Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract How proteins harness mechanical force to control function is a significant biological question. Here we describe a human cell surface receptor that couples ligand binding and force to trigger a chemical event which controls the adhesive properties of the receptor. Our studies of the secreted platelet oxidoreductase, ERp5, have revealed that it mediates release of fibrinogen from activated platelet αIIbβ3 integrin. Protein chemical studies show that ligand binding to extended αIIbβ3 integrin renders the βI-domain Cys177-Cys184 disulfide bond cleavable by ERp5. Fluid shear and force spectroscopy assays indicate that disulfide cleavage is enhanced by mechanical force. Cell adhesion assays and molecular dynamics simulations demonstrate that cleavage of the disulfide induces long-range allosteric effects within the βI-domain, mainly affecting the metal-binding sites, that results in release of fibrinogen. This coupling of ligand binding, force and redox events to control cell adhesion may be employed to regulate other protein-protein interactions. https://doi.org/10.7554/eLife.34843.001 eLife digest Many proteins embedded in a cell’s surface allow the cell to interact with its surroundings. Integrins are a group of cell surface proteins that have many uses in different cells. Integrins become activated when they come into contact with other specific proteins, which like other molecules that bind to proteins are referred to collectively as “ligands”. Much research has focused on how ligands become attached to integrins and how this activates these cell surface proteins. Yet how integrins release ligands and become inactive has not been studied before. One type of integrin, called αIIbβ3, is involved in blood clotting. Found on the surface of blood platelets – the fragments of cells in the blood that play a central role in clotting, this integrin binds to a ligand called fibrinogen. Fibrinogen links platelets together to form clots by building bridges between integrins. Passam, Chiu et al. have now studied platelets from donated human blood to understand how the integrin αIIbβ3 disengages from fibrinogen. The investigation showed that an enzyme called ERp5 aids the release of fibrinogen from the integrin. ERp5 can be released by blood vessel walls and by activated platelets. The experiments revealed that ERp5 breaks a chemical link, called a disulfide bond, in the integrin, but only when the protein is already bound to its ligand. Breaking the disulfide bond (a chemical process known as reduction) changes the integrin’s structure so that it lets go of fibrinogen. Moreover, when physical forces such as blood flow put the integrin under strain, the ERp5 enzyme becomes more effective. These findings show how ligand binding and mechanical force work together to control the breaking of a chemical bond in a human integrin. This chemical event then in turn controls the release of the integrin's ligand. It is possible that other protein-protein interactions may involve similar mechanisms, but this remains to be explored. Lastly, Passam, Chiu et al. suggest that the release of fibrinogen might help to limit the growth of blood clots so they do not block the blood vessels. Further studies should test this hypothesis. Inappropriate clotting can have severe health effects including heart attacks and strokes. As such, this investigation may hint at a more subtle way to regulate clotting through integrin αIIbβ3, such as boosting fibrinogen release to see if it helps slow or reduce clotting without stopping it altogether. https://doi.org/10.7554/eLife.34843.002 Introduction Protein function is controlled by a variety of chemical modifications to amino acid side chains, cleavage or isomerization of peptide bonds and cleavage or formation of disulfide bonds (Butera et al., 2014a; Cook and Hogg, 2013). These chemical reactions are usually facilitated by enzymes and their cofactors. Mechanical force is another factor that is increasingly being recognized to control protein chemical reactions (Garcia-Manyes and Beedle, 2017; Cross, 2016). Mechanical force can markedly reduce the reaction energy