SEARCH

SEARCH BY CITATION

Keywords:

  • Brain tumor;
  • Chloride homeostasis;
  • Epilepsy;
  • GABAA

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Purpose:  Several factors contribute to epileptogenesis in patients with brain tumors, including reduced γ-aminobutyric acid (GABA)ergic inhibition. In particular, changes in Cl homeostasis in peritumoral microenvironment, together with alterations of metabolism, are key processes leading to epileptogenesis in patients afflicted by glioma. It has been recently proposed that alterations of Cl homeostasis could be involved in tumor cell migration and metastasis formation. In neurons, the regulation of intracellular Cl concentration ([Cl]i) is mediated by NKCC1 and KCC2 transporters: NKCC1 increases while KCC2 decreases [Cl]i. Experiments were thus designed to investigate whether, in human epileptic peritumoral cortex, alterations in the balance of NKCC1 and KCC2 activity may decrease the hyperpolarizing effects of GABA, thereby contributing to epileptogenesis in human brain tumors.

Methods:  Membranes from peritumoral cortical tissues of epileptic patients afflicted by gliomas (from II to IV WHO grade) and from cortical tissues of nonepileptic patients were injected into Xenopus oocytes leading to the incorporation of functional GABAA receptors. The GABA-evoked currents were recorded using standard two-microelectrode voltage-clamp technique. In addition, immunoblot analysis and immunohistochemical staining were carried out on membranes and tissues from the same patients.

Key Findings:  We found that in oocytes injected with epileptic peritumoral cerebral cortex, the GABA-evoked currents had a more depolarized reversal potential (EGABA) compared to those from nonepileptic healthy cortex. This difference of EGABA was abolished by the NKCC1 blocker bumetanide or unblocking of KCC2 with the Zn2+ chelator TPEN. Moreover, Western blot analysis revealed an increased expression of NKCC1, and more modestly, of KCC2 transporters in epileptic peritumoral tissues compared to nonepileptic control tissues. In addition, NKCC1 immunoreactivity was strongly increased in peritumoral cortex with respect to nonepileptic cortex, with a prominent expression in neuronal cells.

Significance:  We report that the positive shift of EGABA in epileptic peritumoral human cortex is due to an altered expression of NKCC1 and KCC2, perturbing Cl homeostasis, which might lead to a consequent reduction in GABAergic inhibition. These findings point to a key role of Cl transporters KCC2 and NKCC1 in tumor-related epilepsy, suggesting a more specific drug therapy and surgical approaches for the epileptic patients afflicted by brain tumors.

Epilepsy is common in patients with brain tumors, with frequency depending on tumor type (Hauser et al., 1993) being often an initial presenting symptom (Van Breemen et al., 2007; Sontheimer, 2008). The pathophysiologic mechanism underlying tumor-related seizures is multifactorial, being correlated with tumor histology and peritumoral microenvironment (reviewed in Shamji et al., 2009).

Epilepsies have often been linked to alterations of γ-aminobutyric acid (GABA)ergic functionality (Gibbs et al., 1996; Huberfeld et al., 2007). GABAA receptors are anionic ligand-gated channels, mostly permeable to Cl ions (see for review Kaila, 1994), the permeability to HCO3 being only 18% of Cl permeability (Bormann et al., 1987; Kaila et al., 1993).

In the majority of mature neurons, activation of GABAA receptors leads to stabilization of membrane potential due to a low intracellular chloride concentration ([Cl]i) that sets the equilibrium potential for Cl ions (ECl) at negative potentials. In contrast, in immature neurons [Cl]i is high, resulting in a depolarized ECl as reported in Ben-Ari, 2002. The regulation of chloride homeostasis is mediated by NKCC1 and KCC2 cotransporters; NKCC1 intrudes Cl, leading to high [Cl]i, whereas KCC2 causes an efflux of chloride (Spitzer, 2010). It has been recently proposed that NKCC1 could be involved in maintaining a high [Cl]i in glioma cells (Ernest et al., 2005; Sontheimer, 2008) to support cell shrinkage, motility, and the formation of brain metastasis. Furthermore, recent studies (Sontheimer, 2008; Shamji et al., 2009) suggest that glioma invasion into the healthy parenchyma is at least in part mediated by the activity of Cl channels (Habela et al., 2009). Despite this evidence, to date there are no studies on human subjects leading to a clear understanding of the molecular mechanisms of tumoral epileptogenesis. As previously shown, a useful method to understand the role of GABA in epilepsy is the incorporation of GABAA receptors obtained from human brain into Xenopus oocytes (Miledi et al., 2006; Eusebi et al., 2009). The advantage of this powerful technique is the possibility to investigate human GABAA receptors using the minimal amount of tissue surgically resected from epileptic patients that can be spared after histologic analysis for diagnostic purposes (Miledi et al., 2002; Palma et al., 2004). In this study we injected membranes isolated from peritumoral epileptic tissues of patients afflicted by gliomas (from II to IV WHO grade) to measure the value of EGABA. Our study was aimed at determining whether EGABA is altered in these patients, to obtain evidence supporting the hypothesis that impaired [Cl]i homeostasis contributes to epileptogenesis in brain tumors as previously reported for the hippocampal subiculum of patients with drug-resistant epilepsy (Cohen et al., 2002; Palma et al., 2005, 2006; Muñoz et al.,2007). We found that the human epileptic peritumoral cortex has higher expression of Cl transporters, NKCC1, and KCC2 and a depolarized EGABA when compared to nonepileptic cortex. These findings identify a potentially important mechanism that may result in a decreased inhibitory efficacy of GABA, a feature that may contribute to the precipitation of epileptic seizures in human brain tumors.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Patients

