Purkinje-neuron-specific down-regulation of p 38 protects motoric function from the repeated use of benzodiazepine

Benzodiazepine (BZD) is the most prescribed CNS depressant in America to treat hyper-excitatory disorders such as anxiety and insomnia. However, the chronic use of BZD often creates adverse effects including psychomotor deficit. In this study, we investigated a novel mechanism by which chronic BZD impedes motoric function in female mice. We used female mice because BZD use is much more prevalent in female than male populations. We tested the hypothesis that the accumulation of p38 (stress-activated protein) in cerebellar Purkinje neurons mediates motoric deficit induced by chronic BZD. To test this hypothesis, we generated transgenic mice that lack p38 in cerebellar Purkinje neurons by crossing Pcp2 (Purkinje cell protein 2)-Cre mice with p38 mice. p38-knockdown mice and wild-type mice received BZD (lorazepam, 0.5 mg/kg) for 14 days. During this period, they were tested for motoric performance using Rotarod assay in which a quicker fall from rotating rod indicates poorer motoric performance. Cerebellum was then collected to detect p38 in Purkinje neurons and to measure mitochondrial respiration using immunohistochemistry and real-time XF respirometry, respectively. Compared to vehicletreated mice, BZD-treated mice showed poorer motoric performance, a higher number of Purkinje neurons containing p38, and lower mitochondrial respiration. These effects of BZD were much smaller in p38-knockdown mice. These results suggest that the excessive accumulation of p38 in cerebellar Purkinje neurons contributes to motoric deficit associated with chronic BZD. They also suggest that Purkinje neuronal p38 mediates BZD-induced mitochondrial respiratory inhibition in cerebellum. Our findings may provide a new mechanistic insight into chronic BZD-induced motoric deficit.


INTRODUCTION
Benzodiazepines (BZD)s, inhibitory neurotransmitter enhancers, are by far the most frequently prescribed CNS depressants in Americans [1].Although BZDs are powerfully effective in treating hyper-excitatory CNS disorders, many patients encounter the non-therapeutic effects of BZDs [2].One common problem with BZD therapy is that patients often experience motor incoordination and movement disorders.For example, BZDs increase the risk for falls [3][4][5], automobile accidents [6], slow motor reaction, and the inaccuracy of motor tasks [7][8][9].Animal studies have also shown that diazepam and lorazepam provoke motoric deficit [10,11].The motor-impairing effect of BZD occurs at a therapeutic dose [12], at an acute and chronic dose, across species [12,13], and across genders, limiting the drugs' clinical utility [14].Importantly, the adverse impact of BZD is greater on the population of women than men because they significantly outnumber male BZD users [15,16] even when several factors are normalized [17][18][19][20][21][22].This is a clinically important issue because women outlive men, and thus, extending the period of such problems.Therefore, there is a critical need to develop a protective strategy to prevent this problem.As a step toward the development of a protective strategy, we intend to identify a direct mechanism by which chronic BZD impairs motoric functions and this effect of BZD can be minimized.
p38 is a signaling protein kinase whose aberrant activation is implicated in many pathological conditions [23][24][25].The known members of the p38 family include p38α [26], p38β [27,28], p38γ [29,30] and p38δ [31].Among these isozymes, p38α and p38β are highly expressed in brain areas such as cerebellum and cortex [32][33][34].p38 is activated upon phosphorylation [35], so phosphorylated p38 (pp38) is often measured as an indicator of p38 activation.p38 is also known as a stressactivated protein kinase because p38 is phosphorylated by stress signals, such as inflammatory cytokines, heat shock, or ischemia [36].The pathological activation of p38 has been shown in the brains of Alzheimer's disease patients [37] and in the livers of aged rats after challenged with a prooxidant, H 2 O 2 [38].Purkinje neurons are the major type of cerebellar neurons, responsible for movement control.Our recent study has shown that p38-containing Purkinje neurons are more populated in rats under the stress of abrupt ethanol withdrawal than healthy rats [39].The p38 accumulation in this neuronal population is accompanied by poor motoric performance.As such, studies on p38 or BZD have revealed that both compounds (p38 and BZD) are associated with motoric impairment.This led us to the hypothesis that the up-regulation of Purkinje neuronal p38 mediates motoric deficit induced by chronic BZD.p38 is a cytosolic protein, but studies have demonstrated a link between p38 and mitochondria.For example, mitochondrial reactive O 2 species contribute to the phosphorylation of p38 during hypoxia in cardiomyocytes [40].When cells are treated with a p38 inhibitor (SB203580), their mitochondria are less damaged by oxidative stress than vehicle-treated cells [39].Based on these studies, we also tested whether chronic BZD treatment impairs mitochondrial respiration through Purkinje neuronal p38.To test these hypotheses, we have generated transgenic mice that lack p38 in Purkinje neurons.Here, we report that these p38-knockdown mice show resistance to motoric and mitochondrial impairment induced by chronic BZD.

