Generation and validation of novel adeno-associated viral vectors for the analysis of Ca2+ homeostasis in motor neurons

A finely tuned Ca2+ homeostasis in restricted cell domains is of fundamental importance for neurons, where transient Ca2+ oscillations direct the proper coordination of electro-chemical signals and overall neuronal metabolism. Once such a precise regulation is unbalanced, however, neuronal functions and viability are severely compromised. Accordingly, disturbed Ca2+ metabolism has often been claimed as a major contributor to different neurodegenerative disorders, such as amyotrophic lateral sclerosis that is characterised by selective motor neuron (MN) damage. This notion highlights the need for probes for the specific and precise analysis of local Ca2+ dynamics in MNs. Here, we generated and functionally validated adeno-associated viral vectors for the expression of gene-encoded fluorescent Ca2+ indicators targeted to different cell domains, under the transcriptional control of a MN-specific promoter. We demonstrated that the probes are specifically expressed, and allow reliable local Ca2+ measurements, in MNs from murine primary spinal cord cultures, and can also be expressed in spinal cord MNs in vivo, upon systemic administration to newborn mice. Preliminary analyses using these novel vectors have shown larger cytosolic Ca2+ responses following stimulation of AMPA receptors in the cytosol of primary cultured MNs from a murine genetic model of ALS compared to the healthy counterpart.


Results
Given the plausible implication of perturbed Ca 2+ homeostasis in the pathogenesis of MN diseases [6][7][8] , in this study we sought to generate and validate expression systems for GECI that were suited for the assessment of local Ca 2+ fluctuations in MNs. To this purpose, we have engineered AAV plasmids (pAAV) for the expression of cameleon probes targeted to the cytosol (pAAV-[Hb9_AB]-D1cpv), the mitochondrial matrix (pAAV-[Hb9_ AB]-4mtD3cpv) and the ER lumen (pAAV-[Hb9_AB]-D4ER), under the control of a MN-specific, homeobox Hb9-derived, promoter ( Supplementary Fig. S1). The cameleon probes of choice have great ratiometric sensitivity and large dynamic range, thereby allowing to detect small changes in Ca 2+ concentration over the noise in the target compartment 13 .
To validate such vectors for the specific recording of Ca 2+ fluxes in MNs, we firstly analysed the expression of our AAV-driven probes in the immortalised NSC-34 cell line that -when properly differentiated -displays several typical properties of MNs 16,17 , including the transcriptional activation of the Hb9 gene 18 . We therefore checked the expression of the three cameleon probes in NSC-34 cells, transduced with the AAV vectors, either cultured under proliferating conditions or induced to differentiate by treatment with retinoic acid. Under the latter culturing conditions, all cells were successfully differentiated into a MN phenotype, as determined by both morphological observations (Fig. 1, bright-field images of panels D,H,L) and immunoblot assessment of the MN marker choline acetyl-transferase ( Supplementary Fig. S2). We observed that all cameleons were abundantly present in differentiated cells (>97% cells expressing the probes), but completely absent in cells under active proliferation, suggesting that the Ca 2+ probes were specifically expressed in cells resembling a MN phenotype (Fig. 1).
We then analysed by confocal microscopy the expression of cameleons in primary cultures from mouse spinal cord. After 12 days of growth, such cultures contained different cell types, including mature MNs resembling those present in vivo, both morphologically and for the expression of specific molecular markers, such as the neurofilament protein SMI32, with 4.9% ± 0.7% of SMI32-positive MNs over total cells (n = 35). As described in Fig. 2, after transduction of such cell cultures with the AAV vectors, the cameleons (green signal, panels A,E,I) were present in SMI32-positive cells (red signal, panels B,F,J), while in no case other cell types (highlighted by nuclear staining, blue signal, panels C,G,K) were positive for the expression of cameleons, providing evidence that the Hb9 promoter-driven expression of the Ca 2+ probes is specific for MNs (merge, panels D,H,L). In addition, the transduction efficiency of MNs was rather high, given that in all cases more than 70% of SMI32-positive cells were also positive for the expression of the cameleon probes ( Supplementary Fig. S3).
The MN-specific expression of the Ca 2+ probes was further reinforced by the finding that the AAV vectors were unable to drive the expression of the mitochondria-targeted cameleon in primary cultures of cortical, hippocampal or cerebellar granule neurons, or spinal astrocytes ( Supplementary Fig. S4).
