Using Q5^® Site-Directed Mutagenesis Kit (New England Biolabs), a range of mutations of jGCaMP8f pET30b were generated. Primers (5′-3′) used for mutagenesis were:

After mutagenesis, DNA sequences were sent to Genewiz for checking.

jGCaMP8f and variant proteins which contained His-tag were overexpressed in E. coli B21 Gold cells using 0.5 mM isopropyl thio-β-D-galactoside (IPTG) then purified on a NiNTA column (QIAGEN, ÄKTA Purifier, GE Healthcare) at 4 °C as described previously (Helassa et al., 2015). Protein purity was checked using SDS-PAGE then stored at −80°C.

Protein concentration was calculated from the absorption spectra of the proteins at wavelength 280 nm using the molar extinction coefficient of 26,360 M^–1 cm^–1 for jGCaMP8f and all variants calculated using the ProtParam tool (Expasy).

Fluorescence emission spectra of 0.5–1 μM jGCaMP8f or variant proteins in assay buffer (50 mM HEPES-K⁺, 100 mM KCl and 5 mM EGTA in the presence or absence of 2 mM MgCl₂, pH 7.5) at 20°C were recorded using a Fluorolog-3 spectrofluorometer (Horiba^©) at 492 nm excitation and 500 – 550 nm emission wavelengths. Subsequently, a saturated amount of CaCl₂ was added to obtain the maximum fluorescence emission intensity. Data were normalized and plotted using GraphPad Prism 9 software.

The concentration of jGCaMP8f proteins was adjusted such that the absorbance at the excitation wavelength (492 nm) was between 0.001 and 0.1. A series of dilutions was prepared in a buffered solution (50 mM HEPES, pH 7.5, 100 mM KCl, 2 mM MgCl₂ with either 5 mM EGTA, or 1 mM CaCl₂) and the fluorescence spectra were recorded on a Fluorolog-3 (Horiba^©). GCaMP6f quantum yield measured in Ca²⁺ -saturated buffer was used as a reference (Φ_+Ca2+ = 0.59) (Chen et al., 2013). Data were plotted as integrated fluorescence intensity as a function of absorbance and fitted to linear gradient, S. Quantum yields were obtained using the following equation: Φ_protein = Φ_GCaMP6f × (S_protein/S_GCaMP6f).

To determine the apparent pKa for each jGCaMP8f protein, a series of buffers were prepared. Depending on their respective pH buffering range, an appropriate buffer was used for the measurements (MES for pH 6–6.5, HEPES for pH 7–8, TRIS for pH 8.5–9 and CAPS for pH 10). The pH titrations were performed by recording fluorescence spectra in Ca²⁺ -free (50 mM buffer, 100 mM KCl, 2 mM MgCl₂, 2 mM BAPTA) or Ca²⁺-saturated (50 mM buffer, 100 mM KCl, 2 mM MgCl₂,1 mM CaCl₂) measuring 1 μM protein at 0.5 pH unit intervals at 20°C (Fluorolog-3, Horiba^©). Excitation wavelength was set at 492 nm and emission in the 500–600 nm range was recorded. BAPTA was chosen as Ca²⁺ chelator because of its stable affinity for Ca²⁺ over the pH range.

Ca²⁺ association and dissociation rates were measured using a TgK Scientific KinetAsyst™ double-mixing stopped-flow apparatus in single mixing mode at 20°C. The fluorescence was excited at 492 nm and the emission was collected using a 530 nm cut-off filter. For Ca²⁺ association, 500 nM jGCaMP8f or variant proteins in 50 mM HEPES-K⁺, 100 mM KCl, 2 mM MgCl₂, 5 mM EGTA, pH 7.5 at 20°C were rapidly mixed with increasing concentrations of Ca²⁺ (CaCl₂ in 50 mM HEPES-K⁺, 100 mM KCl, 2 mM MgCl₂, 5 mM EGTA, pH 7.5 at 20°C), calculated free [Ca²⁺] ranged from 108 nM to 252 μM in the mixing chamber). For Ca²⁺ dissociation, 500 nM jGCaMP8f or variant proteins in 50 mM HEPES-K⁺, 100 mM KCl, 2 mM MgCl₂, 200 μM CaCl₂, pH 7.5 at 20°C were rapidly mixed with 50 mM HEPES-K⁺, 100 mM KCl, 2 mM MgCl₂, 10 mM EGTA, pH 7.5 at 20°C. Experiments were repeated at least three times. Kinetic records shown are the average of at least four traces. Averaged data were fitted to a single or double exponential using KinetAsyst software (TgK Scientific) to obtain association or dissociation rates. The error quoted is the standard error of the fit. For the rate plots, the mean values obtained from at least three experiments are shown, the error bars represent S.E.M.

