Differential Pi3K‐pathway Activation in Cortical Tubers and Focal Cortical Dysplasias with Balloon Cells

Abstract

Balloon cells of distinct focal cortical dysplasias type IIb (FCD_IIb) and giant cells of cortical tubers in tuberous sclerosis (TSC) constitute neuropathological hallmarks and cytological similarities. In TSC, frequent mutations in the TSC1 or TSC2 genes result in mTOR‐signaling activity. Here, we addressed whether Pi3K‐pathway activation differentiates balloon cells from giant cells. We used immunohistochemistry with antibodies against p‐PDK1 (S241), p‐Akt (S473), p‐tuberin (T1462), p‐p70^S6K (T389), p‐p70^S6K (T229) and phalloidin‐staining to analyze stress fiber formation in balloon cells of FCD_IIb (n = 23) compared with cortical tuber giant cells (n = 5) and adjacent normal CNS tissue as control. We have further established an in vitro assay to assess potential phosphorylation between Akt and S6. We observed phosphorylated (p‐)PDK1, p‐Akt, p‐tuberin, and p‐p70‐kDa S6‐kinase (p‐p70^S6K; residue T229) in balloon cells, whereas giant cells showed only equivalent levels of p‐tuberin, p‐p70^S6K and stress fibers. Furthermore, Pi3K‐cascade activity in balloon cells may reflect pathway “cross‐talk”. An in vitro assay revealed S6, a major target of p70^S6K, to increase phosphorylation of Akt. Our data suggest recruitment of different Pi3K‐cascade factors in the molecular pathogenesis of giant cells in cortical tubers vs. balloon cells in FCD_IIb and provides new implications for the development of treatment strategies for these cortical malformations.

INTRODUCTION

Cortical dysplasias (CD) are frequently associated with pharmacoresistant epilepsies. A variety of CDs contain dysplastic neuronal components (DNs) (31). In so‐called Taylor type CDs, large balloon cells (BCs) with opaque cytoplasm and eccentric nuclei constitute cytological hallmarks (43). In a recent classification scheme, focal cortical dysplasias type IIb (FCD_IIb) represent the equivalent of Taylor type CDs (31). BCs represent striking neuropathological similarities to giant cells (GCs) in cortical tubers of tuberous sclerosis complex (TSC)‐patients. These aberrantly shaped cells with expression of glial as well as occasionally neuronal markers suggest impaired cellular development and size control to have a pathogenetic role in such glioneuronal lesions (6, 10).

TSC constitutes an autosomal dominant transmitted disorder characterized by hamartomatous lesions in a variety of tissues. In contrast to individuals with TSC, FCD_IIb patients generally lack additional cerebral or extra‐cerebral TSC‐associated stigmata. Over 80% of TSC‐patients harbor mutations in TSC1 (hamartin) or TSC2 (tuberin) (21), whereas FCD_IIb show accumulation of allelic variants but no mutations of TSC1 (5). Recent data suggested hamartin together with tuberin to establish a tumor suppressor complex in the phosphatidylinositol 3‐kinase (Pi3K)‐pathway controlling cell size, neural development and migration (21).

Within this pathway, ligand binding to membrane bound growth factor‐ or insulin‐receptors (GFR, IR) induces Pi3K, which itself activates phosphoinositide‐dependent kinase‐1 (PDK1) (25). PDK1 phosphorylates Akt at S473 (2). p‐Akt was recently observed in FCD_IIb (37). p‐Akt inactivates tuberin, the functional TSC1/TSC2‐complex component, by phosphorylation at four residues including T1462, leading to increased cytoplasmic vs. nuclear distribution of tuberin and subsequent target of rapamycin (mTOR)‐phosphorylation (16, 34, 36). In normal cortex, hamartin/tuberin inhibit mTOR‐mediated signal transduction involving p70^S6K (12, 16). p70^S6K represents a physiological substrate for phosphorylation at T229 by p‐PDK1 (35) and at T389 by p‐mTOR (17).

Recent studies that have addressed modified mTOR‐cascade signaling in cortical tubers and FCD_IIb provided the rationale for our present analyses. Activated p‐mTOR is present in cortical tubers and FCD_IIb (28). Intriguingly, p70^S6K and its substrate S6 are phosphorylated in TSC‐associated lesions such as cortical tubers (4, 13, 28). In contrast, in FCD_IIb BCs p‐S6 is observed whereas p70^S6K is not phosphorylated (4, 28). As the mTOR‐cascade constitutes a downstream compartment of the Pi3K‐pathway, we hypothesized that constitutive Pi3K‐pathway signaling contributes to the pathogenesis of FCD_IIb and pathological differences between FCD_IIb and cortical tubers.

