Conformational equilibria and intrinsic affinities define integrin activation

Abstract

We show that the three conformational states of integrin α₅β₁ have discrete free energies and define activation by measuring intrinsic affinities for ligand of each state and the equilibria linking them. The 5,000-fold higher affinity of the extended-open state than the bent-closed and extended-closed states demonstrates profound regulation of affinity. Free energy requirements for activation are defined with protein fragments and intact α₅β₁ On the surface of K562 cells, α₅β₁ is 99.8% bent-closed. Stabilization of the bent conformation by integrin transmembrane and cytoplasmic domains must be overcome by cellular energy input to stabilize extension. Following extension, headpiece opening is energetically favored. N-glycans and leg domains in each subunit that connect the ligand-binding head to the membrane repel or crowd one another and regulate conformational equilibria in favor of headpiece opening. The results suggest new principles for regulating signaling in the large class of receptors built from extracellular domains in tandem with single-span transmembrane domains.

Keywords: N‐glycan; affinity; conformation; integrin; thermodynamics.

Figure 1. Thermodynamic equilibria of integrin α₅β₁ and relation to other receptors

1. A–C

   Integrins. (A) Integrin structure and arrangement of domains in the three overall states of the conformational ensemble (Luo et al, 2007), with dashed lines representing the flexibility of the lower β‐leg in extended conformations (schematic, upper). In the Pymol representation (lower), structures based on intact, bent‐closed integrin α_Vβ₃ on the cell surface (Zhu et al, 2009), in the same integrin subfamily as α₅β₁, are shown using an ellipsoid or torus for each extracellular domain. The extended‐closed structure is made by rigid body movements at the knees. The extended‐open structure is derived from extended‐closed by superposition on the open headpiece of α_IIbβ₃ (Xiao et al, 2004). (B) Identical to (A, lower) except molecular surfaces are shown for N‐glycans removed by mutation (Fig 6D–F, white) or not removed (blue). Glycans have the structure shown in (E). Glycans were added at all α₅β₁ N‐glycosylation sequons using α₅β₁ headpiece structures (Xia & Springer, 2014) and an α_Vβ₃ homology model. (C) Equations that describe the affinities and conformational equilibria of integrin ensembles.

2. D

   Representative cytokine or growth factor receptor. The ligand‐bound dimer is compact similarly to the bent‐closed integrin conformation, whereas the unassociated monomers show no interaction with one another and thus separate from one another similarly to leg domains in the extended‐open integrin structure. Therefore, inter‐subunit crowding interactions may regulate ligand binding and dimerization in such receptors analogously to regulation of conformational transitions in integrins. Individual domains are shown as ellipses, and N‐glycans are shown as blue molecular surfaces using the same chemical structures as in (B).

3. E

   Representative N‐glycan structures. The complex glycan is common among α₅β₁ N‐glycans (Sieber et al, 2007) and contains the average number of monosaccharide residues per N‐glycosylation sequon found here for the complex glycoform of α₅β₁. The high‐mannose glycan contains the average number of monosaccharide residues in the high‐mannose form of α₅β₁. The shaved glycan is that after Endo H digestion; the mass estimated here for shaved α₅β₁ suggests part of high‐mannose glycans are shaved and the remainder are resistant to Endo H.

Figure 2. Influence of conformation‐selective Fabs and metal ions on ligand binding by the α₅β₁ ectodomain

1. A

   Coomassie‐stained, non‐reducing SDS 12.5% PAGE of Fabs.

2. B, C

   Dependence of unclasped, high‐mannose α₅β₁ ectodomain (20 nM) binding to FITC‐cRGD (5 nM) in FP assays on Fab concentration in 1 mM Mg²⁺ & 1 mM Ca²⁺ (B) and in 1 mM Mg²⁺ & 1 mM Ca²⁺ compared to 1 mM Mn²⁺ & 1 mM Ca²⁺ (C). Values to the right of each plot are from nonlinear least square fits to Appendix Equation S1 (mean ± s.d. of triplicates).

Figure 3. Intrinsic and ensemble affinities of α₅β₁ ectodomain and headpiece preparations for cRGD

1. A, B

   Affinities of high‐mannose glycoforms of the unclasped α₅β₁ ectodomain (A) and headpiece (B) were measured using FP with FITC‐cRGD in the presence of the indicated Fabs. Errors in the plotted datapoints are s.d. from average value of triplicate measurements. Errors for affinities in the inset table are s.d. from nonlinear least square fits of the average values from the triplicate measurements except values with “*” are s.d. from three experiments on different days.

Figure 4. Intrinsic affinities and conformational equilibria of α₅β₁ with three different ligands

Data are for the unclasped, high‐mannose α₅β₁ ectodomain using the indicated Fabs.

1. Affinity for FITC‐RGD measured with FP.

2. Affinity for Fn3_9–10 measured by inhibition of FITC‐cRGD binding to 20 nM α₅β₁ (with open‐stabilizing Fab), 90 nM α₅β₁ (with extension‐stabilizing Fab), or 270 nM α₅β₁ (with no Fab).