The cases included in this study (Table 1) were obtained from the files of the Departments of Neuropathology of the Academic Medical Center (University of Amsterdam) and the University Medical Center in Utrecht (UMCU). For electrophysiology we examined six peritumoral and tumoral specimens removed from epileptic patients afflicted by glial brain tumors (Patients 1–6, Table 1); four histologic normal specimens were from nonepileptic patients afflicted by meningioma (Patients 7–10, Table 1). To increase control experiments, two autoptic specimens from healthy subjects (no. 11–12) were also used. For more details see Table 1 and Supporting Information.

Table 1.   Clinical characteristics and neurophysiologic findings of patients
Patient no. (WHO)GenderAge (years)Epileptogenic zoneSurgeryMean duration of epilepsyHistopathology
  1. Patient 1: astrocytoma II WHO (temporal cortex peritumoral epileptic tissue); Patients 2–5: glioblastoma (temporal cortex peritumoral epileptic tissue); Patient 6: astrocytoma III WHO; Patients 7–10: meningioma III WHO (temporal cortex).

  2. aPatient tissues used for electrophysiology.

  3. bPatient tissues used for immunohistochemistry.

  4. cPatient tissues used for Western blot experiments.

  5. R, right; L, left; T, temporal; ETL, extensive temporal lobectomy; LES, lesionectomy; tumor grading was according to the histological classification (WHO); GBM, glioblastoma; GG, ganglioglioma; MTLE, mesial temporal lobe epilepsy.

 1a,b,cM47R-TETL13 monthsAstrocytoma (II)
 2a,b,cM67R-TETL5 monthsGBM (IV)
 3a,b,cF66L-TETL3 monthsGBM (IV)
 4a,b,cM55R-TETL4 monthsGBM (IV)
 5a,b,cF45L-TETL4 monthsGBM (IV)
 6a,b,cM52R-TETL8 monthsAstrocytoma (III)
 7a,cM50NonepilepticLES
 8a,cF52NonepilepticLES
 9a,cF47NonepilepticLES
10a,cM32NonepilepticLES
11a,bM45NonepilepticAutopsy
12a,bF67NonepilepticAutopsy
13bM55NonepilepticAutopsy
14bM64NonepilepticAutopsy
15bF42NonepilepticAutopsy
16bF29NonepilepticAutopsy
17bF21R-TETL6 yearsGG (I)
18bM23R-TETL3 yearsGG (I)
19bF41L-TETL4 yearsGG (I)
20bF31R-TETL8 yearsMTLE
21bM43L-TETL5 yearsMTLE
22bM35L-TETL9 yearsMTLE

Immunohistochemistry

Frozen tissue from peritumoral epileptic cortex (Patients 1–6, Table 1) and control cortex not infiltrated by tumor (Patients 11–22, Table 1), was used for immunohistochemical analysis as detailed in Supporting Information.

Membrane preparation and injection procedures

Membranes for oocytes injection were prepared as described previously (Miledi et al., 2006) using tissue from tumoral (Patients 2–4, Table 1), and peritumoral areas resected from neocortex of six patients afflicted by gliomas from II to IV WHO grade (Patients 1–6, Table 1) and from cortical tissues obtained from six nonepileptic patients (Patients 7–12, Table 1). For more details see Supporting Information.

Immunoblot analysis

Western blot analysis was performed on membranes extracted from peritumoral tissues surgically resected from neocortex of six epileptic patients (Patients 1–6, Table 1) and from cortical tissues obtained from four nonepileptic patients (Patients 7–10, Table 1) as detailed in Supporting Information.

Electrophysiology

Twelve hours to 48 hours after injection, membrane currents were recorded from voltage clamped oocytes by using two microelectrodes filled with 3 m KCl (Miledi, 1982). The oocytes were placed in a recording chamber (volume, 0.1 ml) and perfused continuously, 9–10 ml/min, with oocyte Ringer’s solution (OR) at room temperature (20–22°C). GABA current rundown was defined as the percentage decrease of the current peak amplitude after six 10-s applications of 500 μm GABA at 40 s intervals (Palma et al., 2004).

