Nicotine protects rat hypoglossal motoneurons from excitotoxic death via downregulation of connexin 36

Motoneuron disease including amyotrophic lateral sclerosis may be due, at an early stage, to deficit in the extracellular clearance of the excitatory transmitter glutamate. A model of glutamate-mediated excitotoxic cell death based on pharmacological inhibition of its uptake was used to investigate how activation of neuronal nicotinic receptors by nicotine may protect motoneurons. Hypoglossal motoneurons (HMs) in neonatal rat brainstem slices were exposed to the glutamate uptake blocker DL-threo-β-benzyloxyaspartate (TBOA) that evoked large Ca2+ transients time locked among nearby HMs, whose number fell by about 30% 4 h later. As nicotine or the gap junction blocker carbenoxolone suppressed bursting, we studied connexin 36 (Cx36), which constitutes gap junctions in neurons and found it largely expressed by HMs. Cx36 was downregulated when nicotine or carbenoxolone was co-applied with TBOA. Expression of Cx36 was preferentially observed in cytosolic rather than membrane fractions after nicotine and TBOA, suggesting protein redistribution with no change in synthesis. Nicotine raised the expression of heat shock protein 70 (Hsp70), a protective factor that binds the apoptotic-inducing factor (AIF) whose nuclear translocation is a cause of cell death. TBOA increased intracellular AIF, an effect blocked by nicotine. These results indicate that activation of neuronal nicotinic receptors is an early tool for protecting motoneurons from excitotoxicity and that this process is carried out via the combined decrease in Cx36 activity, overexpression of Hsp70 and fall in AIF translocation. Thus, retarding or inhibiting HM death may be experimentally achieved by targeting one of these processes leading to motoneuron death.

Amyotrophic lateral sclerosis (ALS) is characterized by motoneuron death in the brainstem and spinal cord. The etiopathology seems to involve a complex interplay among pathogenic factors such as glutamate-mediated excitotoxicity, reactive oxygen species (ROS) and mitochondrial dysfunction. 1 The brainstem nucleus hypoglossus is often affected early, thus leading to dysarthria and dysphagia. 2 Hypoglossal motoneurons (HMs) are very vulnerable because of their high basal intracellular Ca 2+ (ref. 3), expression of Ca 2+ permeable AMPA receptors 4 and low levels of glutamate transporters. 5 In ALS, these properties are exacerbated by impaired glutamate transport 5,6 and increased level of glutamate in cerebrospinal fluid 7 making excitotoxicity an important causative candidate. Motoneuron death shows patchy distribution and slow progression 8 associated with muscle weakness and fasciculations. 9 In the past few years, our laboratory has developed a simple in vitro model of excitotoxic stress applied to rat HMs. [10][11][12] It consists of mimicking the pathological process of motoneuron hyperexcitability and excitotoxicity in the nucleus hypoglossus through pharmacological inhibition of glutamate uptake with DL-threo-β-benzyloxyaspartate (TBOA). Salient characteristics of this model are generation of electrical bursting among HMs interconnected via gap junctions 12 and slow onset of cell death 11 via production of ROS and subsequent mitochondrial energy deficit in a subset of motoneurons. 13,14 One essential property of HMs is their ability to generate 'group bursting' dependent on the membrane expression of gap junctions 15,16 made by connexins, of which connexin 36 (Cx36) is highly represented in the nervous system. 17,18 Gap junctions may also be due to pannexins, particularly pannexin 1 (Panx1), which share, with connexins, similar topology, permeability and gating. 18,19 In case of cell stress, expression of connexins and pannexins may increase the probability of damage diffusion and cell death. 17 We recently observed that the neuronal acetylcholine receptor (nAChR) agonist nicotine prevents HM dysfunction and associated cell death. 13,14 Although these effects are dependent on nAChR activation, 13 it seems likely that nicotine was simply a trigger for intracellular and network events contrasting excitotoxicity. Within this framework, on rat brainstem motoneurons nicotine induces early facilitation of glutamate release followed by depression together with enhancement of inhibitory neurotransmission. 20 As a result of the complexity of the brainstem circuitry and its afferent projections, it is likely that the result of nicotine activity is the outcome of interplay among a variety of mechanisms. To better understand the action of nicotine, the present work investigated the functional topography of responsive neurons first and, subsequently, the processes of their modulation. One lead in this direction was provided by the report that in cultured endothelial cells nicotine downregulates connexins. 21,22 Thus, we wondered if a similar process could account for neuroprotection of HMs. To this aim, we studied how nicotine could modulate Ca 2+ transients, Cx36 and Panx1, and compared it with the effects of the gap junction blocker carbenoxolone. 23 In analogy with previous reports, 24,25 we used expression of heat shock protein 70 (Hsp70) and apoptosis-inducing factor (AIF) as indices of cell protection or death, respectively.
