key: cord-0035068-7t44ufjf
authors: Wingren, Christer; Borrebaeck, Carl AK
title: Protein Microarray Technologies for Detection and Identification of Bacterial and Protein Analytes
date: 2008
journal: Principles of Bacterial Detection: Biosensors, Recognition Receptors and Microsystems
DOI: 10.1007/978-0-387-75113-9_26
sha: 06d5e4e4ac7e80cc8824d665d17528da08b10bfe
doc_id: 35068
cord_uid: 7t44ufjf

Protein-based microarrays is a novel, rapidly evolving proteomic technology with great potential for analysis of complex biological samples. The technology will provide miniaturized set-ups enabling us to perform multiplexed profiling of minute amounts of biological samples in a highly specific, selective, and sensitive manner. In this review, we describe the potential and specific use of protein microarray technology, including both functional protein microarrays and affinity protein microarrays, for the detection and identification of bacteria, bacterial proteins as well as bacterial diseases. To date, the first generations of a variety of set-ups, ranging from small-scale focused biosensors to large-scale semi-dense array layouts for multiplex profiling have been designed. This work has clearly outlined the potential of the technology for a broad range of applications, such as serotyping of bacteria, detection of bacteria and/or toxins, and detection of tentative diagnostic biomarkers. The use of the protein microarray technology for detection and identification of bacterial and protein analytes is likely to increase significantly in the coming years.

Entering the post-genomic era, proteomics-the large-scale analysis of proteins-has become a key discipline (Phizicky et al. 2003; . To this end, the need for (novel) technologies, allowing us to perform rapid and multiplexed analysis of biological samples in a selective, specific, and sensitive manner in various applications, ranging from focused assays to proteome-scale analysis, is tremendous (Yanagida 2002; Hanash 2003; Phizicky et al. 2003; Wingren and Borrebaeck 2004) . The protein-based microarray is a promising and rapidly evolving technology that may provide us with the unique means to perform highthroughput proteomics (Haab 2001; Wingren and Borrebaeck 2004; Kingsmore 2006) . In this review, we will describe the potential and specific use of protein microarray technology for the detection and identification of bacteria, bacterial proteins, and bacterial diseases. The number of such applications is still low, but is likely to increase significantly as the microarray technology develops.

The concept of protein microarrays is based on the arraying of a small amount (pL to nL scale) of protein in discrete positions in an ordered pattern, a microarray, onto a solid support where they will act as probes, or catcher molecules (Fig. 26 .1) (Haab 2001; MacBeath 2002; Wingren and Borrebaeck 2004; Angenendt 2005; Kingsmore 2006) . A minute quantity ( L scale) of the biological sample, e.g., serum, is then incubated on the array and any specifically bound analytes can be detected using mainly fluorescence as a read-out system. Adopting a high-performing setup, assay sensitivities in the pM to fM range can be observed, allowing low-abundance analytes to be readily targeted even in complex samples (Pawlak et al. 2002; Wingren et al. 2005; Wingren et al. 2006) . Depending on the assay setup at hand, the observed microarray binding pattern can then be converted into, for example, a protein-ligand interaction profile, a (differential) protein expression profile, or even a proteomic map revealing the detailed composition of the proteome at the molecular level (MacBeath 2002; Wingren and Borrebaeck 2004; Kingsmore 2006) .

Protein microarrays are frequently divided into two conceptual classes of array approaches, functional protein microarrays (functional proteomics) and affinity protein microarrays (quantitative proteomics) (MacBeath 2002; Poetz et al. 2005) . While functional protein microarrays examine the biochemical activity, such as ligand binding properties and reactivity, of a set of immobilized target proteins (MacBeath and Schreiber 2000; Zhu et al. 2000 Zhu et al. , 2001 Phizicky et al. 2003) ; affinity microarrays utilize affinity reagents as probes to detect and measure the abundance of multiple proteins in a (semi-) quantitative way (Haab 2003; Borrebaeck 2004, 2006a ). As will be described in this review, both classes of protein microarrays have been used for detection and identification of bacteria and bacterial protein analytes, and the different protein array technology platforms have already been shown to display great promise within biomedical and biotechnological applications (Table 26 .1). In this context, it might be of interest to note that protein-based microarray applications have recently emerged in a similar manner also for the detection of viruses, viral proteins, and viral diseases (Perrin et al. 2003; Yuk et al. 2004; Livingston et al. 2005; Lu et al. 2005; Zhu et al. 2006 ).

