Taurine activates glycine and GABA_A receptor currents in anoxia-tolerant painted turtle pyramidal neurons

The majority of vertebrate species cannot survive for more than a few minutes without oxygen; however, seasonal ice cover in the northern hemisphere has selected for the ability to survive for long periods without oxygen in some species, including painted turtles, goldfish and crucian carp (Bickler and Buck, 2007). The western painted turtle (Chrysemys picta bellii) inhabits this region and can tolerate anoxia for at least 12 h at 20°C and 126 days at 3°C by down-regulating ATP-consuming processes (Herbert and Jackson, 1985; Jackson, 2000; Bickler and Buck, 2007; Staples and Buck, 2009). While the mechanisms by which ATP-consuming processes are systemically reduced in anoxic turtle brain are not clear, maintaining and re-establishing homeostatic ion concentrations across the neuronal plasma membrane following an action potential (AP) is an ATP-demanding process and is thus a critical point of control.

The primary mode of synaptic neurotransmission occurs through the release of glutamate or gamma-aminobutyric acid (GABA). In anoxia-intolerant mammals, homeostatic regulation of these neurotransmitters is destabilized after minutes of anoxia and excessive release of glutamate/GABA leads to cell swelling and excitotoxic cell death (Choi, 1992; Allen et al., 2004). This response is absent in the anoxia-tolerant western painted turtle, as extracellular glutamate levels do not change after 4 h of anoxia and glutamatergic receptor currents are arrested: whole-cell α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartate (NMDA) receptor currents are reduced by about 50% after 40 min of anoxia in a sheet of cortical tissue (Pamenter et al., 2008a,b; Zivkovic and Buck, 2010; Shin and Buck, 2003; Bickler et al., 2000). This phenomenon is termed ‘channel arrest’ and corresponds to a reduction in excitatory neurotransmission (Hochachka, 1986; Doll et al., 1991). Conversely, extracellular GABA levels are elevated after 90 min of anoxia and continue to increase up to 4 h, where they can reach 90 times the normoxic level (Nilsson and Lutz, 1991). The elevation in extracellular GABA consequently produces an increase in GABA_A-receptor activity: phasic and tonic post-synaptic GABA_A-receptor currents both double in amplitude and increase whole-cell conductance 1.5-fold within 10 min of anoxia (Pamenter et al., 2011; Hogg et al., 2014). This results in membrane potential moving towards the reversal potential of the GABA receptor and its prolonged conductance clamps membrane potential below threshold in a ‘shunting inhibition’ fashion (Pamenter et al., 2011; Hogg et al., 2015), resulting in ‘spike arrest’ (Sick et al., 1993). Collectively, both channel arrest and spike arrest serve to reduce ATP expenditure in the turtle brain.

The role that neurotransmitters other than GABA and glutamate have in regulating the electrical activity in the anoxic turtle brain is not known. While extracellular GABA levels rise ∼80-fold after 4 h, glycine and taurine levels rise ∼4-fold and ∼24-fold, respectively (Nilsson and Lutz, 1991). Glycine, along with GABA, is considered one of the major inhibitory neurotransmitters in the central nervous system of vertebrates and typically activates ionotropic glycine receptors, allowing Cl⁻ ions to enter the cell and hyperpolarize the membrane potential (Dutertre et al., 2012). However, glycine is not usually considered a predominant inhibitory molecule within the cerebrocortex, as its activity is largely localized to the spinal cord, brainstem and retina (Dutertre et al., 2012). Therefore, the role of taurine in the anoxic turtle cerebral cortex is of greater interest, especially since it has never been investigated in the context of anoxia tolerance.

Since the discovery of taurine in ox (Bos taurus) bile acid, (Tiedemann and Gmelin, 1827), taurine has been found in other organisms and is associated with a number of physiological functions (Huxtable, 1992; Lambert et al., 2015; Ripps and Shen, 2012; Albrecht and Schousboe, 2005). It is the second most abundant amino acid within the central nervous system, after glutamate, suggesting it has an important role to play (Palkovits et al., 1986). One such role is osmoregulation, since it is a relatively inert zwitterion and a non-essential molecule that has a large concentration gradient across the plasma membrane, with an intracellular concentration ∼400-fold higher than its extracellular concentration (Pasantes-Morales and Schousboe, 1997). When osmotic stress occurs, taurine can be transported across the membrane to offset osmotic changes without changes in concentration drastically altering membrane potential, enzymatic activities or other cellular processes (Huxtable, 1992; Lambert et al., 2015). However, neurons in the anoxic turtle brain shrink as a result of water leaving the cell as Cl⁻ exits through GABA_A receptors (Pamenter et al., 2011). Therefore, an increase in taurine for osmoregulation likely represents a mechanism used by anoxia-intolerant species rather than by anoxia-tolerant organisms. Taurine also acts as an antioxidant, as it attenuates formation of reactive oxygen species (ROS) via complex I and inhibits NADPH oxidase, which is an important source of cytosolic ROS (Jong et al., 2012; Li et al., 2009; Miao et al., 2013). Furthermore, elevated levels of taurine are correlated with an attenuation of antioxidant enzyme depletion (Pushpakiran et al., 2004; Zhu et al., 2016; Shimada et al., 2015). In the turtle brain, ROS levels are reduced throughout anoxia (Pamenter et al., 2007; Hogg et al., 2015) and ROS production following re-oxygenation is largely suppressed (Pamenter et al., 2007; Milton et al., 2007). This is in part due to a large antioxidant and heat shock protein reserve, as well as mechanisms that prevent excessive ROS production that may otherwise occur in anoxia-intolerant organisms (Ultsch et al., 1984; Ramaglia and Buck, 2004; Granger and Kvietys, 2015; Larson et al., 2014). Therefore, while taurine may contribute to antioxidant defenses in anoxia-intolerant species, it seems unlikely that it does so in the turtle brain.

