Selectins and integrins are key players in the adhesion and signaling cascade that recruits leukocytes to inflamed tissues. Selectin binding induces β2 integrin binding to slow leukocyte rolling. Here, a micropipette was used to characterize neutrophil adhesion to E-selectin and intercellular adhesion molecule-1 (ICAM-1) at room temperature. The time-dependent adhesion frequency displayed two-stage kinetics, with an E-selectin-mediated fast increase to a low plateau followed by a slow increase to a high plateau mediated by intermediate-affinity binding of integrin αLβ2 to ICAM-1. The αLβ2 activation required more than 5 s contact to E-selectin and spleen tyrosine kinase (Syk) activity. A multi-zone channel was used to analyze αLβ2 activation by P-selectin in separate zones of receptors or antibodies, finding an inverse relationship between the rolling velocity on ICAM-1 and P-selectin dose, and a P-selectin dose-dependent change from bent to extended conformations with a closed headpiece that was faster at 37°C than at room temperature. Activation of αLβ2 exhibited different levels of cooperativity and persistent times depending on the strength and duration of selectin stimulation. These results define the precise timing and kinetics of intermediate activation of αLβ2 by E- and P-selectins.

During inflammation, leukocytes are recruited to infected or injured tissues via a multi-step adhesion and signaling cascade that involves selectins and β2 integrins (McEver and Zhu, 2010). Selectins are regulated by their cell surface expression, including de novo protein synthesis (E-selectin), redistribution from intracellular granules (P-selectin) and shedding via proteolysis (L-selectin) upon activation (McEver, 2015). In addition to changes in expression, integrins are regulated by activation (Luo et al., 2007; Sun et al., 2019), which alters their conformation, as well as their affinity and avidity for ligands. Activation by chemokines induces inside-out signaling to extend β2 integrins and open their headpieces (EO conformation), resulting in high-affinity binding to ligands such as ICAM-1, leading to leukocyte arrest (Bachmann et al., 2019; McEver, 2015; Yago et al., 2018). By comparison, binding of E- or P-selectin to leukocyte ligands, such as P-selectin glycoprotein ligand 1 (PSGL-1) and CD44, extends β2 integrins but keeps their headpieces closed (EC conformation). These EC β2 integrins bind ICAM-1 with an intermediate affinity to slow the leukocyte rolling velocity (Yago et al., 2018; Zarbock et al., 2007). This selectin binding-dependent β2 integrin activation is blocked by mitogen-activated protein kinase (MAPK) inhibitors and involves spleen tyrosine kinase (Syk), indicating that the signaling pathway is different from and independent of the chemokine activation pathway (Chesnutt et al., 2006; Green et al., 2004; McEver, 2015; Miner et al., 2008; Yago et al., 2018; Zarbock et al., 2007).

Neutrophils can be converted rapidly from selectin-mediated rolling to integrin-mediated arrest in 80 or 560 ms upon stimulation by the chemokine interleukin-8 or platelet-activating factor, respectively (Rainger et al., 1997). It has been reported that after crosslinking of L-selectin, the phosphorylation of tyrosine of MAPKs peaks within 1 min, providing the timescale required for the activation and clustering of β2 integrins (Green et al., 2004). Integrin αLβ2 has also been suggested to cluster and be activated within seconds of E-selectin binding (Green et al., 2004). However, the detailed kinetics of αLβ2 upregulation following selectin engagement has not been characterized because the flow chamber and cell aggregation experiments used in previous studies lack sufficient temporal resolution and there is, so far, no precise estimate of the timescale of the activation event and its dependence on the strength of stimulation (Chesnutt et al., 2006; McDonough et al., 2004; Zarbock et al., 2007). Although the signaling cascades stimulated by E- and P-selectins are indistinguishable with proper ligand density control (Stadtmann et al., 2013), E-selectin binds to several glycoprotein ligands, including PSGL-1, CD44 and E-selectin ligand 1 (ESL-1, also known as GLG1), whereas P-selectin signals exclusively through PSGL-1 on neutrophils (McEver, 2015; McEver and Zhu, 2010). Upregulation of the ICAM-1–αLβ2 interaction by the P-selectin–PSGL-1 interaction has been studied in conventional flow chambers and microfluidic channels by measuring the reduced cell rolling velocity on selectin in the presence or absence of ICAM-1 (Kuwano et al., 2010; McEver, 2015; Yago et al., 2018). However, these previous studies were unable to control the time of occurrence and duration of each interaction, making it difficult to dissect the interplay between the two receptor–ligand pairs and characterize the timing, kinetics and cooperativity of β2 integrin activation induced by selectin–ligand binding (Fan and Ley, 2015; McEver, 2015; Yago et al., 2018).

Here, we used a micropipette adhesion frequency assay (Chesla et al., 1998) and a multi-zone microfluidic channel (Zhou et al., 2018) to analyze the crosstalk between β2 integrin and E- or P-selectin. The first technique is highly sensitive and precisely controlled, allowing us to assay binding kinetics of single (Long et al., 2001; Zhang et al., 2005) and dual (Jiang et al., 2011) receptor–ligand species at a single-cell level. The second technique has a much higher throughput and controls cells so that they interact with two immobilized counter molecules sequentially in separate space and time (Chen et al., 2019). Our experiments characterized the timing, kinetics, conformation, cooperativity and reversibility of the β2 integrin activation induced by binding of E- or P-selectin to neutrophil ligands.

Adhesion of neutrophils to red blood cells functionalized with dimeric E-selectin and ICAM-1 exhibits two-stage kinetics

To analyze in situ receptor–ligand interactions on the surface of individual neutrophils using an adhesion frequency assay (Chesla et al., 1998; Long et al., 2001; Zhang et al., 2005) (Fig. 1A), red blood cells (RBCs) were either coated with an anti-Fc antibody to capture chimeric E-selectin–immunoglobulin (E-selectin–Ig), chimeric ICAM-1–immunoglobulin (ICAM-1–Ig) or both (Fig. 1B), or were directly coupled (via CrCl3) with E-selectin–Ig or a monomeric E-selectin and/or ICAM-1–Ig (Fig. 1C). The ligands on neutrophils are illustrated in Fig. 1D (McEver and Zhu, 2010). The adhesion frequency (Pa) versus contact time (tc) curve of neutrophil adhesion to E-selectin–Ig in 1 mM Ca2+ and 1 mM Mg2+ (Ca2+/Mg2+) rapidly increased to a plateau (Fig. 2A). Adhesion was abolished when E-selectin–Ig was not coated or when Ca2+ was chelated by EDTA, demonstrating binding specificity. RBCs coated with low ICAM-1 densities did not bind to neutrophils appreciably in Ca2+/Mg2+, which is known to keep β2 integrins inactive (Fig. 2B) (Chen et al., 2010). However, specific ICAM-1 binding became evident in medium containing 2 mM Mn2+ (Mn2+) or 2 mM Mg2+ plus 2 mM EGTA (Mg2+/EGTA), indicating that the ICAM-1 density was sufficient to support binding to activated β2 integrins (Chen et al., 2010; Jiang et al., 2011). The two ICAM-1 binding curves also displayed monotonic increases to single plateaus, but at much slower rates than that shown in Fig. 2A.

Fig. 1.

Micropipette setup and molecules. (A) Micrograph of an RBC (left) and a neutrophil (right) in a micropipette experiment. The surface molecules studied are shown below as schematics. (B) E-selectin–Ig and ICAM-1–Ig were captured by an anti-Fc antibody (Ab) precoated on the RBC surface using biotin–strepavidin coupling. (C) In some experiments, E-selectin–Ig, ICAM-1–Ig or monomeric E-selectin were directly coated onto the RBC surface via chromium chloride coupling. (D) β2 integrin and E-selectin ligands on neutrophils.

Fig. 1.

Micropipette setup and molecules. (A) Micrograph of an RBC (left) and a neutrophil (right) in a micropipette experiment. The surface molecules studied are shown below as schematics. (B) E-selectin–Ig and ICAM-1–Ig were captured by an anti-Fc antibody (Ab) precoated on the RBC surface using biotin–strepavidin coupling. (C) In some experiments, E-selectin–Ig, ICAM-1–Ig or monomeric E-selectin were directly coated onto the RBC surface via chromium chloride coupling. (D) β2 integrin and E-selectin ligands on neutrophils.

Fig. 2.

Two-stage kinetics of neutrophil adhesion to RBCs co-coated with E-selectin and ICAM-1. (A–C) Frequency of adhesion (Pa; mean±s.e.m. of five pairs of cells) of neutrophils to RBCs bearing (A) E-selectin–Ig (56 μm−2), (B) ICAM-1–Ig or (C) both E-selectin–Ig (42 μm−2) and ICAM-1–Ig (26 μm−2) was measured in the indicated cation conditions and plotted versus contact time tc. Controls in A included the addition of 5 mM EDTA to chelate divalent cations (circles) or using capture antibody-coated RBCs without incubation with E-selectin–Ig (capture Ab, diamonds). ICAM-1–Ig densities in B were ml=12 μm−2 (for Mg2+/EGTA and Ca2+/Mg2+) or 29 μm−2 (for Mn2+). The control in C was the addition of mAb ES1 (Fab) to block E-selectin binding (ES1 blocking, diamonds). Eqn 1 A,B was fitted (curves) to different Pa versus tc data (points) in A and B as well as to the first stage (0<t< 5 s) of the two-stage data (squares) in C. To fit the second stage (t>5 s), the x axis was shifted upward to align with the first plateau and the y axis was shifted rightward by 5 s. (D,E) Product of effective 2D affinity (AcKa) and ligand density (ml) for (D) integrin β2 or (E) E-selectin (left y axis), and the reverse rate (kr; right y-axis), were obtained by fitting (D) the ICAM-1 data in B in Mg2+/EGDA (open bars) or Mn2+ (closed bars) and (E) the E-selectin–Ig data in A (open bars), which was compared with those obtained by fitting the first stage of the two-stage data in C (closed bars). Data are presented as mean±s.e.m. (F) Binding of neutrophils to RBCs bearing E-selectin–Ig (26 μm−2) and ICAM-1–Ig (50 μm−2) in the absence (squares) or presence (circles) of 10 mg/ml β2 integrin-blocking mAb 7E4. Mean±s.e.m. of three pairs of cells. (G) Binding of neutrophils to RBCs bearing E-selectin–Ig (36 μm−2) and ICAM-1–Ig (9 μm−2) in the absence (squares) or presence (circles) of piceatannol or in the presence of 5 mM EDTA (diamonds). Mean±s.e.m. of four pairs of cells. (H) Adhesion frequencies between neutrophils and RBCs coated with E-selectin–Ig and ICAM-1–Ig were measured at contact times of 2 s and 15 s after 30 min incubation and in continuous presence of 10 µg/ml of either anti-β1 mAb TS2/16, anti-αLβ2 blocking mAb TS1/22, or anti-αMβ2 blocking mAb 2LPM19c. Mean±s.e.m. of six pairs of cells. P values shown were calculated using a two-tailed Student's t-test. (I) The adhesion frequency Pa versus contact time tc data from Fig. S1 were transformed to the average number of bonds and normalized by the E-selectin (Es) coating density mr. Mean±s.e.m. of six pairs of cells. (J) Comparison of plateau level (tc=10 s) frequencies of neutrophil adhesion to RBCs bearing ICAM-1–Ig (42 μm−2) in the absence (plain medium) or presence of 10 μg/ml solutions of monomeric E-selectin (Es monomer), E-selectin–Ig (Es–Ig) or human IgG (hIgG). Mean±s.e.m. of six pairs of cells. (K) Binding of neutrophils to RBCs bearing ICAM-1–Ig (65 µm−2) with 1 µg/ml E-selectin–Ig in the medium (squares). Treatment with 50 µM piceatannol (circles) reduced binding to the background level, which was determined using RBCs coated with the anti-Fc capture antibody and incubated with human IgG instead of ICAM-1–Ig (diamonds). Human IgG (10 µg/ml) was added to the medium to block binding of E-selectin–Ig in solution to the capture antibody on RBC surface, which then bound to neutrophils in the adhesion tests. Binding of neutrophils to RBCs not coated with the capture antibody was negligible (triangles). Mean±s.e.m. of four pairs of cells.

