Quantitative detection of conformational transitions in a calcium sensor protein by surface plasmon resonance

Abstract

We determined the conditions under which surface plasmon resonance can be used to monitor at real-time the Ca²⁺-induced conformational transitions of the sensor protein recoverin immobilized over a sensor chip. The equilibrium and the kinetics of conformational transitions were detected and quantified over a physiological range of Ca²⁺ and protein concentrations similar to those found within cells. Structural analysis suggests that the detection principle reflects changes in the hydrodynamic properties of the protein and is not due to a mass effect. The phenomenon appears to be related to changes in the refractive index at the metal/dielectric interface.

---

Calcium plays a central role in many signaling pathways including neuronal and sensory function. Detection of Ca²⁺-induced conformational changes in a Ca²⁺-sensor protein requires a method that is sensitive to its physico-chemical properties in the apo and in the Ca²⁺-loaded form. Ideally, the experiments should be performed under conditions that mimic those in the cell, in terms of relative protein and metal concentration.

Surface Plasmon Resonance (SPR) is a powerful tool to measure biomolecular interactions, but it has only occasionally been used to detect protein isomerization.^1,2 The aim of the present study was to use SPR for monitoring conformational transitions in a Ca²⁺-sensor triggered by physiological [Ca²⁺]. We show for the first time that SPR can be employed to measure the [Ca²⁺] at which the conformational transition is half-maximal by independent analysis of both equilibrium and kinetic data, hence providing useful information as on the timing of the concerted binding/transition events.

The neuronal Ca²⁺-sensor protein recoverin was best suited for testing our approach. It belongs to the EF-hand superfamily of Ca²⁺-binding proteins and binds two Ca²⁺ with an apparent K_D-value of 17–18 μM.^3,4 It operates in photoreceptor cells, where [Ca²⁺] changes dynamically upon light exposure, ranging from about 0.6 μM in dark-adapted cells to 0.1 μM in the light.⁵ Recoverin is myristoylated at the amino-terminus and undergoes a Ca²⁺-myristoyl switch upon Ca²⁺-binding, which is associated with the exposure of a covalently attached myristoyl group.⁶

Myristoylated Rec (mRec) was immobilized on a carboxymethyldextran-layer present on the surface of a commercial sensor chip. The slow immobilization procedure (see ESI†) ensures the full embedding of the ∼4 nm protein within the ∼100 nm extension of the dextran matrix. Site-specific immobilization of mRec was achieved by the single cysteine at position 38 ensuring a well defined orientation of the immobilized protein. The effects of non-site-specific coupling to the matrix were also tested via amine coupling that targets amines in Lys residues and at the N-terminus of the proteins (See ESI†). Control recordings were made with the myristoylated recoverin mutant mRec^E121Q in which an E121Q mutation in the third EF-hand impairs the binding of Ca²⁺ up to >10 mM [Ca²⁺].⁴ Therefore, no conformational change is expected for mRec^E121Q in the range of [Ca²⁺] tested in this study. The two Rec variants were immobilized on the sensor chip in two different flow cells at similar final levels, namely 5.8 ng of mRec and 6.4 ng of mRec^E121Q, which corresponds to ∼12.7 μM and 13.9 μM concentrations within the volume of the dextran layer, and was close to the assumed 17 μM cellular concentration of Rec.⁷ A third flow cell was activated and deactivated in the same way as the other two, but no protein was immobilized, hence constituting a convenient reference cell. Ca²⁺ stocks at final concentrations of 0.4–46.2 μM were prepared in 5 mM Tris/HCl, pH 7.5, 150 mM KCl decalcified buffer and injected over each flow cell in repeated experiments.

