Activation of toll like receptor 4 attenuates GABA synthesis and postsynaptic GABA receptor activities in the spinal dorsal horn via releasing interleukin-1 beta

To study the impact of TLR4 activation on GABAergic synaptic activities, GABAergic mIPSCs were first recorded from neurons located in spinal lamina II in rats in the presence of 1 μM TTX before and after bath perfusion of the well-known TLR4 agonist LPS (1 μg/ml) [26]. As shown in Figure 1A, LPS significantly reduced mIPSC amplitudes and frequencies at the same time. These effects appeared at 4 minutes from the beginning of LPS perfusion and plateaued between 6 to 8 minutes. At 10 minutes after the perfusion, LPS significantly reduced the mean mIPSC amplitudes by 15.35 ± 1.45% (n = 15, P < 0.001) and mean mIPSC frequencies by 29.53 ± 2.91% (n = 15, P < 0.001). Similarly, when neurons in the same area in mice were recorded, LPS (1 μg/ml) significantly reduced the mean mIPSC amplitudes by 14.69 ± 3.28% (n = 7, P < 0.01) and frequencies by 27.11 ± 2.36% (n = 7, P < 0.001). It is generally believed that TLR4 is predominantly expressed in microglia [1,2,27]. If this is the case, the effects induced by LPS on mIPSCs should be abolished when activation of microglia is blocked. Previous studies have shown that minocycline inhibits the production of immune activators by macrophages, microglia [28,29], and monocytes [30]. More specifically, minocycline has been shown to specifically suppress microglial activation in the spinal dorsal horn at the concentration of 100 μM [31]. We then perfused minocycline (100 μM) into the recording for 10 minutes prior to addition of LPS into the bath. In the presence of minocycline, LPS no longer altered the amplitudes or frequencies of GABAergic mIPSCs recorded from rats (Figure 1B) and mice (amplitudes: n = 7, P = 0.82; frequencies: n = 7, P = 0.55, data not shown). Previous studies show that intraperitoneal injection of minocycline (40 mg/kg) suppresses phosphorylation of extracellular signal regulating kinases (ERK) in spinal dorsal horn neurons induced by intraperitoneal injection of acetic acid [32]. To confirm that the minocycline treatment under our experimental conditions did not alter the GABAergic synaptic activities, we examined the effects of minocycline on GABAergic mIPSCs. We found that perfusion of minocycline at 100 μM for 20 minutes did not significantly alter the amplitudes (P = 0.62) or frequencies (P = 0.52) of mIPSCs recorded from neurons in mice (n = 5). These data indicate that LPS suppresses GABAergic synaptic activities through activation of microglia. Several mechanisms may underlie the LPS-induced attenuation of GABAergic synaptic activities, which at least include (a) a reduced release probability of GABA at presynaptic terminals; (b) a decreased function of GABA receptors at postsynaptic neurons; and (c) the decreased synthesis of GABA. We conducted the following experiments to determine which of these mechanisms contributes to the changes of GABAergic synaptic activities induced by LPS.

To determine whether the function of postsynaptic GABA receptors is altered during perfusion of LPS, we recorded GABAergic currents evoked by exogenous GABA injected onto the recorded neuron through a puffing glass pipette (GABA concentration: 100 μM; duration: 25 ms). As shown in Figure 2A, GABAergic currents evoked by exogenous GABA (n = 9, P < 0.001) were significantly reduced during perfusion of LPS, indicating that activation of TLR4 results in attenuation of neuronal GABA receptor activities under our experimental conditions. It has been known that IL-1β is released from microglia activated by LPS [27] and IL-1β receptors are expressed in spinal dorsal horn neurons [33]. We tested if IL-1β mediates the effects induced by LPS on neuronal GABA currents. After recording baseline GABA currents evoked by GABA injected onto the recorded neuron through the puffing glass pipette, we perfused the IL-1β antagonist (IL-1ra, 100 ng/ml) into the recording chamber and re-recorded GABA. GABA currents were not altered by IL-1ra perfusion (Figure 2B), indicating that under normal conditions, activities of GABA receptors in neurons are not under the control of endogenous IL-1β. In the presence of IL-1ra (100 ng/ml) further addition of LPS (1 μg/ml) into the bath did not alter GABA current areas and amplitudes (Figure 2B), indicating that IL-1β mediates the inhibitory effects induced by LPS on neuronal GABA receptor activities. Further, perfusion of IL-1β (10 ng/ml) into the recording bath significantly reduced GABAergic currents (Figure 2C), consistent with a previous report [34]. Together, these data indicate that LPS suppresses neuronal GABA receptor activities via releasing IL-1β.