In general terms, functional protein microarrays have been designed and developed to investigate the biochemical properties, e.g., immunoreactivity ; Hueber Lueking et al. 2005) , and the functional properties, e.g., protein-protein interactions, of the arrayed proteins (MacBeath and Schreiber 2000; MacBeath 2002; LaBaer and Ramachandran 2005; Kingsmore 2006 ). In particular, the use of cDNA expression libraries as the probe source has been very rewarding (Zhu et al. 2000 (Zhu et al. , 2001 Horn et al. 2006) . For example, microarrays composed of 119 of 122 yeast protein kinases (Zhu et al. 2000) , 5,800 of 6,200 yeast proteins (Zhu et al. 2001) , or 37,200 redundant recombinant fetal brain proteins (Horn et al. 2006 ) have been designed, fabricated, and successfully applied to perform global protein-protein interaction studies. To date, mainly water-soluble protein analytes have been targeted, but the first microarray design also targeting membrane proteins in the format of intact mammalian cells has recently been published (Deviren et al. 2007 ). Functional protein arrays have frequently been applied within the academic research community, but several commercial ventures have also developed such microarrays that are now available on the market. For example, focused microarrays targeting cell signaling proteins and tentative cancer-associated proteins have been launched by Sigma (http://www.sigma.com). In addition, comprehensive microarrays based on over 8,000 human recombinant proteins have recently been released by both Protagen (http://protagen.de) and Invitrogen (http://invitrogen.com).

In the case of bacteria and bacterial protein analytes, various designs of functional protein microarrays have been successfully applied (Table 26 .1). In these first examples, microarrays based on a set of human or yeast proteins (Galindo et al. 2006) , recombinant bacterial proteins (Li et al. 2005; Steller et al. 2005; Tong et al. 2005) , biochemically isolated and fractionated bacterial proteins (Tong et al. 2005; Sartain et al. 2006) , neoglycoproteins (oligosaccharides bound to bovine serum albumin) (Tong et al. 2005) , and lipopolysaccharides and saccharides (Tong et al. 2005) have been developed and fabricated. In these studies, the first generation(s) of applications for the identification of toxin modulators/regulators (Galindo et al. 2006) , disease state differentiation (Tong et al. 2005; Galindo et al. 2006; Sartain et al. 2006) , and identification of diagnostic markers (Li et al. 2005; Steller et al. 2005 ) have been outlined.

Affinity protein microarrays have been developed for multiplex protein expression profiling, specifically detecting whether the targeted analytes are expressed and at what (relative) levels (MacBeath 2002; Angenendt 2005; Wingren and Borrebaeck 2006a) . To date, antibodies are the most commonly used probe source for affinity protein microarrays (Wingren and Borrebaeck 2004; Haab 2006; Kingsmore 2006; Wingren and Borrebaeck 2006a) . In more detail, microarrays based on polyclonal and monoclonal antibodies (Sreekumar et al. 2001; Miller et al. 2003; Gao et al. 2005; Sanchez-Carbayo et al. 2006) , as well as recombinant antibody fragments (Wingren et al. 2003; Pavlickova et al. 2004; Wingren et al. 2005; Ellmark et al. 2006b ), have been successfully designed and applied for (disease) proteomics. For example, tentative protein expression profiles associated with disease and clinical parameters have been identified (Gao et al. 2005; Borrebaeck 2006; Ellmark et al. 2006b; Sanchez-Carbayo et al. 2006) .

Intense efforts are currently under way to develop the affinity protein microarray technology even further; for a review see Wingren and Borrebaeck (2004) ; Kingsmore (2006) ; Borrebaeck (2006b, 2006a) . In parallel with all the academic efforts, several commercial products have been launched, mainly for focused assays (protein expression profiling) and point-of-care applications (Wingren and Borrebaeck 2004) ; although larger arrays have also started to emerge on the market (e.g., http://www.clontech.com, http://www.raybiotech.com, http://www.protneteomix.com, http://www.whatman.com, http:// www.sigmaaldrich.com). As for functional protein arrays, mainly water-soluble analytes have been targeted (Wingren and Borrebaeck 2004) , but array designs targeting membrane proteins in the format of intact mammalian cells have recently been published as well (Belov et al. 2001 (Belov et al. , 2003 Ko et al. 2005b Ko et al. , 2005a Campbell et al. 2006; Ellmark et al. 2006a ) (Dexlin, Borrebaeck, and Wingren unpublished) .

