Luteolin inhibits GABAA receptors in HEK cells and brain slices

Modulation of the A type γ-aminobutyric acid receptors (GABAAR) is one of the major drug targets for neurological and psychological diseases. The natural flavonoid compound luteolin (2-(3,4-Dihydroxyphenyl)- 5,7-dihydroxy-4-chromenone) has been reported to have antidepressant, antinociceptive, and anxiolytic-like effects, which possibly involve the mechanisms of modulating GABA signaling. However, as yet detailed studies of the pharmacological effects of luteolin are still lacking, we investigated the effects of luteolin on recombinant and endogenous GABAAR-mediated current responses by electrophysiological approaches. Our results showed that luteolin inhibited GABA-mediated currents and slowed the activation kinetics of recombinant α1β2, α1β2γ2, α5β2, and α5β2γ2 receptors with different degrees of potency and efficacy. The modulatory effect of luteolin was likely dependent on the subunit composition of the receptor complex: the αβ receptors were more sensitive than the αβγ receptors. In hippocampal pyramidal neurons, luteolin significantly reduced the amplitude and slowed the rise time of miniature inhibitory postsynaptic currents (mIPSCs). However, GABAAR-mediated tonic currents were not significantly influenced by luteolin. These data suggested that luteolin has negative modulatory effects on both recombinant and endogenous GABAARs and inhibits phasic rather than tonic inhibition in hippocampus.

Despite accumulating studies suggesting a modulatory effect of luteolin on GABA A Rs in the CNS 7,9 , the pharmacological characterization of luteolin still remains unclear. For instance, studies have shown that luteolin has antidepressant and analgesic effects that are considered to be mediated by enhancing GABA A R functions 6,8 ; however, luteolin did not exhibit any pro-or anti-convulsant effects in various animal models of epilepsy 22 . Furthermore, luteolin has been reported to enhance learning and memory in a neurodegenerative model 10 . Children with autism spectrum disorders showed improvement in adaptive functioning after receiving dietary supplement of flavonoids including luteolin (100 mg/10 kg weight daily) 23 , which produced an estimated concentration of ~8.7 μ mol/L in plasma according to a study on the relationship between oral intake and plasma concentration of luteolin 24 . Hence, these previous studies have led us to hypothesize that luteolin might have varied effects on different subtypes of GABA A Rs to achieve the complex neuropharmacological influences. To elucidate the pharmacological mechanisms underlying these study results, we investigated the sensitivities of α 1-and α 5-containing GABA A Rs to different doses of luteolin in HEK cells. Furthermore, the effect of luteolin on α 1-and α 5-containing GABA A R-mediated phasic and tonic inhibition in hippocampal slices was also studied.

Effects of luteolin on dose-response relationships of GABA A Rs in HEK cells. Inclusion of α and
β subunits is required to create functional GABA A R pentamers and the α β receptors are already sufficient for insertion into the plasma membrane. Despite low levels of expression, the α β receptors naturally exist in the CNS and in peripheral systems 13,25 . The α β receptors are also useful for molecular pharmacology studies using recombinant expression systems like HEK cells. The α β γ receptors are the most abundant form of GABA A Rs in the CNS and are sensitive to benzodiazepine potentiation. Among the α β γ compositions, the α 1β 2γ 2 receptors are the most predominant form and are ubiquitously distributed in inhibitory synapses to mediate phasic inhibition, whereas α 5β 2γ 2 receptors are predominantly located at the extrasynaptic region to mediate the tonic inhibition in hippocampus. In the present study, we selected recombinant α 1β 2, α 1β 2γ 2, α 5β 2, and α 5β 2γ 2 GABA A Rs to characterize the pharmacological effects of luteolin.