Although the correct sub-cellular targeting of the used cameleon chimeric constructs has been already demonstrated in different cell paradigms 14,19,20 , we checked that the probes had the expected localisation also in primary MNs. As shown by the confocal microscopy z-stack reconstructions of Fig. 3, each cameleon (panels A,D,G) largely co-localised (yellow signal, merged images of panels C,F,I) with immuno-labelled markers of the corresponding target compartment, i.e., glyceraldehyde 3-phosphate dehydrogenase (GAPDH) for the cytosol (panel B), Tom20 for mitochondria (panel E), and calreticulin for the ER lumen (panel H), indicating that the probes are correctly processed and delivered to the target site in AAV transduced MNs.
ALS is a fatal neurodegenerative disorder that occur on sporadic or genetic grounds, leading to a selective and progressive loss of MNs in the spinal cord, brainstem and cerebral cortex, thereby resulting in severe motor deficits and -eventually -to death 21,22 . Since perturbations of Ca 2+ homeostasis in MNs have been often claimed as a crucial event in disease progression 7, 23 we tested the applicability of the novel Ca 2+ -probe encoding vectors to the comparative analysis of local Ca 2+ homeostasis in MNs from an ALS mouse model [i.e., mice expressing the ALS-related G93A mutant of human (h) SOD1 and the healthy counterpart (hSOD1(WT)-expressing mice)].
In this work we did not calculated absolute Ca 2+ concentrations, which, however, could be determined by suitable calibration procedures 13 . Instead, we simply reported the ratio between the acceptor and the donor fluorescence of the FRET-based probes, which is, nevertheless, sufficient for comparative analyses. Such experiments demonstrated that all cameleon probes were functional, and indicated that hSOD1(G93A)-expressing MNs have larger cytosolic Ca 2+ transients following AMPA stimulation, compared to SOD1(WT) MNs (Fig. 4A,B). Comparable results for both basal and peak Ca 2+ levels were obtained using the chemical Ca 2+ indicator Fura-2 ( Supplementary Fig. S5), further supporting the suitability of the AAV-mediated, Hb9_AB-driven, expression of the cameleon probes for reliable Ca 2+ measurements in MNs. Conversely, the mitochondrial targeted cameleon revealed no significant difference in AMPA-stimulated Ca 2+ fluxes in the mitochondrial matrix (Fig. 4C,D). Also the ER-targeted cameleon was properly functioning, allowing the measurement of ER Ca 2+ discharge upon stimulation with caffeine, which, however, evidenced no difference between MNs with the two genotypes ( Fig. 4E,F).
Finally, we probed if the developed AAV-based vectors were suited for the in vivo expression of the cameleon probes. To this purpose, we injected the AAV vector coding for the mitochondrial or the ER cameleon into the superficial temporal vein of newborn mice, and (4 weeks later) we evaluated the expression of the probes in spinal cord sections. By use of a fluorescence stereo-microscope, we observed an intense and diffuse signal in tissue samples of mice transduced with either the mitochondrial ( Fig. 5A-C) or the ER (Fig. 5D-F) cameleon, providing a transduction ratio (cameleon-positive cells over total cells) of 16% ± 4% and 9.8% ± 1.7%, respectively (n = 3). That the AAV-based expression system of cameleons was specific for MNs also in vivo was demonstrated by the immunostaining of the MN marker SMI32 in spinal cord slices from ER cameleon-infected mice, followed by confocal microscopy (Fig. 5G-J).

Discussion
Recombinant AAVs are a powerful means for in vivo gene delivery 24,25 and for the transduction of primary cell cultures that are particularly hard to transfect, such as neurons 26,27 . AAVs display low cytotoxicity and immunogenicity, are safe for the operator, and are suitable for long term gene expression in non-dividing cells with relatively high gene delivery efficiency, depending on the cell type 28 .