In Table 26 .1, different applications in which various antibody-based microarray designs have been used to target bacteria and protein analytes thereof are listed. So far, the main applications are the detection of bacteria (Stokes et al. 2001; Huang et al. 2003; Gehring et al. 2006; Oh et al. 2006) , toxins (Wadkins et al. 1998; Ligler et al. 2003; Rubina et al. 2005; Rucker et al. 2005) , or both (Rowe et al. 1999; Rowe-Taitt et al. 2000; Delehanty and Ligler 2002; Taitt et al. 2002; Grow et al. 2003) ; the serotyping of bacteria (Cai et al. 2005; Anjum et al. 2006) ; and the detection of protein expression signatures associated with bacterial infections (Ellmark et al. 2006b ). To date, the degree of multiplexity differs, with array designs based on a few single probes ranging up to a few hundred, depending on the actual assay at hand. In more detail, a low degree of multiplexity (≤ 50 probes) may be sufficient to screen and detect toxins, etc., while a high degree of multiplexity (> 100 probes) may, at least initially, be required to detect, for example, disease-associated protein expression signatures.

Entering the post-genomic era, not only the proteome but also the glycome has gained significant biomedical interest (Shriver et al. 2004; Miyamoto 2006) . In recent work this has been explored to single out pathogenic bacteria (Pohl 2006) , as well as to detect pathogens (Disney and Seeberger 2004) . In these efforts, work has been made to design both lectin microarrays (Hsu et al. 2006; Uchiyama et al. 2006 ) and carbohydrate microarrays (Disney and Seeberger 2004) (Table 26 .1). Cell surface carbohydrates are in fact critical for many seminal interactions that define bacteria as pathogens and symbiotes. By adopting multiplex technologies for carbohydrate profiling, series of bacterial strains could be fingerprinted, based on their carbohydrate patterns, which might be used for identification, etc.

Lectins are sugar-binding proteins of nonimmune origin that contain at least two sugarbinding sites and are commonly used in agglutination tests to screen the bacterial glycome. However, these assays often suffer from inadequate sensitivity and subjective visual read-out. In comparison, MS, NMR, and HPLC-based analysis are alternative but time-consuming assays commonly applied for bacterial glycan analysis. In a recent paper, Hsu et al. presented a lectinbased microarray approach for analyzing the dynamic bacterial glycome (Hsu et al. 2006) . The platform was based on 21 lectins on Nexterion H slides, and the arrays were imaged by fluorescence, since the bacteria were directly labeled with SYT 85. The results showed that (1) closely related E. coli strains could be distinguished based on their glycosylation pattern, i.e., enabling fingerprinting; and (2) dynamic alterations in the bacteria glycome could be observed. The range of specificities displayed by lectins is currently a bottleneck, which is why other carbohydrate binding probes, e.g., antibodies, may provide an alternative route.

Recently, proof-of-concept was reported for an alternative lectin microarray platform by Uchiyama and coworkers (Uchiyama et al. 2006) . They have developed a novel setup that allows observation of lectin-glycoprotein interactions under equlibrium conditions, based on an evanescent-field fluorescence-assisted detection principle. This enables the assay to be performed without washing procedures, a clear advantage considering the relatively weak lectin-glycan interactions.

To examine carbohydrate-cell interactions and to detect pathogens, a carbohydrate microarray has been developed (Disney and Seeberger 2004) . In this context, it may be of interest to note that cell-surface carbohydrates are exploited by many pathogens for tissue adherence and entry into host cells. The carbohydrates (e.g., mannose and fucose) were dispensed and immobilized onto CodeLink slides, and bound bacteria (directly labeled) were detected by fluorescence. Proof-of-principle was shown for this carbohydrate microarray as a means to detect bacteria, as illustrated by different E. coli strains. In addition, these nondestructive arrays allow the bacteria to be harvested and tested for, e.g., antibacterial susceptibility.

Bacteria can be serotyped by determining their somatic (O) and flagellar (H) cell surface antigens. Serotyping is of clinical importance as, for example, many O bacterial serotypes are linked with a number of diseases (syndromes), in that subsets of serotypes, or pathotypes, can cause meningitis, systematic disease, diarrhea, etc. Despite being a primary diagnostic tool, the current methods available, e.g., agglutination tests, suffer from limited throughput, no or low multiplexity, requirement of large sample volumes, and are costly. In two recent publications, the possibility of using antibody-based microarrays for serotyping of E. coli (Anjum et al. 2006) and S. entrecia (Cai et al. 2005 ) strains have been explored and exploited (Table 26 .1).