barrier. Reactions that are too slow become relevant on a biological time scale and others would not occur at all without the input of force. Low forces can trigger bond rotation and rupture of hydrogen bonds, while high forces can break or form covalent bonds. A number of cell surface receptors have been shown to be regulated by mechanical forces (Chen et al., 2017), including the integrins. Vertebrates express 24 different integrins that comprise one of 18 different α-subunits and one of 8 different β-subunits (Hynes, 2002). Integrin-mediated adhesion and signalling events regulate virtually all cell growth and differentiation, while dysregulation of integrins are involved in the pathogenesis of cancer, auto-immune conditions and vascular thrombosis. The integrin heterodimers recognize overlapping but distinct sets of ligands on other cells, extracellular matrix or on pathogens. The integrins are type one trans-membrane proteins that consist of large extracellular segments characterized by various domains and small transmembrane and cytoplasmic segments. Most integrins exist on the cell surface in an inactive state that does not bind ligand or signal. Integrins are activated by intracellular stimuli such as talin binding through a process termed inside-out signalling, and by ligand occupancy that transduces extracellular signals into the cytoplasm through outside-in signalling (Tadokoro et al., 2003; Luo et al., 2007). Affinity for ligands is mostly controlled by global and local conformational rearrangements of the integrin ectodomains. The extracellular segments exist in at least three conformational states: the bent conformation with closed headpiece (low affinity for ligand), the extended conformation with closed headpiece (intermediate affinity) and the extended conformation with open headpiece (high-affinity) (Zhu et al., 2013). While much is known about the structure and function of resting and activated integrins, little is known about how integrins disengage from their ligands. For instance, when cells migrate, integrins adhere at the leading edge and de-adhere at the trailing edge. Force has been suggested as one of the mechanisms to explain both the engagement and disengagement of ligand (Zhu et al., 2008). It was proposed that binding of the integrin β subunit cytoplasmic domain to actin filaments results in lateral translocation of the integrin heterodimer on the cell surface that causes integrin extension. Engagement of immobilised extracellular ligand greatly increases this lateral force that favours the high-affinity, open headpiece conformation. Disassembly of the actin cytoskeleton and dissociation of the β subunit cytoplasmic domain removes the lateral force that results in the closed headpiece conformation and ligand disengagement. In support of this model, cells become highly elongated and stop migrating when integrins are locked in the high-affinity conformation or when actin disassembly is blocked in the uropod (Smith et al., 2007). Here we report a chemical modification of an activated integrin that results in disengagement of ligand. Platelet clumping at sites of blood vessel injury is mediated by cross-linking of platelet αIIbβ3 integrin by the bivalent ligand, fibrinogen. This integrin is critical for thrombus formation and is the target of successful anti-thrombotic agents in routine clinical use for acute coronary syndrome. ERp5 is a protein disulfide isomerase (PDI) family member oxidoreductase released from platelets upon activation (Jordan et al., 2005), and from platelets and endothelial cells at sites of thrombosis in mice (Passam et al., 2015). Human platelets contain about 13,300 molecules of ERp5 per platelet (Burkhart et al., 2012), while mouse platelets contain an estimated 60,000 molecules (Zeiler et al., 2014). Secreted ERp5 binds to the β3 subunit of platelet surface αIIbβ3 integrin (Jordan et al., 2005; Passam et al., 2015). Systemic inhibition of ERp5 with function-blocking antibodies inhibits thrombosis in mice, which implies an essential role for secreted ERp5 in this biology (Passam et al., 2015). Our studies indicate that ERp5 mediates de-adhesion of activated platelet αIIbβ3 integrin