Monitoring the Assembly of a Secreted Bacterial Virulence Factor Using Site-specific Crosslinking

It is often essential to identify and characterize interactions between a protein of interest and other cellular components to define its role in a biological pathway, to understand its assembly, or to elucidate its mechanism of action. A wide variety of approaches are commonly used to study protein-protein interactions, but all have their limitations. The simplest unbiased method to identify proteins that bind to a protein of interest is copurification (or coimmunoprecipitation) but this approach requires that a protein complex remain intact during the purification procedure. Weak or transient protein interactions can be stabilized through chemical cross-linking, but this method typically requires the use of compounds that link proteins through primary amines that are widely scattered in the protein sequence. Complicated cross-linking patterns that are difficult to interpret can be generated, especially when proteins of interest are components of multiprotein complexes. Furthermore, because the spacer arms of chemical cross-linkers generally exceed 10 Å in length, covalent bonds can be formed with proteins that are located in proximity to but do not interact directly with the protein of interest. Methods such as affinity chromatography and two-hybrid screens are also often useful to identify protein-protein interactions. The former approach, however, requires reproducing conditions that promote physiologically significant interactions and cannot be used to detect weak interactions. The latter approach requires that binding interactions can be replicated in the host organism and are maintained when a protein of interest and its binding partners are placed in the context of a fusion protein. Two-hybrid methods also tend to generate false-positive results². Once a protein-protein interaction has been established, considerable work is often required to map the site of interaction. Perhaps the most significant drawback of traditional approaches is that they do not provide any information about the temporal sequence of intermolecular interactions or the kinetics of binding.

This paper describes a simple method that overcomes some of the limitations of other approaches that are used to study protein-protein interactions. Initially a single amber mutation is introduced into the gene that encodes a protein of interest. The amber mutant is then coexpressed with an amber suppressor tRNA and an amino acyl-tRNA synthetase derived from Methanococcus jannaschii that have been engineered to incorporate the photoactivatable amino acid analog p-benzoylphenylalanine (Bpa) only at amber codons³. Irradiation of cells with long wavelength ultraviolet (UV) light (~365 nm) then facilitates formation of a covalent bond between Bpa and proteins that are within ~3-8 Å^4,5. In some experimental contexts, cross-links can also be formed between activated Bpa and nonprotein components of cells such as lipids and nucleic acids⁶. Molecules that are terminated at the amber codon should generally be easily distinguishable from the full-length protein on SDS-PAGE and cannot be cross-linked to other proteins. By subjecting cells to pulse-chase radiolabeling prior to cross-linking and then immunoprecipitating or affinity purifying cross-linking products, the sequence and duration of protein-protein interactions can be established. The ability to generate temporal information about intermolecular interactions is especially valuable to study the progression of a protein of interest through any ordered multistep assembly, trafficking or signaling pathway and to identify pathway intermediates. Like chemical cross-linking, site-specific cross-linking detects protein-protein interactions that occur under physiological conditions, but the introduction of a cross-linker at a single position facilitates the analysis of interactions between individual segments of a protein and other factors. This unique feature of site-specific cross-linking is especially advantageous when the protein of interest has multiple segments or domains that have the potential to interact with distinct binding partners. Furthermore, site-specific cross-linking can be used in combination with methods such as mass spectrometry to map the residues that mediate protein-protein interactions with precision. Although this method is optimized for use in E. coli, the incorporation of photactivatable amino acid analogs into proteins by amber suppression has been reported in Mycobacterium, yeast and mammalian cells^7-11. Thus, in principle our method can be used in other systems.

We have used this method extensively to identify intermolecular interactions that facilitate the biogenesis of EspP, an E. coli O157:H7 virulence factor. EspP is a member of a superfamily of proteins known as "autotransporters" that contain a large N-terminal extracellular domain ("passenger domain") and aC-terminal domain ("b domain") that folds into a cylindrical b barrel structure and anchors the protein to the outer membrane (OM)¹². The mechanism by which the passenger domain is translocated across the OM has been a long-standing mystery. Like all bacterial cell surface proteins, EspP must be transported across the inner membrane, shuttled across the periplasm (the space between the inner and outer membranes), and targeted to the OM in a series of ordered events. Following its translocation across the OM, the EspP passenger domain is also separated from the b domain by a proteolytic cleavage and released from the cell surface. By combining site-specific cross-linking with pulse-chase radiolabeling, we have shown that both domains of the protein initially interact with the molecular chaperone Skp in the periplasm⁶. Subsequently, discrete segments of the b domain interact in a stereospecific fashion with components of a heterooligomer called the Bam complex, which catalyzes the integration of b barrel proteins into the bacterial OM by an unknown mechanism. Perhaps most interestingly, we have found that the passenger domain is in contact with one of the Bam complex subunits (BamA) as it traverses the OM¹³. Our results have enabled us to construct a detailed model for autotransporter assembly¹⁴ and have implicated the Bam complex in the transport of the passenger domain across the OM.