Protein Detecting Microarrays

For clinical diagnostics, the goal is to develop protein-detecting microar-rays with capture agents/ligands that bind specifically to target proteins in complex biological solutions (Figs. 2 and 3) (K2, K3). The efficiency of a samples. AD = adenocarcinoma, SQ = squamous-cell carcinoma, LA = large-cell carcinoma, META = metastases to lung from other sites, REC = recurrent NSCLC, CAR = pulmonary carcinoid, NL = normal lung.1

1 Reprinted with permission from Elsevier (The Lancet, 2003, 362, 433-439).

Biological sample (cell lysate, serum, etc.)

  1. 2. A protein-detecting microarray. Each square in the grid represents a different feature of the array that would be impregnated with a particular protein ligand (blue shapes). When the sample is applied to the chip, each ligand will capture its target protein (orange and red coils in blow-up). The amount of target protein bound to each feature of the array would be quantitated with probes such as fluorescently labeled antibodies against the captured proteins. A fluoresence scanner would then measure the intensity of fluorescence (diVerently shaded green squares) at each spot, which would reflect the level of captured protein.2 (See Color Insert.)
  2. 2. A protein-detecting microarray. Each square in the grid represents a different feature of the array that would be impregnated with a particular protein ligand (blue shapes). When the sample is applied to the chip, each ligand will capture its target protein (orange and red coils in blow-up). The amount of target protein bound to each feature of the array would be quantitated with probes such as fluorescently labeled antibodies against the captured proteins. A fluoresence scanner would then measure the intensity of fluorescence (diVerently shaded green squares) at each spot, which would reflect the level of captured protein.2 (See Color Insert.)

protein-detecting microarray is dependent on a number of factors. The solid support and surface chemistry should minimize the amount of sample needed and optimize the efficiency of protein detection. Immobilized ligands must be stable and retain activity over extended periods. Ligand-binding capability must be validated to ensure that the working range covers the physiologically relevant concentrations of proteins. In addition, methods for signal detection and quantification should have a large dynamic range

2 Reprinted from Trends in Biochemical Sciences, 27, T. Kodadek, Development of protein-detecting microarrays and related devices, 295-300. Copyright (2002), with permission from Elsevier.

  1. 3. A bead-based format for the parallel detection of proteins. Each bead displays a different binding agent directed against a specific protein target (blue shapes). Each bead is color-coded by covalent linkage of two dyes (red and orange shapes) at a characteristic ratio, allowing for uniquely coded beads. Only two beads are shown for clarity. Upon application of the biological sample, the target protein binds to the capture agents. A mixture of secondary binding ligands (in this case, antibodies) conjugated to a fluorescent tag (green) is applied to the mixture of beads. The beads are then passed through a detector where two lasers "read" the ratio (n:m, x:y) of dyes and thus identify the bead, while the fluorescence intensity is read to quantitate the amount of labeled antibodies present (which will reflect the analyte level).2 (See Color Insert.)
  2. 3. A bead-based format for the parallel detection of proteins. Each bead displays a different binding agent directed against a specific protein target (blue shapes). Each bead is color-coded by covalent linkage of two dyes (red and orange shapes) at a characteristic ratio, allowing for uniquely coded beads. Only two beads are shown for clarity. Upon application of the biological sample, the target protein binds to the capture agents. A mixture of secondary binding ligands (in this case, antibodies) conjugated to a fluorescent tag (green) is applied to the mixture of beads. The beads are then passed through a detector where two lasers "read" the ratio (n:m, x:y) of dyes and thus identify the bead, while the fluorescence intensity is read to quantitate the amount of labeled antibodies present (which will reflect the analyte level).2 (See Color Insert.)

appropriate for biological samples. Various aspects of protein arrays, from surface chemistry to detection systems, have been reviewed (C6, E1, F2, K2, K3,M1,T1,Z1). We will focus on the progress in ligand isolation, which is the most crucial feature for the development of clinical protein-detecting microarrays.

