The effect of bilirubin on the excitability of mitral cells in the olfactory bulb of the rat

Olfactory dysfunction is a common clinical phenomenon observed in various liver diseases. Previous studies have shown a correlation between smell disorders and bilirubin levels in patients with hepatic diseases. Bilirubin is a well-known neurotoxin; however, its effect on neurons in the main olfactory bulb (MOB), the first relay in the olfactory system, has not been examined. We investigated the effect of bilirubin (>3 μM) on mitral cells (MCs), the principal output neurons of the MOB. Bilirubin increased the frequency of spontaneous firing and the frequency but not the amplitude of spontaneous excitatory postsynaptic currents (sEPSCs). TTX completely blocked sEPSCs in almost all of the cells tested. Bilirubin activity was partially blocked by N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepro pionic acid (AMPA) receptor antagonists. Furthermore, we found that bilirubin increased the frequency of intrinsic firing independent of synaptic transmission in MCs. Our findings suggest that bilirubin enhances glutamatergic transmission and strengthens intrinsic firing independent of synaptic transmission, all of which cause hyperexcitability in MCs. Our findings provide the basis for further investigation into the mechanisms underlying olfactory dysfunction that are often observed in patients with severe liver disease.

Scientific RepoRts | 6:32872 | DOI: 10.1038/srep32872 We investigated the effect of bilirubin on MC excitability because excitatory output from these cells reflects the encoding of odor information 23 which may be impaired in patients with hepatic disease owing to high serum bilirubin levels. We used patch-clamp techniques to investigate whether bilirubin induced hyperexcitation in MCs and to examine the underlying mechanisms. We found that bilirubin significantly affected MC excitability, providing direct evidence that high bilirubin levels interfere with MC output and contribute to the olfactory dysfunction associated with liver disease.

MC properties.
In the MOB slices, MCs were identified by their morphology and position within the MC layer. MCs typically have a pyramidal soma that is larger than those of other cell types. The synaptic organization of MCs has been reported. The longest dendrite of MCs is the most extensively apical dendrite which extends to the glomerulus layer, where many of dendrites of PGs are present. GCs contact the secondary dendrites of MCs within the external plexiform layer (Fig. 1) 26 . However, because morphological characteristics alone were not sufficient to identify MCs, electrophysiological characteristics are also examined. We observed firing patterns under cell-attached conditions. All MCs showed burst firing. Some fired in bursts with short pauses between bursts while others had relatively longer pauses between bursts of firing. Figure 1c shows the electrophysiological characteristics of two representative MCs, which are consistent with those previously reported 27 .
Bilirubin increased the frequency of sEPSCs. Several factors may alter neuronal excitability, such as changes in synaptic transmission, synaptic connectivity, intrinsic neuronal properties, and intracellular signal transduction. Glutamate, the most prevalent excitatory neurotransmitter in the CNS, can produce long-lasting changes in neuronal excitability. To investigate whether glutamatergic transmission is involved in the bilirubin-induced hyperexcitation of MCs, we recorded sEPSCs in MCs with bicuculline and strychnine to block feedback inhibition from granule and periglomerular cells under voltage-clamp conditions. Because sensory-induced discharge is dependent on multistep excitation rather than direct synaptic connections from the olfactory sensory neurons (OSNs) in MCs 30 , and because MC apical dendrites are relatively longer than those of other neuron types in the MOB 31 , not all MCs have sEPSCs, and their amplitude is relatively small. We were able to record sEPSCs in 7 of 35 MCs. The amplitude and frequency of sEPSCs were 5-20 pA and 2-6 Hz, respectively. The sEPSC frequency increased within 1 min and reached a maximum level about 5 min after bilirubin application (6 μ M; Fig. 3a,c). The average sEPSC frequency during the 5-min bilirubin application was 218.1 ± 5% of the control (n = 7, p < 0.001). After washing with ACSF for 4 min, the sEPSC frequency decreased to 138.6 ± 2% of the control (n = 7, p < 0.001; Fig. 3b). However, bilirubin did not significantly alter the sEPSC amplitude (average: 105.4 ± 3% of the control in the presence of bilirubin, 99 ± 2% of the control after washing for 4 min, n = 7, ns) throughout the bilirubin application period (data shown every minute after bilirubin application; Fig. 3c). These findings suggest that glutamatergic transmission is potentiated by bilirubin in MCs. In the presence of TTX, we found that miniature EPSCs were absent in almost all of the neurons recorded, indicating these events are mainly action potential dependent (Fig. 3d, n = 10).

