TSC and CDIIB cohorts and clinical findings
The TSC pediatric population comprised 20 cases (12 females) ranging in age from 1.1 to 10.1 yr. All TSC cases had one or more cortical tubers by MRI (, left panel). Generalized and partial seizures occurred in all but one case that presented with complex partial seizures, and infantile spasms occurred in 15 cases (75%). The CDIIB group comprised 20 cases (11 females) ranging in age from 0.2 to 9.2 yr. Generalized and partial seizures were seen in all cases and a history of infantile spasms was documented in 7 patients (35%). All CDIIB patients had positive neuroimaging involving cortical malformations (, right panel).
Fig. 1 A. Representative axial MRI scans of children with refractory epilepsy from TSC (left) and CDIIB (right). Both children were 7 yr of age at the time of surgery. There are multiple cortical tubers (dashed circles) in the patient with TSC and one focal (more ...)
Most of the clinical characteristics were similar between TSC and CDIIB groups (). There were no significant differences in age at seizure onset, age at surgery, gender, side of operation, seizure frequency before surgery, number of AEDs at the time of surgery, percent of patients seizure-free after surgery, and duration of follow-up (p>0.07). Epilepsy duration was longer (p=0.026) and the frequency of infantile spasms was higher (p=0.011) in the TSC compared with the CDIIB group. By neuroimaging, 4 patients had hemimegalencephaly, of which 1 was in the TSC and 3 were in the CDIIB group (p=0.373). This association, although rare, is not unprecedented (Guerra et al., 2007
). More patients with CDIIB underwent hemispherectomy compared to the TSC group (p=0.009). Seizure freedom was obtained in 77.5% of this combined cohort without surgical mortality.
At the time of surgery all patients were taking AEDs, and a total of 13 AEDs were used in mono- (n=2) and poly-therapy (n=38). In descending order, the most common AEDs were; topiramate (n=16; 40%), levetiracetam (n=13; 32.5%), carbamazepine (n=13; 32.5%), lamotrigine (n=12; 30%), phenobarbital (n=9; 22.5%), zonisamide (n=8; 20%), vigabatrin (n=7; 17.5%), and clorazepate (n=6; 15%). Comparing patients with TSC and CDIIB, there were no differences in patients taking a particular AED (p>0.08, Chi-square) with one exception. At the time of surgery, more patients with TSC were taking vigabatrin (n=7) compared to those with CDIIB (n=0, p=0.004). However, it should be noted that only 35% of patients with TSC were taking vigabatrin at the time of surgery.
Tissue samples and principal pathologic findings
In the 20 TSC patients, a total of 30 cortical tissue samples were examined (16 from the right side). They were from the frontal (n=9), temporal (n=10), and parietal (n=11) regions. The majority of samples were classified as MA (n=25) and contained tissue from the tuber. Fewer samples (n=5) were classified as LA, and were at the edge of the tuber and part of the planned resection. All children had confirmed tubers upon histopathologic examination using Haematoxylin & Eosin, Nissl and Bielschowsky staining (, left panels). The main findings included cortical dyslamination, pyramidal neuron misorientation, and the presence of dysmorphic-cytomegalic neurons. Cell clustering and heterotopic neurons were also a common finding. Giant cells with pale, glassy, eosinophilic cytoplasm were found in all cases, mostly in white matter but in many cases also extending to the cortical mantle. Astrogliosis and microcalcifications were observed in some TSC cases.
In the 20 CDIIB patients, 29 cortical tissue samples were examined (11 from the right side). The majority of samples were classified as MA (n=21) and a few samples (n=8) were classified as LA. They were from the frontal (n=16), temporal (n=7), and parietal (n=6) regions. There were no differences in sampled regions between the TSC and CDIIB groups (p=0.356, Chi-square, ). Similar to TSC cases, the histopathology of CDIIB cases showed severe architectural abnormalities, neuronal disarray and abnormal polarity, neuronal cytomegaly and balloon-type gemistocytic astrocytes restricted in three cases to the white matter (, right panels).