3.1. Recent Advances in Protein Ligand Isolation

Currently, antibodies are most often used as protein ligands because of their high specificity and affinity (KD in the nM range) for a target protein. However, traditional methods of antibody production are not amenable to high-throughput isolation. Generally, the production of polyclonal antibodies takes 2 to 4 months and requires about 0.2 to 2 mg of purified antigen. It will take another 2 to 4 months to then produce a monoclonal antibody for the particular antigen. Moreover, the high-throughput production of purified antigen is challenging because purifying protein antigens is labor intensive and conditions generally need to be optimized for each protein. Recently, a more high-throughput method has been developed to generate polyclonal antibodies in mice (C2), which uses genetic immunization (T2) rather than purified antigen. Genetic immunization involves directly trans-fecting antigen-presenting cells with genes to express the antigen. Antibody response is enhanced by codon optimization of genes and addition of various elements to enhance antigenicity, such as plasmids encoding genetic adjuvants. Using this method, polyclonal antibodies can be produced within 4 to 8 weeks, even for antigens that failed to produce a response in protein form. The exact affinities of the polyclonal antibodies have not been determined, but are thought to be comparable to other antibodies.

Antibodies may not be optimal ligands for protein-detecting arrays even with high-throughput antibody production. Most commercially available antibodies were found to be unsuitable for microarray-based analysis of cellular lysates (M1). A crucial drawback is that any type of antibody or folded protein is prone to loss of activity upon immobilization and storage. In contrast, small synthetic ligands are more stable and can be produced and purified economically and efficiently in bulk. These synthetic ligands are typically protein aptamers (antibody mimics), peptides, peptide-mimics, and small organic molecules.

Various molecular biology techniques are available to screen for protein aptamer and peptide ligands for specific proteins. Phage display technology, introduced in 1985 (S5), has been used for the isolation of peptide (B5, F1, L5) and antibody fragment (G4) ligands for specific proteins. A 2002 review focuses on the principle of phage display technology and methods for the construction and bio-panning of phage libraries (A3). Libraries are constructed in vitro by inserting foreign DNA into specific locations of the genome of filamentous phage. The encoded protein or peptide is displayed on the surface as a fusion protein with one of the phage coat proteins, generally pill, which displays five copies. Ligands bind to the protein of interest, which is immobilized on a plate. Bound phage are then eluted and amplified for more stringent rounds of panning. The amino acid sequence of the selected ligand can readily be determined by sequencing inserted DNA in the phage genome.

Antibody fragment libraries from immunized and nonimmunized sources can be used in phage display and peptide libraries are commercially available. An alternative technique is to design protein aptamers that consist of a stable protein scaffold on which random peptides are displayed. An example of protein aptamers are affibodies, which present a library of 13 randomized amino acids on the Z domain of Staphylococcus aureus protein A. Crystal structure studies indicate similarity in the binding of an affibody to its target to protein-antibody interactions. However, affibodies have a dissociation constant of approximately 1 yM compared to antibody-antigen complexes of1 nM or less (H3, R1, W1).

The larger the library, the greater the probability of selecting rare high-affinity ligands. Phage display libraries typically contain 108-9 peptides with the limiting factor being the transformation efficiency of bacteria (L6). The in vitro techniques, ribosome and mRNA display, overcome this limitation since more complex libraries up to 1013 can be prepared (R2, W2). During in vitro transcription-translation of random DNA libraries, the encoded peptide remains associated with its mRNA. Either a ribosome complex is formed noncovalently by stalling the ribosome or the peptide is covalently linked to the mRNA through puromycin. Additional advantages of these techniques are that binding of the ligand is monovalent and affinity maturation can be achieved over several rounds of screening by error-prone PCR or DNA shuffling (A3).

