The sequencing of the human genome has shown that there are approximately 20 500 protein encoding genes [1], and has opened up the possibility for global expression profiling of human tissues and cells. However, the proteome is still somewhat under-explored, and up until now large-scale exploration has been restricted by the lack of specific affinity reagents for a large majority of human proteins, highlighting the pressing need for high-quality, well-validated probes. The most commonly used affinity reagents are antibodies, although other forms of probes are becoming useful alternatives.
by Dr C. Kampf, J. Linné and Dr A. Asplund

Antibody-based proteomics requires the systematic generation and use of specific antibodies to explore the proteome [2]. Based on such a strategy, the Human Protein Atlas (HPA) programme has, as its main goal, the generation of a comprehensive atlas of protein profiles in a large number of normal and cancer tissues as well as in a large set of human cell lines [3]. This multi-disciplinary research programme, which began in July 2003, allows for the systematic exploration of the non-redundant set of human proteins, combining high-throughput generation of protein-specific antibodies with protein profiling in human tissues and cells using automated immunohistochemistry and tissue microarrays (TMAs). Based on the human genome sequence (Ensembl), parts of coding sequences called Protein Epitope Signatures Tags (PrESTs), corresponding to 50-150 amino acids, are selected as antigens for each gene, based on their uniqueness relative to the entire proteome [Figure 1]. Following primer synthesis, RT-PCR and cloning into E coli, recombinant PrEST protein fragments are produced and purified. PrEST protein fragments are subsequently used as antigens to obtain polyclonal antibodies, which are affinity purified to generate unique, mono-specific antibodies (msAbs) targeting multiple linear epitopes for each protein [4]. The resulting msAbs have many advantages, as they can be used in various assays where proteins are presented in different states, such as denatured, linearised, native etc.

All antibodies are analysed on TMAs from 48 normal tissues and 20 different forms of tumour tissue [5] as well as 47 in vitro cultured cell lines and 12 clinical blood cells samples [6],  to enable a broad coverage of human organs and tissues. The immunohistochemically stained TMA-sections are then scanned, generating high-resolution images. These images provide the basis for the analysis of protein expression; a pathologist examines each and every image and manually annotates the extent and localisation of protein expression in tissues. For cell microarrays an automated image analysis-based system is used to obtain an objective score that corresponds to the level of immunoreactivity in a given cell line [7]. In addition to protein profiling using immunohistochemistry, confocal microscopy together with fluorescently labelled antibodies is also used for a more detailed determination of the subcellular distribution of proteins. Figure 2 show protein expression of the double-strand break repair protein MRE11A in tissues and cells included in the screening.

Validation of the antibodies is crucial, and within the HPA,  validation strategies are implemented at different levels in the high through-put antibody generation scheme. The sequences of both PrEST clones and PrEST protein fragments are verified using sequencing and electron spray mass spectrometry respectively. Antibody specificity is verified using protein arrays as well as Western blots to check that the generated antibody binds a target of predicted size [Figure 3A and B]. However, the final validation of the antibody is deduced from an evaluation of the immunohistochemical outcome, as the ultimate goal is to generate an immunohistochemistry-based map of protein expression patterns. The outcome of the staining pattern is evaluated with reference to information given in both the literature and bioinformatics, thus giving the antibody a reliability score.