from fibrinogen. Ligand binding to extended αIIbβ3 integrin triggers cleavage of the βI-domain Cys177-Cys184 disulfide by ERp5, which is enhanced by mechanical force. Cleavage of the disulfide results in release of fibrinogen due to allosteric effects at the metal-ion-dependent adhesion (MIDAS) site. Results ERp5 triggers fibrinogen dissociation from αIIbβ3 integrin Immunoblotting of human platelet lysate and releasate indicates that approximately half of the platelet ERp5 molecules are released into the supernatant upon activation (Figure 1—figure supplement 1). ERp5 binds to β3 integrin with a dissociation constant in the low micromolar range (Passam et al., 2015), and an anti-ERp5 antibody has been reported to inhibit fibrinogen binding to activated platelets and platelet aggregation in vitro (Jordan et al., 2005). These findings suggested that ERp5 directly regulates platelet αIIbβ3 integrin function, although the mechanism remains elusive. We examined the effect of soluble ERp5 on platelet αIIbβ3 integrin activation and fibrinogen binding. Incubation of washed platelets with ERp5 did not trigger αIIbβ3 integrin activation, and ERp5 had no effect on integrin activation by ADP (Figure 1A). Integrin activation was measured by binding of PAC-1, an antibody that recognizes the fully activated integrin with an open headpiece (Shattil et al., 1985; Luo et al., 2003). This result indicated that ERp5 is not directly involved in integrin activation, so we explored a role for ERp5 in post-activation events. Effect of soluble ERp5 on the kinetics of adhesion of washed platelets to fibrinogen as a function of fluid shear force was examined (Figure 1B). There was a biphasic effect of ERp5 on platelet binding to fibrinogen at two wall shear stress (Figure 1C). At 10 dyn/cm2 (1000 s−1) platelet adhesion in the first 4 min of flow was unaffected by ERp5 and reduced with time thereafter. The negative effect of ERp5 on platelet adhesion was more pronounced at 30 dyn/cm2 (3000 s−1). Platelet adhesion in the first 1 min of flow was unaffected by ERp5 and reduced significantly with time thereafter. This finding suggested that ERp5 was triggering dissociation of fibrinogen from activated platelet αIIbβ3 integrin. To better understand this mechanosensitive phenomenon, the effect of ERp5 on binding of fibrinogen to αIIbβ3 integrin was characterized using a force spectroscopy technique - biomembrane force probe (BFP) (Ju et al., 2016, Ju et al., 2013). Figure 1 with 1 supplement see all Download asset Open asset ERp5 triggers αIIbβ3 integrin de-adhesion from fibrinogen. (A) ERp5 does not induce αIIbβ3 integrin activation on platelets, or influence integrin activation by ADP. The activated αIIbβ3 is measured by PAC-1 antibody binding. The data points and errors (1 SD) are from measurements of four different healthy donor platelets. (B–D) Platelet adhesion to fibrinogen under flow is impaired by ERp5. Part B are representative images of platelet adhesion to immobilized fibrinogen in the absence of presence of 2 µM ERp5. Parts C and D are kinetics of platelet adhesion to fibrinogen in the absence or presence of 2 µM ERp5 at fluid shear rates of 1000 s−1 or 3000 s−1, respectively. The data points and errors (1 SD) are from three measurements each from three different healthy donor platelets. *p<0.05, ****p<0.001; assessed by unpaired, two-tailed Student’s t-test. https://doi.org/10.7554/eLife.34843.003 The BFP brings an αIIbβ3 integrin coated bead into contact with a fibrinogen bearing force probe (Figure 2A). Upon target retraction, it detects the αIIbβ3 bond from the pico-force signal measured (Figure 2B, red), while a zero force indicates a no-bond event (Figure 2B, black). The intermittent ‘touch and retract’ cycles mimic platelet translocation behavior under shear (Ju et al., 2013; Yago et al., 2008). The adhesion frequencies (the number of ‘bond’ touches (Figure 2F, red) divided by the number of total touches) reflects the binding affinity at zero tensile force. Unexpectedly, we found that soluble ERp5 had no significant effect on the αIIbβ3–fibrinogen adhesion frequencies (Figure 2C). To investigate the force effect, we measured αIIbβ3 integrin–fibrinogen