Laboratoire de Bioénergétique Fondamentale et Appliquée, Université Joseph Fourier, Grenoble, France
U1055 – INSERM, Grenoble, France
Address correspondence and reprint requests to Cécile Batandier, Laboratoire de Bioénergétique Fondamentale et Appliquée, Université Joseph Fourier, INSERM U1055, BP 53X, F-38041 Grenoble Cedex, France. E-mail: email@example.com
In memoriam to Professor Xavier Leverve (1950–2010).
Since emotional stress elicits brain activation, mitochondria should be a key component of stressed brain response. However, few studies have focused on mitochondria functioning in these conditions. In this work, we aimed to determine the effects of an acute restraint stress on rat brain mitochondrial functions, with a focus on permeability transition pore (PTP) functioning. Rats were divided into two groups, submitted or not to an acute 30-min restraint stress (Stress, S-group, vs. Control, C-group). Brain was removed immediately after stress. Mitochondrial respiration and enzymatic activities (complex I, complex II, hexokinase) were measured. Changes in PTP opening were assessed by the Ca2+ retention capacity. Cell signaling pathways relevant to the coupling between mitochondria and cell function (adenosine monophosphate-activated protein kinase, phosphatidylinositol 3-kinase, glycogen synthase kinase 3 beta, MAPK, and cGMP/NO) were measured. The effect of glucocorticoids was also assessed in vitro. Stress delayed (43%) the opening of PTP and resulted in a mild inhibition of complex I respiratory chain. This inhibition was associated with significant stress-induced changes in adenosine monophosphate-activated protein kinase signaling pathway without changes in brain cGMP level. In contrast, glucocorticoids did not modify PTP opening. These data suggest a rapid adaptive mechanism of brain mitochondria in stressed conditions, with a special focus on PTP regulation.
In a rat model of acute restraint stress, we observed substantial changes in brain mitochondria functioning. Stress significantly (i) delays (43%) the opening of permeability transition pore (PTP) by the calcium (Ca2+), its main inductor and (ii) results in an inhibition of complex I in electron transport chain associated with change in AMPK signaling pathway. These data suggest an adaptive mechanism of brain mitochondria in stressed condition, with a special focus on PTP regulation.
Mitochondria are organelles providing the host cell with ATP by oxidative phosphorylation. They are also central to all cell functioning: a hub for intracellular Ca2+ homeostasis, steroid synthesis, generation of free radical species, and apoptosis induction. To sum up, mitochondria are able to lead the cell either to survival or death.
The brain has a very high metabolic rate, so the cell function is particularly sensitive to any mitochondrial dysfunction (Fiskum et al. 1999; Nicholls and Budd 2000) with dramatic consequences as observed in numerous neurological (Rezin et al. 2008, 2009) and neurodegenerative (Schapira 2008) diseases. Surprisingly, few works have considered the brain mitochondria as central object of study in situation of high brain energy demand, such as stress situations. Indeed, stress might alter brain mitochondria (Manoli et al. 2007) with possible consequences on mitochondrial structure (Gong et al. 2011). It has been reported that exposure to chronic stress situations induces activation in juvenile (Krolow et al. 2012) or inhibition of respiratory chain complex in adult (Rezin et al. 2008). In addition, brain mitochondria of chronically stressed animals are less susceptible than those of non-stressed animals to cope with the addition of the Ca2+, the referent permeability transition pore (PTP) activator (Kuchukashvili et al. 2012).
In the present work, we have investigated the consequences of an acute stress exposure on brain mitochondria functions, using a well-described animal model of acute restraint stress (Buynitsky and Mostofsky 2009). We focused on mitochondrial permeability transition, the key component of this phenomenon being the mitochondrial PTP, a multiproteins complex channel (Bernardi et al. 2006; Zorov et al. 2009). Indeed, its opening is associated with an increased mitochondrial free radical production (Batandier et al. 2004), as observed in chronically stressed animals (Kuchukashvili et al. 2012). Furthermore, the PTP is involved in Ca2+ matrix regulation through its transient opening (Bernardi and von Stockum 2012) and apoptosis induction through its final opening (Grimm and Brdiczka 2007). The transient mitochondrial Ca2+ uptake is of importance as it has been shown to control intracellular Ca2+ signaling, cell metabolism, cell survival, and other cell type-specific functions by buffering cytosolic Ca2+ levels (Rizzuto et al. 2012). The final PTP opening is involved in the intrinsic apoptotic pathway through a Ca2+-dependent increase of the outer mitochondrial membrane permeability (Rasola and Bernardi 2011). Consequently, factors modifying PTP opening dynamic could be crucial to cell function and survival. The mechanism by which stress may modify mitochondrial functioning and PTP opening deserves discussion (Manoli et al. 2007). Numerous mechanisms activated in stressful situations might target mitochondria. A high glucocorticoid level (Joels and Karst 2012) alters mitochondria functioning (Du et al. 2009) as do free radicals with altered antioxidant defense systems (Lucca et al. 2009) and nitric oxide (NO) production changes (Echeverry et al. 2004).
