Proteasomal Stimulation by MK886 and Its Derivatives Can Rescue Tau-Induced Neurite Pathology.

Proteasomal degradation of intrinsically disordered proteins, such as tau, is a critical component of proteostasis in both aging and neurodegenerative diseases. In this study, we investigated proteasomal activation by MK886 (MK). We previously identified MK as a lead compound capable of modulating tau oligomerization in a cellular FRET assay and rescuing P301L tau-induced cytotoxicity. We first confirmed robust proteasomal activation by MK using 20S proteasomal assays and a cellular proteasomal tau-GFP cleavage assay. We then show that MK treatment can significantly rescue tau-induced neurite pathology in differentiated SHSY5Y neurospheres. Due to this compelling result, we designed a series of seven MK analogs to determine if proteasomal activity is sensitive to structural permutations. Using the proteasome as the primary MOA, we examined tau aggregation, neurite outgrowth, inflammation, and autophagy assays to identify two essential substituents of MK that are required for compound activity: (1) removal of the N-chlorobenzyl group from MK negated both proteasomal and autophagic activity and reduced neurite outgrowth; and (2) removal of the indole-5-isopropyl group significantly improved neurite outgrowth and autophagy activity but reduced its anti-inflammatory capacity. Overall, our results suggest that the combination of proteasomal/autophagic stimulation and anti-inflammatory properties of MK and its derivatives can decrease tau-tau interactions and help rebalance dysfunctional proteostasis. Further development of MK to optimize its proteasomal, autophagic, and anti-inflammatory targets may lead to a novel therapeutic that would be beneficial in aging and neurodegenerative diseases.

Zafirlukast Is a Promising Scaffold for Selectively Inhibiting TNFR1 Signaling.

Tumor necrosis factor (TNF) plays an important role in the pathogenesis of inflammatory and autoimmune diseases such as rheumatoid arthritis and Crohn's disease. The biological effects of TNF are mediated by binding to TNF receptors, TNF receptor 1 (TNFR1), or TNF receptor 2 (TNFR2), and this coupling makes TNFR1-specific inhibition by small-molecule therapies essential to avoid deleterious side effects. Recently, we engineered a time-resolved fluorescence resonance energy transfer biosensor for high-throughput screening of small molecules that modulate TNFR1 conformational states and identified zafirlukast as a compound that inhibits receptor activation, albeit at low potency. Here, we synthesized 16 analogues of zafirlukast and tested their potency and specificity for TNFR1 signaling. Using cell-based functional assays, we identified three analogues with significantly improved efficacy and potency, each of which induces a conformational change in the receptor (as measured by fluorescence resonance energy transfer (FRET) in cells). The best analogue decreased NF-κB activation by 2.2-fold, IκBα efficiency by 3.3-fold, and relative potency by two orders of magnitude. Importantly, we showed that the analogues do not block TNF binding to TNFR1 and that binding to the receptor's extracellular domain is strongly cooperative. Despite these improvements, the best candidate's maximum inhibition of NF-κB is only 63%, leaving room for further improvements to the zafirlukast scaffold to achieve full inhibition and prove its potential as a therapeutic lead. Interestingly, while we find that the analogues also bind to TNFR2 in vitro, they do not inhibit TNFR2 function in cells or cause any conformational changes upon binding. Thus, these lead compounds should also be used as reagents to study conformational-dependent activation of TNF receptors.

Indirubin Inhibits TRAIL-Induced Activation of Death Receptor 5 in Jurkat Cells.

Death receptor 5 (DR5) is an apoptosis-inducing membrane receptor that mediates cell death in several life-threatening conditions. There is a crucial need for the discovery of DR5 antagonists for the therapeutic intervention of conditions in which the overactivation of DR5 underlies the pathophysiology. DR5 activation mediates cell death in non-alcoholic fatty liver disease (NAFLD) and neurodegenerative processes including amyloid-beta (Aß) accumulation, spinal cord injury (SCI), and brain ischemia. In the current work, we used fluorescence resonance energy transfer (FRET) to monitor the conformational dynamics of DR5 that mediate death signaling. We used a time-resolved FRET screening platform to screen the Selleck library of 2863 U.S. Food and Drug Administration (FDA)-approved compounds. The high-throughput screen (HTS) identified 13 compounds that modulated the FRET between DR5 monomers beyond 5 median absolute deviations (MADs) from the DMSO controls. Of these 13 compounds, indirubin was identified to specifically inhibit tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-induced caspase-8 activity without modulating DR5 surface expression or TRAIL binding. Indirubin inhibited Fas-associated death domain (FADD) oligomerization and increased cellular FLICE-inhibitory protein (c-FLIP) expression; both are molecular mechanisms involved in inhibiting the DR5 signaling cascade. This study has elucidated previously unknown properties of indirubin that make it a promising candidate for therapeutic investigation of diseases in which overactivation of DR5 underlies pathology.

