Generation and Imaging of Transgenic Mice that Express G-CaMP7 under a Tetracycline Response Element

The spatiotemporally controlled expression of G-CaMP fluorescent calcium indicator proteins can facilitate reliable imaging of brain circuit activity. Here, we generated a transgenic mouse line that expresses G-CaMP7 under a tetracycline response element. When crossed with a forebrain-specific tetracycline-controlled transactivator driver line, the mice expressed G-CaMP7 in defined cell populations in a tetracycline-controlled manner, notably in pyramidal neurons in layer 2/3 of the cortex and in the CA1 area of the hippocampus; this expression allowed for imaging of the in vivo activity of these circuits. This mouse line thus provides a useful genetic tool for controlled G-CaMP expression in vivo.


Introduction
Genetically encoded calcium indicators (GECIs), such as green fluorescent protein (GFP)based G-CaMPs, are increasingly used to monitor neural activity in living brains [1][2][3][4][5][6]. GECIs have recently begun to replace synthetic calcium indicator dyes because GECIs can be expressed in genetically defined cell populations to perform long-term repeated imaging of activity in the same cell populations [7][8][9][10]. GECIs can be introduced into cells by gene transfer techniques, such as in utero electroporation, viral vectors and transgenic technologies. Once good lines are established, transgenic mice have significant advantages, such as the capability for GECI labeling of each animal without surgery and for highly reproducible and homogeneous GECI expression; these advantages facilitate a more reliable and efficient collection of imaging data on the activity of large cell populations.
To date, several transgenic mouse lines that express G-CaMPs in the brain have been reported [11][12][13][14][15]. The different G-CaMP variants expressed in these mice have different sensitivities, kinetics and signal-to-noise ratios, which allows researchers to monitor a wide range of neural activity [2,3,5]. Although cell-type-specific expression can be achieved via Cre/lox-mediated recombination, the temporal control of G-CaMP expression, which could circumvent the potential adverse effects of long-term and/or developmental expression of G-CaMPs, is

Generation of TRE-G-CaMP7 mice
The cDNA encoding G-CaMP7 [4] connected to the coding sequence of DsRed2 via the 2A peptide sequence from the Thosea asigna virus (T2A) was subcloned between the BamHI and NotI sites of a modified pTRE-Tight vector (Clontech). The 3.5 kb DNA fragment excised by the ApaI and SphI restriction enzymes was gel-purified and injected into the pronuclei of fertilized eggs of C57BL/6J mice using standard techniques. The mice were genotyped using PCR to amplify DNA samples extracted from their tails. The primers used were 5'-CTGCTGCCCGA CAACCA-3' and 5'-GTCGTCCTTGAAGAAGATGG-3', which amplified a 465-bp fragment of the G-CaMP7 coding sequence from the DNA samples of transgene-positive mice. The mouse line, termed TRE-G-CaMP7, is available at the RIKEN BioResource Center (http:// www.brc.riken.jp/lab/animal/en/; stock number RBRC06510).