In naive rats, we measured mitochondria respiration and related enzymatic activities immediately after acute stressor exposure, as well as calcium retention capacity (CRC) for monitoring PTP functioning. We also investigated the impact of glucocorticoids on mitochondria functioning. Cell signaling pathways activated in the brain of stressed animals, such as cGMP (Sanches et al. 2003), adenosine monophosphate-activated protein kinase (AMPK) (Ronnett et al. 2009), phosphatidylinositol 3-kinase (PI3-K) (Stiles 2009), and MAPK (Subramaniam and Unsicker 2010) were also assessed.
Material and methods
Animals and experimental design
Twenty-five just weaned male Wistar rats (Charles River, l'Arbresle, France) were housed in individual cages at the arrival to laboratory facilities, in a room with 20–22°C ambient temperature, 50 ± 10% relative humidity, less than 50 dB white noise, and 12 h/12 h circadian cycle. All the rats were fed ad libitum for 12 weeks with purina chow purchased from SAFE (Augis, France) and had free access to tap water. Their health status was under survey throughout the experimental period (aspect, body weight evolution, reactivity to handling) and their body weight was monitored once a week. The animals remained in the same environment with the same animal technologist for care throughout the investigation to reduce stress as possible.
All experimental procedures were reviewed and approved by the Institutional Ethic Committee for Animal Care (CRSSA, Protocol N 2008/02.1 accepted in December 2008). The rats were maintained and handled in agreement with the Guide for the Care and Use of Laboratory Rats (NIH 1985).
Stress and killing procedures
The 20 rats were divided into two groups, submitted (S-group, n = 10) or not (C-group, n = 10) to an acute restraint stress. For experiment, the rats were transferred between 08.00 and 09.00 h to the experimental room, where noise was reduced to a minimum. For the restraint stress, a well-established emotional stressor in rodents (Buynitsky and Mostofsky 2009), S-group rats underwent forced immobilization for 30 min in a plastic wire mesh restrainer adjusted to individual size to ensure a psychological strain without compression. They could not move, causing intense discomfort without apparent pain. After stressor exposure, the animals were moved to an adjacent room where they were killed by decapitation without anesthesia and the brain was removed. The control rats stayed undisturbed in their home cage until killing. The dose–response effect of glucocorticoids on brain mitochondria functioning was tested on five additional non-stressed rats.
Forebrain was immediately used for mitochondrial analysis, while the cerebellum was frozen in liquid nitrogen and stored at −80°C until further analysis by molecular methods. Hemolysis because of the decapitation did not allow us to assess glucocorticoids levels (Hiramatsu and Nisula 1991). However, the level of the blood glucocorticoids was measured on another series of animals conditioned in the same conditions (diet, restraint stress, and operator). A strong stress effect (p < 0.001) was revealed by anova for the plasma concentrations of corticosterone (Marissal-Arvy et al. 2014).
Brain mitochondrial functions
Forebrain mitochondria were prepared in the presence of 0.02% digitonin to disrupt synaptosomal plasma membrane in a buffer containing 225 mM mannitol, 75 mM sucrose, 5 mM HEPES, 1 mM EGTA, pH 7.4 [mannitol, sucrose, HEPES, EGTA (MSHE medium)] using differential centrifugation and permeabilization of synaptosomes (Rosenthal et al. 1987). Mitochondrial protein concentration was determined by a bicinchoninic acid assay (BCA Protein Assay; Pierce, Rockford, IL, USA) using bovine serum albumin as standard.