Within the HPA program, 10 new antibodies are processed and validated every day, generating expression data from approximately 3000 new antibodies per year. The current version 5 of the Human Protein Atlas (www.proteinatlas.org) includes protein profiles from over 8.800 antibodies corresponding to 6.844 proteins (approximately 33 % of all human proteins). Altogether more than 7.3 million high-resolution images have been generated and annotated, providing  a knowledge base for protein expression and further functional studies. The current version of the Human Protein Atlas has been developed in a gene-centric way with the inclusion of all human genes and splice variants. The Protein Atlas contains validation data of the antibodies from protein array assays, Western blot analysis, immunohistochemistry and immunofluorescence. The sequence similarity to all other human proteins is displayed [Figure 1] as well as a useful search tool, which enables complex queries about protein expression in normal and/or cancer cells. The next launch of the Protein Atlas will take place in March 2010. An additional portal was recently described, the Antibodypedia [8], in which researchers and affinity reagent distributors can submit protein binders and accompanying validated experimental data from a large number of technology platforms, such as Western blots, ELISA and immunohistochemistry.
The relationship between normal human tissues, as defined by their protein expression profiles, has also been analysed using an unsupervised hierarchical clustering model providing heat maps. The detailed analysis revealed that many proteins were expressed in most tissues and cells. Only very few proteins (<2%) are expressed in any single cell type, although several proteins are expressed in a group of related cells or tissues [9]. Despite this ubiquitous protein expression, the results show that normal cells can be divided into different groups based on global protein expression patterns, and that these groups harmonise well with the current model of normal embryology and histology.

Cancer is a highly complex and heterogeneous disease associated with uncontrolled growth and invasion of adjacent tissues, and one of the most common causes of death in the western world. In addition to providing a knowledge base, the Human Protein Atlas programme facilitates the identification of cancer biomarkers, i.e. proteins with unique expression patterns in defined tumours that can be identified as potentially useful diagnostic, prognostic or treatment predictive clinical biomarkers [10]. Approximately 3% of the analysed proteins show expression patterns that meet the criteria for a potential cancer biomarker. All such proteins, showing high specificity for a given cell type or differentially expressed in any given type of cancer, are further analysed in larger patient cohorts.

In summary, we describe in his article  a publicly available protein atlas, which aims to provide a comprehensive and annotated database of high-resolution images showing protein profiles in normal and cancer tissues. Currently this database visualises the expression and localisation of one third of the human proteome in a multitude of cell types. Analysis involving all non-redundant proteins and most human normal tissues and cancers with the aim of creating global expression profiles at the protein level can thus be envisioned. The antibodies generated from this effort will constitute a valuable resource to define the proteomic landscape in tissues, support the discovery of new diagnostic and therapeutic tools and enhance opportunities for basic biological and medical research. The vision is to extend the Human Protein Atlas portal to generate a first draft of the complete human proteome by 2014.

References
1. Clamp M et al. Distinguishing protein-coding and noncoding genes in the human genome. PNASDec 4 2007; 104:
19428-33.
2. Uhlen M & Ponten F. Antibody-based proteomics for human tissue profiling. Mol Cell Proteomics 2005; 4: 384-93.
3. Uhlen M et al. A human protein atlas for normal and cancer tissues based on antibody proteomics. Mol Cell Proteomics 2005; 4: 1920-32.
4. Nilsson P et al. Towards a human proteome atlas: high-throughput generation of mono-specific antibodies for tissue profiling. Proteomics 2005; 5: 4327-37.
5. Kampf C et al. Antibody-based Tissue Profiling as a Tool for Clinical Proteomics. Clinical Proteomics 2004; 1: 285-299.
6. Andersson AC et al. Analysis of protein expression in cell microarrays: a tool for antibody-based proteomics. J Histochem Cytochem  2006; 54: 1413-23
7. Stromberg S et al. A high-throughput strategy for protein profiling in cell microarrays using automated image analysis. Proteomics 2007; 13: 2142-50.
8. Björling E and Uhlén M. Antibodypedia - a portal for sharing antibody and antigen validation data. Mol Cell Proteomics 2008; 7: 2028-2037.
9. Pontén F et al. A global view of protein expression in human cells, tissues and organs. Mol Syst Biol 2009;5:337
10. Bjorling E et al. A web-based tool for in silico biomarker discovery based on tissue-specific protein profiles in normal and cancer tissues. Mol Cell Proteomics 2008;7(5):825-44.

The authors
Caroline Kampf, Ph.D.*, Jerker Linné, M Sc and Anna Asplund, Ph.D.,Department of Genetics and Pathology
Uppsala University
Sweden

* Corresponding author