A New Transgenic Mouse Line for Imaging Mitochondrial Calcium Signals

Abstract Mitochondria play a key role in cellular calcium (Ca2+) homeostasis. Dysfunction in the organelle Ca2+ handling appears to be involved in several pathological conditions, ranging from neurodegenerative diseases, cardiac failure and malignant transformation. In the past years, several targeted green fluorescent protein (GFP)-based genetically encoded Ca2+ indicators (GECIs) have been developed to study Ca2+ dynamics inside mitochondria of living cells. Surprisingly, while there is a number of transgenic mice expressing different types of cytosolic GECIs, few examples are available expressing mitochondria-localized GECIs, and none of them exhibits adequate spatial resolution. Here we report the generation and characterization of a transgenic mouse line (hereafter called mt-Cam) for the controlled expression of a mitochondria-targeted, Förster resonance energy transfer (FRET)-based Cameleon, 4mtD3cpv. To achieve this goal, we engineered the mouse ROSA26 genomic locus by inserting the optimized sequence of 4mtD3cpv, preceded by a loxP-STOP-loxP sequence. The probe can be readily expressed in a tissue-specific manner upon Cre recombinase-mediated excision, obtainable with a single cross. Upon ubiquitous Cre expression, the Cameleon is specifically localized in the mitochondrial matrix of cells in all the organs and tissues analyzed, from embryos to aged animals. Ca2+ imaging experiments performed in vitro and ex vivo in brain slices confirmed the functionality of the probe in isolated cells and live tissues. This new transgenic mouse line allows the study of mitochondrial Ca2+ dynamics in different tissues with no invasive intervention (such as viral infection or electroporation), potentially allowing simple calibration of the fluorescent signals in terms of mitochondrial Ca2+ concentration ([Ca2+]).