MATERIALS AND METHODS

Surgical specimens. Biopsy samples were collected from patients with chronic pharmacoresistant focal epilepsies in the Epilepsy‐Surgery Program at the University of Bonn. In all epileptic patients, surgical removal of the lesion was indicated in order to achieve seizure control after presurgical evaluation (8). Although genetic data on the mutational status of TSC1 or TSC2 were not available for TSC patients, the diagnosis of tuberous sclerosis could be established as all patients had additional extracerebral stigmata of TSC. In contrast, none of the individuals with FCD_IIb had additional stigmata of tuberous sclerosis. Written, informed consent was obtained from all patients concerning the use of brain tissue for additional studies. All procedures were conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of the University of Bonn. Surgical specimens were fixed in formaldehyde overnight and embedded in paraffin. All FCD_IIb and cortical tubers were reviewed by experienced neuropathologists and classified according to international classification standards (31). Paraffin sections of FCD_IIb (n = 23) and cortical tubers from patients with manifest TSC (n = 5) were appropriate for immunohistochemistry. As controls, we used unaffected central nervous system (CNS) tissue adjacent to FCD_IIb and cortical tubers, which is regularly resected in a small zone around these lesions (controls n = 23 for FCD_IIb, n = 5 for cortical tubers).

Immunohistochemical analysis and staining. We used antibodies directed against p‐PDK1 (Ser241, 1:100 dilution), p‐Akt (Ser473, 1:50 dilution), p‐tuberin/TSC2 (Thr1462, 1:100 dilution), p‐p70^S6K (Thr389, 1:500 dilution, all from Cell Signaling Technology, Beverly, MA, USA) and p‐p70^S6K (Figure 1; pT229, 1:200 dilution, Acris, Germany). Further, biotin‐phalloidin (Invitrogen, UK) was used to stain actin stress fibers (7). Paraffin sections of FCD_IIb and cortical tubers were deparaffinized in xylene, rehydrated in graded alcohols, and washed in Tris‐buffer. Endogenous peroxidase activity was quenched by incubation in phosphate‐buffered saline containing 1% hydrogen‐peroxid. Heat treatment resulted in antigen unmasking, followed by blocking of nonspecific binding performed with 0.5% normal goat serum for 2 h at 37°C. Primary antibodies were added before incubation of slides overnight at room temperature. Sections were washed in phosphate‐buffered saline, covered with diluted biotinylated secondary antibody, and incubated for 2 h at 37°C. An avidin‐biotin‐complex was applied (Vector laboratories, Burlingame, CA, USA) and visualized using a diaminobenzidine‐solution (1:50 DAB, containing 0.05% H₂O₂). Hematoxylin‐counterstained sections were mounted in aqueous media and analyzed by standard light‐microscopy. To analyze the glial or neuronal nature of immunolabeled cells in detail, we carried out hematoxylin and eosin (H&E) stains and immunohistochemical analysis with antibodies against glial fibrillary acid protein, vimentin, neurofilament protein, MAP2c and NeuN as described before (data not shown) (37). GCs and BCs may not only express glial antigens but sometimes show neuronal markers (10). Usually, BCs and GCs can be clearly distinguished from other cell components because of their characteristic cytological properties. Nevertheless, we have used serial sections for immunostaining in order to minimize contamination of the balloon or giant cell groups by other cell types. Only cells, which could be clearly identified as BCs or GCs were included in the analysis. Absolute numbers and labeling indices (labeling index‐LI) were provided for BCs and GCs. The total numbers of labeled cells were calculated for each case. Labeling indices were calculated as ratios of immunolabeled cells related to the entire cell population of interest, that is, immunolabeled cells divided by immunolabeled cells + non‐labeled cells, within the same section. The LI of all FCD_IIb or cortical tubers cases within each group were then calculated to a mean LI including the standard deviation expressed as percentage. The LI for each antibody was calculated within a total microscopic area of 781.250 µm² (200 high power fields of 0.0625 mm × 0.0625 mm width equipped to a Nikon microscope) and expressed as a percentage for all FCD_IIb or cortical tubers. As controls, we have used unaffected CNS tissue adjacent to FCD_IIb and cortical tubers, which is regularly resected in a circumscribed zone around these lesions. LIs in controls were calculated as described for FCD_IIb and cortical tubers. Care was taken to use rather equal amounts of adjacent normal cortex and white matter.