Whole-cell recordings were performed in HEK293T cells expressed with α 1β 2, α 1β 2γ 2, α 5β 2, or α 5β 2γ 2 subunits of GABA A Rs. Fast-perfusion of GABA (0.1-500 μ M) produced inward current responses at a holding membrane potential of − 60 mV. Recombinant expression of GABA A Rs resulted in dose-response curves with EC 50 values within a range of 2 to 4 μ M (Fig. 1, Table 1). Extracellular application of luteolin (50 μ M) inhibited current responses induced by the agonist from medium to saturating doses (1-500 μ M GABA) in all tested forms of GABA A Rs. Although the maximum currents (I max ) of GABA A Rs were substantially reduced by luteolin, the EC 50 was relatively less affected (with 0.8-, 1.7-, 1.6-, and 2.1-fold changes by 50 μ M luteolin in α 1β 2, α 1β 2γ 2, α 5β 2, and α 5β 2γ 2 receptors, respectively, Table 1). These results indicate that luteolin is likely a non-competitive antagonist of GABA A Rs.

Figure 1. Inhibition of GABA-activated currents by luteolin in the recombinant GABA A Rs. Recombinant
GABAARs composed by α 1β 2 (A), α 1β 2γ 2 (B), α 5β 2 (C), and α 5β 2γ 2 (D) were first activated by GABA in the absence of luteolin (circles). Dose-response relationships were calculated from the value of peak whole-cell current amplitudes induced by varying GABA concentrations normalized to the value of I max that were activated by maximum GABA concentration. In the presence of 50 μ M of luteolin (square), the amplitudes of GABAactivated currents were also normalized to the value of I max of the same cell. All data points and bars represent mean values ± s.e.m. More detailed data for the dose-response curves of GABAARs are presented in Table 1 Potency and efficacy of luteolin on different forms of GABA A Rs. We next examined the potency and efficacy of luteolin on the four forms of GABA A Rs by testing the dose-response relationship of luteolin-mediated inhibition. 0.1-100 μ M of luteolin was included in the perfusion solution and was applied onto the same cell in sequence. By using medium doses of GABA to induce current responses, luteolin strongly inhibited GABA currents at concentrations> = 10 μ M (Fig. 2), which exhibited a similar pattern to the effects of other flavonoids like apigenin and quercetin on GABA A Rs 26,27 . By comparing the IC 50 of luteolin on the four types of GABA A Rs, we found that luteolin inhibited GABA currents with similar potency, with the lowest IC 50 in α 1β 2 (10.8 ± 4.46 μ M). In the other forms, the IC 50 values of luteolin are 51.4 ± 242.8 μ M in α 1β 2γ 2, 25.4 ± 9.0 μ M in α 5β 2, and 27.5 ± 13.8 μ M in α 5β 2γ 2. Moreover, the inhibitory efficacy of luteolin was assessed by the extent of reduction in I max . Our results showed that luteolin had better efficacy on α 1β 2 and α 5β 2 receptors ( Table 1). Inclusion of the γ 2 subunit decreased the efficacy of luteolin, suggesting that the γ 2 subunit is not required for forming luteolin binding site for its inhibition.
We also compared the inhibitory efficacy of luteolin on current responses mediated by medium or high doses of GABA. In the α 1β 2 and α 5β 2 receptors, both medium-and high-dose GABA-mediated currents were significantly inhibited by 50 μ M of luteolin (Fig. 3A,C). In the α 1β 2γ 2 receptors, however, luteolin had an inhibitory effect on 500 μ M but not 3 μ M GABA-induced currents (Fig. 3B). In the α 5β 2γ 2 receptors, luteolin showed a better efficacy on 2 μ M compared with 500 μ M GABA-mediated currents (Fig. 3D). Taken together, these data  showed that luteolin exhibited varied degrees of potency and efficacy to reduce current responses of different forms of GABA A Rs.