We recently showed that PKC is an important kinase activated by IL-1β in the spinal dorsal horn [25]. Thus, we determined if the effects induced by LPS are mediated by PKC activation. We found that bath perfusion of a PKC activator (PMA, 2 μM) significantly reduced amplitudes and areas of GABA currents (Figure 3A). The effects induced by PMA under such condition may result from direct activation of PKC in the recorded neuron or from indirect effects induced by activation of PKC in other cell types. To specifically address this issue, PKC in the recorded neuron was blocked by microdialyzing the PKC inhibitor (PKCI 19–30, 5 μM) [35], included in the recording glass pipette, into the recorded neuron. Recordings were made 15 minutes after rupturing the cell. Under such condition, GABAergic currents induced by the puffed GABA (100 μM) remained unchanged when LPS (1 μg/ml) was bath perfused (Figure 3B). Together with the data that the LPS-induced inhibitory effects on GABAergic receptor activities were prevented by minocycline (Figure 1B), these data indicate that LPS causes release of IL-1β from microglia, and IL-1β in turn suppresses neuronal GABA receptor activities through activating PKC in neurons.

To determine whether presynaptic mechanisms contribute to the LPS-induced suppression of GABAergic synaptic activities, the effect induced by LPS on postsynaptic GABA receptors was blocked by microdialyzing the recorded neuron with the PKC inhibitor PKCI (PKCI 19–30, 5 μM) included in the pipette solution. Under such condition, we first recorded GABAergic mIPSCs before and after perfusion of LPS (1 μg/ml). LPS significantly reduced the mean GABAergic mIPSC amplitudes by 5.57 ± 1.64% (n = 8, P < 0.01) and frequencies by 20.12 ± 1.87% (n = 8, P < 0.001) (Figure 4). These data suggest that presynaptic mechanisms contributed to the LPS-induced suppression of GABAergic synaptic activities.

The suppression of mIPSC frequencies could result from a reduction in the presynaptic GABA release probability, whereas suppression of both frequencies and amplitudes of mIPSCs more likely results from a reduction of GABA synthesis at the presynaptic site [20,36,37]. To determine if the GABA release probability at the presynaptic site is altered by LPS, we recorded two evoked GABAergic IPSCs (eIPSCs) elicited by paired pulses separated by 100 ms and analyzed the paired pulse ratio (that is P2:P1 ratio) [18,20]. P1 was the amplitude of the first evoked current and P2 was the second response amplitude measured after subtraction of the remaining P1 “tail current” [20]. An increase or decrease in the P2:P1 ratio is conventionally used to detect a decrease or increase in neurotransmitter release probability at the presynaptic site [21,38-41], including GABAergic synapses in the spinal dorsal horn [20,42]. As shown in Figure 5, although both P2 and P1 values of the GABAergic eIPSCs were reduced after bath perfusion of LPS, the P2:P1 ratios remained unchanged (0.67 ± 0.014 at baseline and 0.66 ± 0.025 during LPS perfusion, n = 11, P = 0.76). These results suggested that the LPS-induced suppression of GABA synaptic activities is not ascribed to the reduced release probability of GABA from presynaptic terminals. Instead, these data suggest that LPS may reduce GABAergic activities through impairing GABA synthesis.