The structural resemblance of taurine, GABA and glycine suggests that these molecules function similarly, as demonstrated by the ability of taurine to mediate neurotransmission through both GABAergic and glycinergic receptors (Albrecht and Schousboe, 2005). Evidence in different brain regions demonstrates that taurine activates ionotropic synaptic and extra-synaptic GABA_A receptors and that receptor blockade via gabazine, bicuculline and/or picrotoxin partially or fully abolishes the effects of taurine (Bureau and Olsen, 1991; del Olmo et al., 2000; Wu et al., 2008; Jia et al., 2008; Nguyen et al., 2013). In addition, taurine and its synthesizing enzyme, cysteinesulfinic acid decarboxylase (CSAD), are present in many brain regions and are localized to both neurons and glial cells, implying that taurine may act as a neuromodulator (Chan-Palay et al., 1982; Magnusson et al., 1988; Reymond et al., 1996; Vitvitsky et al., 2011; Winge et al., 2015). In spite of this knowledge, it is unknown how taurine functions in the anoxia-tolerant turtle brain and if it plays an integral part in the turtle's strategies of neural inhibition. Since GABA plays a central role in spike arrest in the painted turtle and taurine can activate both GABA_A and glycine receptors in mammals, we hypothesize that taurine will: (1) suppress action potential firing, increase whole-cell conductance and depolarize membrane potential in pyramidal neurons; (2) interact with GABA and glycine receptors to mediate shunting inhibition; and (3) act independently of osmotic effects or glutamatergic/GABA_B receptors.

This study was approved by the University of Toronto Animal Care Committee and conforms to the relevant guidelines outlined by the Canadian Council on Animal Care regarding care and use of experimental animals. Adult female Chrysemys picta bellii (Schneider 1783) turtles ∼10 years old and collected between spring and autumn were obtained from Niles Biological Inc. (Sacramento, CA, USA). The turtles were housed together in large indoor tanks equipped with a flow-through dechlorinated freshwater system maintained at ∼18°C, basking platforms, and UV heat lamps. Turtles were maintained on a 12 h:12 h light:dark photocycle and were given access to food regularly.

Turtles were decapitated and whole brains were rapidly excised from the cranium. The entire dorsal cortex was dissected free and bathed in 3–5°C artificial turtle cerebrospinal fluid (aCSF) composed of (in mmol l⁻¹): 107 NaCl, 2.6 KCl, 1.2 CaCl₂, 1.0 MgCl₂, 2.0 NaH₂PO₄, 26.5 NaHCO₃, 10.0 glucose, 5.0 imidazole, pH 7.4; osmolarity 285–290 mosmol l⁻¹. Six cortical sheets were obtained from each whole brain, three from each hemisphere, and placed in vials containing ice-cold aCSF. Tissue was stored at 4°C in aCSF for up to 2 days until used for experiments.

To conduct experiments, a single cortical sheet was placed in an RC-26 chamber with a P1 platform (Warner Instruments) and held in place with a nylon thread anchor. To achieve normoxic conditions, a 500 ml aspirator-style bottle containing room temperature (∼22°C) aCSF was gassed with 95% O₂/5% CO₂ and was used to gravity perfuse the chamber at a rate of 2–3 ml min⁻¹. A second 500 ml bottle of normoxic aCSF was set up in conjunction to bulk-perfuse treatment solutions containing taurine, GABA, glycine or blank controls. In a control experiment, both bottles were filled with normoxic aCSF and flow was switched from one bottle to the other, this did not cause any changes in the electrophysiological parameters measured; therefore, changes were due to the respective treatments and not flow artifacts. A fast-step drug perfusion system (VC-6 model perfusion valve controller, Warner Instruments) was used to perfuse various pharmacological inhibitors directly over the cortical sheets. This system consisted of four 50 ml reservoirs regulated remotely by pinch valves and feeding into a manifold resulting in a single output. The output consists of a piece of electrode glass bent in a 45 deg angle to deliver saline several millimeters up-stream of the patch pipette. Flow through the perfusion system was maintained at all times by rapidly switching between control normoxic saline and saline containing a particular pharmacological modulator. The end of the bent electrode glass was positioned to just make contact with the surface of the bulk saline perfusion solution so that there was smooth flow and no ripples that may dislodge the recording electrode patch. All experiments were conducted at room temperature (∼22°C).