Anjum and coworkers adopted the ArrayTube platform to develop miniaturized antibody arrays on an epoxy-modified glass surface (Anjum et al. 2006) . The authors employed 17 rabbit antisera raised against the most common E. coli pathogens (e.g., O157 and O26) associated with disease syndromes in both humans and animals. After adding the E. coli cell samples, bound cells were detected by secondary antibodies and signal amplification reagents, and the arrays were imaged by monitoring specific changes in red light transmission. This feasibility study showed that 88-100 % of the tested E. coli isolates could be correctly classified. In fact, the observed discrepancy was related to poor sample quality rather than to an inadequate identification. Hence, the results implied that the antibody array setup performed well for O serotyping, providing multiplexed, cost-effective, efficient, and accurate typing.

In comparison, Cai et al. developed an antibody array platform based on 35 polyclonal antibodies (antisera) against 20 common Salmonella serovars using SuperEpoxy slides (Cai et al. 2005) . Numerous fluorescently labeled Salmonella enterica strains were analysed and the arrays were imaged using fluorescence as a read-out system. The results showed that the array setup enabled complete serovar identification of 86 of 117 target strains, and partial identification of 30 of 117. Further, all of the 73 analysed nontarget strains (negative controls) were successfully excluded. Hence, an array platform providing a rapid and cost-effective alternative to the traditional agglutination method for Salmonella serotyping has been developed.

Foodborne pathogenic bacteria, such as E. coli O157:H7, are responsible for about 80 million illnesses in the United states each year, with thousands resulting in death. So far, the analytical approaches applied for the detection of bacteria have included plate culture, ELISA, and PCR. Hence there is a tremendous need for multiplexed technologies enabling combinations of pathogens to be screened and detected in a simple manner, and efforts to address this issue are underway (Table 26 .1).

In recent years, two independent antibody microarray or biochip setups for the detection of E. coli, using O157:H as a test system, have been developed (Stokes et al. 2001; Gehring et al. 2006 ). In the first example, a microfluidics-based antibody biochip-based system was developed for the detection of E. coli (Stokes et al. 2001 ). This reversed affinity protein microarray is based on the exposure of a cellulosic membrane to a sample potentially containing E. coli. The bacteria is then immobilized (bound) to the membrane and detected by fluorescently labeled secondary antibodies. The setup was found to display rapid and selective detection of bacteria with at least three orders of magnitude linear dynamic range. Of note, the assay sensitivity was found to be as low as 20 organisms (in this case E. coli O157:H7). In comparison, Gerhing and coworkers have developed a sandwich fluorescent immunoassay in the microarray format (Gehring et al. 2006) . Biotinylated capture antibodies were immobilized onto streptavidin-modified Superfrost Gold slides, and the setup was evaluated targeting E. coli O157:H7 samples. The applicability of this, so far, low-density setup was outlined and a limit of detection in the 3 × 10 6 cells/ml range was observed.

Similar to the setup by Gerhing et al. (2006) , Oh and colleagues have developed a reversed affinity protein array setup for the detection of bacteria (Oh et al. 2006 ). Using a microfluidic device, the cells were electrokinetically immobilized onto gold electrodes and imaged by fluorescence after adding appropriate secondary reagents (antibodies). It should, however, be noted that this setup was not primarily developed for the detection of bacteria, but rather as a new tool for taking advantage of the bacteria to display a capture protein, e.g., a membrane protein, in its natural environment, and thereby increasing its on-chip functionality.

In this context, it should be noted that designing and fabricating membrane protein microarrays is in general a major challenge that remains to be fully resolved (Fang et al. 2002a (Fang et al. , 2002b Wingren and Borrebaeck 2004) . Hence, this setup might open new avenues not only for the detection of bacteria, but also for designing membrane protein-based microarrays.

In recent work, an antibody microarray setup based on biotinylated monoclonal antibodies was developed and optimized with respect to nonspecific adsorption of bacteria and proteins thereof. The antibodies were immobilized via streptavidin, which in turn was bound to biotinylated bovine serum albumin (BSA) adsorbed onto a C 18 -derivatized SiO 2 surface . The results showed that the dual action of BSA, acting both as a surface blocker and as a probe immobilized, was successful. Directed immobilization and low nonspecific binding were reported.

Rapid, sensitive, and multiplex detection of biological toxins in clinical samples, food, drinking water, and environmental samples is of great importance in revealing possible infections and contaminations, as well as potential bioterrorist threats (Table 26 .1). In the long-term, small, simple, and portable devices (biosensors) would be an attractive format for the instrumentation behind such key applications. In this context, the sample format will also be critical, and assay designs allowing the sample to be directly applied without any significant pretreatment, e.g., fluorescent labeling, would clearly be advantageous.