Current–voltage (I–V) relationships were constructed holding the oocytes at −60 mV and stepping the membrane potential for 2–4 min at the desired value before applying the neurotransmitter.

For these experiments, electrodes were filled with K-Citrate (3 m; Palma et al., 2006) to reduce the leakage of high concentration of Cl from electrodes into the oocytes. However, the experiments gave same results with both filling solutions (not shown). To determine the EGABA, I–V relationships were fitted with a second-order polynomial curve-fitting (pClamp 10) software. In oocytes, almost identical membrane resting potential (VM) values were measured inserting the voltage-recording electrode in different points of cell surface including spots very close to the injection site. The GABA-current desensitization was defined as the time taken for the current to decay from its peak to half-peak value (T0.5). For some experiments, oocytes were pretreated with bumetanide or N,N,N′,N′-tetrakis-(2-pyridylmethyl)-ethylenediamine (TPEN) for 2 h (Hershfinkel et al., 2009), or with acetazolamide (ACT) for at least 15 min (Asiedu et al., 2010) before examining the I–V curves in oocytes injected with epileptic peritumoral or control cortical membranes (Palma et al., 2006). In other experiments, performed to measure the Cl reversal potential (ECl) in different conditions, CaCl2 was pressure-injected into oocytes (Miledi & Parker, 1984) from a pipette containing 50 mm CaCl2, and currents were elicited in response to ≈10 μm calcium. In noninjected oocytes, bumetanide treatment did not affect ECl (−19.2 ± 1.1 and −19.7 ± 1.6 mV, respectively, before and after bumetanide (eight oocytes, one frog). All results are given as mean ± standard error of the mean (SEM). Two data sets were considered statistically different when p < 0.05 [analysis of variance (ANOVA) test].

Chemicals and solutions

OR solution had the following composition (in mm): NaCl 82.5; KCl 2.5; CaCl2 2.5; MgCl2 1; HEPES 5, adjusted to pH 7.4 with NaOH. All drugs were purchased from Sigma (St. Louis, MO, U.S.A.). For further details concerning the bumetanide, TPEN, ACT solutions, and drugs application see Supporting Information.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Human epileptic peritumoral cortical tissues express functional GABAA receptors

Application of GABA (500 μm) to oocytes injected with membranes from tumoral area of epileptic glioblastomas (Patients 2–4) did not elicit significant currents (15 oocytes, two frogs). This result is in agreement with previous observations showing a lack of GABA-evoked responses in cells from most malignant gliomas (Labrakakis et al., 1998; Synowitz et al., 2001). By contrast, in all oocytes injected with membranes from epileptic peritumoral tissues (Patients 1–6, Table 1) and nonepileptic control tissues (Patients 7–10 and autoptic specimens no. 11 and 12, Table 1), GABA elicited inward membrane currents (mean current amplitude, −88 nA, ranging from −5 to −350 nA, 340 oocytes, 24 frogs) within a few hours after injection (Miledi et al., 2002). These currents were caused by activation of transplanted GABAA receptors, since they were activated by GABA and were blocked by the GABAA-receptor antagonist bicuculline (100 μm; data not shown). No current response was evoked by glycine 1 mm (six oocytes, one frog, Patient 5, Table 1) indicating that glycine receptors are not functionally expressed in membrane injected oocytes and that we are recording genuine GABA currents. We next analyzed the desensitization (rundown) of GABA-evoked currents in oocytes injected with membranes obtained from the epileptic peritumoral neocortex of six patients (Patients 1–6, Table 1) and in oocytes injected with nonepileptic neocortex (Patients 7–12, Table 1).

For all samples obtained from epileptic patients, the rundown was small and not different from that of nonepileptic patients (Fig. 1); after repetitive applications of GABA (10 s duration, 40 s interval, see Methods), the current amplitude fell to 77% ± 8% (range 69–90%; 22 oocytes; three frogs; Patients 1–6, Table 1; Fig. 1) of the first pulse, a rundown comparable to that of oocytes injected with membranes from nonepileptic control cortex (78% ± 10%, range 75–94%; 16 oocytes; three frogs; Patients 7–12 Table 1; Fig. 1). In addition, analysis of current decay revealed no differences between peritumoral and control tissues (T0.5 = 7.4 ± 1.1 s nonepileptic vs. T0.5 = 8.1 ± 0.8 s at the first GABA application, p = 0.82) showing that under these experimental conditions, the GABA-current desensitization is similar in both tissues.

image

Figure 1.   GABA-current rundown is not enhanced in peritumoral tissues from patients with glioma. Amplitude of consecutive GABA currents (percent of first response; 500 μm GABA) in oocytes injected with epileptic peritumoral cortex or nonepileptic control cortex. Data points show means ± SEM: peritumoral (22 oocytes, three frogs, six patients); control cortex (16 oocytes, three frogs, six patients). All currents normalized to the first current; Icontrol: −94 ± 9 nA (•); Iperitumoral: −88 ± 12 nA (□); (Inset) sample records of first and sixth GABA currents in oocytes injected with peritumoral and control human membranes. In this and subsequent figures the bars indicate the timing of GABA applications.