Discovery of a Non-competitive TNFR1 Antagonist Affibody with Picomolar Monovalent Potency That Does Not Affect TNFR2 Function.

Tumor necrosis factor (TNF) is a key regulator of immune responses and plays a significant role in the initiation and maintenance of inflammation. Upregulation of TNF expression leads to several inflammatory diseases, such as Crohn's, ulcerative colitis, and rheumatoid arthritis. Despite the clinical success of anti-TNF treatments, the use of these therapies is limited because they can induce adverse side effects through inhibition of TNF biological activity, including blockade of TNF-induced immunosuppressive function of TNFR2. Using yeast display, we identified a synthetic affibody ligand (ABYTNFR1-1) with high binding affinity and specificity for TNFR1. Functional assays showed that the lead affibody potently inhibits TNF-induced NF-κB activation (IC50 of 0.23 nM) and, crucially, does not block the TNFR2 function. Additionally, ABYTNFR1-1 acts non-competitively─it does not block TNF binding or inhibit receptor-receptor interactions in pre-ligand-assembled dimers─thereby enhancing inhibitory robustness. The mechanism, monovalent potency, and affibody scaffold give this lead molecule uniquely strong potential as a therapeutic candidate for inflammatory diseases.

Nimesulide, a COX-2 inhibitor, sensitizes pancreatic cancer cells to TRAIL-induced apoptosis by promoting DR5 clustering †.

Nimesulide is a nonsteroidal anti-inflammatory drug and a COX-2 inhibitor with antitumor and antiproliferative activities that induces apoptosis in oral, esophagus, breast, and pancreatic cancer cells. Despite being removed from the market due to hepatotoxicity, nimesulide is still an important research tool being used to develop new anticancer drugs. Multiple studies have been done to modify the nimesulide skeleton to develop more potent anticancer agents and related compounds are promising scaffolds for future development. As such, establishing a mechanism of action for nimesulide remains an important part of realizing its potential. Here, we show that nimesulide enhances TRAIL-induced apoptosis in resistant pancreatic cancer cells by promoting clustering of DR5 in the plasma membrane. In this way, nimesulide acts like a related compound, DuP-697, which sensitizes TRAIL-resistant colon cancer cells in a similar manner. Our approach applies a time-resolved FRET-based biosensor that monitors DR5 clustering and conformational states in the plasma membrane. We show that this tool can be used for future high-throughput screens to identify novel, nontoxic small molecule scaffolds to overcome TRAIL resistance in cancer cells.

Fluorescence Lifetime Measurement of Prefibrillar Sickle Hemoglobin Oligomers as a Platform for Drug Discovery in Sickle Cell Disease.

The molecular origin of sickle cell disease (SCD) has been known since 1949, but treatments remain limited. We present the first high-throughput screening (HTS) platform for discovering small molecules that directly inhibit sickle hemoglobin (HbS) oligomerization and improve blood flow, potentially overcoming a long-standing bottleneck in SCD drug discovery. We show that at concentrations far below the threshold for nucleation and rapid polymerization, deoxygenated HbS forms small assemblies of multiple α2ß2 tetramers. Our HTS platform leverages high-sensitivity fluorescence lifetime measurements that monitor these temporally stable prefibrillar HbS oligomers. We show that this approach is sensitive to compounds that inhibit HbS polymerization with or without modulating hemoglobin oxygen binding affinity. We also report the results of a pilot small-molecule screen in which we discovered and validated several novel inhibitors of HbS oligomerization.

Noncompetitive Allosteric Antagonism of Death Receptor 5 by a Synthetic Affibody Ligand.

