Human antibody profiling technologies for autoimmune disease

Protein arrays comprise full-length proteins or protein domains being displayed on the solid surface. Some of the first high content protein arrays were established in the year 2000, with over 10,000 full-length human proteins displayed on a glass slide [5]. Since then, protein arrays displaying human proteins have been used for a wide range of applications, including detecting novel antibody biomarkers in autoimmune diseases. Protein arrays can be divided into two categories based on their method of protein synthesis: those that are produced using cellular expression, or those produced using cell-free methods such as in situ synthesis. Cellular produced arrays utilise recombinant proteins expressed in a host organism prior to immobilising on the solid surface. The choice of the expression host is important to consider with common host organisms including bacteria (mostly commonly Escherichia coli), Saccharomyces cerevisiae cells (yeast), insect cells, and human cells. The expression host can affect the protein glycosylation and other post-translational modifications such as citrullination, and it is known that antibodies have the ability to recognise these post-translational modifications [3, 6]. Proteins expressed in yeast and insect cells will be glycosylated, unlike proteins expressed in bacterial hosts such as E. coli, though the glycosylation patterns still differ to that of humans. The number of proteins included in the array is limited to the labour and cost associated with expression and purification of the proteins. This led to the development of cell-free or in situ protein array synthesis, in which proteins are synthesised directly onto the solid surface [7]. DNA or RNA libraries are bound to the solid surface, and just hours prior to running a sample across the array, the proteins are freshly synthesised, minimising the risk of protein denaturation [7].

While custom protein arrays can be produced for a particular purpose, there are a variety of protein arrays available commercially. One of the most widely used for the discovery of novel biomarkers in autoimmune diseases is the Human ProtoArray (Thermo Fisher Scientific), based on over 9000 human proteins expressed with an N-terminal GST-tag in baculovirus, purified from SF9 insect cells and bound to a nitrocellulose-coated glass slide. The N-terminal GST tag allows for the direction of the binding of the protein on the solid surface to be controlled, which may also influence how accessible epitopes within each protein are displayed. The majority of antigens represented on the ProtoArray are metabolism-based proteins (35%), while 16% of displayed proteins are membrane associated or secreted, and 5% are nuclear based proteins. Although the Human ProtoArray was discontinued in 2018, many research groups have employed the platform to discover autoantibodies in autoimmune diseases (Table 1) including type 1 diabetes [8], systemic lupus erythematosus (SLE) [4], multiple sclerosis [9], and acute rheumatic fever [10].

The more recent and most expansive protein array currently available is the HuProt Human Proteome microarray (v4.0, CDI laboratories). The 20,000 proteins cover 81% of the human proteome including 87% of predicted secreted proteins and 78% of plasma membrane proteins based on the Human Protein Atlas. This increased proportion of secreted and plasma membrane proteins offers an advantage compared to the Human ProtoArray for detection of novel autoantibody biomarkers that are more likely to target extracellular and exposed antigens. The HuProt Array has been employed to detect novel autoantibodies in several autoimmune diseases (Table 1) including autoimmune hepatitis [11], multiple sclerosis [12], acute rheumatic fever [10] and SLE [13].

Nucleic Acid Programmable Protein Array (NAPPA) is a popular format of in situ protein synthesis in which proteins are synthesised and captured directly onto the solid surface. For example, a NAPPA developed at the BioDesign Institute in the USA, includes over 12,000 genes encoded to express proteins with a C-terminal GST tag to ensure full translation of the protein prior to local capture on the solid surface [14]. Mammalian cell-free lysates and accessory proteins are added to synthesise the proteins in vitro enabling simple post-translational modifications to be represented such as phosphorylation and citrullination. Autoimmune diseases in which NAPPA has been utilised (Table 1) include type 1 diabetes [15, 16] and rheumatoid arthritis [17].

An emerging human protein array approach is the i-Ome Protein Array Kit (Sengenics), which comprises over 1600 antigens including signalling molecules and cytokines. The proteins are expressed in insect cells with a carboxy-biotin carrier protein signal that marks for correct protein folding to ensure display of conformational epitopes. While the inclusion of structural epitopes is an advantage of the i-Ome array, a limitation is the size, particularly as a discovery technology when the potential breadth of the autoantibody repertoire in autoimmune disease is considered. Examples of autoimmune diseases the i-Ome Protein Array Kit has been utilised to detect novel autoantibodies include SLE [18, 19] and rheumatoid arthritis [20] (Table 1). Another emerging protein array is ImmunoINSIGHTS (formally known as Serotag (Oncimmune)), which differs from the prior examples based on planar solid surfaces, as the over 8000 human proteins are immobilised on Luminex bead-based suspension arrays enabling solution phase antigen binding. Although a relatively new technology, it has been utilised (Table 1) in rheumatoid arthritis [21, 22] and SLE [23].

In contrast to protein arrays, peptide arrays comprise of small protein fragments or peptides, rather than full-length proteins. Peptide arrays with overlapping, or tiled, peptides (6–20 amino acids in length) are generally used for epitope mapping within proteins that have previously been associated with autoimmune diseases [24‐26]. Arrays that employ longer peptide or protein fragments (80–100 amino acids in length) are more often used to discover novel antigens associated with an autoimmune disease. Longer peptides or protein fragments are more likely to have some secondary structure and contain conformational epitopes present in native proteins, compared with short peptides [12, 27]. Like protein arrays, a limiting factor in array production is the labour and cost associated with peptide synthesis. In situ synthesis is also available for peptide arrays, in which the peptides are synthesised in parallel directly onto the solid surface [28]. Commercial peptide array synthesisers such as the Multipep Synthetiser (Invatis) allow for research groups to design and synthesise custom peptide arrays relevant to the autoimmune disease being studied [29].

