Decision letter: Cooperative unfolding of distinctive mechanoreceptor domains transduces force into signals

Abstract

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract How cells sense their mechanical environment and transduce forces into biochemical signals is a crucial yet unresolved question in mechanobiology. Platelets use receptor glycoprotein Ib (GPIb), specifically its α subunit (GPIbα), to signal as they tether and translocate on von Willebrand factor (VWF) of injured arterial surfaces against blood flow. Force elicits catch bonds to slow VWF–GPIbα dissociation and unfolds the GPIbα leucine-rich repeat domain (LRRD) and juxtamembrane mechanosensitive domain (MSD). How these mechanical processes trigger biochemical signals remains unknown. Here we analyze these extracellular events and the resulting intracellular Ca2+ on a single platelet in real time, revealing that LRRD unfolding intensifies Ca2+ signal whereas MSD unfolding affects the type of Ca2+ signal. Therefore, LRRD and MSD are analog and digital force transducers, respectively. The >30 nm macroglycopeptide separating the two domains transmits force on the VWF–GPIbα bond (whose lifetime is prolonged by LRRD unfolding) to the MSD to enhance its unfolding, resulting in unfolding cooperativity at an optimal force. These elements may provide design principles for a generic mechanosensory protein machine. https://doi.org/10.7554/eLife.15447.001 eLife digest Platelets – the blood clotting cells – have the ability to detect, interpret and respond to mechanical forces, such as those generated by the flow of blood. The magnitude and duration of the forces detected by the platelets influences whether they form a blood clot. Understanding how the platelets respond to mechanical forces is therefore crucial for our knowledge of conditions such as thrombosis, where blood clots form inside vessels and block them. Clots that form within arteries are associated with heart attack and stroke, which account for around one third of all deaths worldwide. Cells can sense external forces via individual proteins on their surface and transmit the mechanical information across the cell membrane. This triggers signals within the cell that influence how it responds. However, the molecular details of these “mechanosensory” processes remain poorly understood. To patch up damaged blood vessels, platelets use a protein on their surface named glycoprotein Ibα (GPIbα) to bind to a plasma protein called von Willebrand factor that adheres to the vessel wall. This binding tethers the platelet to the blood vessel and activates it during clot formation. Previous studies suggested that mechanical force affects how this binding triggers the signals that activate platelets. Ju, Chen et al. used a homebuilt nanotool to pull on platelet GPIbα while it was bound to von Willebrand factor. This revealed that two distinct domains of the GPIbα protein unfold to relay information about the force, such as its magnitude and duration, to the platelet to trigger biochemical signalling inside the cell. The unfolding of each GPIbα domain has a distinct role in determining the quantity and quality of the signals. The unfolding events work synergistically – they occur together to produce an effect that’s greater than the sum of their individual effects. However, pulling on GPIbα via a mutant form of von Willebrand factor eliminated the synergy between the two unfolding events, therefore hindering the effective conversion of mechanical forces into biochemical signals. Notably, the two GPIbα domains unfolded by force exist in many protein families, including those involved in mediating cell adhesion and detecting signals. The biophysical tools developed by Ju, Chen et al. could be extended to analyze how mechanical cues are presented, received, transmitted and converted into biochemical signals in other cell types and biological systems. Furthermore, the structural insights gained from the platelet GPIbα system may help to design a generic mechanosensory protein machine. https://doi.org/10.7554/eLife.15447.002 Introduction Platelets can serve as a natural model system