Improved calcium sensor GCaMP-X overcomes the calcium channel perturbations induced by the calmodulin in GCaMP

Abstract

GCaMP, one popular type of genetically-encoded Ca²⁺ indicator, has been associated with various side-effects. Here we unveil the intrinsic problem prevailing over different versions and applications, showing that GCaMP containing CaM (calmodulin) interferes with both gating and signaling of L-type calcium channels (Ca_V1). GCaMP acts as an impaired apoCaM and Ca²⁺/CaM, both critical to Ca_V1, which disrupts Ca²⁺ dynamics and gene expression. We then design and implement GCaMP-X, by incorporating an extra apoCaM-binding motif, effectively protecting Ca_V1-dependent excitation-transcription coupling from perturbations. GCaMP-X resolves the problems of detrimental nuclear accumulation, acute and chronic Ca²⁺ dysregulation, and aberrant transcription signaling and cell morphogenesis, while still demonstrating excellent Ca²⁺-sensing characteristics partly inherited from GCaMP. In summary, CaM/Ca_V1 gating and signaling mechanisms are elucidated for GCaMP side-effects, while allowing the development of GCaMP-X to appropriately monitor cytosolic, submembrane or nuclear Ca²⁺, which is also expected to guide the future design of CaM-based molecular tools.

Fig. 1

Side-effects of GCaMP on cortical neurons. a Abnormal nuclear accumulation of GCaMP. Cultured cortical neurons infected with GCaMP6 (AAV-Syn-GCaMP6f) virus were examined with confocal live-cell imaging. b Analyses indexed with nuclear/cytosolic (N/C) fluorescence ratio indicate time-dependent nuclear accumulation of GCaMP6, in contrast to stable GFP distribution. c GCaMP6 caused apoptosis. For cortical neurons infected with AAV-Syn-EGFP and AAV-Syn-GCaMP6f (green), tracing images of dendritic morphology (black and white) and confocal fluorescence images of Annexin V-kFluor594 (red) are shown. Arrows are to identify apoptotic neurons where both green and red fluorescence are present and the percentage of such neurons (per view) are counted (right, number of views in parentheses). d Subcellular distributions of YFP, GCaMP3 or NLS-GCaMP3-NLS overexpressed in cortical neurons. CFP fluorescence (NLS-CFP-NLS, upper) indicates the nucleus (blue) in cortical neurons. GFP images (lower row) of neurons expressing GCaMP3 (2 days after cDNA transfection) could be categorized into two major subgroups of interest: nuclear-excluded (N/C ratio < 0.6) and nuclear-filled (N/C ratio > 1.0), with the latter mimicked by neurons expressing NLS-GCaMP3-NLS. Such criteria (N/C ratio < 0.6 and >1) were applied throughout this study (unless indicated otherwise). e Based on the above criteria neurons of nuclear-filled group and nuclear-excluded group accounted for ~10% and ~50% of the total number of GCaMP3-expressing neurons, respectively (5 experiments). f Representative images tracing neurite morphology for cortical neurons from different subgroups. g Correlations between N/C ratio of GCaMP3 and neurite outgrowth. The grey line/area represents the total neurite length per neuron (control group). The eclipse enclosing most neurons contains two major areas representing the subgroups of nuclear-filled (cyan) and nuclear-excluded (pink). Nuclear GCaMP accumulation (N/C ratio) and neurite outgrowth (neurite length per neuron) are highly correlated (the correlation coefficient is −0.7). h, i Additional experiments and analyses for neurite morphology. Neurite outgrowth was quantified for the four subgroups by neurite length (h) and Sholl analysis (i), with the total number of cells in parentheses. Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01, and p < 0.001) were calculated when applicable