Whole-cell patch clamp recordings were obtained from pyramidal neurons located in the dorsal cortex using 4–8 MΩ borosilicate glass pipettes (Harvard Apparatus) containing (in mmol l⁻¹): 8.0 NaCl, nominally 0.0001 CaCl₂ (without Ca²⁺ buffering, the concentration cannot be precise), 10.0 Na-Hepes, 130 potassium gluconate, 1.0 MgCl₂, 0.3 NaGTP, and 2.0 NaATP (pH adjusted to 7.4 with methanesulfonic acid; 295–300 mosmol l⁻¹), unless otherwise specified. Glass pipettes with an Ag–AgCl electrode were connected to a CV-4 headstage and Axopatch-1D amplifier with a Digidata 1200 interface (Axon Instruments) and positioned within the tissue using a motorized patch-clamp micromanipulator. Cell-attached 4–10 GΩ seals were obtained using the blind-patch technique as described elsewhere (Blanton et al., 1989). Upon seal formation, the whole-cell patch configuration was achieved by gently applying brief suction. Typical whole-cell access resistance (R_a) was 5–30 MΩ and whole-cell leak was below 20 pA. R_a and leak were checked prior to each experiment and if R_a or leak changed by more than 20% or 30 pA, respectively, during a recording then the recording was discarded.

Pyramidal neurons were identified via their spike frequency adaptative response to an injected current (Connors and Kriegstein, 1986; Ulinski, 2007). Cells were current clamped and a current was injected for 450 ms to cause a 30 mV depolarizing change in membrane potential and action potential firing. Patches from neurons that did not adapt to the injected current were characterized as stellate interneurons and were grouped in a different data set or discarded entirely. Additionally, a rare population of pyramidal neurons displayed spontaneous action potential firing, and therefore when these neurons were patched, the effect of taurine on action potential firing was also accessed, to directly access its inhibitory properties. All data were collected at 5 kHz on the computer using Clampex 7 software (Axon instruments) and later analyzed using Clampfit.

Following whole-cell patch formation, each patch was allowed a 5 min stabilization period prior to recording. Baseline membrane potential was recorded for 5 min following the beginning of the experiment (t=0 min) during which time the tissue was bulk and drip-perfused with normoxic aCSF. Following initial recordings, cells were drip-perfused with either control aCSF or aCSF containing pharmacological agents for an additional 5 min (t=5 to t=10 min) while continuously bulk perfused with normoxic aCSF. After 10 min, bulk perfusion was switched to a treatment solution (i.e. aCSF containing taurine, GABA or glycine) for the next 5 min (t=10 to 15 min). The effects of treatments were typically seen within the first minute of switching the perfusion solution and plateaued within 5 min of treatment. Following treatment conditions, the tissue was bulk re-perfused with normoxic aCSF for 10 min of recovery (t=15 to t=25 min). All membrane potential recordings were recorded passively in current clamp mode. All membrane potential measurements were adjusted for liquid junction potential following data acquisition, which was experimentally determined to be −11 mV and all data have been corrected for this value, including raw traces.

Current–voltage (IV) relationships were obtained at 5 min intervals throughout the 25 min experiment by voltage clamping neurons in sequential steps at −80, −60, −40, −20 and 0 mV in between periods of passive membrane potential recordings. Current measurements at each clamped voltage were assessed and plotted against voltage to create an IV curve, where the slope of the curve at the membrane potential represented the whole-cell conductance (G_w) of the neuron.

Whole-cell patches were obtained as described above; however, the concentration of Cl⁻ within the pipette was increased to 130 mmol l⁻¹ by substituting K-gluconate with equimolar concentrations of CsCl. High concentrations of Cl⁻ were used in order to aid in the detection of currents from Cl⁻ ion channels since taurine was hypothesized to function through these channels and this hypothesis was supported from initial experiments. Additionally, neurons were voltage clamped at a holding potential of −100 mV throughout the experiment in order to further aid in the detection of these currents. Taurine-evoked currents were measured while taurine was perfused for approximately 10 s directly upstream of the area where the electrode penetrated the brain sheet, as 10 s was the typical time it took to reach the peak of the induced current. Normoxic control currents were measured at t=0, t=5, and t=10 min to ensure recording stability. The current recorded at t=0 was used to normalize all subsequent recordings while the measurement made at t=5 was presented as the normoxic control. After demonstrating current stability at t=10 min, the cell was drip-perfused with an inhibitor or control aCSF for 5 min, at the end of which an evoked current recording was taken (t=15 min). The tissue was re-perfused with control normoxic aCSF following experimental treatment and recovery current recordings were taken at t=20 and t=25 min.

Taurine was bulk-perfused when measuring membrane potential and G_w, while it was drip-perfused when measuring current induction. For most experiments, taurine was applied at a concentration of 10 mmol l⁻¹. While the concentration used here is not physiological, as endogenous levels are ∼10–100 μmol l⁻¹ (Lerma et al., 1986; Shibanoki et al., 1993), there are a number of factors that could warrant using a higher concentration, particularly diffusion of the drug into the cortical sheet. To compare the effects of taurine with the other major inhibitory neurotransmitters in the brain, 2 mmol l⁻¹ GABA and 2 mmol l⁻¹ glycine were also applied to the tissue by either bulk perfusion or drip perfusion, as specified for each experiment and the methodology of these experiments was identical to the taurine protocol above.

For experiments measuring GABA_A receptor currents, gabazine at relatively low concentrations was used to competitively block GABA_A synaptic and peri-synaptic receptors, bicuculline was used to competitively block both synaptic and extra-synaptic tonic GABA_A receptors, and picrotoxin was used to block all three receptors because it acts as a non-competitive GABA_A channel blocker. CGP-55845 was used to block GABA_B receptors in the presence of a high concentration (40 mmol l⁻¹) of taurine, which was used in order to aid in the detection GABA_B receptor currents.