for studying cell mechanosensing as they rapidly respond to changes in hydrodynamic forces and substrate stiffness due to vascular pathology (Jackson, 2011; Qiu et al., 2015). Previous studies have suggested the role of GPIbα as a mechanoreceptor, for force exerted on it via its ligand VWF induces platelet signaling (Ruggeri, 2015). Conceptually, this coupled mechanical-biochemical process (mechanosensing) can be broken down into four steps: 1) Mechanopresentation: the receptor binding domain A1 is exposed by structural changes in VWF induced by elongational flow and collagen immobilization (Ju et al., 2015a; Springer, 2014); 2) Mechanoreception: GPIbα LRRD receives the force signal via engaging VWF-A1 to tether the platelet against shear stress; 3) Mechanotransmission: force is propagated from the LRRD through the mucin-like macroglycopeptide (MP) stalk (cf. Figure 2A) (Fox et al., 1988) and the MSD across the membrane to adaptor and signaling molecules (e.g. 14-3-3ζ) inside the platelet (cf. Figure 7G); and 4) Mechanotransduction: force induces mechano-chemical changes to convert mechanical cues to biochemical signals. Some of these steps have been characterized separately. For example, GPIbα forms catch-slip bonds with wild-type (WT) A1 in >15 pN, such that the bond lifetime first increases with force, reaches a maximum at ~25 pN, and decreases thereafter; whereas it forms slip-only bonds with type 2B von Willebrand disease (VWD) mutant (e.g. A1R1450E), such that the bond lifetime decreases monotonically with force (Ju et al., 2013; Yago et al., 2008). As another example, force induces unfolding of the LRRD, which prolongs A1–GPIbα bond lifetime (Ju et al., 2015b), and of the MSD, which is hypothesized to play a role in platelet signaling (Zhang et al., 2015). However, how these inter-connected steps are orchestrated to enable the information encoded by force to be translated into biochemical signals is still poorly understood. We used a biomembrane force probe (BFP) to recapitulate the above process in a single-cell and single-molecular bond level to address the following questions: 1) What molecular events would be induced in GPIbα and how these events are regulated mechanically? 2) Whether, and if so, how changes in presentation of force by VWF-A1 mutation would affect the force reception by GPIbα and its response to force? 3) What features of the force (waveforms) could be sensed by the platelet via GPIbα to initiate intraplatelet calcium fluxes? 4) What proximal events may be responsible for transducing force into a biochemical signal? By manipulating the mechanopresentation and mechanoreception steps then analyzing the resulting mechanotransmission and mechanotransduction steps, we gained insights into the inner workings of this GPIbα-mediated mechanosensory machine. Results In the BFP setup, a probe bead was functionalized with VWF-A1 or an anti-GPIb monoclonal antibody (mAb) to serve as a surrogate subendothelial surface (Figure 1A,B). It was attached to the apex of a micropipette-aspirated red blood cell (RBC) to form an ultrasensitive force transducer (Liu et al., 2014). A platelet was aspired by the target pipette to contact the bead in repetitive force-ramp or force-clamp cycles to mimic the sequential formation, force loading, and dissociation of VWF–GPIbα bonds during the translocation of a platelet on the sub-endothelium (Video 1; Figure 1C and Figure 1—figure supplement 1; Materials and methods). Adhesion frequencies from these cycles were kept low (<20%) by adjusting the ligand or antibody density, a condition required for the platelet to be pulled predominantly (>89%) by a single GPIbα bond (Chesla et al., 1998; Zhu et al., 2002). Control experiments using beads lacking ligand showed no binding, and blocking with mAb AK2 (epitope mapped to leucine-rich repeat 1–2 overlapping the A1 binding site, cf. Figure 2A) eliminated GPIbα binding to A1 but not to mAb WM23 (epitope mapped to the MP below LRRD, cf. Figure 2A) (Figure 1D) (Dong et al., 2001). This confirmed binding specificity and that the binding site of A1 is within LRRD but the binding site of WM23 is