This technique has been used successfully to isolate ligands that retain their high-affinity binding properties when immobilized on a protein microarray. An mRNA library of antibody-mimics was prepared by randomizing three exposed loops on a stable, soluble protein, the tenth fibronectin type III domain. After 10 selection rounds, high-affinity ligands for TNF-a were isolated with dissociation constants between 1 and 24 nM. These ligands were further optimized by random mutagenesis to provide a ligand with a KD of 20 pM (X2).

Disadvantages of both phage and mRNA display are the requirement for numerous rounds of selection and amplification, as well as the need to express and purify target proteins. Selectively infective phage (J3) and bacterial (H4, J2) and yeast (Y2) two-hybrid methods can overcome these obstacles because they are one-step screening assays with in vivo expressed target proteins.

For the selectively infective phage technique, the N-terminal domains of the pIII coat protein is replaced with peptides from a ligand library, resulting in noninfective phage particles. To restore phage infectivity, adaptor molecules consisting of the target protein coupled to the missing N-terminal domains are required. These adaptor molecules can be expressed and exported to the periplasm in E. coli, eliminating the need for purified protein. Interaction between the fused peptide expressed on the phage coat and the adaptor molecule restores infectivity, allowing ligand selection in a single round. Although this method appears to have potential for ligand isolation (I2), few protein ligands are reported in the literature. This technology may be less successful for ligand isolation because of the potential for false positives (I3) or the size restriction of the target or ligand (C1).

The yeast two-hybrid system detects protein-protein or protein-peptide interactions in vivo. The target or ''bait'' protein and the ligand library are fused to either the DNA-binding domain or the transcription activation domain. Yeast cells are transformed with both plasmids and only the transformants expressing the protein-ligand interaction are selected (Y2). The main advantage is the one-step in vivo screening; however, the library size is limited to about 107 because of the transformation efficiency of the cells.

A variation on the yeast two-hybrid technique is the bacterial two-hybrid system. In this case, the target protein and the library-encoded peptide are each fused to a monomer of the DNA binding domain. Only if the target and ligand interact will the DNA binding domain form an active dimeric repres-sor. An activated repressor results in cells immune to phage infection, allowing for one-step selection of immune cells. This method was shown to be capable of selecting a peptide that specifically bound to its target protein with a Kd in the micromolar range and inhibited the activity of the target protein in vivo (Z2). Although specific peptide ligands have been selected by both yeast and bacterial two-hybrid methods, there is little documented evidence for the identification of high-affinity ligands.

An alternative to the biological methodologies for screening protein ligands is the synthetic combinatorial library approach. Chemical libraries are prepared on a solid-support, usually on bead or a microarray format, and encompass a variety of synthetic molecules such as peptides, peptide mimics, and small organic molecules. For feasibility reasons, the libraries are usually limited to a size of 105 to 106, which is several-fold less than libraries developed from other techniques. However, an appealing feature is that a synthetic library is not limited to the 20 natural amino acids, thereby allowing for the inclusion of a variety of chemical properties. One such example are peptoids, N-substituted oligoglycines, which are structurally similar to peptides but are resistant to proteolytic cleavage, easily synthesized on resin, and which have diverse chemical side chains on the nitrogen of the peptoid backbone (A2, F3). Synthetic ligands from a combinatorial library are amenable to high-throughput screening and can easily be prepared in large quantities with little variability.

Numerous protein ligands and inhibitors have been identified using combinatorial libraries. A comparison of combinatorial peptide library approaches has been outlined in a recent review (L7). The one-bead-one-compound (OBOC) approach entails the synthesis of thousands or millions of random compounds on bead. Small molecules as well as peptide mimetics have been identified as ligands to cellular proteins such as protein kinases and intracellular signaling proteins using an OBOC approach (L2). Protein ligands have also been isolated from biased libraries, which include a structural motif or derivatives of initial leads. By incorporating a consensus sequence into a peptide library, ligands were discovered to bind to the SH3 domain of phosphatidylinositol 3-kinase with modest affinity (C5). Inhibitors of aspartyl proteases have been isolated from a peptide library which incorporated chemical functional groups known to interact with essential active site residues (L4). Bead-based libraries are most commonly used for the development of protein ligands. However, a combinatorial small-molecule library on a microarray format was screened and included an inhibitor of the transcription factor Hap3p (K4).