In this work, we have generated pAAV plasmids encoding cameleon Ca 2+ probes, genetically targeted to different cell compartments, under the transcriptional control of a MN-specific promoter. Given that AAV vectors has limited cloning capacity (i.e., large plasmids can hardly be packaged into AAV viral particles), we have chosen a genetically engineered minimal promoter of the Hb9 gene (Hb9_AB), which is about 550 bp in length and has already been demonstrated to specifically drive the expression of a transgene into MNs, both in vitro and in vivo 29 . For the preparation of the viral particles, we have chosen the AAV9 serotype because of its broad tropism, including the capacity to target neurons 28 . We have demonstrated that the engineered AAV vectors are suitable for the specific expression of the cameleons targeted to the cytosol, the mitochondrial matrix and the ER lumen of MNs from spinal cord primary cultures of mice, and allow measuring Ca 2+ responses to proper stimuli in such cellular districts. These vectors have several advantages with respect to other expression systems or the use of ubiquitous promoters. Indeed, they allow Ca 2+ measurements in single primary MNs with no need for difficult Although SMI32-positive MNs were less than 5% of total cells in our primary spinal cord cultures, the transduction efficiency ( >70%) of MNs by the AAV-mediated infection system was largely sufficient for single cell analyses. Indeed, the transduction rate was much higher than that achieved by transfection with plasmidic vectors, with which cameleon-expressing MNs were rarely found in each preparation, also when using a constitutive and ubiquitous (cytomegalovirus), or the pan-neuronal human synapsin1 (hSyn), promoter (data not shown). In addition, in our experimental setting, a substantial number of SMI32-positive cells contained the Ca 2+ probes at an expression level that easily allowed single cell recordings of Ca 2+ dynamics.
Among GECI, the ratiometric cameleon probes have several advantages, including low risk of artifacts, high sensitivity, good dynamic range, and the possibility to provide absolute Ca 2+ concentration values upon proper calibration procedures 13 . The ratiometric nature of these Ca 2+ indicators, as in the case of chemical probes such as Fura-2, prevents any differences due to level of expression or cell thickness. In addition, engineering of mutant cameleons have provided researcher with several cameleons displaying a wide range of K d for Ca 2+ , thereby allowing reliable measurements of basal Ca 2+ levels and evoked Ca 2+ movements depending on the target cell compartment, the cell type and the desired stimulation protocol 14,19 .  . AMPA stimulation results in higher cytosolic Ca 2+ transients in primary spinal cord MNs from ALS mice than in the healthy counterpart. Primary spinal cord cultures from hSOD1(WT) or hSOD1(G93A) mice were transduced with the AAV vectors coding for the cameleon probes targeted to the cytosol (panels A,B), the mitochondrial matrix (panels C,D) or the ER lumen (panels E,F). After 12 days of culturing, FRET measurements were performed on single MNs expressing the different Ca 2+ probes using a computer-assisted fluorescence microscope equipped with a suitable system for double-wavelength recordings. The ratio (R) between the FRET acceptor and donor was calculated by the data acquisition software, allowing the comparison of Ca 2+ mobilisation following stimulation with AMPA (25 μM in the presence of 2 mM CaCl 2 , for the cytosol and the mitochondrial matrix) or caffeine (100 μM, for the ER) at the time-points indicated by arrows, between hSOD1(WT) or hSOD1(G93A) MNs. Panels A, C, E report average traces of the Ca 2+ dynamics (for the sake of clarity, error bars are not reported), while bar diagrams of panels B,D,F report the mean difference between FRET ratios (ΔR = R peak − R baseline ) following the indicated stimulus in the three cell compartments. While no difference was recorded in Ca 2+ movements in the mitochondrial matrix or the ER lumen, the cameleonbased approach highlighted a significantly higher Ca 2+ response following AMPA stimulation in the cytosol For cytosolic Ca 2+ measurements we have used the archetypal D1-based probe 30 , which allowed us to measure cytosolic fluctuations of the ion in MNs following AMPA stimulation.
Mitochondria-Ca 2+ interplay is essential for cell physiology. On the one hand, the fine regulation of mitochondrial Ca 2+ controls ATP production, because different mitochondrial enzymes rely on local Ca 2+ concentration for their activity. On the other hand, mitochondria are a major Ca 2+ -buffering system avoiding noxious cytosolic Ca 2+ overloads. While it is assumed that resting mitochondrial Ca 2+ concentration is in the order of 10 −1 μM, upon stimulation promoting Ca 2+ entry from the extracellular space or its release from the ER, Ca 2+ levels in the mitochondrial matrix may reach up to 10 2 μM, depending on the cell type and the stimulus 31 . Therefore, we have selected the 4mtD3cpv cameleon that, for its K d and dynamic range, has been proved useful for measuring mitochondrial Ca 2+ dynamics 13 .