Morphology of individual cells in TSC and CDIIB
A total of 156 cells were recorded electrophysiologically from TSC tissue and 65% were successfully labeled with biocytin (). Based on previously published criteria (Cepeda et al., 2003
) cells were classified, in descending order of frequency, as normal-appearing pyramidal neurons (n=87, including 13 immature-looking cells), giant cells (n=47), interneurons (n=14) and cytomegalic pyramidal neurons (n=8). As expected, the average area of cytomegalic pyramidal neurons and giant cells was larger than the somatic area of normal-appearing pyramidal neurons (p<0.05, ). The somata of most giant cells were round but others were elongated. They did not have axons and their processes (7-20) had a wavy, hair-like appearance (Supplementary Fig. 1
). Overall these cells resembled enlarged astrocytes (Sosunov et al., 2008
). However, three cells looked more neuronal-like with somata similar to those of pyramidal neurons but whose processes appeared like those of typical giant/balloon cells (). We identified these as intermediate neuronal-glial cells (cf. Blumcke et al., 2011
; Talos et al., 2008
Fig. 2 Morphology and electrophysiology of TSC cell types revealed by biocytin labeling and electrophysiology using whole-cell patch clamp recordings in current clamp mode. Left panels are examples of biocytin-filled normal pyramidal (A), cytomegalic (B), interneuron (more ...)
Fig. 4 A. Camera lucida drawings from biocytin-filled cells with hybrid or neuronal-glial morphology from a TSC and a CDIIB case. Under IR-DIC the somata of these cells looked similar to that of pyramidal neurons. Only the lack of inward currents during electrophysiological (more ...)
In the CDIIB cohort we recorded 129 cells and, as in TSC, the same types of cells were found () but with different frequency. Sampled cells were normal-appearing pyramidal neurons (n=74, including 6 immature-looking pyramidal cells), cytomegalic pyramidal neurons (n=25), balloon cells (n=16, including one intermediate neuronal-glial cell), and interneurons (n=14). Thus, in our experimental conditions, we observed more giant cells in TSC than balloon cells in CDIIB samples (p=0.028), whereas more dysmorphic-cytomegalic neurons were observed in CDIIB than in TSC samples (p=0.032; ). There were no differences in somatic areas and overall morphologic characteristics between cell types from TSC and CDIIB samples based on biocytin measurements.
Fig. 3 Morphology and electrophysiology of CDIIB cell types revealed by biocytin labeling and whole-cell patch clamp recordings in current clamp mode. Left panels are examples of biocytin-filled normal pyramidal (A), cytomegalic (B), interneuron (C), and balloon (more ...)
Passive and active membrane properties of normal and abnormal cell types in TSC and CDIIB
Although there were differences in membrane capacitance, input resistance and time constant among cell types within TSC or CDIIB tissue (p<0.001, one-way ANOVA), these properties did not differ when comparing the same cell types between TSC and CDIIB tissue (). In both TSC and CDIIB, interneurons had the smallest membrane capacitance and highest input resistance, whereas cytomegalic pyramidal neurons had the largest capacitance and lowest input resistance. Giant/balloon cells also had very large capacitance but their input resistance was variable, with some cells having very low (~20 MΩ) and others very high (~1 GΩ) values. This variability may reflect morphological differences and the predominance of a glial (low input resistance) over a neuronal-like (high input resistance) phenotype (Talos et al., 2008
The RMP of normal-appearing pyramidal neurons and interneurons was similar, whereas cytomegalic pyramidal neurons and giant/balloon cells were more hyperpolarized (). Action potentials could be evoked in all neurons but not in giant/balloon cells. The current-voltage (IV) relationships were fairly typical for normal-appearing and cytomegalic pyramidal neurons, with inward rectification in the hyperpolarizing direction. Giant/balloon cells displayed an almost linear relationship in the hyperpolarizing direction but in the depolarizing direction they showed strong rectification, probably caused by the opening of delayed rectifier K+ channels (, arrow).