Free Ca²⁺ concentration [Ca²⁺] was calculated using the two-chelators Maxchelator program (Schoenmakers et al., 1992). Fluo-3 (Biotium) with known apparent dissociation constant for calcium (K_d, the Ca²⁺ concentration at which half-maximum fluorescence intensity is achieved, 450 nM) was used to verify the calculated [Ca²⁺].

Fluorescence emission spectra were recorded using a Fluorolog-3 spectrofluorometer (Horiba^©) at specific excitation and emission wavelengths, 20°C. When bound to Ca²⁺, the Fluo-3 excitation peak is at 506 nm and the emission peak is at 526 nm; the respective wavelengths for jGCaMP8f and all variant proteins are 492 nm and 512 nm. One μM jGCaMP8f or variant proteins in assay buffer (50 mM HEPES-K⁺, 100 mM KCl, 2 mM MgCl₂ and 5 mM EGTA, pH 7.5 at 20°C) were titrated by stepwise addition of a solution of 500 mM CaCl₂ until no further changes in fluorescence emission intensity were observed. Experiments were done in triplicates. Data was corrected for dilution, normalized and expressed as mean ± S.E.M. K_d and Hill coefficient for indication of cooperativity (n) were obtained by fitting the data to the Hill equation using GraphPad Prism 9 software.

HEK293T cells were maintained in Dulbecco’s Modified Eagle’s Media (DMEM, high glucose, from Sigma D6546) supplemented with 10% fetal bovine serum and 4 mM L-glutamine, in 75 cm² flasks. At 90% confluence, cells were grown in poly-D-lysine coated glass-bottomed dishes (MatTek Corporation) at 300,000 cells per dish for 24 h. Cryopreserved pooled primary human umbilical vein endothelial cells (HUVEC) were purchased from PromoCell GmbH (Heidelberg Germany). HUVEC were grown in a human growth medium (HGM) which comprised media M199 supplemented with 20% FCS. A total of 50 μg/mL of the antibiotic gentamicin along with ECGS (endothelial cell growth supplement; 30 μg/mL) and heparin (10 U/mL) were added to the media. HUVEC were grown on 24 well trays or 9 mm glass coverslips that had been pre-treated with porcine skin gelatine (1% w/v) for 30 min at 37°C. Cells were cultured for 4 days in a 5% CO₂ incubator at 37 °C until fully confluent as assessed by visual examination by light microscopy.

Transfection of HEK293T was accomplished using Lipofectamine™ 2000 (Invitrogen): 2.5 μg of plasmid DNA was combined with 5 μL Lipofectamine™ 2000 reagent in Opti-MEM, this was then added to the 2 mL cell culture medium per dish. Transfection of HUVEC was by Nucleofection using an Amaxa Nucleofector 2b device and the HUVEC old transfection reagent (Lonza, Slough, UK) according to manufactures. Experiments were performed 24 h after transfection.

Transfected HEK293T cells or HUVEC were transferred to a physiological saline solution (145 mM NaCl, 5 mM KCl, 1.2 mM CaCl₂, 1 mM MgCl₂, 1 mM NaH₂PO₄, 10 mM glucose, 20 mM Na⁺-HEPES pH 7.4) and placed on the stage of an inverted Olympus IX71 fluorescence microscope. The microscope was equipped with an Olympus UPLSAPO × 100 1.40 NA objective and an OptoLED light source (Cairn Research, Faversham, UK) for EGFP (470 ± 20 nm) fluorescence. Excitation light was directed to the specimen via a 500DCXR dichroic mirror (Chroma Rockingham, USA) and emitted light collected through a 535 ± 30 nm emission filter onto an Ixon3 EMCCD camera (Andor, Belfast, UK) operated in frame transfer mode at full gain and cooled to −68 °C. Images were acquired at 20 frames/s using Winfluor software.¹ Intracellular Ca²⁺ release in HEK293T cells was induced by ATP. ATP was administered as a bolus of 50 μL diluted in imaging buffer to give a final concentration of 50 μM in HEK293T cells and 2 μM for HUVEC. The higher sensitivity of HUVEC to ATP is reflected in the different ATP concentrations used. Full dynamic range was measured following the addition of 10 μM ionomycin. The fluorescence changes during ATP stimulation were analyzed using Winflour software. A square region of interest (ROI) was placed over a peripheral cytoplasmic region of each cell. The mean background fluorescence was measured by placing an ROI with the same dimensions at a point at which no cell was present. GraphPad Prism 9 was used to plot and fit data. Time-to-peak was determined by the time in which the rise amplitude increased from 10 to 90%. Decay t_1/2 was determined from an exponential function fitted for individual records when possible, or by determining the time required for the signal dropping to half the amplitude. Data are expressed as mean ± S.E.M. For further statistical analysis one-way ANOVA tests were carried out with post-hoc Dunnett’s multiple comparison test. Different number of data points could be extracted for the different parameters due to the variability of individual cell responses. For example, following a clear rising phase, the decay phase may include oscillations or plateaus that complicate fitting and analysis. In some cells when ionomycin was added, no clear response was seen.