Fig. 2

GCaMP interferes with Ca_V1.3 gating. a Schematic summary of GCaMP series. Upgrades of GCaMP (from GCaMP3 to GCaMP5 and GCaMP6) were achieved by mutations of the EGFP and CaM domains at the sites indicated by vertical letters (GCaMP3 vs. GCaMP), or by horizontal letters (GCaMP5G/6f/6m vs. GCaMP3). Details see Supplementary Fig. 1. b Effects on I_Ca were examined for neurons infected with AAV-Syn-GCaMP6f. Representative traces of Ca²⁺ current (left), S_Ca and J_Ca analyses (right) for native Ca_V1.3 in cortical neurons expressing GCaMP6. Neurons were treated with a cocktail recipe to isolate Ca_V1 current (mostly Ca_V1.3) and recorded at the membrane potential (V) of −10 mV. S_Ca (quantified as 1−I_Ca,50 /I_{Ca, peak}, where I_Ca,50 and I_{Ca, peak} represent the currents measured at 50 ms and the instantaneous peak, respectively) and J_Ca (pA/pF, the current density of I_{Ca, peak}) serve as the indices for CDI and VGA respectively. c Effects of GCaMP3 on recombinant α_1DL. Representative Ca²⁺ current traces were compared for I_Ca recorded from HEK293 cells expressing long variant α_1DL alone (left), or with GCaMP3 (right) at −10 mV. Ba²⁺ currents (rescaled) and Ca²⁺ currents (I_Ca) were shown as grey and red traces, respectively, with scale bars indicative of I_Ca amplitudes. CDI (S_Ca) and VGA (J_Ca) profiles at different membrane potentials are compared between α_1DL control and α_1DL overexpressed with GCaMP3 (differences highlighted by orange areas). d Effects of GCaMP3 on recombinant α_1DS alone (left), or with GCaMP3 (right), in a similar fashion to c. Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01 and, ***p < 0.001) were calculated when applicable, and n.s. denotes “not significant”

Fig. 3

GCaMP perturbs Ca_V1-dependent E–T coupling. a GCaMP3 effects on CREB signaling. Cortical neurons (DIV 5) were transfected with YFP, GCaMP3 or NLS-GCaMP3-NLS for 2 days, then stained with pCREB antibodies under basal conditions. Green fluorescence represents the distributions of protein expression (upper) and red staining represents the levels of pCREB signals (lower), for all the four groups of neurons: YFP control, GCaMP3 nuclear excluded, GCaMP3 nuclear filled, and NLS-GCaMP3-NLS. b High correlation between subcellular localization of GCaMP3 (N/C ratio) and pCREB signals in cortical neurons expressing GCaMP3 (correlation coefficient = −0.8). Shades of areas indicate the neurons in the nuclear-excluded group (pink) and the nuclear-filled group (cyan). c Statistical summary of basal pCREB intensities. All neurons were normalized by the average value of pCREB fluorescence intensities calculated from the control group. d–f Cytonuclear translocations of CaM and GCaMP3 by confocal imaging. Endogenous CaM of cortical neurons translocated into nucleus under 5 min of 40 mM [K⁺]_o stimulation, and recovered to the resting state when washed back by 2 h of 5 mM [K⁺]_o (d). Representative images of fixed neurons for Hoechst (blue, indicating nucleus), YFP (indicating soma) and endogenous CaM (red, staining of CaM antibodies) at three different time points are shown in order (before 40 mM [K⁺]_o, at the end of 5 min of 40 mM [K⁺]_o stimulation, after washed back to 5 mM [K⁺]_o for 2 h). 1 μM TTX was added throughout the experimental protocol in order to examine subthreshold activities of Ca_V1 channels. For comparison, neurons expressing TagRFP-GCaMP3 were similarly excited and monitored, during which localizations of GCaMP3 were confirmed by fluorescent TagRFP (e). RFP fluorescence at three representative time points (resting, stimulation, and wash-out) was analyzed with the index of N/C ratio and compared (paired Student’s t-test) for GCaMP3 localization, in addition to endogenous CaM (f). Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01, and p < 0.001) were calculated when applicable, and n.s. denotes “not significant”