Fig. 2.

Two-stage kinetics of neutrophil adhesion to RBCs co-coated with E-selectin and ICAM-1. (A–C) Frequency of adhesion (Pa; mean±s.e.m. of five pairs of cells) of neutrophils to RBCs bearing (A) E-selectin–Ig (56 μm−2), (B) ICAM-1–Ig or (C) both E-selectin–Ig (42 μm−2) and ICAM-1–Ig (26 μm−2) was measured in the indicated cation conditions and plotted versus contact time tc. Controls in A included the addition of 5 mM EDTA to chelate divalent cations (circles) or using capture antibody-coated RBCs without incubation with E-selectin–Ig (capture Ab, diamonds). ICAM-1–Ig densities in B were ml=12 μm−2 (for Mg2+/EGTA and Ca2+/Mg2+) or 29 μm−2 (for Mn2+). The control in C was the addition of mAb ES1 (Fab) to block E-selectin binding (ES1 blocking, diamonds). Eqn 1 A,B was fitted (curves) to different Pa versus tc data (points) in A and B as well as to the first stage (0<t< 5 s) of the two-stage data (squares) in C. To fit the second stage (t>5 s), the x axis was shifted upward to align with the first plateau and the y axis was shifted rightward by 5 s. (D,E) Product of effective 2D affinity (AcKa) and ligand density (ml) for (D) integrin β2 or (E) E-selectin (left y axis), and the reverse rate (kr; right y-axis), were obtained by fitting (D) the ICAM-1 data in B in Mg2+/EGDA (open bars) or Mn2+ (closed bars) and (E) the E-selectin–Ig data in A (open bars), which was compared with those obtained by fitting the first stage of the two-stage data in C (closed bars). Data are presented as mean±s.e.m. (F) Binding of neutrophils to RBCs bearing E-selectin–Ig (26 μm−2) and ICAM-1–Ig (50 μm−2) in the absence (squares) or presence (circles) of 10 mg/ml β2 integrin-blocking mAb 7E4. Mean±s.e.m. of three pairs of cells. (G) Binding of neutrophils to RBCs bearing E-selectin–Ig (36 μm−2) and ICAM-1–Ig (9 μm−2) in the absence (squares) or presence (circles) of piceatannol or in the presence of 5 mM EDTA (diamonds). Mean±s.e.m. of four pairs of cells. (H) Adhesion frequencies between neutrophils and RBCs coated with E-selectin–Ig and ICAM-1–Ig were measured at contact times of 2 s and 15 s after 30 min incubation and in continuous presence of 10 µg/ml of either anti-β1 mAb TS2/16, anti-αLβ2 blocking mAb TS1/22, or anti-αMβ2 blocking mAb 2LPM19c. Mean±s.e.m. of six pairs of cells. P values shown were calculated using a two-tailed Student's t-test. (I) The adhesion frequency Pa versus contact time tc data from Fig. S1 were transformed to the average number of bonds and normalized by the E-selectin (Es) coating density mr. Mean±s.e.m. of six pairs of cells. (J) Comparison of plateau level (tc=10 s) frequencies of neutrophil adhesion to RBCs bearing ICAM-1–Ig (42 μm−2) in the absence (plain medium) or presence of 10 μg/ml solutions of monomeric E-selectin (Es monomer), E-selectin–Ig (Es–Ig) or human IgG (hIgG). Mean±s.e.m. of six pairs of cells. (K) Binding of neutrophils to RBCs bearing ICAM-1–Ig (65 µm−2) with 1 µg/ml E-selectin–Ig in the medium (squares). Treatment with 50 µM piceatannol (circles) reduced binding to the background level, which was determined using RBCs coated with the anti-Fc capture antibody and incubated with human IgG instead of ICAM-1–Ig (diamonds). Human IgG (10 µg/ml) was added to the medium to block binding of E-selectin–Ig in solution to the capture antibody on RBC surface, which then bound to neutrophils in the adhesion tests. Binding of neutrophils to RBCs not coated with the capture antibody was negligible (triangles). Mean±s.e.m. of four pairs of cells.

We tested the applicability of a previous model for adhesion mediated by bimolecular interactions (Chesla et al., 1998):
formula
(1A)
formula
(1B)
where 〈n〉 is the average number of bonds per contact, mr and ml (in μm−2) are the densities of receptors on RBCs and ligands on neutrophils, respectively, Ac is the contact area (constant throughout all experiments), Ka (in μm2) is the two-dimensional (2D) affinity and kr (in s−1) is the reverse rate. Eqn 1A,B was fitted to the Pa versus tc data for either the E-selectin–ligand (Fig. 2A; R2=0.81) or the ICAM-1–β2 integrin (Fig. 2B; R2=0.86 for both curves) interaction, returning the reverse-rate constant kr and the effective 2D affinity AcKa (in μm4) multiplied by the density of neutrophil ligands ml. AcKa and ml are estimated as a product because neutrophils express multiple E-selectin ligands of unknown densities (Fig. 1D). For simplicity, we use the same notation for both interactions where their designations are clear and use the subscripts ‘int’ and ‘sel’ to denote the ICAM-1–β2 integrin and E-selectin–ligand interactions, respectively, where it is necessary to avoid confusion. The fast kinetics shown in Fig. 2A corresponded to a rapid kr (5.68 s−1) of E-selectin dissociation from its neutrophil ligands. The much slower kinetics shown in Fig. 2B corresponded to an order of magnitude smaller kr values, 0.49 s−1 and 0.18 s−1, for ICAM-1 dissociation from β2 integrins in Mg2+/EGTA and Mn2+, respectively (Fig. 2D). The higher plateau observed in Mn2+ compared to that in Mg2+/EGTA (Fig. 2B) corresponds to the higher binding affinity of β2 integrins in Mn2+ than in Mg2+/EGTA (Fig. 2D) (Chen et al., 2010) and the higher ICAM-1 density in the Mn2+ (29 μm−2) experiment compared to that in the Mg2+/EGTA experiment (12 μm−2) (Fig. 2B).

Remarkably, the adhesion curve exhibited two-stage kinetics when RBCs were co-coated with E-selectin–Ig and ICAM-1–Ig to mimic activated endothelial cells. In addition to the rapid rise to a low plateau, as observed in Fig. 2A, another slow rise to a high plateau was observed (Fig. 2C). Fitting Eqn 1A,B to the first-stage data (0<t<5 s in Fig. 2C) returned parameters AcKaml and kr indistinguishable from those estimated previously (Fig. 2E), suggesting that the first stage was mediated by E-selectin and that β2 integrins remained inactive. Note that due to the fast increase of adhesion frequency, fitting yielded a minimal value of kr for both cases.

The second-stage data displayed similar kinetics to those shown in Fig. 2B, except that the curve was shifted upwards by an amount that matches the low plateau and shifted rightwards by a ∼5 s lag. We tested the hypothesis that the second-stage increased adhesion resulted from the binding of activated β2 integrins to ICAM-1. Indeed, the second stage was eliminated by an anti-β2 monoclonal antibody (mAb), 7E4, leaving only a monotonic binding curve (Fig. 2F).

Similar two-stage adhesion kinetics have been observed in the cooperation of T cell receptor (TCR) and CD8 for binding of peptide–major histocompatibility complex (pMHC), which requires the activity of the lymphocyte-specific protein tyrosine kinase (also known as Lck) (Casas et al., 2014; Jiang et al., 2011; Liu et al., 2014b). We therefore tested whether the second-stage observed here could be eliminated by the Syk inhibitor piceatannol (Fig. 2G), and the resulting data suggest that Syk activity is required for β2 integrin activation, which is consistent with previous studies (Fan and Ley, 2015; Zarbock et al., 2007). Furthermore, the anti-E-selectin mAb ES1 (Fab) totally abolished adhesion (Fig. 2C), confirming that E-selectin binding is required for β2 integrins in Ca2+/Mg2+ to interact appreciably with ICAM-1 at low densities.

To determine the relative contributions to ICAM-1 binding of two neutrophil ligands (Smith et al., 1989), integrins αLβ2 and αMβ2, we blocked the ligands using mAbs and measured the respective Pa at tc=2 s and 15 s, time points that are localized at the first and second plateaus of the two-stage curve. The adhesion frequency at 15 s was higher than at 2 s in the presence of a control anti-β1 integrin mAb (TS2/16) or an anti-αMβ2 mAb (2LPM19c) but not in the presence of an anti-αLβ2 mAb (TS1/22), identifying the contribution of αLβ2 but not αMβ2 to ICAM-1 binding at 15 s (Fig. 2H). Our observation is consistent with a previous report that αMβ2 binds ICAM-1 at body temperature but not at room temperature, whereas αLβ2 binds ICAM-1 similarly at both temperatures (Diamond et al., 1990). Thus, the upregulated ICAM-1 binding following E-selectin engagement was mediated by the activated αLβ2.

Monomeric E-selectin engagement is unable to activate αLβ2

Dimerization or clustering of selectins has been shown to increase adhesion frequency in the micropipette assay (Zhang et al., 2013) and to enhance tethering strength and stabilize cell rolling under flow (Ramachandran et al., 2001; Setiadi and McEver, 2008). Therefore, we asked whether E-selectin monomers and dimers differ in their ability to activate αLβ2. Monomeric E-selectin and dimeric E-selectin–Ig were directly coupled to RBCs alone or together with ICAM-1–Ig (Fig. 1C). RBCs coated with E-selectin–Ig and ICAM-1–Ig produced a family of two-stage binding curves (Fig. S1, circles), but only monotonic binding curves were observed for monomeric E-selectin and ICAM-1–Ig, indicating that αLβ2 integrins were inactive (Fig. S1, squares). The binding specificity was confirmed using EDTA, which abolished binding (Fig. S1, diamonds). To remove the effect of different E-selectin densities, we used Eqn 1A to transform the adhesion frequency Pa to an average number of bonds per contact, normalized this by the E-selectin site density mr and plotted the new variable (Fig. 2I). It was evident that the first plateau of the E-selectin–Ig curve matched the plateau of the monomeric E-selectin curve, but only the former curve displayed a second plateau. Thus, dimeric, but not monomeric, E-selectin was able to activate αLβ2.

To corroborate the above results, we added 10 µg/ml monomeric E-selectin or dimeric E-selectin–Ig to the Ca2+/Mg2+ medium where neutrophils were tested using RBCs coated with ICAM-1–Ig only. E-selectin–Ig yielded a steady-state adhesion frequency much higher than that observed with monomeric E-selectin, which was similar to that of the control (Fig. 2J). The high neutrophil–ICAM-1 binding induced by soluble E-selectin–Ig was not due to its engagement with neutrophil Fcγ receptors, as 10 µg/ml of irrelevant immunoglobulin G (IgG) produced a baseline adhesion frequency and a much smaller number of bonds per ICAM-1 density (Fig. 2J), confirming that the increased neutrophil adhesion to ICAM-1-coated RBCs was induced by binding of soluble E-selectin–Ig to its ligands, not by binding of the Fc portion of the chimeric protein to the Fcγ receptors.