BZD Injection and Motoric Test
All mice were two months old in the beginning of this study.Among BZDs, we selected lorazepam with an intermediate half-life in human.Lorazepam is currently one of the most frequently used BZDs in clinical settings [41].Clinicians prefer BZD with an intermediate halflife to extensively sedating long half-life BZDs [42,43].Lorazepam has been reported to impair motoric function in humans [8] and mice [11].In the current study, mice were injected with lorazepam (0.5 mg/kg) in the after-noon, and next morning they were tested for motoric function using Rotarod apparatus.This procedure was repeated for 14 days.The dose (0.5 mg/kg) of lorazepam was chosen based on our pilot study and previous studies in which lorazepam was used for anxiolytic effects [44][45][46].Rotarod is a motor driven treadmill (Omnitech Electronics, Columbus, OH) that measures running coordination and motor performance, such that a shorter latency to fall from an accelerating rod indicates poorer motor performance.The rotor consists of four cylinders that are mounted 35.5 cm above a padded surface.Mice were placed on the cylinder and a timer switch was simultaneously activated to rotate the cylinders.Acceleration continued until 44 rpm for maximum 90 seconds or animals fell to the padded surface, which simultaneously stopped the timer.Mice were tested for 3 sessions/day for 14 days with a 20 minutes resting period between sessions [47].

Generation of Purkinje-Neuron-Specific p38-Knockdown Mice
Among p38 isoforms, we selected p38α because it is the most abundant isozyme in the brain and the best characterized isoform [48,49].To avoid lethality, we employed a conditional transgenic mouse system to down-regulate Purkinje p38 genes using the Cre/loxP system and Pcp2 promoter (Purkinje neuron-specific marker) [50,51].Transgenic mice (Pcp2-Cre mice) that express Cre recombinase under the control of the Pcp2 (Jackson Laboratory) were cross-mated with floxed-p38α mice to generate the Pcp2-Cre +/− /p38 loxP/loxP mice.The mice with floxed-p38 were kindly provided by Boehringer Ingelheim Inc.The p38 floxed allele was generated by homologous recombination of embryonic stem cells in which two sites of ATG containing Purkinje p38 sequence were flanked by loxP [52,53] and excised in the presence of Pcp2-Cre.When pups were 21 days old, the tips of the tail were collected for genotype identification.

Immunohistochemical Detection of p38
Mice were anesthetized with isoflorane and perfused with 0.9% saline.The formalin-fixed and paraffin-embedded left hemispheres were cut into 8 μm-thick slices on a microtome.The slices were deparaffinized in xylene, rehydrated through a series of graded ethanol solutions, and washed with PBS.The slices were subsequently moisturized at 95˚C and incubated with primary antibody, polyclonal rabbit anti-phosphorylated p38α (pp38, an active form of p38) overnight at 4˚C.The slices were then incubated with broad spectrum poly HRP conjugate for 40 minutes at room temperature.The antigen-antibody bindings were visualized with a diaminobenzidine color reaction for 10 minutes.The slides were further rinsed, dehydrated through a series of graded ethanol and xylene, and mounted with Permount.All photographs were taken with a compatible Zeiss digital camera.A 20fold magnification was used to take pictures.