Figure 1.

Representative immunohistochemical results in normal CNS tissue, cortical tubers and focal cortical dysplasias type IIb (FCD_IIb). The left column shows representative immunohistochemical reactions in FCD_IIb (A, D, G, J, M, P), whereas in the middle column cortical tubers are presented (B, E, H, K, N, Q). The right column (C, F, I, L, O, R) demonstrates histologically normal cerebral cortex (control). Representative stains are shown for p‐PDK1 (A, B, C), p‐Akt (D, E, F), p‐tuberin (G, H, I), p‐p70^S6K (T229; J, K, L), p‐p70^S6K (T389; M, N, O) and phalloidin (P, Q, R). Whereas no significant expression of phosphorylated proteins under study is found in histologically normal cerebral cortex, substantial expression of p‐PDK1, p‐Akt, p‐tuberin, p‐p70^S6K (T229) and phalloidin is observed in FCD_IIb balloon cells (BCs), which can be identified because of abnormal cell size and eccentric nuclei. Only occasionally FCD_IIb BCs express p‐p70^S6K (T389). In contrast, cortical tubers lack significant expression of p‐PDK1 and p‐Akt, whereas we find substantial staining for p‐tuberin, p‐p70^S6K (T229), p‐p70^S6K (T389) and phalloidin (in all figures scale bar—200 µm).

Statistical analysis. Student’s t‐test was used for statistical analysis of the data. In order to obtain a high power of significance we included as many specimens as possible. However, our series was larger for FCDs compared with cortical tubers. Nonetheless, t‐test analysis appears applicable for the number of cases compared here in the cortical tuber and FCD_IIb groups, as t‐test computation immanently accounts for smaller sample sizes by augmenting the requirements for significance above average, either by a higher degree of difference and/or a higher intra‐group homogeneity.

In vitro phosphorylation assay. To purify S6 protein from bacteria, the S6 coding sequence was subcloned in pGEX‐KG. The S6 coding region was amplified using human cDNA as template and oligonucleotide primers for S6 (forward‐4F: 5′‐GCG GGA TCC AAG CTG AAC ATC TCC TTC CCA G‐3′ and reverse‐750R: 5′‐GCG AAG CTT TTA TTT CTG ACT GGA TTC AGA CTT AGA A‐3′) using standard polymerase chain reaction procedures described before (37). The polymerase chain reaction‐product was cut with BamHI and HindIII and cloned into the corresponding sites of pGEX‐KG. For protein purification, recombinant proteins were expressed in bacteria (BL21‐DE3) and purified following standard procedures (40). To assess the potential phosphorylation of S6 protein in vitro, S6‐GST‐fusion protein and GST bound to glutathione‐beads and GSK‐3‐fusion protein (product ♯9278, Cell Signaling, Frankfurt, Germany) were phosphorylated in a reaction (50 µL) containing 25 mM Tris (pH7.5), 5 mM β‐Glycerolphosphate, 2 mM DTT, o.1 mM Na₃VO₄, 10 mM MgCl₂, 10 mM ATP (2.5 mM ³²P‐γ‐ATP, 6000 cpm/pmol) and 25ng Akt1 protein‐kinase (product ♯7502, Cell Signaling) for 10 minutes at 32°C. Reactions were stopped with 12 µL 5× SDS‐PAGE buffer and analyzed by SDS‐PAGE (standard PageRuler™ Prestained Protein Ladder SM0671; Fermentas, Germany) and autoradiography (Figure 2).

Figure 2.

Analysis of phosphorylation events between S6 protein and Akt. A. Sequence of the putative Akt phosphorylation site in the S6 protein aligned with the corresponding sequence of GSK‐3 (positive control) and an optimal Akt phosphorylation motif (29). The alignment shows corresponding sequences in the key positions. Amino acid positions –1/0 in the alignment correspond to amino acid positions 253/236 of the S6 protein, which have been previously found to be phosphorylated in focal cortical dysplasias type IIb (28). B. The autoradiographic analysis of phosphorylated proteins reveals a substantially increased autophosphorylation of Akt at 79 kDa (45) in the presence of S6 protein (lane 1) in contrast to the same reaction in absence of S6 protein (lane 3). A phosphorylation of S6 protein (55 kDa) by Akt is not observed (lane 1). GSK3 phosphorylation (31 kDa) by Akt is used as positive control (lane 4). C. Staining of the SDS‐PAGE with Coomassie blue shows the S6 protein band (55 kDa, arrows in lanes 1, 2) as well as protein degradation products and contaminating bacterial proteins (*). In lane 3, only GST‐protein is present (negative control). In lane 4, GSK3 protein is visible as a distinct band at 31 kDa.