With 1 µg/ml E-selectin–Ig in the medium, the binding of ICAM-1–Ig-coated RBCs to neutrophils exhibited a monotonic curve (Fig. 2K). This is consistent with our model, because E-selectin–Ig was added to the medium prior to the micropipette test and was present throughout the experiment. Thus, αLβ2 integrins had been activated long before being tested by the micropipette, similar to their upregulation in Mg2+/EGTA or in Mn2+ (Fig. 2B), which eliminated the 5 s lag. Consistent with the surface-bound E-selectin–Ig results (Fig. 2G), αLβ2 activation by solution E-selectin–Ig also required Syk activity, as piceatannol decreased the adhesion frequency to a level comparable to that of neutrophil binding to RBCs not coated with ICAM-1–Ig (IgG was also added to prevent capturing of E-selectin–Ig from the medium onto the RBCs) (Fig. 2K). These data further indicate that crosslinking of E-selectin ligands on neutrophils is required to activate αLβ2.

The fast speed by which adhesion was induced suggests that E-selectin–Ig induced αLβ2 activation, rather than upregulating αLβ2 expression. To test this hypothesis, we measured β2 integrin expression using flow cytometry after incubating neutrophils with E-selectin–Ig, IgG or N-formylmethionine leucyl-phenylalanine (fMLP) for 10 min. The expression of β2 integrin was increased by the positive control fMLP, but not by E-selectin–Ig or the negative control IgG (Fig. S2).

Prior P-selectin–PSGL-1 binding decreases neutrophil rolling velocity on ICAM-1

We next applied a multi-zone channel (Zhou et al., 2018) to separate the P-selectin–PSGL-1 interaction and ICAM-1–αLβ2 interaction in space and time (Fig. 3A). The upstream preprocessing zone was used to confirm that αLβ2 on the incoming neutrophils was inactive and hence unable to bind ICAM-1. The downstream reporting zone was used to measure the effect on neutrophil rolling velocity on ICAM-1 following P-selectin-PSGL-1 binding in the stimulating zone. We performed four experiments to measure neutrophil rolling over a range of wall shear stresses on surfaces of single-, dual- or triple-zone channels: (1) a single-zone experiment in which neutrophil rolling was measured on a surface coated with ICAM-1 only; (2) a dual-zone experiment in which the preprocessing zone was coated with ICAM-1 and the stimulating zone, where neutrophil rolling was measured, was coated with P-selectin; (3) a single-zone experiment in which neutrophil rolling was measured on a surface co-coated with both P-selectin and ICAM-1; and (4) a triple-zone experiment in which the preprocessing zone was coated with ICAM-1, the stimulating zone was coated with P-selectin, and the reporting zone, where neutrophil rolling was measured, was coated with ICAM-1 (Fig. 3B).

Fig. 3.

Timing and reversibility of crosstalk between selectin ligand and integrin αLβ2. (A) Schematic of a triple-zone microfluidic device consisting of a preprocessing zone to confirm that αLβ2 is inactive, a stimulating zone to stimulate cells and a reporting zone to report αLβ2 activation. Arrows indicate the flow direction. (B) Scheme of protein coatings used in the four experiments. Experiment 1 (Exp #1), ICAM-1 coating on a single-zone channel; experiment 2 (Exp #2), sequential coating of ICAM-1 and P-selectin (P-sel) on a dual-zone channel; experiment 3 (Exp #3), co-coating of ICAM-1 and P-selectin on a single-zone channel; and experiment 4 (Exp #4), sequential coating of ICAM-1, P-selectin and ICAM-1 on a triple-zone channel. Black boxes indicate zones where velocities were measured to obtain the data shown in C. (C) Plots of rolling velocity (v) versus wall shear stress for experiments 2–4. Quiescent neutrophils did not interact with the ICAM-1 surface (20 µg/ml; Exp #1, data not plotted), but rolled on surfaces coated with P-selectin alone (20 µg/ml; Exp #2, circles), co-coated with P-selectin and ICAM-1 (20 µg/ml each; Exp #3, squares), and coated with ICAM-1 (20 µg/ml) on the reporting zone (Exp #4, triangles) after rolling across the P-selectin-coated (20 µg/ml) stimulating zone. Data (points) were fitted (curves) using a quadratic equation of log (x) (see Statistics section in Materials and Methods). Data are presented as mean±s.e.m. of 15–20 cells. P values were calculated using a two-way ANOVA with a post hoc Tukey's test to compare groups. (N.S., not significant). Inset graph shows the normalized density of neutrophil with two types of behaviors, slow rolling or firmly adhered, in the various experiments. (D) Time (calculated based on data in C) required for a neutrophil to travel from the beginning to the middle of the reporting zone (triangles) or across the 800 μm distance of the P-selectin-coated stimulating zone (circles) is plotted versus wall shear stress. Mean±s.e.m. of 15–20 cells. (E) Adhesion frequencies of neutrophil binding to RBCs bearing E-selectin–Ig (36 μm−2) and ICAM-1–Ig (9 µm−2) measured for 100 repeated contacts that alternated between 2 s and 15 s of contact time using the same cell pair (open bars), compared with those measured for 50 consecutive contacts of either 2 s or 15 s using the same cell pair (closed bars). Mean±s.e.m. of five pairs of cells for the consecutive experiment and two pairs of cells for the alternating experiment. See Movie 1 for one representative experiment. (F) A zone coated with the blocking protein BSA was added between the stimulating zone (P-selectin coated) and reporting zone (ICAM-1 coating concentration 100 μg/ml or BSA), allowing stimulated neutrophils to deactivate. The adherent neutrophils per unit area is plotted against the length of this zone, and the resulting time that cells were allowed for deactivation is indicated. Control experiment with BSA coated on both the deactivation and reporting zone is shown for comparison. Mean±s.e.m. of two individual experiments.

Fig. 3.

Timing and reversibility of crosstalk between selectin ligand and integrin αLβ2. (A) Schematic of a triple-zone microfluidic device consisting of a preprocessing zone to confirm that αLβ2 is inactive, a stimulating zone to stimulate cells and a reporting zone to report αLβ2 activation. Arrows indicate the flow direction. (B) Scheme of protein coatings used in the four experiments. Experiment 1 (Exp #1), ICAM-1 coating on a single-zone channel; experiment 2 (Exp #2), sequential coating of ICAM-1 and P-selectin (P-sel) on a dual-zone channel; experiment 3 (Exp #3), co-coating of ICAM-1 and P-selectin on a single-zone channel; and experiment 4 (Exp #4), sequential coating of ICAM-1, P-selectin and ICAM-1 on a triple-zone channel. Black boxes indicate zones where velocities were measured to obtain the data shown in C. (C) Plots of rolling velocity (v) versus wall shear stress for experiments 2–4. Quiescent neutrophils did not interact with the ICAM-1 surface (20 µg/ml; Exp #1, data not plotted), but rolled on surfaces coated with P-selectin alone (20 µg/ml; Exp #2, circles), co-coated with P-selectin and ICAM-1 (20 µg/ml each; Exp #3, squares), and coated with ICAM-1 (20 µg/ml) on the reporting zone (Exp #4, triangles) after rolling across the P-selectin-coated (20 µg/ml) stimulating zone. Data (points) were fitted (curves) using a quadratic equation of log (x) (see Statistics section in Materials and Methods). Data are presented as mean±s.e.m. of 15–20 cells. P values were calculated using a two-way ANOVA with a post hoc Tukey's test to compare groups. (N.S., not significant). Inset graph shows the normalized density of neutrophil with two types of behaviors, slow rolling or firmly adhered, in the various experiments. (D) Time (calculated based on data in C) required for a neutrophil to travel from the beginning to the middle of the reporting zone (triangles) or across the 800 μm distance of the P-selectin-coated stimulating zone (circles) is plotted versus wall shear stress. Mean±s.e.m. of 15–20 cells. (E) Adhesion frequencies of neutrophil binding to RBCs bearing E-selectin–Ig (36 μm−2) and ICAM-1–Ig (9 µm−2) measured for 100 repeated contacts that alternated between 2 s and 15 s of contact time using the same cell pair (open bars), compared with those measured for 50 consecutive contacts of either 2 s or 15 s using the same cell pair (closed bars). Mean±s.e.m. of five pairs of cells for the consecutive experiment and two pairs of cells for the alternating experiment. See Movie 1 for one representative experiment. (F) A zone coated with the blocking protein BSA was added between the stimulating zone (P-selectin coated) and reporting zone (ICAM-1 coating concentration 100 μg/ml or BSA), allowing stimulated neutrophils to deactivate. The adherent neutrophils per unit area is plotted against the length of this zone, and the resulting time that cells were allowed for deactivation is indicated. Control experiment with BSA coated on both the deactivation and reporting zone is shown for comparison. Mean±s.e.m. of two individual experiments.

Experiment 1 was a negative control with no rolling observed, confirming that the neutrophils were quiescent and their αLβ2 was inactive. Experiment 2 was a positive control and, apart from an additional preprocessing zone, it was identical to the flow chamber previously used to study neutrophil rolling on P-selectin (Miner et al., 2008; Yago et al., 2002, 2018). The neutrophil rolling velocity first decreased, reached a minimum of ∼3.5 μm/s at ∼0.2–0.3 dyn/cm2 (0.02–0.03 Pa) and increased with further increase in wall shear stress (Fig. 3C, black circles), consistent with previous reports that the rolling behavior is governed by a P-selectin–PSGL-1 catch-slip bond (Beste and Hammer, 2008; Marshall et al., 2003; Yago et al., 2004). Experiment 3 was another positive control, identical to the flow chamber previously used to study slow rolling of neutrophils on both P-selectin and ICAM-1 (Miner et al., 2008; Yago et al., 2018). The rolling velocity curve was shifted downward (Fig. 3C, red squares), indicating that an induced interaction between activated αLβ2 and ICAM-1 slowed the neutrophil rolling velocity, in agreement with the findings of previous studies (Kuwano et al., 2010; Miner et al., 2008). Experiment 4 was designed to analyze the crosstalk between PSGL-1 signaling induced by P-selectin binding and inside-out signaling of αLβ2 as reported by ICAM-1 binding. Remarkably, the rolling velocity of neutrophils on ICAM-1 followed a shear-dependent curve that was indistinguishable from that observed in experiment 3 at wall shear stresses less than 1 dyn/cm2 (0.1 Pa) and slightly faster at higher wall shear stresses (Fig. 3C, cyan triangles). This result suggests that the earlier P-selectin–PSGL-1 interaction in the stimulating zone activated αLβ2 to allow subsequent binding to ICAM-1 in the reporting zone. Since neutrophil rolling was mediated by ICAM-1–αLβ2 interactions only, the U-shaped velocity versus wall shear stress curve is consistent with catch-slip bond behavior between αLβ2 and ICAM-1 (Beste and Hammer, 2008; Yago et al., 2004), which is in agreement with our previous publications (Chen et al., 2010; Rosetti et al., 2015).

Duration and reversibility of selectin–ligand binding-induced αLβ2 activation

The results described above indicate that αLβ2 remains active with the ability to bind ICAM-1 after P-selectin–PSGL-1 dissociation. To estimate how long αLβ2 could stay active, we plotted the time taken for a neutrophil to travel from the end of the stimulating zone to where cell rolling was observed (∼50 μm downstream of the demarcation between the P-selectin and ICAM-1 zones) versus wall shear stress (Fig. 3D), which was calculated by dividing 50 μm by the rolling velocity. The results indicate that αLβ2 remains activated for as long as 20 s after leaving the stimulating zone.

We next examined whether activated αLβ2 could reverse to the inactive state in the micropipette experiment, and if so, how rapidly. We programmed the micropipette movement to alternate the durations of RBC–neutrophil contact between 2 s and 15 s in 100 repetitive cycles with a 0.5 s gap between the current separation and the next encounter of the RBC and neutrophil (Jiang et al., 2011) (Movie 1). The contacts were sorted into two groups according to their durations (2 s or 15 s). These two contact durations reside in the first and second plateaus of the two-stage binding curve (Fig. 2C). Adhesion events in each group were enumerated to calculate an adhesion frequency for that contact duration. The two Pa values so measured were significantly different (P=0.004) (Fig. 3E). However, they were indistinguishable from the respective adhesion frequencies obtained from 50 consecutive repeated contacts of 2 s and 15 s for both groups (Fig. 3E). This result indicates that αLβ2 was activated in a 15 s contact, reversed back to the inactive state within the 0.5 s cycling time, remained inactive in the next 2 s contact because the duration was too short, but were activated again when the contact duration was switched back to 15 s.

