Rundown of GABA type A receptors is a dysfunction associated with human drug-resistant mesial temporal lobe epilepsy

Abstract

Pharmacotherapeutic strategies have been difficult to develop for several forms of temporal lobe epilepsy, which are consequently treated by surgical resection. To examine this problem, we have studied the properties of transmitter receptors of tissues removed during surgical treatment. We find that when cell membranes, isolated from the temporal neocortex of patients afflicted with drug-resistant mesial temporal lobe epilepsy (TLE), are injected into frog oocytes they acquire GABA type A receptors (GABA_A-receptors) that display a marked rundown during repetitive applications of GABA. In contrast, GABA_A-receptor function is stable in oocytes injected with cell membranes isolated from the temporal lobe of TLE patients afflicted with neoplastic, dysgenetic, traumatic, or ischemic temporal lesions (lesional TLE, LTLE). Use-dependent GABA_A-receptor rundown is also found in the pyramidal neurons of TLE neocortical slices and is antagonized by BDNF. Pyramidal neurons in cortical slices of a traumatic LTLE patient did not show GABA_A-receptor rundown. However, the apparent affinity of GABA_A-receptor in oocytes microtransplanted with membranes from all of the epileptic patients studied was smaller than the affinity of receptors transplanted from the nonepileptic brain. We conclude that the use-dependent rundown of neocortical GABA_A-receptor represents a TLE-specific dysfunction, whereas the reduced affinity may be a general feature of brains of both TLE and LTLE patients, and we speculate that our findings may help to develop new treatments for TLE and LTLE.

Epilepsy is a brain disorder affecting ≈1% of the world population. Although many forms of epilepsy can be medically suppressed, a small percentage of epileptic patients experience intractable seizures. To date, the surgical treatment offered to patients with intractable epilepsy has generally yielded beneficial outcomes. Postoperative seizure-free patients score better than the best medically treated patients on a variety of measures designed to assess the impact of epilepsy surgery on health and quality of life. Although surgery is a beneficial procedure, one of the main challenges for both basic and clinical research of intractable epilepsy is to discover effective drugs to treat dysfunctions related to epileptogenesis.

Of all the surgically remediable temporal lobe (TL) epilepsy syndromes, that with mesial temporal sclerosis [mesial TL epilepsy (TLE)] and lesional TL epilepsy (LTLE) with neoplastic, dysgenetic, traumatic, or ischemic temporal lesions are the forms of focal epilepsy that more frequently undergo resective surgery. In the last years, intractable TLE has been studied through the analyses of surgically resected TL and mesial structures by using molecular and electrophysiological approaches. This has led to the discovery of alterations of excitatory and inhibitory amino acid receptor binding and of their transporter systems (1–4), altered expression of neurotransmitter receptors (5–7) and of GABA inhibitory function (8–11), and overexpression of BDNF (12). Interestingly, GABA type A receptors (GABA_A-receptors) originating from the TLE temporal neocortex exhibit a surprisingly low apparent affinity for GABA and a marked current elicited by GABA (GABA_A-current) rundown upon repetitive stimulation with GABA. This rundown is strongly reduced by phosphatase inhibitors (10) and by BDNF (13). Thus GABA_A-receptor instability (GABA_A-current rundown) may be a dysfunction of the GABAergic system that could be treated with exogenous (phosphatase inhibitors) (10) or endogenous (BDNF) substances (13). If proven to be highly related to an epileptic syndrome, treating the GABA_A-current rundown could be a promising route for devising medical alternatives to surgical resection.

To address that issue directly in human TLE, we have used two experimental approaches. In the first, we have microtransplanted preassembled GABA_A-receptors from the human brain to the plasma membrane of Xenopus oocytes, by injecting the oocytes with cell membranes prepared from surgically resected epileptic nervous tissues (14–16). Our second approach has been to record whole-cell GABA_A-currents from neurons in slices from TL neocortex of patients with TLE. Using these approaches we have analyzed the behavior of GABA_A-receptors transplanted to oocytes or receptors expressed in native nervous tissue, from the same epileptic patients. We find that the GABA_A-current rundown differs between TLE and LTLE receptors, whereas the GABA_A-receptor affinity is similar for TLE and LTLE receptors, and both are lower than that of receptors from the nonepileptic brain.

