2.1.

Preparation

Transverse organotypic hippocampal slices ($350\mu \mathrm{m}$) were made from brains of Wistar rat pups aged 6 to 8 postnatal days and cultured on Millicell CM membranes for 5 to 6 days prior to transfection as described previously.^1,24

GECIs constructs were verified by DNA sequencing and introduced into target cells by biolistic (“gene-gun”) transfection.²⁵ Briefly, complementary DNAs encoding all GECIs were placed under cytomegalovirus (CMV) promoter, except for Rex-GECO0.9, Rex-GECO1, and Y-GECO, which used the human synapsin I promoter, as a CMV promoter yielded inadequate expression with these sensors (Table 1). Mammalian expression plasmids for all GECIs were constructed using pcDNA32 (Life Technologies). Most GECIs were too dim at resting conditions for reliable identification of expressing cells. To facilitate identification of transfected cells, GCaMPs and GECOs with dim basal fluorescence (all except GCaMP3 and GCaMP6f) were cotransfected with spectrally distinct cytoplasm-filling markers, i.e., blue fluorescent protein (BFP) or GFP, as indicated in Table 2. GECI plasmid DNA ($50\mu \mathrm{g}$), mixed with an equal amount of cytoplasmic FP vector DNA (total of $100\mu \mathrm{g}$ DNA) as required, was coated on $1.0\mu \mathrm{m}$ gold particles.^25,26 GECI plasmid DNAs were delivered to organotypic hippocampal slices after five to six days in vitro by biolistics using a pressure of 120 psi with a customized barrel at a distance of 4 cm above the slice through a Millipore filter membrane ($10\mu \mathrm{m}$ pore size) directly above the target. Imaging was carried out 1 to 4 days post-transfection.

Table 1

Measured fluorescence on- and reported off-kinetics, reported Kd for Ca2+, promoters used for each GECI, and references.

Table 2

Optimal two-photon excitation wavelengths, specifications and references for each genetically encoded calcium indicator (GECI).

Prior to imaging, slice preparations were transferred to the recording chamber for at least 30 min, where they were superfused with oxygenated ($95\%{\mathrm{O}}_2/5\%{\mathrm{CO}}_2$ gas mixture) artificial cerebrospinal fluid (ACSF) at 30 to 32°C. The ACSF contained, in mM: 120 NaCl, 3 KCl, 2 ${\mathrm{MgSO}}_4$, 4 ${\mathrm{CaCl}}_2$, 1.2 ${\mathrm{NaH}}_2{\mathrm{PO}}_4$, 23 ${\mathrm{NaHCO}}_3$, 11 glucose.

2.2.

Imaging and Electrophysiology

Slices were viewed on an upright focusing-nosepiece epifluorescence microscope (Olympus, BX51WI) using a $60\times$ NA 1.0 IR water immersion objective (Olympus) and an MRC1024MP laser scanner (Bio-Rad Microscience, Hemel Hempstead, United Kingdom) with nondescanned photomultiplier tube detectors. Available two-photon excitation wavelengths ranged between 710 to 970 nm (coherent MIRA or Spectra-Physics MaiTai titanium:sapphire laser). Average energy levels measured at the specimen plane were 17 mW at 710 nm, 33 mW at 800 nm, and 10.5 mW at 970 nm. Specific excitation wavelengths used for each sensor are listed in Table 2.

Pyramidal cells expressing GECIs in the CA1 region were identified by single-photon widefield epifluorescence and impaled under visual control with sharp microelectrodes (60 to $80\mathrm{M}\mathrm{\Omega }$) filled with 3 M KCl in 20 mM 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid. Electrophysiological data were acquired and analyzed using AxoGraph software (AxoGraph Scientific, Sydney, Australia), and images were collected and analyzed using LaserSharp (Bio-rad Microscience) and ImageJ²⁹ software. In brief, one or more action potentials were evoked by 8 ms, 600 to 1200 pA current injections at 100 Hz via the intracellular electrode. Ten line scan images (each extending over 400 to 500 ms) were collected for each stimulus condition (one to eight action potentials) from two to five cells. Calcium transients were measured by collecting line scans across the proximal apical dendrite and expressed as a percent fractional change in fluorescence: $\%\mathrm{\Delta }F/F=100$ $(F_{\text{transient}}-F_{\text{initial}})/(F_{\text{initial}}-F_{\text{background}})$, where $F_{\text{initial}}$ is the mean fluorescence intensity of the imaged cell over a 70 ms period prior to stimulation, and $F_{\text{transient}}$ was measured over 330 ms for one to four action potentials and 430 ms for eight action potentials (Fig. 1).