Bilirubin-evoked firing of MCs is partially mediated by glutamatergic transmission. Previously,
we showed that bilirubin stimulated glutamate release in lateral superior olive neurons 15 . However, no studies have investigated the role of glutamatergic neurotransmission in bilirubin-induced hyperexcitation in MCs. We found that, under current-clamp conditions, bilirubin perfusion (6 μ M) for 5-min markedly increased the MC firing frequency to 160.4 ± 7% of the control (n = 7, p < 0.01). Then, we applied AMPA and NMDA receptor antagonists during bilirubin treatment for 5-min. The results showed that the application of 50 μ M APV and 10 μ M NBQX reduced the firing frequency to 65.6 ± 4% of the control, which was significantly lower than that observed with bilirubin alone (n = 7, p < 0.001; Fig. 4a,b). Next, application of the receptor antagonists, 50 μ M l-2-amino-5-phosphonovaleri -cacid (APV) and 10 μ M 2,3-dihydroxy-6-nitro-7-sulfamoyl-ben-zoquinoxaline -2,3-dione (NBQX) reduced the cell firing frequency to 48.2 ± 3% of the control. The addition of bilirubin to the APV and NBQX solution increased the firing rate to 130.5 ± 6% of the antagonists, which was significantly Average spontaneous firing frequency before, during, and after application of bilirubin. Frequencies were averaged every minute and normalized to the mean value during the 5-min control period (n = 6). (c) Bar graphs show the average spontaneous firing frequency before, during, and after the application of bilirubin (1, 3, 6, and 10 μ M). Vertical error bars represent SE. *p < 0.05; **p < 0.01; ***p < 0.001; ns: no significant difference. different from the response to the antagonists alone (n = 5, p < 0.001; 6 μ M; Fig. 4c,d). Figure 4e shows the effects of bilirubin alone and bilirubin following pretreatment with APV and NBQX. Although bilirubin increased the firing rate in the presence of NBQX and APV, the effect was significantly suppressed compared with that under the bilirubin-alone condition. These results indicate that bilirubin-induced hyperexcitation in MCs is mediated, in part, by glutamatergic neurotransmission.
Bilirubin increased the intrinsic firing rate of MCs. Intrinsic firing patterns, which generate spontaneous rhythmic firing in the absence of synaptic input, occur in a number of neurons [32][33][34][35][36][37][38][39] . This characteristic may contribute to the function of these neurons 40 . These previous findings suggest that intrinsic firing may play a role in the bilirubin-induced hyperexcitation of MCs. To address this issue, we first recorded the spontaneous firing of MCs in the cell-attached configuration and then pharmacologically blocked synaptic-evoked activity using AMPA, NMDA, GABA A , and glycine receptor antagonists. We found that the spontaneous firing frequency was reduced to 58.9 ± 3% of the control (n = 5, p < 0.05) after the application of 50 μ M bicuculline, 3 μ M strychnine, 10 μ M NBQX, and 50 μ M APV (Fig. 5a,b). However, the remaining firing was robust and regular with bursts, just as before the synaptic blockade (Fig. 5a). These findings are similar to those of a previous study 41 . Interestingly, despite the blockade of all synaptic transmission, the cell firing rate increased to 125.1 ± 4% of all blockers during the 5-min bilirubin application, which was significantly different from the results under the blockers-alone condition (n = 5, p < 0.05; Fig. 5b).

Discussion
Excitotoxicity is a generally accepted consequence of bilirubin-induced hyperexcitation 13 , which is caused by the free fraction of unconjugated bilirubin that is not bound to plasma proteins and can freely enter the brain 42 .  7) before and during the application of bilirubin, and during APV and NBQX co-application with bilirubin. (c) Traces of spontaneous firing recorded from a cell before and during the application of APV and NBQX and subsequent co-application of bilirubin with APV and NBQX. (d) Bar graphs show the average spontaneous action potential frequency of MCs (n = 5) before and during the application APV and NBQX and co-application of bilirubin with APV and NBQX. (e) Bar graphs show the MC firing rate in response to bilirubin alone and bilirubin following pretreatment with APV and NBQX. Vertical error bars represent SE. *p < 0.05; **p < 0.01; ***p < 0.001.
