Synthesis and evaluation of heteroaryl substituted diazaspirocycles as scaffolds to probe the ATP-binding site of protein kinases

With the success of protein kinase inhibitors as drugs to target cancer, there is a continued need for new kinase inhibitor scaffolds. We have investigated the synthesis and kinase inhibition of new heteroaryl-substituted diazaspirocyclic compounds that mimic ATP. Versatile syntheses of substituted diazaspirocycles through ring-closing metathesis were demonstrated. Diazaspirocycles directly linked to heteroaromatic hinge binder groups provided ligand efficient inhibitors of multiple kinases, suitable as starting points for further optimization. The binding modes of representative diazaspirocyclic motifs were confirmed by protein crystallography. Selectivity profiles were influenced by the hinge binder group and the interactions of basic nitrogen atoms in the scaffold with acidic side-chains of residues in the ATP pocket. The introduction of more complex substitution to the diazaspirocycles increased potency and varied the selectivity profiles of these initial hits through engagement of the P-loop and changes to the spirocycle conformation, demonstrating the potential of these core scaffolds for future application to kinase inhibitor discovery.     

A novel mode of protein kinase inhibition exploiting hydrophobic motifs of autoinhibited kinases: discovery of ATP-independent inhibitors of fibroblast growth factor receptor

Protein kinase inhibitors with enhanced selectivity can be designed by optimizing binding interactions with less conserved inactive conformations because such inhibitors will be less likely to compete with ATP for binding and therefore may be less impacted by high intracellular concentrations of ATP. Analysis of the ATP-binding cleft in a number of inactive protein kinases, particularly in the autoinhibited conformation, led to the identification of a previously undisclosed non-polar region in this cleft. This ATP-incompatible hydrophobic region is distinct from the previously characterized hydrophobic allosteric back pocket, as well as the main pocket. Generalized hypothetical models of inactive kinases were constructed and, for the work described here, we selected the fibroblast growth factor receptor (FGFR) tyrosine kinase family as a case study. Initial optimization of a FGFR2 inhibitor identified from a library of commercial compounds was guided using structural information from the model. We describe the inhibitory characteristics of this compound in biophysical, biochemical, and cell-based assays, and have characterized the binding mode using x-ray crystallographic studies. The results demonstrate, as expected, that these inhibitors prevent activation of the autoinhibited conformation, retain full inhibitory potency in the presence of physiological concentrations of ATP, and have favorable inhibitory activity in cancer cells. Given the widespread regulation of kinases by autoinhibitory mechanisms, the approach described herein provides a new paradigm for the discovery of inhibitors by targeting inactive conformations of protein kinases.     

A loop-mimetic inhibitor of the HCV-NS3 protease derived from a minibody

We have been interested for some time in establishing a strategy for deriving lead compounds from macromolecule ligands such as minibody variants. A minibody is a minimized antibody variable domain whose two loops are amenable to combinatorial mutagenesis. This approach can be especially useful when dealing with 'difficult' targets. One such target is the NS3 protease of hepatitis C virus (HCV), a human pathogen that is believed to infect about 100 million individuals worldwide and for which an effective therapy is not yet available. Based on known inhibitor specificity (residues P6-P1) of NS3 protease, we screened a number of minibodies from our collection and we were able to identify a competitive inhibitor of this enzyme. We thus validated an aspect of recognition by HCV NS3 protease, namely that an acid anchor is necessary for inhibitor activity. In addition, the characterization of the minibody inhibitor led to the synthesis of a constrained hexapeptide mimicking the bioactive loop of the parent macromolecule. The cyclic peptide is a lead compound prone to rapid optimization through solid phase combinatorial chemistry. We therefore confirmed that the potential of turning a protein ligand into a low molecular weight active compound for lead discovery is achievable and can complement more traditional drug discovery approaches.     