Materials and Methods

Patients. Surgically removed tissue specimens were obtained from 14 patients with drug-resistant TLE, who were operated on at the Neuromed Neurosurgery Center for Epilepsy (Pozzilli, Italy) (Table 1). Informed consent to use part of the biopsy material for our experiments was obtained from all patients, and the Ethics Committees at the Neuromed Neurosurgery Center and at the University of Rome “La Sapienza” approved the selection procedures. The resective surgery in seven TLE patients was the “standard” anteromedial selective amygdalohippocampectomy. The histopathology showed the typical neuropathological features of Ammon's horn sclerosis in five of seven patients, with no sclerosis in the TL. The remaining two patients showed no sclerosis throughout the resected tissues. The resective surgery in three of seven patients afflicted with lesional TL epilepsy was the total resection of the TL lesion along with adjacent epileptogenic tissue (lesionectomy plus), whereas in the remaining four patients, mesial structures were not spared. Two additional patients were also studied: one (patient 15 in Table 1) was afflicted with medically intractable epilepsy, with the TL showing neither epileptic foci nor seizure propagation; and two other patients (patients 16 and 17 in Table 1) were nonepileptic but afflicted with a TL tumor. For these reasons, the resected tissues (neocortices) were considered as “control.” The neurophysiological assessment, including presurgical diagnostic protocol, neuroradiological evaluation, long-term intensive video-electroencephalogram monitoring, and the resective surgery criteria adopted, is in Table 2, which is published as supporting information on the PNAS web site. All tissue specimens were (i) frozen in liquid nitrogen and processed for membrane preparation or RNA extraction and (ii) used for neuropathology. Portions of the TL specimens from the same tissues to be analyzed with patch–clamp techniques were placed in an oxygenated chamber with sucrose-based artificial cerebrospinal fluid (ACSF) (see below) for preparing brain slices.

Table 1. Clinical characteristics and neurophysiological findings in patients with temporal lobe epilepsy.

Patient	Sex	Age, yrs	Epilepsy onset, yrs	MRI	Epileptogenic zone	Surgery	Histopathology	Seizure outcome	Followup, months
1	M	31	6	RTMes	RTMeslat	ETL	None	1a	18
2	M	36	9	RTMes	RTMesLat	ETL	None	2a	26
3	M	19	6	LTMes	LTMesLat	ETL	MES	1a	1
4	M	34	4	LTMes	LTMesLat	ETL	MES	1a	2
5	F	39	28	LTMes	LTMesLat	ETL	MES	1a	6
6	M	42	10	LTMes	LTMes	AMTL	MES	1a	18
7	F	39	12	RTMes	RTMes	AMTL	MES	1a	26
8	F	34	8	RTMesLat	RTMesLat	ETL	Ischemic lesion	1a	21
9	M	8	6	RTMes	RTMes	Lesionectomy plus	DNET	1a	12
10	M	7	1	RTLat	RTMesLat	Lesionectomy plus ETL	FCD	2a	14
11	F	38	19	RTMesBas	RTMesLat	ETL	FCD	1a	24
12	F	34	Birth	RTMes	RTMes	Lesionectomy + AMTL	Ganglioglioma (I WHO)	1a	24
13	M	16	11	LTLat	LTLat	Lesionectomy plus	Ganglioglioma (III WHO)	1a	3
14	M	29	13	RTMes	RTMes	Lesionectomy plus	Cranial trauma	1a	3
15	F	37	22	LTMes	LTMes	Amygdalohyppocamectomy	MES	1a	2
16	M	66	Absent	LT-insular lesion	None	Lesionectomy plus	Glioblastoma (IV WHO)	-	-
17	F	27	Absent	LTMes	None	Lesionectomy plus	Oligodendroglioma (II WHO)	-	-

Patients 1-7, afflicted with TLE; patients 8-14, afflicted with LTLE; patient 15, afflicted with focal hippocampal epilepsy with seizure-free TL; patients 16 and 17, afflicted with nonepileptic tumor. Tumor grading was according to the World Health Organization (WHO) histological classification. Seizure outcome: 1a (Engel class), completely seizure-free since surgery; 2a (Engel class), initially free of disabling seizures and only rare seizures at present. R, right; L, left; T, temporal; Lat, lateral; Bas, basolateral; MES, hippocampal mesial selerosis; ETL, extensive temporal lobectomy; AMTL, anteromesial temporal lobectomy; DNET, dysembryoplastic neuroepithelial tumors; FCD, focal cortical dysplasia; M, male; F, female.