To explain the apparent discrepancy between the results shown in Fig. 3D and E, we note that the time for a neutrophil to travel across the stimulating zone of 800 μm, calculated from the rolling velocity on P-selectin (Fig. 3C), ranges from 50 s to 200 s depending on the wall shear stress (Fig. 3D). Using a simple rolling model of the cell alternating between one and two bonds to maintain continuous contact with the substrate (McEver and Zhu, 2010; Yago et al., 2004), which greatly underestimates the number of bonds involved in supporting persistent rolling, the cumulative duration of P-selectin–PSGL-1 engagement would be 1.5 times greater than the neutrophil transit time across the stimulating zone. By comparison, a ∼0.5 value of the adhesion frequency Pa at tc=15 s (Fig. 3E) translates to an average number of bonds per contact of 〈n〉=0.67 (Eqn 1A). Multiplying this value by the 15 s contact duration, the cumulative duration of P-selectin–PSGL-1 engagement in the micropipette assay is ∼10 s, which is 1/7.5–1/30 of the values estimated for the multi-zone channel experiment. These calculations suggest that the time required for the activated αLβ2 to return to the inactive state depends on the time of the prior selectin–ligand engagement.

To further validate this hypothesis, we included an additional zone in the microfluidic channel between the stimulating and reporting zone, which was coated with the blocking protein bovine serum albumin (BSA) to allow the stimulated neutrophils to rest (Fig. 3F). The activation status of αLβ2 on neutrophils was measured at 37°C as the number of adherent neutrophils per unit area in the reporting zone coated with either a high concentration of ICAM-1 (100 μg/ml) or BSA as control. The adherent cell number remained the same from 0.5 s to 3 s, indicating a persistent activation of neutrophils in the microfluidic channel. The apparent discrepancy between the results in Fig. 3E and F may be explained by the different levels of stimulation in the two different experiments; stimulation in the micropipette experiment was much weaker (by alternative contacts, where each contact might form only a very low number of bonds) than that in the flow chamber experiment (by continuous rolling, which alternates at a minimum between one and two bonds).

Cooperativity and dose response of selectin–ligand binding-induced αLβ2 activation

We next asked two questions: (1) How many selectin–ligand bonds are needed to signal for αLβ2 activation? (2) What is the level of cooperativity between the extent of selectin–ligand engagement and αLβ2 activation? It is reasonable to expect that, as the available selectin increases, the level of αLβ2 activation would increase initially but approach an eventual plateau. For a low level of cooperativity, such a ‘dose-response curve’ would rise gradually to a plateau. For a high level of cooperativity, this curve would rise sharply and produce a threshold of E-selectin density below which αLβ2­ activation is minimal but above which the level of upregulation rapidly achieves maximum.

To answer these questions, we examined how the upregulated ICAM-1–αLβ2 binding affinity depended on the E-selectin site density. It seems reasonable to assume that activated αLβ2 binds to ICAM-1 independently of E-selectin–ligand binding, since there is no evidence of cross-reactivity. Therefore, we could apply a previously published model of concurrent but independent binding of dual receptor–ligand species (Williams et al., 2000a,b; Zhu and Williams, 2000). The model predicts that in the micropipette experiment, the average total number of bonds equals the sum of the average numbers of bonds of the two species. At steady state, the average number of bonds for either species is 〈n〉=AcKamrml, which equals −ln(1−Pa) (Eqn 1A) if adhesion is mediated by a single species (Chesla et al., 1998). When both species contribute, −ln(1−Pa) equals the sum of the average number of bonds of each interaction species, 〈n〉=〈nsel+〈nint (Williams et al., 2000a,b; Zhu and Williams, 2000).

The above model was a good fit for the first stage of the two-stage data, when 〈nint represents bonds of inactive integrins (Fig. 2C), suggesting that the concurrent E-selectin–ligand binding and ICAM-1–αLβ2 binding were independent. This was clearly not the case for the second-stage data, indicating that the dual receptor became cooperative. The presence of cooperativity can be revealed by comparing the average number of bonds formed by both receptors, 〈nsel+int, to the sum of the average number of bonds formed by each of the two receptors individually, 〈nsel+〈nint. The level of cooperativity can be quantified by the excess of the whole over the sum of the parts, that is, (〈nsel+int−〈nsel−〈nint)/(〈nsel+〈nint) (Fiore et al., 2014; Li et al., 2021). Since the contribution from inactive integrin binding is negligible, this cooperativity index can be expressed using 〈nsel calculated from the Pa in the first plateau (tc=3 s) and 〈nsel+int calculated from the Pa in the second plateau (tc=15 s) of the two-stage curve, as follows:
formula
(2A)
Since the cooperativity results from inside-out signaling induced by E-selectin–ligand engagement that activates αLβ2, the numerator of Eqn 2A is the average number of ICAM-1 bonds with activated αLβ2. Dividing it by the ICAM-1 density (mr)ICAM would yield the effective 2D affinity of ICAM-1 binding to activated αLβ2 multiplied by the density of the activated integrin αLβ2, (AcKaml)int:
formula
(2B)

Note that the effective 2D affinity AcKa is a parameter defined by the conformation state of the integrin. Therefore, an increase in the (AcKaml)int value reflects an increase in the density of intermediate activated αLβ2 on the neutrophil surface, because no change was observed in the total β2 integrin expression (Fig. S2).

We measured the Pa(3) and Pa(15) values using RBCs coated with different densities of E-selectin–Ig and ICAM-1–Ig. At each combination of E-selectin–Ig and ICAM-1–Ig densities we calculated the cooperativity index and (AcKaml)int values using Eqn 2A and B, respectively, and plotted them against E-selectin-Ig site density (mr)sel (Fig. 4A,B). The cooperativity index was positive across all (mr)sel values, demonstrating a strong and biphasic E-selectin density-dependent cooperativity (Fig. 4A). By comparison, the (AcKaml)int value rose initially as E-selectin–Ig density increased up to ∼18 µm−2 and approached a saturation level of ∼0.016 µm2 (Fig. 4B). The plateau value of 0.016 µm2 predicts on average 0.32 ICAM-1–αLβ2 bonds at steady state for 20 ICAM-1–Ig/µm2. Induction of this saturation level of αLβ2 activation required on average less than 0.3 E-selectin–ligand bonds during the contact. This was calculated by multiplying the threshold E-selectin density, (mr)sel=18 µm−2, by (AcKaml)sel=0.016 µm2 from Fig. 2F.

Fig. 4.

The selectin dose-dependency of αLβ2 activation. (A,B) Adhesion frequencies of neutrophils to RBCs bearing various densities of ICAM-1–Ig and E-selectin–Ig were measured at tc=3 and 15 s and used to calculate (A) the cooperativity index and (B) (AcKaml)int using Eqn 2A and B, respectively. The values obtained are plotted against E-selectin–Ig site density (msel). A second-order polynomial (y=P1x2+P2x+P3) was fitted to the data in A (line). Eqn 3 was fitted to the data in B (line), which returned the half-saturation E-selectin–Ig density msel,½ (vertical dashed line) and the Hill coefficient nH. Outlier points (red) were excluded. Each data point represents 1–2 pairs of cells. (C) Steady-state (tc=10 s) adhesion frequency of neutrophils to RBCs bearing ICAM-1–Ig (49 μm−2) was measured in Ca2+/Mg2+ with various concentrations of E-selectin–Ig added in the medium. The (AcKaml)int values were calculated using Eqn 4 and plotted versus the concentration of E-selectin–Ig in the medium. Eqn 3 was fitted to the data, which returned the half saturation E-selectin–Ig concentration C½ (vertical dashed line) and the Hill coefficient nH. One outlier point was excluded (red). Mean±s.e.m. of four pairs of cells. (D) The activation of αLβ2 on neutrophils in multi-zone channels was measured as the number of adherent neutrophils per unit area of reporting zone coated with 100 μg/ml ICAM-1, after rolling across a stimulating zone of variable lengths coated with 20 µg/ml P-selectin. Distances on the x axis indicate the length of the stimulating zone. Positive controls (ctrl) include adhesion in a single-zone channel co-coated with P-selectin and ICAM-1 (mixed), as well as adhesion to ICAM-1 of neutrophils preincubated with Mn2+ for 10 min. Mean±s.e.m. of 3–5 individual experiments. *P<0.05; **P<0.01; N.S., not significant (Student's t-test).

Fig. 4.

The selectin dose-dependency of αLβ2 activation. (A,B) Adhesion frequencies of neutrophils to RBCs bearing various densities of ICAM-1–Ig and E-selectin–Ig were measured at tc=3 and 15 s and used to calculate (A) the cooperativity index and (B) (AcKaml)int using Eqn 2A and B, respectively. The values obtained are plotted against E-selectin–Ig site density (msel). A second-order polynomial (y=P1x2+P2x+P3) was fitted to the data in A (line). Eqn 3 was fitted to the data in B (line), which returned the half-saturation E-selectin–Ig density msel,½ (vertical dashed line) and the Hill coefficient nH. Outlier points (red) were excluded. Each data point represents 1–2 pairs of cells. (C) Steady-state (tc=10 s) adhesion frequency of neutrophils to RBCs bearing ICAM-1–Ig (49 μm−2) was measured in Ca2+/Mg2+ with various concentrations of E-selectin–Ig added in the medium. The (AcKaml)int values were calculated using Eqn 4 and plotted versus the concentration of E-selectin–Ig in the medium. Eqn 3 was fitted to the data, which returned the half saturation E-selectin–Ig concentration C½ (vertical dashed line) and the Hill coefficient nH. One outlier point was excluded (red). Mean±s.e.m. of four pairs of cells. (D) The activation of αLβ2 on neutrophils in multi-zone channels was measured as the number of adherent neutrophils per unit area of reporting zone coated with 100 μg/ml ICAM-1, after rolling across a stimulating zone of variable lengths coated with 20 µg/ml P-selectin. Distances on the x axis indicate the length of the stimulating zone. Positive controls (ctrl) include adhesion in a single-zone channel co-coated with P-selectin and ICAM-1 (mixed), as well as adhesion to ICAM-1 of neutrophils preincubated with Mn2+ for 10 min. Mean±s.e.m. of 3–5 individual experiments. *P<0.05; **P<0.01; N.S., not significant (Student's t-test).

To further estimate the level of cooperativity, we fitted the (AcKaml)int versus (mr)sel data in Fig. 4B by the Hill equation:
formula
(3)
where y denotes the saturated level, x½ denotes the half saturation E-selectin density and nH is the Hill coefficient. The best-fit values were (AcKaml)int,∞=0.015 μm2, (mr)sel,½=9.9 μm−2 and nH=2.2.
To obtain a separate estimate of the cooperativity level, we used RBCs coated with ICAM-1–Ig alone with soluble E-selectin–Ig present in the medium. The adhesion frequency was measured at a sufficiently long contact duration (10 s) to estimate the effective 2D binding affinity per ICAM-1 density only, which could be done using the steady-state (tc→∞) form of Eqn 1A,B:
formula
(4)

The (AcKaml)int value increased with E-selectin–Ig concentration (C) and reached saturation at 1 µg/ml with a plateau (AcKaml)int value similar to that obtained using RBC-coated E-selectin–Ig (Fig. 4C). Fitting Eqn 3 to the data yielded the best-fit values of (AcKaml)int,∞=0.018 μm2, C½=1.0 μg/ml and nH=0.97.