Effective Therapeutic Approach for Head and Neck Cancer by an Engineered Minibody Targeting the EGFR Receptor

Cetuximab, a chimeric monoclonal antibody developed for targeting the Epidermal Growth Factor Receptor (EGFR), has been intensively used to treat cancer patients with metastatic colorectal cancer and head and neck cancer. Intact immunoglobulin G (IgG) antibody like cetuximab, however, has some limitations such as high production cost and low penetration rate from vasculature into solid tumor mass due to its large size. In attempt to overcome these limitations, we engineered cetuximab to create single chain variable fragments (scFv-CH3; Minibody) that were expressed in bacterial system. Among three engineered minibodies, we found that MI061 minibody, which is composed of the variable heavy (V_H) and light (V_L) region joined by an 18-residue peptide linker, displays higher solubility and better extraction properties from bacterial lysate. In addition, we validated that purified MI061 significantly interferes ligand binding to EGFR and blocks EGFR''s phosphorylation. By using a protein microarray composed of 16,368 unique human proteins covering around 2,400 plasma membrane associated proteins such as receptors and channels, we also demonstrated that MI061 only recognizes the EGFR but not other proteins as compared with cetuximab. These results indicated that engineered MI061 retains both binding specificity and affinity of cetuximab for EGFR. Although it had relatively short half-life in serum, it was shown to be highly significant anti-tumor effect by inhibiting ERK pathway in A431 xenograft model. Taken together, our present study provides compelling evidence that engineered minibody is more effective and promising agent for in vivo targeting of solid tumors.     

Modular peptide binding: From a comparison of natural binders to designed armadillo repeat proteins

Several binding scaffolds that are not based on immunoglobulins have been designed as alternatives to traditional monoclonal antibodies. Many of them have been developed to bind to folded proteins, yet cellular networks for signaling and protein trafficking often depend on binding to unfolded regions of proteins. This type of binding can thus be well described as a peptide–protein interaction. In this review, we compare different peptide-binding scaffolds, highlighting that armadillo repeat proteins (ArmRP) offer an attractive modular system, as they bind a stretch of extended peptide in a repeat-wise manner. Instead of generating each new binding molecule by an independent selection, preselected repeats – each complementary to a piece of the target peptide – could be designed and assembled on demand into a new protein, which then binds the prescribed complete peptide. Stacked armadillo repeats (ArmR), each typically consisting of 42 amino acids arranged in three α-helices, build an elongated superhelical structure which enables binding of peptides in extended conformation. A consensus-based design approach, complemented with molecular dynamics simulations and rational engineering, resulted in well-expressed monomeric proteins with high stability. Peptide binders were selected and several structures were determined, forming the basis for the future development of modular peptide-binding scaffolds.     

Specialized Hsp70 Chaperone (HscA) Binds Preferentially to the Disordered Form, whereas J-protein (HscB) Binds Preferentially to the Structured Form of the Iron-Sulfur Cluster Scaffold Protein (IscU)

The Escherichia coli protein IscU serves as the scaffold for Fe-S cluster assembly and the vehicle for Fe-S cluster transfer to acceptor proteins, such as apoferredoxin. IscU populates two conformational states in solution, a structured conformation (S) that resembles the conformation of the holoprotein IscU-[2Fe-2S] and a dynamically disordered conformation (D) that does not bind metal ions. NMR spectroscopic results presented here show that the specialized Hsp70 chaperone (HscA), alone or as the HscA-ADP complex, preferentially binds to and stabilizes the D-state of IscU. IscU is released when HscA binds ATP. By contrast, the J-protein HscB binds preferentially to the S-state of IscU. Consistent with these findings, we propose a mechanism in which cluster transfer is coupled to hydrolysis of ATP bound to HscA, conversion of IscU to the D-state, and release of HscB.     

Protein scaffolds can enhance the bistability of multisite phosphorylation systems

The phosphorylation of a substrate at multiple sites is a common protein modification that can give rise to important structural and electrostatic changes. Scaffold proteins can enhance protein phosphorylation by facilitating an interaction between a protein kinase enzyme and its target substrate. In this work we consider a simple mathematical model of a scaffold protein and show that under specific conditions, the presence of the scaffold can substantially raise the likelihood that the resulting system will exhibit bistable behavior. This phenomenon is especially pronounced when the enzymatic reactions have sufficiently large K(M), compared to the concentration of the target substrate. We also find for a closely related model that bistable systems tend to have a specific kinetic conformation. Using deficiency theory and other methods, we provide a number of necessary conditions for bistability, such as the presence of multiple phosphorylation sites and the dependence of the scaffold binding/unbinding rates on the number of phosphorylated sites.     

