Signaling mechanism for modulation by ATP of glycine receptors on rat retinal ganglion cells

ATP modulates voltage- and ligand-gated channels in the CNS via the activation of ionotropic P2X and metabotropic P2Y receptors. While P2Y receptors are expressed in retinal neurons, the function of these receptors in the retina is largely unknown. Using whole-cell patch-clamp techniques in rat retinal slice preparations, we demonstrated that ATP suppressed glycine receptor-mediated currents of OFF type ganglion cells (OFF-GCs) dose-dependently, and the effect was in part mediated by P2Y1 and P2Y11, but not by P2X. The ATP effect was abolished by intracellular dialysis of a Gq/11 protein inhibitor and phosphatidylinositol (PI)-phospholipase C (PLC) inhibitor, but not phosphatidylcholine (PC)-PLC inhibitor. The ATP effect was accompanied by an increase in [Ca2+]i through the IP3-sensitive pathway and was blocked by intracellular Ca2+-free solution. Furthermore, the ATP effect was eliminated in the presence of PKC inhibitors. Neither PKA nor PKG system was involved. These results suggest that the ATP-induced suppression may be mediated by a distinct Gq/11/PI-PLC/IP3/Ca2+/PKC signaling pathway, following the activation of P2Y1,11 and other P2Y subtypes. Consistently, ATP suppressed glycine receptor-mediated light-evoked inhibitory postsynaptic currents of OFF-GCs. These results suggest that ATP may modify the ON-to-OFF crossover inhibition, thus changing action potential patterns of OFF-GCs.

Scientific RepoRts | 6:28938 | DOI: 10.1038/srep28938 modulation of glycine-receptor mediated responses of OFF-GCs. By using whole-cell patch-clamp techniques in rat retinal slice preparations, we characterized how ATP modulated glycine currents of OFF-GCs, by activating P2Y receptors and explored the intracellular signaling pathway mediating such a modulation. Our results clearly show that a distinct G q/11 /phosphatidylinostiol (PI)-phospholipase C (PLC)/inositol-1,4,5-trisphosphate (IP 3 )/ Ca 2+ /protein kinase C (PKC) signaling pathway is responsible for the ATP effect. Consistent with this, we also found that ATP suppressed light-evoked glycine receptor-mediated inhibitory postsynaptic currents (L-IPSCs) of OFF-GCs via P2Y receptors.

ATP suppresses glycine currents of OFF-GCs. We first characterized glycine-induced currents in rat
GCs. Glycine receptor-mediated currents were pharmacologically isolated by adding D-AP5, CNQX, bicuculline and TTX to bath Ringer's (see Methods for details). Figure 1A shows that the current of a GC clamped at − 60 mV, which was induced by local puff of 100 μ M glycine to the dendrites of the cell in Ringer's containing the above antagonists. The current was almost completely abolished by 1 μ M strychnine, a specific antagonist of glycine receptors 42 (7.36 ± 1.62% of control, n = 5, P < 0.001; Fig. 1B). The current response returned to the control level after washout with Ringer's (83.0 ± 4.42% of control, P > 0.05). The current-voltage relationship of the currents was linear, with a reversal potential of 0.15 ± 1.4 mV (n = 5; Fig. 1C), which is very close to E Cl -(0 mV), calculated according to the Nernst equation.
Application of 100 μ M ATP elicited no detectable current in OFF-GCs (data not shown). When 100 μ M ATP was bath-applied, as shown in Fig. 2A, the current induced by 100 μ M glycine was suppressed in a progressive manner during the first 6 min after ATP application, and the current became stable in about 8 min and was kept at a similar level thereafter. ATP-induced suppression of glycine currents was observed in most of the OFF-GCs tested (19 out of 23, 82.61%). The average current amplitudes, following 14 min perfusion of 100 μ M ATP, were reduced to 67.3 ± 4.05% of control (n = 19, P < 0.001; Fig. 2B,C). In the remaining four cells, ATP had no effects on the glycine currents (4/23, 17.39%).