In normal-appearing and cytomegalic pyramidal neurons repetitive action potentials were readily elicited once the threshold for action potential firing was reached. In interneurons the action potential duration was generally shorter than in normal-appearing pyramidal neurons and the amplitude of the action potential afterhyperpolarization was larger (, arrow). Action potentials could never be elicited in balloon cells, even at highly depolarized membrane potentials. There were no significant differences in resting membrane potentials, firing properties, or current-voltage relationships of normal pyramidal neurons from TSC and CDIIB cases (-).
In voltage clamp recordings, normal-appearing and cytomegalic pyramidal neurons had prominent inward currents including persistent and fast Na+ currents, as well as transient and/or slowly inactivating Ca2+ currents. Most interneurons had negligible Ca2+ currents and fast-firing interneurons displayed repetitive Na+ currents (not shown). Giant/balloon and intermediate cells () did not display inward currents. In contrast, small outward currents were commonly observed in these cells.
Spontaneous glutamatergic currents in TSC and CDIIB
Except for the giant cells, all neurons from TSC cases displayed spontaneous EPSCs. Normal-appearing pyramidal neurons and interneurons displayed the highest frequencies of spontaneous EPSCs (average 2.95±0.5 and 4.6±0.9 Hz, respectively), whereas cytomegalic pyramidal neurons generally displayed low frequencies (average 1.8±0.5 Hz). The frequency of spontaneous EPSCs was significantly higher in normal-appearing pyramidal neurons from TSC compared with CDIIB cases (p=0.029, Mann-Whitney Rank Sum test) ( inset). Amplitude-frequency histograms showed that increased frequencies occurred at most amplitude bins, encompassing low-, medium- and large-amplitudes (). Cumulative inter-event interval (the reciprocal of frequency) probability distributions between normal pyramidal neurons from TSC and CDIIB cases also were different (p<0.05, K-S, ). In contrast, cumulative amplitude probability histograms were similar, suggesting differences were probably due to presynaptic mechanisms.
Fig. 5 A. Spontaneous EPSC activity recorded in voltage clamp mode at a holding potential of -70 mV (mostly glutamatergic) in different cell types from TSC and CDIIB tissue samples. Normal pyramidal neurons and interneurons from TSC cases displayed higher frequencies (more ...)
The EPSC frequency of interneurons from TSC was significantly higher compared to interneurons from CDIIB (p=0.012, Student's t-test). Amplitude-frequency histograms demonstrated that differences occurred across most amplitude bins and cumulative inter-event interval distributions were significantly different (p<0.0001, K-S). Although the average frequency of spontaneous EPSCs in cytomegalic neurons from TSC and CDIIB was not different (p=0.995) due to the low number of cytomegalic pyramidal neurons sampled from TSC cases, the cumulative inter-event interval probability distributions were different (p=0.001, K-S). The kinetics of spontaneous synaptic events, including rise time, decay time and half-amplitude duration, of normal and cytomegalic pyramidal neurons and interneurons were similar between TSC and CDIIB groups ().
Spontaneous GABAergic currents in TSC and CDIIB
GABAA receptor-mediated spontaneous currents were relatively abundant in all neuronal types from TSC cases (normal pyramidal 8.0±1.3 Hz, cytomegalic 10.8±3.6 Hz, and interneurons 6.7±0.6 Hz) (). As expected, giant/balloon cells displayed no spontaneous GABA synaptic activity. The average frequency of spontaneous IPSCs in normal-appearing pyramidal neurons from CDIIB cases was higher than in cells from TSC cases (p=0.018, Mann-Whitney Rank Sum test). Amplitude-frequency histograms showed that increased frequencies occurred at most amplitude bins, but were more evident in the larger amplitude bins (>25 pA) and were significantly different for events >50 pA (). Similarly, the average IPSC frequencies in cytomegalic pyramidal neurons and interneurons from CDIIB tended to be higher than in TSC samples, but because of the low numbers of cells sampled and high variability between cells the differences did not reach statistical significance. Reflecting increased frequencies, cumulative inter-event interval probability distributions demonstrated a significant leftward shift in pyramidal neurons (p<0.0001, K-S) and interneurons (p=0.0152, K-S), but only marginal in cytomegalic pyramidal neurons (p=0.0699), from CDIIB compared to TSC cases (). In contrast, cumulative amplitude probability histograms were similar (not shown). Finally, the kinetic properties of spontaneous IPSCs were not different in cells from TSC and CDIIB cases although there was a trend for normal and cytomegalic pyramidal neurons from CDIIB cases to display slower decay times ().