Introduction
Precise regulation of cytosolic [Ca 2þ ] in terms of amplitude, frequency, and spatiotemporal patterns contributes to fundamental processes such as synaptic transmission, exocytosis, muscle contraction, and cell death. [1][2][3][4] These processes are finely orchestrated and even subtle changes give rise to a disequilibrium in the system and correlate with pathologies ranging from cardiovascular diseases to neurological disorders and carcinogenesis. [5][6][7] Cytosolic [Ca 2þ ] is not only controlled by the action of plasma membrane (PM) pumps, channels, and exchangers, but also by the presence of numerous Ca 2þ buffers and by the activity of organelles, i.e. primarily, the endo/sarcoplasmic reticulum (ER/SR), the Golgi apparatus and mitochondria. [8][9][10][11] Best known for their role in energy metabolism and cell death, 12 mitochondria are nowadays well accepted as central players of the cellular Ca 2þ handling toolkit. 8,[13][14][15][16][17][18] Beside contributing to shaping cytosolic Ca 2þ signals, mitochondria rely on Ca 2þ for their specific metabolic functions [19][20][21] and dysfunctions of the mitochondrial Ca 2þ handling machinery are linked to several pathological conditions. [22][23][24][25][26][27][28][29] Advances in understanding of biological phenomena have almost invariably been dependent on improvements in the techniques to measure them and this is particularly true in the case of Ca 2þ signaling. [30][31][32] In particular, the introduction in the last decades of more and more sensitive and sophisticated Ca 2þ indicators, both synthetic and genetically encoded, has been essential for unraveling the role of Ca 2þ in cell pathophysiology. 32,33 Mitochondria are a classical example of how technology and physiology are intimately linked. Until the 90s, the lack of methods to measure Ca 2þ dynamics in the mitochondrial matrix led to the wrong conclusion that these organelles were not involved in cellular Ca 2þ signaling. The development of new methods that allowed the monitoring of mitochondrial matrix Ca 2þ in living cells completely changed this scenario. Today, a large number of methodologies to monitor mitochondrial Ca 2þ in living cells is available, from chemical dyes to GECIs. 4,[34][35][36][37][38][39] Mitochondria-targeted GECIs, in particular, have extensively been used in cell lines where transfection is, in most cases, a simple and efficient method to express the sensors. More complex is the expression of GECIs in primary cultures, acute slices, and especially in vivo, where infection with viral vectors or electroporation protocols are the exclusive choice. We and others have exploited in vivo virus-mediated delivery or electroporation of mitochondria-targeted recombinant Ca 2þ probes. 4,[39][40][41][42][43][44] Although of undeniable value, these approaches present important limitations, for example, invasive surgical delivery that is time-consuming and stressful for the experimental animal, limited spread of the virus in the region of interest, uneven expression in the target cell population, potential cytotoxicity, or high gene expression. 45 Ideally, to overcome these issues, the use of transgenic mice expressing Ca 2þ sensors in a controlled way appears the best solution and, indeed, several mouse lines expressing cytosolic GECIs have become available in the last years. [46][47][48][49][50][51][52] Transgenic animals expressing GECIs offer a number of advantages over viral injection (or other techniques such as in situ electroporation and bulk injection), guaranteeing less invasive experimental procedures, consistent expression level among cells and during aging, full access to all body regions. Surprisingly, however, no such transgenic mouse lines are available yet for mitochondria. A couple of lines have been generated for the expression of mitochondria-targeted bioluminescence-based sensors (GFPaequorin, 53 c-Photina 54 ) but the spatial resolution of these sensors is quite limited. Moreover, these probes are non-ratiometric and thus are sensitive to motion artifacts of the sample due to breathing, heartbeat, and blood flow or movements of the organelles themselves. Anecdotical information from different laboratories tells that, although mitochondria-targeted sensor sequences can be easily inserted in the mouse genome, the level of expression of the Ca 2þ indicators are so low to be practically useless. The reason for these repetitive failures is presently unknown.
We here describe the generation and characterization of a transgenic mouse line expressing the mitochondria-targeted Cameleon Ca 2þ sensor 4mtD3cpv [55][56][57] under the control of Cre recombinase. Among GECIs, we focused on this sensor because, though not the most brilliant among the available mitochondrial Ca 2þ sensors, it is based on FRET and accordingly: (1) it allows ratiometric measurement of signal dynamics and thus, unlike intensity-based indicators, permits easily to transform the fluorescence changes into absolute [Ca 2þ ] values; (2) it guarantees that the output measurement is independent of movement artifacts, probe expression levels, focus, and intensity changes. 4,58 Mitochondria, in particular, are highly dynamic organelles that undergo fission and fusion processes and move inside the cell. For these reasons, ratiometric biosensors are particularly recommended for qualitative and quantitative Ca 2þ measurements in these organelles, ensuring independency from movements and shape changes. Cameleon 4mtD3cpv is characterized by a single affinity constant for Ca 2þ (K d % 3.2 mM at pH 8) and a dynamic range in response to Ca 2þ changes (R max /R min ) calculated around 2.4 in living cells, 39 features that make this probe suitable to measure mitochondrial Ca 2þ changes in different organs and tissues.
To generate the transgenic mouse line mt-Cam, the Cameleon sequence, preceded by a loxP-STOP-loxP sequence, was inserted in the ROSA26 genomic locus, well characterized for ubiquitous expression of transgenes at moderate levels without interrupting essential genes. 59,60 The probe can be expressed in a tissue-specific, controlled manner upon Cre recombinase-mediated excision, easily obtainable with a single genetic cross. Following crossing with ubiquitous or tissuespecific Cre-expressing mice, positive offspring showed expression of the mitochondria-targeted probe in different cell types, from newborn to aged animals. Moreover, a series of classical Ca 2þ imaging experiments in primary cultures and in acute brain slices are presented here, demonstrating the functionality of the probe and its potential applicability to other relevant experimental paradigms.

Construction of the Probe
To improve the transduction efficiency in mammalian cells, the mitochondria-targeted Cameleon sequence was optimized for mouse codon usage (GeneArt Gene Optimizer service from Life Technology). The codon-optimized version of 4mtD3cpv cDNA sequence, modified as previously reported, 39 was obtained by chemical synthesis (GeneArt Gene Synthesis service from Life Technology). Briefly, a Kozak consensus sequence (GCCACC) was placed before the starting codon and the cDNA sequence encoding the 33 aminoacid-long N-terminal sequence derived from COX VIII was repeated 4 times to guarantee mitochondrial matrix import. The resulting nucleotide sequence (named mt-Cam) is available upon request.