3. Affinity for Fn3_9–10 measured by inhibition of FITC‐cRGD binding to a range of α₅β₁ concentrations (with closure‐stabilizing Fab) at three Fn3_9–10 concentrations.

4. Tabulation of results.

5. Energy landscape plots showing valleys for the ΔG values determined here as lines, and hills for the transition state ΔG values, which remain to be determined, as dashed lines.

Data information: (A–E) Errors in the plotted datapoints are s.d. from the average value of triplicate measurements. Errors in the affinity values are s.d. from nonlinear least square fits of the average value of triplicate measurements except values with “*” are s.d. from three experiments on different days and values with “#” represent s.d. of measurements with distinct Fabs stabilizing the same conformation. Errors for P and ΔG values are propagated from affinities values as described in the Materials and Methods.

Figure 5. Glycoform and C‐terminal clasp regulate α₅β₁ ectodomain conformational equilibria and have little effect on intrinsic affinities

1. A

   SDS 7.5% PAGE of three α₅β₁ ectodomain glycoforms with or without C‐terminal clasp stained with Coomassie blue.

2. B, C

   Tabulation of results (B) and energy landscape plots (C). See Appendix Fig S4 for representative FP results. Error definitions and explanations of values marked with “*” and “#” are as in the Fig 4 legend.

3. D, E

   Effect of complex (D) and shaved (E) N‐glycans on clasped α₅β₁ ectodomain conformation in EM. Representative, well‐resolved class averages are shown together with their rank among 35 class averages and the % of total particles in that class (inset, upper left and right, respectively). Below each en face class average are the best cross‐correlating projections of open and closed headpiece crystal structures to the headpiece portion of the EM class average. Insets show correlation coefficients with the best correlating conformation in yellow. Scale bars are 10 nm. All class averages are shown in Appendix Fig S6.

Figure 6. Regulation of α₅β₁ conformational equilibria by the lower legs and N‐glycans

1. A–C

   The semi‐truncated α₅β₁ ectodomain. (A) Coomassie‐stained SDS 7.5% PAGE. (B) Affinity of FITC‐cRGD for the semi‐truncated, high‐mannose α₅β₁ ectodomain in the presence of the indicated Fabs measured with FP. (C) Comparison of the semi‐truncated ectodomain to the unclasped α₅β₁ ectodomain and headpiece, all with high‐mannose glycans.

2. D–F

   Effect of N‐glycosylation sequon mutation. The ΔN‐α₅ and ΔN‐β₁ mutants have 6 of 14 and 5 of 12 predicted N‐glycosylation sites mutated, respectively. The indicated mutants were tested in the clasped α₅β₁ ectodomain with complex N‐glycosylation. (D) Coomassie‐stained SDS 7.5% PAGE. (E) Tabulation of results. See Appendix Fig S4 for representative FP results. ND, not determined; limited solubility of the N‐glycosylation sequon mutants prevented their use at the high concentrations required for intrinsic affinity measurements of the BC+EC ensemble. Therefore, thermodynamic parameters were calculated by assuming that the intrinsic affinities of the BC and EC conformations were identical to those determined for the WT ectodomain. (F) Energy landscape plots are as described in the Fig 4E legend.

Data information: (B, C, E, F) Error definitions are as described in the legend to Fig 4.

Figure 7. Conformational equilibria and intrinsic affinity of intact α₅β₁ on cell surfaces

Mean fluorescence intensity (MFI) after background subtraction is from quantitative fluorescence flow cytometry of K562 cells without washing.

1. Determination of EC₅₀ values for conformation‐selective Fabs from enhancement of 10 nM Alexa488‐Fn3_9–10 binding and fitting to Appendix Equation S2. Mean ± s.d. of least square fits to triplicates.

2. Affinity of α₅β₁ for Alexa488‐Fn3_9–10 in the presence of indicated Fabs.

3. Affinity of α₅β₁ for Fn3_9–10 by enhancement of 0.4 nM Alexa488‐12G10 Fab binding.

4. Affinities of α₅β₁ for Alexa488‐12G10 Fab in the presence of indicated Fabs.

5. Thermodynamics and intrinsic affinities of α₅β₁ conformational states on K562 cells and Jurkat cells (Appendix Fig S8). Affinity for Fn3_9–10 of the closed conformations of α₅β₁ on K562 cells was estimated from $K_{\mathrm{d}}^{\mathrm{EO}}$ using the same fold‐difference as found with Fn3_9–10 for the α₅β₁ ectodomain (Fig 4D). Since 12G10 stabilizes the open conformation only, thermodynamic calculations use K _a = 0 for the closed conformations.

6. Energy landscape plots, as described in the Fig 4E legend, comparing K562 α₅β₁ to the clasped and unclasped α₅β₁ ectodomain with complex glycosylation (Fig 5B).

7. Summary of the intrinsic affinities of α₅β₁ conformational states on K562 cells and comparison of the conformational equilibria for α₅β₁ on K562 cells and for the unclasped α₅β₁ ectodomain with complex glycosylation.

Data information: (B–F) Error definitions are as described in the legend to Fig 4.