Luteolin affected activation kinetics of GABA A Rs. Activation kinetics of GABA currents can influence the time course of the neurotransmitter-evoked responses at the synapses 28 . In the present experiments, currents were elicited by fast perfusion of GABA at concentrations close to EC 50 , and the activation kinetics were characterized by the 10-90% activation time of the current after fast perfusion of GABA. Luteolin at high concentrations Representative current traces showed medium or high doses of GABA-activated current responses in α 1β 2 (A), α 1β 2γ 2 (C), α 5β 2 (E), and α 5β 2γ 2 receptors (G) before and after 50 μ M of luteolin. The quantitative results of luteolin inhibition were calculated from the value of GABA currents in the presence of luteolin normalized to the value before luteolin in α 1β 2 (n = 9) (B), α 1β 2γ 2 (n = 9) (D), α 5β 2 (n = 10) (F), and α 5β 2γ 2 receptors (n = 10) (H). All data points and bars represent mean values ± s.e.m. * P < 0.05; * * P < 0.01; * * * P < 0.001; n.s., no significance using student's t-test. (100 μ M) slowed the activation time in all tested forms of GABA A Rs (Fig. 4). In the presence of 100 μ M luteolin, the activation time of α 1β 2, α 1β 2γ 2, α 5β 2, and α 5β 2γ 2 GABA A Rs after luteolin treatment was increased by 3.5-, 2.4-, 5.8-, and 4.1-fold, respectively. Luteolin concentration lower than 10 μ M had no significant effects on

Effects of luteolin on phasic currents mediated by GABA A Rs in hippocampal slices. The hip-
pocampus is the critical locus for learning and memory. The α 1β 2γ 2 GABA A Rs are located at the postsynaptic site of inhibitory synapses and predominantly mediate the fast inhibitory synaptic currents in hippocampus. To further investigate the pharmacological modulation of luteolin on endogenous α 1β 2γ 2 GABA A Rs under physiological conditions, we tested the effect of luteolin on phasic inhibition in hippocampal slices. The GABA A R-mediated mIPSCs were pharmacologically isolated by inclusion of the sodium channel blocker TTX (0.5 μ M) and the AMPA/kainate receptor blocker CNQX (20 μ M). At the end of each experiment, the GABA A R antagonist bicuculline (10 μ M) was applied to confirm that the mIPSCs were abolished (Fig. 5A). Here we selected two representative doses (0.1 and 100 μ M) to test the effect of luteolin. We showed that 100 μ M luteolin significantly decreased the amplitude (87.7 ± 3.4% normalized to baseline) but not the frequency (103.6 ± 6.2%) of mIPSCs ( Fig. 5A-D).
The rise time of mIPSCs was also significantly slower after high dose of luteolin treatments (108.6 ± 2.0%) represent mean values ± s.e.m. * P < 0.05, * * P < 0.01compared with control using one-way ANOVA.
Effects of luteolin on tonic currents mediated by GABA A Rs in hippocampal slices. The α 5β 2γ 2 GABA A Rs are located at the extrasynaptic site and mediate tonic inhibition in hippocampus. Our data have shown that, in the recombinant α 5β 2γ 2 GABA A Rs, luteolin inhibited medium to high doses GABA-mediated responses.
Considering that tonic inhibitory currents in the CNS are ascribed to the low concentration of ambient GABA in the extracellular space (from 0.2 to 2.5 μ M) 29 , it still remained to be determined whether luteolin has any modulatory effect on the low-dose GABA-mediated responses in slices. In our experiments, tonic currents were defined as the shift of holding currents after application bicuculline (10 μ M). In some experiments, we also included a low dose of GABA in the bath solution (0.5 μ M) to unify the ambient GABA concentration. Treatments with 0.1 or 100 μ M luteolin did not significantly affect the amplitudes of tonic currents (P > 0.05 for both 0.1 and 100 μ M groups) (Fig. 6). These results suggested that luteolin did not influence extrasynaptic GABA A R-mediated tonic inhibitory currents.