The excitotoxic effects of bilirubin have been documented in various CNS areas, including the ventral cochlear nuclei, subthalamic nucleus, and hippocampus 13 . The central processing of odors is disrupted by elevated serum bilirubin levels. Therefore, we hypothesized that the administration of bilirubin would significantly interfere with the activity of MCs, the principal output neurons of the MOB, and their role in encoding odor information. MCs were identified using morphological and electrophysiological criteria, which were consistent with those described previously. We recorded spontaneous firing in MCs using the cell-attached patch-clamp technique. We found that bilirubin at a concentration of 1 μ M did not significantly increase the MC firing rate, while at concentrations of 3, 6, and 10 μ M, bilirubin significantly increased the frequency of spontaneous firing in a concentration-dependent manner, indicating that bilirubin induced hyperexcitation in MOB neurons at concentrations comparable to those of hyperbilirubinemia. Amin et al. 43 reported similar findings in auditory brainstem nuclei. The MC firing rate is associated with the encoding of odor information 23 . Therefore, our finding that bilirubin concentrations exceeding 3 μ M changed MC activity suggests a possible mechanism for odor dysfunction in patients with liver disease.
Glutamate is the primary excitatory neurotransmitter in the CNS 28,30 . Glutamate excitotoxicity is hypothesized to damage neurons via excessive synaptic release of glutamate and consequent overstimulation of glutamate receptors 44 . We previously showed that bilirubin stimulated presynaptic glutamate release, causing increased activity in lateral superior olive neurons, which is consistent with bilirubin-induced glutamate excitotoxicity 15 . The excitatory activity in the MCs is mediated by glutamate via ionotropic glutamate AMPA, and NMDA receptors 45 . Therefore, we tested the hypothesis that glutamate and glutamate receptors are involved in the bilirubin-induced hyperexcitation of MCs. We found that bilirubin increased the frequency but not the amplitude of glutamatergic sEPSCs. These findings suggest that bilirubin can potentiate glutamatergic synaptic transmission without influencing the sensitivity of postsynaptic glutamate receptors. Glutamatergic EPSCs can originate either olfactory nerve axons or the dendrodendritic release of glutamate from other mitral/tufted cells. To clarify whether the action of bilirubin depends on action-potential, 1 μ M TTX was added to ACSF. This showed that sEPSCs were blocked completely in almost all of the cells tested. Furthermore, the bilirubin-induced neuronal hyperexcitation was not completely blocked by glutamate receptor antagonists in our study (Fig. 4a,b), implying that the increase in glutamatergic neurotransmission is not the only mediator of bilirubin-induced hyperexcitation in MCs. It is possible that bilirubin also increases firing by acting on the ion channels in the soma of mitral/tufted cells leading to increase in firing and subsequent increase in release of glutamate from their dendrites. Our findings support the notion that enhanced glutamatergic neurotransmission is associated with the abnormal coding of olfactory information 46 , which may occur in patients with hepatic encephalopathy, a severe liver disease 47 .
Intrinsic firing independent of synaptic transmission has been detected in neurons in several areas of the CNS. This phenomenon depends on intrinsic membrane properties and has been shown to play an essential role in the refinement of neural connections, particularly during development 8,48,49 . To determine whether MCs exhibited intrinsic firing patterns and whether they were affected by bilirubin, we recorded neuronal firing in the presence of excitatory and inhibitory synaptic receptor inhibitors. We found that the MCs exhibited intrinsic rhythmic firing. Moreover, bilirubin increased the frequency of intrinsic firing. Balu et al. 50 hypothesized that the intrinsic membrane properties that drive intermittent spike clusters in MCs are important for the precise timing of the spikes evoked by odors for the encoding of olfactory information. MC firing activity is perceived as encoded odor messages. Thus, our finding that bilirubin significantly increased the frequency of intrinsic firing in MCs suggests that it may play a key role in olfactory function as a biochemical modulator. Intrinsic firing in MCs is controlled by several ion channels (e.g., HCN, 4-AP-sensitive K-currents, I D -like currents, and TTX-sensitive Na-current) 50 . The exact biophysical mechanisms involved in intrinsic MC activity are not known. Moreover, those ion channels controlling intrinsic MC firing that are affected by bilirubin remain to be identified.