Membrane Preparation, Injection Procedure, and Electrophysiological Recording from Oocytes. Membranes from human nerve tissue were prepared and injected into oocytes with procedures as described (14, 15). GABA_A-currents were elicited by GABA at a concentration of 1 mM (10, 11, 16). GABA-current rundown was defined as the decrease of the peak current amplitude after six consecutive applications of GABA (10 s duration, 40 s interval). Dose–current response relationships were estimated by fitting the data to a Hill equation (10, 11, 16). Further details are in Supporting Text, which is published as supporting information on the PNAS web site.

Whole-Cell Recording from TL Slices. Neocortical slices were prepared as shown (17) from a 1-cm³ block of surgically resected TL neocortex. The tissue was placed immediately in low-Na⁺ ice-cold oxygenated (95% O₂/5% CO₂) transport solution (sucrose-based ACSF) and transported in an airtight cooled container filled with chilled oxygenated transport solution. Upon arriving at the laboratory 90–120 min later, 250- to 300-μm transverse slices were cut in transport solution with a vibratome (DTK-1000, Ted Pella, Redding, CA). Slices were placed in a slice incubation chamber at room temperature with oxygenated ACSF and transferred to a recording chamber within 2–24 h after slice preparation. Whole-cell patch–clamp recordings were performed at 24–25°C, as described for pyramidal neurons (10, 18). Membrane currents were recorded by using glass electrodes (3–4 MΩ) filled with 130 mM CsCl, 10 mM Hepes, 5 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetate, and 2 mM Mg-ATP (pH 7.3, with CsOH). GABA (100 μM) was delivered to cells by short pressure applications (10–20 psi; 5–50 ms; Picospritzer II, General Valve, Fairfield, NJ) from glass micropipettes positioned above whole-cell voltage-clamped neurons. In this way, we obtained stable whole-cell currents and rapid drug wash before applying the rundown protocol. The current rundown protocol adopted was the following (10): after current amplitude stabilization with applications every 120 s, a sequence of 10 applications of GABA once every 15 s were delivered, after which the control rate (every 120 s) was resumed to monitor the recovery. The reduction in peak amplitude current was expressed as percentage amplitude of current at the end of the rundown protocol vs. control (% I_tenth/I_first).

Chemicals and Solutions. The oocyte Ringer solution had the following composition: 82.5 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl₂, 1 mM MgCl₂, and 5 mM Hepes, adjusted to pH 7.4, with NaOH at 24–25°C. Except when indicated, all drugs were purchased from Sigma. ACSF contained 125 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 1.25 mM NaH₂PO₄, 1 mM MgCl₂, 26 mM NaHCO₃, and 10 mM glucose (pH 7.35). Transport sucrose-based ACSF contained 3 mM KCl, 2 mM MgCl₂, 1.6 mM CaCl₂, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 0.4 mM ascorbic acid, 26 mM NaHCO₃, 10 mM glucose, and 200 mM sucrose (pH 7.35).

Statistics. The results were expressed as means ± SEM. One-way ANOVA was used to examine statistical significance. Values were considered significantly different when P < 0.01.

Results

Rundown of GABA_A-Receptors Transplanted to Oocytes from the TL Neocortex of TLE Patients. After injecting membranes isolated from specimens of epileptic TL, we determined the properties of GABA_A-receptors microtransplanted from both glial and nerve cells to the oocyte plasma membrane. In agreement with previous experiments (10, 11, 16), applications of GABA (1 mM) to oocytes injected with membranes from seven TLE patients elicited bicuculline-sensitive (100 μM) GABA_A-receptor-mediated currents of variable amplitudes, some as large as -450 nA, others as low as -20 nA (mean = -212 ± 40 nA, 63 oocytes from 11 frogs, 63/11) depending on oocytes, patients, or frogs. These currents exhibited a considerable rundown after repetitive GABA applications, the GABA_A-current falling to 26.8 ± 3% of control (range, 13–38%; 63/10; patients 1–7 in Table 1), with partial recovery after a few minutes of washout (Fig. 1). Full recovery was attained 8–12 h after the rundown protocol (not shown). In contrast, much less rundown occurred in oocytes injected with membranes derived from the seizure-free TL of patient 15 afflicted with hippocampal focal epilepsy (see Fig. 1 and Table 1). These findings suggest that the increased GABA_A-current rundown is related to seizure development and represents a GABA_A-receptor dysfunction associated with TLE.