As a result of the scattered onset of [Ca 2+ ] i transients (Supplementary Movies 1-3), we studied whether topographical distance between two motoneurons was predictive of the closely spaced activity. Thus, Figure 2 compares the interneuronal distance with the latency between the first calcium transient of each cell pair as exemplified in Figure 2a in which the long interval between transients indicates lack of coupling. The top plot (Figure 2b), which refers to TBOA-treated cells only, shows how the latency between transients was significantly related to their distance (ρ = 0.30, Po0.001) and that the vast majority of latency values were below 100 s. Conversely, when nicotine or carbenoxolone was co-applied with TBOA (Figures 2c and d), the plots comprised a cloud of widely scattered points with latency values larger than 200 s even for closely located HMs. These data demonstrated the asynchronous occurrence of [Ca 2+ ] i changes when nicotine or carbenoxolone were co-applied.
Nicotine or carbenoxolone modulates Cx36 expression. As HM bursting is supported by a variety of mechanisms including gap junctions, 15,16,26,27 the above results raised the hypothesis that nicotine (like carbenoxolone) may impair intercellular communication via Cx36 and Panx1. As mitochondrial dysfunction and motoneuron death are observed after 4-h treatment with TBOA, 13 we evaluated Cx36 expression (Figures 3a and b; red) in relation to immunoreactivity of single HMs (identified with SMI 32 immunostaining; green). DAPI was used for nuclear staining (blue). In accordance with previous studies, 28,29 Cx36 immunoreactivity was distributed through the soma of HMs as shown in Figure 3a (middle column) and quantified in Figure 3b. There was also weak staining of the nucleus in accordance with the report by Belluardo et al. 30 who have detected nuclear Cx36 in rat brain neurons.
In the present experiments, the Cx36 signal was not significantly altered by TBOA alone in the cells that remained after the excitotoxic stimulation, whereas the motoneuronal number after TBOA (Figure 3c) fell by approximately 30%. When nicotine was co-applied with TBOA, the Cx36 immunoreactivity was significantly decreased (Figures 3a and b) and the average number of motoneurons was similar to sham ( Figure 3c). An analogous result was observed when carbenoxolone was co-applied with TBOA (Figures 3a-c). Nicotine per se did not change the average Cx36 signal (Figure 3a and b) or motoneuron survival ( Figure 3c). 13 To further investigate the amount of Cx36 expressed under various experimental conditions, western blot experiments were performed as depicted in Figure 4 using brainstem tissue blocks in order to collect a sufficient amount of protein. Even if the nucleus hypoglossus represents only a minor region of this tissue, this approach was earlier useful to detect changes in gene profiling, protein synthesis and expression, and mitochondrial energy production 13,14 which, although cannot be directly attributed to HMs (or other brainstem motoneurons), yet they provide an index of the metabolic state of the network and of any ongoing neuroprotective process.
