Oxytocin modulates GABAAR subunits to confer neuroprotection in stroke in vitro

Oxytocin protects against ischemia-induced inflammation and oxidative stress, and is associated with GABA (γ-aminobutyric acid, an inhibitory neurotransmitter) signaling transduction in neurons. However, the molecular mechanism by which oxytocin affords neuroprotection, especially the interaction between oxytocin receptor and GABAA receptor (GABAAR), remains to be elucidated. Primary rat neural cells were exposed to oxytocin before induction of experimental acute stroke model via oxygen-glucose deprivation-reperfusion (OGD/R) injury. Pretreatment with oxytocin increased cell viability, decreased the cell damage against oxidative stress, and prevented the release of high mobility group box1 during OGD/R. However, introduction of oxytocin during OGD/R did not induce neuroprotection. Although oxytocin did not affect the glutathione-related cellular metabolism before OGD, oxytocin modulated the expression levels of GABAAR subunits, which function to remove excessive neuronal excitability via chloride ion influx. Oxytocin-pretreated cells significantly increased the chloride ion influx in response to GABA and THIP (δ-GABAAR specific agonist). This study provides evidence that oxytocin regulated GABAAR subunits in affording neuroprotection against OGD/R injury.

suggesting that the number of membrane-bound GABA A Rs could be a pivotal process in the progression of ischemic-induced neuronal cell death 12 . Oxytocin regulates GABA A R-mediated synaptic signaling in the fetal brain during delivery, and reduces brain vulnerability to hypoxic damage 13 . Although oxytocin-induced neuroprotection has been demonstrated in ischemic-reperfusion injury models, the molecular mechanisms underlying such therapeutic benefit, especially how oxytocin interacts with individual GABA A R subtypes 14 , are still unknown.
In this study, we demonstrated that administration of oxytocin in primary rat neural cells (PRNCs) before OGD resulted in robust neuroprotective effects, but not when oxytocin was initiated during OGD/R. We also showed that oxytocin shifted the expression patterns of GABA A R subunit on the cells, accompanied by increased chloride ion influx. These observations provide evidence that oxytocin modulated GABA A R in exerting its neuroprotective effects against ischemia-induced neuronal cell death.
Biological activity readouts across treatments. Oxytocin acts as an anabolic hormone, and exhibits cell growth 5,6 and anti-oxidative properties, suggesting its potential therapeutic application in stroke 18 . However,  (A) Cell viability tested by Calcine-AM/EthD-1 florescence and trypan blue dyes. (B) Mitochondrial activity by MTT assay. (C) GSSG activity. (D) Extracellular HMGB1 levels. *P < 0.05, **P < 0.01, and ***P < 0.001. Experiments were conducted in triplicate, with n = 6 per treatment condition in each run. oxytocin administration did not alter cell growth of PRNCs compared with control treatment (Fig. 3A). Next, because peripheral oxytocin participates in glucose metabolism in modulating reactive oxygen species (ROS) production via NADPH (nicotinamide adenine dinucleotide phosphate) pathway, we examined the effects of oxytocin on glutathione (a major antioxidant), and glucose 6-phosphate dehydrogenase (G6PD) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (key enzymes that regulate NADPH production) 19 . Results revealed that G6PD, GAPDH, and glutathione disulfide (GSSG) activity levels did not significantly differ across treatment conditions ( Fig. 3B-D). Pretreating the cells with oxytocin did not change the protein expression levels of G6PD and GAPDH in comparison with control values (Fig. 4). Oxytocin had no significant effects on the cell growth, although oxytocin has been shown to contribute to the differentiation of bone marrow-derived mesenchymal stem cells 6 (B) G6PD activity. G6PD catalyzes the rate-determining step in the pentose phosphate pathway and produces NADPH to regulate GSH/GSSG levels. G6PD activities across treatments were not significantly different. (C) GAPDH activity. GAPDH regulates the ATP generation phase of glycolysis-derived NAD and functions as a reversible metabolic switch under oxidative stress. GAPDH activities across treatments did not significantly differ. (D) GSSG activity. GSSG is a biomarker of oxidative stress, and is generated by oxidized GSH with reduced NADPH. There were no statistical differences across conditions. Experiments were conducted in triplicate, with n = 6 per treatment condition in each run. Blue bars represent significantly increased protein expression levels, red bar showed significantly decreased levels, and white bars indicate no statistical differences between oxytocin-treated cells and control. *P < 0.05 and **P < 0.01. The dotted red line represents combined data from control and OXT + ATS, since these two groups did not significantly differ. Experiments were independently conducted in 3~6 times. Oxytocin pretreatment shifts the GABA A R subunit expression patterns. Oxytocin has been found to alter the subtype expression patterns 20 and function of GABA A Rs 13,21 . We therefore assessed whether oxytocin modulated the expression patterns of GABA A Rs subunit on PRNCs (Fig. 4). Treatment of PRNCs with oxytocin significantly increased α 4 , β 3 , δ , and ε GABA A R subunit expression levels, but decreased γ 2 GABA A R subunit (Fig. 4). Although oxytocin receptor expression has been reported to be increased following oxytocin treatment 4 , we could not detect any differences in oxytocin receptor upregulation between control and oxytocin treatment in this experiment (Fig. 4). Moreover, oxytocin treatment did not affect the expression levels of Bestrophin-1 (BST1), a calcium-activated chloride ion channel normally distributed on synapses adjacent to soma 22 and shown to mediate GABA release from astrocytes 23 .