Four studies have been published, in which the efforts at developing antibody-based microarray biosensors for the detection mainly of toxins have been successfully described (Wadkins et al. 1998; Ligler et al. 2003; Rubina et al. 2005; Rucker et al. 2005) . In an early study by Wadkins et al. a planar array immunosensor for the detection of multiple toxic agents was fabricated (Wadkins et al. 1998) . Polyclonal antibodies were covalently coupled to derivatized glass slides, and bound toxins, e.g., ricin and staphylococcal enterotoxin B, were monitored using fluorescently labeled secondary antibodies. Assay sensitivities in the 5-25 ng/ml range were observed.

In comparison, an array biosensor, based on monoclonal and polyclonal antibodies, capable of detecting multiple targets on the borosilicate glass surface of a single waveguide, was more recently designed (Ligler et al. 2003) . Both competitive and sandwich fluoroimmuno setups were developed to enable small, as well as large, molecular weight toxins (e.g., ricin, botulinum toxoids, and trinitrotoluene) to be detected in complex samples, such as food or clinical specimens. Notably, the setup was capable of addressing up to 12 samples at the same time. The results showed that specific and sensitive (≥0.5 ng/ml) detection of target analytes was accomplished. With additional development of the sensor instrument, this may in the end provide a rapid, fieldable, and low-tech assay for the detection of toxins.

Similarly, Rucker and colleagues have developed competitive and noncompetitive antibody microarray setups for native toxin detection (e.g., diphtheria toxin and anthrax lethal factor) in serum samples (Rucker et al. 2005) . In this case, monoclonal antibodies were immobilized on epoxy-slides, and the arrays were imaged using fluorescence. While the competitive assay setup was favored for not having to label the sample, the direct assay benefited from superior sensitivity (low ng/ml vs. high ng/ml). In the end, the choice of setup may be dependent on whether the assay is run in field trials, where a simple assay is desired, or in the laboratory, where more complex assay principles providing higher sensitivity can be applied.

Further, a hydrogel-based monoclonal antibody microchip was recently designed and fabricated by Rubina et al. (2005) . The platform was developed with the aim of performing a quantitative immunoassay of a series of plant toxins (e.g., ricin and viscumin) and bacterial toxins (e.g., diphtheria toxin and tetanus toxin). Direct, competitive, and sandwich assay setups were successfully developed and found to be compatible with the platform. In contrast to the previous studies, this system was interfaced with either a fluorescent-, chemiluminescent-, or MS-based read-out system, providing high flexibility. In all cases, the assay sensitivities were found to be in the low ng/ml range, i.e., within the range of sensitivity required in order to be able to perform clinical applications.

Similar to the work described in Section 26.2.3, antibody-based microarray biosensors have also been used for the simultaneous detection of bacteria and bacterial proteins (e.g., toxins) (Table 26 .1), where again technologies for rapid and multiplexed detection will play a key role.

An antibody-based array biosensor composed of three parts, the antibody array (recognition element), an image capture and processing part, and an automated fluidics unit, has been developed by Ligler et al. (Rowe et al. 1999; Rowe-Taitt et al. 2000; Taitt et al. 2002) . The capture polyclonal and/or monoclonal antibodies were biotinylated and immobilized on neutravidin-derivatized waveguides. Bound analytes (proteins, glycoproteins, Gram-negative, and Gram-positive bacteria) were detected using labeled secondary (tracer) antibodies. The results showed that assay sensitivities in the mid ng/ml range were readily observed targeting, for example, cholera toxin and B. globigii. Moreover, the assay was demonstrated to be rapid (<15 min) and easy to execute. Taken together, these studies have demonstrated proof-of-concept for an inexpensive and multiplex device for simple detection of bacteria and bacterial analytes. In addition, the setup is in a format amenable to automation and portability. In these first studies, the capture antibody spots were in the 2.5 mm 2 size range and generated by physically isolated patterning using polymer flow cells. In recent work, the spot size of the biotinylated capture antibodies has been considerably reduced (0.04 mm 2 ) by adopting a noncontact piezo-based dispenser to fabricate the arrays (Delehanty and Ligler 2002) . Using confocal microscopy for detection, an assay sensitivity in the low ng/ml range was still obtained. Hence the latter study outlined a way of fabricating high-density arrays for bacterial detection, while maintaining assay sensitivity.