outside (Zhang et al., 2015). Figure 1 with 2 supplements see all Download asset Open asset BFP analysis of ligand binding kinetics and domain unfolding mechanics of platelet GPIbα. (A) BFP micrograph. A micropipette-aspirated RBC with a probe bead attached to the apex (left) was aligned against a platelet held by an opposing micropipette (right). (B) BFP functionalization. The probe bead was coated with streptavidin (SA, for attachment to the biotinylated RBC) and VWF-A1 or mAb (left) for interaction with platelet GPIb (right). (C) Representative force vs. time traces of repetitive force-clamp cycles over a 200-s period. Cycles produced different results are color-coded (black: no bond; blue: bond-rupture; red: bond-lifetime). (D) Mean ± s.e.m. of adhesion frequencies (n ≥ 3) of platelets binding to beads functionalized with indicated proteins in the absence (open) or presence (closed) of 50 µg/ml AK2. The coating densities are 131 and 95 µm-2 for A1 and WM23 respectively. *** = p < 0.001 by Student t-test. (E) Force vs. time trace of a representative BFP cycle showing unfolding signatures in both ramping and clamping phases. The inset zooms in the ramped unfolding signature and indicates the unfolding length. (F) Zoom-in view of the clamped unfolding signature in (E). Higher displacement resolutions were obtained after smoothing the raw data (points) by the Savitzky-Golay method (curves). Time to unfolding (tu) is indicated. https://doi.org/10.7554/eLife.15447.003 Video 1 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg BFP experiment mimics platelet translocation on sub-endothelium. This animation (produced by Adobe Flash; 12 fps) illustrates the resemblance between platelet translocation on the sub-endothelium (a collagen network covered with VWF on the surface, upper panel) and the repetitive BFP experiment cycle (middle panel) with synchronized molecular interaction between a GPIbα and a VWF-A1 domain (lower panel). Two zoom-in platelet signaling scenarios are inserted following a short- and a long-lived VWF–GPIbα bond respectively. It starts with platelet translocation along the shear force direction, mimicked by the first two no-adhesion BFP cycles. An A1–GPIbα binding event with no lifetime in the BFP cycle results in a transient deceleration in the platelet translocation. After two more no-bond cycles, another bond event survives for a short lifetime without GPIbα unfolding. This triggers a signal (represented by purple stars) displaying a β-type Ca2+. Later, another bond event survives for a long lifetime, during which both LRRD and MSD unfold. This triggers a signal (represented by blue stars) displaying an α-type Ca2+. https://doi.org/10.7554/eLife.15447.006 Figure 2 with 1 supplement see all Download asset Open asset Identification and characterization of unfolding of LRRD and MSD. (A–C) Schematics of GPIbα on the platelet membrane (A), highlighting the folded (−) and unfolded (+) LRRD (B) and MSD (C). Different regions and binding sites for VWF-A1 and mAbs are indicated. (D–G) Normalized histograms (bar) and their multimodal Gaussian fits (curve) of GPIbα (or GC) unfolding lengths pulled by engaged A1 (D,E) or AN51 (F,G) in indicated probe–target pairs. Peak values and percentages of unfolding lengths are indicated along with the identified unfolding domains. (H,I) Validation of MSD (H) and LRRD (I) unfolding. The WLC model was fit (curves) to the unfolding force vs. length data (black circles, mean ± s.e.m. of 15–25 measurements per point) from the WM23 vs. platelet experiments where only MSD unfolding was possible (H) or A1 vs. GC experiments where only LRRD unfolding was possible (I), yielding a contour length of 25.99 ± 0.85 nm or 70.29 ± 3.56 nm, respectively. Overlying on the two panels are corresponding unfolding force vs. length data (red triangles, mean ± s.e.m. of 20–30 measurements per point) from A1 vs. platelet ramped experiments where unfolding of MSD, LRRD or both were all possible, but were segregated into putative MSD (H) and LRRD (I) unfolding groups based on our decision rules in Figure 3—source data 1A. (J,K) Mean ± s.e.m. (n ≥ 20) of unfolding