Discussion
The pharmacological effect of luteolin on recombinant and endogenous GABA A Rs. Luteolin has been widely studied for its pharmacological effects, including anti-inflammatory, anti-oxidant, and anticarcinogenic activities. Recent studies have suggested that luteolin might enhance the function of GABA A Rs, thereby producing antihyperalgesic, anxiolytic, and antidepressant-like effects in the CNS 5,6,8 . However, it still remains unclear whether luteolin takes these effects by directly targeting on GABA A Rs. In the brain, modulation of distinct subunit compositions of GABA A Rs is associated with different neurological and behavioral outcomes. Enhancements of α 1β 2γ 2 GABA A Rs that mainly mediate fast synaptic inhibition generally produce sedative effects, while inhibition of α 5β 2γ 2 GABA A Rs by either pharmacological or transgenic approaches can promote learning and memory. In the present study, we showed that low concentration of luteolin (0.1 μ M) did not affect either GABA A R-mediated phasic or tonic currents. In contrast, high concentration of luteolin (100 μ M) reduced the amplitude of mIPSCs and prolonged the activation time course, but had no effect on mIPSC frequency. This indicated that the effect of luteolin was likely due to postsynaptic rather than presynaptic modulation. The inhibition of mIPSCs by luteolin was consistent with the observation from HEK cells: high concentration of luteolin suppressed the amplitude and prolonged the activation kinetics in recombinant GABA A Rs including the α 1β 2γ 2 form. Although we have shown that luteolin reduced current responses that were induced by high dose but not medium dose of GABA in recombinant α 1β 2γ 2 receptors (Fig. 3D), it was still in line with the results from slices since vesicular release of GABA at synapses normally reaches a very high concentration 30 . Furthermore, our results are consistent with previous studies that luteolin did not affect the threshold of seizure induction by pilocarpine or electrical stimuli in animal models 22 , suggesting that luteolin cannot produce obvious sedative effect through enhancement of the function of α 1β 2γ 2 GABA A Rs. The tonic inhibition in hippocampus was unaffected by luteolin even at concentrations as high as 100 μ M. However, the α 5β 2γ 2 GABA A R-mediated responses in HEK cells were significantly reduced. This could be ascribed to the low concentration of ambient GABA in the extracellular space in hippocampal slices, since luteolin showed very weak efficacy to modulate low dose of GABA-mediated responses (at [GABA] < 1 μ M) (Fig. 1D). Therefore, our findings indicated that luteolin is a negative modulator for GABA A Rs and did not have any potentiation effect on phasic or tonic inhibition in hippocampus.
Previous studies of luteolin mostly focused on the in vivo effects. One of the major differences between the in vivo and in vitro environments is the temperature. In our study, we performed in vitro electrophysiological experiments in room temperature (23-25 °C), which was lower than the body temperature (37 °C). Notably, some allosteric modulators, such as zolpidem, can modulate GABA A Rs in a temperature-dependent manner 31 . The affinity of zolpidem to GABA A Rs increased along with the increasing temperature from 16, 26 to 36 °C 31 . Previous studies showed that luteolin was stable at 37 °C in culture medium for 24 hours 32 . We thereby predict that luteolin might consistently take effects and show increased inhibition on GABA A Rs in vivo. Moreover, our data revealed that luteolin showed stronger inhibitory effects on recombinant GABA A Rs in HEK cells than in the endogenous GABA A Rs in brain slices. It was possibly due to the lack of synaptic scaffolding proteins in HEK cells and this might affect luteolin-mediated inhibition on GABA A Rs as suggested by previous studies 33 . Luteolin likely targets at non-benzodiazepine binding sites of GABA A Rs. Benzodiazepines are one of the most potent positive modulators of γ -containing GABA A Rs and the binding site is located at the α (+ )/γ (− ) interface. Previous studies have indicated the structural similarity between flavones and benzodiazepine ligands 34 . Among different types of flavones, the presence of electronegative groups at the C6 and C3′position are critical determinants for high affinity to the benzodiazepine-binding site 35 . For instance, hispidulin (4′ ,5,7-trihydroxy-6-methoxyflavone) bearing a methoxyl group at C6 is a potent benzodiazepine site ligand and potentiates GABA A R-mediated responses in α 1β 2γ 2 form but not in α 1β 2 form receptors 36 . The