Gateway Vectors Construction and Gateway Reaction
MultiSite Gateway Cloning was used to create a conditional ROSA26-targeting vector. The vector consisted in a CAG promoter floxed bgeo-STOP cassette 59 placed before mt-Cam coding sequence, inserted between a 5 0 and a 3 0 ROSA26 locus homology arm in anti-sense orientation relative to the ROSA26 sense promoter. The resulting pROSA26-mt-Cam vector is depicted in Supplementary Figure 1 and its construction was obtained through the recombination of pROSA26 destination and 3 Entry vectors, as follows.
The attL1-attL2 Entry vector was created by cloning the codon-optimized version of 4mtD3cpv cDNA sequence mt-Cam in pENTR1A (Thermo Fisher Scientific) using restriction enzymes KpnI-XhoI.
MultiSite Gateway LR reaction was performed using Gateway LR Clonase II Enzyme mix (Thermo Fisher Scientific) according to the suppliers' instructions. Briefly, 50 ng of each of the 3 Entry vectors were first incubated with the LR Clonase II enzyme mix for 6 h at room temperature (RT), then 150 ng of pROSA26-DV2 Destination vector was added with additional Clonase II enzyme mix and incubated overnight (O/N) at RT. LR reaction was desalted using 0.025 mm membrane (Millipore) prior to electroporation (1.5 kV) in One Shot TOP10 electrocompetent bacteria (Thermo Fisher Scientific).
Bacteria were recovered in lysogeny broth LB medium without antibiotics at 28 C for 90 min shaking and plated on LB-AMP agar plates also grown at 28 C to avoid recombination. Correctly recombined clones were selected by restriction analysis with KpnI or XhoI, followed by sequencing.
For electroporation, ES cells were harvested by trypsinization with 0.25% trypsin plus 1 mM Ethylenediaminetetraacetic acid (EDTA) (Thermo Fisher Scientific) and washed with PBS (Thermo Fisher Scientific). The cell suspension was diluted with PBS to a final concentration of 0.75 Â 10 7 cellsÁmL À1 . 800 mL cell suspension (0.6 Â 10 7 cells) were mixed with 25 mg of linearized pROSA26-mt-Cam targeting vector in an electroporation cuvette (0.4 cm gap, Bio-Rad Laboratories). Electroporation was performed using a gene pulser II electroporator (Bio-Rad Laboratories), setting 200 V and 500 mF. After electroporation, the cells were seeded onto a 10 cm culture dish containing neomycin-resistant, mitomycin-C inactivated feeder cells. Twenty-four hours after electroporation, G418 (150 mgÁmL À1 ) (Thermo Fisher Scientific) and 4 days after the electroporation, ganciclovir (2 mM) (Sigma-Aldrich) were applied to the ES cells. Nine days after the electroporation, surviving colonies were picked.

PCR and Southern Blot Analysis
Polymerase chain reaction (PCR)-based screening of ROSA26targeted ES cell clones was performed using mt-Cam specific primers (Primer F: 5 0 -TTCCTGTGTGAAAGGGCGAA-3 0 , Primer R: 5 0 -CAGAGAGCCTCGGCTAGGTA-3 0 ) to generate a 1.3 kb PCR fragment. PCR positive clones were confirmed by Southern blotting for 5 0 -integration using a 5 0 -external probe (550 bp) after EcoRI-KpnI digestion of genomic DNA (4.2 kb fragment in correctly targeted allele, 11 kb fragment for the un-targeted wild-type allele). 3 0 -integration was confirmed using a 3 0 -external probe (800 bp) after EcoRI-KpnI digestion of genomic DNA (7.8 kb fragment for correctly targeted allele, 11 kb fragment for the un-targeted wild-type allele). Both the 5 0 -and 3 0 -external Southern probes were generated by Michael Taschner and Christine Hartmann (IMP, Vienna). An internal probe on mt-Cam (800 bp) was used to detect a single 4.9 kb fragment in mt-Cam ROSA26-targeted allele, upon BamHI digestion of genomic DNA. Southern blots were performed following standard procedures, using Digoxigenin (DIG)-High Prime DNA Labeling and Detection Kit (Roche, #11585614910) for random-primed labeling of DNA probes and chemiluminescent detection of the DIG-labeled hybrids.

