Expression and Cellular Distribution of Vascular Endothelial Growth Factor‐C System in Cortical Tubers of the Tuberous Sclerosis Complex

Abstract

Cortical tubers are malformations of cortical development in patients with tuberous sclerosis complex (TSC), and highly associated with pediatric intractable epilepsy. Recent evidence has shown that signaling mediated through vascular endothelial growth factor‐C (VEGF‐C) and its receptors, VEGFR‐2 and VEGFR‐3, has direct effects on both neurons and glial cells. To understand the potential role of VEGF‐C system in the pathogenesis of cortical tubers, we investigated the expression patterns of VEGF‐C signaling in cortical tubers compared with age‐matched normal control cortex (CTX). We found that VEGF‐C, VEGFR‐2 and VEGFR‐3 were clearly upregulated in tubers at both the mRNA and protein levels, compared with CTX. The in situ hybridization and immunostaining results demonstrated that VEGF‐C, VEGFR‐2 and VEGFR‐3 were highly expressed in dysplastic neurons (DNs), giant cells (GCs) and reactive astrocytes within tubers. Most DNs/GCs expressing VEGF‐C and its receptors co‐labeled with neuronal rather than astrocytic markers, suggesting a neuronal lineage. In addition, protein levels of Akt‐1, p‐Bad and ERK1/2, the important downstream factors of the VEGF‐C pathway, were significantly increased in cortical tubers, indicating involvement of VEGF‐C–dependent prosurvival signaling in cortical tubers. Taken together, our results suggest a putative role for the VEGF‐C signaling pathway in the pathogenesis of cortical tubers.

INTRODUCTION

Tuberous sclerosis complex (TSC) is a bigenic autosomal dominant disorder caused by loss‐of‐function mutations in one of two genes, TSC1 (which encodes hamartin) and TSC2 (which encodes tuberin), and is pathologically characterized by the growth of benign tumors called hamartomas in multiple organs (9). Central nervous system (CNS) involvement is frequent in TSC and results in prominent lesions, including cortical tubers, subependymal nodules, subependymal giant cell tumors (SGCTs), and white matter abnormalities 34, 35. Cortical tubers, a typical pathological hallmark of TSC in the brain, are regions of focal cortical dysplasia that exhibit (i) disorganization or lack of the normal six‐layered cortical lamination structure; and (ii) abnormal cells known as dysplastic neurons (DNs), with aberrant somato‐dendritic morphologies, and giant cells (GCs, also named TS‐cells) with short thickened processes. TSC with cortical tubers represents a well‐recognized cause of epilepsy, which occurs in more than 70% to 80% of patients with TSC. Virtually all subtypes of seizure have been observed (9), with a poor response to the currently available anti‐epileptic drugs (AEDs). Clinically, cortical tubers are often identified as epileptogenic foci (30), which act as the source of interictal discharges and require surgical intervention (29).

Vascular endothelial growth factor (VEGF) has been found to play a significant role in angiogenesis and vasculogenesis during embryogenesis and in pathological states. Moreover, a wealth of functional analyses has indicated that VEGF also act as a neuroprotective and neurotrophic factor, supporting neuronal survival and neuronal regeneration. Among known VEGF proteins, the great majority of studies concentrate on the biological effects of VEGF‐A, the most well‐characterized member of the VEGF family, in the brain. In addition to VEGF‐A, the VEGF family currently comprises six other members: VEGF‐B, VEGF‐C, VEGF‐D, VEGF‐E, PIGF (placenta growth factor), and snake venom VEGF. VEGF‐C, a well‐recognized regulator of lymphangiogenesis, exhibits ∼30% homology to VEGF‐A and is structurally closely related to VEGF‐D. VEGF‐C binds to and activates the specific receptors VEGFR‐2 (flk‐1) and VEGFR‐3 (flt‐4). Because of a deficiency of the lymphatic system in the brain (16), relatively few studies have focused on the expression and function of VEGF‐C in the CNS.