structure of luteolin (5,7,3′ ,4′ -Tetrahydroxyfavone) is similar to apigenin (5,7,4′ -Trihydroxyflavone) and quercetin (3,5,7,3′ ,4′ -Pentahydroxyfavone), all of which lack electronegative moieties at C6. Previous studies have demonstrated that luteolin and quercetin have weak affinities for the benzodiazepine site, with K i values over 100 μ M for the [3H] flunitrazepam binding competition 35 . Apigenin exhibited a higher affinity for central benzodiazepine receptors (with a K i of 4 μ M to compete [3H]flunitrazepam binding) 37 and exerted anxiolytic and antidepressant effects in in vivo animal models. However, whether a direct involvement of GABA A Rs in the CNS is responsible of the effects of apigenin remains questionable. Electrophysiological studies showed that apigenin and quercetin similarly inhibited GABA-induced currents, while the inhibition of apigenin on α 1β 2γ 2 GABA A R-mediated responses were not prevented by the benzodiazepine site antagonist flumazenil 26 . These studies indicated that, different from the traditional anxiolytic chemical benzodiazepine, the CNS effects of apigenin and quercetin are not likely due to their direct interaction and potentiation of GABA A Rs. In the present study, we found that luteolin negatively modulated GABA A Rs that lacked γ subunits. In agreement with previous studies using the [3H] flunitrazepam binding assay, our results indicated that luteolin inhibited GABA A Rs through non-benzodiazepine site of GABA A Rs. Such a modulatory effect of luteolin is in resemblance of that of apigenin and quercetin. We did not observe apparent potentiation effects of luteolin on either α 1-or α 5-containing GABA A Rs that were reported for hispidulin, likely due to the lack of a hydroxyl group at the C6 position of flavone 9 . Therefore, although we did not exclude the possible interaction between benzodiazepine sites and luteolin's metabolites, the direct effect of luteolin on GABA A Rs did not involve binding to central benzodiazepine receptors. In other words, the CNS effects of luteolin are likely obtained through other pharmacological mechanisms, possibly similar to those of apigenin and quercetin due to their structural similarity.

Different modulation on αβ and αβγ forms of GABA A Rs indicated possible luteolin-binding sites.
During the development of benzodiazepine ligands, researchers have discovered new allosteric modulation sites on GABA A Rs. Ramerstorfer et al. have revealed a new ligand-binding site at the α (+ )/β (− ) interface that was independent from the benzodiazepine site at the α (+ )/γ (− ) interface 38 . This new binding site was discovered by screening of benzodiazepine site ligand and was determined by CGS 9895 that could potentiate GABA A Rs in a flumazenil insensitive manner regardless of the incorporation of γ subunits, indicating that CGS 9895 targets on non-BZ-binding sites of GABA A Rs 38 . Interestingly, CGS 9896, which is a structural analog of CGS 9895, exhibited a similar manner to 6-Methylflavone in a pharmacophore model of benzodiazepine site binding 34 . Given that the inhibition by luteolin, apigenin, and quercetin is independent from γ incorporation and is insensitive to flumazenil, it is reasonable to speculate the possibility that these flavones might interact with the newly identified CGS 9895-binding site at the α (+ )/β (− ) interface. Our results showed that luteolin had more potent effects on the α β compared with the α β γ receptors, agreeing with the fact that the α β receptors embrace two α (+ )/β (− ) interfaces while the α β γ form only contains one α (+ )/β (− ) interface. However, the limited information about the exact molecular location of CGS 9895-binding site precludes further investigation of this hypothesis. In addition, we cannot rule out the possibility that luteolin targets on other modulation sites like the neurosteroid-binding site, which is located at the transmembrane domains of GABA A Rs. Methods cDNA constructs and transfection. Rat GABA A R α 1, α 5, β 2, and γ 2 subunits were subcloned into the pCDNA3.1 expression vector. HEK293T cells (6 × 10 5 ) were transfected with purified plasmids encoding GABA A Rs (total plasmid amount 2-2.5 μ g) by electroporation (NEPA21, NEPA GENE). A small amount (0.2 μ g) of pcDNA3-GFP was co-transfected along with GABA A Rs to act as a transfection marker and facilitate the visualization of transfected cells during electrophysiological experiments. After transfection, cells were re-plated on poly-(D-lysine)-coated glass coverslips and were grown in DMEM in 24-well plates for 16-24 h before patchclamp recordings.