Localization of GABA A R subunits. Electric current and localization patterns of GABA A R vary depending on the region of interest within the neuron, because glycogen (the main storage form of glucose in the body) is predominantly preserved in the soma where the main production of ATP occurs. In the ischemic brain, the rate of glycogen metabolism is significantly increased 24 . We observed the localization of α 4 , β 3 , γ 2 , δ , and ε GABA A R subunit expression within subcellular compartments of the neuron. Immunocytochemichal analysis showed that (i) δ GABA A R subunit was mainly located at the axon (Fig. 5A, indicated with box), (ii) ε GABA A R subunit predominantly resided in the soma (Fig. 5B, indicated with box), (iii) γ 2 , α 4 , and β 3 GABA A R subunits were broadly expressed in the whole neuron ( Fig. 5A,B,E,F), (iv) α 4 /δ -and β 3 /δ -GABA A R subunits showed co-localization ( Fig. 5C,D, indicated with arrow), and (v) α 4 /ε -and β 3 /ε -GABA A R subunits were also co-localized in PRNCs (Fig. 5E,F, indicated with arrow).

Intracellular chloride ion influx kinetics.
After binding with GABA, GABA A R engages a chloride ion selective pore, resulting in chloride ion influx that inhibits the firing of neuron action potentials. The kinetic property of GABA A R depends on receptor subunit compositions, thereby providing a mechanism for neurons to regulate individual biological activities. We performed a time course study to reveal any differences in GABA-elicited chloride ion influxes between control-and oxytocin-treated PRNCs. Figure 6A revealed that chloride ion influx reached equilibrium at 10 min, but treatment with oxytocin significantly increased the influx at the 20 min period. To assess the differences of both GABA A R antagonistic conditions, we compared the inhibition dynamics of GABA-induced chloride ion influx in the presence of flumazenil (GABA A R antagonist, GABA + FLU) or picrotoxin (GABA A R channel blocker, GABA + PIC). Both reagents inhibited the GABA-induced chloride ion influx (control; Fig. 6B and oxytocin treatment; Fig. 6C). Interestingly, oxytocin-treated cells were more sensitive to picrotoxin inhibition, as evidenced by the Δ value of control cells = 14.0 ± 2.20 AU (Fig. 6B), and that of oxytocin-treated cells = 24.1 ± 1.70 AU (Fig. 6C), P < 0.001. Because δ GABA A Rs display increased sensitivity to THIP (δ -GABA A R specific agonist) 25 , we tested whether oxytocin-treated cells additively increased THIP-induced chloride ion influx. Results revealed that THIP-evoked chloride ion influx of oxytocin-treated cells was significantly higher than that of control (Fig. 6D).

Discussion
The present study revealed a novel molecular mechanism underlying oxytocin-mediated neuroprotection against ischemic stroke in a cell culture paradigm. We found that oxytocin-induced GABA A R subunit modification is a predominant factor in conferring neuroprotection against OGD. GABA is the principal inhibitory transmitter in the brain, and its functions are mediated by ubiquitously expressed ligand-opened chloride ion channel GABA A Rs 26 . Aberrant GABAergic inhibition is a key pathological feature displayed by ischemic neurons in the peri-infarct area (secondary damaged region) after stroke 26 . Our present results demonstrated that oxytocin reduced ischemic stroke deficits likely by modulating specific GABA A R subtype signal transduction 14 , which parallels studies showing that oxytocin improves stroke outcomes via social interaction pathways 18 .