In addition to its potent effects on the vasculature, it has become clear that the VEGF‐C signal axis has specific effects on both neurons and glia. Notably, Le Bras et al (28) demonstrated that VEGF‐C and VEGFR‐3 are present in the neural progenitor cells of Xenopus laevis and mouse embryos. Furthermore, a lack of VEGF‐C results in a severe defect in the proliferation of neuronal progenitor cells expressing VEGFR‐3, suggesting that VEGF‐C is a trophic factor in neuronal progenitor cells in the vertebrate brain 7, 28. More recently, activation of the VEGF‐C/VEGFR‐3 signaling pathway was shown to mediate proliferation and chemotaxis in glial precursor cells (26). In addition, the VEGF‐C signaling pathway has been implicated in neurological diseases, primarily brain tumors 17, 18, 19, 20, 23. Additionally, Shin et al showed that the expression of VEGF‐C and its receptors varies, with region‐ and cell‐dependent patterns in ischemic brain of rats, suggesting a potential role of VEGF‐C in ischemic insults 46, 47, 48. Notwithstanding, several of the VEGF family members (eg, VEGF‐A and VEGF‐B) are implicated in epilepsy 3, 10. More recently, Parker et al (37) revealed that VEGF‐A was upregulated in cortical tuber and subependymal giant cell astrocytoma (SEGA) specimens and TSC1 conditional knockout mouse cortex. However, no reports to date have investigated whether VEGF‐C signaling is also involved in cortical tubers of TSC.

In the present study, we analyzed the expression of VEGF‐C, its receptors and downstream factors in cortical tubers from patients with medically intractable epilepsy, at both the mRNA and protein levels. Moreover, we investigated the specific cellular distribution of VEGF‐C and its receptors in both the neuronal and glial components of this lesion.

MATERIALS AND METHODS

Surgical specimens

The biopsy samples included in this study were all obtained from the files of the Departments of Neurosurgery of Xinqiao Hospital (Third Military Medical University). Informed written consent was obtained for the use of brain tissue and for access to medical records for research purposes. TSC tissues were obtained from patients who underwent surgical resection for medically refractory epilepsy and were used in a manner compliant with the Declaration of Helsinki and the Guidelines for the Conduct of Research Involving Human Subjects, as established by Third Military Medical University. No tissue was resected solely for experimental purposes.

All patients who underwent surgical treatment met the criteria for epilepsy surgery (29). Before surgery, the epileptic lesion was localized in all patients by brain magnetic resonance imaging (MRI) or computed tomography (CT), as well as routine and ambulatory electroencephalogram (EEG) or long‐term video EEG monitoring. Single‐photon emission computed tomography (SPECT) scanning and invasive intracranial monitoring with subdural plates were also sometimes employed to delineate the epileptic zone further. Sphenoidal electrode monitoring and intraoperative electrocorticography (ECOG) were performed to localize the epileptic lesion before resection in all patients. Epilepsy surgery was most often performed in TSC patients in whom the ictal onset zone could be localized to a single corresponding tuber, or a single region surrounding several tubers, and if the potential epileptogenic foci were surgically accessible and identified outside functional cortex. Seizure outcome was assessed using Engel's et al criteria (14). All patients underwent surgery and had a follow‐up at least 1 year later.

Two neuropathologists reviewed all cases independently. All patients with TSC fulfilled the diagnostic criteria for TSC (9). The diagnosis of TSC was always confirmed by subsequent neuropathological examination. Furthermore, clinical mutation analysis of the TSC1 or TSC2 loci was performed by means of denaturing high‐performance liquid chromatography (DHPLC). We examined a total of 18 specimens removed from patients with cortical tubers, and the detailed clinical data were listed in Table 1. Information concerning the exact time of the last seizure occurrence prior to surgical resection was not available. However, all of the patients included in our series did not have seizure activity in the last 24 h before surgery.

Table 1.

Clinical and neuropathological characteristics of patients with TSC. Abbreviations: TSC = tuberous sclerosis complex; NA: no available information on mutation; NMI = no mutation identified; M = male; F = female; PO = postoperative outcome (Engel's class); PS = partial seizure; GTCS = generalized tonic‐clonic seizure; IS = infantile spasm; FR = frontal; P = parietal; O = occipital; T = temporal; RT‐PCR = reverse transcriptase‐polymerase chain reaction; WB = Western blot; IHC = immunohistochemistry (including immunofluorescence); ISH = in situ hybridization.