Blastocyst Injection
Blastocyst injection was carried out by injecting 15-25 correctly targeted ES cells into host blastocysts of C57BL/6J mice. After injection, blastocysts were re-implanted into 2.5-day pseudopregnant Swiss Webster females, previously mated to vasectomized males. Chimeric animals were mated to C57BL/6J mice to generate heterozygous germline offspring. Resulting offspring were genotyped to check for the presence of the targeted insertion.

Mouse Lines
All animal procedures were performed following a protocol approved by the ethic committee of the University of Padua and the Ministry of Health (287/2015-PR). ROSA26-mt-Cam female mice were crossed with male mice expressing Cre recombinase under the control of a cytomegalovirus immediate early enhancer chicken b-actin hybrid (CAG) promoter (CAG-Cre mice 62 ) or the tissues specific promoter of Choline Acetyltransferase (ChAT-IRES-Cre mice, stock #006410 from The Jackson Laboratory). The F1 generation was used for experiments.

PCR Analysis on Genomic DNA
Genomic DNA extraction from mouse tissues was performed with KAPA Express Extract Kit (Kapa Biosystems) following the manufacturer's instruction. PCR reactions were performed with the same kit following KAPA2G Robust PCR Protocol (Kapa Biosystems).
The following primers were used: • to assess the presence of the Cameleon (Primer R:

Quantitative PCR
Total RNA was extracted from 20 mg of each tissue using the NucleoSpin RNA purification kit (Macherey-Nagel, 740955.250) and quantified with NanoDrop 2000 (Thermo Fisher Scientific). For each sample, 1 lg of total RNA was reverse-transcribed into cDNA using the SuperScript II Reverse Transcriptase (Thermo Fisher Scientific, 18064071) following the manufacturer's instructions. PCR amplification was performed in a CFX96 Touch Real-Time PCR Detection System (BioRad), using Applied Biosystems PowerUp SYBR Green Master Mix (Thermo Fisher Scientific, #A25777). The transcript of TOM20 was used as an internal control. The following primers were employed: mt-Cam: 5 0 -TCAGGCACAACATCGAGGAC-3 0 and 5 0 -GTGGTCCCTC TTCTCGTTGG-3 0 TOM20: 5 0 -CTGACAAATGCCATTGCTGT-3 0 and 5 0 -CACATCATCTT CAGCCAAGC-3 0 Data are expressed as percentage change relative to the internal control transcript.

Mouse Dimension and Preparation of Mouse Tissues
The mouse length was measured from the nose to the back of P13 and P75 individuals, using a ruler.
P0 mice ubiquitously expressing mt-Cam were identified upon illumination with a Dual Fluorescent Protein Flashlight lamp (Nightsea). Fluorescent pups were killed by decapitation, brain was extracted, rapidly cut at the vibratome (Leica, VT 1200S) in PBS and visualized by confocal microscopy (Leica TCS SP5 II confocal system).
To mark nuclei, fixed slices were incubated with the nuclear marker TO-PRO 3 (1:10000, Sigma-Aldrich) for 10 min at RT and rinsed in PBS.
Slices were mounted and coverslipped with Mowiol, then imaged using a Leica TCS SP5 II confocal system equipped with a HCX PL Fluotar 20Â/0.50 or a HCX PL APO 100Â/1.40 objective. The Argon laser line (488 nm) was used to excite fluorescent proteins (FPs) of the Cameleon sensor. Confocal microscopy imaging was performed at 1024 Â 1024 pixels per image, with a 100 Hz acquisition rate, average of two frames.
The analysis of the area occupied by mitochondria expressing the probe was performed on 3 slices from 3 animals using ImageJ software (US National Institutes of Health). For each image, a threshold was imposed and the fluorescent area was determined.
Cortical astrocytes were isolated using the same dissociation protocol described above for cortical neurons and plated at a density of 0.2 Â 10 6 cells/coverslip on glass coverslips (18 mm) coated with poly-L-lysine (100 mgÁmL À1 ). Twenty-four hours after dissociation, the growth media was replaced with DMEM with high glucose (Sigma-Aldrich, #5671) with the addition of FBS 10%, L-Glutamine 0.5 mM, penicillin 100 UÁmL À1 , and streptomycin 100 lgÁmL À1 . Astrocytes were imaged after 7 DIV.
Stained neurons, astrocytes or fibers were imaged at a Leica TCS SP5 II confocal system, equipped with a PlanApo 100Â/1.4 objective. The Argon laser line (488 nm) was used to excite FPs of the Cameleon sensors and the He/Ne 543 nm laser was used to excite the Alexa 555 secondary antibody. Confocal microscopy imaging was performed at described above.
In Vitro Ca 21 Imaging For FDB fibers, real-time imaging was performed the day after plating using a DM6000 inverted microscope (Leica, Wetzlar, Germany) with a HCX Plan Apo 40Â/1.25 oil objective. The excitation light, produced by a 405 nm LED (LED Engin, #LZ1-00UA00) was filtered by an ET410/20 band pass filter and 455 dichroic mirror. The emitted light was collected through a custom-made beam splitter (ET480/40 and ET535/30 emission filters for ECFP and cpV, respectively, and a 515DCXR dichroic mirror (Chroma). All filters and dichroic mirrors were from Chroma Technologies. Images were acquired using an IM 1.4C cool camera (Jenoptik Optical Systems) attached to a 12-bit frame grabber. Synchronization of the excitation source and the cool camera was obtained through a control unit run by custom-made software package Roboscope (developed by Ciubotaru C. D., Padua, Italy). Images were acquired at 1 Hz with 100-200 ms exposure time. The coverslip was mounted into an open-topped chamber and maintained in a modified KRB solution of the following composition (in mM): NaCl 135, KCl 5 , MgCl 2 1, HEPES 20, MgSO 4 1, KH 2 PO 4 0.4, CaCl 2 1, glucose 5, pH 7.4 at RT, in the presence of 75 lM BTS (N-benzyl-P-toluene sulphonamide) (TOCRIS, #1870) to avoid fiber contraction. In a typical experiment, after recording basal Ca 2þ level, caffeine (20 mM, Sigma-Aldrich, #C0750) was added to elicit Ca 2þ release from intracellular stores.