We showed that oxytocin protected PRNCs against OGD (Fig. 2). Ischemic injury is mediated by ROS, generated primarily by damaged mitochondria 27 , which leads to apoptosis and necrosis. During OGD, cell viability and mitochondrial activity were decreased, and the GSSG activity and extracellular HMGB1 levels were increased. HMGB1, a non-histone DNA-binding protein, is released from necrotic neurons after 2 h OGD 17 , and its concentrations in serum are significantly increased in stroke patients due to blood brain barrier (BBB) disruption associated with the disease progression 11,17 . That oxytocin exerted neuroprotection in OGD, but not in the OGD/R model is consistent with in vivo evidence, demonstrating that the subsequent reperfusion after ischemia exacerbates neuronal functions and causes massive brain injuries when oxygen-saturated and nutrient-rich blood suddenly returns to the lesion after a period of ischemia 11 , suggesting that OGD/R is worse than OGD. Under the OGD condition, pretreatment with oxytocin increased cell viability and mitochondrial activity, decreased the GSSG activity, and prevented HMGB1 secretion from the cells. In the presence of atosiban, this neuroprotection was abolished, indicating that the therapeutic effect was likely mediated by oxytocin receptor signal transduction. HMGB1 is phosphorylated by protein kinase C 28 and calcium/calmodulin-dependent protein kinase 29 . Although oxytocin is capable of activating both kinases 3 , we could not detect extracellular HMGB1 despite incubating the cells with oxytocin prior OGD. Altogether, these observations suggest that oxytocin could serve as a neuroprotective agent in the acute phase of stroke by acting as an ischemic preconditioning factor in modulating therapeutic protein synthesis.
Oxytocin regulates glucose uptake that is critical for stem cell growth 6 and antioxidant activity 30 . However, cell growth of PRNCs was not affected by oxytocin (Fig. 3A). We thus tested whether oxytocin utilized its receptor signal transduction in regulating glutathione-related proteins (G6PD, GAPDH, GSSG). G6PD regulates the antioxidant activity of NADPH 31 , facilitating NADPH to maintain glutathione/GSSG recycling 32 . GAPDH is not only a key enzyme in glycolysis, but also phosphorylates the α 1 GABA A R subunit for sustaining the GABA A R structure and stability, thereby establishing the link of GABAergic inhibition with glucose metabolism. Under ischemic condition, glycolytic flux increases, while GAPDH activity is reversibly decreased, inhibiting neuronal cells from producing NADPH 31 as a result of increased G6PD activity. We speculated that oxytocin would normalize G6PD, GAPDH, GSSG activity, but no significant difference were detected across treatment conditions (Fig. 3B-D), suggesting that oxytocin's neuroprotection is largely independent of G6PD, GAPDH, and GSSG signal transduction pathways.
Maternal oxytocin exerts neuroprotective action on fetal neurons during parturition (a perturbed physiological environment similar to hypoxic-ischemic brain) mediated by GABA A R signaling pathway 13 . In the present study, oxytocin likely engaged the GABA A R subunit expression patterns by enhancing α 4 , β 3 , δ , and ε GABA A R subunit expression levels while reducing γ 2 GABA A R subunit on PRNCs (Fig. 4). Although most GABA A R subunit expressions were not significantly influenced in the presence of atosiban (OXT + ATS), α 2 GABA A R subunit expression decreased and γ 1 and π GABA A R subunits increased (Fig. 4A), implicating that redundant signal impedance of oxytocin receptor could activate alternative signal transduction of these subunit expressions. Conversely, an on/off signaling switch of oxytocin may be tightly regulated by engagement with GABA A R subunits.