Download figure to PowerPoint

Epileptic peritumoral cortical tissues show altered reversal potential of GABA-evoked currents (EGABA)

To test whether Cl homeostasis was altered in epileptic tissues, we measured EGABA in oocytes injected with membranes isolated from epileptic peritumoral neocortex (Patients 1–6, Table 1). Results obtained demonstrated that the GABA-evoked currents reversed at an average potential of −18.5 ± 0.2 mV (range: −10 to −24 mV; 250 oocytes, 24 frogs, see Fig. 2A). These EGABA values were significantly different (p < 0.001) from those measured in oocytes injected with nonepileptic control membranes in the same set of experiments (−24.7 ± 0.2 mV; range: −21 to −29 mV; 90 oocytes, 10 frogs, Patients 7–12, Table 1; Fig. 2A). EGABA were similar in oocytes injected with membranes from autopsic tissues (−24.0 ± 0.4 mV, eight oocytes, one frog, Patients 11 and 12) compared to membranes from frozen surgically resected tissues (−24.9 ± 0.4 mV, eight oocytes, one frog, Patients 7 and 8). Current amplitudes were also similar for autopsic and “fresh” tissues in this same set of experiments (40 ± 21 nA vs. 45 ± 15 nA), suggesting that postmortem delay does not affect EGABA values and the functional properties of incorporated GABAA receptors.

image

Figure 2.   Epileptic peritumoral cortical tissues show an altered EGABA. (A) Current–voltage (I–V) relationships from oocytes injected with membranes of epileptic peritumoral cortex (•) and control nonepileptic cortex (○). Points represent means (±SEM) of peak GABA currents that inverted at −24.7 mV (○; 90 oocytes, 10 frogs, six patients) and at −18.5 mV (•; 250 oocytes, 24 frogs, six patients). The GABA concentration applied was 500 μm. (Inset) Sample currents from the same experiments at the holding potentials as indicated (in millivolts). (B) Membrane currents elicited by injections of Ca2+ into oocytes injected with peritumoral (▪) or nonepileptic cortex membranes (□), both inverting at ≈−20 mV and GABA currents from the same oocytes (nine oocytes each plot, two frogs, two patients). In this case, (•) refers to the peritumoral membranes and (○) to the nonepileptic cortex, which inverted at −18.6 and −25.0 mV, respectively. (Inset) Sample Ca2+ activated Cl currents from the same experiment.

Download figure to PowerPoint

Because we found no significant difference of VM in oocytes injected with peritumoral epileptic tissues (−35.5 ± 1.4 mV; 16 oocytes, four frogs, Patients 4 and 5 Table 1) or with control nonepileptic tissues (−34.5 ± 2.3 mV; 16 oocytes, four frogs, Patients 7 and 8 Table 1; same set of experiments), we can exclude that the altered value of EGABA in epileptic peritumoral tissues was caused by a differential passive distribution of Cl into the cells.

To exclude a possible contribution of HCO3 ions in EGABA (Kaila & Voipio, 1987), we performed experiments with ACT to block carbonic anhydrase (CA) activity (Pasternack et al., 1993). In oocytes injected with nonepileptic membranes after incubation with ACT (15–45 min), EGABA was −26.2 ± 0.6 mV (eight oocytes, one frog, Patient 8, Table 1), a value not different from that found before ACT treatment in the same cells (−25.5 ± 0.5; p = 0.2). Also in the oocytes injected with peritumoral epileptic tissue, EGABA was unaffected before and after ACT treatment (−17.6 ± 0.5 and −18.0 ± 0.4 mV, respectively, p = 0.3; nine oocytes, one frog, Patient 5, Table 1), indicating that the value of EGABA is not due to alteration of HCO3 permeability.

To determine whether the value of EGABA in epileptic peritumoral tissues was caused by an altered Cl gradient across the oocyte membrane in toto, we measured the Cl reversal potential (ECl) of the native Ca2+ activated Cl current (Kusano et al., 1982). This is a slow response elicited by injecting Ca2+ ions directly into the oocytes (Miledi & Parker, 1984). ECl was found to be the same in oocytes injected with membranes from epileptic peritumoral cortex (−20.2 ± 0.6 mV; nine oocytes, two frogs, Patients 2–3, VM = −35.0 ± 1.2 mV; Table 1; Fig. 2B), nonepileptic cortex (−20.4 ± 0.4 mV; nine oocytes, two frogs, patients no. 7–8, VM = −34.1 ± 1.8 mV; Table 1; Fig. 2B) or in noninjected oocytes (−18.7 ± 0.4 mV; nine oocytes, two frogs, VM = −34.5 ± 3.8 mV; not shown), indicating that the observed differences in EGABA are not due to change in Cl membrane gradient of the whole oocyte.

