytokeratinpeket och Pia för att du bla hjälpte till med kloning och odling när jag var mammaleding. Ett speciellt tack till Kennet Wester för trevliga avbrott när jag kommit till Uppsala och för alltid lika fina IHC bilder och roliga mikroskoptittningar.

Alla i HPR projektet! Proteinfabriken och Molbio för hjälp, buffertdealar och svar på en massa frågor. Konfokal gruppen: Emma för svar på alla mina frågor om immunofluorescence, KARIN LARSSON 65 speciellt under avhandlingsskrivandet, Marie för att du kan alltid tar dig tid att svara på allt från om jag borde splitta mina celler till hur man blandar PFA och för nya A-431 celler när jag råkat blasticidina mina, Charlotte för att du hjälpte mig framkalla finfina konfokalbilder. Alla i Uppsala för roliga julfester mm. Nya och gamla immunotechare för att jag alltid får rena på xpresserna och för att jag fått sticka emellan och framkalla mina blottar: Pia förutom all hjälp för att vi är original immunotechare! Erik för att du höjde stämningen och villigt svetsade i vår xbox, Cajsa för alla snygga blotbilder mm du skickat under avhandlingsskrivandet och för en hel del skvallrande och snack på ditt nya kontor;-)

Gamla och nya rumskompisar för alla roliga pratstunder och behövliga pauser: Tove, Emma, Cecilia L, Esther, Pawel (för att du är ett bajenfan och för att det alltid är lika roligt att höra om alla konspirationer som finns mot hammarby), Julia, Erik (har varit lite ont om djurgårdare att tracka sen du slutade), Johan L, Markus, bonus-Danne, Joel, Ulrika, Sebastian, Johanna (för hjälp med stavning när inte ens Word förstått vilket ord jag vill skriva och för lite sista-minuten-nödlösningar) och Sverker (för hjälp med diverse Illustrator problem).

Jag vill även tacka Ronny och Stisse för excellent exjobbshandledning (känns som att det var ganska länge sen nu…)

Johanna för att det är mycket roligare när du är här och lite tomt när du är borta. Det har blivit många roliga luncher på Phiphi, boxnings/avreageringspass på KTH-hallen (även om det är väldigt länge sen nu), i alla fall en konferens (värsta flygresan) och en mammaledighet ihop.

Cilla och Hanna bla för roligt sällskap på diverse konferenser och för att ni fick mig att förstå hur roligt det är att sticka och Anna för lite skitsnack i korridorerna och för att du (väldigt målande) förklarat ungefär hundra ggr hur man kalibrerar Luminexen, Marica och Anna för att ni är så trevliga och också varit här nästan lika länge som jag;-)

Fia och Johan för alltid lika roliga fikapauser, luncher mm. Det har känts bra att ha er nära, vi har ju faktiskt som Johan påpekade känt varandra mer än en tredjedel av våra liv…

66 GENERATION AND CHARACTERIZATION OF ANTIBODIES FOR PROTEOMICS RESEARCH

Alla andra på plan 3 för att det alltid är roligt när jag kommer till jobbet även när labbande och skrivande har känts mindre upplyftande.

Och såklart mina vänner utanför jobbet som förgyller min fritid (även om det inte funnits så mycket av den varan på sista tiden). Det är alltid lika roligt att träffas!

Och sist men absolut inte minst familjen: Mamma och Pappa för massor av stöd och uppmuntrande under den här tiden, speciellt under hösten, och tack för alla dagishämtningar och inhopp när jag har behövt jobba. Alla mina syskon: Git, Tobbe, Jenny och Pelle för att ni har varit sådana superbra föregångare. Min egen lilla familj: Oggi för att du alltid lugnar mig när paniken kryper närmare och för att du kommer med konstruktiva idéer och lösningar på mina ofta påhittade och ibland verkliga problem och William för att du kom och lärde mig vad som är viktigt i livet!