Evaluation of Cx36 total lysate (approximately 36 kDa) samples confirmed unchanged protein quantity after 4-h treatment with TBOA (0.99 ± 0.07, n = 9 brainstems) or nicotine (0.94 ± 0.07, n = 6 brainstems; Figure 4a). In support of the immunohistochemical data (Figures 3a and b), we observed that, when samples were treated with nicotine + TBOA or carbenoxolone + TBOA, Cx36 expression was significantly reduced to 0.89 ± 0.07 (n = 9 brainstems) or 0.82 ± 0.08 (n = 5 brainstems; P = 0.015 among groups), respectively. We further explored the distribution of Cx36 in membrane and cytoplasmic compartments. In the case of nicotine exposure, there was a small, yet significant decrease in Cx36 membrane expression (Figure 4b; P = 0.008 for sham versus nicotine; n = 5 brainstems) concomitant with a corresponding increase in the cytoplasmic fraction (Figure 4c; Po0.05 for sham versus nicotine; n = 5 brainstems). The fall in membrane expression of Cx36 was intensified when nicotine was co-applied with TBOA ( Figure 4b; P = 0.016 for sham versus nicotine + TBOA; n = 5 brainstems) in association with a significant rise in the cytoplasmic expression (Figure 4c; Po0.05 for sham versus nicotine + TBOA; n = 5 brainstems). Interestingly, no compartmental expression was modified by TBOA alone, perhaps suggesting that the manifestation of the action of this drug required an unchanged Cx36 expression. These effects of Cx36 compartmentalization were not due to changes in its synthesis as qPCR Each data point represents the latency value calculated for a pair of cells. For data points shown in b, a significant linear correlation (ρ = 0.30, Po0.001) between signal latency and cell distance was observed with the analysis of deviance (see Materials and Methods section) in which the regression line for the plot of latency (taken as y) is equal to 21.08 + 0.13 inter-cell distance (taken as x): this result differs in a significant manner (Po0.001) from the null linear model, that is, the mean value latency = 36.60. Therefore, the regression line appears to be the minimal adequate model to establish that distance has an effect on latency in a statistical sense Hsp70 and AIF contrasting expression in HMs. In case of excitotoxicity, motoneuronal survival may depend on the relative intracellular expression of Hsp70 and AIF with cellprotecting or cell-damaging properties, respectively. 24,25 Figures 5b and c show that, after TBOA or nicotine treatment, Hsp70 expression was unchanged compared with basal sham conditions. When nicotine and TBOA were co-applied, Hsp70 expression was significantly enhanced (P = 0.027 among groups; Po0.05 for sham versus nicotine + TBOA; n = 5 brainstems). Thus, during excitotoxic stimulation, application of nicotine enhanced Hsp70 expression, thereby strengthening the neuroprotective processes of motoneurons. In fact, protection of motoneurons from excitotoxicity depends to a large extent on the binding by Hsp70 of AIF, a mitochondrial factor released by metabolically damaged motoneurons. 24,25 When, during excitotoxicity, Hsp70 expression is insufficient to bind AIF, the latter migrates to the cell nucleus and inactivates DNA. 31 We, therefore, studied immunohistochemical expression of AIF when TBOA was applied alone or together with nicotine. Figure 6a shows examples of AIF immunoreactivity increased after TBOA administration with broad distribution within nuclear (delineated by the blue line in Figure 6b for DAPI staining) and non-nuclear compartments as indicated by the line scan red trace that had higher value throughout (Figure 6b; note different ordinate scale for the various treatment protocols). Vice versa, the level of AIF was low in the presence of nicotine with or without TBOA (Figure 6b), suggesting that nicotine kept the AIF level at sham-like condition. These data are quantified in the bar graph of Figure 6c in which the average fluorescence signal (AU) of AIF in basal condition was 16.1 ± 3.2 (n = 9 slices) and significantly increased to 60 ± 7.8 (Po0.05, n = 19 slices) after TBOA exposure, whereas it was low when nicotine was co-applied with TBOA (n = 16 slices). It should be noted that nicotine (n = 6 slices) per se left unchanged the AIF immunoreactivity compared with sham.
Unchanged expression of Panx1. We sought to understand whether nicotine or carbenoxolone could change other proteins such as Panx1 reputed to make gap junctions. To this end, immunohistochemical and western blot experiments were performed as illustrated in Figures 7a-c. Hence, Panx1 immunoreactivity was readily detected in HMs as depicted in Figure 7a and remained unchanged following TBOA with or without nicotine protocols (Figure 7b). Similarly, Panx1 protein expression was very similar among all these treatments, indicating that Panx1 was not involved in nicotine neuroprotection at least within the 4-h experimental timeframe.

Discussion
The novel finding of this study was the demonstration that, during an excitotoxic stimulus, nicotine perturbed the emergence of coordinated Ca 2+ transients, decreased the These data were critical components of the neuroprotective action by this alkaloid. As many effects of nicotine were replicated by carbenoxolone, it is likely that inhibition of Cx36 was essential to uncouple motoneurons from their collective bursting behavior that was prodromic to cell distress and death.