We further examined the concentration dependence of the ATP effect. For these experiments, data were pooled only from the cells, in which peak amplitudes to 100 μ M glycine applied at intervals of 2 min were altered less than 5% during a period of 8 min prior to the experiment. For each cell, only a single concentration of ATP was tested, and the response amplitude obtained after 14-min incubation of ATP was normalized to the level recorded prior to the incubation (control). Following the 8-min washout, the response commonly returned to the control level (Fig. 2C). The data were discarded in case the response amplitudes (after 8-min washout) were changed by more than 5% of control. Figure 2D shows how the ATP effect depended upon ATP concentration. No suppression was seen with 10 μ M ATP (104.2 ± 4.67%, n = 4, P > 0.05), but the glycine currents were respectively reduced to 86.3 ± 1.18% (n = 5, P < 0.01), 67.3 ± 4.05% (n = 19, P < 0.001) and 38.3 ± 3.53% (n = 5, P < 0.001) of control following ATP incubation at concentrations of 50 μ M, 100 μ M and 300 μ M. Based on these data, ATP of 100 μ M was chosen for all experiments to be subsequently described.
Since extracellular ATP can be converted into adenosine by ectonucleotidases 14 , we tested whether ATP or adenosine induced the suppression. ARL67156 (100 μ M), an ectonucleotidase antagonist, which blocks the degradation of ATP 43 , did not change the glycine currents of OFF-GCs (105.1 ± 2.33% of control, n = 7, P > 0.05).
Given P2Y receptors being G-protein-coupled 44,45 , the ATP effect should be eliminated when G-protein activity is inhibited. This was experimentally demonstrated. ATP did not suppress glycine currents recorded from OFF-GCs which were intracellularly dialyzed with the G-protein inhibitor GDP-β -S (3 mM) for more than 8 min (98.4 ± 6.25% of control, n = 5, P > 0.05; Fig. 4A,B). Recent evidence suggests that P2Y 1,2,4,6,11 and P2Y 12-14 are mainly coupled to G q/11 and G i/o proteins, respectively 51,52 . We further examined which subtype(s) of G-proteins may mediate the ATP effect. Internal dialysis with 10 μ M mastoparan, a peptide activator of G i and G o 53 , for 8 min appeared to slow down the decay phases of the glycine currents of OFF-GCs, but did not change glycine current amplitudes (100.7 ± 3.69% of control, n = 5, P > 0.05), and addition of ATP persisted to suppress the currents to 63.7 ± 7.66% of control (P < 0.001; Fig. 4C). In contrast, during internal infusion of 30 μ M GPAnt-2a, a specific G q/11 protein inhibitor 51 , application of ATP no longer suppressed the glycine currents of OFF-GCs (95.3 ± 2.48% of control, n = 9, P > 0.05; Fig. 4D).

PI-PLC, but not PC-PLC, signaling pathway mediates ATP-induced suppression of glycine currents.
Activation of P2Y receptors could regulate several second messengers, and the PLC-PKC signaling pathway is a major downstream effector following P2Y receptor activation 52,[54][55][56] . To test whether the PLC pathway may be involved, we investigated how U73122 (PI-PLC inhibitor) or D609 (PC-PLC inhibitor) 57 changed the ATP effect. Figure 5A shows that internal infusion of 10 μ M U73122 for 8 min did not change glycine currents of OFF-GCs (104.4 ± 4.40%, n = 7, P > 0.05), then addition of ATP for 8 min no longer reduced the currents (107.6 ± 8.18%, P > 0.05). In contrast, during internal infusion of 30 μ M D609, application of ATP reduced the currents to 65.5 ± 1.42% (n = 5, P < 0.001; Fig. 5B) of those obtained before the ATP application. These results suggest the involvement of the PI-PLC pathway, but not the PC-PLC one. Intracellular Ca 2+ is involved in ATP-induced suppression of glycine currents. As Ca 2+ is considered to be a mediator between PI-PLC and PKC 58 , whether the ATP effect is dependent on [Ca 2+ ] i was further examined. We first monitored ATP-induced changes in [Ca 2+ ] i in isolated GCs via calcium imaging. Because it is impossible to make a distinction between ON-and OFF-GC when they were isolated, the cells on which calcium imaging was performed should contain both the types. As shown in a representative result (Fig. 6A), application of ATP induced a significant increase in [Ca 2+ ] i of the GC in a reversible manner, represented as the ratio of fura-2 (340/380). The CCD images (Fig. 6a-c) show the changes in [Ca 2+ ] i induced by ATP in the soma of a GC. In eight GCs tested the averaged peak ratio value of fura-2 (340/380) obtained with the perfusion of ATP was 2.34 ± 0.23, which was significantly higher compared with the value obtained in Ringer's (1.25 ± 0.07, P < 0.001; Fig. 6B). Consistently, internal infusion of Ca 2+ -free solution containing 10 mM BAPTA, a calcium chelator 59 , the application of ATP for 8 min no longer suppressed the glycine currents (95.9 ± 2.49% of control, n = 5, P > 0.05; Fig. 6C). ATP-induced increase in [Ca 2+ ] i may be induced by an increase in extracellular Ca 2+ influx across the plasma membrane via Ca 2+ channels and/or an increase in Ca 2+ release from intracellular calcium stores. When OFF-GCs were bathed in Ca 2+ -free extracellular solution containing 1 mM EGTA, a calcium chelator 60 , ATP still suppressed the glycine currents to 62.1 ± 1.56% of control (n = 6, P < 0.001; Fig. 6D), suggesting that the ATP effect was independent of changes in extracellular calcium levels ([Ca 2+ ] o ).