Fatty acid-induced upregulation of death receptor 5 (DR5) and its cognate ligand, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), promotes hepatocyte lipoapoptosis, which is a key mechanism in the progression of fatty liver disease. Accordingly, inhibition of DR5 signaling represents an attractive strategy for treating fatty liver disease. Ligand competition strategies are prevalent in tumor necrosis factor receptor antagonism, but recent studies have suggested that noncompetitive inhibition through perturbation of the receptor conformation may be a compelling alternative. To this end, we used yeast display and a designed combinatorial library to identify a synthetic 58-amino acid affibody ligand that specifically binds DR5. Biophysical and biochemical studies show that the affibody neither blocks TRAIL binding nor prevents the receptor-receptor interaction. Live-cell fluorescence lifetime measurements indicate that the affibody induces a conformational change in transmembrane dimers of DR5 and favors an inactive state of the receptor. The affibody inhibits apoptosis in TRAIL-treated Huh-7 cells, an in vitro model of fatty liver disease. Thus, this lead affibody serves as a potential drug candidate, with a unique mechanism of action, for fatty liver disease.

Soluble Extracellular Domain of Death Receptor 5 Inhibits TRAIL-Induced Apoptosis by Disrupting Receptor-Receptor Interactions.

Dysregulation of tumor necrosis factor (TNF) receptor signaling is a key feature of various inflammatory disorders. Current treatments for TNF-related diseases function either by sequestering ligand or blocking ligand-receptor interactions, which can cause dangerous side effects by inhibiting the receptors that are not involved in the disease condition. Thus, alternate strategies that target receptor-receptor interactions are needed. We hypothesized that the soluble extracellular domain (ECD) of long isoform of death receptor 5 (DR5) could block endogenous receptor assembly, mimicking the biological effect of decoy receptors that lack the death domain to trigger apoptosis. Using live-cell fluorescence resonance energy transfer studies, we demonstrated that soluble ECD disrupts endogenous DR5-DR5 interactions. Cell viability assays were used to demonstrate the complete inhibition of TNF-related apoptosis-inducing ligand (TRAIL)-induced apoptosis by the ECD, although TRAIL is still able to bind to the receptor. Importantly, we used mutagenesis to prove that the inhibition of TRAIL-induced apoptosis by the ECD predominantly comes from the disruption of DR5 oligomerization and not ligand sequestration. Inhibition of death receptor activation should have important therapeutic applications in diseases such as nonalcoholic fatty liver disease. More generally, this approach should be generalized to enable the inhibition of other TNF receptor signaling mechanisms that are associated in a wide range of clinical conditions.

Death Receptor 5 Activation Is Energetically Coupled to Opening of the Transmembrane Domain Dimer.

The precise mechanism by which binding of tumor necrosis factor ligands to the extracellular domain of their corresponding receptors transmits signals across the plasma membrane has remained elusive. Recent studies have proposed that activation of several tumor necrosis factor receptors, including Death Receptor 5, involves a scissorlike opening of the disulfide-linked transmembrane (TM) dimer. Using time-resolved fluorescence resonance energy transfer, we provide, to our knowledge, the first direct biophysical evidence that Death Receptor 5 TM-dimers open in response to ligand binding. Then, to probe the importance of the closed-to-open TM domain transition in the overall energetics of receptor activation, we designed point-mutants (alanine to phenylalanine) in the predicted, tightly packed TM domain dimer interface. We hypothesized that the bulky residues should destabilize the closed conformation and eliminate the ∼3 kcal/mol energy barrier to TM domain opening and the âˆ¼2 kcal/mol energy difference between the closed and open states, thus oversensitizing the receptor. To test this, we used all-atom molecular dynamics simulations of the isolated TM domain in explicit lipid bilayers coupled to thermodynamic potential of mean force calculations. We showed that single point mutants at the interface altered the energy landscape as predicted, but were not enough to completely eliminate the barrier to opening. However, the computational model did predict that a double mutation at i, i+4 positions at the center of the TM domain dimer eliminates the barrier and stabilizes the open conformation relative to the closed. We tested these mutants in cells with time-resolved fluorescence resonance energy transfer and death assays, and show remarkable agreement with the calculations. The single mutants had a small effect on TM domain separation and cell death, whereas the double mutant significantly increased the TM domain separation and more than doubled the sensitivity of cells to ligand stimulation.

An Innovative High-Throughput Screening Approach for Discovery of Small Molecules That Inhibit TNF Receptors.