Fig. 1

Simultaneous calcium imaging and electrophysiological recordings. (a) G-CaMP7 expressing CA1 pyramidal neuron; location of line scan is indicated by the white line across the proximal apical dendrite. Increasing fluorescence intensity in this and the following line scan images is represented by colors from black through red and yellow to green, blue and purple. (b) Representative line scans during one action potential and (c) eight action potentials with GECO1.2. The vertical white lines in (b) and (c) and corresponding deflections in (d) and (e) are time stamps, the first indicating the start of current injection, and the second, 50 ms later, serving as an internal time marker. Traces above the line scans show the simultaneously recorded action potentials on the same time scale; these are visible in more detail in panels (f) and (g). (d) and (e) Quantifications of line scans: (d) one action potential and (e) eight action potentials. $X$ axis shows time, $Y$ axis shows $\mathrm{\Delta }F/F$ values. Traces are representations of single line scans. (f) and (g) Simultaneously recorded membrane potentials showing one or eight action potentials.

3.1.

Expression and Basal Fluorescence of GECIs

Of the sensors tested here, GCaMP3 had the brightest fluorescence at resting calcium levels, which allowed for reliable detection of expressing cells. GCaMP6f also displayed adequate basal fluorescence for identification of transfected cells. Cotransfection with bright spectrally distinct cytoplasmic markers, such as BFP and GFP, was required for all other sensors tested (Table 2) in order to identify transfected cells. Expressing R-GECO in CA1 pyramidal cells often resulted in bright punctate, rather than diffuse, fluorescence throughout the cell; although unique to R-GECO among the sensors tested here, this pattern is a common phenomenon with red FPs.^30,31

Our reported comparison for each indicator is based on the values of fractional change in fluorescence and is, therefore, independent of the different excitation energies available at different optimized wavelengths. We found that the optimal wavelengths (over the range available with our laser) for maximal fractional change in intensity in response to action potentials for various GECIs were as follows: 910 nm for GCaMP3, G-CaMP7, and G-GECO 1.0, 1.1 and 1.2; 930 nm for GCaMP6f; 800 nm for B-GECO; 888 nm for R-GECO, Rex-GECO0.9, and Rex-GECO1; 970 nm for CAR-GECO and O-GECO; and 810 nm for Y-GECO1s (Table 2). (When imaging Y-GECO1s with 520 nm single-photon excitation or with 990 nm two-photon excitation, the fluorescence of resting cells is bright but decreases as ambient calcium ion concentration increases.²⁸ Conversely, with 410 nm single-photon excitation or with two-photon excitation below 850 nm, the resting fluorescence is dimmer but increases in proportion to the calcium ion concentration.²⁸)

3.2.

Comparison of the two-photon performances of selected G-CaMPs and GECOs as demonstrated by their ability to report apical dendritic calcium transients associated with action potentials

After optimizing the excitation wavelength for each GECI, fluorescence responses to one, two, three, four, and eight action potentials were determined. In general, G-CaMPs detected low numbers of action potentials (one to four) more reliably than GECOs. Of the sensors tested here, G-CaMP7, GCaMP6f, and G-GECO1.2 were able to detect single action potentials with high reliability (Fig. 2).

Fig. 2

(a) Summary of changes in fluorescence intensity corresponding to one to eight action potentials at 100 Hz for all GCaMPs and GECOs tested. (b) Summary of changes in fluorescence intensity corresponding to one to eight action potentials at 100 Hz for all genetically encoded calcium indicators tested, excluding GCaMP6f and G-CaMP7. (c) Summary of changes in fluorescence intensity corresponding to one to eight action potentials at 100 Hz for blue-, orange-, carmine-, red-, and yellow-GECOs. Each data point corresponds to $n=10$ measurements from multiple neurons. Error bars represent standard error of the mean.