The β 3 and γ 2 GABA A R subunits on the neuronal membrane are vulnerable to ischemic stroke 10,33 . In the late stage of rat pregnancy the maternal brain displays increased expression of the ε GABA A R subunit, which is responsible for the respiratory function 34 . The α 4 GABA A R subunit, on the other hand, is associated with dendritic development 35 , and is well co-expressed with δ -subunit in the brain 8,36 . The α 4 , δ , and ε GABA A R subtypes are extrasynaptic GABA A Rs 8,36 , which mediate tonic inhibition upon activation by GABA spillover from synaptic sites, as well as by ambient GABA in the extracellular space. The majority of GABA A Rs are α 1 β 2 γ 2 , α 2 β 3 γ 2 , and α 3 β 2/3 γ 2 , approximately occupying 80% of total GABA A R expressions in the brain 8 , in contrast, α 4 , δ , and ε GABA A Rs represent less than 5% 8 . Oxytocin-treated cells significantly increased α 4 , δ , and ε GABA A R subunit expression levels and decreased γ 2 GABA A R subunit on PRNCs (Fig. 4). It is conceivable that oxytocin pretreatment led to an upregulation of specific GABA A R subtype expression levels, which in turn might have modified the neuronal networks towards neuroprotection.
Electrical properties of a neuron vary along the segments of subcellular organization (soma, dendrites, and axon), which is essential for orchestrating cellular function and structure preservation 37 . GABA A R-mediated chloride ion fluctuation substantially changes the intracellular chloride ion concentration in the soma and spreads into the dendrites 38 . Hypoxic ischemia causes retrograde neurodegeneration, which shortens the axonal and dendritic lengths and swells the soma, and produces a rapid and significant loss of axon in the acute phase of injury 39 . Therefore, it is an important to elucidate the GABA A R subunit localizations at the subcellular neuron. Interestingly, δ GABA A R subunit is mainly expressed at axon, and ε GABA A R subunit is primarily distributed on soma which concurred with previous report 40 . In contrast, α 4 , β 3 and γ 2 GABA A R subunits were broadly expressed throughout the whole neuron 8 . Here, we found that α 4 /δ -and β 3 /δ -GABA A R subunits, and α 4 /ε -and Scientific RepoRts | 6:35659 | DOI: 10.1038/srep35659 β 3 /ε -GABA A R subunits were well co-localized on neurons (Fig. 5). That appropriate distribution and specific expression of GABA A Rs subtypes exist in neurons suggest that oxytocin could elicit neuroprotection by subcellularly targeting specific GABA A R subunits within ischemic neurons.
Under hypoxic ischemia, the extracellular GABA concentration on/around the synaptic cleft increases and elevates neuronal intracellular chloride ion, which functions as a counter-reaction of depolarization 41 . We demonstrated here that oxytocin modulated discrete GABA A R subunits tasked to monitor chloride ion influx. The kinetic of GABA-stimulated chloride ion influx on oxytocin-treated cells was altered (Fig. 6A), in that while the response to GABA by the control cells was saturated at 10 min, oxytocin-treated cells continuously evoked the ion influx for 20 minutes, suggesting that boosting GABA A Rs-mediated neuronal inhibition can afford substantial protection while minimizing the extent of neuronal cell loss during OGD. Following incubation of cells with 50 μ M GABA for 10 min (Fig. 6A), the values of chloride ion influx were similar. We next assessed GABA A R antagonism and found that Flumazenil inhibited the γ contained GABA A Rs in response to GABA, but was not able to antagonize the δ and ε GABA A Rs 42 . Picrotoxin directly binds the ion pore of GABA A R, thereby regulating the influx of chloride ion, and inhibiting the whole GABA A R channel activity 42 . The inhibition ability of flumazenil (GABA + FLU) did not significantly differ between conditions, implicating that oxytocin administration had no effect on the expression levels of γ GABA A R subtypes (Fig. 6B,C). In contrast, oxytocin-treated cells exhibited significant inhibition of chloride ion influx following picrotoxin treatment (GABA + PIC) compared to control, suggesting that oxytocin significantly increased total GABA A R expression, especially δ and ε GABA A R subtypes. Of note, to date, there is no specific antagonist for δ and ε GABA A R subunits. The present observation of specialized GABA A Rs antagonism is also supported by δ -GABA A R specific agonist THIP significantly elevating the chloride ion influx of oxytocin-treated cells compared to control (Fig. 6D). In summary, oxytocin induced the shift of GABA A R subunit expression in cultured PRNCs, which likely changed the kinetics of chloride ion influx in response to GABA.