Analysis of transgene expression
Double transgenic mice obtained by breeding TRE-G-CaMP7 mice with driver mice expressing tTA under the forebrain-specific calcium/calmodulin-dependent protein kinase IIα promoter (CaMKII-tTA mice; stock number 003010, Jackson Laboratory) [19] were deeply anesthetized with Avertin and were perfused transcardially with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA) in PBS. Brains were removed and further fixed in 4% PFA at 4°C overnight. Coronal or parasagittal sections were cut on a vibratome to a thickness of 50 or 100 μm, respectively. For immunolabeling, the sections were incubated at 4°C overnight with mouse anti-calcium/calmodulin-dependent protein kinase IIα (CaMKIIα) antibody (1:1000, Clone 6G9, Millipore), rabbit anti-Iba1 antibody (1:2000, 019-19741, Wako Pure Chemical Industries, Ltd., Japan), rabbit anti-glial fibrillary acidic protein (GFAP) antibody (1:1000, N1506, Dako) or anti-c-fos antibody (1:5000, PC38, Calbiochem) diluted in PBS containing 2% normal goat serum, 1% BSA and 0.1% Triton X-100, followed by Alexa Fluor 647-labeled goat anti-rabbit or anti-mouse IgG antibody (1:700-1:1000, A-21245 or A-21236, Life Technologies) diluted in the same buffer at room temperature for 1 h. Nuclear counterstaining was performed in PBS containing 10 μg/ml Hoechst 33258 (Calbiochem) and 0.1% Triton X-100 at room temperature for 5 min. Fluorescence images were acquired using a Keyence BZ-9000 epifluorescence microscope equipped with a 4x objective or an Olympus FV1000 or FV1200 laserscanning confocal microscope equipped with a 4x, 10x, 20x dry or 60x oil immersion objective. The G-CaMP7 and DsRed2 signals shown in all the images represent their native fluorescence and were not enhanced by immunolabeling. For cell counting, the fraction of G-CaMP7-labeled cells was calculated against the number of total neuronal cells stained with NeuroTrace 435/455 Blue Fluorescent Nissl Stain (1:100, Life Technologies) diluted in PBS containing 0.1% Triton X-100 at room temperature for 20 min. G-CaMP7-labeled cells and NeuroTrace-stained cells were counted in upper layer 2/3 (< 250 μm from the pia), lower layer 2/3 (> 250 μm from the pia) and layer 5 of the neocortex as well as the CA1 area of the hippocampus. More than 200 cells in 6-8 fields per animal (field size 212 x 212 μm for the hippocampus and 318 x 318 μm for the cortex; 16-58 cells per field) were counted for each layer or area. For doxycycline (Dox) treatment, mice were housed with ad libitum access to 0.1% saccharin water with or without 2 mg/ml doxycycline hyclate (Sigma-Aldrich) for 2-4 weeks. For withdrawal from Dox, the mice treated with Dox for 2 weeks were then given water without Dox for an additional 2 weeks. Light-induced c-fos expression in the visual cortex was examined in mice kept in darkness for 24 h and then exposed to ambient light for 1 h before perfusion. Control mice were perfused immediately after 24 h of adaptation to darkness.

In vivo two-photon imaging and data analysis
Adult double-transgenic mice were anesthetized with isoflurane (3% induction, 1.5% maintenance) and were placed in a stereotaxic frame. Atropine (0.3 mg/kg, s.c.) and dexamethasone (2 mg/kg, s.c.) were administered prior to the anesthesia to reduce respiratory secretions and brain edema, respectively [20]. A stainless steel head plate with a circular opening (7 mm diameter) was placed over the left parietal bone and attached to the skull with dental acrylic. Craniotomy and window preparation for cortical and hippocampal imaging were performed essentially as described by Holtmaat et al. and Dombeck et al., respectively [7,21]. For hippocampal imaging, a small volume of cortical tissue overlying the dorsal CA1 region of the hippocampus was surgically removed by aspiration before the implantation of an imaging window [7]. After surgery, the mice were returned to their home cages and were allowed to recover until the imaging experiments.
On the day of imaging, each mouse was re-anesthetized with isoflurane (3% induction, 1% maintenance) supplemented with chlorprothixene (1 mg/kg, i.p.) and was placed under the microscope objective via the head plate. Body temperature was maintained at 37°C with a heating pad throughout the imaging sessions. G-CaMP7 was excited using a Ti:Sapphire laser (Mai Tai DeepSee eHP, Spectra-Physics) at 910 nm, and time-series fluorescence changes in neurons during spontaneous network activity in the dorsal posterior cortex (visual cortex) [20] or the dorsal CA1 hippocampus [7] were imaged using a 495-540 nm bandpass filter and a GaAsP photomultiplier tube. The laser power under the objective was 10-28 mW. Images were acquired using an Olympus FV1000MPE microscope equipped with a 25x NA 1.05 objective (Olympus) in 256 x 256 pixels (field size 254 x 254 μm for the cortex and 169 x 169 μm for the hippocampus) or 128 x 128 pixels (12.6 x 12.6 μm for dendritic imaging) at a frame rate of 2.3 or 5.3 Hz (pixel dwell time 2 μs/pixel), respectively. For dual color imaging, DsRed2 was simultaneously excited at 910 nm, and the signal was separated by a 570 nm dichroic mirror and a 575-630 nm bandpass filter.