Tumor necrosis factor receptor 1 (TNFR1) is a transmembrane receptor that binds tumor necrosis factor or lymphotoxin-alpha and plays a critical role in regulating the inflammatory response. Upregulation of these ligands is associated with inflammatory and autoimmune diseases. Current treatments reduce symptoms by sequestering free ligands, but this can cause adverse side effects by unintentionally inhibiting ligand binding to off-target receptors. Hence, there is a need for new small molecules that specifically target the receptors, rather than the ligands. Here, we developed a TNFR1 FRET biosensor expressed in living cells to screen compounds from the NIH Clinical Collection. We used an innovative high-throughput fluorescence lifetime screening platform that has exquisite spatial and temporal resolution to identify two small-molecule compounds, zafirlukast and triclabendazole, that inhibit the TNFR1-induced IκBα degradation and NF-κB activation. Biochemical and computational docking methods were used to show that zafirlukast disrupts the interactions between TNFR1 pre-ligand assembly domain (PLAD), whereas triclabendazole acts allosterically. Importantly, neither compound inhibits ligand binding, proving for the first time that it is possible to inhibit receptor activation by targeting TNF receptor-receptor interactions. This strategy should be generally applicable to other members of the TNFR superfamily, as well as to oligomeric receptors in general.

Basic residue at position 14 is not required for fast assembly and disassembly kinetics in neural cadherin.

In spite of their structural similarities, epithelial (E-) and neural (N-) cadherin are expressed at different types of synapses and differ significantly in their dimerization kinetics. Recent studies proposed a transient intermediate in E-cadherin as the key requirement for rapid disassembly kinetics of the adhesive dimer. This E-cadherin intermediate comprises four intermolecular ionic and H-bonding interactions between adhesive partners. These interactions are not preserved in N-cadherin except for a basic residue at the 14th position, which could stabilize the intermediate through either H-bonding or ionic interactions with the partner protomer. To investigate the origin of the rapid dimerization kinetics of N-cadherin in the presence of calcium, studies reported here systematically test the role of ionic and H-bonding interactions in dimerization kinetics using R14S, R14A, and R14E mutants of N-cadherin. Analytical size-exclusion chromatographic and bead aggregation studies showed two primary results. First, N-cadherin/R14S and N-cadherin/R14A mutants showed fast assembly and disassembly kinetics in the calcium-saturated state similar to that of wild-type N-cadherin. These results indicate that the fast disassembly of the calcium-saturated dimer of N-cadherin does not require a basic residue at the 14th position. Second, the dimerization kinetics of N-cadherin/R14E were slow in the calcium-saturated state, indicating that negative charge destabilizes the intermediate state. Taken together, these results indicate that the basic residue at the 14th position does not promote rapid dimerization kinetics but that an acidic amino acid in that position significantly impairs dimerization kinetics.

Impact of pH on the structure and function of neural cadherin.

Neural (N-) cadherin is a transmembrane protein within adherens junctions that mediates cell-cell adhesion. It has 5 modular extracellular domains (EC1-EC5) that bind 3 calcium ions between each of the modules. Calcium binding is required for dimerization. N-Cadherin is involved in diverse processes including tissue morphogenesis, excitatory synapse formation and dynamics, and metastasis of cancer. During neurotransmission and tumorigenesis, fluctuations in extracellular pH occur, causing tissue acidosis with associated physiological consequences. Studies reported here aim to determine the effect of pH on the dimerization properties of a truncated construct of N-cadherin containing EC1-EC2. Since N-cadherin is an anionic protein, we hypothesized that acidification of solution would cause an increase in stability of the apo protein, a decrease in the calcium-binding affinity, and a concomitant decrease in the formation of adhesive dimer. The stability of the apo monomer was increased and the calcium-binding affinity was decreased at reduced pH, consistent with our hypothesis. Surprisingly, analytical SEC studies showed an increase in calcium-induced dimerization as solution pH decreased from 7.4 to 5.0. Salt-dependent dimerization studies indicated that electrostatic repulsion attenuates dimerization affinity. These results point to a possible electrostatic mechanism for moderating dimerization affinity of the Type I cadherin family. Extrapolating these results to cell adhesion in vivo leads to the assertion that decreased pH promotes adhesion by N-cadherin, thereby stabilizing synaptic junctions.