Cl transporters are responsible for the altered value of EGABA in epileptic peritumoral cortical tissues

To gain evidence supporting the hypothesis that the value of EGABA in peritumoral tissues was related to alteration in the expression and/or function of Cl transporters, we measured EGABA after inhibiting the Cl transporter NKCC1 by using low concentrations of bumetanide (Dzhala et al., 2005, 2010). Oocytes injected with peritumoral membranes showed an EGABA of −18.7 ± 0.3 mV (40 oocytes, eight frogs, Patients 1–5, Fig. 3A), which shifted to −26.4 ± 0.3 mV (see for single patients Table 2; Fig. 3A) after treatment of the same oocytes with 12 μm bumetanide (2 h; Palma et al., 2004). The latter value was similar to the EGABA determined in oocytes injected with nonepileptic cortex membranes (see above and Fig. 2A). EGABA recovered to the original value after bumetanide withdrawal (1–3 h wash; not shown). By contrast, in bumetanide-treated oocytes injected with control nonepileptic tissues, EGABA was −23.6 ± 0.5 mV (12 oocytes, four frogs, Patients 8–9, Tables 1 and 2; not shown), comparable to the value measured without bumetanide pretreatment (see above).

image

Figure 3.   Cl transporters are responsible for the positive shift of EGABA in epileptic peritumoral cortical tissues. (A) Current–voltage (I–V) relationships from oocytes injected with membranes of epileptic peritumoral cortex before (•) and after treatment with bumetanide 12 μm (○). Points represent means (±SEM) of peak GABA currents that inverted at −18.9 ± 0.9 mV (•; 28 oocytes, five frogs, three patients representative of 5) and −26.8 ± 0.5 mV (○). (Inset) Sample currents from the same experiments, at the holding potentials indicated (in millivolts). (B) Current–voltage (I–V) relationships from oocytes injected with membranes of epileptic peritumoral cortex before (•) and after treatment with TPEN 100 μm (○). Points represent means (±SEM) of peak GABA currents that inverted at −19.0 ± 0.8 mV (•; 40 oocytes, 12 frogs, three patients representative of 5) and at −24.4 ± 0.3 mV (○) (Inset) Sample currents from the same experiments at the holding potentials as indicated.

Download figure to PowerPoint

Table 2.   Effects of bumetanide or TPEN on EGABA
Patient no.EGABA [cells/frogs] (mV)+bumetanide 12 μm [cells/frogs] (mV)+TPEN 100 μm [Cells/frogs] (mV)
  1. EGABA represents the mean ± SEM of the values obtained from a given patient before and after treatment with bumetanide or TPEN as indicated. Patient numbers as Table 1.

  2. *p ≤ 0.001; **p ≤ 0.05 between EGABA before and after bumetanide or TPEN treatment.

1−17.6 ± 0.6 [28/6]−26.0 ± 0.5 [11/2]*−24.5 ± 0.5 [17/4]*
2−20.0 ± 0.5 [24/6]−27.0 ± 0.4 [9/2]*−24.8 ± 0.3 [15/4]*
3−18.8 ± 0.6 [16/4]−27.6 ± 0.2 [8/2]*−23.2 ± 0.6 [8/2]*
4−18.0 ± 0.7 [12/3]−23.8 ± 0.4 [6/1]*−23.7 ± 0.3 [6/2]*
5−18.7 ± 0.4 [12/2]−26.8 ± 0.6 [6/1]*−23.5 ± 0.5 [6/1]*
8−25.4 ± 0.6 [13/4]−23.4 ± 0.7 [5/2]−28.1 ± 0.5 [8/2]**
9−24.0 ± 0.4 [18/5]−23.7 ± 0.6 [7/2]−27.6 ± 0.4 [11/3]*

In mature neurons, under normal conditions, NKCC1 is downregulated compared to KCC2, a transporter that extrudes Cl from cells (Ben-Ari, 2002). Because endogenous Zn2+ is reported to tonically inhibit KCC2 transporter activity (Hershfinkel et al., 2009), we modified the intracellular concentration of Cl by unblocking KCC2 with the zinc chelator TPEN. In oocytes injected with peritumoral epileptic membranes, TPEN treatment induced a negative shift of EGABA, from −18.7 ± 0.4 to −24.3 ± 0.3 mV (52 oocytes, 13 frogs, Patients 1–5, see for single patients Table 2; Fig. 3B), a value similar to that found in oocytes injected with control membranes (see above and Fig. 2A). In control experiments, vehicle (1% EtOH) failed to alter EGABA (−16.3 ± 0.4 mV; seven cells, one frog, Patient 3, Table 1).