We recently demonstrated that IL-1β reduces glial glutamate transporter activities [25] and the impairment of glial glutamate transporters results in a reduction of GABA synthesis through the glutamate-glutamine cycle [20]. Thus, we extrapolated that the endogenous IL-1β release induced by LPS reduces GABA synthesis by suppressing glial glutamate transporter activities. We then investigated the effects of LPS on glial glutamate transporter activities in astrocytes. The uptake of glutamate by glial glutamate transporters is accompanied by the cotransport of two to three Na⁺ with one H⁺ and the countertransport of one K⁺ [43-45]. Because of the translocation of a net positive charge during each transport cycle, glutamate uptake generates a current called glutamate transporter current (GTC) [43-46]. The size of GTC reflects the amount of transported glutamate, which has been widely used as an effective tool to study the function of glial glutamate transporters [47-49]. GTCs were recorded from GFAP-GFP transgenic mice as we described previously [25]. Astrocytes in these mice were easily identified by the expression of GFP under a fluorescent microscope. We found that LPS (1 μg/ml) bath perfusion reduced amplitudes and charge transfers of GTCs (Figure 6A). This effect was blocked in the presence of the selective IL-1β inhibitor IL-1ra (100 ng/ml) (Figure 6B). We recently showed that IL-1β reduces glial glutamate transporter activities through activating PKC. Here, we further confirmed these conclusions by determining the effect of LPS on GTCs in the presence of the PKC inhibitor GF109203X (4 μM). As expected, LPS no longer attenuated GTC amplitudes and charge transfers under such condition (Figure 6C). Furthermore, when PKC in the recorded astrocyte was blocked by microdialyzing the recorded neuron with the PKC inhibitor PKCI (PKCI 19–30, 5 μM) included in the pipette solution, GTC amplitudes (n = 8, P = 0.44) and charge transfers (n = 8, P = 0.91) were not significantly altered by LPS (1 μg/ml) bath perfusion (Figure 6D). Taking together, these data indicate that LPS suppresses glial glutamate transporter activities through releasing IL-1β and activating PKC.

To further dissect the presynaptic mechanism underlying the changes in GABAergic synaptic activities induced by LPS, the results shown in Figures 7 and 8 were collected under the condition when the PKC inhibitor PKCI (PKCI 19–30, 5 μM) was included in the pipette solution. If the LPS-induced suppression of glial glutamate transporter activity leads to a reduction of GABA synthesis, the amount of GABA released from presynaptic terminal vesicles should be reduced. In other words, the concentration in the GABAergic synaptic cleft is reduced [18,20]. To prove this, we examined the inhibition induced by TPMPA on GABAergic IPSCs in the absence and presence of LPS. Because TPMPA is a low-affinity competitive antagonist of GABA_A receptors, the binding of TPMPA to GABA_A receptors is readily replaced by synaptic GABA neurotransmitter. When GABA_A receptors are exposed to TPMPA, the degree of inhibition caused by TPMPA on GABAergic IPSCs is inversely related to the concentration of GABA transients (the amount of GABA release) in the synaptic cleft [18,50]. As shown in Figure 7, the percentage of inhibition by TPMPA (60 μM) on GABAergic IPSCs in the presence of LPS (37.04 ± 1.72%, n = 14) was significantly higher (P < 0.01) than that in the absence of LPS (19.90 ± 0.62%, n = 15). These data indicate that inhibition induced by LPS on GABAergic IPSCs was indeed due to a decrease, at the synaptic cleft, in the amount of GABA release from presynaptic neurons.

Based on the glutamate-glutamine cycle theory (see introduction [16]), we reasoned that if the inhibitory effects induced by LPS on GABAergic IPSCs is due to its inhibition on glial glutamate transporters and subsequent inhibition of the glutamate-glutamine cycle, glutamine concentrations in the extracellular space in the presence of LPS should be reduced. These would result in a deficient glutamine supply to GABAergic neurons and a reduction in GABA synthesis. If this holds, supplementing exogenous glutamine in the recording bath should prevent the reduction induced by LPS in GABAergic eIPSCs [18,20]. As expected, in the presence of 1 mM glutamine, LPS no longer significantly altered amplitudes and frequencies of GABAergic mIPSCs (Figure 8).

As IL-1β and PKC mediate the suppressive effects of LPS on GTCs, we reasoned that inhibition of either IL-1β or PKC should prevent the suppressive effects of LPS on GABAergic IPSCs. As expected, we found that in the presence of the IL-1β inhibitor IL-1ra (100 ng/ml) (Figure 9A and C) or PKC inhibitor GF109203X (4 μM) (Figure 9B and D), perfusion of LPS did not alter amplitudes and frequencies of GABAergic mIPSCs. Taking together data in Figures 5, 6, 7 and 8, we conclude that: a. activation of TLR4 by LPS reduces GABAergic synaptic activities through suppressing the glutamate-glutamine cycle-dependent GABA synthesis; and b. the suppressed GABA synthesis is mediated by IL-1β, which attenuates glutamate transporter activities by activating PKC.