In a recent review by Grow et al. a new biochip technology for label-free detection of pathogens and their toxins was presented and discussed (Grow et al. 2003 ). The biochip is composed of spots of capture probes (e.g., antibodies) immobilized on a surface-enhanced Raman scattering (SERS) active metal surface. Once the chip has been subjected to sampling and target analytes have been bound, a Raman microscope is applied to collect SERS fingerprints from the spots (pixels) on the chip. This interesting technology has been named SERS, as it couples SERS with microscopy. The identification is based on SERS fingerprints, and the authors demonstrated that both Gram-positive and Gram-negative bacteria often could be detected at the strain/subspecies level based on their SERS fingerprints. Further, the SERS fingerprints could also be used to differentiate viable vs. nonviable, e.g., heator UV-killed, microorganisms; different physiological states of the bacteria cells, e.g., when cultured under conditions known to affect virulence; and to detect toxins in a specific and sensitive manner. Work is currently under way to develop the Raman microscope instrumentation even further, to enable the simultaneous collection of hundreds or thousands of spectra from discrete positions on the chip with a spatial resolution of 250 nm to 1.5 m. Future experiments will unravel the potential of this read-out system for protein microarray-based applications.

In two recent publications (Li et al. 2005; Steller et al. 2005) , the possibility of using protein microarrays to identify novel potential diagnostic markers and/or vaccine candidates was explored and outlined (Table 26 .1). Again, the array format was critical in order to enable sufficient multiplexity and throughput to gain success.

In the first study, a recombinant bacterial protein microarray was fabricated on FASTslides and used for identification of new potential diagnostic markers for Neisseria meningitides (Steller et al. 2005) . The authors succeeded in expressing 67 of 102 known phase-variable genes from N. meningitides serogroup B strain MC58 as recombinant proteins in E. coli. Subsequently, these proteins were used as probes in the array format and applied to screen sera from healthy controls vs. patients suffering from meningitis. The results showed that 47 of these proteins were immunogenic, i.e., that an antibody response had been mounted. Nine proteins were found to be immunogenic in at least 3 of 20 meningitis sera tested, while 1 protein showed a response in 11 of 20 sera. The potential of these N. meningitis proteins for diagnostic purposes remains to be elucidated, but this study clearly outlines the potential of the approach.

Yersina pestis causes plague, which is one of the most feared diseases. Work is ongoing to identify novel vaccine candidates to improve the current plague vaccines. In these efforts, Li et al. have developed a 149-recombinant Yersina pestis protein microarray to profile the antibody response in immunized rabbits, providing a new tool in the search for vaccine candidates and/or diagnostic antigens (Li et al. 2005) . The authors found that an antibody response had been elicited against about 50 of the arrayed Y. pestis proteins. Among these 50, 11 proteins to which the predominant antibody response was directed were identified. Taken together, these 11 new proteins show promise for further evaluation as candidates for vaccines and/or diagnostic antigens.

The evaluation of serological reactivity from healthy vs. nonhealthy patients, in order to allow disease state differentiation and the identification of tentative diagnostic markers, has gained significant attention within the field of disease proteomics (Hanash 2003; Wingren and Borrebaeck 2004; Borrebaeck 2006) . Focusing on bacterial related diseases, two independent protein microarray setups, focusing on tuberculosis, have been developed (Tong et al. 2005; Sartain et al. 2006 ) and applied to perform serological tuberculosis assays (Table 26 .1). Tuberculosis can be diagnosed by microscopy and culture of mycobacteria of the Mycobacterium tuberculosis complex from clinical samples. Still, these approaches are associated with limitations and technical hurdles. To be proven valuable, a serodiagnostic approach should (1) display a specificity > 90 % (i.e., comparable to microscopy and bacterial cultures), and (2) be able to differentiate/detect multiple disease states.

Interestingly, Tong and coworkers have developed a protein microarray setup based on 54 M. tuberculosis antigens on epoxy-slides, and the arrays were imaged by fluorescence (Tong et al. 2005) . The probe antigens were obtained from five sources, including biochemical fractionation of M. tuberculosis cells/culture fluids, oligosaccharides bound to BSA, purified lipopolysaccharides, purified polysaccharides, and recombinant antigens. The clinical serum samples from healthy controls (non-TB) and tuberculosis (TB) patients were screened for IgG antibodies specific for any of these antigens, e.g., for a serum-specific IgG profile. Based on the analysis of 20 TB sera and 80 non-TB sera, combinations of TB antigens were ranked with respect to specificity and sensitivity of TB detection. The results showed that the highestranking TB antigen combination displayed a receiver operator curve (ROC) with an area under the curve (AUC) of 0.95. Of note, a single antigen, Ara 6 -BSA, was found to give an AUC value of 0.90. The authors concluded that the TB antigen microarray provided a rapid and efficient means of finding TB antigens that could be used to discriminate between TB and non-TB patients.