Fluorescence changes from G-CaMPs in response to a single action potential were $26\pm 7\%$ (all values $\pm$ standard error of the mean) with GCaMP3, $120\pm 58\%$ with GCaMP6f, and $302\pm 39\%$ with G-CaMP7 [Fig. 2(a)]. Among the GECOs, G-GECO1.2 performed best in reporting low numbers of action potentials. The performance of G-GECO1.2 was superior to that of GCaMP3 both in the $\mathrm{\Delta }F/F$ values and in the linearity of responses to increasing numbers of action potentials [Figs. 2(a) and 2(b)]. G-GECO 1.2 showed a $51\pm 7\%$ $\mathrm{\Delta }F/F$ for a single action potential, and $384.5\pm 26\%$ $\mathrm{\Delta }F/F$ for eight action potentials, whereas these values were $26\pm 7.7\%$ and $329\pm 33\%$ for GCaMP3 [Fig. 2(b)]. G-GECO 1.0 and 1.1 displayed smaller fluorescence transients ($22.5\pm 18\%$ and $16.5\pm 9\%$, respectively) than G-GECO 1.2 in response to single action potentials [Fig. 2(b)]. B-GECO demonstrated a $35\pm 11\%$ fluorescent change upon a single action potential, but the signals saturated beyond three action potentials and the maximal $\mathrm{\Delta }F/F$ values only reached $52\pm 10\%$ with eight action potentials [Fig. 2(c)].

Performance of the longer-wavelength GECOs was modest. O-GECO and CAR-GECO performed similarly, with CAR-GECO displaying slightly higher $\mathrm{\Delta }F/F$ values. The fluorescence transients ranged from $9.5\pm 4.5\%$ to $22\pm 11\%$ for CAR-GECO in response to one to eight evoked action potentials, respectively, versus from $7.5\pm 10\%$ to $18.7\pm 15\%$ for O-GECO [Fig. 2(c)]. R-GECO responded with a slight increase in fluorescence of $6.5\%\pm 7$ and $12.5\%\pm 15$ to a single or eight action potentials, respectively [Fig. 2(c)]. Two other red indicators, Rex-GECO0.9 and Rex-GECO1, displayed greater fluorescence changes than R-GECO. The maximum $\mathrm{\Delta }F/F$ values in response to eight action potentials were $65\pm 56\%$ with Rex-GECO0.9 and $47\pm 28\%$ with Rex-GECO1. Moreover, Rex-GECO0.9 showed a linear increase in $\mathrm{\Delta }F/F$ values over one to eight action potentials at 100 Hz. On the other hand, Rex-GECO1 fluorescence shows a larger fluorescence increase in response to one to three action potentials.

The performance of the yellow sensor, Y-GECO1s, was comparable to that of Rex-GECO0.9 when imaged at 810 nm. Both indicators have similar $\mathrm{\Delta }F/F$ values as well as linear responses over the range of one to eight action potentials, though Rex-GECO0.9 offers a slightly higher fluorescent response than Y-GECO1s to eight action potentials but is considerably more variable ($65\pm 56\%$ versus $53\pm 11\%$, respectively).

The most robust calcium transients by far were obtained with G-CaMP7, which showed an immense $302\pm 39\%$ $\mathrm{\Delta }F/F$ in response to individual action potentials [Fig 2(a)], sufficient to render spontaneous electrical activity clearly visible by eye. Beyond four action potentials at 100 Hz, however, this sensor began to deviate from linearity, and it displayed a $1351\pm 147\%$ change with eight action potentials [Fig. 2(a)]. The G-CaMP7 $\mathrm{\Delta }F/F$ was significantly larger than those of GCaMP6f over the entire one to eight action potential range ($p<0.0001$, analysis of variance), but GCaMP6f provided a more linear response over this range with $120\pm 58\%$ $\mathrm{\Delta }F/F$ in response to single action potentials and $876\pm 145\%$ $\mathrm{\Delta }F/F$ in response to eight action potentials.

On- and off-kinetics of all these sensors were too slow to follow individual action potentials reliably at frequencies above approximately 20 Hz, as the fluorescence in general rose with a time constant around 100 ms and decayed with a time constant around 400 to 700 ms, with the exception of Rex-GECO0.9 with $\sim 320\mathrm{ms}$ and O-GECO with $\sim 350\mathrm{ms}$ tau-off values (Table 1). Our data acquisition epoch was too short for reliable determination of tau-off values, but our observation that G-CaMP7 fluorescence transients evoked by single action potentials decayed at a rate intermediate between that of GCaMP3 and Rex-GECO0.9 allowed us to estimate that its decay time constant was $\sim 470\mathrm{ms}$. Our data did permit determination of tau-on values for all sensors, and these are reported in Table 1.