For experiments measuring glutamatergic receptor currents, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) was used to block AMPA receptors, dl-2-amino-5-phosphonovaleric acid (APV) was used to block NMDA receptors and tetrodotoxin (TTX) was used to block Na⁺ channel-mediated action potentials and eliminate excitatory glutamatergic inputs.

Membrane potential measurements were averaged over 5 min before treatment and for another 5 min after treatment when a new steady state was reached. Currents were recorded prior to drug application and subsequently at the peak of the response. The 10–90% rise and decay times were determined using Clampfit v. 10.2 (Molecular Devices). All data passed normality and equal variance tests and were analyzed using SigmaPlot software package v. 11.0 (Systat Software Inc.). Membrane potential and G_w data were analyzed by a one-way ANOVA followed by a Tukey's post hoc test. Taurine-induced whole-cell current amplitudes and charge transfer measurements as well as rise times and decay times were analyzed using one-way repeated-measures ANOVA. Significance for all data was determined at P<0.05. All data are expressed as the means±s.e.m.

Application of taurine to an actively firing turtle cortical pyramidal neuron resulted in the reversible inhibition of action potential firing (Fig. 1A). Membrane potential in pyramidal neurons was maintained over the 20 min experimental period (t=5: −90.0±0.8 mV; t=10: −91.0±1.1 mV; t=20: −91.0±0.8 mV, Fig. 1B,C) while application of taurine (10 mmol l⁻¹) at t=5 min resulted in a significant depolarization of membrane potential from −91.4±0.5 mV to −83.1±0.7 mV at t=10 min, corresponding to a change of 8.2±0.5 mV, which reversed to 91.4±0.7 mV following aCSF washout at t=20 min (P≤0.001, Fig. 1B–D). This effect was concentration dependent, where the degree of membrane depolarization began to plateau at concentrations above 20 mmol l⁻¹ (Fig. 1E) and had long-lasting effects at 40 mmol l⁻¹ (Fig. 1F). The shift in membrane potential is similar to the 8 mV depolarization that occurs following an anoxic transition, which is attributed to shifting membrane potential towards the reversal potential of GABA (Pamenter et al., 2011), indicating that taurine mimics anoxia-mediated membrane depolarization.

Since inhibitory GABAergic neurotransmission underlies anoxia-tolerance in the turtle brain, the impact of taurine on membrane potential was compared with that of the major inhibitory neurotransmitters GABA and glycine. Similarly to taurine, GABA and glycine both resulted in membrane depolarization: application of 2 mmol l⁻¹ GABA at t=5 min significantly depolarized the membrane potential from −90.8±0.9 mV to −83.6±1.1 mV at t=10 min, a change of 7.3±0.9 mV, and this change reversed to 90.3±1.1 mV following aCSF washout at t=20 min (P≤0.001), while application of 2 mmol l⁻¹ glycine at t=5 min resulted in a significant depolarization from −89.4±0.9 mV to −84.6±0.8 mV at t=10 min, a change of 4.8±0.4 mV, and also recovered to 89.5±1.1 mV following aCSF washout at t=20 min (P≤0.005, Fig. 1B–D). Together, these data indicate that taurine has an inhibitory and depolarizing effect on the membrane potential of pyramidal neurons that is similar to other inhibitory molecules within the turtle cortex.

To determine if taurine affected ion movement across the membrane, whole-cell conductance (G_w) was measured. Over the course of the experimental period, whole-cell conductance was maintained (t=5: 6.3±1.2 pS; t=10: 7.0±1.4 pS; t=20, 6.4±1.1 pS, Fig. 2A). Consistent with the observed impact of taurine on membrane potential, application of taurine at t=5 min significantly increased neuronal G_w from 5.0±0.5 pS to 10.9±1.3 pS at t=10 min, an increase of 219.8±16.3% from baseline, and this was reversible following 10 min of aCSF washout (4.9±0.5 pS, P≤0.001, Fig. 2A–C). This increase in whole-cell conductance was concentration dependent, where G_w increased with taurine levels until it plateaued at concentrations above 20 mmol l⁻¹ (Fig. 2D).

Similar to the effects of taurine, application of 2 mmol l⁻¹ GABA significantly increased G_w from 4.7±0.4 pS to 12.3±2.0 pS, an increase of 253.9±29.0% from baseline, and this was reversed to 4.6±0.5 pS following aCSF washout (P≤0.001), while application of 2 mmol l⁻¹ glycine did not significantly change G_w (from 5.0±1.0 pS to 6.7±1.5 pS) (Fig. 2A–C). Similarly to the change in membrane potential, GABA had the largest effect on G_w, although it was not significantly different from the change in G_w observed following taurine application. However, glycine had a significantly smaller effect at the given concentration than either GABA or taurine (P≤0.005 and P≤0.04, respectively).

Given the suspected function of taurine in ion channel activation (particularly Cl⁻), its ability to induce whole-cell currents was directly assessed using a high [Cl⁻] pipette solution (130 mmol l⁻¹). Application of taurine induced inward whole-cell currents (Fig. 3A) and successive application of taurine in 5 min intervals did not significantly affect the amplitude of these currents (94.9±4.1% to 94.8±6.6%, Fig. 3B,C) or their charge transfer (82.1±12.1% to 87.2±9.0%, Fig. 3B,C), indicating that taurine application was not desensitizing and was repeatable. The size of this current did, however, vary depending on the concentration of taurine used. Higher concentrations of taurine increased both the amplitude and charge transfer of the current (Fig. 3D–F). Unlike the effect on membrane potential and G_w, concentrations above 20 mmol l⁻¹ did not show a plateauing effect. This difference may be due to the high concentration of Cl⁻ included in the pipette solution to enhance the current. Additionally, fast inward currents were observed in subsequent current recordings and were likely AMPA currents, as we have shown in the past (Pamenter et al., 2008a).