Both the biological methods and the chemical combinatorial libraries usually yield low to modest affinity protein-ligands (KD in micromolar range), which are insufficient to capture low abundance proteins from complex biological mixtures. Rather than designing and synthesizing larger libraries, an alternative approach is to synthesize multivalent ligands. Two modest affinity ligands can be linked to yield a new ligand with an affinity that theoretically equals the product affinity for the two individual molecules.

One example of this "pincer" strategy was demonstrated when nuclear magnetic resonance was used to identify small molecules that bind to diVer-ent surfaces on FK506-binding protein and determine the appropriate linker for the two compounds. The chimeric molecule of the individual FK506-binding protein binders had a KD of 19 nM (S3). A potent inhibitor of cSrc kinase was designed using a similar strategy. Initial lead molecules for kinase inhibition were identified from aldehyde-derived oxime compounds. Screening of a small library of chimeric compounds of the initial hits yielded a much more eVective kinase inhibitor with an inhibition constant, Kl5 of -60 nM.

The pincer approach oVers a unique opportunity for creating chemically diverse protein-ligands, though designing optimal linkers for the pincer molecule requires some experimental eVort. Instead of linking two solution binders, protein ligands can be immobilized onto solid support, providing a wide variety of combinations of the two ligands. Appropriately positioned ligands will bind different surfaces of the same protein, increasing the overall affinity. Therefore, two noncompetitive, modest-affinity ligands can be synthesized on solid support without a linker to provide a high-affinity chimeric molecule, also known as the mixed-element capture agent (MECA) (Fig. 4) (B1). To demonstrate this, a MECA of two specific protein ligands was synthesized on resin. Each peptide of the MECA was specific for monomeric protein, either MBP or Mdm2. The MECA was determined to have a slightly higher affinity for the MBP-Mdm2 fusion protein as compared to the individual peptide ligands in solution, but was a much more eVective capture agent on solid support.

Using a similar idea, high-affinity protein capture agents can be designed simply by immobilizing modest-affinity ligands for multimeric proteins.

  1. 4. Schematic representation of the MECA concept. Two noncompeting ligands (red and blue shapes) could be immobilized individually (left) or as a linear fusion (right), allowing for two appropriately positioned molecules to cooperate in the capture of the target protein. Reprinted with permission from (B1). Copyright (2003) American Chemical Society.
  2. 4. Schematic representation of the MECA concept. Two noncompeting ligands (red and blue shapes) could be immobilized individually (left) or as a linear fusion (right), allowing for two appropriately positioned molecules to cooperate in the capture of the target protein. Reprinted with permission from (B1). Copyright (2003) American Chemical Society.

Some fractions of these ligands should be appropriately oriented on the surface to promote binding of multiple ligands to one protein (i.e., one ligand bound per monomer). High-affinity capture agents were created by synthesizing peptide ligands to dimeric proteins on Tentagel resin (N1). These immobilized ligands dramatically increased the half-life of the peptide-target protein complex when compared to random peptide ligands. The creation of high-affinity capture agents from modest affinity solution binders suggests that protein-detecting arrays may be more readily available than previously expected.

3.2. Recent Applications of Protein-Detecting Microarrays

Due to the limited availability of well-characterized ligands, protein-detecting arrays are not ideal as a signature/biomarker discovery tool. However, protein-detecting microarrays are being developed as research tools and for diagnostics (L1). The first generation of protein microarrays are constructed with antibodies as capture agents. Although antibodies are less stable, very few synthetic ligands are currently available. An array of 368 antibodies was developed to identify the proteins present in the tissue of a single case of oral cavity cancer (K1). Antibody suspensions were spotted onto a thin film of nitrocellulose bonded to a glass slide. Protein lysates were biotinylated and bound protein was detected and quantified by an enzyme-linked colorimetric assay. The antibody arrays were capable of detecting cancer-related proteins, as three of the eleven proteins detected were previously identified in tissue culture models of oral cavity cancer. Although this is the largest array to date, the assay needs to be validated, since the antibodies were not characterized with respect to affinities, concentration, and cross-reactivity. Furthermore, the detected proteins were in non-native form.