Periplasmic expression,purification, and characterization of an anti-epidermal growth factor receptor antibody fragment in <Emphasis Type="Italic">Escherichia coli</Emphasis>

New structural designs of antibody fragments have considerable biotechnological and therapeutic potential. In this study, we describe the construction and functional expression of a cetuximab-based antibody fragment (scFv-C_H3, minibody) that exhibits activity against human colon cancer. Heterologous expression in Escherichia coli (E. coli) was improved by optimizing the host cells, signal peptides, induction conditions, and culture media. The recombinant minibody was expressed successfully in the periplasm of E. coli BL21(DE3) and purified by immobilized metal affinity chromatography using a Ni²⁺-NTA resin. The purified minibody showed high binding affinity to cell-surface epidermal growth factor receptor (EGFR) and exhibited inhibition of EGFR-mediated signal transduction in the human colon cancer cell line HT29 in a similar way by the cetuximab. The minibody also showed significant level of anti-cancer ability in the HT29 colorectal cancer xenograft model, which was lower than that by the cetuximab.     

Structure-guided engineering of Anticalins with improved binding behavior and biochemical characteristics for application in radio-immuno imaging and/or therapy

Modern strategies in radio-immuno therapy and in vivo imaging require robust, small, and specific ligand-binding proteins. In this context we have previously developed artificial lipocalins, so-called Anticalins, with high binding activity toward rare-earth metal–chelate complexes using combinatorial protein design. Here we describe further improvement of the Anticalin C26 via in vitro affinity maturation to yield CL31, which has a fourfold slower dissociation half-life above 2 h. Also, we present the crystallographic analyses of both the initial and the improved Anticalin, providing insight into the molecular mechanism of chelated metal binding and the role of amino acid substitutions during the step-wise affinity maturation. Notably, one of the four structurally variable loops that form the ligand pocket in the lipocalin scaffold undergoes a significant conformational change from C26 to CL31, acting as a lid that closes over the accommodated metal–chelate ligand. A systematic mutational study indicated that further improvement of ligand affinity is difficult to achieve while providing clues on the contribution of relevant side chains in the engineered binding pocket. Unexpectedly, some of the amino acid replacements led to strong increases – more then 10-fold – in the yield of soluble protein from periplasmic secretion in Escherichia coli.     

Molecular engineering of respiratory syncytial virus immunogen for rational redesign of prophylactic peptide vaccines against pediatric viral pneumonia

The immunogen FFL_001 of respiratory syncytial virus (RSV), a widely spread virus that infects the lungs and breathing passages, is a three-helix bundle protein of 115 amino acids long that consists of three helical arms H1, H2 and H3. Here, we attempted to perform molecular engineering of the immunogen, aiming to considerably reduce the protein scaffold of FFL_001 with only moderate activity loss. Structural analysis and molecular dynamics (MD) simulations revealed that two helices H2 and H3 of the FFL_001 three-helix bundle can directly interact with its cognate monoclonal antibody Motavizumab, while the remaining helix H1 plays a crucial role in stabilisation of the three-helix bundle conformation. Binding of the two split peptide segments separately representing FFL_001 two-helix bundle H2-H3 and Motavizumab-binding site to the antibody would incur considerable entropy penalty as compared to binding of the intact FFL_001, suggesting that peptide segments are highly flexible that exhibit a strong intrinsic disorder in solution. In this respect, a scheme was proposed to rationally redesign the immunogen protein scaffold by truncation and cyclisation of the three-helix bundle. The cyclisation was conducted on the spatially vicinal residue pairs in H2 and H3 helical arms by mutating the residue to cysteine and introducing a disulphide bond across them. Consequently, we obtained three cyclic peptides that were theoretically predicted to have strong binding potency towards Motavizumab. In order to substantiate the computational finding, binding affinity of the designed cyclic peptide and its linear counterpart was determined. Consistently, no binding can be found between the linear peptide and Motavizumab (K_d = n.d.), while a moderately high affinity was observed for cyclic peptide (K_d = 16.2 nM).     