The mitochondrial oxygen consumption was measured polarographically at 30°C using a Clark-type oxygen electrode in a Mitocell S200® (Strathkelvin Instruments Limited, North Lanarkshire, Scotland). Mitochondria (1 mg) were suspended in 1 mL KET medium: 125 mM KCl, 1 mM EGTA, 20 mM Tris pH 7.2 with 0.1% bovine serum albumin, and 5 mM Pi. Oxygen consumption rate was measured after the addition of 300 μM ADP (respiration state 3), 0.5 μg/mL oligomycin (respiration state 4), or 75 μM dinitrophenol (DNP) (uncoupling state: uC) with 5 mM glutamate + 2.5 mM malate (GM) or 5 mM succinate (S) as substrates of complex I and complex II, respectively, or 0.5 μM antimycine A plus 2 mM ascorbate and 250 μM tetramethyl phenylenediamine (TMPD) as synthetic substrates of complex IV. Respiratory control ratio (RCR) was calculated by dividing state 3 respiratory level by state 4 respiratory level.
Reactive oxygen species (ROS) production was assessed by monitoring H2O2-induced fluorescence of 1 μM Amplex Red® (Life Technologies SAS, Saint Aubin, France) (excitation: 560 nm, emission: 584 nm) in the presence of horseradish peroxidase (10 IU) and 5 mM succinate as mitochondrial substrate. The quantification of H2O2 production was obtained by the addition of a known amount of H2O2. Assays were performed at 30°C with a PTI Quantamaster C61 fluorimeter® (Photon Technology International, Edison, NJ, USA), as described previously (Batandier et al. 2006).
The complex I activity was assessed by measuring the oxidation rate of NADH in the presence of an artificial acceptor, decylubiquinone, as described previously (Batandier et al. 2004). We checked that there was no rotenone-insensitive activity in our samples that allowed us to only consider the rotenone-inhibitable complex I activity. The complex II activity was assessed by measuring the reduction rate of dichloroindophenol (DCIP) (Hatefi 1978). The absorbance change of NADH or DCIP was measured at 37°C in a Specord 210® (Analytik Jena AG, Jena, Germany) spectrophotometer equipped with magnetic stirring and thermostatic control. The mitochondrial linked hexokinase activity, as potential regulator of PTP by binding to voltage-dependent anion channel was also measured at 37°C under continuous stirring as described elsewhere (Scheer et al. 1978).
The changes in PTP opening were assessed by the CRC test measuring the capacity of mitochondria to store repeated amount of calcium until PTP opening. Results are expressed as the amount of Ca2+ added. The free Ca2+ concentration was measured fluorimetrically in a buffer containing 250 mM sucrose, 10 mM Tris-MOPS (4-morpholinepropanesulfonic acid), 1 mM Pi, pH 7.4 (CRC buffer) in the presence of 0.25 μM Calcium Green-5N® (Life Technologies SAS) with stirring and thermostatic controls (30°C) as described previously (Fontaine et al. 1998). 5 mM succinate was added as substrate for electron transport chain. Measurement started after 1-min incubation and Ca2+ pulses (10 nmol/mg protein) were successively added after 1-min signal stabilization at 1-min intervals until PTP opening, as indicated by the release of Ca2+ into the medium. The increase in CRC was quantified in three experimental conditions: without modulator, with 1 μM cyclosporine A (CsA) as the competitive inhibitor of cyclophylin D (the referent inhibitor of PTP), and with 1 μM rotenone, as a specific complex I inhibitor. They were combined to calculate the average value of changes. The effect of glucocorticoids on CRC was assessed by adding in the buffer increasing concentrations of corticosterone (30 and 300 nM; 3 and 30 μM).
The cGMP levels were measured in frozen cerebellum samples (50 mg). They were homogenized in 6% (w/v) trichloroacetic acid at 2–8°C to obtain a 10% (w/v) homogenate. After centrifugation, the supernatant was recovered and washed four times with diethyl ether saturated water. The remaining aqueous extract was dried and then dissolved in 1 mL of assay buffer prior to analysis. The analysis was done using the cGMP Enzyme Immuno Assay Biotrak System® (RPN226; GE Healthcare Bio-Sciences, Pittsburgh, PA, USA).