We performed image analysis using custom software written in MATLAB (MathWorks). The center of a region of interest (ROI) in each cell was determined manually on the time-averaged G-CaMP7 fluorescence image so that the ROI was defined as a circle with a 6-pixel (6.0 μm for cortical neurons) or 10-pixel (6.6 μm for hippocampal neurons) radius inscribed within the cell body. The fluorescence intensity values of the pixels within the ROI in each frame of the time-lapse image sequences were then averaged to represent the cellular signal F for that cell. For each F, a baseline value (F0) was defined as the mean of the 75th (for cortical imaging) or 80th percentile (for hippocampal imaging) of all the data points. F0 was then used to calculate ΔF/F as (F-F0)/F0. The timing and magnitude of neuronal activity were defined as the local maxima of ΔF/F above the threshold (4 standard deviations from F0). The center of an ROI for the cortical neuropil was also defined manually in image areas that were visually devoid of cell bodies; this ROI had the same size and shape as the ROI used in the analysis of cellular signals. An ROI for a basal dendritic segment of CA1 pyramidal cells was defined manually. A pair-wise cross-correlation analysis of ΔF/F signals was performed using custom software written in MATLAB. For hippocampal imaging, the sequential activation of a pair of cells was quantified by counting the number of events in which one cell's activity preceded or followed another's by one image frame. Quasi-ratiometric measurement was performed by calculating the normalized ratio of G-CaMP7 to DsRed2 signals. To simulate motion artifacts, randomly generated image displacements of 8-12 pixels (0.79-1.18 μm) in distance and 0.19-0.56 s in duration were added to the real data at an average frequency of 0.2 Hz. The direction of the artificial motion artifacts was chosen randomly and remained fixed throughout the simulation.

Results and Discussion
We generated a transgenic mouse line that expresses G-CaMP7 and DsRed2 via 2A peptidemediated bicistronic expression under the control of a TRE (Fig 1A). G-CaMP7 is a recently improved G-CaMP variant that exhibits large fluorescence changes in response to a broad range of intracellular calcium concentrations and is thus suitable for reliable in vivo calcium imaging [4]. The coexpressed DsRed2 serves as a calcium-independent fluorescent marker protein of a different color and aids the identification of G-CaMP7-expressing cells in vivo because the basal fluorescence of G-CaMP7 is relatively low [4]. In addition, this DsRed2 signal can be used as a normalization factor for the quasi-ratiometric measurement of calcium responses and to estimate the extent of image motion artifacts, especially when imaging is performed under awake behaving conditions (for an example of ratiometric measurements, see S5 Fig). From 774 DNA-injected eggs, 12 transgene-positive founder lines were obtained, 7 of which showed detectable levels of G-CaMP7 fluorescence in brain sections of offspring crossed with the CaMKII-tTA line. By contrast, our parallel effort to generate CaMKII-G-CaMP7 transgenic mice by inserting the same bicistronic expression cassette under the CaMKIIα-promoter [19] produced 7 transgene-positive mice from 1151 DNA-injected eggs. None of these mice showed detectable G-CaMP7 fluorescence, suggesting that the tTA/TRE system may have achieved higher expression levels by transcriptional amplification.