Motoneurons wired together, die together. Inhibition of glutamate uptake recruits clusters of motoneurons in group bursting that, if it continues unabated, will lead to significant neuronal death. 12 As our recent work indicates that nicotine largely suppresses bursting, 13 we investigated how this phenomenon was expressed within the HM population (recorded as Ca 2+ transients) and its spatio-temporal It has been demonstrated that carbenoxolone acting as a pharmacological blocker of gap junctions, inhibits motoneuron synchronization in vertebrates and invertebrates. 32,33 Thus, following application of carbenoxolone as well as nicotine, bursting became sparse and, importantly, this event was followed by significant protection of HMs from death. This realization implies that the severity of damage was perhaps dependent on the collective activation and recruitment of clusters of HMs into pathological discharges, and led us to study the role of gap junctions (known to exist among HMs; 15,16) mediated by Cx36 in this process. Even if carbenoxolone is not a very specific blocker of gap junctions, the salient contribution by these structures to the collective bursting behavior and their inhibition by nicotine were validated with immunohistochemistry data as shown in Figures 3a and b. Connexins and cell death. Cx36 is the most prevalent gap junction protein expressed by neurons. 17,18 It electrically couples neighboring cells by allowing transcellular communication and exchange of Ca 2+ , Na + , K + , and small (o1-1.5 kDa) hydrophilic molecules. 18,34 Many factors regulate the activity of gap junction channels, including changes in voltage, [Ca 2+ ] i , pH, 35 connexin phosphorylation 35,36 and ROS. 37 In the past few years, many other components of intra/extracellular signaling have been associated with plasma membrane hemichannels such as facilitation of glutamate release from astrocytes 38 and apoptosis. 34 Indeed, gap junction-related apoptosis has been demonstrated by carbenoxolone-mediated prevention, 39 and by the spreading of death signals within a cell cluster via gap junctions. 39,40 This mode of propagation from injured to uninjured close neighbor cells is classified as bystander killing (the 'kiss of death'). 17,34 An interaction among connexins and mitochondrial AIF in cardiomyocytes has been described with potential effects on mitochondrial respiration and ROS signaling. 41 Whether this process could also occur in motoneurons remained unclear.
During NMDA receptor-mediated excitotoxicity, a strong reduction in neuronal death of cultured cortical neurons is obtained when Cx36 is pharmacologically blocked or genetically ablated, indicating that the expression level of Cx36 critically modulates neuronal death. 27,42 Thus, the observation of basal expression of Cx36 by HMs made this protein a likely candidate in the excitotoxic cell death. It was of interest that, following glutamate uptake block and loss of a number of HMs, surviving cells did not show impaired Cx36 expression, alluding to the possibility that these cells had relied on certain intrinsic mechanisms to withstand the injury process.
Nicotine prevents HM excitotoxic death via Cx36 downregulation. In endothelial cell cultures, nicotine, via membrane ACh receptors, downregulates various connexins through intracellular pathways 21,22 that impact their turnover. 22 In excitotoxic stress, we actually detected a fall in Cx36 immunoreactivity of HMs when nicotine (or carbenoxolone) was applied with TBOA, a result that was accompanied by no significant loss of HMs. Although nicotine can decrease excitatory synaptic transmission on HMs, 13,20 the analogy with the effects by carbenoxolone and the   44 Thus, it is feasible that blocking Cx36 directly with carbenoxolone or via intracellular signaling with nicotine led to decreased immunosignal of this protein at HM level because degradation was enhanced and/or replenishment impaired. In the case of nicotine, one parsimonious hypothesis is that this drug delayed Cx36 transportation from intracellular stores to the plasma membrane so that ongoing Cx36 degradation could not be efficiently compensated. Our view is consistent with the notion of nicotine uncoupled cells via increased proteolysis of connexins. 22 Circumstantial evidence was sought with tissue fractionation experiments in which we sought the relative distribution of Cx36 among membrane and cytosolic fractions. Despite the limitation inherent in the use of brainstem tissue blocks containing a heterogeneous cell population, we detected a significant decrease in the Cx36 membrane fraction together with a notable rise in its cytoplasmic fraction that also comprises contributions from intracellular stores. These data implied that the expression and perhaps the function of Cx36 at membrane level were impaired with nicotine application. Although the present data cannot, of course, be attributed to discrete changes in HMs alone because of the use of brainstem tissue, they allowed us to point to Cx36 as an important target for nicotine neuroprotection: should this process extend to other brainstem neurons (or glia), it may reinforce the widespread impact of nicotine effect against excitotoxicity. Our model also indicated that the effect by nicotine had a degree of specificity as Panx1 expression was unchanged by nicotine. We cannot, however, rule out that the long life-cycle of Panx1 45 had precluded detecting a later alteration following nicotine application.