Western blot analyses were carried out on cerebellum samples as described previously (Drouet et al. 2012). After electrophoresis and transfer, nitrocellulose membranes were incubated overnight at 4°C with the appropriate primary antibody for the detection of phosphorylated protein and their relative total proteins: (i) the PI3-K pathway: phospho-AKT (Ser473, ref. #4060), AKT (ref. #9272), Phospho-glycogen synthase kinase 3 beta (GSK-3β) (Ser9, ref. #9323), GSK-3β (ref. #9315), (ii) the MAPK pathway: phospho-ERK 1/2 (Thr202/Tyr204, ref. #4370), extracellular signal-regulated kinases (ERK) 1/2 (ref. #9102), phospho-P38 (Thr180/Tyr182, ref. #4511), P38 (ref. #9212), (iii) the mammalian target of rapamycin (mTOR) pathway: phospho-mTOR (Ser2448, ref. #2971), mTOR (ref. #2972), and (iv) the AMPK pathway: phospho-AMPKα (Thr172, ref. #2535), AMPKα (ref. #2603). Phospho-acetyl-coA carboxylase (phospho-ACC, Ser79, ref. #3661) and ACC (ref. #3676). (Cell Signaling Technology®, Danvers, MA, USA).
Membranes were stripped of the phosphorylated antibody before reprobing with the corresponding total antibody, as recommended (Amersham Pharmacia Biotech, Piscataway, NJ, USA). The semiquantitative analyses of band intensity were performed using Quantity One Software® (Bio-Rad Laboratories, Hercules, CA, USA). Results were expressed as phosphorylated/total protein ratio (P/T).
The data were presented as mean ± SEM. Comparisons between groups were done using Student's t-test and Mann–Whitney test according to the distribution of the results. A p-value < 0.05 was considered statistically significant.
Respiration related to complex I monitored by GM was altered in stressed rats (Fig. 1a). Acute stress reduced significantly O2 consumption in state 3 (p <0.05, C-group: 56.1 ± 2.1 vs. S-group: 48.1 ± 2.0 nmol/min/mg prot) and uncoupled state (p <0.05, C-group: 62.5 ± 1.8 vs. S-group: 55.1 ± 2.4 nmol/min/mg prot). No stress effect was observed on O2 consumption with succinate as complex II substrate or with TMPD/ascorbate as synthetic complex IV substrates (Fig. 1a). As exemplified in Fig. 1(d) and (e), the RCR was 23.6 ± 2.0 (n = 10) and 6.7 ± 0.5 (n = 10) when control mitochondria were energized with GM and succinate, respectively. The same high coupling level was measured with mitochondria isolated from stressed animals [26.2 ± 2.3 (n = 10) and 7.0 ± 0.7 (n = 10)] in the presence of GM and succinate, respectively.
Mitochondrial enzymatic activities
Consistent with oxygraphic results, acute stress significantly p <0.05 decreased complex I activity [C-group: 556 ± 15 (n = 10) vs. S-group 515 ± 11 (n = 10) μmol NADH/min/mg prot] (Fig. 1b). The activity of the isolated complex II remained unchanged [C-group: 285 ± 3 (n = 10) vs. S-group: 277 ± 3 (n = 10) μmol DCIP/min/mg prot] (Fig. 1c). The activity of linked mitochondrial hexokinase was not modified by stress (Table 1).
Table 1. H2O2 production (pmol/min/mg prot) and hexokinase activity (μmol NADPH/min/mg prot) in the isolated brain mitochondria samples and cGMP content in cerebellum (fmol/g tissue)
Results are expressed as mean ± SEM. The significance level was set at p < 0.05. Experimental conditions are described in the 'Material and methods' section.
H2O2 (pmol/min/mg prot)
838 ± 82
(n = 10)
627 ± 90
(n = 10)
Hexokinase (µmol NADPH/min/mg prot)
376 ± 4
(n = 9)
368 ± 3
(n = 10)
cGMP (fmol/g tissue)
81.5 ± 6.8
(n = 6)
66.7 ± 6.2
(n = 4)
Maximal mitochondrial ROS generation assessed by H2O2 production by reverse electron flow through complex I was not significantly (Mann–Whitney test) modified by acute stress (Table 1).