We more thoroughly characterized one founder line that expressed G-CaMP7 and DsRed2 prominently in the forebrain by crossing the founder line with the CaMKII-tTA line (TRE-G--CaMP7 x CaMKII-tTA mice; Fig 1B). Higher-magnification images revealed that pyramidal neurons in layer 2/3 and layer 5 of the neocortex as well as pyramidal neurons in the CA1 area of the hippocampus strongly expressed G-CaMP7 with varying degrees of DsRed2 coexpression ( Fig 1C). The coexpressed DsRed2 exhibited cytoplasmic localization in most cells, although it was also localized to the nucleus in a subset of cells, particularly in layer 2/3 of the cortex (Fig 1C). The fractions of G-CaMP7-expressing cells among the total neurons were 49.4±4.7, 62.5±6.5, 45.1±6.9 and 71.5±5.2% in upper layer 2/3, lower layer 2/3, and layer 5 in the cortex and the CA1 area of the hippocampus, respectively (mean±SD, n = 13-15 fields, 2 mice). Immunofluorescence labeling confirmed that G-CaMP7-expressing neurons in these mice were CaMKIIα-positive, consistent with the expression pattern of tTA defined by the cell-type-specific promoter of the driver line (Fig 1D) [19].
In the hippocampus, a subset of CA3 pyramidal cells and dentate gyrus granule cells appeared to be less strongly labeled with G-CaMP7 than the CA1 pyramidal cells, whereas subicular pyramidal neurons were more strongly labeled (S1A Fig). Strikingly, the stratum lacunosum-moleculare and the molecular layer of the dentate gyrus were intensely labeled with G-CaMP7 and DsRed2 (S1A Fig). This finding was in accordance with the strong labeling of the neurons in the entorhinal cortex, which projects perforant path fibers to these layers (S1B Fig).
Previous studies have reported that G-CaMPs expressed by adeno-associated viral vectors accumulated within the cytoplasm and the nucleus over time; these cells exhibited abnormal calcium responses [2,7]. The G-CaMP7-expressing cells in the TRE-G-CaMP7 x CaMKII-tTA mice showed no apparent abnormal morphology, and their nuclei were devoid of G-CaMP7 fluorescence even at 7 months of age (S2 Fig).
The functional influence of long-term G-CaMP7 expression was examined histologically by visualizing the light-induced expression of the immediate early gene product c-fos in the visual cortices of TRE-G-CaMP7 x CaMKII-tTA and wild-type mice at 8 months of age (S3 Fig). Exposing mice of both genotypes to light for 1 h caused robust induction of c-fos protein expression in the nuclei of many neurons in the visual cortex. The levels of induction were equivalent in both genotypes, suggesting that neuronal function remained undisturbed after long-term transgenic G-CaMP7 expression in these mice.
The possibility of neuronal toxicity associated with long-term transgenic labeling with G-CaMP7 was further assessed by immunofluorescence microscopy using antibodies against the reactive astrocyte marker GFAP and the microglial marker Iba1 (S4 Fig). Most of the GFAP-positive astrocytes were located in the hippocampus, but only a small number of them were observed in the neocortex in both TRE-G-CaMP7 x CaMKII-tTA and wild-type mice at 7 months of age (S4A Fig), whereas Iba1-positive microglia were diffusely distributed throughout the neocortex and hippocampus (S4B Fig). The finding that the distribution patterns of GFAPpositive astrocytes and Iba1-positive microglia were indistinguishable between both genotypes of mice provides evidence that long-term transgene expression is non-toxic in older TRE-G--CaMP7 x CaMKII-tTA mice. Moreover, the absence of G-CaMP7 expression in these two types of glial cells further confirms the neuronal expression of G-CaMP7 in the TRE-G-CaMP7 x CaMKII-tTA mice (S4C Fig). We next tested whether G-CaMP7 expression could be reversibly suppressed by the administration of the tetracycline derivative doxycycline (Dox; Fig 1A). G-CaMP7 and DsRed2 fluorescence signals in the forebrain were reduced nearly to the background level in TRE-G-CaMP7 x CaMKII-tTA mice treated with Dox (2 mg/ml in 0.1% saccharin water) for 2-4 weeks (Fig 1E). Withdrawal from Dox for 2 weeks after 2 weeks of treatment restored the G-CaMP7 and DsRed2 signals to the control level (Fig 1E), demonstrating that temporal control of G-CaMP7 expression is possible in these mice.