Slice preparation. Hippocampal slices were prepared from 12-21-day-old ICR mice. Treatments of animals were evaluated and approved by the Institutional Animal Care and Use Committee of China Medical University according to Care of the animals and surgical procedures of China Medical University Protocols. Mice were anaesthetized with Urethane and decapitated. Brains were removed into ice-cold slicing solution containing (in mM): 230 sucrose, 26 NaHCO 3 , 10 D-glucose, 2.5 KCl, 1.25 NaH 2 PO 4 , 0.5 CaCl 2 , and 10 MgSO 4 . 350 μ m-thick transverse hemi-sections from hippocampus were sliced (Leica vibratome) in the slicing solution. Then the slices were transferred to a storage chamber or the recording chamber with fresh artificial cerebrospinal fluid (ACSF) containing the following (in mM): 128 NaCl, 2.5 KCl, 2.0 MgCl 2 , 2.0 CaCl 2 , 1.25 NaH 2 PO 4 , 26 NaHCO 3 , and 10 D-glucose, and were incubated at room temperature for > 1 h before recording. All solutions were saturated with 95% O 2 /5% CO 2 .
Electrophysiology. For recording HEK cells, whole-cell patch clamp recordings were performed under voltage-clamp mode using the AXOPATCH 200B amplifier (Molecular Devices, USA). Whole-cell currents were recorded with a holding potential of − 60 mV and signals were acquired via a Digidata 1440A analog-to-digital interface and were low-pass filtered at 2 kHz and digitized at 10 kHz. Patch electrodes (3-7 MΩ) were pulled from 1.5 mm outer diameter thin-walled glass capillaries in three stages on a Flaming-Brown micropipette puller and were filled with intracellular solutions (ICS), which contained (in mM) 140 CsCl, 10 HEPES, 4 Mg-ATP and 0.5 BAPTA (pH 7.20, osmolarity, 290-295 mOsm). The coverslips were continuously superfused with the extracellular solution containing (in mM): 140 NaCl, 5.4 KCl, 10 HEPES, 1.0 MgCl 2 , 1.3 CaCl 2 and 20 glucose (pH 7.4, 305-315 mOsm). To evoke GABA currents, we used fast perfusion of GABA with a computer-controlled multibarrel fast perfusion system (Warner Instruments, CT, USA). For recording hippocampal neurons, slices were perfused with ACSF. Hippocampal CA1 pyramidal neurons were recorded with whole-cell patch clamp with a holding potential of − 60 mV under voltage-clamp model using the MultiClamp 700B amplifier (Molecular Devices, USA). Recording pipettes (3-7 MΩ) were filled with the ICS that was mentioned above. For testing the effects of luteolin, slices were pretreated with luteolin for 5-10 min. For recording miniature inhibitory postsynaptic currents (IPSCs), all bath solutions (ACSF) contained 0.5 μ M TTX and 20 μ M CNQX. The seal tests were performed through the application of a − 5 mV step about every 5 min to monitor the changes in access resistance. Data were collected only if the whole-cell access resistance was consistent throughout the recording (changes < 15%). All experiments were performed at 23-25 °C.

Data analysis.
Values are expressed as mean ± s.e.m. One-way ANOVA or a two-tailed Student's t-test was used for statistical analysis and P values less than 0.05 were considered to be statistically significant. Peak current amplitude and 10-90% rises time of recombinant GABA A R-mediated response was measured by Clmapfit 10. Maximum currents (I max ) were determined as the amplitude of peak currents induced by saturated concentration or indicated concentration of agonists. Dose-response curves were created by fitting data to the Hill equation: I = I max /[1 + (EC 50 /[A]) nH ], where I is the current, [A] is a given concentration of agonist and n H is the Hill coefficient (GraphPad Prism 6, CA, USA). The amplitude, frequency, and rise time of mIPSCs were measured by MiniAnalysis. The amplitude of tonic currents was revealed by measuring the change in the holding current evoked by applying the GABA A R antagonist bicuculline. The baseline current of 10 s was selected for each treatment and was analyzed by generating an all-points histogram and fitting a Gaussian distribution to the positive side of the histogram by Clampfit 10. The means of the fitted Gaussian were used to determine the holding current before and after drug application 39 .