RESULTS

Here, we addressed Pi3K‐pathway signaling in BCs of FCD_IIb (n = 23) compared with cortical tuber GCs (n = 5) and adjacent normal CNS tissue of the respective lesions as control with antibodies against p‐PDK1 (S241), p‐Akt (S473), p‐tuberin (T1462), p‐p70^S6K (T389), p‐p70^S6K (T229) and phalloidin‐staining (1, 3). We identified BCs and GCs by their characteristic cytological and immunohistochemical profile in serial sections (data not shown) (37). The relative numbers of labeled BCs and GCs proved to be rather consistent within samples. p‐PDK1 labeled BCs were present in all FCD_IIb cases. Quantitative cell counts revealed 70 ± 9% immunolabeled BCs [control (FCD_IIb): 8 ± 4%]. Also in cortical tubers p‐PDK1 expression was increased vs. controls, that is, 29 ± 9% of GCs were p‐PDK1 immunoreactive vs. 11 ± 6% cells in control tissue. There was only occasional expression of p‐PDK1 in control tissue samples. To our best knowledge based on cytological characteristics and immunohistochemical analyses of serial sections, these cells were mainly identified as reactive glial cells. Significant expression of p‐PDK1 by preexisting cells in control tissue was not observed. However, the vast majority of reactive glial cells did not show expression of p‐PDK1. With respect to cortical tubers and FCD_IIb, p‐PDK1 expression was significantly higher in BCs compared with GCs.

Figure 3.

Quantitative expression analysis of the phosphorylation status of Pi3K‐pathway components and derived signaling models in normal CNS tissue, cortical tubers and focal cortical dysplasias type IIb (FCD_IIb). A. With respect to the labeling index [LI, percentage of immunoreactive balloon cells (BCs), giant cells (GCs)] of p‐PDK1, significantly more FCD_IIb components show expression than corresponding controls and elements of cortical tubers. However, there is to some degree more expression of p‐PDK1 in cortical tuber GCs than in corresponding normal control tissue. Whereas p‐Akt is significantly more frequently expressed in BCs than in GCs and controls, similarly increased LIs of p‐tuberin and p‐p70^S6K (T229) are found in FCD_IIb and cortical tubers compared with controls. For p‐p70^S6K (T389), the number of expressing GCs is significantly increased compared with controls and BCs. In contrast, there is no significant difference in p‐p70^S6K (T389) expression between FCD_IIb components and controls. There are similar levels of phalloidin‐stained BCs and GCs, significantly increased vs. controls (Co) [LIs for BCs (from FCD_IIb n = 23) and GCs (from cortical tubers n = 5; white columns] are compared with corresponding histologically normal cerebral cortex adjacent to the lesions (co‐control; n = 23 as well as n = 5, respectively; t‐test *P < 0.05, ***P < 0.001). B. Schematic representation of Pi3K‐pathway signaling in normal CNS tissue, that is, no significant induction of growth factor receptor (GFR) or insulin receptor (IR) activity is present. There is intact activity of the TSC1/TSC2 complex inhibiting mTOR signaling. No significant phosphorylation of individual pathway components is observed. C. In contrast, we find activation/phosphorylation of Pi3K‐pathway components upstream of TSC1/TSC2, that is, p‐PDK1 and p‐Akt in FCD_IIb. We suggest, that this activation induces the formation of p‐tuberin. However, there may be less phosphorylation of mTOR in FCD_IIb compared with cortical tubers as previously reported (28), which in concert with the inactivation of p70^S6K by actin stress fibers can result in the detection of minimally phosphorylated p70^S6K (T389). p‐PDK1 contributes to the phosphorylation of p70^S6K (T229) in FCD_IIb, which itself may be involved in the phosphorylation of S6. As we show here, S6 besides increasing translation is potentially involved in a positive feedback loop for phosphorylation of Akt. D. In cortical tubers, our results suggest significantly less Pi3K‐pathway activation upstream of TSC1/TSC2 compared with FCD_IIb. However, putative mutations of TSC1 or TSC2 in concert with phosphorylation of tuberin, which can be induced by factors external of the Pi3K‐pathway, will lead to stronger activation of the downstream pathway compartment including increased phosphorylation of mTOR compared with FCD_IIb (28). This can result in substantial phosphorylation of p‐p70^S6K, which in contrast to the situation in FCD_IIb overrules the inactivation drive of stress fiber formation. We suggest such “cross talk” to constitute a negative feedback mechanism within the Pi3K‐pathway involved in aberrant migration and neural network integration of GCs/BCs (26). Also in cortical tubers, p‐p70^S6K may contribute to phosphorylation of S6, which itself increases translation.