We next used the triple-zone channel to recapitulate the stimulation dose-dependency of αLβ2 activation induced by P-selectin–PSGL-1 binding (Fig. 3A). We kept the wall shear stress constant while varying the length of the P-selectin-coated zone to control the ‘stimulation dose’, as this length is proportional to the number of P-selectin molecules a neutrophil might encounter in its transit across the stimulation zone. Neutrophils were first perfused through the preprocessing zone to confirm the inactivity prior to stimulation, then rolled on a P-selectin-coated stimulating zone of variable length and finally entered the reporting zone for activated αLβ2 to bind ICAM-1, which arrested neutrophils due to the high ICAM-1 coating (100 μg/ml). The number of neutrophils captured per unit area in the reporting zone was plotted versus the length of the stimulating zone (S), which increased slowly when S was less than 800 μm but increased significantly when S was greater than 2000 μm (Fig. 4D). Two positive controls were used: (1) co-coating of P-selectin and ICAM-1 on a single-zone channel, similar to experiment 3 shown in Fig. 3 (but with a much higher ICAM-1 coating concentration to produce firm adhesion); and (2) pretreating neutrophils with Mn2+ for 10 min and perfusing them to a single-zone channel, similar to experiment 1 shown in Fig. 3. High numbers of adherent neutrophils were observed in both positive controls (Fig. 4D).

Neutrophil rolling velocity on ICAM-1 is inversely related to P-selectin dose

Next, we measured the rolling velocity of neutrophils, as in experiment 4 shown in Fig. 3, and subjected the cells to various stimulating zones of different lengths coated with P-selectin at different concentrations (Fig. 5). Since the neutrophil rolling velocity on P-selectin was constant at a given wall shear stress, the number of transient bonds (mb) between the neutrophil PSGL-1 (ml) and the P-selectin (msel) coated on the channel floor remained constant. We therefore modeled this interaction using a steady-state kinetic equation:
formula
where ml,T denotes the total PSGL-1 site density and ml,Tmb is the density of unbound PSGL-1 in the contact area between a neutrophil and the channel floor available for binding. Dividing by msel, mb, ml,T and the forward rate kf, and rearranging the terms, the above equation becomes:
formula
(5)
where Ka=kf/kr is the binding affinity.
Fig. 5.

Inverse relationship between activated αLβ2-mediated neutrophil rolling velocity and dose of P-selectin stimulation. (A–D) Neutrophils were perfused into multi-zone devices with a wall shear rate of 50 s−1. The velocity of neutrophils rolling on ICAM-1 (20 µg/ml) after rolling across the stimulating zone is plotted against the P-selectin coating concentration (A) and the zone length (C) for the indicated P-selectin length (A) and concentration (C) conditions. A linear equation between y and 1/x (i.e. of the form of Eqn 5) was fitted to each data curve. The parameters from fitting the data in A were plotted versus the P-selectin zone length S in B, and those from fitting the data in C were plotted versus the P-selectin coating concentration in D. (E,F) The neutrophil rolling velocity (E) and its reciprocal (F) were plotted against the combined selectin dose (product of the length of stimulating zone and the coating concentration, S*C) for four (E) or three (F) different coating concentrations. Eqn 3 was fitted to the data from three different coating conditions pooled together in F, which returned the half-saturation P-selectin dose (SC)½ (vertical dashed line) and the Hill coefficient, nH. Data in A–E are presented as mean±s.e.m. of 15–20 cells.

Fig. 5.

Inverse relationship between activated αLβ2-mediated neutrophil rolling velocity and dose of P-selectin stimulation. (A–D) Neutrophils were perfused into multi-zone devices with a wall shear rate of 50 s−1. The velocity of neutrophils rolling on ICAM-1 (20 µg/ml) after rolling across the stimulating zone is plotted against the P-selectin coating concentration (A) and the zone length (C) for the indicated P-selectin length (A) and concentration (C) conditions. A linear equation between y and 1/x (i.e. of the form of Eqn 5) was fitted to each data curve. The parameters from fitting the data in A were plotted versus the P-selectin zone length S in B, and those from fitting the data in C were plotted versus the P-selectin coating concentration in D. (E,F) The neutrophil rolling velocity (E) and its reciprocal (F) were plotted against the combined selectin dose (product of the length of stimulating zone and the coating concentration, S*C) for four (E) or three (F) different coating concentrations. Eqn 3 was fitted to the data from three different coating conditions pooled together in F, which returned the half-saturation P-selectin dose (SC)½ (vertical dashed line) and the Hill coefficient, nH. Data in A–E are presented as mean±s.e.m. of 15–20 cells.

It seems reasonable to assume that the higher the number of steady-state P-selectin–PSGL-1 bonds, the more intensive stimulation the neutrophil receives, the greater the number of activated αLβ2 molecules (or the greater level of their activation) on the neutrophil surface, and the slower the neutrophil rolling velocity on ICAM-1 (Lefort and Ley, 2012; Yago et al., 2018). Therefore, we assume the neutrophil rolling velocity v on ICAM-1 to be proportional to 1/mb. Thus, v should be a linear function of the reciprocal ligand density, v=a/msel+b, which has the same form as Eqn 5.

Since msel increases with the P-selectin coating concentration C until saturation, we expect the relationship between v and C to have the same form as Eqn 5, which was indeed observed for all four lengths of the stimulating zone (Fig. 5A). As the P-selectin coating concentration increased, the neutrophil rolling velocity on ICAM-1 decreased initially and approached a constant. The different rates of decrease and constant levels are captured by the fitting parameters a and b, respectively. These parameters were evaluated by fitting the data at each condition in Fig. 5A and were plotted versus the stimulation zone length S (Fig. 5B). Interestingly, both relationships follow the same form as Eqn 5 [note the use of different subscripts to indicate different parameters a and b for different conditions, e.g. aA(C) means that parameter a in A (aA) is a function of C].

The decreases of aA and bA with increasing S make sense because the longer the zone length, the longer the contact time, the more cumulative P-selectin–PSGL-1 bonds over the stimulating zone, the greater the stimulation received by the neutrophil, the higher the fraction of activated αLβ2 (or the higher level of their activation) on the neutrophil surface, and the slower the neutrophil rolling velocity on ICAM-1 (Lefort and Ley, 2012; Yago et al., 2018). Plotting the rolling velocity against the stimulating zone length revealed that an inverse relationship also describes the dependence of v on S (Fig. 5C). As we observed in Fig. 5A, coating density was saturated at around 50 μg/ml, which explained the collapse of the curves for the 20 μg/ml and 50 μg/ml conditions. Furthermore, the dependence on C of the parameters aC and bC evaluated from fitting the four sets of v versus S data also seemed to follow the form of Eqn 5 (Fig. 5D).

The fact that the dependences of v on C and S have the same form suggests that v may depend on the product of C and S. This makes sense, because this product corresponds to the total amount of ligands a traveling cell can bind to. To test this hypothesis, we plotted the neutrophil rolling velocity on ICAM-1 against the P-selectin ‘stimulation dose’ SC (Fig. 5E). Curves of 5–20 µg/ml collapsed into one single curve, which was distinct from the 50 µg/ml curve, indicating a departure from the linear relationship between v and 1/SC. We note that such a linear relationship relies on the proportional relationships between the P-selectin site density and coating concentration, between P-selectin stimulation and αLβ2 activation, between αLβ2 activation and ICAM-1 binding, and between ICAM-1–αLβ2 binding and the reciprocal neutrophil rolling velocity. Any of these proportionality relationships can become saturated at high values, giving rise to nonlinearity in the end result.

The reciprocal of Eqn 5 has the form of the Hill equation (Eqn 3). We therefore fitted the Hill equation to the reciprocal rolling velocity 1/v versus SC data (except for the 50 µg/ml curve) (Fig. 5F). The best-fit values were v=0.86 μm/s, (SC)½=36,090 μm·μg/ml and nH=1.2.

PSGL-1 signaling activates αLβ2 to the extended-closed conformation

We next coated the reporting zone of the multi-zone channel with specific reporter mAbs for different integrin conformations (Fig. 6A) and measured the number of firmly adherent neutrophils per area as a function of the length of P-selectin stimulation zone (Fig. 6C). The number of Mn2+-pretreated neutrophils captured on these reporter mAbs was measured as the fully activated condition of β2 (Fig. 6B). The number of neutrophils captured on the Kim127 mAb (reporting extended conformation of β2 integrins; Fig. 6A) (Chen et al., 2010; Yago et al., 2018) initially increased at 800 μm and approached a plateau at 3000 μm, consistent with the saturation dose in Fig. 5E. Conversely, the curve of neutrophil binding on the HI-111 mAb (reporting the closed conformation of the αL I-domain) (Chen et al., 2010; Ma et al., 2002) showed an opposite trend. By comparison, the number of neutrophils captured was uniformly low on the MEM148 mAb (reporting the hybrid domain swing-out of β2 integrins) (Yago et al., 2018) but was uniformly high on the MEM25 mAb (reporting all conformations of αLβ2) (Li et al., 2013) (Fig. 6C), showing no sign of dose dependency (Fig. 6E). These results obtained at room temperature were confirmed by repeated experiments at 37°C (Fig. 6D). The trend of cell adherence to the four conformation-specific antibodies was preserved, but with a faster kinetics at 37°C (Fig. 6D), as expected. The distances required for neutrophils to roll on P-selectin to achieve half-plateau level at 37°C were significantly longer than at room temperature (Fig. 6E). These data suggest that integrin αLβ2 had an extended conformation with no further swing-out upon P-selectin activation, confirming an EC conformation and intermediate-affinity state of αLβ2 (Hogg et al., 2011; Lefort and Ley, 2012).

Fig. 6.

Characterization of the αLβ conformation induced by selectin binding. (A) Schematic of the bent-closed (BC), extended-closed (EC) and extended-open (EO) conformations of β2 integrin and epitopes of conformation reporter mAbs (reporter mAb binding is indicated by stars). MEM25 recognizes all conformations of αL. HI-111 recognizes the inactive conformation of the I-domain. Kim127 recognizes the extended β2 conformation. MEM148 recognizes the hybrid domain swing-out conformation (open headpiece). LFA-1, integrin αLβ2. (B) After being pretreated with Mn2+ for 5 min, neutrophils at 2×106 cells/ml were perfused through microfluidic devices for 2 min, and adherent cells per unit area of surfaces coated with the indicated antibodies (100 µg/ml) were measured. Each point represents one device. Three devices were used per condition per experiment. P value was calculated using a two-tailed unpaired Student's t-test (N.S., not significant). (C,D) Neutrophils (2×106 cells/ml) were perfused through the multi-zone devices for 2 min, rolled across the stimulating zone coated with P-selectin (20 µg/ml) and entered the reporting zone coated with indicated antibodies (100 µg/ml). The number of adherent neutrophils per unit area on the reporting zone is plotted versus the length of the stimulating zone. Each point represents one individual experiment. Experiments were performed at 25°C (C) and 37°C (D). Data with increasing or decreasing trends were fitted (solid curves) using y=A1+B1S/(K1+S) (Kim127: R2=0.95 for C, R2=0.93 for D) and y=A2B2S/(K2+S) (HI-111: R2=0.77 for C, R2=0.93 for D), respectively, to estimate S½, the distance of P-selectin dose S required to achieve half-plateau level. Dashed lines indicate the mean. ***P<0.001; ****P<0.0001; N.S., not significant (one-way ANOVA with a post hoc Tukey's test to compare groups). (E) S½, calculated from fitting curves in C and D, is shown for Kim127 and HI-111 antibodies at the two temperatures. Data are mean±95% c.i. of predictions by fitted curves in C and D. P value by two-tailed paired Student’s t-test.

Fig. 6.