Acute restraint stress delayed the opening of PTP. As shown in Fig. 2(a) and (b), whatever the conditions used, rat brain mitochondria exhibited significantly higher CRC in S-group (Fig. 2b) than in C-group (Fig. 2a). Note that the CRC were significantly different between the two groups in control condition (patterns a, Fig. 2c, S-group: 72.2 ± 2.8 vs. C-group: 51.0 ± 3.5 nmol/mg prot), in the presence of CsA (patterns b, Fig. 2d, S-group: 125.0 ± 2.7 vs. C-group: 87.0 ± 5.4 nmol/mg prot), as well as in the presence of rotenone (patterns c, Fig. 2d, S-group: 137.8 ± 4.6 vs. C-group: 96.0 ± 3.1 nmol/mg prot). The mean change expressed as percent increase was 43% (without PTP modulator: 42%; with CsA: 44%; and with rotenone: 44%). As shown in Fig. 3, we did not observe any dose–response effect of corticosterone on PTP opening under our experimental conditions.
No significant (Mann–Whitney test, p = 0.07) changes in cGMP levels were induced by stress (Table 1).
AMPK signaling pathway
The P/T AMPK ratio was significantly increased by acute stress [p <0.05, S-group: 4.53 ± 0.82 (n = 8) vs. C-group: 2.73 ± 0.57 (n = 10)]. The P/T ACC ratio remained unchanged (S-group: 1.12 ± 0.11 vs. C-group: 1.29 ± 0.12) (Fig. 4).
PI3-K signaling pathway
Stress did not modify the P/T AKT (S-group: 1.16 ± 0.06 vs. C-group: 1.20 ± 0.17), P/T mTOR (S-group: 1.09 ± 0.15 vs. C-group: 1.03 ± 0.12), and P/T GSK-3β ratio (S-group: 1.14 ± 0.19 vs. C-group: 1.03 ± 0.09) (Fig. 5).
MAPK signaling pathway
P/T MAPK ratio (S-group: 0.67 ± 0.08 vs. C-group: 0.86 ± 0.12) and P/T ERK (1/2) ratio (S-group: 1.04 ± 0.11 vs. C-group: 1.15 ± 0.08) remained unchanged in stressed animals (Fig. 5).
Although some authors have recently pointed out that chronic stress could modify brain mitochondria functioning (Rezin et al. 2008; Krolow et al. 2012; Kuchukashvili et al. 2012); so far, few studies have investigated this effect in acute stress conditions. In the present work, we observed that naive rats submitted to acute restraint stress exhibited, within 30 min, several major mitochondrial alterations such as respiration and related enzymatic activities and CRC related to PTP opening.
Stressor exposure is followed by an inhibition of mitochondrial electron transport chain at the level of complex I. Stress (i) inhibits the O2 consumption when using complex I substrates G/M (state 3 and uncoupled state), while the RCR was unchanged and (ii) decreases the activity of isolated complex I. This effect seems to be specific to complex I since stress did modify neither respiration driven by succinate oxidation nor isolated complex II activity. Furthermore, we did not find any inhibition at the level of complex IV. These results are in agreement with those previously reported in chronic stress for complex I inhibition and complex II steadiness but not for complexes III and IV inhibition as observed in the cortex and the cerebellum of stressed animals (Rezin et al. 2008). Obviously, these differences might be explained by topography of the samples as evidenced by these authors (global forebrain vs. cortex and cerebellum) or the type of stress (acute vs. chronic).
Consistent with our data, ROS production by reverse flux in complex I trends to decrease, suggesting a possible relationship between the amount of ROS produced by isolated brain mitochondria and the amount of complex I as hypothesized elsewhere (Chinta et al. 2009). In addition, Liu reported a link between the amount of produced ROS and the reverse flux inhibition in complex I (Liu et al. 2002). This inhibition is well established in the pathophysiology of several neurodegenerative brain diseases (Schapira 2010). However, it does not fit with the increased production of free radicals during chronic (Lucca et al. 2009) and acute (Pal et al. 2006) stressor exposures.
Acute restraint stress also results in a large increase in CRC test (43%) because of a delay in brain PTP opening. The extent of stress effect is relevant since it is of the same order to that observed in the presence of CsA. Furthermore, the observed effect on PTP opening occurs within 30 min after the beginning of stressor exposure, suggesting that signaling or a post-translational modifications might be involved rather than an implementation of a process of protein synthesis.