We observed strong immunoreactivity for p‐Akt in the cytoplasm of BCs in FCD_IIb (77 ± 11%). In contrast, no significant expression of p‐Akt was present in neuronal and preexisting glial cell components in normal CNS tissue adjacent to the FCD_IIb lesion (4 ± 3%) or GCs in cortical tubers (11 ± 4%; controls 6 ± 2%). There were occasionally reactive glial cells positive for p‐Akt in the control sections. Nevertheless, the majority of reactive glial cells did not show expression of p‐Akt. As previously described (25), we also observed expression of p‐Akt by dysplastic neuronal cell components in FCD_IIb. As BCs as well as GCs constitute unique cell types for FCD_IIb and cortical tubers, we concentrated on these cellular elements in the present study. p‐Tuberin showed similar LIs in BCs and GCs (BC 66 ± 5%, GC 62 ± 17%) increased vs. controls [control (FCD_IIb): 5 ± 2%, control (cortical tuber): 8 ± 4%]. We also observed similar expression levels of p‐p70^S6K (T229) in BCs and GCs compared with controls [BC 82 ± 5%, GC 80 ± 6%; control (FCD_IIb): 7 ± 3%, control (cortical tuber): 7 ± 2%]. As previously reported by others (28), we found nonspecific nuclear p‐p70^S6K (Thr389) expression in all groups. Significantly more GCs (71 ± 19%, control: 9 ± 5%) revealed cytoplasmic p‐p70^S6K (T389) than BCs (17 ± 6%, control: 6 ± 2%) similar to what has been reported before (4, 28). Using phalloidin staining, we observed many BCs and GCs containing actin stress fibers compared with controls [BC 73 ± 7%, GC 78 ± 14 %; control (FCD_IIb): 9 ± 8%, control (cortical tuber): 7 ± 4%]. For all proteins under study, we found significantly less phosphorylation in adjacent histopathologically normal cortex and white matter (1, 3), which virtually excluded substantial phosphorylation as a consequence of seizure activity. Some reactive glial cells showed expression of phosphorylated proteins under study. We did not observe expression of phosphorylated proteins analyzed here by the majority of cells that exhibit reactive astrocytic morphology. Occasionally, cells morphologically defined as DNs showed expression of phosphorylated proteins. This pattern appeared rather heterogeneous within and between individual FCD_IIb and cortical tuber specimens. However, we could not entirely exclude that the plane of section may have misidentified BCs or GCs partially hit at apical or basal cellular areas. We did not detect correlations between the expression intensity or numbers of p‐proteins expressing cells and clinical parameters such as number and type of seizures, age at seizure onset or patient sex and age (data not shown).

S6 protein has been previously reported to be activated in FCD_IIb. Therefore, we analyzed a potential “cross‐talk” between S6 and Akt in an in vitro phosphorylation assay (4, 28). We examined whether (i) S6 was phosphorylated by the Akt protein kinase or (ii) autophosphorylation of Akt was stimulated in the presence of S6 in vitro. S6 contains a potential consensus sequence for phosphorylation by Akt (residues 235/236) (Figure 2). To determine whether this residue is phosphorylated by Akt, we examined a purified recombinant GST‐fusion protein containing the complete coding region of S6. Fusion proteins were incubated with ³²P‐γ‐ATP and Akt and subsequently analyzed by SDS‐PAGE, Coomassie blue staining, and autoradiography. Intriguingly, autoradiographic analysis of phosphorylated proteins showed a substantially increased autophosphorylation of Akt at 79 kDa (45) in the presence of S6. In contrast, phosphorylation of S6 (55 kDa) by Akt was not observed in our in vitro assay, whereas the positive control GSK‐3, a known Akt substrate, was phosphorylated as expected (Figure 2). These data were replicated in three independent experimental assays. Our data suggest a positive feedback loop of the Pi3K‐pathway potentially activated in FCD_IIb.