Ca 2+ release from intracellular calcium stores could be mediated by ryanodine-and/or IP 3 -sensitive pathways. With intracellular dialysis of 50 μ M ryanodine, which depletes ryanodine-sensitive calcium sites 61 , ATP still significantly suppressed the currents (64.1 ± 3.64% of control, n = 6, P < 0.001; Fig. 6E). In contrast, during internal infusion of 20 μ M xestospongin-C (Xe-C), an IP 3 receptor antagonist, addition of ATP no longer suppressed the glycine currents of OFF-GCs (97.9 ± 5.81% of control, n = 5, P > 0.05; Fig. 6F). Similar results were obtained with internal infusion of heparin (5 mg/ml), another IP 3 receptor antagonist (data not shown).  Fig. 7). Internal infusion of 2 μ M Gö6976, an inhibitor of conventional Ca 2+ -dependent PKCα and β 1 isozymes, yielded a similar result. That is, in the presence of Gö6976 the mean current amplitude during ATP perfusion was hardly changed (102.7 ± 2.61% of control, n = 6, P > 0.05; Fig. 7). Moreover, perfusion of 1 μ M PMA, a PKC activator, suppressed the glycine currents in a progressive manner, with the amplitudes at 8 min being 62.8 ± 4.68% of control (n = 8, P < 0.001; Fig. 7), thus mimicking the ATP effect. During the perfusion of PMA, ATP did not cause a further suppression of the currents (103.9 ± 2.59% of the currents obtained before ATP application, P > 0.05 vs. ATP).

Role of PKC activity. Changes in [Ca
No involvement of cAMP-PKA and cGMP-PKG signaling pathways in the action of ATP. Finally, we tested whether cAMP-PKA and cGMP-PKG signaling pathways may be involved in the ATP effect on OFF-GCs. During intracellular application of 3 mM cAMP or 4 mM cGMP, ATP still suppressed the glycine currents in all OFF-GCs tested (65.7 ± 2.38% of control, n = 5, P < 0.001 for cAMP; 64.3 ± 9.21% of control, n = 5, P < 0.001 for cGMP). Furthermore, internal infusion of Rp-cAMP (50 μ M), a PKA inhibitor or KT5823 (30 μ M), a PKG inhibitor, did not change the ATP effect on glycine currents respectively (62.1 ± 5.09% for Rp-cAMP, n = 6, P < 0.001; 65.1 ± 3.42% for KT5823, n = 5, P < 0.001). These results suggest the involvement of neither cAMP-PKA nor cGMP-PKG pathway.
ATP suppresses glycine receptor-mediated L-IPSCs of OFF-GCs through P2Y receptors. To further explore physiological implication of the ATP-induced suppression of glycine currents in OFF-GCs, we examined the effects of ATP on glycine receptor-mediated L-IPSCs in retinal slice preparations. Whole cell light responses were recorded from 48 OFF-GCs. In 22 of these cells light-induced responses were completely blocked by co-application of bicuculline and TTX (see Methods for details) and no glycine receptor-mediated L-IPSCs could be recorded. In the remaining 26 GCs, glycine receptor-mediated L-IPSCs were recorded, which were abolished by co-application of 1 μ M strychnine. As shown in Fig. 8A, perfusion of 100 μ M ATP significantly suppressed the glycine receptor-mediated L-IPSC and the response returned to the control level after washout. Similar results obtained in 8 out 9 OFF-GCs tested, and the average peak amplitudes of the L-IPSCs following  ATP application were reduced to 61.8 ± 2.19% of control (P < 0.001). For the remaining one cell, ATP had no effect on the currents.