Evaluation of lumazine synthase from Bacillus anthracis as a presentation platform for polyvalent antigen display

Polyvalent antigen display is an effective strategy to enhance the immunogenicity of subunit vaccines by clustering them in an array‐like manner on a scaffold system. This strategy results in a higher local density of antigens, increased high avidity interactions with B cells and other antigen presenting cells, and therefore a more effective presentation of vaccine antigens. In this study, we used lumazine synthase (LS), an icosahedral symmetry capsid derived from Bacillus anthracis, as a scaffold to present 60 copies of a linear B cell epitope (PB10) from the ricin toxin fused to the C terminus of LS via four different linkers. We then investigated the effects of linker length, linker rigidity and formaldehyde crosslinking on the protein assembly, conformational integrity, thermal stability, in vitro antibody binding, and immunogenicity in mice. Fusion of the PB10 peptide onto LS, with varying linker lengths, did not affect protein assembly, thermal stability or exposure of the epitope, but had a minor impact on protein conformation. Formaldehyde crosslinking considerably improved protein thermal stability with only minor impact on protein conformation. All LS_PB10 constructs, when administered to mice by injection without adjuvant, elicited measurable anti‐ricin serum IgG titers, although the titers were not sufficient to confer protection against a 10× lethal dose ricin challenge. This work sheds light on the biophysical properties, immunogenicity and potential feasibility of LS from B. anthracis as a scaffold system for polyvalent antigen display.     

Cooperation between an intrinsically disordered region and a helical segment is required for ubiquitin-independent degradation by the proteasome

The 26 S proteasomal complex, which is responsible for the bulk of protein degradation within the cell, recognizes its target substrates via covalently linked polyubiquitin moieties. However, a small but growing number of proteasomal substrates are degraded without a requirement for ubiquitinylation. One such substrate is the pyrimidine biosynthetic enzyme thymidylate synthase (EC 2.1.1.45), which catalyzes the synthesis of TMP and is the sole de novo source of TTP for DNA replication and repair. Previous work showed that intracellular proteolysis of human thymidylate synthase is directed by a degron at the polypeptide's N-terminal end, composed of an intrinsically disordered region (IDR) followed by a highly conserved amphipathic α-helix (hA). In the present report, we show that the hA helix does not function simply as an extension or scaffold for the IDR; rather, it provides a specific structural component that is necessary for degradation. Furthermore, its helical conformation is required for this function. We demonstrate that small domains from heterologous proteins can substitute for the IDR and the hA helix of human thymidylate synthase, indicating that the degradation-promoting function of these regions is not sequence-specific. The results, in general, indicate that cooperation between intrinsically disordered domains and α-helical segments is required for ubiquitin-independent degradation by the proteasome. There appears to be little sequence constraint on the ability of these regions to function as degron constituents. Rather, it is the overall conformation (or lack thereof) that is critical.     

Identification of novel and distinct binding sites within tenascin-C for soluble and fibrillar fibronectin

Interactions between fibronectin and tenascin-C within the extracellular matrix provide specific environmental cues that dictate tissue structure and cell function. The major binding site for fibronectin lies within the fibronectin type III-like repeats (TNfn) of tenascin-C. Here, we systematically screened TNfn domains for their ability to bind to both soluble and fibrillar fibronectin. All TNfn domains containing the TNfn3 module interact with soluble fibronectin. However, TNfn domains bind differentially to fibrillar fibronectin. This distinct binding pattern is dictated by the fibrillar conformation of FN. TNfn1-3, but not TNfn3-5, binds to immature fibronectin fibrils, and additional TNfn domains are required for binding to mature fibrils. Multiple binding sites for distinct regions of fibronectin exist within tenascin-C. TNfn domains comprise a binding site for the N-terminal 70-kDa domain of fibronectin that is freely available and a binding site for the central binding domain of fibronectin that is cryptic in full-length tenascin-C. The 70-kDa and central binding domain regions are key for fibronectin matrix assembly; accordingly, binding of several TNfn domains to these regions inhibits fibronectin fibrillogenesis. These data highlight the complexity of protein-protein binding, the importance of protein conformation on these interactions, and the implications for the physiological assembly of complex three-dimensional matrices.     

Structural consequences of binding of UO2(2+) to apotransferrin: can this protein account for entry of uranium into human cells?

It has been established that transferrin binds a variety of metals. These include toxic uranyl ions which form rather stable uranyl-transferrin derivatives. We determined the extent to which the iron binding sites might accommodate the peculiar topographic profile of the uranyl ion and the consequences of its binding on protein conformation. Indeed, metal intake via endocytosis of the transferrin/transferrin receptor depends on the adequate coordination of the metal in its site, which controls protein conformation and receptor binding. Using UV-vis and Fourier transform infrared difference spectroscopy coupled to a microdialysis system, we showed that at both metal binding sites two tyrosines are uranyl ligands, while histidine does not participate with its coordination sphere. Analysis by circular dichroism and differential scanning calorimetry (DSC) showed major differences between structural changes associated with interactions of iron or uranyl with apotransferrin. Uranyl coordination reduces the level of protein stabilization compared to iron, but this may be simply related to partial lobe closure. The lack of interaction between uranyl-TF and its receptor was shown by flow cytometry using Alexa 488-labeled holotransferrin. We propose a structural model summarizing our conclusion that the uranyl-TF complex adopts an open conformation that is not appropriate for optimal binding to the transferrin receptor.     