Application of GABA and glycine similarly induced an inward current consistent with the activation of Cl⁻ channels and the efflux of Cl⁻ to depolarize the membrane potential (Fig. 3G). As expected, GABA application induced the largest current, as both amplitude and charge transfer were larger than that induced by taurine or glycine. With respect to GABA-induced currents, glycine application induced a current with an amplitude (74.3±13.9%) that was significantly larger than that evoked by taurine (31.4±4.1%, P≤0.04), but not significantly different than the GABA-evoked current (Fig. 3H). Yet, despite the differences in amplitude, the total amount of current carried over time following treatment with respect to the GABA current was not significantly different between taurine (76.8±5.7%), glycine (85.4±26.0%) and GABA, although GABA and glycine appeared slightly higher than taurine (Fig. 3I). This large difference in amplitude but small difference in charge transfer between taurine, GABA and glycine was likely due to the slower activation and recovery from taurine application (Fig. 3G). Together, these results indicate that taurine increases ion flux and induces a similar current to GABA and glycine.

Although taurine, GABA and glycine all affected pyramidal neurons similarly, the mechanism of action for taurine appeared slightly different with respect to membrane potential and G_w. Most notably, the latency between taurine current induction and recovery from taurine application was slower (Fig. 1B and Fig. 3G). To assess this directly and to determine if there was a difference in the kinetics between taurine, GABA and glycine, the rise time and decay time of whole-cell currents induced by each molecule were analyzed (Fig. 3J–K). GABA and glycine displayed similar rise and decay times that were not significantly different from each other. The average rise time for GABA and glycine was 3.0±0.6 s and 3.8±0.9 s, respectively (Fig. 3J). Decay times were slower at 8.6±1.4 s and 7.9±0.8 s for GABA and glycine, respectively (Fig. 3K). However, taurine exhibited a significantly longer rise time (7.8±1.3 s, P≤0.04) compared with GABA but not glycine, and a significantly longer decay time (31.2±5.2 s, P≤0.04) with respect to the values obtained for GABA or glycine. Overall, these data suggest that taurine acts slower than GABA or glycine, and although taurine may affect the neuron similarly to GABA and glycine, its functionality (i.e. binding kinetics) is different.

In addition to functioning as a neuromodulator, taurine has a prominent role in cell volume regulation as an osmolyte. Given the role of taurine in controlling cell osmolality (Schaffer et al., 2010), taurine-induced changes in osmolality were investigated to determine if it influenced membrane potential or whole-cell conductance. The addition of taurine to the aCSF (10 mmol l⁻¹) solution only increased the osmolality from 287.0±1.8 mmol kg⁻¹ to 293.0±2.9 mmol kg⁻¹ (Fig. 4A). To test if small changes in osmolality affect the membrane potential and G_w of the neuron, sucrose was added to aCSF to increase the osmolality by ∼10 mmol kg⁻¹. An increase in osmolality did not significantly change membrane potential (−90.7±1.4 mV to −91.4±1.3 mV) or G_w (9.2±0.8 pS to 9.5±0.5 pS, Fig. 4B–D). These data suggest that a change in osmolality is not responsible for the depolarization or the increase in G_w caused by taurine.

To investigate the potential contribution of glycine receptors to the taurine response, the glycine antagonist strychnine (25 μmol l⁻¹) was used. Application of strychnine alone did not significantly affect either the membrane potential (−89.4±1.0 mV to −89.4±1.1 mV, Δ=0±0.5 mV) or G_w (4.84±0.62 pS to 3.75±0.49 pS, a non-significant reduction to 78.4±5.5% of baseline, Fig. 5A,B) of the neuron. When taurine (10 mmol l⁻¹) was applied in the presence of strychnine, it depolarized the membrane potential from −89.7±0.8 mV to −86.6±0.8 mV, for a change of 3.1±0.6 mV, and recovered to −90.0±0.7 mV following 10 min of aCSF washout (P≤0.03, Fig. 5A). Furthermore, G_w increased from 4.8±0.5 pS to 6.7±0.7 pS, which is 141.3±11.6% of baseline values, and recovered to 5.5±0.7 pS following 10 min of aCSF washout (P≤0.03, Fig. 5B). However, both of these variables were significantly reduced from values obtained via taurine application alone, suggesting that glycine receptor activation is a component of the taurine-induced change in membrane potential and G_w (P≤0.001 and P≤0.03, respectively).

To further assess the role of glycine receptors, the ability of strychnine to block whole-cell taurine-induced currents was analyzed. The amplitude of the taurine-evoked whole-cell current significantly decreased to 29.2±11.5% of baseline values and this recovered to 86.9±19.3% of baseline values following aCSF washout (P≤0.05), while charge transfer values significantly decreased to 49.9±16.2% of baseline values and this recovered to 130.1±40.9% of baseline values following aCSF washout, Fig. 5C,E,F). To further demonstrate that strychnine effectively blocks glycine receptors, 2 mmol l⁻¹ glycine was applied in the presence and absence of strychnine, which completely blocked glycine-mediated currents (Fig. 5D). Together, these data indicate that glycine receptor activation is a component of the taurine-evoked current in pyramidal neurons.