To exclude possible direct effects of TPEN on GABAA receptors due to its chelating action on Zn2+, we also pressure-injected TPEN into oocytes injected with peritumoral epileptic membranes, finding no differences with extracellular application of TPEN (EGABA = −23.1 ± 0.3 mV; seven cells, two frogs, Patient 3). TPEN treatment induced a negative shift of EGABA (≈3 mV) also in oocytes injected with control membranes (Patients 8–9, Table 2; not shown), indicating that KCC2 is blocked by intracellular endogenous levels of Zn2+ as previously reported in rodent neurons (Hershfinkel et al., 2009).

Altogether these findings indicate that the positive shift of EGABA in epileptic peritumoral human cortex is due to unbalance of NKCC1 and KCC2 activity, leading to altered Cl homeostasis and a subsequent deficit in GABAergic inhibition. In particular our data suggest that the EGABA from oocytes injected with peritumoral membranes is depolarized because of overactive NKCC1 that accumulates Cl inside the cells. To determine whether altered transporter functionality was associated with changes of expression of NKCC1 and KCC2, we performed Western blot analysis and immunohistochemical staining in tissues from the same patients.

The expression of NKCC1 in epileptic peritumoral cortex is altered compared to normal brain

Immunoblot analysis

Western blot analyses were performed on equal amounts (20 μg/lane) of proteins extracted from peritumoral tissues resected from six epileptic patients (Patients 1–6, Table 1) and from cortical tissues obtained from four nonepileptic patients (Patients 7–10, Table 1). NKCC1 was detected as a single band of 170 kDa (Fig. 4A), and densitometric analysis of signal intensity, normalized for total actin, revealed higher NKCC1 expression in peritumoral tissues compared to nonepileptic cortical tissues (representative data shown in Fig. 4A, inset). Western blot analysis of KCC2 resulted in a band of 140 kDa (representative data shown in Fig. 4C, inset) and also revealed a higher expression level in peritumoral membranes (Fig. 4C).

image

Figure 4.   NKCC1 and KCC2 expression in epileptic peritumoral tissues. Western blot analysis of NKCC1 (A) and KCC2 (B) expression in cell membranes isolated from peritumoral tissues of patients with gliomas and from nonepileptic cortex (control). The graphs represent the densitometric analysis of NKCC1 (A) and KCC2 (C) expression in control (four patients, dark bar) and peritumoral (six patients, gray bar) tissues. The inserts are representative immunoblots of NKCC1 (170 kDa) and KCC2 (140 kDa). Total actin is shown as control of equal protein loading. Panels (C) and (D): representative photomicrographs of immunohistochemical staining for NKCC1 (C) and KCC2 (D). (B): NKCC1 staining showing very weak immunoreactivity (IR) in histologically normal cortex (a,b) and strong neuronal IR in the epileptic peritumoral cortex (c,d). Inset in b represents a high magnification of one of the neurons shown in the slice by the black arrow indicating no lack of neurons in normal tissue (D): KCC2 staining in histologically normal cortex (a,b) and epileptic peritumoral cortex (c,d) showing diffuse neuropil immunoreactivity (IR). The peritumoral cortex shows strong neuropil staining (b,d), with occasionally KCC2 intrasomatic IR in pyramidal neurons (d). Sections are counterstained with hematoxylin. Scale bars and calibration as indicated.

Download figure to PowerPoint

Immunohistochemical staining

Low NKCC1 neuronal expression was found in normal control nonepileptic cortex (Fig. 4B, a,b, Patients 11–16; 17–22; Tables 1 and 3). The expression pattern was similar in both frontal and temporal specimens. Resting glial cells did not show detectable levels of NKCC1 immunoreactivity (IR). NKCC1 IR was encountered in all peritumoral (infiltrated) specimens examined (Patients 1–6, Table 1), with prominent expression in neuronal cells (Fig. 4B, c,d, see Table 3). NKCC1 IR was also observed in tumor cells (not shown). In agreement with previous observations (Aronica et al., 2007) histologically normal (noninfiltrated) peritumoral cortex (from Patients 17–19, Table 1) and histologically normal surgical temporal cortex (Patients 20–22, Table 1) displayed a pattern of IR similar to that observed in control cortex, with weak or no detectable NKCC1 IR (not shown).

Table 3.   Immunoreactivity in control and peritumoral tissues
 ControlPeritumoral
  1. Values represent the mean ± SEM and are expressed as the ratio of relative optical density (ODR) of NKCC1- and KCC2-immunoreactivity.