bond lifetimes at multiple clamped forces in the absence or presence of ERp5. This interaction is characterized by slip-bond behaviors in which force accelerates bond dissociation, consistent with the previous optical tweezer study (Litvinov et al., 2011). Surprisingly, at the low force regime of 5–15 pN, ERp5 displayed a similarly non-significant effect on the αIIbβ3–fibrinogen bond lifetimes as the adhesion frequency assay, whereas as force goes beyond 15 pN, it greatly enhanced αIIbβ3–fibrinogen dissociation with reduced bond lifetimes (Figure 2D). In accordance with the perfusion experiments, these findings indicate that ERp5 has a force-dependent de-adhesive effect on the αIIbβ3–fibrinogen interaction. Figure 2 Download asset Open asset Mechanical force accelerates ERp5-induced fibrinogen de-adhesion. (A) Biomembrane-Force-Probe (BFP) detection of αIIbβ3–fibrinogen (FBG) binding in a purified system. The probe bead was coated with fibrinogen and streptavidin (SA) for attachment of the bead to biotinylated red blood cell (RBC). Fibrinogen is the focus for interaction with αIIbβ3 on the target bead. The effect of soluble ERp5 on this interaction was measured. (B) Force versus time traces from two representative test cycles. The target bead was driven to approach and contact the probe bead, then retracted. In a 'no bond' event (black), the cycle ended after the probe–target separation. In a 'bond' event (red), the target was clamped (marked by *) at a preset force until dissociation. Lifetime is measured from the point when the clamped force (i.e. 30 pN) was reached to the point when the bond dissociated, signified by a force drop to zero. (C) Soluble ERp5 has no significant effect on αIIbβ3–fibrinogen adhesion frequency. The adhesion frequencies represent mean ± SEM of n = 3 independent experiments in the absence or presence of 2 µM ERp5. For each experiment, five random probe–target pairs with 50 touches were analyzed and averaged. (D) Lifetime of fibrinogen–αIIbβ3 integrin bonds vs. clamp force in the absence or presence of 2 µM ERp5. Results represent mean ±SEM of 10–30 measurements at each force bin. *p<0.05, **p<0.01; assessed by unpaired, two-tailed Student’s t-test. https://doi.org/10.7554/eLife.34843.005 ERp5 cleaves the βI-domain Cys177-Cys184 disulfide bond ERp5 is an oxidoreductase that can cleave, form or potentially rearrange disulfide bonds in protein substrates. From crystal structures of the complete ectodomain of αIIbβ3 integrin (Zhu et al., 2008), 28 disulfide bonds have been defined in the β3 subunit. No unpaired cysteines within a few Angstoms of each other were identified, which implied that all possible disulfide bonds in the subunit are intact. We hypothesized that the functional effect of ERp5 on fibrinogen binding was a result of its cleavage of one or more of the 28 β3 disulfide bonds. This was tested by measuring the presence of unpaired cysteine thiols in platelet surface β3 before and after activation by ADP. Platelet activation resulted in increased labelling of β3 by a thiol-specific probe (Figure 3A), indicating that one or more disulfide bonds were cleaved in the subunit upon platelet activation. Figure 3 with 2 supplements see all Download asset Open asset ERp5 cleaves the βI-domain Cys177-Cys184 disulfide bond. (A) One or more disulfide bonds are cleaved in platelet β3 integrin upon platelet activation. Washed platelets were untreated or activated with ADP and unpaired cysteine thiols in platelet surface proteins labelled with the biotin-linked maleimide, MPB. The β3 integrin was immunoprecipitated from the platelet lysate and blotted for MPB. (B) Differential cysteine alkylation and mass spectrometry method of measuring the redox state of the β3 integrin cysteines. Unpaired cysteine thiols in purified β3 integrin are alkylated with 12C-IPA and the disulfide bonded cysteine thiols with 13C-IPA following reduction with DTT. Sixty-eight peptides (Supplementary file 1) encompassing cysteines representing 24 of the 28 β3 integrin disulfide bonds were analyzed. (C) Positions of the β3 integrin disulfide bonds (yellow and grey spheres) in a modelled open structure of the complete αIIbβ3 integrin