2.1. Interfaces revealed by UAA incorporation and crosslinking
Incorporation of a photo-reactive UAA into a target protein in live cells followed by controlled crosslinking in situ provides a great advantage of identifying target protein-specific binding partner(s) in a native environment (Figure 2A). Importantly, it circumvents the need for target protein extraction and purification in aqueous solution before crosslinking, which is almost impossible for some proteins, especially those in cellular membranes like G protein-coupled receptors (GPCRs). GPCRs are the largest family of integral membrane proteins with a characteristic structure comprising seven transmembrane helix folds, three intracellular loops together with the N-terminus, and three extracellular loops together with the C-terminus (Cvicek et al., 2016). Membranous and flexible natures of GPCRs usually compromise the sample preparation for classical structure analyses that require a substantial sample amount and purity. Even though dozens of GPCR structures have been determined, most of them are believed to represent a single snapshot of GPCR that otherwise would wiggle vigorously in a native environment.
CXC chemokine receptor 4 (CXCR4) is a pathologically and clinically important GPCR in cancers, inflammation, and viral infection. To probe the binding site of T140, a 14-residue cyclic peptide HIV-1 entry blocker, CXCR4 was site-specifically mutated with a photoactivatable UAA, p-benzoyl-l-phenylalanine (BzF), at various positions one-by-one in live cells (Grunbeck et al., 2011). Co-incubation of the fluorescently labeled T140 with CXCR4-transfected cells was followed by UV exposure to trap a CXCR4-T140 complex. A series of experiments for each CXCR4 mutant showed that the crosslinked complex was formed only when BzF had been incorporated at Phe189. Based on a characteristic 3 Å-long reactivity distance of BzF as well as the positional information gained from the crosslinking analyses, molecular modeling could successfully derive a more accurate model of the CXCR4-T140 interface modified from available crystal structures. Similarly, strategic incorporation of photo-crosslinking UAA allowed identification of contact residues of CCR5 in complex with a small molecule HIV-1 entry inhibitor, maraviroc (Grunbeck et al., 2012). Newly discovered contact residues previously not recognized in the computational modeling could serve as a cue to reconstruct the interface taking the dynamic allosteric binding of maraviroc into account (Grunbeck et al., 2012). Although the UAA-mediated crosslinking studies mentioned above lacked the mass spectrometric analysis, a genetically well-defined position of crosslinking and the ability to incorporate the UAA in any position of interest could make a significant contribution to remodeling GPCR-ligand structures to display the biological interface. The most attractive aspect of the method in studying the structural biology of GPCRs lies in that important contact sites can be identified even when GPCR behaves dynamically in a native cellular environment in response to different biological stimulation. By analyzing 25 different mutants of the angiotensin II type 1 receptor (AT1R) containing a photo-reactive UAA, Gagnon et al. demonstrated that AT1R exhibits structurally differential binding modes, i.e., distinct conformations or residues, with an intracellular β-arrestin regulator depending on the presence of an extracellular ligand, angiotensin II (Gagnon et al., 2019).
Photo-reactive UAA incorporation technique in live cells readily allows investigation of a dynamic interface hidden in the transmembrane domain which is much more challenging to recombinantly prepare in vitrothan the extracellular domain. CGRP is a ligand that binds to the CLR/RAMP1 receptor. While the crystal structure of the extracellular domain of CLR/RAMP1 in complex with CGRP and corresponding contact residues had been available, the interface located in the transmembrane domain remained to be probed. Crosslinked residue screening by photo-reactive UAA crosslinker incorporation at multiple potential contact sites in the CLR transmembrane helix domain identified major contact residues, providing insight into the extent of CGRP penetration into the transmembrane core of CLR (Simms et al., 2018). It should be noted that, besides GPCR, most protein complexes recombinantly expressible in E. coli or mammalian cells can benefit from techniques reviewed above for investigation of biological interfaces regardless of their size, cellular location, and complexity (Bridge et al., 2019; Owens et al., 2019; Rubino et al., 2020).
Since biochemical data obtained from the UAA-mediated crosslinking are an array of band shift patterns resulting from various crosslinked-target proteins or fragments in the immunoblotted gel electrophoresis, the output is generally moderate or low in three-dimensional conformation and resolution. The crosslinking data can be best utilized when integrated with other structural information to give a refined structure with high resolution and fidelity. For example, the class B corticotropin-releasing factor receptor type 1 (CRF1R) was mapped by genetically incorporating an UAA crosslinker systematically to characterize the difference in a conformational change and an interfacial landscape, when CRF1R was stimulated by an agonist or antagonist. Differences in band shift patterns were evident in western blotting, and the inter-residue distance constraints, a characteristic of the crosslinked chemical structure, estimated for all of the crosslinked residues were applied to computationally generate conformational models of the agonist- and the antagonist-bound CRF1R complex. Extensive sampling of conformations led to optimized structure models for both complexes, revealing that CRF1R adopted distinct binding interfaces and local conformations to engage the agonist and the antagonist, respectively (Seidel et al., 2017). Remarkably, the predicted models have been found to be very compatible with the high-resolution cryo-EM structures obtained years later (Liang et al., 2020; Ma et al., 2020). A unique hormone-binding motif of the insulin receptor could be recognized from the crosslinking data in conjunction with the preexisting crystal structure of the apo-insulin receptor. As was expected from the crystal contacts, a photo-reactive UAA incorporated in the typical β-strands 2 and 3 in the L1 domain of the insulin receptor α-subunit was found to be crosslinked to the other α-subunit in an apo-state. However, in a holo-state where the insulin receptor was bound to the insulin, the crosslinks were made to the insulin, implying a dynamic interface change caused by the induced fit of the insulin (Whittaker et al., 2012). Similarly, a recent study performed a crystal structure-based photo-crosslinking analysis of the macromolecular complex of nuclear pore proteins, Nup82, Nup116, and Nup159. Of note, crosslinking patterns demonstrated that, in comparison to the crystal interfaces, the interfaces of Nup82 in contact with Nup116 and 159 were significantly different in areas and residues involved (Shin and Lim 2020). Interestingly, some contact residues in the interface engaging Nup82 and Nup116 in the ternary complex seemed to lose contacts when Nup82 and Nup116 formed the binary complex by themselves without Nup159, indicating a dynamic nature of the interface not locally isolated but wriggling in conjunction with overall structural changes. Abovementioned and related studies are summarized in the header row of ‘photo-crosslinking UAA in live cells’ (Table 1).