The SPR response in terms of resonance units (RU) was shown to clearly depend on [Ca²⁺] for mRec (Fig. 1a), while the same Ca²⁺ injections over the control (immobilized mRec^E121Q) led to virtually no response (Fig. 1b). These results were highly reproducible in independent sets of experiments (Fig. 1c), which also indicates a functional regeneration of the sensor chip surface upon removal of Ca²⁺. The amplitudes of the sensorgrams saturated at Ca²⁺-injections of 14–50 μM showing relative responses of >150 RU. These exceeded the theoretical Ca²⁺-binding capacity of the protein-coated surfaces, which was 20.3 RU and 11.1 RU for mRec and mRec^E121Q, respectively. Therefore, the SPR response was a consequence of the concerted Ca²⁺-binding and -induced protein conformational change rather than being due to a mere mass effect. This overall process caused a change in refractive index in the vicinity of the protein–dielectric interface and strongly depended on Ca²⁺. Our interpretation is supported by the analysis of the NMR structures of apo and Ca²⁺-mRec, reported as average structures (Fig. 1d).^6,8 Ca²⁺-free recoverin has a tense (T), compact conformation in which the myristoyl group is sequestered in a hydrophobic pocket. It undergoes a transition to a released (R), more extended conformation in the Ca²⁺-loaded form, in which the myristoyl is solvent-exposed.⁶ The transition is characterized by both an increase in the radius of gyration and total solvent-accessible surface area (SAS) (Fig. 1d). In particular, for the R conformation the increase in hydrophobic SAS significantly exceeds that in hydrophilic SAS (22% vs. 8%, respectively). It is therefore plausible that the solvation shell of the R conformation differs significantly from that of the T conformation, having consequences on the surrounding water structure and the overall refractive index. Accordingly, no change in resonance units was observed when Ca²⁺-binding was impaired and the conformational transition was thus prevented, such as in the case of mRec^E121Q (Fig. 1b and c).

Fig. 1 Surface plasmon resonance detection of conformational changes in mRec at T = 25 °C. Repeated injections of increasing [Ca²⁺] over a sensor chip in which mRec (a) and mRec^E121Q (b) were immobilized, corrected after blank subtraction of the reference flow cell that lacked immobilized protein. (c) Results from four independent replicates of the experiment described in (a) and (b) for mRec (○) and mRec^E121Q (□) (mean ± s.d. of the maximum response). The fitting curve is a Hill sigmoid with K^app_D = 1.9 ± 0.4 μM. (d) Average NMR structures of the T (left, PDB entry: 1IKU⁸) and R (right, PDB entry: 1JSA⁶) states of mRec. The protein backbone is displayed in tubes, Ca²⁺ ions (red) and myristoyl group (magenta) are shown in spheres and SAS is represented in transparency by gray (hydrophobic) and blue (hydrophilic) surface. The residue Cys38 target of site-specific thiol–disulfide exchange for protein immobilization on the sensor chip is marked by orange sticks. The radius of gyration is 16.5 Å and 18.3 Å for T and R, respectively. The total SAS for the two conformations are 10 333 Ų and 12 014 Ų, respectively. (e) Results of injections of higher [Ca²⁺] over the same two flow cells, in which mRec (○) and mRec^E121Q (□) were immobilized.

We noticed that for increasing Ca²⁺-concentration in the mM range, the SPR signal significantly increased also for the mRec^E121Q variant to reach similar levels as for mRec (Fig. 1e). This phenomenon was not related to protein–Ca²⁺-binding, since control experiments with other proteins suggest that the effect is independent of what protein coats the chip (results not shown). Instead, it very likely affects the dextran matrix by swelling or shrinking. Therefore, conformational changes triggered by Ca²⁺ should not be monitored by SPR at [Ca²⁺] exceeding 1 mM. The high sensitivity of the dextran form to factors such as ionic strength is in fact supported by previous studies.^2,9

Fitting SPR Ca²⁺-titration data with a sigmoid Hill function led to a half-maximal response at [Ca²⁺] (K^app_D) of 1.9 ± 0.4 μM (Fig. 1c), hence significantly below the known K_D for Ca²⁺-binding to mRec, which is about 17–18 μM.^3,4 The overall equilibrium in the [Ca²⁺]-dependent transition of T to R is the consequence of concerted binding and conformational change events. Therefore, the overall equilibrium constant (K_tot) will result from the product of the pure binding constant (K_A) and the constant of isomerization (K_isom): K_tot= K_AK_isom. If K_tot is assumed to equal 1/K^app_D, as assessed in this study, and K_A is taken from the literature,^3,4 then the isomerization term reasonably results to account for a factor of ∼9 compared to the pure binding term. Moreover, the total free energy change associated with the transition would be −7.8 ± 0.1 kcal mol⁻¹.