We demonstrated that oxytocin exerts neuroprotection against ischemic stroke, but requires its treatment initiation prior to injury induction. Oxytocin may serve as a pharmacological ischemic preconditioning factor that can engage GABA A R towards neuroprotection. The present results provide evidence that oxytocin altered the expression patterns of GABA A R subunit and the kinetics of GABA-induced chloride ion influx. Our study highlights a close interaction between oxytocin and GABA A R that should aid in our understanding of stroke pathology and its treatment.

Methods
Cell culture and oxygen-glucose deprivation-reperfusion (OGD/R) progression. Primary rat neural cells (PRNCs; consisted of 40% neurons and 60% astrocytes) were obtained from BrainBit (E18 rat cortex; Springfield, IL, USA). As described elsewhere 15 , cells (4 × 10 4 cells/well) were suspended in 200 μ l Neural Medium (NbActive 4, BrainBit) containing 2 mM l-glutamine and 2% B27 in the absence of antibiotics and grown in poly-l-lysine-coated 96-well plates at 37 °C in humidified atmosphere containing 95% O 2 and 5% CO 2 . After 3 days in culture, PRNCs were exposed to 1 μ M oxytocin (O4375, Sigma-Aldrich, St. Louis, MI, USA), 1 μ M oxytocin + 10 μ M atosiban (A3480, Sigma-Aldrich), 10 μ M atosiban, and the absence of regents (control) for 3 days at 37 °C. After 6 days in culture (Fig. 1), RPNCs were exposed to OGD as described previously 15 . The cells were initially exposed to OGD medium (glucose-free Dulbecco's Modified Eagle Medium), then placed in an anaerobic chamber containing 95% N 2 and 5% CO 2 for 15 min at 37 °C (preincubation), for 90 min at 37 °C (culture medium pH 6.7~6.8; mimicking the acidic environment of ischemic brain in vivo). OGD was terminated by adding 5 mM glucose to medium and cell cultures were re-introduced to the regular CO 2 incubator at 37 °C for 2 h. Control cells were incubated in the same buffer containing 5 mM glucose at 37 °C in a regular 95% O 2 and 5% CO 2 incubator. Measurement of cell viability. Measurement of cell viability was performed using fluorescent live/ dead cell assay and trypan blue exclusion method 15,43 . Following treatment, the cells were incubated with 2 μ M Calcein-AM and 4 μ M EthD-1 (L3224 Invitrogen, Waltham, MA, USA) for 45 min at room temperature (RT) in dark. After washing once with phosphate buffer saline (PBS), the green fluorescence of the live cells was measured by the Gemini EX florescence plate reader (Ex/Em = 490/520; Molecular Devices, Sunnyvale, CA). In addition, trypan blue (15250, Gibco, Waltham, MS, USA) exclusion method was conducted and mean viable cell counts were calculated in 16 randomly selected areas (1 mm 2 , n = 10) to reveal the cell viability. Briefly, within 5 min after adding trypan blue, we digitally captured under microscope (200x) 10 pictures (approximately 100 cells/picture) for each condition, then randomly selected 5 pictures, and counted the number of cells for each individual treatment condition. Normalized cell viability was calculated from the following equation: viable cells (%) = [1.00 -(Number of blue cells /Number of total cells)] × 100. To precisely calibrate the cell viability, the values were standardized from fluorescence intensity and trypan blue data.
Measurement of extracellular high mobility group box1 (HMGB1) levels and glutathione disulfide (GSSH) activity. After OGD/R, culture medium was centrifuged at 3,000 g, 4 °C for 15 min, and the supernatant was processed for detection of HMGB1 using an ELISA kit (amin416082, Antibody, Atlanta, GA, USA) with absorbance measured at 450 nm on a Synergy HT plate reader (Bio-Tex). Cells were treated with oxidized glutathione lysis reagent (V6611, Promega, Fitchburg, WI, USA), and GSSG activity, a biomarker of reactive Scientific RepoRts | 6:35659 | DOI: 10.1038/srep35659 Data analysis. Data were evaluated using one-way analysis of variance (ANOVA) followed by post hoc compromised t-tests (GraphPad Prism 6 ® software). Statistical significance was preset at P < 0.05. Data are represented as means ± SD from quintuplicates of each treatment condition.