X-interface is not the explanation for the slow disassembly of N-cadherin dimers in the apo state.

In spite of structural similarities Epithelial- (E-) and Neural- (N-) cadherins are expressed at two types of synapses and differ significantly in dimer disassembly kinetics. Recent studies suggested that the formation of an X-dimer intermediate in E-cadherin is the key requirement for rapid disassembly of the adhesive dimer (Harrison et al., Nat Struct Mol Biol 2010;17:348-357 and Hong et al., J Cell Biol 2011;192:1073-1083). The X-interface in E-cadherin involves three noncovalent interactions, none of which is conserved in N-cadherin. Dimer disassembly is slow at low calcium concentration in N-cadherin, which may be due to the differences in the X-interface residues. To investigate the origin of the slow disassembly kinetics we introduced three point mutations into N-cadherin to provide the opportunity for the formation of X-interface interactions. Spectroscopic studies showed that the triple mutation did not affect the stability or the calcium-binding affinity of the X-enabled N-cadherin mutant. Analytical size exclusion chromatography was used to assay for the effect of the mutation on the rate of dimer disassembly. Contrary to our expectation, the disassembly of dimers of the X-enabled N-cadherin mutant was as slow as seen for wild-type N-cadherin in the apo-state. Thus, the differences in the X-interface residues are not the origin of slow disassembly kinetics of N-cadherin in the apo-state.

Calcium-induced strain in the monomer promotes dimerization in neural cadherin.

Cadherins are cell adhesion proteins that are important for tissue formation and integrity. Cell-cell adhesion occurs through the formation of the strand-crossover dimer between identical cadherins on the surface of neighboring cells. The strand-crossover dimer forms exclusively between their EC1 domains via swapping of the ßA sheet by undocking the conserved tryptophan 2, W2, from its own hydrophobic pocket and docking it into the hydrophobic pocket of its adhesive partner. An interesting aspect of the system is the fact that critical noncovalent interactions in the monomer re-form in the dimer. Thus, as these noncovalent interactions are conserved, what drives the formation of dimer? Moreover, why is dimer formation calcium-dependent? Thus, to probe the structural and energetic effects of calcium on the noncovalent interactions that are necessary for dimer formation, we performed spectroscopic, stability, and assembly studies of wild-type and two mutants, W2A and E89A, of neural (N-) cadherin. We find that while the ionic interaction involving E89 has a minimal effect on the general stability of the closed conformation of the ßA sheet, the hydrophobic interaction involving W2 is the source of the calcium requirement for adhesive dimer formation. The binding of calcium creates strain in the W2-hydrophbic pocket interaction through direct connection of E11 at the C-terminus of the ßA sheet to calcium. To overcome this unfavorable condition in the monomer, N-cadherin forms a dimer. Taken together, our data provide a thermodynamic basis for the calcium dependence of strand-crossover dimer formation in N-cadherin.

Stability studies of extracellular domain two of neural-cadherin.

Neural- (NCAD) and epithelial- (ECAD) cadherin are calcium-dependent cell-adhesive molecules, and are localized at excitatory and inhibitory synapses respectively. They play an important role in synaptogenesis, synapse maintenance and plasticity. The extracellular region plays a critical role in cadherin-mediated cell adhesion, and has five tandemly repeated ectodomains (EC1-EC5). Calcium binding is required for dimer formation between first two N-terminal domains (EC1-EC2). Despite similarity in the primary structure, the extracellular domains of NCAD and ECAD have different intrinsic stability, dimerization affinity and kinetics of disassembly. To investigate the origin of these differences, we are characterizing the modular domains individually. Here, we report studies of NCAD2, EC2 of NCAD. This domain is important for calcium binding and is the physical linkage between the dimerization interface in EC1 and the membrane proximal modular domains. Thermal-denaturation studies show that NCAD2 is less stable than ECAD2 and less influenced by the adjoining 7-residue, N- and C-terminal linker segments. In addition the NCAD2 constructs are less influenced by added salt. This difference is likely due to variation in the overall number and distribution of charges on these anionic proteins. Our studies indicate that despite their sequence similarity and apparently passive role in adhesive dimer formation, EC2 of E- and N-cadherins are distinctly different and may contribute to the differences in energetics and kinetics of dimerization.