In another study, a screen of potential serum biomarkers of human prostrate cancer identified five proteins with significantly diVerent expression levels between 33 prostate cancer samples and 20 healthy controls (M3). These proteins were detected by antibodies spotted on either microscope slides coated with poly-L-lysine/N-hydroxysuccinimide-4-azidobenzoate (HSAB) or acrylamide-based HydrogelTM-coated slides. One hundred and eighty-four antibodies to target serum proteins and intracellular proteins were spotted in quadruplicate. The sample proteins were labeled directly with fluorescent tags and compared to a reference sample consisting of equal volumes of all serum samples. Labeling the sample eliminates the need for paired antibodies able to detect noncompeting epitopes of a protein, but can lead to bias. To control for labeling bias, the samples and reference were alternately labeled with different fluorescent tags (reverse labeling). Labeling bias due to diVerent fluorophores can be controlled this way, but bias will be introduced if the presence of any label interferes with binding of the labeled protein to its specific ligand. Hydrogels were considered superior to the poly-L-lysine/HSAB-coated slides because of the lower background and more antibodies with a measurable signal (78 compared to 23).

A number of companies have developed antibody arrays for research purposes. For example, a bead-based assay to screen up to 17 cytokines is available from Bio-Rad Laboratories (Hercules, CA, USA). In addition to the bead-based assays, Zyomyx, Inc. (Hayward, CA, USA) offers a Human Cytokine Biochip for profiling of 30 biologically relevant cytokines. BD Biosciences Clontech (Palo Alto, CA, USA) has developed the largest commercially available antibody array for research purposes. This array includes over 500 antibodies with affinity for proteins involved in a range of biological functions such as signal transduction, cancer, cell cycle regulation, cell structure, apoptosis, and neurobiology. Furthermore, the company reports that proteins present in the low pg/ml range can be detected in complex protein mixtures.

The first antibody array with diagnostic potential was produced for immunotyping of leukemia (B3). Sixty antibodies were adhered to a film of nitrocellulose bound to a glass slide. Leukocytes from leukemia patients and healthy controls were incubated on arrays and bound leukocytes were visualized by dark-field microscopy. Relative densities of subpopulations of cells with distinct immunophenotypes were determined by eye. Distinctive and reproducible patterns were obtained for five leukemia types, indicating the potential for accurate diagnosis. Flow cytometric analysis of samples from two patients with chronic lymphocytic leukemia correlated closely with the array analyses for antigens expressed at high levels. A comparison of samples from 20 patients with chronic lymphocytic leukemia and 20 healthy controls indicated that leukocyte expression levels for 7 of the 60 cell-surface antigens could discriminate between the two sets. Although the results are only semiquantitative, this study suggests that leukemia types can be differentiated rapidly using a simple technique without specialized, expensive equipment.

An alternative to the antibody array is the immobilization of proteins and detection of specific antibodies in sera. These antibody-detecting arrays will probably be the first to be routinely available in the clinic for serodiag-nosis of autoimmune and infectious diseases. Serum samples from 60 individuals were tested with an array of microbial antigens printed on silanized glass microscope slides (M2). Anti-human IgG and IgM detection antibodies were labeled with fluorophores and quantified using confocal scanning microcopy. Comparison with commercially available ELISAs indicated that the microarray assay could identify positive and negative sera with similar efficiency. In this experiment, only 5 microbial antigens were arrayed.

However, it is conceivable that arrays could be developed with ligands for a range of infectious disease agents and microbial toxins, allowing for rapid serodiagnosis in a clinical setting.

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook


Post a comment