Data Analysis and Statistics
Off-line analysis of Ca 2þ imaging experiments was performed using ImageJ software (US National Institutes of Health). Proper regions of interest (ROIs) were selected for each image and fluorescence signals F cpV and F ECFP were subtracted (in vitro experiments) or not (ex vivo experiments) of background. The ratio of the emitted fluorescence intensities (R ¼ F cpV /F ECFP ) was calculated for each ROI and each timeframe. Data are presented as the relative change in F or R normalized to the value measured before stimulus addition (DF/F 0 ; DR/R 0 ).
Data were analyzed using Microsoft Excel. Values are expressed as mean 6 SEM with n ¼ number of animals in immunohistochemistry, or SD with n ¼ number of lanes loaded for western blotting. Statistical analyses were performed using unpaired Student's t-test with a confidence interval of 95% (*p < 0.05, **p < 0.01, and ***p < 0.001).

Generation of the Transgenic ROSA26-mt-Cam Mouse Line
In our effort to improve current strategies for in vivo expression of Ca 2þ sensors, we developed a transgenic mouse line for the controlled expression of a mitochondria-targeted probe. Among others, we selected the FRET-based Cameleon 4mtD3cpv because it allows ratiometric imaging, with the great advantage of reducing sensitivity to motion artifacts and probe expression levels. To avoid toxicity due to constitutive expression of the exogenous protein and to guarantee the possibility of accessing specific cell types using selected promoters, we exploited a well-established Cre/loxP recombination system. 63 We combined this approach with site-directed transgenesis at the ROSA26 locus, in order to prevent the issue of unpredictable expression due to the effect of random genomic integration site or transgene copy number. 64 Indeed, this site provides high targeting efficiency and transgene transcription at adequate levels. 60,65 When directed by a CAG promoter, expression is observed in all organs/tissues analyzed, although the proportion of positive cells substantially varies among them. 59,66 A MultiSite Gateway approach was exploited to build a conditional ROSA26targeting vector for transgenesis. 59 The vector consisted of the CAG promoter 67 floxed bgeo-STOP cassette 59 followed by mt-Cam, a codon-optimized version of 4mtD3cpv cDNA sequence. The obtained targeting vector ensures the insertion of mt-Cam sequence into the ROSA26 locus in anti-sense orientation, relative to the ROSA26 promoter ( Figure S1A). Anti-sense orientation was chosen because it was previously shown to yield higher levels of expression, compared to the sense. 68 The vector was used for ES cell electroporation. Correct ROSA26 locus targeting was screened by Southern blot analysis of genomic DNA extracted from surviving colonies using 5 0 (550 bp) and 3 0 (800 bp) external probes together with an internal probe designed on mt-Cam sequence (800 bp) to confirm 5 0 -and 3 0 -integration, as well as single-copy insertion ( Figure S1B). Correctly targeted ES cells were injected into host blastocysts of C57BL/6J mice, which were re-implanted into pseudo-pregnant Swiss Webster females. Chimeric animals were mated to C57BL/6J mice to generate heterozygous germline offspring. Resulting offspring were genotyped to check for the presence of mt-Cam sequence.