Fig. 1.

GABA_A-receptor rundown in oocytes injected with TLE membranes. (Upper) Sample current responses to GABA applied (horizontal bars) before the rundown protocol (1st), at the end of the rundown protocol (6th), and during recovery (4 min after the end of the rundown protocol) from an oocyte injected with membranes isolated from the TL neocortex of TLE patient 3 (Table 1). (Lower) Time course of current rundown in oocytes injected with TL membranes from seven TLE patients (circles; patients 1–7 in Table 1; 63/11) and from one patient with seizure-free TL (squares; patient 15 in Table 1; 8/2). Here and in Figs. 2, 3, 4, 5, filled symbols represent mean current amplitudes (±SEM) before and after the rundown protocol; open symbols, during rundown protocol; all normalized to the current (*) just before rundown. The dashed line indicates the 100% level.

To investigate whether the enhanced GABA_A-current rundown is restricted to TLE or extends also to LTLE, oocytes were injected with membranes isolated from the TL neocortex of seven LTLE patients (patients 8–14 in Table 1), and the rundown of the microtransplanted GABA_A-receptors was determined. For all these patients, the rundown was small and no different from that of two nonepileptic brain tissues (Fig. 2). The LTLE patients were afflicted with dysgenetic (two patients), tumor (three patients), traumatic (one patient), and ischemic (one patient) temporal lesions. After repetitive applications of GABA, the currents (-183 ± 44 nA; 57/7) fell to 76 ± 3% (range, 68–92%), a rundown comparable to that of oocytes injected with membranes from nonepileptic TL cortex (“control”) obtained as margins of cerebral tumors (see Fig. 2; patients 16 and 17 in Table 1).

Fig. 2.

Lack of GABA current rundown in oocytes injected with LTLE membranes. Oocytes were injected with membranes isolated from the TL neocortex of patients afflicted with LTLE (triangles; patients 8–14 in Table 1; 57/7) or with nonepileptic TL tumor (squares; patients 16 and 17 in Table 1; 18/4).

Rundown of GABA_A-Receptors in the Pyramidal Neurons. Because the microtransplanted membranes originated from both glial and nerve cells, the GABA_A-currents recorded in oocytes were not caused exclusively by activation of neuronal GABA_A-receptors. To overcome this problem, recordings were performed directly from pyramidal neurons in human brain slices obtained from the same TLE patients (patients 1–7 in Table 1) as those analyzed in oocytes (Fig. 1). GABA (100 μM), applied to pyramidal neurons via a puffer pipette positioned above the TLE brain slice, elicited a rapid inward current due to activation of postsynaptic GABA_A-receptors. Stable inward currents ranging from -150 pA to -700 pA (-362 ± 52 pA; n = 13) were obtained with low-frequency (0.5 per min) applications. In contrast, applications of GABA at four per minute caused a progressive decrease in current amplitude. At the end of such GABA applications (rundown protocol; see Materials and Methods), the current fell to 66 ± 3% (P < 0.01) in 13 neurons examined. Within 8 min after the rundown protocol, the GABA current returned to 93 ± 7% of the control. In contrast, pyramidal neurons in TL neocortical slices of patient 15 (see above and Table 1) displayed GABA_A-currents that remained stable during the rundown protocol, as seen in oocytes injected with TL membranes from the same patient 15 (Table 1). Fig. 3A summarizes the results obtained from the seven TLE patients and TL seizure-free patient 15 (Table 1). The absence of GABA_A-current rundown in five pyramidal neurons in TL slices of a patient with traumatic lesion (patient 14 in Table 1) illustrated in Fig. 3B, a unique LTLE patient from which we had the opportunity to perform a full analysis with the patch–clamp techniques, provides an additional control.

Fig. 3.