Fig. 4

Design principles and basic validations of new GCaMP-X sensors. a Design principles of GCaMP-X. CaM binding motif (CBM) was fused onto N-terminus of GCaMP to tightly bind apoCaM at rest but with relatively low affinity (e.g., much lower than the M13 motif) to Ca²⁺/CaM, as the central strategy for GCaMP-X design. Also, additional tags of NES (nuclear export signal), NLS (nuclear localization signal) and MTS (membrane targeting signal) motifs could be appended onto C-terminus (or N-terminus), to control specific subcellular localization of GCaMP. Accordingly, a series of GCaMP-X sensors were developed, in addition to GCaMP-X_O without any tag, including GCaMP-X_C, GCaMP-X_N, and GCaMP-X_M, tagged with NES, NLS, and MTS respectively. b, c Design validation with GCaMP3 and recombinant Ca_V1.3 channels. Neither CDI nor VGA of α_1DL channels (b) and α_1DS channels (c) were altered by GCaMP3-X_O or GCaMP3-X_C overexpressed in HEK293 cells, as demonstrated by I_Ca exemplars (top) and voltage-dependent CDI (middle) and VGA (bottom) profiles. d Validation of GCaMP6m-X_C with neuronal Ca_V1.3 channels. Effects of GCaMP6-X_C on α_1DL gating were examined in cortical neurons infected with AAV-Syn-GCaMP6m-X_C, indexed with S_Ca and J_Ca. Dotted lines represent GCaMP6f profiles (Fig. 2b). e, f GCaMP-X eliminated perturbations on subthreshold Ca²⁺ dynamics. For cortical neurons infected with AAV-Syn-GCaMP6m or AAV-Syn-GCaMP6m-X_C, confocal fluorescent images depict Ca²⁺ dynamics in response to extracellular stimuli of 20 mM [K⁺]_o, indexed by normalized fluorescence changes (ΔF/F₀). Exemplar images (left), averaged curves of ΔF/F₀ (middle) and statistical summary (right) are shown for GCaMP6m and GCaMP6m-X_C. Particularly, peak ΔF/F₀ and total calcium influx (rightmost, area integral within 300 s, in the units of ΔF/F₀·s) were evaluated and compared (e). To confirm, similar experiments and analyses were performed with Fura-2 ratiometric Ca²⁺ imaging, indexed with fluorescence emission ratio F₃₄₀/F₃₈₀ achieved from 340 nm and 380 nm excitation (f). Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01, and ***p < 0.001) were calculated when applicable, and n.s. denotes “not significant”

Fig. 5

Additional validations of GCaMP-X. a Nuclear accumulation was substantially relieved for cortical neurons transfected with GCaMP-X. N/C fluorescence ratios indicative of GCaMP distributions in cortical neurons (DIV 5) were quantified 2 days after being transfected with GCaMP3, GCaMP3-X_O or GCaMP3-X_C. b GCaMP-X sensors no longer perturbed pCREB signals. For cortical neurons transfected with YFP, GCaMP3-X_O or GCaMP3-X_C, fluorescence of pCREB immunostaining (left) was normalized and compared (right). Dotted lines represent the mean values from nuclear-excluded GCaMP3 (pink) or nuclear-filled GCaMP3 (cyan) neurons, respectively (adopted from Fig. 3c). c Dysregulations of neurite outgrowth were diminished with GCaMP-X. Based on neurite tracing images (left), Sholl analysis (middle) and measurement of neurite length (right) were performed for cortical neurons transfected with YFP, GCaMP3-X_O or GCaMP3-X_C. Similar to b, dotted lines depict mean values of neurite length for nuclear-excluded GCaMP3 (pink) or nuclear-filled GCaMP3 (cyan) neurons (adopted from Fig. 1h). d Problematic nuclear accumulations were strongly attenuated for long-term viral expression of GCaMP6m-X_C. Fluorescence indicative of sensor localization was mostly restricted to the cytosol of neurons infected with AAV-Syn-GCaMP6m-X_C (GCaMP6-X_C), as shown by representative images at different time points (upper panel). The temporal trend of N/C ratio for GCaMP6-X_C (red) is distinct from that for GCaMP6f (black, GCaMP6 data adopted from Fig. 1b). e Apoptotic damages arising from GCaMP were strongly attenuated in GCaMP-X_C expressing neurons. Fluorescence signals of apoptosis (red) were detected by Annexin V assay, to correlate with the expression of sensors (green). AAV-Syn-GCaMP6m-X_C (GCaMP6-X_C), EGFP and GCaMP6f (data adopted from Fig. 1c) were compared in a similar fashion to Fig. 1c. Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01, and ***p < 0.001) were calculated when applicable, and n.s. denotes “not significant”