ectodomain (blue ribbon). We were able to map 24 of the 28 β3 disulfide bonds. The mapped disulfides are in yellow and the unmapped bonds are in grey. (D–E) Redox state of 24 of the 28 β3 integrin disulfide bonds in the absence or presence of 2- or 10-fold molar excess of ERp5 or redox inactive ERp5 (D), or PDI or redox inactive PDI (E). (F–G) Redox state of 24 of the 28 β3 integrin disulfide bonds in the absence or presence of 2- or 10-fold molar excess of ERp5 or redox inactive ERp5 (D), or PDI or redox inactive PDI (E), and 1 mM RGD peptide. The bars and errors (1 SD) are for 5–15 measurements from three different integrin preparations. The β3 Cys177-Cys184 disulfide bond (indicated by red arrow) is significantly cleaved by 10-fold molar excess of ERp5 (p<0.05) (H) The βI-domain Cys177-Cys184 disulfide bond is cleaved on the platelet surface by platelet ERp5. Washed platelets were incubated with function-blocking anti-ERp5 antibodies or isotype control antibodies and the redox state of the βI-domain disulfide determined in the integrin immunoprecipitated from lysate. The bars and errors (1 SD) are from three different platelet preparations from healthy donors. ***p<0.005; ****p<0.001; assessed by unpaired, two-tailed Student’s t-test. https://doi.org/10.7554/eLife.34843.006 To identity the β3 disulfide(s) cleaved by ERp5, it was critical that we accurately quantify the redox state of the subunits disulfide bonds. This was achieved using a differential cysteine alkylation and mass spectrometry technique (Pasquarello et al., 2004; Bekendam et al., 2016). Briefly, reduced disulfide bond cysteines in purified platelet β3 integrin were alkylated with 2-iodo-N-phenylacetamide (IPA), and the oxidized disulfide bond cysteines with a stable carbon-13 isotope of IPA following reduction with dithiothreitol (Figure 3B). Sixty-eight cysteine containing peptides (Supplementary file 1) reporting on 24 of the 28 β3 integrin disulfides were resolved by mass spectrometry and quantified (Figure 3—figure supplement 1). The four disulfides we were unable to map are the Cys528-Cys542 and Cys536-Cys547 bonds of the EGF-3 domain, Cys575-Cys586 of the EGF-4 domain and Cys601-Cys604 of the Ankle domain (Figure 3C). Purified platelet αIIbβ3 integrin was incubated with 2- or 10-fold molar excess of ERp5 or PDI and the redox state of the disulfides quantified. PDI was used to test the substrate specificity of ERp5. PDI is the archetypal member of the PDI family of oxidoreductases, which includes ERp5, and is also secreted by platelets and endothelial cells at sites of thrombosis in mice and binds to surface β3 integrin (Cho et al., 2008; Cho et al., 2012). Reactions were also performed with redox-inactive ERp5 and PDI to test the redox dependence of their action. Redox-inactive ERp5 and PDI were produced by replacing the active-site cysteines with alanines. In accordance with crystal structures of the integrin ectodomain, all 24 disulfide bonds in untreated β3 integrin were >90% oxidized, with one exception (Figure 3D and E). Approximately 10% of the βI-domain Cys177-Cys184 bond was reduced in the β3 preparations. Incubation of the integrin with ERp5 or PDI did not significantly change the redox sate of any of the 24 disulfide bonds (Figure 3D and E). Our fibrinogen binding studies suggested that ERp5 was having an effect on the extended/activated integrin. The native integrin exists predominantly in a bent conformation with closed headpiece. Soaking of the αIIbβ3 integrin headpiece with RGD peptide ligand results in variably extended configurations (Zhu et al., 2013). Six intermediate conformations and fully extended conformation with open headpiece have been described. Incubation of RGD-bound αIIbβ3 integrin with ERp5 resulted in significant cleavage of only one of the 24 β3 disulfides: the βI-domain Cys177-Cys184 bond. There was dose-dependent cleavage of the disulfide by ERp5 (Figure 3F) and the bond was not cleaved by PDI (Figure 3G), indicating selectivity for ERp5. Control RGE peptide that does not bind the integrin did not facilitate cleavage of the disulfide by ERp5 (Figure 3—figure supplement 2). As anticipated, redox-inactive ERp5 did not cleave