The stress-induced biological effects might target the PTP structure or its regulation (Adzic et al. 2009). PTP is functionally well characterized, but remains an elusive molecular entity, despite numerous regulatory elements identified (Zorov et al. 2009; Rasola et al. 2010) and recent assumptions (Bernardi 2013) with some evidence (Giorgio et al. 2013). With such a structure of ATP synthase dimers (Giorgio et al. 2013), the possibility that the observed partial inhibition of complex I might be linked to PTP inhibition cannot be ruled out. This hypothesis, however, does not fit with our data since we observed: (i) a complex I inhibition persisting in uncoupling condition and (ii) an absence of complexes II and IV inhibition in phosphorylating conditions.
The mechanisms by which stress induces mitochondria dysfunctions deserve discussion. Corticosterone addition did not modify the PTP opening despite the presence of specific mitochondria receptors (Du et al. 2009). Although chronic stress is known to activate NO production (Kuchukashvili et al. 2012), cGMP was reduced in S-group as compared to C-group. Consequently, the stress-induced delayed PTP opening is unlikely to be solely because of these two mediators. Although oxidative stress has been reported to occur in chronic stress and to participate in PTP opening (Kuchukashvili et al. 2012), we did not find evidence of an increased ROS production in isolated mitochondria after acute stress. Our data are consistent with the delayed opening of PTP since oxidative stress was reported to facilitate PTP opening (Zorov et al. 2009).
Stress is unlikely to delay PTP opening through changes in hexokinase level. Although hexokinase level could modify PTP opening by interacting with B-cell lymphoma 2 (Bcl2) family proteins (Pastorino and Hoek 2008), we did not observe stress-induced changes in hexokinase enzymatic activity.
We used complex II substrate (succinate) rather than complex I substrate (glutamate/malate) to generate membrane potential in the CRC test. Hence, we avoided the direct involvement of the complex I, closely linked to PTP regulation (Fontaine et al. 1998). Indeed, inhibition of complex I by rotenone (Chauvin et al. 2001; Li et al. 2012) and other complex I inhibitors (Guigas et al. 2004; Lablanche et al. 2011) inhibits PTP opening. The stress-induced complex I inhibition might also explain the delay in PTP opening. But the inhibitory effect of stress persisted in the presence of rotenone. Al together, these observations suggest that restraint stress did not just act via the inhibition of complex I activity.
Stress might also affect the binding of Cyclophilin D to the PTP. Cyclophilin D is known to favor PTP opening (Giorgio et al. 2010), while CsA inhibits opening resulting in the detachment of Cyclophilin D from the PTP. The inhibitory effect of a restraint stress persisted in the presence of CsA suggesting that restraint stress did not affect the binding of Cyclophilin D to the PTP.
The stress-induced cell signaling may also be involved in mitochondria dysfunction. The phosphorylation of AMPK was significantly increased in S-group, reflecting an increased brain energy requirement in stressed conditions (Ronnett et al. 2009). In agreement with these results, an increased hypothalamic AMPK activity was observed after restraint stress but not after surgical stress under anesthesia by Marques et al. (2012). AMPK acts as a cellular energy sensor responding to low ATP levels. The increased phosphorylation of AMPK might suggest a response to the partial inhibition of complex I to counteract the defective energy level.
In spite of the involvement of PI3-K pathway in regulating mitochondrial functions, of GSK-3β on PTP opening regulation (Stiles 2009; Rasola et al. 2010), and those of MAPKs pathways on cell survival regulation (Subramaniam and Unsicker 2010), these metabolic ways are unlikely implicated in mitochondria dysfunction as we did not observe any change on those pathways.
In summary, the present work reports a delay in PTP opening in brain mitochondria in stressed condition. This change is associated with a partial inhibition of complex I respiratory chain and modulations in AMPK signaling pathway. Brain mitochondria are dramatically involved in neuronal survival, especially as essential regulators of cellular Ca2+ homeostasis. Our data suggest that cell survival process could be stimulated by restraint stress. The mechanisms involved in these changes are still to be clarified since they do not match our understanding of PTP regulation. Furthermore, additional studies are needed to estimate the impact of emotional stress on mitochondria functioning with a special focus on the differential sensibility of brain areas and cell types.
Acknowledgments and conflict of interest disclosure
We thank Manar Awada, Mireille Osman, Nadine Fidier, and Renaud Maury for their excellent technical assistance. This work is a part of the CERVIRMIT Study funded by the French National Agency for Research (PNRA 007) and ANR – Paris – France (grant number ANR-07-PNRA-003-01). We have no conflicts of interest to declare.
All experiments were conducted in compliance with the ARRIVE guidelines.