Fig. 6

Schematic summary and sensor performance for GCaMP and GCaMP-X. a GCaMP defects and advantages of GCaMP-X. In neurons, GCaMP sensors behave as CaM-like proteins perturbing Ca_V1 gating and distorting Ca²⁺ dynamics (1). The abnormal enhancement of Ca_V1 by cytosolic GCaMP underlies overgrown neurites. Different from endogenous CaM, GCaMP itself could no longer properly translocate into the nucleus in response to membrane excitation, which also dominant-negatively impairs the acute mobility of endogenous CaM. Once present in the nucleus (by accumulation or NLS), nuclear GCaMP perturbs transcription signaling and gene expression potentially through its aberrant CaM motif, impairing general health of neurons including Ca²⁺ signals (reflected as abnormal sensor readouts) and neurite outgrowth (2). The new GCaMP-X sensors are designed with apoCaM protection and explicit localization (e.g., GCaMP-X_C for cytosolic Ca²⁺) to eliminate GCaMP perturbations on Ca_V1 gating (3) and signaling (4). b Sensor performance of GCaMP-X_C validated by single-AP like Ca²⁺ transients in HEK293 cells. Briefly, glass electrodes formed the Giga-Ohm seal with cell membrane (resting); then 3 ms ZAP for break-in generated a rapid Ca²⁺ influx due to transient membrane rupture, immediately followed by a fast decay arising from strong Ca²⁺ chelators of 10 mM BAPTA included in the pipette solution. Key indices of peak ΔF/F₀, rise time t_r, decay time t_d and SNR were compared between GCaMP6m and GCaMP6m-X_C. Epi-fluorescence images were acquired at the sampling rate of 100 Hz or higher. c–e Sensor performance of GCaMP-X_C validated with mechanosensing outer hair cells. Cells were transfected with GCaMP6m or GCaMP6m-X_C at P1 by electroporation, and cultured for 1 day (DIV 1) or 3 days (DIV 3). Fluid-jet pulses of 100, 300 or 500 ms were applied to cells (c). Representative images, normalized fluorescence changes (ΔF/F₀) indicative of Ca²⁺ dynamics and statistical summary of peak ΔF/F₀ were compared between DIV 1 (d) and DIV 3 (e) cells expressing GCaMP6m or GCaM6m-X_C. Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01, and ***p < 0.001) were calculated when applicable, and n.s. denotes “not significant”

Fig. 7

Characterizations and validations of GCaMP-X_N targeting nuclear Ca²⁺. a Design of GCaMP-X_N. Based on the design principle of GCaMP-X, similar to GCaMP-X_C, CBM was fused into N-terminus of GCaMP; and a nuclear localization signal (NLS) was tagged onto C-terminus of GCaMP. b, c Basic validations of GCaMP3-X_N with pCREB signals and neurite outgrowth. Representative images of pCREB immunostaining (b, left), statistical summary of pCREB intensities (b, right), tracing of neurite morphology (c, upper), Sholl analysis and statistical summary of neurite length (c, lower) were compared among neurons transfected with YFP, NLS-GCaMP3-NLS or GCaMP3-X_N. d Different Ca²⁺ dynamics resulted from neurons expressing NLS-GCaMP3-NLS or GCaMP3-X_N. Confocal images representing Ca²⁺ fluorescence at three phases: before, during and at the end of extracellular stimuli of 40 mM [K⁺]_o (upper). Ca²⁺ response (ΔF/F₀) (lower left) and its rising speed (ΔF/F₀·s⁻¹, normalized change of fluorescence per second, lower right) were averaged from multiple neurons (number indicated within parentheses), to compare NLS-GCaMP3-NLS with GCaMP3-X_N. e, f Simultaneous monitoring of cytosolic and nuclear Ca²⁺ dynamics. Representative confocal images indicative of Ca²⁺ fluorescence (upper in green) and time-dependent responses (ΔF/F₀, lower) are to compare Ca²⁺ dynamics in the cytosol vs. the nucleus of the same neuron upon 40 mM [K⁺]_o stimuli. Cytosolic Ca²⁺ was monitored by GCaMP3-X_C for both cases, whereas nuclear Ca²⁺ was either by GCaMP3-X_N (e) or by NLS-GCaMP3-NLS (f). Neurons were loaded with Hoechst 33342 to label the nuclei of neurons (upper, blue). Standard error of the mean (S.E.M.) and Student’s t-test (two-tailed unpaired with criteria of significance: *p < 0.05; **p < 0.01, and ***p < 0.001) were calculated when applicable, and n.s. denotes “not significant”