Ubiquitous Expression of mt-Cam
The ROSA26-mt-Cam mice obtained were crossed with mice ubiquitously expressing Cre recombinase (CAG-Cre mice 62 ; Figure S1C) and positive offspring was selected by PCR ( Figure  S1D). PCR analysis on genomic DNA confirmed the complete excision of the bgeo-STOP cassette in male and female offspring positive for both mt-Cam and Cre ( Figure 1A and B). Noteworthy, the probe is expressed in early stages of development and it is easy to score also in intact P0-P1 pups observed under a fluorescence lamp or microscope ( Figure 1E). However, when observed at the fluorescence microscope, high variability of fluorescence intensity level was observed among animals, even siblings ( Figure 1E; Figure S1E). Moreover, we observed that the expression of the probe is variable in the different tissues analyzed ( Figure 1C). Western blotting analysis of proteins extracted from different P30 organs confirmed the presence of variable amounts of the protein ( Figure 1D). Significantly low expression was observed in some organs, for example, in the liver. The results were confirmed by fluorescence analysis of organ slices, showing that in liver sections the average area occupied by the fluorescent signals, which gives an estimate of the percentage of positive cells, was calculated as 0.9 6 0.4% (mean 6 SD) ( Figure S2A and S2B), while in brain slices it was 18.5 6 5.1% (mean 6 SD) ( Figure S2C and S2D). Differences in protein expression were recapitulated when mRNA relative expression was estimated by qPCR ( Figure S1F) and are consistent among animals (data not shown).
Confocal imaging revealed the presence of positive fluorescent cells in all tissues/organs analyzed, e.g., brain (Figure 2A-a), skeletal muscles ( Figure 2B Figure 2H-h). Moreover, to check whether the expression of the probe is stable during development and aging, brain slices from mice at different ages, from embryonic stage E10 to P360, were analyzed at the confocal microscope. The expression of the probe is maintained during aging ( Figure 2I).
The proper mitochondrial localization of the expressed probe was verified in the tissues analyzed. As an example, Figure 3 shows the immunohistochemical analysis of fixed pancreatic and brain slices stained with the mitochondrial marker TOM20. Noteworthy, the expression of the Cameleon probe does not alter the aspect and size of mice, which were similar to control littermates ( Figure S1G).