In creating jGCaMP8f, the RS20 peptide used in most previous GCaMP designs was substituted for the CaM binding domain of nitric oxide synthase (eNOS) (Zhang et al., 2023). The relative positions of the Ca²⁺/CaM–peptide complex and cpEGFP are revealed in an adaptation of the crystal structure of jGCaMP8f (PDB ID: 7ST4) (Figure 1A).

We generated variants by introducing mutations in the CaM EF-hands and selected sites in the eNOS peptide (Figure 1B). Mutation at the EF-hands gave the D294A (EF-1), D330A (EF-2), D367A (EF-3), and D403A (EF-4) variants. In the eNOS peptide we made the A18S, L27A and in CaM the F366H and F366A mutations.

Mutations in the Ca²⁺ binding sites of CaM and/or in the eNOS peptide of jGCaMP8f will weaken the binding affinity resulting in faster calcium off-kinetics [L27A, A18S, D294A (EF-1), D330A (EF-2), D367A (EF-3), D403A (EF-4)] (Figure 1B). In designing the D294A (EF-1), D330A (EF-2), D367A (EF-3), and D403A (EF-4) variants, mutation of the first Asp residue of the EF-hand, known to disable Ca²⁺ binding (Jama et al., 2011), was expected to increase the Ca²⁺ off-rate. Likewise, the L27A and A18S mutations in the eNOS peptide were made to destabilize hydrophobic interactions with Ca²⁺/CaM with the expectation of increasing the off-rate, albeit to a lesser extent.

In turn, mutations that stabilize the Ca²⁺/CaM – eNOS peptide complex may increase the brightness and fluorescence dynamic range (F366A, F366H). The F366 residue in CaM is adjacent to EF-3 and is involved in hydrophobic interactions with the peptide. We designed two mutations, F366H and F366A, with the view of reducing hydrophobicity and size to increase dynamic range.

We were interested to see how these mutations affected the brightness and dynamic range, apparent Ca²⁺ affinity and kinetic responses of the variants in comparison with jGCaMP8f. We furthermore tested the response of jGCaMP8f, its novel variants and previously generated fast GCaMP variants to ATP stimulation in HEK293T cells and HUVEC.

jGCaMP8f is reported to have a 79-fold fluorescence enhancement upon Ca²⁺ binding and an off-rate of 37 s^–1 at 20°C (Zhang et al., 2023). Surprisingly, we measured an only sevenfold increase for jGCaMP8f (Figure 2A). Puzzled by the large discrepancy, we checked the experimental conditions. We found that the very high dynamic range was measured in the absence of Mg²⁺ and at pH 7.2 as opposed to our buffer that contains 2 mM Mg²⁺ and is adjusted to pH 7.5. Remeasuring the dynamic range of jGCaMP8f in the absence of Mg²⁺ (pH 7.5), we got a remarkably different result: the fluorescence of jGCaMP8f increased 31-fold upon Ca²⁺ binding. We then compared fluorescence intensities with or without added Mg²⁺ in the presence of EGTA and in the presence of Ca²⁺ (Figure 2A). We found equal intensities in Ca²⁺ with or without Mg²⁺. In the absence of Ca²⁺, however, the addition of 2 mM Mg²⁺ increased the fluorescence intensity of jGCaMP8f ∼4-fold.