DISCUSSION

Recent data have suggested differential activation of mTOR and individual downstream pathway components in FCD_IIb and cortical tubers (4, 25, 28). The finding of individual activated components of the cascade including p‐S6 in BCs prompted us to study the mode of their activation. Our data provide a comprehensive analysis of differential Pi3K‐pathway signaling in FCD_IIb BCs and cortical tuber GCs.

The presence of p‐PDK1 in BCs suggests active signaling within the upstream compartment of the Pi3K‐cascade in FCD_IIb (1, 3). p‐PDK1 has been demonstrated to phosphorylate p70^S6K at the T229 residue (35). Accordingly, we observe phosphorylation of p70^S6K in BCs of FCD_IIb at T229 (Figure 1). What underlies the activation of PDK1 in FCD_IIb? A potential explanation is based on specific microenvironmental conditions. Growth or other neurotrophic factors secreted by adjacent interstitial cells or the lesion itself potentially activate the Pi3K‐pathway (Figure 3C and D). Vascular endothelial growth factor and basic fibroblast growth factor were shown to activate the Pi3K‐pathway dependent on PKC and extracellular signal‐regulated kinase (42). Also endogenously produced neurotrophins (NT) such as NT‐3 and BDNF signal via tyrosine kinase receptors to activate the Pi3K‐pathway (3). Expression of NT‐4 and tyrosine kinase receptor C mRNA is elevated in dysplastic neurons and GCs of cortical tubers (22) and could also be relevant in FCD_IIb. Also in GCs, p‐PDK1 expression is higher than in control tissue, which may relate to microenvironmental conditions in the lesion (22). Notably, we observe significantly less p‐PDK1 expression in GCs compared with BCs, which suggests different factors promoting Pi3K‐pathway signaling in cortical tubers and FCD_IIb (1, 3). Our data further argue for p‐PDK1 to phosphorylate Akt in BCs, which has been recently demonstrated in FCD_IIb independently of mutations in the tumor suppressor genes PTEN and CTMP (37). In a study by Miyata et al, Akt phosphorylation was not observed in cortical tubers of patients with manifest TSC as well as in a FCD_IIb series (28). This discrepancy to our data may be partially explained by the fact that expression of p‐Akt is somewhat heterogeneous. In contrast to the approach by Miyata et al, who addressed expression of p‐Akt on tissue microarrays, we used larger FCD_IIb tissue specimens.

Downstream of Akt, phosphorylation (i.e. inhibition) of tuberin occurs in BCs (Figure 1), a mechanism previously termed “posttranslational silencing” in TSC‐associated tumors, that is, subependymal giant cell astrocytomas (SEGAs) (14). In TSC brain lesions, unlike kidney lesions for example, harbor only a low incidence of loss of heterozygosity in concert with mutations of the second allele of TSC1 or TSC2 (15). Similar to what has been reported for SEGAs (14), our finding of p‐tuberin in GCs (1, 3) suggests post‐translational inactivation of tuberin also in cortical tubers. Thereby, the somatic inactivation of the tuberin/hamartin complex involves other mechanisms than bi‐allelic inactivation because of concerted mutation and loss of heterozygosity (20). We did not address correlations between potential mutations of TSC2 and impaired expression of tuberin as the status on mutations in TSC1 or TSC2 was not available for the TSC patients with cortical tubers in the present study. However, recent data have suggested that tuberin and hamartin are expressed in similar populations of neuroglial cells of TSC tubers, even in the presence of TSC1 or TSC2 germline mutations (18). We do not observe significant expression of p‐Akt in GCs. Other factors outside the Pi3K‐pathway such as p38‐activated kinase MK2 and mitogen‐activated protein kinase could contribute to phosphorylation of tuberin in GCs (23, 44).

p‐Tuberin present in BCs (Figure 1) has the potential to induce at least some phosphorylation of mTOR. Notably, a certain degree of mTOR phosphorylation—certainly less than in cortical tubers with mutations of TSC1 or TSC2 and thereby functional ablation of one allele—has previously been shown in FCD_IIb (Figure 3C and D) (28). These previous findings in concert with our present analysis may provide an explanation for differential phosphorylation of p70^S6K in BCs, that is, phosphorylated by p‐PDK1 at the T229 residue (35) and non‐phosphorylated at the T389 residue, which is in contrast to substantial phosphorylation of p70^S6K at both sites in GCs. Subsequently, p‐p70^S6K can phosphorylate S6. p‐S6 has been previously observed in BCs and GCs (Figure 3C and D) (4, 28). Nevertheless, it may also be that the recently described p70^S6K2 or a novel unidentified kinase contributes to phosphorylation of S6 in BCs (32).