frequency (J) and length (K) of LRRD (A1 vs. GC) and MSD (WM23 vs. platelet). Force waveforms indicated as ramped force (1000 pN/s) and clamped force (25 pN) were generated with force-ramp and force-clamp experiment modes respectively. N.D. = Not detected. https://doi.org/10.7554/eLife.15447.007 Identification of LRRD and MSD unfolding Using an optical trap, Zhang et al. observed force-induced MSD unfolding in purified recombinant full-length GPIb-IX and a GPIbα stalk region construct (Zhang et al., 2015). Using a BFP, we observed LRRD unfolding in glycocalicin (GC) (Ju et al., 2015b), the extracellular segment of GPIbα lacking the MSD (Liang et al., 2013) (Figure 2A–C). Here we pulled GPIbα on platelets via A1 and observed two unfolding signatures, one in the ramping and the other in the clamping phases of the force trace (Figure 1E). Unfolding that occurred in the ramping phase is termed ramped unfolding, which is featured by a sudden force kink at 5–20 pN as observed in previous studies of GPIbα unfolding (Ju et al., 2015b; Zhang et al., 2015). Similar to findings of protein unfolding studies (Kellermayer et al., 1997; Rief et al., 1997; Tskhovrebova et al., 1997; Zhang et al., 2009a, 2015), both the force-extension curves before and after unfolding were well fitted by the worm-like chain (WLC) model (Figure 1—figure supplement 2). Unfolding that occurred in the clamping phase is termed clamped unfolding, which is featured by an abrupt force drop (Figure 1F). Although not observed in the previous studies of GPIbα unfolding (Ju et al., 2015b; Zhang et al., 2015), this feature has been described in protein unfolding studies using force-clamp experiments (Oberhauser et al., 2001; Tskhovrebova et al., 1997). Unfolding lengths derived from both signatures were measured from the probe bead position vs. time data (Figure 1E insert, 1F and Figure 1—figure supplement 2). The lengths of individual ramped unfolding events distributed tri-modally with three subpopulations (Figure 2D and Figure 2—figure supplement 1; Materials and methods). The first subpopulation coincides with the ramped unfolding length distribution from WM23 vs. platelet experiments (Figure 2E, white bars). WM23 binds the MP region below the LRRD (Figure 2A), hence could unfold MSD only. The average unfolding force vs. length data from the WM23 experiment was well fitted by the WLC model, yielding a contour length of 25.99 ± 0.85 nm (Figure 2H) that matches the previously reported MSD contour length (Zhang et al., 2015). The average unfolding force vs. length data from the A1 experiment overlaid well on the same WLC model fit (Figure 2H). These results identify the first subpopulation in Figure 2D as MSD unfolding. The second subpopulation in Figure 2D matches the histogram of ramped unfolding lengths of GC pulled via A1 (Figure 2E, blue bars) that ranges from 18–56 nm and peaks at 36 nm (length of leucine-rich repeats 3–6). The average unfolding force vs. length plots derived from the A1 vs. platelet and A1 vs. GC experiments overlaid well on the same WLC model fit (Figure 2I). The best-fit contour length (70.29 ± 3.56 nm) matches the length of LRRD, calculated using a 4-Å contour length per residue (Ju et al., 2015b). These results identify the second subpopulation in Figure 2D as LRRD unfolding. The third subpopulation in Figure 2D can be identified as concurrent unfolding of both MSD and LRRD that occurred within too short a time elapse to be distinguished by our BFP as two separate events, because its maximum unfolding length (85 nm) matches the sum of the observed maximum MSD and LRRD unfolding lengths. Similar tri-modally distributed ramped unfolding lengths were obtained by using mAb AN51 (epitope mapped to the N-terminal flanking region above LRRD, cf. Figure 2A) instead of A1 to pull the platelet GPIbα (Figure 2F), and the second subpopulation also matches the ramped unfolding length distribution obtained using AN51 to pull GC (Figure 2G, blue bars). These results are expected because the unfolding lengths are determined by the respective primary structures of the LRRD and MSD, and as such should not depend on