Nicotine modulated Hsp70 expression as a gateway to cell survival against AIF-mediated death. One further result of interest was the observation that nicotine and TBOA application evoked a significant increase in Hsp70 expression, whereas nicotine or TBOA alone did not change this protein. As the intracellular expression of Hsp70 is an important biomarker of the ability of motoneurons to resist to excitotoxicity, 24 we surmise that surviving cells had an adequate Hsp70 expression although these data were obtained from brainstem tissue rather than single HMs. The perturbation triggered by TBOA plus the effects of nicotine probably synergized to raise Hsp70 expression and extend neuroprotection. These observations, therefore, provided an interesting clue to further explore the mechanism of nicotine neuroprotection. Our view was in line with reports of nicotine ability to enhance Hsp70 expression in lung vessels and ovarian tissue culture, 46,47 and to block AIF lethal translocation into the nucleus. 31 The immunofluorescent data showed that, at HM level, TBOA clearly enhanced the expression of AIF, a factor released by distressed mitochondria, and that this biomarker was distributed throughout the motoneuron including its nucleus. Co-application of TBOA and nicotine normalized the rise in AIF expression and restituted shamlike condition, suggesting that the effect of this death factor was inhibited likely as a consequence of the raised Hsp70 expression in the cytoplasm. These results indicate that neuroprotection by nicotine was, at least in vitro, more systematic than the effect produced for example by pharmacological depression of glutamate release with riluzole. 11 We propose that the additional target of Cx36 uncoupling by nicotine is important to confer prevention of excitotoxicity.
A scenario for motoneuron protection by nicotine. Figure 8 depicts an idealized diagram to account for the process of HM death evoked by glutamate uptake block and Figure 8 Idealized diagram to account for the toxic effects by TBOA and their antagonism by nicotine on HMs. (a) In the presence of TBOA, glutamate (filled red circles) released from presynaptic terminals acts on glutamatergic receptors to depolarize HMs. TBOA-mediated inhibition of glutamate uptake by astrocytes leads to intensification of glutamate effects with build-up of intracellular free Ca 2+ , which acts on mitochondria and favors production of ROS, release of AIF and DNA damage. This process is greatly amplified by the bidirectional (red arrows) operation of gap junctions (Cx36; filled blue cylinders) that spread and recruit HMs into a hyperexcitable state with deleterious consequences on cell survival. (b) In the presence of nicotine, the neurotoxic effect of TBOA is largely attenuated. Nicotine exerts multiple effects via nAChRs located on presynaptical terminals (with consequent decrease in glutamate release, and increase in the inhibitory response), HM membrane and even mitochondria. Through these effects, nicotine strongly decreases the motoneuron coupling via gap junctions through redistribution of Cx36 subunits, therefore disjoining motoneurons from their collective excitatory behavior. Our hypothesis is that this process is accompanied by inhibition of Cx36 activity, reduction in production of ROS and facilitation of the effect of Hsp70 to sequester AIF and prevent its nuclear migration. For references, see text the intervention levels exerted by nicotine. Excitotoxicity, induced by block of excitatory amino-acid transporters, 48 severely damages motoneurons by excessive glutamate receptor stimulation that induces strong network bursting [11][12][13]26 among electrically coupled HMs (Figure 8a). This phenomenon elicits excessive Ca 2+ influx, which depolarizes the neuronal membrane potential to activate further Ca 2+ influx 49,50 and intracellular second messengers in a death cell cascade. 51,52 The role by mitochondria in Ca 2+ buffering may lead to mitochondrial membrane depolarization 49,50,53 and perturbation of the respiratory chain. 49,50 This process is compounded by ROS-evoked oxidative stress. 13,54,55 In this vicious sequence, cell damage is extended via gap junctions 17,27,56 activated by ROS. 37 To contrast excitotoxic damage, a potential device is activation of nAChRs expressed on interneurons and HMs. 13,14,20,57 nAChRs operate through a cycle of activation, desensitization and reactivation 58 that, in the case of brainstem synaptic transmission with diffuse inputs impinging on HMs, include early facilitation of glutamate release followed by depression, whereas synaptic inhibition is enhanced 13,20 (Figure 8b). Direct activation of nAChRs on motoneurons reduces ROS production, enhances Hsp70 expression and uncouples Cx36-mediated intercellular communication. The mechanism linking nAChR activation to inhibition of Cx36 activity will require future investigation. We can posit that, as nicotine enhances protein kinase C (PKC), 59-61 PKC-mediated connexin phosphorylation is a putative process to reduce gap junction-dependent communication. 62 Finally, our data do not intend to support a neuroprotecting role of smoking, rather to prompt a strategy for further investigation of inhibiting motoneuron disease in vitro and in vivo with novel tools to stimulate nAChRs.

