Background

Protein microarrays are an excellent tool for identifying disease-associated antibody reactivity patterns since they allow the simultaneous detection of antibody binding to a large number of antigens, up to entire bacterial proteomes (Hufnagel et al., 2018) and beyond (Syafrizayanti et al., 2017). Protein microarrays used to be generated with proteins that were isolated from recombinant cellular libraries. This approach is work-intensive, time-consuming and costly, however, since it requires the individual handling of very many cells. Each gene of interest is amplified by a polymerase chain reaction (PCR), followed by cloning into an expression vector. After transformation into a cellular system–frequently Escherichia coli (E. coli)–the cell clones are grown separately from others and characterized. Subsequently, the respective overexpressed protein is isolated and purified, followed by immobilization on the microarray. Our largest array of a non-mammalian gene set consisted of 14,000 proteins (Syafrizayanti et al., 2017), for example. Therefore, producing this array would have required going through the above process 14,000 times.

In order to overcome the complex procedure, means have been developed for the production of protein microarrays by cell-free protein expression directly on the microarray surface. After the initial publication of the method (Protein in situ Arrays, PISA) (He and Taussig, 2001), several adaptations have been reported (Ramachandran et al., 2004; Angenendt et al., 2006; He et al., 2008). All have in common that genes are copied into DNA expression constructs by PCR, which are directly placed onto the microarray surface. Proteins are then synthesized in situ using cell lysates containing all elements and ingredients required for transcription and translation. All steps are done in parallel and in a largely automated manner. No cells are involved throughout. In addition, the approach eliminates the need for protein purification.

Besides many minor variations, our protocol differs from the others mainly by the fact that multiple spotting is applied to produce protein microarrays (Angenendt et al., 2006; Syafrizayanti et al., 2017). First, the DNA expression constructs are placed on planar glass slides. In a second step, each position is revisited to deposit in a second spotting event a cell-free transcription and translation mixture on top of the DNA spots. Proteins are expressed directly on the microarray slide and bind immediately. The double-spotting protocol decreases substantially the volume of cell lysate needed for protein expression since the empty surface in between spots is not covered. Concomitantly, there is less background signal in between the spots eventually. More importantly, however, the reaction space is restricted to the locations of the DNA expression constructs. In consequence, expressed protein cannot float away but binds in situ only. No additional means, such as antibodies, are required to keep the proteins attached to their exact positions. Nevertheless, even extensive washing will not remove them. Previous studies have shown that most proteins are expressed in full-length, and very many fold into a functionally active conformation. Also, the process is versatile and can be adapted to fit to a variety of applications (Syafrizayanti et al., 2017).

Besides studies on microarrays with human or parasite proteins, we have used arrays that represent entire bacterial proteomes to screen clinically characterized patient (and control) sera in order to identify antibody binding patterns that are disease-specific. New marker molecules for disease diagnosis and prognosis were discovered in this way, including markers for cancer entities that are associated with bacterial infections. In a recent study, microarrays were produced that present all proteins of Chlamydia trachomatis (Ct), for example (Hufnagel et al., 2018). Through comparison of the antibody binding patterns on the microarrays, we identified antigens that are reproducibly recognized by the antibodies in Ct-seropositive samples, in addition to molecules that were patient-specific. With samples from cervical cancer patients, we found the common Ct-specific markers and additional antigens, which were shared between the cancer patients only. Large-scale validation experiments using high-throughput suspension bead array serology on hundreds of samples confirmed the significance and relevance of the newly identified markers for both general Ct infection and related cervical cancer. Next to providing meaningful marker molecules, the results strongly support the hypothesis that there is an association of Ct infection with cervical cancer.

The quick and efficient proteome-wide immunoassay can easily be adapted to other microorganisms in all areas of infection research. By reverse transcribing RNA to DNA-templates, the protocol is applicable to any transcript isolate. We also have utilized established ORF libraries, such as the Human ORFeome library (The ORFeome Collaboration, 2016), by which the process of microarray production is simplified further since only one vector-specific primer pair is required for the initial PCR. Here, we provide a detailed protocol for the production of whole-proteome bacterial microarrays and their application to screening patient sera for the immune response in infected people.


Figure 1. Images of protein microarrays made from bacterial genes. About 1,500 proteins were produced on each array. Blue boxes indicate controls. A-B. Two examples of analyses are presented that were performed using the protocol described below. Signals that were common to a group of patients are labeled by green boxes. C-E. Few examples of technical problems are presented. C. An artifact is shown, which was due to a technical fault of the spotting device, resulting in a signal gradient across the microarray. D. In a close-up, background signals on 40 microarray positions are shown. The only difference between the top and bottom half was the quality of the DNA-polymerase used for PCR. After spotting unpurified PCR-products, incubation with patient serum and subsequent labeling produced significant differences in background signal. E. A complete protein microarray is shown after incubation with patient serum and detection of antibody binding. Because of inadequate blocking, a relatively high level of background signal was produced.