the 'grabbing handle' used to pull GPIbα. The consistence of the A1 and AN51 results imparts confidence in our identification of the three subpopulations as unfolding of MSD, LRRD, and both, respectively. Interestingly, the two force waveforms induced unfolding of different GPIbα domains. Clamped forces unfolded only MSD as the lengths of clamped unfolding distribute as a single peak at 20 nm (Figure 2E,G, red bars), matching the first subpopulation in Figure 2D,F, respectively, regardless of whether platelet GPIbα was engaged by A1 or AN51. Furthermore, unfolding of LRRD in GC was induced only by ramped forces but not clamped forces (Figure 2J,K). By comparison, pulling platelet GPIbα via WM23 with both ramped and clamped forces induced MSD unfolding events with similar occurrence frequencies and unfolding lengths (Figure 2J,K). These results indicate that MSD can be unfolded by increasing forces as well as constant forces. By comparison, LRRD unfolding requires increasing forces. Some force-clamp cycles (Figure 1C; Video 2) generated two consecutive unfolding events, one in the ramping and the other in the clamping phase (Figure 1E). The respective unfolding lengths of the ramped and clamped unfolding events were 34–55 nm and 13–25 nm that totaled 47–80 nm, agreeing with those of the LRRD, MSD, and MSD+LRRD subpopulations in Figure 2D,F. Together, these results provide criteria to determine whether and which GPIbα domain(s) is unfolded (Figure 3—source data 1A). Video 2 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Force-clamp experiment mode with a bond lifetime event. The video consists of two parts in series. Part I is an animation (produced by Adobe Flash; 12 fps), and part II is a video recording of a representative fluorescence BFP experiment (recorded by a customized LabView program; 25 fps). Both parts show BFP force-clamp measurement cycles. In part I, the synchronized BFP illustration (upper panel), A1–GPIbα interaction (middle panel) and 'Force vs. Time' signal (lower panel) of the same force-clamp cycle with a lifetime event are displayed in parallel. Phases of the BFP cycle are indicated in the lower panel. Part II shows two BFP cycles, which sequentially render a no bond event and a bond lifetime event. The pseudo-color epifluorescence images (acquired at 1 fps) are interpolated and superimposed onto the brightfield images to reflect the real-time intraplatelet Ca2+ level (in a progressive sequence: blue, green, yellow, orange and red). Following the long lifetime event, calcium first rapidly elevates and then quickly decays, manifesting an α-type Ca2+. https://doi.org/10.7554/eLife.15447.009 Force- and ligand-dependent cooperativity between LRRD and MSD unfolding To characterize the mechanical response of GPIbα, we measured the frequency, force and length of LRRD and MSD unfolding induced by a range of clamped forces exerted on platelet GPIbα or GC by A1WT or a type 2B VWD mutant A1R1450E. The ramped unfolding frequencies of both domains were extremely low at ≤10 pN but increased with the higher levels of clamped forces (Figure 3A,B). Interestingly, LRRD, but not MSD, unfolded more frequently when platelet GPIbα (Figure 3A) and GC (Figure 3—figure supplement 1) were pulled by A1WT than A1R1450E. The ramped unfolding forces of both domains increased with the clamped force and were indifferent to whether force was applied via WT or R1450E mutant of A1 (Figure 3C,D). In general, a higher force was required to unfold LRRD than MSD. Surprisingly, pulling platelet GPIbα via different ligands generated distinctive MSD clamped unfolding frequency vs. force plots: increasing initially and decreasing after reaching maximal at 25 pN when pulled by A1WT, but decreasing monotonically when pulled by A1R1450E (Figure 3E). These data suggest that the mechanoreceptor GPIbα may be able to interpret mechanical cues and discriminate ligands by responding to different force waveforms applied via different ligands with distinct LRRD and MSD unfolding frequencies. In addition, the distinctive force-dependences of two subpopulations of events that we deemed as respective LRRD and