Characterization of the αLβ conformation induced by selectin binding. (A) Schematic of the bent-closed (BC), extended-closed (EC) and extended-open (EO) conformations of β2 integrin and epitopes of conformation reporter mAbs (reporter mAb binding is indicated by stars). MEM25 recognizes all conformations of αL. HI-111 recognizes the inactive conformation of the I-domain. Kim127 recognizes the extended β2 conformation. MEM148 recognizes the hybrid domain swing-out conformation (open headpiece). LFA-1, integrin αLβ2. (B) After being pretreated with Mn2+ for 5 min, neutrophils at 2×106 cells/ml were perfused through microfluidic devices for 2 min, and adherent cells per unit area of surfaces coated with the indicated antibodies (100 µg/ml) were measured. Each point represents one device. Three devices were used per condition per experiment. P value was calculated using a two-tailed unpaired Student's t-test (N.S., not significant). (C,D) Neutrophils (2×106 cells/ml) were perfused through the multi-zone devices for 2 min, rolled across the stimulating zone coated with P-selectin (20 µg/ml) and entered the reporting zone coated with indicated antibodies (100 µg/ml). The number of adherent neutrophils per unit area on the reporting zone is plotted versus the length of the stimulating zone. Each point represents one individual experiment. Experiments were performed at 25°C (C) and 37°C (D). Data with increasing or decreasing trends were fitted (solid curves) using y=A1+B1S/(K1+S) (Kim127: R2=0.95 for C, R2=0.93 for D) and y=A2B2S/(K2+S) (HI-111: R2=0.77 for C, R2=0.93 for D), respectively, to estimate S½, the distance of P-selectin dose S required to achieve half-plateau level. Dashed lines indicate the mean. ***P<0.001; ****P<0.0001; N.S., not significant (one-way ANOVA with a post hoc Tukey's test to compare groups). (E) S½, calculated from fitting curves in C and D, is shown for Kim127 and HI-111 antibodies at the two temperatures. Data are mean±95% c.i. of predictions by fitted curves in C and D. P value by two-tailed paired Student’s t-test.

In this study, we used a micropipette adhesion frequency assay at room temperature and multi-zone microfluidic devices to perform detailed analyses of the activation of neutrophil integrin αLβ2 following selectin–ligand engagement. This topic has previously been studied using flow chamber and cell aggregation assays (Chesnutt et al., 2006; Green et al., 2004; McEver and Zhu, 2007; Miner et al., 2008; Woolf et al., 2007; Zarbock et al., 2007). Consistent with previous work, we found that activated αLβ2 was the major β2 integrin that mediates the upregulated binding to ICAM-1 and that Syk activity was required for αLβ2 activation (Chesnutt et al., 2006; Zarbock et al., 2007). We showed that ligand binding of dimeric, but not monomeric, E-selectin activated αLβ2, suggesting a requirement for dimeric E-selectin–ligand interaction (Zhang et al., 2013) for αLβ2 activation. Using our micropipette assay, which possesses sufficient temporal resolution to determine how fast selectin binding-induced signaling activates αLβ2, we found two-stage binding kinetics between neutrophils and RBCs co-presenting ICAM-1 and E-selectin. This is similar to the two-stage binding kinetics between T cells and RBCs co-presenting ICAM-1 and cognate pMHC (Ju et al., 2017), which is consistent with the similarity between the signaling pathways induced by the TCR and by PSGL-1 for αLβ2 activation (Fan and Ley, 2015). The time needed to activate αLβ2 after selectin–PSGL-1 contact was found to be ∼5 s, longer than the ∼2 s required for TCR–pMHC contact (Ju et al., 2017). Interestingly, the αLβ2 activated by selectin signaling has EC conformation and an intermediate binding affinity, AcKa ∼10−4 μm4, calculated by dividing the (AcKaml)int,∞ value from Fig. 4B by the β2 integrin site density ml (∼75 μm−2; Fig. S2), which is threefold lower than that of αLβ2 activated by Mn2+ (calculated using the data in Fig. 2B), and is between the high (∼10−3 μm4) and low (∼10−7 μm4) 2D affinity values of the ICAM-1–αLβ2 interactions we previously reported (Zhang et al., 2005). This intermediate binding affinity is similar to that activated by Mg2+/EGTA (calculated using the data in Fig. 2B), which is consistent with the previous finding that Mg2+/EGTA is not a highly activated condition (Chen et al., 2012). The findings of EC conformation and intermediate affinity are consistent with previous reports that the intermediate activation of αLβ2 induced by selectin is different from the full activation of αLβ2 triggered by chemokines, where αLβ2 has an EO conformation and high affinity (Constantin et al., 2000; Yago et al., 2018).

Using the multi-zone microfluidic assay as a complementary approach, we found that selectin binding induced global αLβ2 activation and that such activation could be sustained for ∼20 s following sufficiently strong and long stimulation; although for weak and short stimulations used in the micropipette assay, the αLβ2 activation could be reversed in as little as ∼0.5 s. The findings that αLβ2 can remain activated after moving out of the P-selectin zone and that activation can be reversed for weak and short stimulations are similar to our recent results describing integrin αIIbβ3 activation induced by platelet glycoprotein Ib binding (Chen et al., 2019), but are different from a previous suggestion that αLβ2 activation requires the concurrent stimulation of selectin and ICAM (Zarbock et al., 2007).

We analyzed the cooperativity between selectin and integrin using two approaches. The first approach defined cooperativity as the excess of the whole over the sum of the parts. The total number of bonds of E-selectin and ICAM-1 together (i.e. the whole) was much more than the sum of the bonds of E-selectin and ICAM-1 separately (i.e. sum of the parts), indicating synergy, which results from inside-out signaling induced by E-selectin–ligand engagement that activates αLβ2.

In the second approach, we analyzed the dose response of activated αLβ2 on a neutrophil. This was done in four ways: coating the RBCs with variable E-selectin–Ig densities, incubating neutrophils with variable E-selectin–Ig concentrations, coating the stimulating zone with variable P-selectin concentrations and adjusting the P-selectin zone length. The first two used the micropipette adhesion frequency assay to measure the fraction of activated αLβ2 integrins directly. The last two used the multi-zone channels and measured neutrophil rolling velocity as a proxy for integrin activation. We demonstrated that the product of coating concentration and zone length is a single stimulation-dose variable for αLβ2 activation. We used the Hill equation to characterize these dose response curves. For short-term stimulation (≤15 s), a Hill coefficient of ∼2 was found, suggesting significant cooperativity. However, for long-term stimulation (minutes), the Hill coefficient was ∼1, indicating no cooperativity.

Our results also provide new insights into the signaling model of selectin-induced integrin activation­. In the microfluidic assay, a longer stimulation zone resulted in a higher level of integrin activation, indicating that it is the cumulative bonds from different copies of PSGL-1 on different regions of a neutrophil surface that altogether contribute to the global activation of αLβ2. This suggests a model similar to that seen in T cell Ca2+ signaling triggered by accumulation of serial TCR bonds (Liu et al., 2014a; Pryshchep et al., 2014): integrin αLβ2 can be activated by the accumulation of a sufficient number of consecutive selectin–ligand bonds or of a sufficient duration of cumulative bond lifetimes. Our data and model thus further our understanding of integrin activation.

Cells

Human neutrophils and RBCs were isolated from whole blood using protocols approved by the Institutional Review Board of The Georgia Institute of Technology. Informed consent was obtained from all donors. For micropipette experiments, neutrophils were isolated from a drop of whole blood obtained via a finger prick. After RBCs were lysed by a brief hypotonic shock, cells were spun down and resuspended in Hank's Balanced Salt Solution (HBSS; Sigma-Aldrich, St Louis, MO) with 1% human serum albumin (HAS; ZLB Plasma, Boca Raton, FL). For flow cytometry measurements, 40 ml of heparin-anticoagulated peripheral venous blood from healthy donors was mixed with 6% Dextran 70 in 0.9% NaCl. After 60 min sedimentation, nucleated cells were collected and centrifuged at 211 g for 10 min. Remaining RBCs were lysed using 0.2% NaCl for 30 s with subsequent addition of 1.6% NaCl. Cells were spun down and resuspended in HBSS without Ca2+ and Mg2+. Neutrophils were isolated from these cells by centrifugation (30 min, 123 g, room temperature) through Ficoll-Hypaque lymphocyte separation medium (Cellgro-Mediatech, Herndon, VA). For microfluidics experiments, blood was drawn using a 19G butterfly needle into a K2EDTA tube containing 1 mM EDTA. Whole blood was then added to 5 ml polystyrene round-bottom tubes, with 2 ml whole blood added per tube. Isolation cocktail (50 µl/ml; EasySep™ Direct Human Neutrophil Isolation Kit, STEMCELL Technologies) and well vortexed RapidSpheres (50 µl/ml; STEMCELL Technologies) were then added to the whole blood and incubated at room temperature for 5 min. Separation buffer (Ca2+ and Mg2+-free PBS with 1 mM EDTA) was added to top up the volume in each tube to 4 ml, with gentle mixing, before the tubes were placed (without lid) into a magnetic rack and incubated for 5 min. The enriched cell suspension was collected into new tubes with additional RapidSpheres (same volume as before). The purification steps were repeated twice before the cells were ready for use. To purify RBCs for micropipette experiments, whole blood was layered over Histopaque 1119 (Sigma-Aldrich) and centrifuged (30 min, 2000 g, room temperature). The pelleted RBCs were washed and stored in EAS45 solution (Dumaswala et al., 1996).

Proteins, antibodies and inhibitors

Human ICAM-1–Ig and E-selectin–Ig were from R&D Systems (Minneapolis, MN) and Glycotech (Gaithersburg, MD), respectively. Human soluble monomeric E-selectin and the anti-E-selectin mAb ES1 have been described previously (Patel et al., 1995). Human P-selectin dimer has been described previously (Ramachandran et al., 2001). Anti-αLβ2 mAb TS1/22 and control mAb TS2/16 (anti-β1 integrin) were from ThermoFisher (Waltham, MA; MA11A10 and 12-0299-42, respectively). Anti-αMβ2 mAb 2LPM19c was from ThermoFisher (Waltham, MA; MA1-91659). Anti-β2 integrin mAb 7E4 (PE-conjugated and unconjugated) and PE-conjugated isotype control mouse IgG1 were from Beckman Coulter (Fullerton, CA). Two mouse anti-human ICAM-1 mAbs were used to determine ICAM-1 site density: FITC-conjugated MEM-111 was from Caltag Laboratories (Carlsbad, CA) and PE-conjugated anti-human CD54 was from eBioscience (San Diego, CA). FITC-conjugated goat anti-mouse IgG (Pierce) was used as secondary antibody to determine E-selectin site densities. Biotinylated goat anti-human Fc antibody (A18821) for capturing ICAM-1–Ig and E-selectin–Ig was from ThermoFisher. The Syk inhibitor piceatannol was from Calbiochem (San Diego, CA). IgG from human serum was from Sigma-Aldrich (St Louis, MO). EDTA was from BD Biosciences (San Jose, CA). HI-111 was from Fisher Scientific (Hampton, NH; 50-124-28). MEM-148 (anti-β2-integrin) was from Santa Cruz Biotechnology (Dallas, TX; sc-8420). The mAb MEM25 was from BioXcell (West Lebanon, NH; BE0048). Kim127 was purified from the supernatant of culture medium of KIM127 B cell hybridoma, which was purchased from ATCC (Manassas, VA; CRL-2838). Recombinant human ICAM-1–Fc chimera was from R&D Systems (Minneapolis, MN). The EasySep human neutrophil isolation kit was from STEMCELL Technology (Cambridge, MA).

Protein coupling to RBCs and site density determination

For the antibody capturing method, RBCs were biotinylated using biotin-X-NHS (Calbiochem), then incubated subsequently with 10 µg/ml streptavidin (Pierce), 10 µg/ml biotinylated goat anti-human Fc antibody, and different concentrations of E-selectin–Ig, ICAM-1–Ig or both at desired ratios. CrCl3 coupling was used to directly couple soluble monomeric E-selectin, E-selectin–Ig and ICAM-1-Ig onto RBCs as described previously (Chesla et al., 1998).