Discussion
ATP has been found to regulate both voltage-and ligand-gated channels in central neurons [4][5][6][7][8][9][10] . As far as ligand-gated channels are concerned, NMDA receptors are the only ones which are reported to be modulated by ATP. In the layer V pyramidal neurons of rat prefrontal cortex, for instance, ATP enhances NMDA responses via P2Y 2 activation 6 .
In the retina it has been previously reported that ATP released from Müller cells and retinal neurons modulates the activity of GCs. Newman (2003Newman ( , 2004 shows that ATP released from rat Müller cells could mediate interaction between these cells and GCs, suggesting that Müller cells contribute to information processing in the inner retina. Such neuromodulatory actions of ATP on GCs, however, are not due to the activation of P2 receptors, but may be resulted when neuronal adenosine receptors (P1 receptors) are activated by adenosine, which is hydrolyzed from ATP. In the mouse retina Kaneda et al. 41 show that ATP differentially modulates ON-GCs and OFF-GCs, but these authors did not identify the P2 receptor subtypes mediating this modulation of ATP. This work demonstrated, for the first time, that ATP suppresses glycine currents via P2Y receptors. This effect of ATP is neither mediated by P2X receptors (Fig. 3B) nor by P1 receptors (Fig. 2D,E), which is quite different from the effect of ATP released from Müller cells on a subset of GCs that is mediated by P1 receptors. Among the P2Y subtypes, it seems likely that P2Y 1,11 may be involved, as evidenced by the fact that MRS2365/ NF546 induced suppression of glycine currents (Fig. 3E,F). The involvement of P2Y 1,11 was further suggested by the partial blockade of the ATP effect by the antagonists (MRS2500 and NF157) of these receptor subtypes (Fig. 3C,D). Furthermore, the partial blockade due to either MRS2500 (2 μ M) or NF157 (50 μ M) raises a possibility that the subtypes (P2Y 2,4,6 ) other than P2Y 1,11 could be also involved. In fact, P2Y 2,4,6 , just like P2Y 1,11 , are also mainly coupled to G q/11 57 . The involvement of the P2Y 6 subtype seems unlikely since it only responds to UDP and UTP, but not ATP 44,45 . Whether P2Y 2,4 may work together with P2Y 1,11 to mediate the ATP effect remains to be further explored when antagonists for these subtypes are available. Modulation by ATP of glycine responses of retinal GCs observed in this work should be the first report about purinoceptor-mediated modulation of strychnine-sensitive glycine receptors, not only in the retina, but also in the CNS. In the inner retina cholinergic ACs may most likely be the cell type that releases ATP acting on OFF-GCs 26,27 .
By pharmacological dissections, we provided evidence showing that a distinct PI-PLC/PKC signaling pathway, following P2Y receptor activation, may be responsible for the ATP effect on glycine responses of OFF-GCs. Actually, this signaling pathway in P2Y 1,2,4,6,11 receptor-mediated effects has been demonstrated in both neurons and non-neuronal cells 46,52,55,56,63 . Furthermore, the ATP effect on glycine currents is dependent on calcium released from intracellular stores via the IP 3 -sensitive pathway. This is consistent with the observation that activation of P2Y receptors by ATP induces IP 3 -mediated calcium release in astrocytes and spinal dorsal horn 44,64,65 . Consistently, Ca 2+ -dependent PKC (possibly PKCα and β 1 isozymes) was involved in the ATP effect (Fig. 7). This is the first work reporting the intracellular signaling pathway, which is schematically depicted in Fig. S1, responsible for the modulation of ligand-gated channels caused by the activation of P2Y receptors.
It is of interest that glycine responses of rat GCs could be modulated due to activation of other G-protein-coupled receptors via a totally different signaling pathway. In isolated rat retinal GCs melatonin activates the G i/o protein-coupled MT2 receptor, thus potentiating glycine responses of rat GCs through a PC-PLC/Ca 2+ -independent PKC signaling pathway 66 .