A similar pattern was seen for the L27, F366H, and F366A (Figures 2B–D; Table 1). For the L27A variant only a 5.5-fold increase (comparable to the sevenfold measured for jGCaMP8f) was found in the presence of Mg²⁺, this, however, increased to 27-fold in the absence of Mg²⁺ (Figure 2B; Table 1). In the presence of Mg²⁺, Ca²⁺ binding induced 25- and 27-fold increases for the F366H and F366A variants, respectively, ∼3-fold greater than jGCaMP8f (Figures 2C, D). When measuring the dynamic range of the F366H and F366A jGCaMP8f variants in the absence of Mg²⁺ (pH 7.5), we obtained a 42-fold and a 37-fold increase upon Ca²⁺ binding, respectively, compared to 31-fold for jGCaMP8f (Figure 2; Table 1). Thus, Mg²⁺ binding to apo-jGCaMP8f and its variants significantly affects their fluorescence intensity in vitro and hence it is likely significant in determining the fluorescence dynamic range in resting cells. Moreover, novel variants jGCaMP8 F366H and F366A have significantly increased fluorescence dynamic range compared to jGCaMP8f (Table 1). The other mutations that were introduced had little effect on the dynamic range of jGCaMP8f (Supplementary Table 2) and thus were not investigated in detail for the effect of Mg²⁺.

We measured the molecular brightness of jGCaMP8f and the L27A and F366H variants, in the presence of 2 mM Mg²⁺ (Supplementary Table 1). Measurement of molar extinction coefficient values at 492 nm [ε_{o(492 nm)}] in saturating Ca²⁺ of jGCaMP8f, 54,226 M^–1cm^–1 and a quantum yield of 0.52 gave a brightness value of 28.3 mM^–1cm^–1. This compares well with 29.7 mM^–1cm^–1 measured for GCaMP6f in the absence of Mg²⁺ (Chen et al., 2013) [N.B. in the presence of Ca²⁺, Mg²⁺ has no effect on brightness (Figure 2)]. The values measured for L27A, ε_{o(492 nm)} 48,107 M^–1cm^–1 (quantum yield 0.54) and F366H, ε_{o(492 nm)} 21,203 M^–1s^–1 (quantum yield 0.53) suggest that lower brightness of the apo-form rather than increased brightness of the Ca²⁺-bound form of the F366H variant underlies its increased fluorescence dynamic range compared to jGCaMP8f. In contrast, for the L27A variant, the Mg²⁺ induced large increase in fluorescence intensity in the apo form accounts for the loss in dynamic range. Neither fluorescence intensity nor ε_{o(492 nm)} is affected by Mg²⁺ in saturating Ca²⁺. The only available data for comparison of the apo form with and without Mg²⁺ is for jGCaMP8f (Supplementary Table 1) showing that while in Mg²⁺ ε_{o(492 nm)} is 8,413 M^–1cm^–1, in the absence of Mg²⁺it is reduced to 1,930 M^–1cm^–1, consistent with the 4-fold increase in fluorescence intensity by Mg²⁺ in the absence of Ca²⁺ (Figure 2; Table 1).

The pH dependence of the fluorescence intensity with or without Ca²⁺ (in the presence of 2 mM Mg²⁺) present revealed that for jGCaMP8f and the L27A and F366H variant, the intensities increased for both the Ca²⁺-bound and apo-forms, the dynamic ranges decreased with increasing pH, peaking at pH 7 (Supplementary Figure 1). pK_a values for the Ca²⁺-bound form were 7.2, 7 and 7.5, respectively.

The Ca²⁺ association kinetics of jGCaMP8f measured in the presence of 2 mM Mg²⁺ at pH 7.5 proved remarkably complex, occurring in four phases. The pattern of the four phases was dependent on [Ca²⁺]. The four phases are illustrated in records taken at 1.4 μM [Ca²⁺] at a short timescale of 35 ms (Figure 3A) and a long timescale of 4.5 s (Figure 3B). The observed rates for phases 1, 2, 3, and 4 are denoted R1, R2, R3, and R4. In phases 1, 3, and 4 the fluorescence increases, however, in phase 2 the fluorescence intensity decreased. Phase 2 is no longer detected at [Ca²⁺] > 2.5 μM (Figure 3C). Phase 1 had a fast, ∼500 s^–1 observed rate (k_obs), R1 (Figure 3D). With an off-rate of 33.5 s^–1, t_1/2 for fluorescence decay upon Ca²⁺ sequestration was 22 ms (Figure 3E).