To determine if GABA_A receptor currents are a component of taurine-evoked conductance, the GABA_A receptor blockers gabazine (25 μmol l⁻¹), bicuculline (10 μmol l⁻¹), and picrotoxin (100 μmol l⁻¹) were used. Neither the change in membrane potential (ΔV_m) nor the change in whole-cell conductance (ΔG_w) were significantly changed following application of gabazine (ΔV_m=1.0±0.3 mV, ΔG_w=86.6±6.0% of baseline, Fig. 6A,B), bicuculline (ΔV_m=0.0±0.8 mV, ΔG_w=99.9±23.4% of baseline, Fig. 6A,B) or picrotoxin (ΔV_m=0.0±0.6 mV, G_w=87.1±9.8% of baseline, Fig. 6A,B). When taurine (10 mmol l⁻¹) was applied in the presence of gabazine, the membrane potential depolarized from −90.9±0.7 mV to −88.1±0.6 mV (ΔV_m=2.8±0.3 mV, P≤0.03) while G_w increased from 4.27±0.50 pS to 5.38±0.77 pS (ΔG_w=126.6±11.1% of baseline, P≤0.03, Fig. 6A–C). Taurine, in the presence of bicuculline, similarly depolarized the membrane potential from −91.4±1.3 mV to −89.3±2.0 mV (ΔV_m=2.1±0.8 mV, P≤0.03) and increased ΔG_w from 4.36±0.76 pS to 5.06±pS (ΔG_w=120.1±8.2% of baseline, P≤0.03, Fig. 6A–C). Results observed with bicuculline treatment did not significantly differ from those seen with gabazine treatment, suggesting that taurine primarily affects synaptic GABA_A receptors. When treated with picrotoxin, the membrane potential depolarized from 89.2±0.4 mV to −86.0±1.0 mV (ΔV_m=3.0±1.01 mV, P≤0.03) while G_w increased from 4.61±0.68 pS to 5.35±0.65 pS (ΔG_w=119.3±10.2% of baseline, P≤0.03, Fig. 6A–C) following taurine treatment. Overall, similarly to strychnine, taurine decreased the membrane potential, and G_w following GABA_A receptor blockade also decreased by ∼50%, indicating that a substantial component of the taurine response is due to the activation of GABA_A receptors.

The role of GABA_A receptors was further assessed by testing the effect of each GABA_A receptor inhibitor on taurine-induced whole-cell currents. Following gabazine treatment, the current amplitude significantly decreased from −348±134 pA to −201±84 pA, a reduction to 53.9±7.1% of the baseline value (P≤0.01, Fig. 6C,D) and the current charge transfer decreased from 4142.8±1494.0 pC to 3233.5±1059.1 pC, which is 85.3±9.4% of the baseline value (Fig. 6C,E). Picrotoxin also resulted in a significant decrease in the current amplitude from 270.0±98.9 pA to 80.8±23.2 pA, a reduction to 37.5±11.8% of the baseline value (Fig. 6C,D, P≤0.04), and a significant decrease in the charge transfer from 3969.4±1429.4 pC to 1604.9±303.7 pC, which is 49.8±12.2% of the baseline value (P≤0.03). Interestingly, following bicuculline treatment, the current amplitude (448.6±217.8 pA to 423.7±197.6 pA) and charge transfer (7034.7±2734.1 pC to 6645.82±2668.7 pC) remained unchanged (92.8±18% and 97.4±10.6% of the original current, respectively, Fig. 6D,E). Collectively, these results suggest that GABA_A receptors constitute a component of the taurine-induced current in pyramidal neurons.

To determine if receptors other than glycine and GABA_A mediate the effects of taurine, the contribution of GABA_B, NMDA and AMPA receptors were also investigated. Application of CGP-55845, a GABA_B receptor inhibitor, alone did not significantly affect the membrane potential (ΔV_m=1.0±1.0 mV) or G_w (ΔG_w=90.7±7.0% of baseline) of the neuron (Fig. 7A,B). When 40 mmol l⁻¹ taurine was applied following CGP-55845 treatment, there was a significant decrease in the degree of membrane depolarization compared with the 40 mmol l⁻¹ taurine treatment alone (ΔV_m=5.8±0.9 mV versus ΔV_m=10.1±0.9 mV, Fig. 7A,C; P≤0.01, respectively). Compared with 10 mmol l⁻¹ taurine treatment, the effect of 40 mmol l⁻¹ taurine in the presence of CGP-55845 was not significant. In the presence of CGP-55845, taurine significantly increased G_w from 6.82±0.81 pS to 16.58±3.44 pS, a percentage change of 247.5±50.4% from control values (P≤0.04, Fig. 7B) and this increase in G_w was not significantly different between 40 mmol l⁻¹ and 10 mmol l⁻¹ taurine treatments. CGP-55845 also had no effect on the amplitude or charge transfer of the taurine-induced currents in the presence of CGP-55845 (109.4±9.5% and 126.1±17.0% of the baseline values, respectively; Fig. 7D,E). These data suggest that GABA_B receptors do not play a significant role in mediating taurine-mediated conductance.