GABA_A-receptor rundown in pyramidal neurons from epilepsy patients. (A Upper) Sample currents elicited by repetitive GABA applications (triangles) to a pyramidal neuron in a TL slice of TLE patient 2 (Table 1) before the rundown protocol (1st), at the end of the rundown protocol (10th), and after recovery (6 min after the 10th stimulus). (A Lower) Averaged time course of GABA current rundown and recovery from 13 pyramidal neurons of the seven TLE patients (circles; patients 1–7) and lack of rundown in three TL pyramidal neurons of a patient with seizure-free neocortex (squares; patient 15, Table 1). (B) Time course of GABA current amplitude from five pyramidal neurons in TL slices of a TLE patient afflicted with traumatic lesion (triangles; patient 14, Table 1). (Inset) Sample currents elicited by repetitive GABA applications to a pyramidal neuron from this patient.

BDNF Decreases GABA_A-Current Rundown in TL Pyramidal Neurons of TLE Patients. We reported previously that the increased rundown of epileptic GABA_A-receptors transplanted to oocytes is greatly reduced by BDNF (13). To determine whether BDNF reduces the GABA_A-current rundown in TLE pyramidal neurons as well as in the expression system, experiments were performed on TL slices from three TLE patients (patients 4–6 in Table 1). GABA_A-currents (-103 ± 35 pA; n = 5) fell to 71 ± 3% (range: 65–78%; five pyramidal neurons) after repetitive GABA applications, in agreement with findings reported in Fig. 3A. Although there was no obvious effect on the GABA_A-current rundown when BDNF was acutely applied (not shown), a 15-min preincubation of epileptic slices with 200 ng/ml BDNF significantly reduced the rundown to 89 ± 4% (range: 83–97%; four pyramidal neurons; P < 0.01). Fig. 4 illustrates the BDNF-induced reduction of GABA_A-current rundown in a pyramidal neuron (patient 5). These findings show that the effect of BDNF on GABA_A-current rundown, first discovered in oocytes microtransplanted with membranes from TLE patients, occurs also in native pyramidal neurons from TLE patients.

Fig. 4.

BDNF-induced inhibition of GABA_A-current rundown in TLE pyramidal neurons. The records are representative sample currents elicited by GABA applied to a pyramidal neuron before (1st), at the end of the rundown protocol (10^th), and after 6 min recovery. The graphs show rundown of GABA_A-currents from the same neuron in control conditions, after 15 min of BDNF treatment and after 20 min wash. GABA_A-current amplitudes normalized to the current (*) just before the rundown: 180 pA (control), 188 pA (BDNF), and 189 pA (wash). TL slices from patient 4 (Table 1).

GABA_A-Receptor Affinity Is Similar in All Epileptic Patients. We have reported that TLE GABA_A-receptors microtransplanted from resected brain tissues into oocytes show a smaller apparent affinity to GABA than that of nonepileptic GABA_A-receptors (10, 13). To determine whether the affinity differs between TLE (patients 1–7 in Table 1) and LTLE patients (patients 8–14 in Table 1), GABA dose–current response relations were constructed from oocytes injected with membranes extracted from the TL neocortex of each patient. The membrane currents elicited by different concentrations of GABA yielded GABA dose–current response relations with similar mean EC₅₀ (half-maximal concentration) values, ranging from 106 to 126 μM, and mean Hill number (n_H) values, ranging from 0.9 to 2 (EC₅₀ = 120 ± 2 μM, n_H = 1.26 ± 0.06, 36/3, patients 1–7 in Table 1; EC₅₀ = 112 ± 4 μM, n_H = 1.2 ± 0.1, 39/3, patients 8–14 in Table 1; P > 0.05) (Fig. 5). In contrast, the EC₅₀ value of oocytes injected with TL membranes derived from margins of two nonepileptic tumors (patients 16 and 17 in Table 1) was significantly smaller than that of the epileptic patients (EC₅₀ = 36.7 ± 1.5 μM; n_H = 1.2 ± 0.1; 16/3; P < 0.01) (Fig. 5), in agreement with previous observations (11). Thus, these findings indicate that, in contrast to the different GABA_A-current rundown, the apparent affinity for GABA is significantly similar for TLE and LTLE patients, and this in turn is significantly different from that of nonepileptic patients.

Fig. 5.