The N-Terminal Domain of the Tomato Immune Protein Prf Contains Multiple Homotypic and Pto Kinase Interaction Sites

Resistance to Pseudomonas syringae bacteria in tomato (Solanum lycopersicum) is conferred by the Prf recognition complex, composed of the nucleotide-binding leucine-rich repeats protein Prf and the protein kinase Pto. The complex is activated by recognition of the P. syringae effectors AvrPto and AvrPtoB. The N-terminal domain is responsible for Prf homodimerization, which brings two Pto kinases into close proximity and holds them in inactive conformation in the absence of either effector. Negative regulation is lost by effector binding to the catalytic cleft of Pto, leading to disruption of its P+1 loop within the activation segment. This change is translated through Prf to a second Pto molecule in the complex. Here we describe a schematic model of the unique Prf N-terminal domain dimer and its interaction with the effector binding determinant Pto. Using heterologous expression in Nicotiana benthamiana, we define multiple sites of N domain homotypic interaction and infer that it forms a parallel dimer folded centrally to enable contact between the N and C termini. Furthermore, we found independent binding sites for Pto at either end of the N-terminal domain. Using the constitutively active mutant ptoL205D, we identify a potential repression site for Pto in the first ∼100 amino acids of Prf. Finally, we find that the Prf leucine-rich repeats domain also binds the N-terminal region, highlighting a possible mechanism for transfer of the effector binding signal to the NB-LRR regulatory unit (consisting of a central nucleotide binding and C-terminal leucine-rich repeats).     

Solution NMR Structure of the C-terminal DNA Binding Domain of Mcm10 Reveals a Conserved MCM Motif

The eukaryotic DNA replication protein Mcm10 associates with chromatin in early S-phase and is required for assembly and function of the replication fork protein machinery. Xenopus laevis (X) Mcm10 binds DNA via a highly conserved internal domain (ID) and a C-terminal domain (CTD) that is unique to higher eukaryotes. Although the structural basis of the interactions of the ID with DNA and polymerase α is known, little information is available for the CTD. We have identified the minimal DNA binding region of the XMcm10-CTD and determined its three-dimensional structure by solution NMR. The CTD contains a globular domain composed of two zinc binding motifs. NMR chemical shift perturbation and mutational analysis show that ssDNA binds only to the N-terminal (CCCH-type) zinc motif, whose structure is unique to Mcm10. The second (CCCC-type) zinc motif is not involved in DNA binding. However, it is structurally similar to the CCCC zinc ribbon in the N-terminal oligomerization domain of eukaryotic and archaeal MCM helicases. NMR analysis of a construct spanning both the ID and CTD reveals that the two DNA binding domains are structurally independent in solution, supporting a modular architecture for vertebrate Mcm10. Our results provide insight in the action of Mcm10 in the replisome and support a model in which it serves as a central scaffold through coupling of interactions with partner proteins and the DNA.     

Using physical potentials and learned models to distinguish native binding interfaces from de novo designed interfaces that do not bind

Protein–protein interactions are a fundamental aspect of many biological processes. The advent of recombinant protein and computational techniques has allowed for the rational design of proteins with novel binding capabilities. It is therefore desirable to predict which designed proteins are capable of binding in vitro. To this end, we have developed a learned classification model that combines energetic and non‐energetic features. Our feature set is adapted from specialized potentials for aromatic interactions, hydrogen bonds, electrostatics, shape, and desolvation. A binding model built on these features was initially developed for CAPRI Round 21, achieving top results in the independent assessment. Here, we present a more thoroughly trained and validated model, and compare various support‐vector machine kernels. The Gaussian kernel model classified both high‐resolution complexes and designed nonbinders with 79–86% accuracy on independent test data. We also observe that multiple physical potentials for dielectric‐dependent electrostatics and hydrogen bonding contribute to the enhanced predictive accuracy, suggesting that their combined information is much greater than that of any single energetics model. We also study the change in predictive performance as the model features or training data are varied, observing unusual patterns of prediction in designed interfaces as compared with other data types. Proteins 2013; 81:1919–1930. © 2013 Wiley Periodicals, Inc.     