Site densities of ICAM-1–Ig, E-selectin–Ig and monomeric E-selectin on RBCs, as well as of β2 integrin on neutrophils, were determined by flow cytometry using a standard immunofluorescence staining protocol (Chesla et al., 1998). Samples were read on a BD LSR flow cytometer (Becton-Dickinson Immunocytometry Systems, San Jose, CA) and site densities of cell-surface molecules were determined using standard beads [Quantum 25 FITC High Level (Bangs Laboratories, Fishers, IN), or QuantiBRITE PE beads (BD Biosciences, San Jose, CA)] as per the manufacturer's instructions.

Micropipette adhesion frequency assay

As described previously (Chesla et al., 1998), the adhesion frequency assay used a micropipette-aspirated RBC (Fig. 1A, left) driven by a piezoelectric translator to repeatedly contact a neutrophil aspirated by an apposing pipette (Fig. 1A, right) for a controlled duration tc and area Ac. Cells at room temperature were added into a chamber (preblocked with HBSS containing 1% HSA for 30 min to prevent neutrophils from adhering to the bottom of the chamber) mounted on the stage of an inverted microscope (Leica DM-IRBE microscope and 20× objective magnification). The presence or absence of adhesion was detected by membrane deflections of the RBC upon its retraction (Movie 1). The adhesion frequency Pa was the ratio of the number of adhesions to the number of contacts from a single pair of cells for a given tc. The specific frequency was determined by removing the nonspecific frequency from the total frequency observed (Zhu and Williams, 2000), which was determined using 3–5 pairs of cells per time point to obtain the mean and s.e.m. Contact time tc was varied from 0.2–20 s to obtain a Pa versus tc curve. Two to three curves were generated by varying the densities of ICAM-1 and/or E-selectin. The probability data were fitted to a probabilistic model (Eqn 1A,B) that describes a second-order forward and first-order reverse single-step interaction between a single species of receptors and a single species of ligands (Chesla et al., 1998). An independent thermal fluctuation assay supports the accuracy of these parameters (Chen et al., 2008).

Microfluidic assay and device fabrication

As described previously (Zhou et al., 2018), the cells in the multi-zone channels flowed and encountered the surfaces coated with various proteins in a sequential manner. Device surfaces can be functionalized as single, double, triple or quadruple zones with various proteins. The auto-alignment of protein zones was achieved using two matched polydimethylsiloxane (PDMS) stamps (Zhou et al., 2018). The first carries blocking protein to contact and transfer blocking protein to the second stamp, which automatically creates a blocked zone (for relaxation), a blank zone (for the next coating step) and two channels (for channel coating) when the second stamp makes firm contact with the coverslip. We delivered ICAM-1 and/or antibodies to these two channels for protein coating and then blocking protein to prevent contamination at later steps. A third PDMS channel was aligned perpendicular to the protein pattern to coat the blank area with P-selectin before running experiments. Neutrophil rolling was imaged in zones of P-selectin or ICAM-1 (low coating concentration 20 μg/ml) and firm adhesion was captured on ICAM-1-coated surfaces (high coating concentration 100 μg/ml). Standard soft lithography was used to fabricate devices in PDMS (Sylgard 184; Corning, Midland, MI). Two matched silicon wafers were fabricated (SU8-2025; MicroChem, Round Rock, TX). The first was a one-layer mold with a 25 μm-thick single layer for microchannel PDMS. The second was a two-layer mold, 25-μm thick each for stamping PDMS. A mixture of PDMS (PDMS base and crosslinker in a 10:1 ratio) was poured onto the mold, and the whole preparation was left for curing for 2 h at 75°C. After peeling off the PDMS, the devices were cut into shape, and access holes were punched using 19G needles (McMaster-Carr, Elmhurst, IL). Plastic clamp holders were laser-cut with holes matching the inlets and outlets. Plastic holders and screws were applied to hold the PDMS device and glass substrate together. After assembly, the device was primed in a vacuum to remove bubbles. The chip was connected to syringe pumps to deliver flow for protein coating.

Statistics

Statistical analysis was performed using Graphpad Prism 8.0. Differences between data or samples were analyzed using two-tailed unpaired Student's t-tests (for points) or two-way ANOVA with a post hoc Tukey's test to compare groups (for curves). Data are presented as mean±s.e.m. The equation v=B0+B1(logτ)+B2(logτ)2 was used to fit the data in Fig. 3C. For experiment 2, B2 was 11.01 (95% c.i. of 6.43–15.6), indicating a significant positive quadratic curve of the velocity and log of shear rate τ. Fitted results of experiment 2: V=5.7+9.4(log τ)+11(log τ)2. Fitted results of experiment 3: V=2.3+2.6(log τ)+3.8(log τ)2. Fitted results of experiment 4: V=3.2+3.7(log τ)+4.5(log τ)2. A second-order polynomial (y=P1x2+P2x+P3) was fitted to the data in Fig. 4A, with all P1 values negative across the 95% c.i., indicating a significant biphasic trend. (P1=−0.002; 95% c.i. −0.003 to −0.0006).

Micropipette adhesion frequency data are presented as mean±s.e.m. of five pairs of cells for each condition at each time point. Microfluidic channel data are presented as mean±s.e.m. of 15–20 cells, or as individual point measurements that are representative of consistent repeated results.

Author contributions

Conceptualization: F. Zhou, F. Zhang, C.Z.; Methodology: F. Zhou, F. Zhang, V.I.Z., H.L., C.Z.; Validation: F. Zhou, F. Zhang, V.I.Z., L.D., Z.Y., K.L.; Formal analysis: F. Zhou, F. Zhang, V.I.Z.; Investigation: F. Zhou, F. Zhang, V.I.Z., L.D.; Resources: R.P.M.; Data curation: F. Zhou, F. Zhang, V.I.Z., Z.Y., K.L.; Writing - original draft: F. Zhou, F. Zhang; Writing - review & editing: F. Zhou, V.I.Z., R.P.M., H.L., C.Z.; Supervision: H.L., C.Z.; Project administration: C.Z.; Funding acquisition: H.L., C.Z.

Funding

This work was supported by the National Institutes of Health grants R01HL18671, U01CA214354 and U01CA250040 to C.Z. Deposited in PMC for release after 12 months.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258046