Semi-Quantitative Analysis of p38-Positive Purkinje Neurons
Brain slice samples were evaluated using the Carl Zeiss microscope, the image analysis program AxioVision 4 (Carl Zeiss, Thornwood, NY) and a previous method [54] that was modified for our purpose.Three mice per treatment group were evaluated.Six microscopic fields per mouse were selected, such that two microscopic fields were randomly selected from each of anterior (lobes I-V), medius (VI-VII), and posterior (VIII-X) regions of the cerebellar cortex [39,55].All Purkinje cells with visible p38-positive deposits were individually counted per microscopic field, and the length of the Purkinje layer per field was measured using a software program, Image Pro Plus (Media Cybernetics, Silver Spring, MD) to normalize the cell counts.Data were presented as the average number of p38-positive Purkinje neurons/Purkinje layer (mm) from the 18 data points (3 mice/group, 6 fields/mouse).

Mitochondrial Respiration
Mitochondrial respiration was assessed by measuring mitochondrial O 2 consumption rate (pmoles/minutes) according to a method provided by the XF respirometer manufacturer (Seahorse Bioscience).XF sensor cartridge was hydrated overnight in XF calibration buffer (at 37˚C, no CO 2 ).Isolated mitochondria were diluted with mitochondrial assay solution (Seahorse Bioscience, North Billerica, MA) to yield a final concentration of 200 µg/ml.Diluted mitochondria (50 µl) were transferred into each well of XF microplate and spun down at 4˚C for 10 -20 minutes at 2000 -3600 g.A consistent monolayer of mitochondrial adhesion to well bottom was visually ensured.A volume of 450 µl of succinate (5.5 mM) and rotenone (2.2 µM) was then added to each well.The XF microplate was warmed at 37˚C for 8 -10 minutes and placed in XF respirometer.Real-time (data are obtained while mitochondria respire) mitochondrial respiration was subsequently recorded every 5 -7 minutes.

Statistical Analysis
All numerical data are expressed as mean ± standard error of mean (SEM).The data were analyzed by oneway ANOVA as a factor of treatment or two-way ANOVA as a factor of treatment and days of testing.When a significant difference was observed, post-hoc Tukey's test was conducted to detect an individual group difference.p value was set less than 0.05 to indicate statistical significance.

Chronic BZD Inhibits Motoric Function
Figure 1 illustrates the motoric performance of mice that were injected with BZD or vehicle (methyl cellulose) for 14 days.Two-way ANOVA was conducted as a factor of treatment and days of testing.Compared to vehicle-injected mice, BZD-injected mice fell quicker (a shorter latency, poor performance) from Rotarod from day 9 to the end of the test (day 14) ( * p < 0.01).There was no significant difference in Rotarod performance between day 1 and 8 of the test.These data indicate that chronic rather than acute BZD results in motoric deficit.

Chronic BZD Increases the Number of p38-Containing Purkinje Neurons
We tested whether chronic BZD affects p38 expression in Purkinje neurons.Although p38 immunoreactivity was found in other types of neurons, such as granular neurons, we focused on Purkinje neurons because Purkinje neurons are the sole output of the cerebellar cortex and play a major role in motor function [56].The immunohistochemical photographs revealed that p38 (as a form of pp38) immunoreactivity (dark deposits) was more distinctively visualized in the Purkinje neurons along the Purkinje layer of BZD-injected mice than vehicle-injected mice (Figure 2).