Prediction of methionine oxidation risk in monoclonal antibodies using a machine learning method

ABSTRACT

Monoclonal antibodies (mAbs) have become a major class of protein therapeutics that target a spectrum of diseases ranging from cancers to infectious diseases. Similar to any protein molecule, mAbs are susceptible to chemical modifications during the manufacturing process, long-term storage, and in vivo circulation that can impair their potency. One such modification is the oxidation of methionine residues. Chemical modifications that occur in the complementarity-determining regions (CDRs) of mAbs can lead to the abrogation of antigen binding and reduce the drug’s potency and efficacy. Thus, it is highly desirable to identify and eliminate any chemically unstable residues in the CDRs during the therapeutic antibody discovery process. To provide increased throughput over experimental methods, we extracted features from the mAbs’ sequences, structures, and dynamics, used random forests to identify important features and develop a quantitative and highly predictive in silico methionine oxidation model.     

Myosin binding protein C interaction with actin: characterization and mapping of the binding site

Myosin binding protein C (MyBPC) is a multidomain protein associated with the thick filaments of striated muscle. Although both structural and regulatory roles have been proposed for MyBPC, its interactions with other sarcomeric proteins remain obscure. The current study was designed to examine the actin-binding properties of MyBPC and to define MyBPC domain regions involved in actin interaction. Here, we have expressed full-length mouse cardiac MyBPC (cMyBPC) in a baculovirus system and shown that purified cMyBPC binds actin filaments with an affinity of 4.3 ± 1.1 μM and a 1:1 molar ratio with regard to an actin protomer. The actin binding by cMyBPC is independent of protein phosphorylation status and is not significantly affected by the presence of tropomyosin and troponin on the actin filament. In addition, cMyBPC-actin interaction is not modulated by calmodulin. To determine the region of cMyBPC that is responsible for its interaction with actin, we have expressed and characterized five recombinant proteins encoding fragments of the cMyBPC sequence. Recombinant N-terminal fragments such as C0-C1, C0-C4, and C0-C5 cosediment with actin in a linear, nonsaturable manner. At the same time, MyBPC fragments lacking either the C0-C1 or C0-C4 region bind F-actin with essentially the same properties as full-length protein. Together, our results indicate that cMyBPC interacts with actin via a single, moderate affinity site localized to the C-terminal region of the protein. In contrast, certain basic regions of the N-terminal domains of MyBPC may act as small polycations and therefore bind actin via nonspecific electrostatic interactions.     

Conformational Selection and Equilibrium Governs the Ability of Retinals to Bind Opsin

Despite extensive study, how retinal enters and exits the visual G protein-coupled receptor rhodopsin remains unclear. One clue may lie in two openings between transmembrane helix 1 (TM1) and TM7 and between TM5 and TM6 in the active receptor structure. Recently, retinal has been proposed to enter the inactive apoprotein opsin (ops) through these holes when the receptor transiently adopts the active opsin conformation (ops*). Here, we directly test this “transient activation” hypothesis using a fluorescence-based approach to measure rates of retinal binding to samples containing differing relative fractions of ops and ops*. In contrast to what the transient activation hypothesis model would predict, we found that binding for the inverse agonist, 11-cis-retinal (11CR), slowed when the sample contained more ops* (produced using M257Y, a constitutively activating mutation). Interestingly, the increased presence of ops* allowed for binding of the agonist, all-trans-retinal (ATR), whereas WT opsin showed no binding. Shifting the conformational equilibrium toward even more ops* using a G protein peptide mimic (either free in solution or fused to the receptor) accelerated the rate of ATR binding and slowed 11CR binding. An arrestin peptide mimic showed little effect on 11CR binding; however, it stabilized opsin·ATR complexes. The TM5/TM6 hole is apparently not involved in this conformational selection. Increasing its size by mutagenesis did not enable ATR binding but instead slowed 11CR binding, suggesting that it may play a role in trapping 11CR. In summary, our results indicate that conformational selection dictates stable retinal binding, which we propose involves ATR and 11CR binding to different states, the latter a previously unidentified, open-but-inactive conformation.