To assess the role that glutamate receptors have on taurine-induced currents, AMPA and NMDA receptor blockers were utilized. Application of taurine (10 mmol l⁻¹) in the presence of CNQX, APV and TTX still significantly depolarized the membrane potential from −80.0±1.3 mV to −74.0±1.3 mV (ΔV_m=6.0±1.1 mV, P≤0.003, Fig. 8A,C) and significantly increased G_w from 5.5±1.0 pS to 9.2±1.6 pS (ΔG_w=237.3±50.7% of baseline, P≤0.04, Fig. 8B). Similarly, the size of the current amplitude and charge transfer did not differ significantly following drug treatment (97.1±4.5% and 113.1±12.2% of the baseline values, respectively, Fig. 8D,E). From these data, we conclude that taurine does not affect glutamatergic neurotransmission through NMDA or AMPA receptors.

We investigated the potential protective role of taurine in an anoxia-tolerant western painted turtle. Prior to this, the majority of research on taurine focused on identifying the roles this molecule has from osmoregulation and early development to neurotransmission within mammalian models (Moran et al., 1994; Pasantes-Morales and Schousboe, 1997; Sturman et al., 1985; Shivaraj et al., 2012; Albrecht and Schousboe, 2005). Within the realm of ischemia and neuroprotection, taurine suppresses glutamate-induced toxicity, regulates intracellular Ca²⁺ levels, prevents excessive ROS formation, and prevents cell death, which contribute to its ability to reduce cellular damage during ischemia and oxygen reperfusion in anoxia-intolerant species (Leon et al., 2009; Chen et al., 2001; Jong et al., 2012; Takatani et al., 2004). However, the role of taurine in anoxia-tolerant models has not been investigated before. Given the important role that neuronal inhibition plays in the anoxia-tolerant western painted turtle (Buck et al., 2012; Pamenter et al., 2011), we focused on the potential inhibitory role of taurine on pyramidal neurons in the turtle cortical brain sheet model. Taurine application on turtle pyramidal neurons inhibited action potential firing and led to an approximate 8 mV depolarization of membrane potential, a doubling of whole-cell conductance and induction of an outward Cl⁻ current, all of which also occur following the onset of anoxia in response to increased GABAergic neurotransmission (Pamenter et al., 2011). Furthermore, the rise and decay times of taurine-induced currents were slower than those elicited by GABA or glycine, suggesting that the prolonged inhibitory effects of taurine are unique and may help explain why taurine is needed in addition to GABA during anoxia. We also found that inhibition of glycine or GABA_A receptors decreased taurine-induced currents and its effects on membrane potential and G_w, but neither receptor was solely responsible for any taurine-mediated responses.

The former model of the anoxic turtle cortex posits that GABA levels increase with anoxia and mediate inhibition of pyramidal neurons through a shunting inhibition mechanism (Pamenter et al., 2011). Extracellular taurine levels also increase during anoxia and accumulate in synaptic regions, although not to the same extent as GABA (Nilsson and Lutz, 1991). We can now add that as GABA binds to and activates GABA_A and GABA_B receptors, taurine, either directly or indirectly, also leads to the activation of glycine and GABA_A receptors and contributes to anoxia-mediated inhibition. Based on this model, it would seem that taurine and GABA have redundant functions and this therefore raises questions regarding the role of taurine-mediated neurotransmission. Taurine has slower rise and decay times in turtle cortical neurons than GABA, even at higher concentrations, which suggests that taurine may function differently. For instance, while GABA may mediate fast inhibitory synaptic neurotransmission and the majority of inhibition in the anoxic turtle brain, taurine may be responsible for a longer-lasting, tonic form of inhibition. A longer time frame may indeed be required since taurine had no effect on neural activity at the 2 mmol l⁻¹ concentrations that were effective for GABA and glycine to elicit a response.

An important difference to consider between GABA and taurine that may impact their function during anoxia is how each molecule is synthesized. While GABA is synthesized from glutamate via glutamate decarboxylase (GAD) and does not require oxygen, taurine is synthesized from cysteine in a series of steps that do require oxygen (Martin and Rimvall, 1993; Tappaz, 2004; Tappaz et al., 1994; Vitvitsky et al., 2011). In the predominant taurine synthesis pathway, cysteine is first oxidized to cysteinesulfinate and then converted into hypotaurine via CSAD before undergoing its final oxidation from hypotaurine to taurine. This reliance on oxygen in taurine synthesis ultimately means that taurine cannot be synthesized during anoxia and the extent of its release is limited by what is already present within the cell. Thus, the function of taurine in the anoxic brain may be largely restricted by its existing intracellular pool and ability to be recycled within neurons. However, this does not negate its potential role in anoxia-mediated neural inhibition but rather suggests that taurine may be particularly important in the early stages of anoxia. In this way, taurine may contribute an extra level of complexity and inhibitory support during the critical transition to anoxia.