  2. *p < 0.05.

NKCC19.6 ± 2.313.9 ± 0.7*
KCC264.2 ± 3.870.5 ± 0.8

KCC2 expression, with prominent neuropil staining, was found in normal control nonepileptic cortex (Fig. 4D a,b; Table 3). The expression pattern was similar in both frontal and temporal specimens. Resting glial cells did not show detectable levels of KCC2 IR. The peritumoral (infiltrated) cortex showed strong neuropil staining, with occasional KCC2 intrasomatic IR in pyramidal neurons (Fig. 4D c,d). NKCC1 IR was not observed in tumor cells (not shown). Again, histologically normal (noninfiltrated) peritumoral cortex (Patients 17–19) and histologically normal surgical temporal cortex (Patients 20–22,Table 1) displayed a pattern of IR similar to that observed in control cortex, with neuropil IR (not shown; Aronica et al., 2007).

All these findings show that the level of expression of NKCC1 in epileptic peritumoral cortex is significantly increased compared to normal brain as occurs in epileptic subiculum of patients with mesial temporal lobe epilepsy (MTLE) (Palma et al., 2006). The level of expression of KCC2 is not statistically different in peritumoral and normal areas, even if a slight increase of the expression in peritumoral cortex was detectable by Western blot analysis.

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Although numerous articles have been published on the role of Cl transporters in epilepsy (Palma et al., 2006; Aronica et al., 2007; Huberfeld et al., 2007; Muñoz et al., 2007), this study is the first to investigate whether alterations of Cl homeostasis occur specifically in epileptic brain tumors in human subjects. To address this issue, we injected membranes from surgically resected peritumoral tissues into Xenopus oocytes, to verify the hypothesis that alterations of GABAergic transmission underlie the hyperexcitability of the epileptic peritumoral tissue. The microtransplantation approach (Miledi et al., 2002; Eusebi et al., 2009) offers the advantage that functional receptors can be studied starting from limited amounts of surgically removed native tissues remaining after diagnostic procedures. These specimens are often too small to obtain brain slices or even isolated neurons. Moreover, autopsic tissue, which represents the only source of human brain tissue from patients not affected by neurologic diseases, can be used for control experiments as it yields no relevant differences from the fresh tissues. The main limitation of the method is that the proteins transplanted into oocytes might have different cellular origin. Despite this limitation, the functional properties of native neurotransmitter receptors are retained after transplantation into oocytes (Palma et al., 2003). Moreover, the reliability of the microtransplantation approach, at least in terms of qualitative information, has been previously demonstrated by the observation that similar variations of EGABA are observed in experiments on oocytes injected with membranes from subiculum of patients with MTLE (Palma et al., 2005, 2006) and in human brain slices (Cohen et al., 2002).

We found that tumoral tissues were not responsive to GABA, as previously reported for cultured glioma cells (Labrakakis et al., 1998; Synowitz et al., 2001), whereas GABA-evoked currents were detected in oocytes injected with epileptic peritumoral tissues or nonepileptic control tissues.

These currents undergo a limited desensitization (i.e., rundown) upon repetitive stimulation, not different between epileptic and control tissues, in agreement with previous studies on lesional epilepsies (Ragozzino et al., 2005; Janigro, 2006), contributing to indicate that this specific GABAergic dysfunction does not have a key role in tumor-related seizures. By contrast, the enhanced rundown is an event linked to epilepsy in MTLE patients (Palma et al., 2004).

A depolarized value of EGABA was the most consistently observed difference between GABA-evoked responses in oocytes injected with membranes from peritumoral epileptic cortex and oocytes injected with membranes from nonepileptic patients. This depolarizing shift in reversal of potential may cause a less efficient GABAergic inhibition in the peritumoral areas and, thus, play a role in epileptogenesis. The difference in EGABA is likely linked to altered balance of Cl transporters, as blockade of NKCC1 (using bumetanide) or unblock of KCC2 (using TPEN), reverted the positive shift in EGABA in oocytes injected with epileptic peritumoral tissue making it similar to that observed in oocytes injected with control tissue. Because the brain homogenate we injected in oocytes contained proteins from different cell types, one could argue that the transplanted GABA receptors and transporters might not reflect the situation of native neurons. Future experiments using patch-clamp recordings on peritumoral human slices, provided that enough tissue is available, will be necessary to clarify this point. Noteworthy, our hypothesis is supported by histochemical and Western blot analyses showing an increased expression of NKCC1 in epileptic peritumoral tissue as compared to controls. Control experiments also revealed that HCO3 is not relevant to the value of EGABA in epileptic tissue, since it is not present in the solutions used in this study and blockade of carbonic anhydrase by ACT (Pasternack et al., 1993; Asiedu et al., 2010) did not affect EGABA in all types of injected oocytes.