MSD unfolding provide further support for our criteria for their identification and classification. Figure 3 with 1 supplement see all Download asset Open asset Force- and ligand-dependent cooperative unfolding of GPIbα LRRD and MSD. (A–D) Frequency (A,B) and force (C,D) of LRRD (A,C) or MSD (B,D) unfolding events occurred in the ramping phase induced by pulling via A1WT (blue) or A1R1450E (red) with indicated preset clamped forces. (E) Occurrence frequencies of MSD clamped unfolding induced by holding at indicated clamped forces with A1WT or A1R1450E bonds. (F) The degree of cooperativity, quantified by ∆P/P = P(MSD+LRRD)/[P(MSD)×P(LRRD)] -1, is plotted vs. clamped force. P(LRRD), P(MSD) and P(LRRD+MSD) are the observed occurrence frequencies of unfolding events of LRRD alone, MSD alone and LRRD+MSD, respectively. (G,H) Significance of cooperativity assessed by (negative log10 of) p-value of the χ2 test of the null hypothesis H0: MSD unfolding and LRRD unfolding are independent. The χ2 test was not performed at 10 pN since under this force LRRD unfolding did not occur and hence no unfolding cooperativity. N.D. = not detected (A,C) or not done (F–H). https://doi.org/10.7554/eLife.15447.010 Figure 3—source data 1 Statistics and cooperativity evaluation of the GPIbα domains unfolding. (A) Decision rules for and statistical summary of GPIbα domain unfolding in force-clamp experiment mode. Criteria for deciding whether or not (+ or −) and which (LRRD, MSD, or both) GPIbα domain(s) was (were) unfolded are based on BFP profile signatures and the unfolding lengths. YES = observed, NO = not observed. NA = not applicable. (B) Related to Figure 3F–H. Evaluation of LRRD and MSD unfolding cooperativity. All probabilities were calculated from occurrence data in (A). Observed joint probabilities were compared to their predicted counterparts based on the assumption that LRRD and MSD unfolded independently. For example, in 'WT A1 vs. Platelet' under 25 pN: The probability of LRRD unfolding is P(LRRD) = 3.4% + 6.9% + 2.76% = 13.06%. The probability of MSD unfolding is P(MSD) = 7.6% + 13.8% + 6.9% + 2.76% = 31.06%. The probability of MSD ramped unfolding is P(MSD, ramp) = 7.6% + 2.76% = 10.36%. The probability of MSD clamped unfolding is P(MSD, clamp) = 13.8% + 6.9% = 20.7%. https://doi.org/10.7554/eLife.15447.011 Download elife-15447-fig3-data1-v2.docx The spatial separation of LRRD and MSD by the >30 nm long MP stalk and the distinctive dependences of their unfolding on the force waveform would seem to favor these two GPIbα domains to unfold independently. This hypothesis predicts that the probability for LRRD and MSD to unfold concurrently should be equal to the product of the respective probabilities for LRRD and MSD to unfold separately. To test this hypothesis, we estimated these probabilities from the observed unfolding occurrence frequencies. At 25 pN, the 34.5% of BFP force-clamp cycles with unfolding events consist of 7.6, 17.2, 6.9, and 2.8% of unfolding of LRRD alone, MSD alone, LRRD and MSD sequentially, and concurrently (Figure 3—source data 1A). Significantly, the frequency of observing both LRRD and MSD unfolding in the same binding cycle, calculated by pooling together both cases of two domains unfolding sequentially and concurrently, P(MSD+LRRD), is much higher than the product of their respective occurrence frequencies, P(MSD)×P(LRRD), which is the joint probability for both to unfold assuming that they were independent (Figure 3—source data 1B). These data suggest that the two GPIbα domains may unfold cooperatively, i.e., one domain unfolding may increase the likelihood for the other to unfold. To quantify the degree of such cooperativity, we defined a relative probability difference, ∆P/P = [P(MSD+LRRD) - P(MSD)×P(LRRD)]/[P(MSD)×P(LRRD)]. ∆P/P > 0 indicates positive cooperativity between LRRD and MSD unfolding. No cooperativity was observed at 10 pN because this force was insufficient to induce appreciable LRRD unfolding. Pulling with A1WT by a 25 pN clamped force generated high cooperativity, and further increase in force decreased cooperativity (Figure 3F). Remarkably, unfolding