Finally, for the measurements of Ca 2+ homeostasis in the ER lumen, we have chosen the recently generated D4ER probe 14 that has been proved useful for comparative Ca 2+ measurements in an experimental setting of neurodegenerative disorders 32 .
Notably, cameleons are mostly expressed in SMI32-positive α-MNs, enriched in the anterior ventral horn of the spinal cord 33,34 , which are particularly vulnerable to (AMPA-mediated) excitotoxic challenge and ALS-related damage 6,7 . Thus, to prove the efficacy of the probes to monitor local Ca 2+ fluctuations in MNs, we have compared the response of MNs from mice expressing the ALS-related hSOD1(G93A) mutant, or the hSOD(WT)-expressing counterpart, to selected stimuli. In particular, because MNs express high levels of AMPA receptors through which they process strong glutamatergic inputs, and because excitoxicity has been proposed to contribute crucially to MN injury in ALS, probably as a consequence of glutamate-triggered Ca 2+ overload 7, 35 , we investigated Ca 2+ responses to AMPA stimulation. The cytosolic and mitochondrial probes allowed the assessment of AMPA-mediated cell Ca 2+ responses, thereby providing proof-of-principle for the functional application of these tools. Such conclusion is further corroborated by the finding that Fura-2-based measurements provided very similar results for cytosolic Ca 2+ responses to AMPA stimulation. In addition, both cameleon-and Fura-2-based strategies underscored a larger cytosolic Ca 2+ response in ALS MNs compared to the healthy controls supporting the idea that Ca 2+ overloads may be involved in ALS MN damage 7,36 . Surprisingly, in spite of increased cytosolic Ca 2+ load, no difference was observed in the mitochondrial matrix, suggesting a possible deficit of mitochondrial Ca 2+ upload in ALS MNs that deserves further investigation. With this respect, it must be noted that, to the best of our knowledge, no direct measuring of mitochondrial Ca 2+ fluctuations under our experimental settings has been previously reported, and that mitochondrial Ca 2+ handling deficits could be stage-dependent during ALS disease course 37 . We also provided evidence that the ER cameleon was properly functioning and monitored the ER Ca 2+ discharge promoted by caffeine 38 , although no difference was observed between the two SOD1 genotypes. These notions further highlight the importance of the here-presented tools for deepening our understanding of disease-associated local Ca 2+ alterations in MNs, in in vitro and in vivo models.
In conclusion, the novel vectors we generated in this work allowed for the first time to directly monitoring local Ca 2+ fluctuations in sub-cellular compartments of primary mouse MNs, thereby providing valuable and versatile tools for expanding our knowledge on Ca 2+ -related pathogenic routes in ALS and other neuromuscular diseases. With this respect, it is worth underlining that other GECI-encoding sequences can be easily cloned into the generated MN promoter-containing pAAV scaffold, thus further increasing the toolset of Ca 2+ probes with different properties (e.g., K d ) for suitable Ca 2+ measurements under different experimental protocols in MNs.
Importantly, we also provided evidence that the AAV vectors can drive the selective in vivo expression of the Ca 2+ probes in MNs. This result was achieved by systemic administration of the virus, which is by far less invasive and arduous than local injection in the spinal cord of live mice. Thus, by means of suitable fluorescence microscopy equipment (e.g., two-photon microscopy), this system could be used for Ca 2+ measurements in more physiologic environments than primary cultures, such as tissue slices or even live animals.

Materials and Methods
Plasmid construction. Multiple pAAVs have been generated, for the expression of the FRET-based cameleon Ca 2+ probes targeted to different cellular compartment (i.e. cytosol, mitochondrial matrix and ER lumen 13,14,19,20 under the control of either a minimal promoter (Hb9_AB) derived from the (MN)-specific promoter of the Hb9 gene 29 , or the pan-neuronal hSyn promoter.