In comparison, Sartain et al. fabricated a TB antigen microarray based on 960 unique fractions, obtained from M. tuberculosis cytosol and culture filtrates, by multidimensional protein fractionation (Sartain et al. 2006) . TB antigen arrays were fabricated on FAST slides and interfaced with a fluorescent read-out system. Next, serum samples from 12 healthy individuals, 9 noncavitary TB patients, 11 cavitary TB patients, 10 HIV-positive TB patients, and 6 HIVposititve TB-negative patients were analysed. The authors demonstrated that the TB antigen microarray setup provided them with a novel means of assessing antigen recognition profiles (e.g., specific IgG profiles) discriminating between different disease states. In more detail, the different sera were found to display partly overlapping reactivity patterns, e.g., containing antibodies specific for material in the arrayed subfractions, but also distinctly unique patterns. Hence, the results indicated that the setup could be useful for differentiating the different disease states, thus demonstrating the potential of array-based serodiagnostics for tuberculosis.

The use of high-density protein microarrays to examine the protein-protein interaction patterns for bacterial toxins in a multiplex high-throughput manner, is very appealing (Table 26 .1). In the end, this may allow for identification of novel toxin modulators and/or regulators.

In a recent paper by Galindo and colleagues, the potential of cytotoxic enterotoxin (Act) of Aeromonas hydrophila to bind to human and yeast proteins was investigated by adopting a protein microarray approach (Galindo et al. 2006) . To this end, the human and yeast ProtoArrays composed of 1869 human proteins and 4319 yeast proteins, respectively, on nitrocellulose coated slides were used. The study showed that Act was capable of binding nine human proteins and 4 yeast proteins. For three of the interactions, a confirming Western blot analysis was performed. Next, a set of experiments, including small interfering RNA, was performed in order to explore the relevance of the observed interactions. These efforts indicated a potential involvement of galectin-3 and SNAP23 in A. hydrophila cytotoxic enterotoxin-induced host cell apoptosis. Hence, by adopting a high-density protein microarray screening approach, the authors were able to present the first report of tentative protein binding partners for Act, as well as potential mediators/regulators for Act-induced apoptosis.

To date, a major focus has been placed upon using protein microarrays, and in particular antibody-based microarrays for oncoproteomics, with the aim of finding disease-specific (serum or tissue) protein signatures for diagnostics and biomarker discovery, etc. (Wingren and Borrebaeck 2004; Borrebaeck 2006; Kingsmore 2006) . In a similar fashion, protein microarrays could be used to find protein expression signatures associated with bacterial infections and conditions (Table 26 .1).

In a recent study by Ellmark et al. the authors examined stomach tissue samples from gastric adenoma carcinoma patients using a 127-human recombinant scFv antibody microarray on black polymer Maxisorb slides, interfaced with a fluorescent read-out system (Ellmark et al. 2006b ). Of note, these cancer patients are often associated with Helicobacter pylori infections. The proteins were extracted from the tissue samples, biotinylated, and analysed on the recombinant antibody microarrays. The platform was found to display an assay sensitivity in the low pg/ml range. Further, the results showed that a 14-protein expression signature associated with H. pylori infection could be identified, where 10 analytes were distinctly different from the corresponding protein signature found to be associated with adenoma carcinoma. Taken together, these studies clearly demonstrate the use and potential of antibody (protein) microarray technology for rapid, sensitive, and multiplexed expression profiling of complex samples in order to identify disease-associated protein signatures.

Taken together, the first generations of protein-based microarray technology platforms for detection and identification of bacteria, bacterial proteins and bacterial diseases, have in recent years been developed. These miniaturized assay platforms include functional protein microarrays as well as affinity protein microarrays, and the designs range from small-scale focused biosensors targeting a few analytes to large-scale semi-dense microarray set-ups for multiplex screening. A broad range of applications, such as serotyping of bacteria, detection of bacteria, identification of toxins, disease state differentiation, and discovery of diseaseassociated biomarkers, have so far been demonstrated, clearly outlining the potential of the technology. Still, the number of applications is low, but is likely to increase significantly as the microarray technology progress and develops into a robust proteomic technology. In future, protein microarray based applications are likely to play an important role for detection and identification of bacterial and protein analytes.