Fig. 6 A. Spontaneous synaptic activity recorded in voltage clamp mode at a holding potential of +10 mV (mostly GABAergic) in different cell types from TSC and CDIIB tissue samples. In contrast to glutamatergic activity, the frequencies of spontaneous GABAergic (more ...)
In a few normal pyramidal cells from TSC (n=4) and CDIIB (n=4) cases mIPSCs were isolated by addition of TTX to the external solution (not shown). The difference in average frequency between both groups remained after TTX (TSC 1.5±0.3 Hz and CDIIB 2.6±0.4 Hz, p<0.05). Cumulative inter-event interval probability distributions were also different (p<0.0001, K-S, not shown). In contrast, the cumulative amplitude probability histogram and kinetics were similar, suggesting the involvement of presynaptic mechanisms (Redman, 1990
GABA-glutamate synaptic ratio
In the subset of cases where the frequency of glutamate and GABA synaptic activities could be measured in the same cell, an index ratio was obtained by dividing the average GABA frequency by the average glutamate frequency for each patient (TSC, n=7 and CDIIB, n=9). This analysis included 56 cells from TSC (50 normal pyramidal, 2 cytomegalic, and 4 interneurons) and 72 cells from CDIIB (48 normal pyramidal, 13 cytomegalic, and 11 interneurons). The average GABA-glutamate ratio in TSC cases was lower (4.2±0.9) than in CDIIB cases (10.2±0.8, p<0.001) confirming that, relative to glutamate, GABA synaptic activity is higher in CDIIB compared with TSC cases.
GABAA receptor function in dissociated normal-appearing pyramidal neurons from TSC and CDIIB cases
To examine postsynaptic GABAA receptor sensitivity in TSC and CDIIB cases we used the isolated cell preparation. After recording the membrane properties, GABA was exogenously applied to normal-appearing pyramidal neurons (n=15, from 5 TSC cases and n=13, from 5 CDIIB cases). In all cells a clear apical dendrite and some basilar dendrites were present (, left panels). No differences in capacitance, input resistance or time constant were found between cells from TSC and CDIIB cases (). Application of increasing concentrations of GABA induced typical outward currents (Vhold=-40 mV) with a fast peak followed by a rapidly desensitizing current more evident at the higher concentrations (, right panel). Peak current amplitudes and current densities at different GABA concentrations were larger in CDIIB compared with TSC pyramidal neurons at all concentrations tested (1, 10, 100 and 1000 μM) (). Furthermore, the EC50 was significantly lower in cells from CDIIB (17.6±2.2 μM) compared with TSC cases (30.3±4.5 μM), indicating increased GABAA receptor sensitivity in CDIIB pyramidal neurons ().
Fig. 7 A. Responses to exogenous application of GABA in dissociated normal pyramidal neurons from TSC and CDIIB samples. Top left panels are examples of pyramidal neurons from a TSC and a CDIIB case. Traces on the right are outward currents induced by increasing (more ...)
Correlation with Clinical Data
Clinical variables that differed between TSC and CDIIB cases, including epilepsy duration, history of infantile spasms and use of vigabatrin (), were compared with glutamate and GABA synaptic activities and GABAA receptor sensitivity. Results showed that none of these clinical variables was associated with differences in synaptic activity or GABAA receptor sensitivity (p>0.08). In other words, the observed differences were not confounded by identified clinical variables, including use of AEDs, between TSC and CDIIB cases.