Materials and Methods
Ethical approval. All experiments were performed following the ethical guidelines for the use and the care of laboratory animals of National Institutes of Health. The Scuola Internazionale Superiore di Studi Avanzati (SISSA) ethics committee (prot. 3599, 28 May 2012) approved all treatment protocols, which were in agreement with the European Union rules for animal experimentation. We made all the effort to minimize the use and the suffering of the animals, and reduce their number for experimentation.
Western blot. Analysis of quantitative and qualitative protein expression was performed with a standard western blot technique on whole brainstem treated as described above in accordance with the previously published protocol. 70 Total lysates, cytoplasmic and membrane fractions were analyzed. Total lysates were obtained by homogenizing samples in CHAPS buffer solution (0.5% CHAPS, 50 mM Tris pH 7.5, 1 mM EDTA, 150 mM NaCl, 10% glycerol plus protease inhibitors mixture and reducing agents; Complete, Roche Applied Science, Basel, Switzerland; and Sigma-Aldrich). A hypotonic lysis buffer (10 mM HEPES pH 7.9 with 1.5 mM MgCl 2 and 10 mM KCl plus protease inhibitors mixture and reducing agents) was used to process samples for membrane and cytoplasmic extraction, which were then centrifuged for 5 min at 4000 r.p.m. at 4°C and the supernatant transferred to ultracentrifuge tubes for centrifugation (1 h) at 100 000 g. Cytoplasmic proteins were considered those in the supernant, whereas the pellet was formed by membrane proteins, which were re-suspended in the extraction buffer (20 mM HEPES pH 7.9, with 1.5 mM MgCl 2 , 0.2 mM EDTA, 25% glycerol, 1% SDS plus protease inhibitors mixture and reducing agents). Samples were then immunoblotted with rabbit anti-Cx36 (1:200, Alomone), rabbit anti-Panx1 (1:400, Alomone), mouse anti-Hsp70 (1:5000, Abcam, Cambridge, UK), mouse anti-β-actin (1:2000, Sigma-Aldrich), mouse anti-β-tubulin (1:2000, Sigma-Aldrich), or mouse antisynaptophysin (1:10 000, Millipore) antibodies. As both Cx36 and Panx1 encode consensus sites for phosphorylation and glycosilation, it is, thus, common to observe more than one band in the immunoblotting gels 45,71 as reported in this study as well. The enhanced chemiluminescence light system (ECL, Amersham Bioscience, Piscataway, NJ, USA) was used to detect signals recorded with the digital imaging system Alliance 4.7 (UVItech, Cambridge, UK) and quantified with the software Alliance LD2-77-WL (UVItec).
Statistics. Results were expressed as means ± S.E.M. and collected from at least three different experiments, where n refers to the number of slices or brainstems for each independent experiment, as indicated, and N refers to the number of times an experiment was repeated. Using the standard software SigmaStat 3.5 (Systat Software, Inc., Chicago, IL, USA), we first run an ANOVA test and the normality test to discriminate between parametric and non-parametric data. Subsequently, as directed by the software, the Mann-Whitney test was used for comparing two non-parametric groups. Multiple groups were first analyzed with the Kruskal-Wallis one-way ANOVA on ranks when data were non-parametric and further analysis was carried out with a software directed test (Dunn's method, Holm-Sidak method or Tukey test). The use of each test for data analysis is indicated in the corresponding Figure legends. Cumulative probabilities were compared with the Kolmogorov-Smirnov test. To analyze the relation between Ca 2+ transient latency and cell topographical distribution, we applied the linear correlation method based on the analysis of deviance. 75 In this way, we observed that that the regression line latency for the experiments with TBOA alone differed in a significant manner (Po0.001) from the null linear model, that is, the mean value latency = 36.60. Thus, the regression line appears to be the minimal adequate model. 76 Groups of data were accepted as statistically different if P ≤ 0.05.

Conflict of Interest
The authors declare no conflict of interest.