The L27A variant had similarly complex on-kinetics to jGCaMP8f (Figure 4A), with fast (∼300 s^–1) initial on-rate (R1) (Figure 4B). The off-rate of the L27A variant was 84.4 s^–1 (decay t_1/2 of 8.2 ms) (Figure 4C) threefold faster than that of jGCaMP8f.

jGCaMP8f F366H showed a similar pattern of association kinetics to jGCaMP8f and the L27A variant (Figure 5A), with > 500 s^–1 initial on-rates (R1) measurable in the μM [Ca²⁺] range (Figure 5B). The F366H variant had a similar off-rate of 33 s^–1 to jGCaMP8f (Figure 5C). The F366A variant also had improved fluorescence enhancement (27-fold), however, both the on- and the off-rates of this variant are slowed down to 171 s^–1 and 14 s^–1, respectively (Supplementary Figures 3G, H; Table 1).

Supplementary Figure 3 shows the on- and off-kinetic responses of the other variants. Notably, the D367A (EF-3) and D403A (EF-4) mutations had very little effect on either the kinetic or fluorescence responses of jGCaMP8f (Supplementary Figures 3I, J; Supplementary Table 1). The A18S variant had a slightly increased, eightfold dynamic range and slightly slower off-rate (25 s^–1) than jGCaMP8f (Supplementary Table 2).

Measurement of the Ca²⁺ on- and off-response rates of GCaMPs informs on the interaction mechanisms. Furthermore, simple rate and amplitude measurements are usually sufficient as screening tools for judging the sensors’ potential for cell imaging. However, given the complexity of their on-kinetics, for jGCaMP8f and its variants detailed analysis of the rate and amplitude patterns of the four phases of the on-response was required to see if the [Ca²⁺] range where the fast on-phase is the most significant is compatible for intracellular applications. Such analysis helped predict if our novel variants perform well in the cellular environment. As examples, the rate and amplitude patterns of the four phases of the on-response of jGCaMP8f, the L27A, D294A (EF-1), D330A (EF-2), and F366H variants were studied.

Figure 6 shows the Ca²⁺ dependence of the four phases of the on-fluorescence response of jGCaMP8f and its L27A, D294A (EF-1), D330A (EF-2), and F366H variants, revealing that rates (R1 ↑) of the fast, rising phase (400–1,000 s^–1) are highest at around 1–2.5 μM [Ca²⁺] for all five proteins (Figure 6A). In the second phase R2 ↓ of 50–250 s^–1, the fluorescence change is inverted and shows little Ca²⁺ dependence (Figure 6B). The third phase is upward with a slow k_obs (R3 ↑) of 5 s^–1 which increases to ∼30 s^–1 in the 1–2.5 μM [Ca²⁺] range for the F366H variant, while only at high [Ca²⁺] for the other four proteins (Figure 6C). k_obs for the fourth phase (R4 ↑) is slow ∼1 s^–1 (Figure 6D). The low μM [Ca²⁺] range, in which the response rates and patterns are considered most relevant to function in the cellular environment, is highlighted by a red box.

Analysis of the amplitudes (A1, A2, A3, and A4, corresponding to phases 1, 2, 3, and 4 with rates R1, R2, R3, and R4) shows that for the jGCaMP8f, the L27A and F366H variants the rapid first phase represents the major fraction of the amplitude (Figures 7A, B, E). In contrast, for the D294A (EF-1) and D330A (EF-2) variants the major part of the amplitude derives from the slow third phase, but even that only becomes significant at [Ca²⁺] of 20 μM and greater, with time to peak of ∼200 ms (Figures 7C, D). Therefore, even though, the off-rates for these two variants are quite fast (153 and 303 s^–1, Supplementary Figures 3C–F; Supplementary Table 2), they would only function as useful sensors at [Ca²⁺] concentrations that exist in organelles rather than the cytoplasm and only in time-insensitive experimental setups. The F366H variant scores best in terms of reaching maximum amplitude in the first, fastest phase (incidentally, also in phase 3 and 4 (Figure 7E). The low μM [Ca²⁺] range, in which the response rates and patterns are considered most relevant to function in the cellular environment, is highlighted by a red box.

jGCaMP8f and all its variants had apparent K_d for Ca²⁺ in the low 1–4 μM range with Hill coefficient (n) values of 2–5 (Figures 3F, 4D, 5D; Supplementary Figure 2; Table 1; Supplementary Table 2). In our hands, the apparent K_d of jGCaMP8f for Ca²⁺ was 1.80 ± 0.03 μM (Figure 3F), in contrast with a K_d of 334 ± 18 nM measured in the absence of Mg²⁺ (Zhang et al., 2023). The Hill coefficient for Ca²⁺ binding was 2.1 ± 0.1–0.2 in both sets of measurements (Table 1). For the L27A variant a K_d for Ca²⁺ of 2.52 ± 0.02 μM with n of 4.6 ± 0.2 was measured (Figure 4D; Table 1). The F366H and F366A variants had K_d for Ca²⁺ of 1.06 ± 0.03 μM and 0.90 ± 0.2 μM, respectively, with n of 2.3 ± 0.1 and 2.8 ± 0.1, respectively (Figure 5D; Table 1; Supplementary Figure 2).