Despite the evidence that taurine is involved in inhibitory transmission in the turtle cortex and has a potential role in anoxia, there are limitations to the current study. The most obvious limitation is whether these findings translate to physiological concentrations of taurine. Although taurine is one of the most abundant amino acids in the brain, with intracellular concentrations in the millimolar range, extracellular levels typically are 400-times lower, in the 10–100 μmol l⁻¹ range (Pasantes-Morales and Schousboe, 1997) This is below the concentrations used in this study – even after neural stimulation when taurine levels increase (Lerma et al., 1986; Shibanoki et al., 1993). Although not shown, GABA and/or glycine gave optimal responses at a 5-times lower concentration of 2 mmol l⁻¹, the reason for this difference is not clear. However, the large inwardly directed uptake gradient for taurine may be partly responsible for the higher concentrations required to generate a current. Finally, synaptic taurine concentrations would be expected to be higher than at the whole brain level, where taurine levels were measured (Nilsson and Lutz, 1991). But even with this reasoning, there is still a discrepancy between the physiological and experimental concentrations. As a result, it remains to be determined whether physiological concentrations of taurine are capable of activating glycine and GABA_A receptors.

Another limitation of the current study is that although we know that taurine has the potential to contribute to neural inhibition during anoxia, whether it actually does play a substantial role is unknown. While taurine levels do increase following anoxia, it is unclear if this increase is enough to contribute to neural inhibition. Typically, to determine the role of something during anoxia, its actions are inhibited. Unfortunately, inhibiting taurine during anoxia without affecting other processes is difficult; blocking the targets of taurine, the glycine and GABA_A receptors, also blocks the ability of GABA and glycine to function. Alternatively, taurine production via its major enzyme CSAD or cysteine dioxygenase (CDO) could be inhibited, but disturbing this could affect multiple biological processes other than neurotransmission in which taurine has been implicated. As taurine production also relies on oxygen, production would already be abolished during anoxia and inhibiting taurine production via CSAD or CDO would not affect pre-existing taurine within cells. Genetic knockouts of CSAD and CDO exist in mice and display significantly decreased taurine levels, but genetic manipulation of turtles is limited (Park et al., 2014; Ueki et al., 2011). To date, the only well-established way to affect taurine is to inhibit its uptake transporter (TauT) (Tappaz, 2004; Tappaz et al., 1994). Inhibition of TauT with the taurine analog guanidinoethyl sulfonate (GES) can inhibit taurine reuptake as well as deplete taurine levels in tissues by 50–80% through a mechanism involving taurine transport inhibition and synthesis interference (Huxtable, 1992; Suárez et al., 2016). However, using GES in systems that also contain GABA_A receptors is difficult and must be used cautiously because GES is also a GABA_A receptor agonist (Mellor et al., 2000). Furthermore, even if taurine could be manipulated during anoxia, the large contribution of GABA can still make it difficult to specifically isolate taurine-mediated effects. Therefore, in addition to manipulating endogenous taurine during anoxia, it is critical to be able to remove the effects of GABA without impacting the taurine response. Removing GABA also has its challenges, because of its importance within the brain, and for these reasons it can be difficult to directly assess the role of taurine in anoxia.

There are also a number of questions regarding the role of taurine in the anoxic turtle brain that have yet to be addressed. One important question is whether taurine directly or indirectly activates glycine and GABA_A receptors. Based on our findings, we cannot eliminate the possibility that taurine leads to changes in the anoxic turtle brain that then lead to receptor activation rather than taurine binding directly to each receptor. Indirect changes may include increased GABA and/or glycine release, increased receptor sensitivity to the ligand, and strengthening of the inhibitory response. Interestingly, taurine uptake via TauT is substantially decreased in the presence of GABA as well as GABA uptake in the presence of taurine, thus promoting extracellular increases in each molecule (Sivakami et al., 1992). Additionally, taurine enhances the effects of both glycine and GABA application on neurons. In neurons of the rat anteroventral cochlear nucleus, simultaneous application of taurine and glycine or taurine and GABA produces a larger current than glycine or GABA application alone (Song et al., 2012). Even longer-term application of taurine affects the GABAergic system: mice fed taurine in their drinking water for 4 weeks display elevated levels of GABAergic interneurons, [GABA] and the GABA-synthesizing enzyme GAD (Levinskaya et al., 2006; El Idrissi and Trenkner, 2004). All of these changes to the GABAergic system suggest that taurine may have an indirect effect on turtle cortical neurons, leading to an enhanced inhibitory response.

In addition to its role in neurotransmission, taurine is also involved in a number of other biological processes in the body, all of which may be important in anoxia tolerance. Therefore, it is prudent to ask what other functions taurine may have within the anoxia-tolerant turtle. We narrowed our focus to just one possible function of taurine because of the importance of neural inhibition in anoxia tolerance. Within the brain, taurine may also guard neurons against a rise in apoptotic factors (Ripps and Shen, 2012). In the heart, taurine is particularly important in maintaining normal contractile function as taurine depletion leads to cardiomyopathy. Here, it may play an important role in Ca²⁺ homeostasis and even serve as a modulator of protein kinases and phosphatases (Schaffer et al., 2010). At a systemic level, taurine produced from the liver may enhance glucose transport into cells and also serve as an antioxidant (De la Puerta et al., 2010; Shimada et al., 2015).

In summary, we can conclude that taurine is capable of acting as an inhibitory molecule within the turtle brain and can do so by activating glycine and GABA_A receptors. The inhibitory effect of taurine is thought to complement the large increase in GABAergic neurotransmission with the onset of anoxia in painted turtle brain and to contribute to anoxia tolerance.

We thank Dr D. Hogg for the original suggestion to investigate the role of taurine in the anoxia-tolerant turtle brain.