p38-Knockdown Mice Show Better Motoric Performance than Wild-Type Mice after Chronic BZD Injection
If Purkinje-neuron-specific p38 contributes to BZD-induced motoric deficit, the down-regulation of the p38 would attenuate such effect of BZD.To test this hypothesis, p38-knockdown and wild-type mice were injected with BZD and tested on Rotarod, and this procedure was repeated for 14 days.BZD-injected p38knockdown mice were able to run on Rotarod longer than BZD-injected wild-type mice throughout the test period (Figure 4(a), * p < 0.01), an indicator of better motoric performance.There was no significant difference in motoric performance between BZD-injected and vehicle-injected p38-knockdown mice (data not shown).When p38-containing Purkinje neurons were examined, chronic BZD treatment failed to increase the number of p38-positive Purkinje neurons in p38-knockdown mice compared to wild-type mice (Figure 4(b)).These data indicate that Purkinje neuronal p38 mediates BZD-induced motoric deficit.
When the number of p38-positive Purkinje neurons were computed, the number was higher in BZD-injected mice than vehicle-injected mice ( * p < 0.001).These data indicate that chronic BZD use results in p38 accumulation in cerebellar major neurons.

p38-Knockdown Mice Lack p38 in Purkinje Neurons
We generated transgenic mice that were designed to lack p38 in Purkinje neurons to test the role of Purkinje neuronal p38 in BZD-induced motoric deficit.Our recent study reported the DNA results from genotyping, such as the DNA images of Pcp2-Cre at 100 base pair, homozygous p38 loxP/loxP at 414 base pair, and an internal positive

Chronic BZD Suppresses Mitochondrial Respiration in a Manner that Is Prevented by p38-Knockdown
The effect of chronic BZD on mitochondrial respiration was tested in p38-knockdown and wile-type mice.As seen in Figure 5, mice injected with chronic BZD show significantly lower mitochondrial respiration than vehicle-injected mice ( * p < 0.01).This difference was almost .Chronic BZD suppresses mitochondrial respiration in a manner that is prevented by p38-knockdown.p38-knockdown mice and wild-type mice were injected with lorazepam (0.5 mg/kg, I.P.) for 14 days.Next morning, these mice were euthanized under anesthetics and mitochondria were isolated from cerebellum to assess mitochondrial respiration using XF respirometer.Among mice groups, mitochondrial respiration was decreased only in BZD-injected wild-type mice.* p < 0.01 vs. three other groups.Data are presented as mean ± SEM out of all data collected at four time points of measurements with 5 -7 minutes interval between time points for 5 or 6 mice/group.
completely abolished in p38-knockdown mice injected with BZD or vehicle.These data indicate that Purkinje neuronal p38 mediates BZD-induced mitochondrial respiratory suppression.