Prolines in ßA-sheet of neural cadherin act as a switch to control the dynamics of the equilibrium between monomer and dimer.

Neural cadherins dimerize through the formation of calcium-dependent strand-crossover structures. Dimerization of cadherins leads to cell-cell adhesion in multicellular organisms. Strand-crossover dimer forms exclusively between the first N-terminal extracellular modules (EC1) of the adhesive partners via swapping of their ßA-sheets and docking of tryptophan-2 in the hydrophobic pocket. In the apo-state wild-type cadherin is predominantly monomer, which indicates that the dimerization is energetically unfavorable in the absence of calcium. Addition of calcium favors dimer formation by creating strain in the monomer and lowering the energetic barrier between monomer and dimer. Dynamics of the monomer-dimer equilibrium is vital for plasticity of synapses. Prolines recurrently occur in proteins that form strand-crossover dimer and are believed to be the source of the strain in the monomer. N-cadherins have two proline residues in the ßA-sheet. We focused our studies on the role of these two prolines in calcium-dependent dimerization. Spectroscopic, electrophoretic, and chromatopgraphic studies showed that mutations of both prolines to alanines increased the dimerization affinity by ~20-fold and relieved the requirement of calcium in dimerization. The P5A and P6A mutant formed very stable dimers that required denaturation of protein to disassemble in the apo conditions. In summary, the proline residues act as a switch to control the dynamics of the equilibrium between monomer and dimer which is crucial for the plasticity of synapses.

Dimeric states of neural- and epithelial-cadherins are distinguished by the rate of disassembly.

Epithelial- and neural-cadherins are specifically localized at synapses in neurons which can change the shape and contact surface on a time scale of seconds to months. We have focused our studies on the role of the extracellular domains of cadherins in the dynamics of synapses. The kinetics of dimer disassembly of the first two extracellular domains of E- and N-cadherin, ECAD12 and NCAD12, were studied with analytical size exclusion chromatography and sedimentation velocity. NCAD12 forms three different dimers that are distinguished by assembly conditions and kinetics of dissociation. ECAD12 dimer disassembles rapidly regardless of the calcium concentration, whereas the disassembly of NCAD12 dimers was strongly dependent on calcium concentration. In addition to the apo- and saturated-dimeric forms of NCAD12, there is a third dimeric form that is a slow exchange dimer. This third dimeric form for NCAD12, formed by decalcification of the calcium-saturated dimer, was kinetically trapped in apo-conditions and did not disassemble over a period of months. Sedimentation velocity experiments showed that this dimer, upon addition of calcium, had similar weighted averages as a calcium-saturated dimer. These studies provide evidence that the kinetics of dimer disassembly of the extracellular domains may be a major contributor to the morphological dynamics of synapses in vivo.

Sequential binding of calcium leads to dimerization in neural cadherin.

Neural cadherin (N-cadherin) is a calcium-dependent homophilic cell-adhesive molecule and critical for synaptogenesis and synapse maintenance. The extracellular region plays an important role in cadherin-mediated cell adhesion and has five tandemly repeated ectodomains (EC1-EC5) with three calcium-binding sites situated between each of these domains. Adhesive dimer formation is significantly dependent on binding of calcium such that mutations in the calcium-binding sites adversely affect cell adhesion. To investigate the relative significance of the calcium-binding sites at the EC1-EC2 interface in calcium-induced dimerization, we mutated three important amino acids, D134, D136, and D103, in NCAD12, a construct containing EC1 and EC2. Spectroscopic and chromatographic experiments showed that all three mutations affected calcium binding and dimerization. Mutation of D134, a bidentate chelator in site 3, severely impaired the binding of calcium to all three sites. These findings confirm that binding to site 3 is required for binding to occur at site 2 and site 1. Interestingly, while the D103A mutation diminished only the affinity for calcium, it completely eliminated dimerization. Equilibrium dialysis experiments showed a stoichiometry of 3 at 2 mM calcium for D103A, but no dimerization was apparent even at 10 mM calcium. These results indicate that calcium binding alone is not sufficient for dimerization but requires cooperativity between calcium-binding sites. In summary, our findings confirm that the calcium-binding sites are occupied sequentially in the order of site 3, then site 2 and site 1, and that cooperativity between site 2 and site 1 is essential for formation of adhesive dimers by N-cadherin.