From in vitro experiments, compared with jGCaMP8f, L27A has faster kinetics and F366H has greater dynamic range whilst kinetics is preserved therefore these variants were tested in cells at 37°C. [Ca²⁺] elevation was induced in HEK293T cells by stimulation with 50 μM ATP (Supplementary Videos 1–3). Figure 8A shows a montage of images at the indicated points of the ΔF/F_o time course for each jGCaMP8f, the L27A and F366H variants at the indicated region of interest. In each case, the first Ca²⁺ transient is followed by one or more smaller further Ca²⁺ rises, indicating oscillating [Ca²⁺]. This can often be seen in response to ATP stimulation. jGCaMP8f detected intracellular Ca²⁺ elevation with a fluorescence increase, ΔF/F_o of 2.4 (ΔF/F_o measured in response to 2 μM ionomycin was 3.2). The F366H variant responded with a ΔF/F_o of 9 (ΔF/F_o measured in response to 2 μM ionomycin was 12), while for L27A the ΔF/F_o was 1 (ΔF/F_o measured in response to 2 μM ionomycin was 2.8). Time to peak values were 1.4 s, 1.8 s and 1 s for jGCaMP8f and its F366H and L27A variants, respectively. Thus, the on-kinetic responses of the three proteins were similar, however, the decay t_1/2 for the L27A variant of 4.6 s was 2.3-fold smaller than for jGCaMP8f (10 s) and F366H variant (11 s). Moreover, the amplitude of fluorescence increase for jGCaMP8f F366H was 3.75-fold greater than for jGCaMP8f (Figure 8; Table 2), while preserving the fast off-kinetics of jGCaMP8f.

jGCaMP8f has 13-fold faster decay rate than GCaMP6f (in vitro decay t_1/2 22 ms at 20°C). Previously, GCaMP3_fast and GCaMP6f_u were developed which had in vitro decay t_1/2 of 11 and 8 ms, respectively, at 20°C (Helassa et al., 2015, 2016). We were interested to see how these differences in in vitro off-rates translate to differences in cell responses at 37°C. Therefore, we compared the kinetics of the responses of jGCaMP8f and three previously developed fast decay sensors GCaMP3_fast and GCaMP6f_u and GCaMP6f, in addition to the two novel jGCaMP8f variants, L27A and F366H, to 2 μM ATP in HUVEC (Figures 9A, B). As seen in the record for GCaMP6f_u, the first Ca²⁺ transient is followed by a smaller second Ca²⁺ rise, indicating oscillating [Ca²⁺]. This can often be seen in response to ATP stimulation and can distort the response kinetics. Comparison of the rise times revealed no significant difference between the sensors when compared to jGCaMP8f except for GCaMP3_fast the rise time for which was significantly shorter than for the others, with a p-value of 0.1 (Figure 9C). The decay times of the responses suggest similar kinetics for the sensors, however, accurate comparison was hindered by the saturation effect most noticeable for GCaMP6f and the jGCaMP8f F366H variant (Figures 9B, D; Table 3). In terms of response dynamic ranges, jGCaMP8f sensitively detected intracellular Ca²⁺ elevation with a fluorescence increase, ΔF/F_o of 2.1, similar to the 2.4 measured in HEK293T cells and significantly higher than 0.23 (p > 0.0001) for GCaMP3_fast and 0.44 (p = 0.003) for GCaMP6f_u (Figure 9E; Table 3). There was no significant difference between jGCaMP8f and its L27A variant, however, both the jGCaMP8f F366H variant and GCaMP6f responded with 1.9- (p = 0.0009) and 3.5-fold (p < 0.0001) greater ΔF/F_o than jGCaMP8f. The full dynamic ranges (ΔF/F_o measured in response to 10 μM ionomycin) of the sensors were similar with, interestingly, GCaMP6f exceeding even that of jGCaMP8f F366H (p = 0.002) (Figure 9F; Table 3), suggesting that a lower apparent K_d for Ca²⁺ compared to that of GCaMP3_fast and GCaMP6f_u is likely to underlie the much greater amplitude of jGCaMP8f in response to 2 μM ATP.