DISCUSSION
The key finding of the current study is that chronic BZD treatment impedes motoric function and mitochondrial respiration in a manner that is protected by the downregulation of Purkinje neuron-specific p38.These results suggest that chronic BZD acts as a stressor, provoking the excessive accumulation of p38 in vulnerable neurons accompanied by behavioral deficit.The most studied mechanism of BZDs or BZD analogues is the enhancement of inhibitory GABA neurotransmission through GABA-A receptors, thereby exerting therapeutic effects such as anxiolytic, sedative, or anti-seizure effects [57].While BZD is powerfully effective and largely safe, evidence indicates that the prolonged use of BZDs often creates adverse effects including motoric deficit [2,7,9,58].Nearly all the disadvantages of BZDs result from long-term use [59], but most mechanistic studies on motor deficit have focused on BZD's acute effect.Savic et al. [60] have studied the effects of a single dose of BZD analogues on the motor performance of rats and reported that GABA-A receptor α1 subunit mediates the ataxic potential of BZDs.Similarly, Korpi et al. [10] have reported that GABA-A receptor α6 subunit mediates motoric deficit associated with an acute dose of diazepam.In a study done by Stanley et al. [11], the motor incoordination induced by a single dose of diazepam and lorazepam required a high occupancy of a GABA-A receptor.In comparison with these studies that employed an acute dose regimen of BZD, we studied the effect of a prolonged BZD use on motoric function.The majority of patients who receive BZD therapy continue to take BZD for an extended period of time.This phenomenon is more prevalent in women than men [9], which increases their risk of motoric deficit associated with the prolonged use of BZDs [2,7,9,58].As shown in Figure 1, mice treated with BZD showed no significant difference in motoric performance compared to vehicle-treated mice during the early stage of BZD injection (day 1 to 8).However, BZD-injected mice began to show poorer performance than vehicletreated mice upon the further treatment of BZD (from day 9).The motoric impairment seen in BZD-injected mice is unlikely due to the sedative effects of BZD for at least two reasons.Firstly, the motoric behavior of these mice did not differ from control mice during the initial several days of BZD injection.Secondly, mice were tested on Rotarod 20 hours after the injection of lorazepam of which half-life is 70 minutes [44].The motoric impairment induced by repeated BZD injection may suggest that chronic BZD provokes permanent or long lasting changes in the brain.Our data indicate that one of the BZD-induced changes occurs at the level of Purkinje neuronal p38; BZD-treated mice show a higher number of Purkinje neurons containing p38 than vehicle-treated mice.
We chose Purkinje neurons among multiple types of cerebellar neurons.Purkinje neurons constitute the sole output of the cerebellar cortex, governing all movement [56] and coordination [61].Damage to these neurons inevitably provokes the impairment of motor behavior and coordination [55,56,62].Our recent study reveals that motoric deficit occurs in ethanol withdrawn rats accompanied by p38 accumulation in Purkinje neurons.This finding prompted us to test whether BZD-associated motoric deficit is mediated through p38 accumulation in this neuronal population.Our behavioral and immunohistochemical tests demonstrate that mice treated with chronic BZD show poor motoric performance and a significantly increased number of p38-positive Purkinje neurons.However, this observation is only correlational and does not necessarily indicate that the accumulation of Purkinje-specific p38 is responsible for the motoric deficit.We, therefore, generated mice that lack p38 specifically in Purkinje neurons.These mice indeed show substantially better motoric performance than wild-type mice under chronic BZD therapy, proving that Purkinje p38 contributes to BZD-induced motoric impairment.A previous study showed that a p38 inhibitor protects immature, vulnerable Purkinje neurons from neurotoxic stress [63].Mice with Purkinje neuron degeneration have shown a spatial learning deficit in the water maze test [61,64,65].These studies along with our current findings suggest that aberrant p38 activation/accumulation perturbs the movement-controlling function of Purkinje neurons.
Mitochondria are the main source of energy supply for cellular and neuronal survival.Mitochondria consume most of inhaled O 2 , and such O 2 -consuming mitochondrial respiration is obligatorily coupled with ATP production.In humans or experimental animals, mitochondrial defects are often observed in a variety of CNS disorders and brain aging [66].BZD has been reported for both beneficial [67] and adverse [68,69] effects on mitochondria.BZD (diazepam) blocked the apoptotic release of cytochrome c from mitochondria [67], suggesting that BZD has a mito-protective effect.In contrast, several BZDs including diazepam, suppressed mitochondrial respiratory control ratio (State III/State IV) in rat kidney [70].Diazepam also induced mitochondrial swelling [71], suggesting that BZDs have adverse effects on mitochondria.It is not clear what underlies the beneficial and adverse effects of BZD on mitochondria.One possibility is that BZD exerts differential effects on mitochondria depending upon the target area or function of mitochondria.It is also possible that the effects of BZD on mitochondria vary depending upon the duration or dose of BZD treatment.As far as the effect of chronic BZD on mitochondrial respiration is concerned, our current study demonstrates that chronic BZD hinders mitochondrial respiration accompanied by motoric impairment.
Although p38 is mainly located in cytosol, studies have demonstrated the effects of p38 on mitochondria.For instance, p38 inhibits the adverse effect of a nuclear receptor (PPARγ) coactivator on mitochondrial respiretion in muscle [72].p38 activation is associated with free radical production [53].p38 inhibitor (SB203580) mitigates the swelling of mitochondrial membrane induced by ethanol withdrawal stress (unpublished observation).These studies suggest that there is a crosstalk between cytosolic p38 and mitochondria.Our results strengthen this view in that the down-regulation of Purkinje p38 attenuates the suppressing effect of BZD on mitochondrial respiration (Figure 5).How the inhibition of cytosolic protein p38 mediates mitochondrial respiratory enhancement is an open question.p38 activation induces BAX (apoptotic protein) translocation to mitochondria, and BAX subsequently impedes mitochondrial respiration [73].Therefore, p38 depletion would result in an opposite effect, the enhancement of mitochondrial respiration.Mitochondria are one of the major targets of estrogen, a neuroprotective steroid that directly acts on cytosol and mitochondria [74].Thus, it is also possible that estrogen may interfere between p38 and BAX, pro-tecting mitochondria.Alternatively, BZD inhibits ventilatory response [75,76], reducing the amount of O 2 available for mitochondrial respiration.Although this possibility cannot be ruled out, this is a remote possibility because in general, BZD-induced ventilatory inhibition occurs at a much higher dose than an anxiolytic dose used in the current study.Unfortunately, we were not able to determine a direct relationship among BZD-induced mitochondrial respiratory deficits, motoric impairment, and excessive p38.The administration of mitochondrial respiratory inhibitors to brain could test whether the inhibitors blunts the motor-improving effects of p38-down-regulation.However, we encountered a technical limitation of delivering mitochondrial respiratory inhibitors to brain without creating the toxic or non-specific effects of the inhibitors.
In conclusion, our findings provide empirical evidence that chronic BZD impedes motoric function at least partly through the accumulation of p38 in Purkinje neurons.The current study may provide a new mechanistic insight involving p38 into BZD-induced motoric deficit.