Progress in protein and antibody microarray technology

Use of miniaturized protein arrays for Escherichia coli O serotyping

Immunophenotyping of leukemias using a cluster of differentiation antibody microarray

Identification of repertoires of surface antigens on leukemias using an antibody microarray

Antibody microarray-based oncoproteomics

Development of a novel protein microarray method for serotyping Salmonella enterica strains

Cell interaction microarray for blood phenotyping

A microarray immunoassay for simultaneous detection of proteins and bacteria

Detection of antigen-specific T cells on p/MHC microarrays

The use of carbohydrate microarrays to study carbohydrate-cell interactions and to detect pathogens

Multiplex detection of surface molecules on colorectal cancers

Identification of protein expression signatures associated with H. pylori infection and gastric adenocarcinoma using recombinant antibody microarrays

Membrane protein microarrays

Membrane biochips

Potential involvement of galectin-3 and SNAP23 in Aeromonas hydrophila cytotoxic enterotoxin-induced host cell apoptosis

Distinctive serum protein profiles involving abundant proteins in lung cancer patients based upon antibody microarray analysis

Antibody microarray detection of Escherichia coli O157:H7: quantification, assay limitations, and capture efficiency

New biochip technology for label-free detection of pathogens and their toxins

Advances in protein microarray technology for protein expression and interaction profiling

Methods and applications of antibody microarrays in cancer research

Applications of antibody array platforms

Disease proteomics

Profiling humoral autoimmune repertoire of dilated cardiomyopathy (DCM) patients and development of a disease-associated protein chip

Analyzing the dynamic bacterial glycome with a lectin microarray approach

Composite surface for blocking bacterial adsorption on protein biochips

Antigen microarray profiling of autoantibodies in rheumatoid arthritis

Multiplexed protein measurement: technologies and applications of protein and antibody arrays

Antibody microarray for correlating cell phenotype with surface marker

Parallel analysis of multiple surface markers expressed on rat neural stem cells using antibody microarrays

Protein microarrays as tools for functional proteomics

Protein microarray for profiling antibody responses to Yersinia pestis live vaccine

Array biosensor for detection of toxins

Biochip sensors for the rapid and sensitive detection of viral disease

Screening of specific antigens for SARS clinical diagnosis using a protein microarray

Protein biochips: A new and versatile platform technology for molecular medicine

Protein microarrays and proteomics

Printing proteins as microarrays for high-throughput function determination

Antibody microarray profiling of human prostate cancer sera: antibody screening and identification of potential biomarkers

Clinical applications of glycomic approaches for the detection of cancer and other diseases

Microfluidic protein detection through genetically engineered bacterial cells

Zeptosens protein microarrays: a novel high performance microarray platform for low abundance protein analysis

Advances in recombinant antibody microarrays

A combined oligonucleotide and protein microarray for the codetection of nucleic acids and antibodies associated with human immunodeficiency virus, hepatitis B virus, and hepatitis C virus infections

Protein analysis on a proteomic scale

Protein microarrays: catching the proteome

Array methodology singles out pathogenic bacteria

Protein arrays for autoantibody profiling and fine-specificity mapping

Array biosensor for simultaneous identification of bacterial, viral, and protein analytes

Array biosensor for detection of biohazards

Quantitative immunoassay of biotoxins on hydrogel-based protein microchips

Antibody microarrays for native toxin detection

Profiling bladder cancer using targeted antibody arrays

Disease state differentiation and identification of tuberculosis biomarkers via native antigen array profiling

Glycomics: a pathway to a class of new and improved therapeutics

Profiling of cancer cells using protein microarrays: discovery of novel radiation-regulated proteins

Bacterial protein microarrays for identification of new potential diagnostic markers for Neisseria meningitidis infections

Detection of E. coli using a microfluidics-based antibody biochip detection system

Nine-analyte detection using an array-based biosensor

A multiplexed and miniaturized serological tuberculosis assay identifies antigens that discriminate maximally between TB and non-TB sera

Development of a lectin microarray based on an evanescent-field fluorescence principle

Detection of multiple toxic agents using a planar array immunosensor

Antibody microarrays-current status and key technological advances

Recombinant antibody microarrays

High-throughput proteomics using antibody microarrays

Design of recombinant antibody microarrays for complex proteome analysis: choice of sample labeling-tag and solid support

Recombinant antibody microarrays-a viable option

Microarrays based on affinitytagged single-chain Fv antibodies: sensitive detection of analyte in complex proteomes

Functional proteomics: current achievements

Development and evaluation of a protein microarray chip for diagnosis of hepatitis C virus

Global analysis of protein activities using proteome chips

Severe acute respiratory syndrome diagnostics using a coronavirus protein microarray

Analysis of yeast protein kinases using protein chips

Protein chip technology

This study was supported by grants from the Swedish National Science Council (VR-NT), the SSF Strategic Center for Translational Cancer Research (CREATE Health), the Alfred Österlund Foundation, and the Great and Johan Kock Foundation.