the bond (Figure 3F). These results indicate that ERp5 specifically cleaves the βI-domain Cys177-Cys184 disulfide in one or more of the extended/open αIIbβ3 conformations. Approximately 30% of the Cys177-Cys184 disulfide bond in the purified integrin preparation is cleaved by 10-fold molar excess of ERp5 under static conditions, which is a ~20% increase over baseline. The force spectroscopy findings suggest that extent of cleavage is likely to be higher when the integrin is subject to mechanical shearing. ERp5-mediated dissociation of fibrinogen from αIIbβ3 is greatly enhanced when force goes beyond 15 pN and is complete at 40 pN (Figure 2D). The βI-domain Cys177-Cys184 disulfide bond is also cleaved on the platelet surface by platelet ERp5. Washed human platelets were incubated function-blocking anti-ERp5 antibodies or isotype control antibodies and the redox state of the βI-domain disulfide determined in the integrin immunoprecipitated from lysate. Lysis of platelets releases stored ERp5 and fibrinogen that was predicted to mediate cleavage of the Cys177-Cys184 disulfide bond. The Cys177-Cys184 disulfide bond was reduced in ~10% of the integrin population in untreated or control antibody treated platelet lysate (Figure 3H), which is in accordance the redox state of this bond in purified platelet αIIbβ3 (Figure 3D–G). Incubation of platelets with function-blocking anti-ERp5 antibodies during lysis inhibited reduction of the Cys177-Cys184 bond (p<0.001) (Figure 3H). The Cys177-Cys184 disulfide has an allosteric configuration The Cys177-Cys184 disulfide is a –/+RHhook in all crystal structures of the protein, which includes bent, extended, ligand-free and ligand-bound structures (Supplementary file 2). Disulfide bonds are classified based on the geometry of the five dihedral angles that define the cystine residue (Schmidt et al., 2006). Twenty possible disulfide bond configurations are possible using this classification scheme and all 20 are represented in protein structures. Some disulfide bonds are cleaved in the mature protein to control function (Hogg, 2003), the so-called allosteric bonds (Schmidt et al., 2006), and these disulfides are increasingly recognized to have one of three configurations (Butera et al., 2014a). The –/+RHhook is one of these configurations, along with –RHstaple and –LHhook bonds (Butera et al., 2014b). The conformational constraints imposed on the –/+RHhook and –RHstaple disulfides by topological features stress the bonds via direct stretching of the sulfur-sulfur bond and neighbouring angles (Schmidt et al., 2006; Zhou et al., 2014). Stretching of sulfur-sulfur bonds increases their susceptibility to cleavage (Baldus and Gräter, 2012; Li and Gräter, 2010; Wiita et al., 2006; Wiita et al., 2007), so this internal stress fine tunes bond cleavage and thus the function of the protein in which the bond resides. Both catalytic domains of ERp5 are required for substrate specificity The position of the Cys177-Cys184 disulfide relative to the ligand binding pocket and access to the bond by ERp5 was examined in the crystal structure of the extended holo headpiece (PDB code 2vdo). Cys184 of the bond is surface exposed (Supplementary file 2) on a face of the βI-domain that is remote from the ligand binding pocket (Figure 4A). This suggests that the Cys184 sulfur atom of the disulfide is attacked by ERp5 to cleave the bond and explains why ERp5 access is not blocked by RGD ligand binding. Figure 4 Download asset Open asset Both catalytic domains of ERp5 are required for substrate specificity. (A) Surface structure of the extended holo conformation headpiece (PDB code 2vdo) showing the βI-domain (blue), β-propeller domain of the α2b subunit (wheat) and fibrinogen γC peptide (red). Cys184 of the Cys177-Cys184 disulfide (yellow) is surface exposed in a pocket that is removed from the ligand binding pocket. (B) Redox potentials of the a (Cys55-Cys58) and a' (Cys190-Cys193) active-site dithiols/disulfides of ERp5. Plots of the fraction of reduced ERp5 as a function of the ratio of GSH to GSSG. The lines represent the best non-linear least squares fit of the data to Equation 1. The calculated equilibrium constants were