The ORFeome of S. aureus
that we constructed in this work consists of 1753 ORFs maintained as homogeneous plasmid DNAs purified from single colonies plus 809 ORFs purified from separate 'pools' of entry clones. These 2562 cloned ORFs represent 95% of the S. aureus
(Mu50) ORFeome, a proportion of coverage similar to those reported for the ORFeomes of three other bacterial pathogens, Brucella melitensis
], Treponema pallidum
], and Francisella tularensis
]. PCR amplification of the ORFs produced products not only of the expected size, but also shorter ones, which were probably primer dimers [20
]. Although we did not purify the PCR products as in other ORFeome projects [21
], enough material was produced for individual BP reactions, all of which resulted in colonies after transformation. Testing of the effectiveness of both purified and unpurified PCR products in BP reactions showed that cloning could be achieved to the same extent with both types of PCR products. To minimize the occurrence of PCR-induced mutations, we optimized the PCR conditions, used a proofreading-capable DNA polymerase, and kept the number of amplification cycles low.
Entry clones are the basis for further experiments using recombinational cloning, including genome-wide protein-protein interaction (PPI) analysis, protein expression, and protein localization studies [22
]. The single entry clones are of higher quality than the pools of entry clones because they are homogeneous, and LR reactions performed subsequently with the pure clones will be more efficient. However, for reasons of economy and because of time constraints, it was not possible to screen every colony to obtain a pure isolate of the appropriate ORF-containing entry clone. When the first colony-PCR failed to produce the anticipated ORF, we screened the remaining pool of colonies to identify the presence of the expected entry clone. We failed to produce PCR products for 135 ORFs. This could be due to possible errors in annotation of the Mu50 genome or, more likely, because of nucleotide differences between our clinical S. aureus
isolate and the sequenced Mu50 genome or for technical reasons related to primer design and effectiveness specific to each missing ORF.
Different strains of S. aureus
can exhibit important differences in gene content owing to the presence/absence of plasmids and/or mobile elements integrated into the chromosome [e.g., staphylococcal cassette chromosome (SSC) elements, prophage], usually at specific loci termed "genomic islands" [27
]. Such factors are important determinants of MRSA virulence and drug resistance, so that an ideal ORFeome would include ORFs for these genes in addition to the bacterial core ORFs. Our ORFeome is certainly incomplete in this regard, because it is a composite of only two S. aureus
strains; however, ORFs for virulence and resistance factors of interest could easily be added to the ORFeome as the resource is further characterized and developed.
We assessed the quality of the ORFeome by PCR and agarose gel-electrophoresis of the PCR-products for all the entry clones obtained. Sequencing a sample of 300 PCR products confirmed the identity of the entry clones. Functional validation was done by recombinant protein production and PPI analysis using the Y2H system. In addition to identifying 22 new homodimeric interactions, the experiments also showed the functionality of the ORFeome and its suitability for further applications, including automated high-throughput protein purification or ORFeome-wide PPI analysis. Indeed, generation of an S. aureus
intrapathogen 'interactome' is conceivable through ORFeome-wide PPI screening of 3.3 million (2562 × 2562/2) protein-protein interactions by automated procedures. Likewise, a system-wide screen for host-pathogen PPIs could be conducted by screening against other available ORFeomes, e.g. the human ORFeome [28
], which would produce a detailed description of the host-pathogen interface at the molecular level, leading to improved understanding of S. aureus
pathogenesis in humans.
Production of recombinant proteins in the present system is achieved with the two well-known protein fusion tags, 6 × His and GST. These tags could be used for further protein purification or affinity studies such as GST Pull-Down [29
] for validation of protein-protein interactions. We successfully produced fusion proteins with both types of tags, although the GST-tagged proteins were more strongly expressed on average than the His-tagged ones in our experiments. His-tagged proteins could not be detected in three out of 20 experiments by Western blotting, due to high background staining caused by the His-specific antibody used or because of low expression rates, or both. Although, we loaded equal amounts of total E. coli
protein lysates on protein gels, the intensity of the expressed proteins varied because of degradation in the crude cell lysates. It is interesting to note the presence in some experiments of a second band double the size of the expected molecular weight, suggesting homodimerization, although it is unlikely that dimerization would occur under the denaturation conditions used.
We tested 150 proteins for homodimerization using the Y2H-system and found a total of 22 such interactions (Table ) after ruling out 37 autoactivators. All Y2H-experiments, including the autoactivator tests, were performed in triplicate, and each PPI was reproducible by independent analysis. It is important to stress that even in the cases of autoactivation, an ORF had to have been expressed in the yeast reporter strain, which again demonstrates the functionality of the ORFeome. The 22 homodimers observed here imply that at least ~15% (22 of 150) of the tested proteins can undergo homodimerization, consistent with the levels found in previous large-scale studies (2–20%) of different organisms [S. cerevisiae
3.6%, C. elegans
2.8%, D. melanogaster
2.2%, and H. sapiens
]; 143 homodimers among 1546 PPIs (9.25%, Y2H-dataset) in H. sapiens
]; 33 homodimers in 1301 human PPIs (2.5%) [32
]. Moreover, in the literature we found reports of biochemical and/or structural evidence for homodimerization among homologs (Interologs) of 6 of our 22 dimer-forming proteins (Table ). For example, RNase III, the product of the SAV1233 clone, is a homodimer in numerous species [33
]. In S. aureus
, RNase III is an essential regulator of expression of the spa
gene, which encodes the virulence factor Staphylococcal protein A[34