For the generation of the ER lumen-targeted expression vector, the sequence coding for the D4ER cameleon 14 was amplified by PCR using the pcDNA3-D4ER plasmid (kindly provided by Dr. Paola Pizzo, Dept. of Biomedical Science, Univ. of Padova) as template, and the primers ER-KpnI-F and ER-HindIII-R (sequences are reported in the Supplementary Information). The PCR product was digested with KpnI-HindIII, and purified DNA fragments were ligated into the KpnI-HindIII sites of the pAAV-[hSyn]-ChR2/EYFP as indicated above, generating the pAAV-[hSyn]-D4ER plasmid ( Supplementary Fig. S1, panel B).
of MNs expressing the ALS-related hSOD1(G93A) mutant compared to the hSOD1(WT) counterpart. Data are reported as mean ± standard error of the mean (SEM); n = 6 (A and B), 12 (C and D), 9 (E and F), for each genotype, from at least 3 different MN cultures for each condition; *p < 0.01, Student's t-test. After transformation of E. coli TOP10 cells with the generated constructs and selection of positive clones, recombinant plasmids were amplified and validated by sequencing. All enzymes were from New England Biolabs, excepted the high-fidelity DNA polymerase (KAPA Biosystems), and DNA primers (Life Technologies). DNA manipulations have been performed accordingly to standard methods 39 and to manufacturer's instructions. Detailed maps and plasmids sequences are all available upon request to the corresponding author.
Animals. Tg mice expressing WT, or the ALS-related G93A mutant of hSOD1, (B6SJL(Tg-SOD1)2Gur/J and B6SJL(Tg-SOD1*G93A)1Gur/J mice, respectively) were purchased from The Jackson Laboratories. The colonies were maintained by breeding hemizygote (Tg) males to wild-type B6SJLF1/J hybrid females. Embryos and newborns were genotyped (as described in Supplementary Information), and used for subsequent experiments. All aspects of animal care and experimentation were performed in compliance with European and Italian (D.L. Molecular Medicine, Padova, Italy), which was also used for image acquisition, with an exposure time of 100 ms. The software recorded with-time variations (sampling rate = 1 s −1 ) of the ratio (R) between the acceptor and the donor fluorescence intensity, which was taken as a relative measurement of Ca 2+ concentration. Peak values of Ca 2+ transients were reported as the difference between the peak and the baseline R value (ΔR = R peak − R baseline ).
In vivo AAV-mediated delivery of cameleons. For in vivo delivery of AAV vectors, we have used P1 mice, in light of previous reports demonstrating much higher transduction rates in newborns than in adult mice 43,44 . After anaesthesia by topical administration of lidocaine, neonatal hSOD1(WT)-expressing mice (P1) received 50 μl of viral suspension containing 3 × 10 11 GC/ml of AAV9-Hb9_AB-4mtD3cpv or AAV9-Hb9_AB-D4ER, or vehicle (PBS), into the temporal vein using a Hamilton syringe with a 32-gauge needle. The injections were performed in a Biosafety Level 2 (BL2) laboratory.
Four weeks after the injection, animals were anesthetized by CO 2 before killing by cervical dislocation, and spinal cords were collected and successively fixed in PFA [24 h, 4 °C, 4% (w/v)]. Fixed tissues were transferred to sucrose [30% (w/v) in PBS, overnight, 4 °C] for cryoprotection. 20 μm-thick sections were serially cut on a cryostat (Leica Microsystems), immunostained for the MN marker SMI32 and counter-stained with the nuclear dye Hoechst 33342, as described previously for immunocytochemical analyses. Finally, slices were mounted in montage solution on a glass coverslip, and observed with a fluorescence stereo-microscope (Leica M205-FA), or with the above described inverted confocal microscope system.

Statistical analysis.
Off-line analysis of FRET data was performed with the ImageJ software. YFP and CFP images were subtracted for the respective background signals, and distinctly analysed after choosing proper regions of interest on selected cells. Subsequently, the ratio between YFP and CFP emission fluorescence intensity (F) was calculated (R = F 535 /F 480 ) and reported as %. All the data are representative of at least n (indicated in the figures and/or figure legends) independent experiments, and are reported as mean ± standard error of the mean (SEM). Statistics were performed by unpaired Student's t test, with a p value < 0.05 being considered statistically significant.
For mouse genotyping, NSC-34 cell culturing, and methods for Western blot analysis and Fura-2-based Ca 2+ measurements, see Supplementary Information.