It is noteworthy that a consistent value for the total energetics is achieved when the kinetics of the observed transitions are considered rather than the amplitudes after each injection (Fig. 2a). The kinetic analysis was performed on 11 different injections, assuming a simple transition scheme in which the R to T transition was modeled as a first order process, while the T to R transition as a pseudo-first order kinetics. The equilibrium constant obtained from the ratio between the rate constants (k_on = (1.1 ± 0.5) × 10⁴ M⁻¹ s⁻¹ and k_off = (5 ± 2) × 10⁻² s⁻¹) would lead to a free-energy of transition of −7.3 ± 0.5 kcal mol⁻¹, hence consistent with that obtained from the equilibrium analysis. These values for the rate constants significantly differ from those observed in stopped-flow and NMR experiments of Ca²⁺-dissociation from other proteins^10–13 and clearly indicate a slower process, in line with the conceptual distinction between metal-binding and dissociation events and concerted conformational changes. With the method presented here, therefore, we observed the concerted phenomenon and not the metal binding/unbinding alone. Such difference was further confirmed by isothermal titration calorimetry (ITC) experiments to investigate mRec energetics upon Ca²⁺-binding. We employed the same decalcified buffer as well as a number of slightly different settings in terms of protein concentration, [Ca²⁺] provided in each injection and temperature, resulting in quantitatively comparable results (Fig. 2b). The calorimetric curves were fitted by one set of Ca²⁺-binding sites. The apparent K^app_D (obtained as an average of 8 measurements) was found to be 0.9 ± 0.3 μM and the overall thermodynamics parameters were favorable for the transition (Fig. 2b). Although this value obtained by ITC did not perfectly match the determination of K^app_D by SPR, it is significantly lower than the binding constant of 17–18 μM attributed to pure Ca²⁺-binding.^3,4 In this respect, ITC experiments confirm our SPR finding.

Fig. 2 Kinetic analysis of mRec conformational transitions and ITC experiments at T = 25 °C. (a) SPR real-time monitoring of concerted Ca²⁺-binding/T to R conformational transition (upper panel) and Ca²⁺-dissociation/R to T relaxation (lower panel) for 12.7 μM immobilized mRec and 2.5 μM Ca²⁺. Ca²⁺ was injected for 120 s and the relaxation was followed for 120 s after injection, in which only decalcified buffer was flowed over the sensor chip. The first points (4 to 10 s) of both phases are omitted by the fitting due to detection mechanical limitations (see Fig. S1, ESI†) and the time-scale in each panel has been adjusted for the sake of fitting. Data were fitted (black lines) according to a simple concerted model, which separately considers the dissociation/relaxation (k_off) and the association/conformational change processes (k_on). The obtained kinetic parameters for this example are k_off = 4.3 × 10⁻² s⁻¹ and k_on = 19 × 10⁴ M⁻¹ s⁻¹. Mean and standard deviations for kinetic parameters were assessed from 11 different experiments in which [Ca²⁺] varied from 1.58 μM to 14 μM and injection/dissociation times were varied in the 120–240 s range. (b) Conformational changes and Ca²⁺-binding in mRec monitored by ITC. Repetitive injections of 5 μl, 0.5 mM Ca²⁺ into 19.8 μM mRec dissolved in decalcified buffer. The heat response is shown in the upper panel and the heat-per-mole injectant in the lower panel. Data were fitted to a single set of Ca²⁺-binding sites, which in this example led to 1.77 ± 0.07 binding sites, K^app_A = (1.1 ± 0.2) × 10⁵ M⁻¹, ΔH = −3.6 ± 0.2 kcal mol⁻¹ and apparent ΔS = +11 cal mol⁻¹ K⁻¹.