Glycine is predominantly released by narrow-field ACs, including AII ACs, and these glycinergic ACs mediate the crossover inhibition between ON and OFF pathways [30][31][32][33][34][35][36][37][38][39] , initiated in the ON pathway, that provides an inhibitory conductance to OFF-GCs by controlling glutamate release from presynaptic OFF cone bipolar cells and directly shapes temporal properties of light-evoked responses of OFF-GCs 32,35,36,40 . The suppression by ATP of glycine responses of OFF-GCs suggests that ATP weakens the crossover inhibition, thus resulting in a regulation of spike patterns in OFF-GCs.
There is evidence, showing that ligand-gated receptors may interact with each other, which is mediated either through direct protein-protein interaction 8,67 or via protein phosphorylation 60,66 . This direct interaction occurs between P2X and GABA A receptors in neurons in the ventromedial nucleus of the hypothalamus 67 and between P2Y and NMDA receptors in layer V pyramidal neurons of the prefrontal and parietal cortex 8 . The ATP effect on glycine responses of OFF-GCs should not be a consequence of such a direct crosstalk between P2Y and glycine receptors, as the effect was abolished by G-protein inhibitors (Fig. 4). Nevertheless, the possibility that a direct crosstalk between P2X and GABA A receptors in OFF-GCs could not be ruled out. If this were the case, it would provide a sophisticated way in that the ATP-induced modulation of inhibitory inputs from ACs, mediated respectively by glycine and GABA receptors, comes into play in two different manners, by activating two distinct purinoceptor subtypes: P2Y and P2X.

Methods
Ethical approval. All animal protocols were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by Animal Care and Use Committee of Shanghai Medical College, Fudan University. Male albino rats (Sprague-Dawley, 15-18 days of age) were used in this study. During this study, all efforts were made to minimize the number of animals used and their pain and discomfort.
Retinal slice preparations. Retinal slices were prepared following the procedures described previously 68 , with minor modifications. Briefly, following deep anesthesia with 25 mg/ml urethane, the eyes were enucleated, and the retinas were removed. The isolated retinas were vertically cut into 200 μ m-thick slices in Ringer's using a manual cutter (ST-20, Narishige, Tokyo, Japan). The slices were transferred into a recording chamber with the cut side up and held mechanically in place by a grid of parallel nylon strings glued onto a U-shape frame of platinum wire. They were then viewed through a fixed-stage upright microscope (BX51WI, Olympus, Tokyo, Japan) equipped with a 60X water-immersion ceramic objective and DIC optics. Unless described otherwise, retinal slices were perfused continuously with oxygenated and carbogen-bubbled Ringer's, which contained (in mM) NaCl 125, KCl 2.5, CaCl 2 2, MgCl 2 1, NaH 2 PO 4 1.25, NaHCO 3  Whole-cell patch-clamp recording. Whole-cell membrane currents of GCs, clamped at − 60 mV, were recorded with pipettes of 6-8 MΩ resistance in voltage-clamp modes filled with the internal solution containing (in mM) CsCl 120, CaCl 2 1, MgCl 2 2, EGTA 10, HEPES 10, ATP-Mg 2, GTP-Na 0.4, NaCl 5 and phosphocreatine 10; adjusted to pH 7.2 with CsOH. Pipettes were mounted on a motor-driven micromanipulator (MP-285, Sutter, Novato, CA, USA), and connected to an EPC10 patch clamp amplifier (HEKA, Lambrecht, Germany). Fast capacitance was fully cancelled and cell capacitance was partially cancelled by the circuits of the amplifier as much as possible. Sixty percent of the series resistance was compensated. Data were acquired at a sampling rate of 5 kHz, and then stored for further analysis. Drug-containing extracellular Ringer's was either locally applied through a puff pipette (tip diameter ~2 μ m), using a pressure micro-injector (PMI-100, DAGAN, Minneapolis, MN, USA), which applied a pressure of 35 kPa (5 p.s.i.) to the top of the pipette, or administrated in bath medium through another inlet by gravity, depending on the purpose of an experiment. Some drugs (GDP-β -S, GPAnt-2a, U73122, Scientific RepoRts | 6:28938 | DOI: 10.1038/srep28938 D609, mastoparan, BAPTA, heparin, xestospongin-C, ryanodine, Bis IV, Gö6976, cAMP, cGMP, Rp-cAMP and KT5823) were dialyzed into neurons after membrane rapture by including them into the patch electrodes. All experiments were performed at room temperature (20-25 °C).