Examples of autoimmune diseases for which peptide arrays have been utilised to map epitopes within previously identified autoantigens include multiple sclerosis [26] and SLE [25] (Table 1). The largest peptide array to date contains approximately 2.2 million overlapping 12 amino acid long peptides that represent the entire human proteome and has been used to discover novel autoantibodies in multiple sclerosis [24]. Other examples of commercially available short peptide arrays include PEPperCHIP (PEPperPRINT), BioSynth (BioSynth), and PEPstar and Pepspots (JPT Innovative Peptide Solutions), which can provide custom-designed or standard arrays based on protein sequences.

The Human Peptide Array (SciLifeLab) utilises longer peptides and protein fragments ranging 80–100 amino acids in length allowing for novel autoantigen discovery. The array is based on unique Protein Epitope Signature Tags (PrESTs) designed by the Human Protein Atlas. These PrESTs have low homology to other human protein sequences so that every fragment is unique. The 42,000 fragments, representing 94% of the human proteome, are routinely expressed in E. coli, purified, and immobilised on microarrays to create the Human Peptide Arrays. Examples of autoimmune diseases in which these peptide arrays have been used include multiple sclerosis [12, 27], rheumatoid arthritis [17, 30], and sarcoidosis [31] (Table 1).

Phage immunoprecipitation sequencing (PhIP-Seq) involves displaying a synthetic library of peptide or protein fragments of between 40 and 90 amino acids on bacteriophage. An oligonucleotide library encoding the fragments is amplified and cloned into a T7 phage display system [32, 33]. The phage library is incubated with sera, and phage bound by serum IgG is precipitated with protein A/G coated magnetic beads [33, 34]. The fragment displayed (phenotype) is linked with the fragment encoded within the genome of each phage (genotype) so that next-generation sequencing of precipitated phage provides insight into serum antigen specificity. Once a phage library has been constructed, the immunoprecipitation and sequence data can be collected within 1 week, making the technique quick and affordable [31].

The affordability and depth of data that can be obtained from PhIP-seq has led to its increasing use in autoimmune disease research (Table 1) including type-1 diabetes, multiple sclerosis, rheumatoid arthritis [32], autoimmune encephalitis [35], and most recently the post-COVID-19, multi-inflammatory syndrome in children (MIS-C) [36]. As display technologies are adaptable to a 96-well format, sample throughput is high. The depth of data obtained, as result of the incorporation of next-generation sequencing approaches, has highlighted the huge diversity of autoantibody reactivity between individuals. These unique autoantibody fingerprints maybe a hallmark of autoimmune disease that is only beginning to be appreciated [27, 37].

Molecular Indexing of Proteins by Self-Assembly (MIPSA) was recently developed to complement peptide display approaches, in particular PhIP-Seq. In MIPSA over 11,000 proteins are displayed on ribosomes [38], and each protein has a unique, conjugated amplifiable DNA-barcode that enables paralleled sequencing. MIPSA was first described at the time of writing and has been used to investigate autoreactive antibodies in severe SARS-CoV-2 infections as a proof of concept [38]. As MIPSA utilises full-length antigens, this technique can be used to detect discontinuous and conformational epitopes, whereas the use of protein fragments in PhIP-Seq is more likely to detect linear epitopes [38]. The complementary nature of these approaches, including assay conditions and amplification and sequencing primers, provides opportunities for PhIP-Seq and MIPSA to be performed together in a single reaction in the future [38].

Rapid extracellular antigen profiling (REAP) was recently developed to increase the sensitivity of autoantibody detection to extracellular proteins, limited in other technologies by complex folding and post-translational modification requirements [39]. Human extracellular and secreted proteins, selected as those most accessible to autoantibodies in vivo, are displayed on the surface of yeast cells. Unlike phage produced by E. coli, yeast is a eukaryotic organism capable of producing post-translational modifications, although the glycosylation patterns differ slightly to that of humans [39]. Each displayed antigen is genetically barcoded and like PhIP-Seq, next-generation sequencing is used to identify autoantigens following incubation with sera and magnetic isolation of IgG bound yeast. The REAP library consists of 2688 extracellular proteins, encompassing a wide range of protein families and representing 87% of all human exoproteins. The length of the extracellular regions displayed ranges between 50 and 600 amino acids [40]. REAP is a very new technique, and proof of concept was shown in the autoimmune diseases APS-1 and SLE [40] (Table 1). Of note, the technology was also utilised to show that patients with severe SARS-CoV-2 infections develop autoantibodies, indicative of immune dysregulation [41]. The increased sensitivity for REAP to identify extracellular autoantigens was illustrated by the identification of autoantibodies to chemokines, cytokines, and growth factors in SLE patients [40]. This adds to the autoantibodies to intranuclear proteins identified with the Human Protoarray [4] and provides new avenues for research into SLE pathogenesis.