Figure 1 .
Figure 1.Chronic BZD inhibits motoric function.Mice were injected with lorazepam (0.5 mg/kg, I.P.) at 2 PM and tested on Rotarod (3 sessions/day) the next morning (10 AM).This procedure was repeated daily for 14 days.Compared to vehicle-injected mice, BZD-injected mice show poorer motoric performance from day 9 of testing.* p < 0.01 vs. vehicleinjected mice.Data are presented as mean ± SEM out of all data collected during the time period indicated in X axis.N = 4 -7 mice/group.

Figure 2 .Figure 3 .Figure 4 .
Figure 2. Chronic BZD increases the number of p38-containing Purkinje neurons.Mice were injected with lorazepam (0.5 mg/kg, I.P.) for 14 days.Next morning, they were deeply anesthetized and perfused intracardially with saline followed by paraformaldehyde.Left hemisphere containing the cerebellar vermis was processed for immunohistochemical analysis.All photographs were taken of the cerebellar cortex area containing Purkinje layers that showed a clear image across all treatment groups.Dark deposits marked with arrows indicate p38 immunoreactivity in Purkinje neurons along the Purkinje layer, and they more distinctively appeared in BZD-injected mice than vehicle-injected mice.p38-containing Purkinje neurons were counted using a software program Image Pro Plus, revealing that BZD-injected mice had a higher number of p38-positive Purkinje neurons/Purkinje layer (mm) than vehicle-injected mice.* p < 0.001 vs. vehicle-injected mice.The scale bar indicates an actual length of 100 µm.Depicted are mean ± SEM for 6 microscopic fields/mouse for 3 mice/group.

Figure 5
Figure5.Chronic BZD suppresses mitochondrial respiration in a manner that is prevented by p38-knockdown.p38-knockdown mice and wild-type mice were injected with lorazepam (0.5 mg/kg, I.P.) for 14 days.Next morning, these mice were euthanized under anesthetics and mitochondria were isolated from cerebellum to assess mitochondrial respiration using XF respirometer.Among mice groups, mitochondrial respiration was decreased only in BZD-injected wild-type mice.* p < 0.01 vs. three other groups.Data are presented as mean ± SEM out of all data collected at four time points of measurements with 5 -7 minutes interval between time points for 5 or 6 mice/group.