Mitochondrial Ca 21 Dynamics in the Central Nervous System of mt-Cam-expressing Mice
We then focused our analysis on the nervous system, where, at present, expression of mitochondria-targeted GECIs relies on viral infection. 39 Immunohistochemical analysis of brain slices from positive animals, using neuron-specific NeuN or astrocyte-specific GFAP staining, revealed that both neurons and astrocytes expressed the fluorescent sensor ( Figure 4). The Cameleon signal was clearly excluded from the nucleus (labeled by NeuN, in cortical neurons) and reached the finest processes ( Figure 4).
To investigate the ability of mt-Cam of reporting mitochondrial Ca 2þ dynamics in neurons and astrocytes, we first set up a series of in vitro experiments. Primary neuronal cultures were prepared from the cortex or hippocampus of P0-P1 mice. The mitochondrial localization of mt-Cam was further confirmed by loading isolated cortical neurons ( Figure S3A) with the cell-permeant dye TMRM, which is selectively sequestered by polarized mitochondria (Figure 5A). To verify whether ROSA26-mt-Cam mice express a functional probe, we performed Ca 2þ imaging experiments on isolated hippocampal and cortical neurons. When Cameleon binds Ca 2þ , it undergoes a conformational change that increases FRET and, accordingly, we observe an increase in the 540 nm and a decrease in the 480 nm fluorescence (F) intensities, which result in an increased F 540 /F 480 ratio. Both cortical ( Figure 5A-C) and hippocampal neurons ( Figure S4) were challenged with NMDA plus glycine to maximally activate NMDA receptors ( Figure 5B; Figure S4A) or with a mix of stimuli, i.e. DHPG, carbachol, ATP, and glutamate ( Figure 5C; Supplementary Figure 4B), in the absence of external Ca 2þ and in the presence of EGTA, in order to induce maximal store depletion. In both cases a clear decrease of the ECFP emission signal (480 nm) and a concomitant increase of the cpV signal (540 nm) were observed ( Figure 5B and C; Figure S4). In the latter experiment, the increase in the F 540 /F 480 ratio is exclusively due to mitochondrial uptake of Ca 2þ released from internal stores, thus resulting in a sharper peak, compared to that caused by NMDA plus glycine, whose Ca 2þ rise is dominated by Ca 2þ influx across the PM. After the peak, mitochondrial [Ca 2þ ] returned rapidly to the basal level ( Figure 5C; Figure S4B).
Similar results were obtained on primary cortical astrocytes isolated from P0-P1 animals. The presence of positive astrocytes is confirmed by the immunostaining with the specific marker GFAP ( Figure S3B). TMRM loading confirmed the proper localization of the probe also in this cell type ( Figure 5D). When stimulated with the purinergic agonist ATP, astrocytes showed the expected FRET response ( Figure 5E), which was abolished if cells were pre-treated with the mitochondrial uncoupler FCCP ( Figure 5F), further confirming the exclusive mitochondrial localization of the probe. A small drop is detected in ECFP and cpV fluorescence upon addition of FCCP ( Figure 5F, upper panel, arrow), likely due to a change in pH and/or to mitochondrial swelling. Notably, this is efficiently corrected by ratio calculation ( Figure 5F, lower panel, arrow).
To extend the validity of our approach, we analyzed mitochondrial Ca 2þ dynamics of neurons in acute brain slices. We performed ex vivo two-photon Ca 2þ imaging experiments in coronal slices of P40 mt-Cam-expressing mice ( Figure 5G). After recording the basal ratio at the cortical level, we perfused a mix of stimuli (ATP, DHPG, carbachol, and glutamate) aimed at activating both ionotropic and metabotropic receptors, thus triggering major mitochondrial [Ca 2þ ] increases. Again, Ca 2þ entry into the mitochondrial matrix was readily revealed by a decrease of the ECFP signal and a concomitant increase of the cpV signal, causing an increase in F 540 /F 480 ratio R ( Figure 5H).
Finally, as a proof of principle example of tissue-specific expression of the probe, ROSA26-mt-Cam mice were crossed with ChAT-Cre mice, expressing Cre recombinase under the control of the ChAT promoter. Analysis of sagittal brain sections confirmed the presence of clusters of cholinergic neurons expressing the mt-Cam probe, mostly localized in the midbrain, close to the cerebellum ( Figure S5).

Mitochondrial Ca 21 Dynamics in FDB Muscle Fibers Isolated from mt-Cam Mice
Increases in mitochondrial matrix [Ca 2þ ] are crucially important in skeletal muscle where, by stimulating Ca 2þ -sensitive key enzymes such as pyruvate dehydrogenase, 19,20,70 they ensure the optimal production of ATP enabling the cross-bridge cycle of actomyosin. Currently, the use of GECIs to measure mitochondrial Ca 2þ in fibers requires great expertise, since it relies on in vivo plasmid DNA electroporation. When ROSA26-mt-Cam mice are crossed with CAG-Cre mice, 4mtD3cpv is expressed in both cardiac and skeletal muscles in the offspring (Figure 2). We thus isolated FDB muscle fibers from mt-Cam expressing mice and verified the correct mitochondrial localization with anti-TOM20 antibody staining ( Figure 6A). In the isolated muscle fibers, probe functionality was assessed by applying to the fibers caffeine, a ryanodine receptor agonist, in order to induce Ca 2þ release from the SR, rapidly triggering mitochondrial Ca 2þ uptake ( Figure 6B). The same type of experiment was then carried out in the absence of extracellular Ca 2þ and in the presence of EGTA, thus avoiding the influx of Ca 2þ across the PM ( Figure 6C). When cells were pre-treated with FCCP, the mitochondrial response to caffeine stimulation was completely abolished ( Figure 6D).