cooperativity was completely abolished at all forces when applied via the VWD mutant A1R1450E (Figure 3F). We used χ2 test to determine if the hypothesis that MSD and LRRD unfolded independently should be rejected (Materials and methods). At 25 pN, LRRD unfolding significantly enhanced MSD unfolding (p = 3.09 × 10–4). The χ2 test results are depicted as negative log p-values vs. force plots in Figure 3G,H for A1WT and A1R1450E, respectively. Interestingly, significant (p = 0.05, dashed horizontal lines) unfolding cooperativity was observed only for A1WT at 25 and 40 pN. These data show that the cooperativity between LRRD and MSD unfolding is force- and ligand-dependent. Model for cooperativity between LRRD and MSD unfolding To elucidate the mechanism underlying the force- and ligand-dependent unfolding cooperativity, we note that when the MSD unfolding events were separately analyzed according to their occurrence in the ramping or clamping phase, MSD clamped, but not ramped, unfolding was significantly (p= 8.79 × 10–3 vs. 0.076 at 25 pN) enhanced by LRRD unfolding (Figure 3G), which occurred in the ramping phase only. This dominance of cooperativity by sequential rather than concurrent unfolding suggests a model for LRRD unfolding to impact MSD unfolding, which includes three ideas. The first idea has to do with the MSD time-to-unfold, tu (cf. Figure 1F). Our force-clamp measurements revealed similar tu values induced by A1WT or A1R1450E pulling (Figure 4A). The only exception is at 10 pN where a shorter tu was induced by A1WT than A1R1450E. This can be explained by their differential bond lifetimes (Figure 4B,C). Compared to A1R1450E, the much shorter lifetime of GPIbα bond with A1WT at 10 pN may underestimate tu because early dissociation of GPIbα would prevent observation of slow MSD unfolding events. This reasoning provides the second idea for our model: MSD clamped unfolding should occur before A1–GPIbα dissociation. The third idea comes from our previous observation (Ju et al., 2015b) that LRRD unfolding significantly prolongs GPIbα bond lifetime with A1WT (Figure 4B) but not A1R1450E (Figure 4C). Combining these three ideas, our model proposes that the A1–GPIbα bond lifetime, regulated by force and prolonged by LRRD unfolding in respective ligand-specific manners, determines the occurrence of MSD clamped unfolding, which, despite its ligand-independent unfolding kinetics, generates a cooperativity pattern that maximizes at the optimal force of 25 pN for A1WT but not for A1R1450E. Figure 4 with 1 supplement see all Download asset Open asset LRRD unfolding prolongs A1–GPIbα bond lifetime and facilitates MSD clamped unfolding. (A–C) Mean ± s.e.m. of MSD time-to-unfold (tu, A) and GPIbα bond lifetimes (tb, B,C) with A1WT (blue) or A1R1450E (red) were measured in the clamping phase at different forces in the absence (−) or presence (+) of LRRD unfolding in the same BFP cycle. No LRRD unfolding occurred at 10 pN; hence no bond lifetime was measured under the LRRD+ at this force. (D) 3D plot of the surface of joint probability density (z-axis) of GPIbα to dissociate from A1WT at tb (x-axis) and MSD to unfold at tu (y-axis) (Materials and methods). Three planes, tu = 1, 3, and 5 s, under the probability density surface (gray) are shown in green or red, depending on whether they are on the left or right side of the tu = tb plane (yellow). (E,F) Measured (solid bars) and predicted (open bars) frequency of MSD unfolding events occurred in the clamping phase induced by the indicated force exerted via A1WT (E) or A1R1450E (F) in the presence (+) or absence (−) of LRRD unfolding in the same BFP cycle. N.D. = not detected. Error bar = s.e.m. estimated by the multinomial distribution of events. https://doi.org/10.7554/eLife.15447.013 Figure 4—source data 1 MSD unfolding rates (ku) and the fraction (w1) and off-rates (k1, k2) of GPIbα dissociating from A1WT or A1R1450E under different forces. w1 represents the fraction of binding events that dissociate with the off-rate k1. The fraction of events that dissociate with the off-rate of k2 is simply calculated as w2 = 1-w1. NA = not applicabl