Furthermore, we here elucidate a negative feedback signaling component that can influence differential phosphorylation of p70^S6K in BCs vs. GCs. In mammalian cells such as fibroblasts, in vitro polymerization and depolymerization of filamentous actin (F‐actin) constitute the basis of migration. Stress fibers represent actin bundles that contribute to the structural organization of the cytoskeleton and modulate attachment through focal adhesions (39). p70^S6K has been demonstrated to co‐localize with actin stress fibers in Swiss 3T3 cells and rapamycin to disrupt the organization of the actin cytoskeleton (9). Actin stress fiber formation was linked to inactivation of p70^S6K (7). p‐mTOR is enriched in cellular compartments with high p‐p70^S6K levels, in contrast to cellular sites with stress fibers and inactive p70^S6K (19). Actin stress fibers have also been observed in Hib5, hippocampal neural progenitor cells, with a similar immature phenotype as BCs (24), as well as in developing glia cells (1). Ezrin, moesin, radixin (ERM) proteins, that show aberrant expression in FCD_IIb, interact with actin stress fibers (26, 41). Here, we detected stress fiber formation by phalloidin staining in BCs and GCs (Figure 1). This notion suggests a potential feedback inactivation of p70^S6K in FCD_IIb (Figure 3C) similar to what has been reported in an in vitro model before (7). In contrast to BCs, the stronger activation drive by p‐mTOR in cortical tubers (28) may overrule negative signaling of stress fibers on p‐p70^S6K (Figure 3D). This aberrant negative feedback signal provided by stress fibers in BCs would be deleterious at the cellular compartment with particular emphasis for providing focal adhesions and integration in the neural network.

It has been shown that S6 protein activation plays a significant role in transcriptional and translational regulation and increased cell size (27, 30). Furthermore, we detect a substantial increase of p‐Akt in the presence of S6 protein (Figure 2). This finding points to a positive feedback loop within the Pi3K pathway, and can be relevant for aberrant Pi3K‐pathway activation in FCD_IIb. p‐Akt is not observed in cortical tubers although S6 is phosphorylated, and may suggest S6 as cofactor supporting the auto‐phosphorylation of Akt (Figure 2) (45).

Are there implications for treatment approaches of patients with cortical tubers and FCD_IIb by our data? Recently, SEGAs in TSC patients were successfully treated by the mTOR antagonist rapamycin (11). Rapamycin has been shown to block mTOR mediated phosphorylation of p70^S6K at T389 (33, 46). Our data together with previous reports argue in favor of Pi3K‐cascade activation downstream of p‐tuberin in cortical tubers, that should render them similarly susceptible to rapamycin treatment as SEGAs (4, 25, 28). Considering active mTOR‐pathway signaling in FCD_IIb (4, 25, 28), rapamycin treatment provides a potential therapy perspective in patients with drug‐refractory epilepsy, who are not optimal candidates for epilepsy surgery. However, one has to be aware, that (i) significantly lower p‐mTOR activation is present in FCD_IIb than cortical tubers (4), and (ii) we here show that the Pi3K‐pathway is substantially activated also in more upstream compartments in FCD_IIb. By aberrant Pi3K‐cascade upstream signaling along, for example, cell cycle control cascades (38) or ERMs (26), FCD_IIb may “escape” rapamycin treatment targeted on p‐mTOR. Furthermore, although substantial regression of SEGAs was observed after rapamycin treatment (11), it is unclear, whether this may also be an antiepileptic treatment option, as not only the lesion itself but also the surrounding tissue is potentially epileptogenic. The development of pharmacological strategies for FCD_IIb may have to target more upstream compartments of the Pi3K‐cascade. However, there could be additional epigenetic events as well as—considering FCD_IIb as sporadic disorder—potential somatic mutations in tumor suppressor genes that relate to the activation of the Pi3K‐pathway in FCD_IIb, which still have to be unraveled.