Bachmann
,
M.
,
Kukkurainen
,
S.
,
Hytönen
,
V. P.
and
Wehrle-Haller
,
B.
(
2019
).
Cell Adhesion by integrins
.
Physiol. Rev.
99
,
1655
-
1699
.
Beste
,
M. T.
and
Hammer
,
D. A.
(
2008
).
Selectin catch–slip kinetics encode shear threshold adhesive behavior of rolling leukocytes
.
Proc. Natl Acad. Sci. USA
105
,
20716
-
20721
.
Casas
,
J.
,
Brzostek
,
J.
,
Zarnitsyna
,
V. I.
,
Hong
,
J.-S.
,
Wei
,
Q.
,
Hoerter
,
J. A. H.
,
Fu
,
G.
,
Ampudia
,
J.
,
Zamoyska
,
R.
,
Zhu
,
C.
et al. 
(
2014
).
Ligand-engaged TCR is triggered by Lck not associated with CD8 coreceptor
.
Nat. Commun.
5
,
5624
.
Chen
,
W.
,
Evans
,
E. A.
,
McEver
,
R. P.
and
Zhu
,
C.
(
2008
).
Monitoring receptor-ligand interactions between surfaces by thermal fluctuations
.
Biophys. J.
94
,
694
-
701
.
Chen
,
W.
,
Lou
,
J.
and
Zhu
,
C.
(
2010
).
Forcing switch from short- to intermediate- and long-lived states of the αA domain generates LFA-1/ICAM-1 catch bonds
.
J. Biol. Chem.
285
,
35967
-
35978
.
Chen
,
W.
,
Lou
,
J.
,
Evans
,
E. A.
and
Zhu
,
C.
(
2012
).
Observing force-regulated conformational changes and ligand dissociation from a single integrin on cells
.
J. Cell Biol.
199
,
497
-
512
.
Chen
,
Y.
,
Ju
,
L. A.
,
Zhou
,
F.
,
Liao
,
J.
,
Xue
,
L.
,
Su
,
Q. P.
,
Jin
,
D.
,
Yuan
,
Y.
,
Lu
,
H.
,
Jackson
,
S. P.
et al. 
(
2019
).
An integrin αIIbβ3 intermediate affinity state mediates biomechanical platelet aggregation
.
Nat. Mater.
18
,
760
-
769
.
Chesla
,
S. E.
,
Selvaraj
,
P.
and
Zhu
,
C.
(
1998
).
Measuring two-dimensional receptor-ligand binding kinetics by micropipette
.
Biophys. J.
75
,
1553
-
1572
.
Chesnutt
,
B. C.
,
Smith
,
D. F.
,
Raffler
,
N. A.
,
Smith
,
M. L.
,
White
,
E. J.
and
Ley
,
K.
(
2006
).
Induction of LFA-1-dependent neutrophil rolling on ICAM-1 by engagement of E-selectin
.
Microcirculation
13
,
99
-
109
.
Constantin
,
G.
,
Majeed
,
M.
,
Giagulli
,
C.
,
Piccio
,
L.
,
Kim
,
J. Y.
,
Butcher
,
E. C.
and
Laudanna
,
C.
(
2000
).
Chemokines trigger immediate β2 integrin affinity and mobility changes: differential regulation and roles in lymphocyte arrest under flow
.
Immunity
13
,
759
-
769
.
Diamond
,
M. S.
,
Staunton
,
D. E.
,
de Fougerolles
,
A. R.
,
Stacker
,
S. A.
,
Garcia-Aguilar
,
J.
,
Hibbs
,
M. L.
and
Springer
,
T. A.
(
1990
).
ICAM-1 (CD54): a counter-receptor for Mac-1 (CD11b/CD18)
.
J. Cell Biol.
111
,
3129
-
3139
.
Dumaswala
,
U. J.
,
Wilson
,
M. J.
,
Jose
,
T.
and
Daleke
,
D. L.
(
1996
).
Glutamine- and phosphate-containing hypotonic storage media better maintain erythrocyte membrane physical properties
.
Blood
88
,
697
-
704
.
Fan
,
Z.
and
Ley
,
K.
(
2015
).
Leukocyte arrest: biomechanics and molecular mechanisms of β2 integrin activation
.
Biorheology
52
,
353
-
377
.
Fiore
,
V. F.
,
Ju
,
L.
,
Chen
,
Y.
,
Zhu
,
C.
and
Barker
,
T. H.
(
2014
).
Dynamic catch of a Thy-1–α5β1+syndecan-4 trimolecular complex
.
Nat. Commun.
5
,
4886
.
Green
,
C. E.
,
Pearson
,
D. N.
,
Camphausen
,
R. T.
,
Staunton
,
D. E.
and
Simon
,
S. I.
(
2004
).
Shear-dependent capping of L-selectin and P-selectin glycoprotein ligand 1 by E-selectin signals activation of high-avidity β2-integrin on neutrophils
.
J. Immunol.
172
,
7780
-
7790
.
Hogg
,
N.
,
Patzak
,
I.
and
Willenbrock
,
F.
(
2011
).
The insider's guide to leukocyte integrin signalling and function
.
Nat. Rev. Immunol.
11
,
416
-
426
.
Jiang
,
N.
,
Huang
,
J.
,
Edwards
,
L. J.
,
Liu
,
B.
,
Zhang
,
Y.
,
Beal
,
C. D.
,
Evavold
,
B. D.
and
Zhu
,
C.
(
2011
).
Two-stage cooperative T cell receptor-peptide major histocompatibility complex-CD8 trimolecular interactions amplify antigen discrimination
.
Immunity
34
,
13
-
23
.
Ju
,
L.
,
Chen
,
Y.
,
Li
,
K.
,
Yuan
,
Z.
,
Liu
,
B.
,
Jackson
,
S. P.
and
Zhu
,
C.
(
2017
).
Dual biomembrane force probe enables single-cell mechanical analysis of signal crosstalk between multiple molecular species
.
Sci. Rep.
7
,
14185
.
Kuwano
,
Y.
,
Spelten
,
O.
,
Zhang
,
H.
,
Ley
,
K.
and
Zarbock
,
A.
(
2010
).
Rolling on E- or P-selectin induces the extended but not high-affinity conformation of LFA-1 in neutrophils
.
Blood
116
,
617
-
624
.
Lefort
,
C. T.
and
Ley
,
K.
(
2012
).
Neutrophil arrest by LFA-1 activation
.
Front. Immunol.
3
,
157
.
Li
,
N.
,
Mao
,
D.
,
,
S.
,
Tong
,
C.
,
Zhang
,
Y.
and
Long
,
M.
(
2013
).
Distinct binding affinities of Mac-1 and LFA-1 in neutrophil activation
.
J. Immunol.
190
,
4371
-
4381
.
Li
,
K.
,
Yuan
,
Z.
,
Lyu
,
J.
,
Ahn
,
E.
,
Davis
,
S. J.
,
Ahmed
,
R.
and
Zhu
,
C.
(
2021
).
PD-1 suppresses TCR-CD8 cooperativity during T-cell antigen recognition
.
Nat. Commun.
12
,
2746
.
Liu
,
B.
,
Chen
,
W.
,
Evavold
,
B. D.
and
Zhu
,
C.
(
2014a
).
Accumulation of dynamic catch bonds between TCR and agonist peptide-MHC triggers T cell signaling
.
Cell
157
,
357
-
368
.
Liu
,
B.
,
Zhong
,
S.
,
Malecek
,
K.
,
Johnson
,
L. A.
,
Rosenberg
,
S. A.
,
Zhu
,
C.
and
Krogsgaard
,
M.
(
2014b
).
2D TCR-pMHC-CD8 kinetics determines T-cell responses in a self-antigen-specific TCR system
.
Eur. J. Immunol.
44
,
239
-
250
.
Long
,
M.
,
Zhao
,
H.
,
Huang
,
K.-S.
and
Zhu
,
C.
(
2001
).
Kinetic measurements of cell surface E-Selectin/carbohydrate ligand interactions
.
Ann. Biomed. Eng.
29
,
935
-
946
.
Luo
,
B.-H.
,
Carman
,
C. V.
and
Springer
,
T. A.
(
2007
).
Structural basis of integrin regulation and signaling
.
Annu. Rev. Immunol.
25
,
619
-
647
.
Ma
,
Q.
,
Shimaoka
,
M.
,
Lu
,
C.
,
Jing
,
H.
,
Carman
,
C. V.
and
Springer
,
T. A.
(
2002
).
Activation-induced conformational changes in the I domain region of lymphocyte function-associated antigen 1
.
J. Biol. Chem.
277
,
10638
-
10641
.
Marshall
,
B. T.
,
Long
,
M.
,
Piper
,
J. W.
,
Yago
,
T.
,
McEver
,
R. P.
and
Zhu
,
C.
(
2003
).
Direct observation of catch bonds involving cell-adhesion molecules
.
Nature
423
,
190
-
193
.
McDonough
,
D. B.
,
McIntosh
,
F. A.
,
Spanos
,
C.
,
Neelamegham
,
S.
,
Goldsmith
,
H. L.
and
Simon
,
S. I.
(
2004
).
Cooperativity between selectins and β2-integrins define neutrophil capture and stable adhesion in shear flow
.
Ann. Biomed. Eng.
32
,
1179
-
1192
.
McEver
,
R. P.
(
2015
).
Selectins: initiators of leucocyte adhesion and signalling at the vascular wall
.
Cardiovasc. Res.
107
,
331
-
339
.
McEver
,
R. P.
and
Zhu
,
C.
(
2007
).
A catch to integrin activation
.
Nat. Immunol.
8
,
1035
-
1037
.
McEver
,
R. P.
and
Zhu
,
C.
(
2010
).
Rolling cell adhesion
.
Annu. Rev. Cell Dev. Biol.
26
,
363
-
396
.
Miner
,
J. J.
,
Xia
,
L.
,
Yago
,
T.
,
Kappelmayer
,
J.
,
Liu
,
Z.
,
Klopocki
,
A. G.
,
Shao
,
B.
,
McDaniel
,
J. M.
,
Setiadi
,
H.
,
Schmidtke
,
D. W.
et al. 
(
2008
).
Separable requirements for cytoplasmic domain of PSGL-1 in leukocyte rolling and signaling under flow
.
Blood
112
,
2035
-
2045
.
Patel
,
K. D.
,
Moore
,
K. L.
,
Nollert
,
M. U.
and
McEver
,
R. P.
(
1995
).
Neutrophils use both shared and distinct mechanisms to adhere to selectins under static and flow conditions
.
J. Clin. Invest.
96
,
1887
-
1896
.
Pryshchep
,
S.
,
Zarnitsyna
,
V. I.
,
Hong
,
J.
,
Evavold
,
B. D.
and
Zhu
,
C.
(
2014
).
Accumulation of serial forces on TCR and CD8 frequently applied by agonist antigenic peptides embedded in MHC molecules triggers calcium in T cells
.
J. Immunol.
193
,
68
-
76
.
Rainger
,
G. E.
,
Fisher
,
A. C.
and
Nash
,
G. B.
(
1997
).
Endothelial-borne platelet-activating factor and interleukin-8 rapidly immobilize rolling neutrophils
.
Am. J. Physiol. Heart Circ. Physiol.
272
,
H114
-
H122
.
Ramachandran
,
V.
,
Yago
,
T.
,
Epperson
,
T. K.
,
Kobzdej
,
M. M. A.
,
Nollert
,
M. U.
,
Cummings
,
R. D.
,
Zhu
,
C.
and
McEver
,
R. P.
(
2001
).
Dimerization of a selectin and its ligand stabilizes cell rolling and enhances tether strength in shear flow
.
Proc. Natl. Acad. Sci. USA
98
,
10166
-
10171
.
Rosetti
,
F.
,
Chen
,
Y.
,
Sen
,
M.
,
Thayer
,
E.
,
Azcutia
,
V.
,
Herter
,
J. M.
,
Luscinskas
,
F. W.
,
Cullere
,
X.
,
Zhu
,
C.
and
Mayadas
,
T. N.
(
2015
).
A lupus-associated mac-1 variant has defects in integrin allostery and interaction with ligands under force
.
Cell Rep.
10
,
1655
-
1664
.
Setiadi
,
H.
and
McEver
,
R. P.
(
2008
).
Clustering endothelial E-selectin in clathrin-coated pits and lipid rafts enhances leukocyte adhesion under flow
.
Blood
111
,
1989
-
1998
.
Smith
,
C. W.
,
Marlin
,
S. D.
,
Rothlein
,
R.
,
Toman
,
C.
and
Anderson
,
D. C.
(
1989
).
Cooperative interactions of LFA-1 and Mac-1 with intercellular adhesion molecule-1 in facilitating adherence and transendothelial migration of human neutrophils in vitro
.
J. Clin. Invest.
83
,
2008
-
2017
.
Stadtmann
,
A.
,
Germena
,
G.
,
Block
,
H.
,
Boras
,
M.
,
Rossaint
,
J.
,
Sundd
,
P.
,
Lefort
,
C.
,
Fisher
,
C. I.
,
Buscher
,
K.
,
Gelschefarth
,
B.
et al. 
(
2013
).
The PSGL-1-L-selectin signaling complex regulates neutrophil adhesion under flow
.
J. Exp. Med.
210
,
2171
-
2180
.
Sun
,
Z.
,
Costell
,
M.
and
Fässler
,
R.
(
2019
).
Integrin activation by talin, kindlin and mechanical forces
.
Nat. Cell Biol.
21
,
25
-
31
.
Williams
,
T. E.
,
Nagarajan
,
S.
,
Selvaraj
,
P.
and
Zhu
,
C.
(
2000a
).
Concurrent and independent binding of Fcγ receptors IIa and IIIb to surface-bound IgG
.
Biophys. J.
79
,
1867
-
1875
.
Williams
,
T. E.
,
Selvaraj
,
P.
and
Zhu
,
C.
(
2000b
).
Concurrent binding to multiple ligands: kinetic rates of CD16b for membrane-bound IgG1 and IgG2
.
Biophys. J.
79
,
1858
-
1866
.
Woolf
,
E.
,
Grigorova
,
I.
,
Sagiv
,
A.
,
Grabovsky
,
V.
,
Feigelson
,
S. W.
,
Shulman
,
Z.
,
Hartmann
,
T.
,
Sixt
,
M.
,
Cyster
,
J. G.
and
Alon
,
R.
(
2007
).
Lymph node chemokines promote sustained T lymphocyte motility without triggering stable integrin adhesiveness in the absence of shear forces
.
Nat. Immunol.
8
,
1076
-
1085
.
Yago
,
T.
,
Leppänen
,
A.
,
Qiu
,
H.
,
Marcus
,
W. D.
,
Nollert
,
M. U.
,
Zhu
,
C.
,
Cummings
,
R. D.
and
McEver
,
R. P.
(
2002
).
Distinct molecular and cellular contributions to stabilizing selectin-mediated rolling under flow
.
J. Cell Biol.
158
,
787
-
799
.
Yago
,
T.
,
Wu
,
J.
,
Wey
,
C. D.
,
Klopocki
,
A. G.
,
Zhu
,
C.
and
McEver
,
R. P.
(
2004
).
Catch bonds govern adhesion through L-selectin at threshold shear
.
J. Cell Biol.
166
,
913
-
923
.
Yago
,
T.
,
Zhang
,
N.
,
Zhao
,
L.
,
Abrams
,
C. S.
and
McEver
,
R. P.
(
2018
).
Selectins and chemokines use shared and distinct signals to activate β2 integrins in neutrophils
.
Blood Adv.
2
,
731
-
744
.
Zarbock
,
A.
,
Lowell
,
C. A.
and
Ley
,
K.
(
2007
).
Spleen tyrosine kinase Syk is necessary for E-selectin-induced αLβ2 integrin-mediated rolling on intercellular adhesion molecule-1
.
Immunity
26
,
773
-
783
.
Zhang
,
F.
,
Marcus
,
W. D.
,
Goyal
,
N. H.
,
Selvaraj
,
P.
,
Springer
,
T. A.
and
Zhu
,
C.
(
2005
).
Two-dimensional kinetics regulation of αLβ2-ICAM-1 interaction by conformational changes of the αL-inserted domain
.
J. Biol. Chem.
280
,
42207
-
42218
.
Zhang
,
Y.
,
Jiang
,
N.
,
Zarnitsyna
,
V. I.
,
Klopocki
,
A. G.
,
McEver
,
R. P.
and
Zhu
,
C.
(
2013
).
P-selectin glycoprotein ligand-1 forms dimeric interactions with E-selectin but monomeric interactions with L-selectin on cell surfaces
.
PLoS ONE
8
,
e57202
.
Zhou
,
F.
,
Chen
,
Y.
,
Felner
,
E. I.
,
Zhu
,
C.
and
Lu
,
H.
(
2018
).
Microfluidic auto-alignment of protein patterns for dissecting multi-receptor crosstalk in platelets
.
Lab. Chip
18
,
2966
-
2974
.
Zhu
,
C.
and
Williams
,
T. E.
(
2000
).
Modeling concurrent binding of multiple molecular species in cell adhesion
.
Biophys. J.
79
,
1850
-
1857
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information