It could be argued that despite clear evidence of the involvement of NKCC1 and KCC2 in determining the reported EGABA value, the global ECl is not altered, in agreement with findings in oocytes injected with epileptic subicular membranes (Palma et al., 2005). However, it should be noted that in this experimental preparation, a few patches of injected tissue are incorporated into oocyte membrane, mostly around the injection site (Eusebi et al., 2009). These patches carry both GABAA receptors and Cl transporters originally present in the native tissue. Therefore, GABA applications activate receptors present only in discrete zones of the oocyte membrane, and colocalized with exogenous transporters, so that GABA-evoked currents may be reasonably affected by local perturbations of Cl homeostasis. Therefore, we hypothesize the existence of Cl microdomains under the plasma membrane close to transplanted GABAARs.

In cultured neurons GABA-evoked currents dynamically alter ECl (Akaike et al., 1987; Grassi, 1992), despite intracellular dialysis through patch-pipette, which is supposed to control [Cl]i, indicating that Cl ions may undergo localized changes of concentration. The Cl gradients in neurons are attributable to differences of Cl transporters expression (Rivera et al., 1999). Furthermore the existence of [Cl]i gradients has been shown in acute brain slices and in cultured neurons by loading a fixed amount of Cl via patch pipette (Khirug et al., 2005). The alteration of EGABA appears to be specifically related to epileptic tissue. It could be argued that postmortem delays inflict damage-related alterations of Cl transporters functionality and expression (Payne et al., 2003) to tissues obtained at autopsy. However, EGABA, which is influenced by these transporters, is almost identical in oocytes injected with membranes derived from nonepileptic surgical specimens and autoptic samples.

In developing, as well as in the adult brain, NKCC1 is the most important transporter mediating neuronal chloride accumulation (Blaesse et al., 2009). In the present study we show a strong increase of both the expression and staining of NKCC1 in epileptic peritumoral cortex compared to normal cortex. Increased levels of NKCC1 mRNA have been observed in the subiculum of patients with MTLE (Palma et al., 2006) and an abnormal expression of NKCC1 and KCC2 has been shown in malformations of cortical development (Aronica et al., 2007). In this article we described a neuronal IR increase of NKCC1 in tumor-related epileptic tissues. In addition, we observed a very low percentage of glial infiltration in peritumoral areas, suggesting that the increase of NKCC1 expression is mainly neuronal. This last piece of evidence provides quantitative information reinforcing the electrophysiologic results that might not properly take into account glial contamination in the membrane preparation. However, further experiments will be needed to better investigate the altered cellular and subcellular expression of NKCC1. We also report an increase of KCC2 expression and a unique neuropil staining in epileptic peritumoral areas. We hypothesized that the increased expression of KCC2 may be a compensatory mechanism underlying NKCC1 overactivity as result of neurologic disorders or developmental modifications (Gulyás et al., 2001; Aronica et al., 2007). Altogether, the data reported in this work suggest that, in peritumoral epileptic areas, the altered expression of Cl transporters renders GABAA receptor–mediated inhibitory neurotransmission less efficient, and potentially excitatory, which might lead to seizures in adult brain which, at variance with developing brain, is wired assuming a potent GABA-mediated inhibition (Ben-Ari et al., 2007; Staley, 2008; Spitzer, 2010). This hypothesis is in line with previous reports indicating that local [Cl-]i is altered by dysfunctional Cl transporters (Cohen et al., 2002; Blaesse et al., 2009) and suggesting an anticonvulsant action of bumetanide, which suppresses epileptic activity in hippocampal slices from MTLE patients (Huberfeld et al., 2007) and reduces seizures in the developing brain, inhibiting chloride accumulation and the subsequent increase of neonatal epileptic activity (Dzhala et al., 2005, 2010).

Therefore, our results may help to identify new treatments targeted at restoring the normal inhibitory action of GABA, possibly leading to the development of adjuvant antiepileptic drugs in human brain tumors.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

This article is dedicated to the memory of our colleague Prof. Fabrizio Eusebi. This work was supported by the “Mariani Foundation” of Milan (Grant R-09-76 to C.L.) and Ministero della Sanità (Ricerca finalizzata, 5 per mille, Antidoping to C.L. and E.P.), Cenci Bolognetti (to C.L.) and National Epilepsy Fund, NEF 05-11; NEF 09-05 (to E.A). We thank Dr. Agenor Limon-Ruiz and Dr. Francesca Grassi for reading the manuscript and Dr. Ricardo Miledi for suggestions. We are grateful to the patients for allowing us to carry out this work. L.C. is supported by the Neurophysiology Ph.D. program University of Rome La Sapienza.

Disclosure

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Data S1. Methods.

FilenameFormatSizeDescription
EPI_3111_sm_SupportingInformation.doc54KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.