In addition to the neocortex and hippocampus, TRE-G-CaMP7 x CaMKII-tTA mice exhibited notable G-CaMP7 expression in various neuronal populations in forebrain structures, such as the axon terminals of olfactory receptor neurons in olfactory glomeruli, layer 2 neurons in the piriform cortex and neurons in the basolateral amygdala (Fig 2). A subset of neurons in the lateral part of the striatum was also found to be labeled with G-CaMP7 (Fig 2). These expression patterns further expand the applicability of these mice in in vitro and in vivo imaging studies of neural circuit activity in widespread forebrain areas.
The functionality and utility of the TRE-G-CaMP7 x CaMKII-tTA mice were further demonstrated using two-photon imaging of spontaneous circuit activity in the neocortex and hippocampus of anesthetized mice (Figs 3 and 4). In the neocortex, a large population of neurons with homogeneous G-CaMP7 and DsRed2 labeling were imaged in superficial layers of different depths (Fig 3A and S1 Movie). Time-lapse imaging revealed that a subpopulation of layer 2/3 neurons exhibited large, spontaneous calcium transients, which often occurred near the peaks of synchronous slow baseline oscillations (Fig 3B and 3C and S2 Movie). These correlated slow calcium oscillations were also observed in G-CaMP7 but not in DsRed2 signals in the neuropil and were similar to those observed in a cortex loaded with the synthetic calcium dye Oregon Green BAPTA-1 AM [22] (Fig 3C). A pair-wise correlation analysis of cellular G-CaMP7 signals resulted in an average correlation coefficient of 0.308±0.143 (Fig 3D; mean±SD, n = 129 cell pairs, 3 mice), which was indicative of a weak overall correlation, and the identification of a small number of cell pairs that were anatomically remote but exhibited highly correlated spontaneous activity (Fig 3E).
In the CA1 area of the hippocampus, different subcellular compartments, such as axons, basal dendrites and cell bodies, of G-CaMP7-and DsRed2-labeled pyramidal cells were imaged at different depths corresponding to the alveus, stratum oriens and stratum pyramidale (Fig 4A  and S3 Movie). Time-lapse imaging revealed that a subset of pyramidal cells exhibited large, spontaneous calcium transients with no baseline oscillations (Fig 4B and 4C and S4 Movie). The average pair-wise correlation coefficient was 0.030±0.066 (mean +/-SD, n = 355 cell pairs, 3 mice), suggesting that spontaneous network activity was much less correlated in the hippocampus than in the cortex (Fig 4D). However, the sequential activation of particular cell pairs was repeatedly observed in the hippocampus (Fig 4E), suggesting the presence of hidden temporal structures [23][24][25]. Calcium transients could be imaged not only in cell bodies but also in the basal dendrites of the CA1 pyramidal cells in these mice (S5 Fig) [26]. Collectively, the results shown in Figs 3 and 4 demonstrate that TRE-G-CaMP7 x CaMKII-tTA mice can be used to study the activity patterns of neuronal ensembles in vivo. A full elucidation of the patterns observed in this work will require future in-depth studies.
In conclusion, we generated and characterized TRE-G-CaMP7 transgenic mice expressing a recently improved G-CaMP7 under a TRE. The expression of G-CaMP7 in these mice is homogeneous, stable and functional, and it can be controlled spatially and temporally by celltype-specific tTA driver lines and Dox treatment, respectively. Together with a recently reported transgenic mouse line that expresses a Förster resonance energy transfer (FRET)-based GECI via a tTA-mediated strategy [27], our TRE-G-CaMP7 mice offer a useful alternative to Cre/lox-mediated cell-type-specific GECI labeling and will provide a valuable genetic tool for the reliable imaging of neural circuit activity in vivo.