Design and mechanistic insight into ultrafast calcium indicators for monitoring intracellular calcium dynamics

Abstract

Calmodulin-based genetically encoded fluorescent calcium indicators (GCaMP-s) are powerful tools of imaging calcium dynamics from cells to freely moving animals. High affinity indicators with slow kinetics however distort the temporal profile of calcium transients. Here we report the development of reduced affinity ultrafast variants of GCaMP6s and GCaMP6f. We hypothesized that GCaMP-s have a common kinetic mechanism with a rate-limiting process in the interaction of the RS20 peptide and calcium-calmodulin. Therefore we targeted specific residues in the binding interface by rational design generating improved indicators with GCaMP6f_u displaying fluorescence rise and decay times (t_1/2) of 1 and 3 ms (37 °C) in vitro, 9 and 22-fold faster than GCaMP6f respectively. In HEK293T cells, GCaMP6f_u revealed a 4-fold faster decay of ATP-evoked intracellular calcium transients than GCaMP6f. Stimulation of hippocampal CA1 pyramidal neurons with five action potentials fired at 100 Hz resulted in a single dendritic calcium transient with a 2-fold faster rise and 7-fold faster decay time (t_1/2 of 40 ms) than GCaMP6f, indicating that tracking high frequency action potentials may be limited by calcium dynamics. We propose that the design strategy used for generating GCaMP6f_u is applicable for the acceleration of the response kinetics of GCaMP-type calcium indicators.

Figure 1. Targeted mutagenesis and biophysical characterisation of GCaMP6f and GCaMP6f_u.

(a) Crystal structure of monomeric GCaMP6m in a Ca²⁺-bound form with cpEGFP (green), Ca²⁺ ions (yellow), CaM (blue) and the RS20 peptide (light brown) (adapted from Ding et al., PDB 3WLD). The positions of the mutated amino acid residues in the CaM EF-hand Ca²⁺-binding sites and in the RS20 peptide are highlighted. Equilibrium Ca²⁺ titrations for GCaMP6f (•) and GCaMP6f_u (), (b) at 20 °C (c) at 37 °C. Fluorescence changes are normalised to F₀ of 0 and F_max of 1 and fitted to the Hill equation. Fitted curves are represented by solid lines overlaying the data points. Ca²⁺ dissociation kinetics of (d) GCaMP6f and (e) GCaMP6f_u at 20 °C () and 37 °C (—); Experimental data are overlaid by fitted curves using parameters for the kinetic model for GCaMPs (Supplementary Table S3). (f) Arrhenius plots of the observed rates for Ca²⁺ dissociation of GCaMP6f (•) and GCaMP6f_u ().

Figure 2

Ca²⁺ association kinetics of GCaMP6f at (a) 20 °C and (b) 37 °C; (c) GCaMP6f_u at 20 °C. Left hand panels show stopped-flow records at the specified final [Ca²⁺] values. Fluorescence changes are normalised to F₀ of 0 and maximum of 1. Experimental data (dotted lines) are overlayed with fitted curves (solid lines) generated using the parameters shown in Supplementary Table S3 fitted to the kinetic model for GCaMP-s. Right hand panels show plots of the [Ca²⁺] dependence of the observed rate(s) (k_obs). GCaMP6f has biphasic association kinetics at 20 °C; the fast phase () corresponds to 30% and the slow phase (•) to 70% of the fluorescence amplitude. At 37 °C, GCaMP6f Ca²⁺ association kinetics are monophasic with the fluorescence response shown only by the slow phase. (d) Stopped-flow record of GCaMP6f_u Ca²⁺ association at 10 μM final [Ca²⁺] measured at 30 °C, fitted to the model. (e) Arrhenius plots of the observed rates for Ca²⁺ association of GCaMP6f fast phase (), slow phase (•), GCaMP6f_u (). For GCaMP6f the amplitude of the fast phase diminishes with increasing temperature: relative amplitudes are at 20 °C, fast (0.3) slow (0.7); at 25 °C, fast (0.2) slow (0.8); at 30 °C, fast (0.1) slow (0.9) and at 37 °C, fast (0) slow (1.0).

Figure 3. Schematic model of the kinetic mechanism of GCaMPs.

General model for the mechanism of Ca²⁺-induced fluorescence of GCaMP-type probes depicting Ca²⁺ binding to the CaM N- and C-lobe followed by peptide binding and isomerisation leading to fluorescent states. Fluorescence is proposed to derive from the equilibrium of two states: in the first, Ca²⁺-bound CaM N-lobe is with the P1 CaM binding site of RS20; in the second, both the N- and C-lobes of CaM are Ca²⁺-bound binding to both the P1 and P2 sites of the peptide. Formation of the fluorescent complexes involves essentially irreversible processes. Peptide release is triggered by Ca²⁺ sequestration returning the complex to the apo-state. This general model for GCaMPs is applicable to EF-hand mutants by omitting the appropriate Ca²⁺ binding steps e.g. eq. 1 for the EF-1 mutant, eq. 2 for EF-2 and so on.

Figure 4. Ca²⁺ response of GCaMP6f and GCaMP6f_u in ATP-stimulated HEK293T cells and post-synaptic CA1 neurons in hippocampal slices.

(a) Ca²⁺ transients were triggered by exposure of HEK293T cells to 100 μM ATP. Time courses of response of GCaMP6f and GCaMP6f_u are shown. t_1/2 for GCaMP6f_u 2.3 ± 0.04 s was significantly different from that for GCaMP6f (9.4 ± 0.2 s). (b) Ca²⁺ response kinetics in post-synaptic CA1 neurons in hippocampal slices of GCaMP6f (28 °C) to stimulation by 5 action potentials (AP-s). Grey shaded areas indicate the duration of stimulation. GCaMP6f stimulated at (3 cells, n number of recordings in brackets) 10 Hz (n = 28), 20 Hz (n = 28), 40 Hz (n = 29), 50 Hz (n = 27), 75 Hz (n = 30) and 100 Hz (n = 19); (c) GCaMP6f_u stimulated (4 cells, n number of recordings in brackets) at 10 Hz (n = 29), 20 Hz (n = 30), 40 Hz (n = 29), 50 Hz (n = 29), 75 Hz (n = 30) and 100 Hz (n = 26). The achieved maximum ΔF/F₀ values are plotted against time. Inset images: representative images baseline expression of GCaMP6f and GCaMP6f_u in CA1 pyramidal neurons with white line in the position of the line scan.

Figure 5. Comparison of fast GCaMP probes (GCaMP3_fast and GCaMP6f_u) and their parent variants (GCaMP3 and GCaMP6f).

(a) Position of the mutations that gave rise to GCaMP3_fast and GCaMP6f_u variants, relative to GCaMP3. (b) Crystal structure of monomeric GCaMP6m in a Ca²⁺-bound form with cpEGFP (green), Ca²⁺ ions (yellow), CaM (blue) and the RS20 peptide (light brown) (adapted from Ding et al., PDB 3WLD). The amino acid residues highlighted in red are those that generated GCaMP6f_u relative to GCaMP3. Schematic representation of (c) Ca²⁺-bound GCaMP3/GCaMP6f showing “open” conformation of EF-hand 3 (helices V and VI); (d) Ca²⁺-bound GCaMP3_fast/GCaMP6f_u showing “closed” conformation of disabled EF-hand 3 (helices V and VI).