Decreased GABA_A-receptor affinity in epileptic vs. nonepileptic patients. GABA dose–current relationships in oocytes injected with membranes from (○) TLE patients (patients 1–7 in Table 1; 36/3), (▾) LTLE patients (patients 8–14 in Table 1; 39/3), and (□) two nonepileptic TL tissues resected as tumor margins (patients 16 and 17 in Table 1; 16/3). Peak currents were normalized to the current obtained with 2 mM GABA.

Discussion

A general view is that seizures are generated by a failure of GABA-mediated inhibition, and that diminished GABAergic function plays a pivotal role in epileptogenesis (19). For example, in animal models, it has been postulated that the decreased inhibition during status epilepticus is due, at least partly, to alterations of postsynaptic GABA_A-receptors caused by the seizures (20–22). In humans, many forms of epilepsies and epileptic syndromes are susceptible to antiepileptic medications that affect GABAergic function. Nevertheless, despite an optimal use of antiepileptic drugs, at least 25% of epilepsy patients do not gain complete seizure control. The most frequently evaluated epilepsy syndrome is focal refractory TLE, which is often the result of mesial temporal sclerosis, sometimes of temporal lesions. It is noteworthy that a significant number of patients afflicted with TLE benefit from resective epilepsy. However, discovery of medicinal alternatives to surgery is a main goal for developing better treatments of refractory epilepsy. Therefore, the principal aim of this work was to identify properties of GABA_A-receptors that are altered in refractory epilepsy, and that could be targets for medical antiepileptic treatments.

We reported previously (10) that “epileptic” GABA_A-receptors microtransplanted from TLE patients to oocytes display an increased GABA_A-current rundown, and that this putative epileptogenic dysfunction can be reversed by inhibiting phosphatase activity or by stimulating tyrosine kinase receptor B, TrkB, with BDNF (13). Here, we report that the native GABA_A-receptors present in TLE pyramidal neurons display a similar rundown (11). This rundown is related to the occurrence of seizures in the TL neocortex, because it was absent in membranes from patient 15 (Table 1; see Fig. 3), and this rundown was significantly reduced by treating the brain slices with BDNF. It is not yet clear why GABA_A-current rundown develops during repetitive receptor activation. Because the GABA_A-current recovers fully several hours after the rundown, internalization of receptors and recycling back to the cell surface (23) are worthy of study. The possible clinical improvement of GABA_A-receptor stability by exposing neocortical pyramidal neurons to BDNF requires further investigation, because this may help design novel treatments for intractable TL epilepsy.

Lesions of various types (neoplastic, ischemic, dysgenetic, or traumatic) are identified in 20–30% of patients with intractable epilepsy. We found that, in contrast to GABA_A-receptors transplanted into oocytes from TL epileptic neocortex, GABA_A-receptors transplanted from the epileptic neocortex of patients with TL lesions do not exhibit increased rundown. Our findings thus indicate that altered GABA_A-receptor stability during repetitive activation by GABA is a use-dependent dysfunction of focal epilepsy specifically coupled to TLE and not associated with LTLE.

We have shown previously that the efficacy of GABA in TLE is altered due to a reduced receptor affinity for GABA (11). We confirm here the reduced affinity of GABA_A-receptors throughout TLE tissues (10, 13) and show further that the affinity for GABA is also reduced in receptors microtransplanted to oocytes from the TL of LTLE patients, with lesions of various types. This reduced affinity would result in a diminished inhibitory action of GABA and thus lead to the development of seizures. This view is strongly supported by our previous finding that epileptic GABA_A-receptors display the lowest affinity for GABA in the subiculum, which is the origin of spontaneous field potential discharges resembling those in the electroencephalogram of TLE patients (24).

Conclusion

Although GABAergic interneurons and synapses are heterogeneous, and epileptogenesis cannot be ascribed to a single GABA signaling mechanism, this work describes a use-dependent dysfunction of GABA_A-receptors specific for intractable TL epilepsy without temporal lesions, which is partially antagonized by BDNF. In addition to the GABA_A-receptor rundown, we have discovered that another GABA_A-receptor dysfunction, the relatively low affinity for GABA, is similar in both TLE and LTLE. Such dysfunctions could be targets for treatments of medically refractory focal epilepsy.