used to determine the standard redox potentials from Equation 2. Data points and errors are the mean of 2–4 peptides encompassing the active site cysteine residues. (C) Cleavage of the Cys177-Cys184 disulfide bond by 2- or 10-fold molar excess of full-length ERp5, ERp5 N-terminal domain, or ERp5 C-terminal domain. The bars and errors (1 SD) are for 2–6 measurements. *p<0.05, ****p<0.001; assessed by unpaired, two-tailed Student's t-test. (D) Platelet adhesion to fibrinogen at 4 min in the absence or presence of 2 µM redox inactive ERp5 (riERp5), ERp5 or the N- or C-terminal ERp5 domains at a fluid shear rate of 1000 s−1. The inset is a representative image of platelet adhesion to immobilized fibrinogen in the presence of 2 µM ERp5 CTD. The bars and errors (1 SD) are from three measurements each from six different healthy donor platelets. (E) Aggregation of washed platelets activated with the PAR-1 agonist, TRAP (7 µM), in the absence or presence of 2 µM ERp5 or ERp5 CTD. The data points and errors (1 SEM) are from three different healthy donor platelets. https://doi.org/10.7554/eLife.34843.009 The ERp5 N-terminal part consists of two thioredoxin-like domains containing a catalytic dithiol/disulfide in CysGlyHisCys motifs, a and a’, separated by an x segment (Figure 4B). These domains are followed by a possible substrate binding domain, b. The redox potentials of the a (Cys55-Cys58) and a' (Cys190-Cys193) catalytic disulfides of ERp5 were determined using differential cysteine alkylation and mass spectrometry. The equilibrium data is shown in Figure 4B. The standard redox potentials of the a and a' domain disulfides of ERp5 are −206 mV and −211 mV, respectively. These redox potentials are about mid-way between the potentials of the PDI (Bekendam et al., 2016) and thioredoxin (Lundström and Holmgren, 1993) catalytic disulfides. N- and C-terminal fragments of ERp5 containing a single active-site were tested for cleavage of the Cys177-Cys184 disulfide. Both fragments cleaved the bond with the same efficiency as full-length protein (Figure 4C). This is in agreement with the equivalent redox potentials of the active-site dithiols/disulfides (Figure 4B). It also indicates that the substrate binding domain of ERp5 is not required for access to and cleavage of the Cys177-Cys184 disulfide. It was possible, though, that separating the two catalytic domains of ERp5 would influence substrate specificity, that is, the disulfide bond or bonds cleaved by ERp5. This was tested by examining the effect of the ERp5 fragments on adhesion of washed human platelets to fibrinogen in the first 4 min of flow at a shear rate of 1000 s−1. These conditions were chosen as full-length ERp5 has no effect over this time frame at this shear rate (Figure 1C). As for full-length ERp5 and redox-inactive ERp5, where the active site cysteines of both thioredoxin-like domains are replaced with serines, the N-terminal catalytic domain of ERp5 had no effect on platelet adhesion to fibrinogen under these conditions. In marked contrast, the C-terminal domain enhanced platelet adhesion to fibrinogen by ~50 fold (Figure 4D). Large platelet aggregates adhered to the fibrinogen-coated slides (Figure 4D inset). The C-terminal domain also enhanced platelet aggregation is response to a PAR-1 agonist, while full-length ERp5 had no effect (Figure 4E). These findings indicate that both catalytic domains in full-length ERp5 are required for specificity of cleavage of the Cys177-Cys184 disulfide. The result implies that separating the domains leads to cleavage of other disulfide bonds in the system and different functional effects. Cleavage of the βI disulfide results in reduced affinity for fibrinogen due to increased βI-domain flexibility and high stresses at the MIDAS site An intact Cys177-Cys184 disulfide bond was found to be required for normal fibrinogen binding in a 2004 structure/function study of the disulfide bonds in β3 integrin (Kamata et al., 2004). We confirmed this result by measuring binding of soluble or immobilized fibrinogen to BHK cells expressing wild-type or disulfide mutant (C177,184S) αIIbβ3 integrin in the absence or presence of the integrin activator, Mn2+. E