In conclusion, our findings show that bilirubin facilitates glutamatergic synaptic transmission and intrinsic firing in MCs, which may contribute to bilirubin-induced hyperexcitation. Our results suggest a possible cellular mechanism for explaining the loss of the ability to identify odors experienced by patients with liver disease. Furthermore, our findings provide the basis for further research into the prevention and treatment of olfactory dysfunction in patients with liver disease.

Materials and Methods
Our study was approved by the Ethics Review Committee for Animal Experimentation at Shanghai Jiao Tong University. The experiments were performed in accordance with the Guiding Principles for the Care and Use of Animals, and all efforts were made to minimize the possible pain and discomfort of the animals during the experiments.
Reagents. The reagents used in the experiments included free bilirubin, bicuculline, strychnine, NBQX, and APV and TTX (all from Sigma, St. Louis, MO, USA). As a stock solution, bilirubin was dissolved in 0.1 M NaOH at 1 mM, stored in single-use aliquots in the dark at − 20 °C (< 48 h), and diluted to the final solution concentration before use. The bilirubin solution was protected from light at all times. Bicuculline and APV were prepared as stock solutions using 100% dimethyl sulfoxide (DMSO; Sigma) and diluted to the required concentration in ACSF immediately before use, resulting in a maximum DMSO concentration of < 0.1%. Other drugs were prepared as stock solutions in distilled water and diluted to the required concentrations in ACSF immediately before use. All drugs were applied locally via a square glass capillary (0.4 mm in width, Cat. #64-0121, Warner Instrument, New Haven, CT, USA), which was placed on the slice as close to the MC as possible.
Electrophysiological measurements and data analyses. Cell-attached and whole-cell recordings were performed using a patch-clamp amplifier (EPC10; HEKA, Elektronik, Lambrecht, Germany). The electrode capacitance and liquid-junction potential were compensated. Data were filtered at 1-3 kHz and sampled at 3-10 kHz using a Dell computer equipped with PATCHMASTER software (HEKA, Elektronik). The patch pipettes were pulled from borosilicate capillary glass through two stages using a vertical pipette puller (P-9; Narishige, Tokyo, Japan). The resistance of the electrode was 3-7 MΩ . Patch electrodes were filled with the following solutions for cell-attached recordings (in mM): 97.5 K-gluconate, 32.5 KCl, 0.5 ethyleneglycol-bi s(2-aminoethyl)-N,N,N′,N′,-tetraacetic acid (EGTA), 40 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), and 1 MgCl 2 , 290-300 mOsm, pH 7.3; they were filled with the following solutions for whole-cell recordings (in mM): 120 K-gluconate, 5 KCl, 2 MgCl 2 , 0.05 EGTA, 10 HEPES, 2 Mg-ATP, 0.4 Mg-GTP, 10 creatine phosphate, at 290-300 mOsm and pH 7.3. Some of the recordings used to assess spontaneous excitatory postsynaptic currents (sEPSCs) in MCs were performed using low Cl − based with cesium in the pipette solution (in mM): 127 cesium methanesulfonate, 20 NaCl, 20 HEPES, 0.4 EGTA, 5 TEA-Cl, 3 QX314-Cl, 2.5 MgATP and 0.33 NaGTP at 300-310 mOsm, and pH 7.3. We conducted one recording per brain slice. All experiments were performed at room temperature (24-26 °C). Differences in the frequency of action potential currents and sEPSCs were analyzed, using Wilcoxon signed-ranks tests for differences in the frequency of action potential currents and sEPSCs intra-group comparisons, using Mann-Whitney U-test for independent concentration groups comparisons. The Statistical Package for the Social Sciences version 17.0 (SPSS, Chicago, IL, USA) was used to conduct the statistical tests. The results are expressed as means ± standard errors (SEs), and p-values < 0.05 were deemed to indicate statistical significance.