Electrophysiological recordings of light-evoked responses were performed on retinal slices. Dark-adapted (3 h) rats were deeply anaesthetized and retinal slices were prepared under dim red illumination. Slices were transferred to a recording chamber and superfused constantly with oxygenated bicarbonate-buffered Ringer's at 30-32 °C. The pipette solution consisted of (in mM): CsCH 3 SO 3 120, TEA-Cl 10, Hepes 10, CaCl 2 0.1, EGTA 1, phosphocreatine 12, ATP-Mg 3, GTP-Na 0.5; adjusted to pH 7.2 with CsOH. Whole cell light-evoked glycine receptor-mediated IPSCs of GCs were recorded with an EPC 10 amplifier. The cell was held at 0 mV. Bicuculline (10 μ M) and TTX (0.5 μ M) were added to the perfusion solution to block GABA A receptor and voltage-gated sodium channels, respectively. Light stimuli were generated using an LED (λ = 525 nm). Full-field illumination was delivered from the LED, which was controlled by Pulse software (HEKA Elektronik) and delivered to the retina through the microscope condenser. Photon fluxes on the surface of the superfusion chamber were measured with a linear/log optometer (S350, UDT Instruments, San Diego, CA, USA). Light stimuli of 0.5 μ W/cm 2 were provided for 3 s at 60 s intervals.
GCs were distinguished from displaced ACs in the ganglion cell layer (GCL) according to soma diameters and physiological criteria 68,[70][71][72] . ON type GCs (ON-GCs) and OFF-GCs were further identified according to well-established morphological and physiological criteria 30,68,70,73 . Morphologically, ON-and OFF-GCs, revealed by Lucifer yellow, were characterized by their dendrites terminating in proximal and distal parts of the inner plexiform layer, respectively 30,68,70 . Physiologically, 500-ms negative current injection in the current-clamp mode led to rebound burst firing in the OFF-GCs, but not in the ON-GCs 68,70,73 . Figure S2 shows an OFF-GC intracellularly stained by Lucifer yellow (A) and its response to a 500-ms negative current injection (B).
Calcium imaging. Changes in intracellular calcium concentration ([Ca 2+ ] i ) were assessed using the membrane permeable indicator fura-2 AM (Dojindo, Kumamoto, Japan). Fura-2 AM was dissolved in 20% Pluronic F-127 (w/v, DMSO) and added to a chamber that contained Ringer's, with a final fura-2 AM concentration of 2 μ M. Isolated GCs were incubated in the dye solution for 30 min at room temperature and then perfused with dye-free Ringer's for at least 15 min. Digital fluorescence images were acquired with an inverted microscope (IX-70, Olympus) furnished with a digital CCD camera (ORCA-ER; Hamamatsu Photonics, Shizuoka, Japan). A high-speed continuously scanning monochromatic light source (Polychrome V; Till Photonics, Gräfeling, Germany) was used for the excitations at wavelengths of 340 nm and 380 nm. Fluorescence intensities at both wavelengths (F 340 and F 380 ) were measured every 3-10 s, and images were obtained using PC-based software (C-imaging systems; Hamamatsu Photonic). The ratio between the two images was proportional to [Ca 2+ ] i of the cell under study. Prior to an experiment, a background level of fluorescence (attributable to autofluorescence and camera noise) was measured and subtracted from all the obtained data.
Chemicals. D-AP5, PPADS, suramin, Evans blue, NF157, NF546, MRS2500, MRS2365, ARL67156, GPAnt-2a and ryanodine were purchased from Tocris Bioscience (Ellisville, MO, USA). All other chemicals were obtained from Sigma-Aldrich (St. Louis, MO, USA). Adenosine, U73122, ryanodine, Bis-IV, PMA and Gö6976 were initially dissolved in DMSO for stock and then diluted in solutions to final working concentrations. The final DMSO concentration was less than 0.1%, with no effects on glycine-induced currents of GCs. All other drug solutions were prepared in ion-free water, stored at − 20 °C and freshly diluted to the working concentrations using normal solutions.
Statistical analysis. The data are presented as means ± SEM. Student's t test (paired data) and one way analysis of variance (ANOVA) followed by post hoc Tukey's tests (multiple comparisons) were used to identify significant differences. In all cases, P < 0.05 was considered to be statistically significant.