The conditions determined in this work to detect conformational transitions in Ca²⁺-sensor proteins appear to be valid independent of the effective orientation of the proteins embedded in the dextran matrix. Indeed, when performing the same Ca²⁺-titrations on mRec and mRec^E121Q immobilized via non-site-specific amine coupling, the same behaviour as in Fig. 1 was reproduced, though with a roughly one-third reduction of the observed intensity (Fig. S2, ESI†). This suggests that the site-specific homogeneous immobilization of the proteins under investigation enhances the intensity of the phenomenon, but the refractive index changes induced by concerted Ca²⁺-binding/conformational transitions are essentially isotropic. Therefore, conformational changes can be detected even via the p-polarized resonance excited by the commercial SPR systems (Biacore, GE). This seems to be a direct result of the likely heterogeneous orientation of immobilized proteins within the dextran matrix. A different case is for instance the anisotropic immobilization of membrane receptors in lipid bilayers that require both s- and p-polarizations to be fully characterized as in Plasmon Waveguide Resonance spectroscopy.¹⁴

Our results have interesting implications for investigating the biological function of mRec and other Ca²⁺-sensor proteins as well. By considering the binding constant of each individual site in mRec³ (see ESI†) and the concentrations used in our SPR experiments, the occupancy of the high affinity site at a [Ca²⁺] of 1.9 μM (=K^app_D) is 99.5% and that of the second binding site is 22%. Hence, the K^app_D is a snapshot of the sensor protein undergoing a Ca²⁺-triggered conformational transition.

This work was supported by a Research Fellowship from the Alexander von Humboldt Foundation (DD0).

Notes and references

1. T. Christopeit, T. Gossas and U. H. Danielson, Anal. Biochem., 2009, 391, 39–44.
2. D. J. Winzor, Anal. Biochem., 2003, 318, 1–12.
3. J. B. Ames, T. Porumb, T. Tanaka, M. Ikura and L. Stryer, J. Biol. Chem., 1995, 270, 4526–4533.
4. I. I. Senin, T. Fischer, K. E. Komolov, D. V. Zinchenko, P. P. Philippov and K. W. Koch, J. Biol. Chem., 2002, 277, 50365–50372.
5. E. N. Pugh, Jr. and T. D. Lamb, Phototransduction in Vertebrate Rods and Cones: Molecular Mechanisms of Amplification, Recovery and Light Adaptation, Elsevier Science B.V., 2000.
6. J. B. Ames, R. Ishima, T. Tanaka, J. I. Gordon, L. Stryer and M. Ikura, Nature, 1997, 389, 198–202.
7. V. A. Klenchin, P. D. Calvert and M. D. Bownds, J. Biol. Chem., 1995, 270, 16147–16152.
8. T. Tanaka, J. B. Ames, T. S. Harvey, L. Stryer and M. Ikura, Nature, 1995, 376, 444–447.
9. S. Paynter and D. A. Russell, Anal. Biochem., 2002, 309, 85–95.
10. W. Dong, S. S. Rosenfeld, C. K. Wang, A. M. Gordon and H. C. Cheung, J. Biol. Chem., 1996, 271, 688–694.
11. T. Drakenberg, T. Andersson, S. Forsen and T. Wieloch, Biochemistry, 1984, 23, 2387–2392.
12. Y. Ogawa and M. Tanokura, J. Biochem., 1986, 99, 81–89.
13. I. Sokal, A. E. Otto-Bruc, I. Surgucheva, C. L. Verlinde, C. K. Wang, W. Baehr and K. Palczewski, J. Biol. Chem., 1999, 274, 19829–19837.
14. V. J. Hruby and G. Tollin, Curr. Opin. Pharmacol., 2007, 7, 507–514.

---

Footnote

† Electronic supplementary information (ESI) available: Supporting methods and materials, Figure S1, Figure S2 and supporting references. See DOI: 10.1039/c0cc02086a

---