Discussion
The pivotal role of Ca 2þ signaling in regulating central aspects of physiology and pathology, as well as the importance of mitochondria in this signaling, is widely accepted. 13,20,23,[25][26][27][28][29]71 Mitochondrial Ca 2þ dynamics can now be routinely studied in cell cultures thanks to an extended palette of recently developed chemical 35 and genetically encoded probes. 4,72,73 In contrast, only a few transgenic mouse lines have been developed expressing mitochondria-targeted Ca 2þ sensors. 53,54 These latter are nonratiometric, bioluminescence-based sensors that suffer from low spatial resolution and are not suitable to perform single cell mitochondrial Ca 2þ imaging. As a result, in vivo/ex vivo experiments still rely on viral injection or in situ electroporation, with obvious drawbacks such as limited access to areas of interest, laborious procedures, and potential toxicity.
Single fluorophore-based indicators GCaMPs are widely used to study Ca 2þ dynamics in living organisms and several transgenic mouse line expressing cytosolic versions of these probes have been created 46,48 (https://www.jax.org; https:// www.janelia.org). Although the dynamic range of GCaMPs is in general higher, FRET-based indicators appear superior to correct movement artifacts, photobleaching, pH sensitivity and, most relevantly, they allow the quantitative estimation of [Ca 2þ ]. The insertion of a FRET-based Cameleon probe in the ROSA26 locus under the control of a Cre/loxP expression system, allowed the generation of mice expressing a FRET-based probe upon a single cross leading to Cre-mediated excision. The great potential of Cre/loxP expression system resides in the possibility of controlling the transgene expression in a temporal and cellspecific manner. 63 Indeed, the transgenic model we generated allows the expression of a mitochondria-targeted Ca 2þ sensor in regions otherwise difficult to inject/electroporate, such as deep brain structures or pancreas, upon crossing with a transgenic mouse expressing Cre recombinase under the control of a generic or cell-specific promoter.
As a proof of principle of the functionality of the tool generated, a ubiquitous Cre (CAG-Cre) was exploited to drive the expression of the Cameleon probe. The protein was detected in all the tissues/organs analyzed, throughout development and aging, confirming a widespread and stable expression of the probe. The CAG promoter is one of the most active promoters 74 ; however, its high variability in transgene expression in vivo has been reported, 75 even when inserted in the ROSA26 locus. 59,66 Our results confirm this observation, indeed we found an uneven expression of the probe within the analyzed tissues and among animals, even if littermates.
The expressed Cameleon is correctly localized in mitochondria of tissues from P0 to adulthood. Two-photon Ca 2þ imaging experiments in ex vivo brain slices demonstrated the proper functionality of the probe. Moreover, we also demonstrated the feasibility of the generated tool for in vitro imaging. Indeed, Ca 2þ imaging on primary cultures is often challenging due to poor transfection of some dissociated cells (e.g., neurons) or the need of electroporation (e.g., muscle fibers). Here we show that cultured neurons or astrocytes from newborn mice and muscle fibers from adult mice display a robust expression of the Cameleon probe and can thus be readily imaged without further procedures, that is viral infection. Noteworthy, the decay of matrix Ca 2þ transient we recorded after ATP stimulation of isolated astrocytes is similar to that previously observed in isolated astrocytes transfected with 4mtD3cpv, 76 suggesting the absence of adaptive effects on matrix Ca 2þ dynamics arising in the transgenic animal. Crossing with cholinergic neuron-specific ChAT-Cre mice demonstrated that ROSA26-mt-Cam mice can easily be used for studying specific subpopulations of cells, upon crossing with any specific Cre mice line of choice.
We created the first line of transgenic mice of our knowledge that allows expression in a cell type of choice of a ratiometric Ca 2þ probe, targeted to mitochondria. The great advantage of our system is its high versatility: Cre recombinase expression under the control of specific promoters allows the expression of the probe in virtually any tissue or cell type, ensuring the possibility of performing in vitro, ex vivo, and in vivo Ca 2þ imaging experiments (alone or in combination with other techniques, e.g., electrophysiology, optogenetics), both in physiological and pathological conditions.

Supplementary Material
Supplementary material is available at the APS Function online.