Case No.	Gender	Genotype	Age at surgery (year)	Seizure type	Tubers location	Epilepsy duration (year)	Seizure frequency (per month)	PO	Application in the present study
1*	M	TSC2	1.2	IS	FR	0.8	10	I	RT‐PCR; WB; IHC; ISH
2	M	TSC1	1.8	PS	P	0.9	25	I	RT‐PCR; WB; IHC;
3	M	TSC2	1.9	PS; Tonic	FR	0.9	15	I	RT‐PCR; WB; IHC; ISH
4*	F	TSC2	2.5	PS;IS	O	2.0	135	II	RT‐PCR; WB; IHC; ISH
5	M	TSC2	2.5	PS	P	0.8	10	I	RT‐PCR; WB; IHC; ISH
6	M	NA	3.9	GTCS	P	2	10	II	WB; IHC
7	F	TSC1	4.6	PS;GTCS	O	3.5	30	I	RT‐PCR; WB; IHC; ISH
8*	F	TSC2	4.6	PS; IS	T	3.5	120	III	RT‐PCR; WB; IHC; ISH
9*	M	TSC2	5.6	GTCS	T	4.5	15	I	RT‐PCR; WB; IHC; ISH
10	M	NA	7.5	PS	FR	4.5	15	I	RT‐PCR; WB
11*	F	TSC2	7.6	PS; GTCS	FR	6.0	10	I	RT‐PCR; WB; IHC; ISH
12	F	TSC2	8.5	PS; Tonic	T	6.5	30	I	RT‐PCR; WB; IHC
13	F	TSC1	9.6	PS,	FR	2.0	3	I	RT‐PCR; WB; IHC; ISH
14	M	TSC1	9.7	PS;	FR	5.3	10	I	WB; IHC
15	M	TSC1	11.2	GTCS	P	5.5	5	I	RT‐PCR; WB; IHC; ISH
16*	M	TSC2	11.3	Tonic	FR	4.9	5	IV	RT‐PCR; WB; IHC; ISH
17	F	NMI	11.5	PS	P	5.0	5	I	RT‐PCR; WB
18	M	NMI	11.5	PS; GTCS; Tonic	T	8.0	10	III	RT‐PCR; WB; IHC

^* TSC patients contained normal‐appearing cortex/white matter adjacent to the lesion (control).

For comparison, we used normal‐appearing cortex/white matter obtained at autopsy from 10 patients (male/female: 6/4; mean age 5.4, range 1.3–10.8), and the detailed clinical data were summarized in Table 2. The two neuropathologists also reviewed these cases; gross and microscopic examination revealed no history of seizures or other neurologic disorders. We also selected six surgical cases of TSC that contained sufficient amounts of perilesional zone (normal‐appearing cortex/white matter adjacent to the lesion) for comparison with the autopsy specimens. The clinical information of perilesional tissue was also listed in Table 1. Normal‐appearing tissue adjacent to the lesional zone represents an excellent disease control tissue because it is exposed to the same seizure activity, drugs and fixation time, and the age and gender are the same.

Table 2.

Clinical and neuropathological characteristics of autopsy subjects. Abbreviations: M = male; F = female; FR = frontal; P = parietal; o = occipital; T = temporal; RT‐PCR = reverse transcriptase‐polymerase chain reaction; WB = Western blot; IHC = immunohistochemistry (including immunofluorescence); PMI = post‐mortem interval (the interval between death of a patient and removal of the brain prior to freezing or fixation); ISH = in situ hybridization.

Subject No.	Gender	Pathology examination	Age at autopsy	PMI	Cause of death	Brain region	Application in the present study
1	M	Normal	1.3	1.5	Choking/suffocation	FR; P; O; T	RT‐PCR; WB; IHC; ISH
2	F	Normal	1.8	5	Drowning	P; O	RT‐PCR; WB; IHC; ISH
3	M	Normal	2.1	0.5	Non‐neurological disease	P;O;T	RT‐PCR; WB; IHC; ISH
4	M	Normal	3.5	1.2	Electric shock	FR	RT‐PCR; WB; IHC; ISH
5	F	Normal	4.0	1	Non‐neurological disease	FR; T	RT‐PCR; WB; IHC; ISH
6	F	Normal	6.5	2	Motor vehicle accident	T; O	RT‐PCR; WB; IHC; ISH
7	F	Normal	7.1	6	Drowning	FR; P	RT‐PCR; WB; IHC; ISH
8	M	Normal	8.2	2.5	Drowning	FR; P; O	RT‐PCR; WB; IHC; ISH
9	M	Normal	8.5	3	Drowning	P; O; T	RT‐PCR; WB; IHC; ISH
10	M	Normal	10.8	2.5	Motor vehicle accident	FR; T	RT‐PCR; WB; IHC; ISH

Tissue preparation

All brain samples obtained at surgery or autopsy were directly divided into two portions. One portion was snap‐frozen in liquid nitrogen and maintained at −80°C until used for mRNA analysis, Western blotting and double‐labeling immunofluorescence, as described in the subsequent text. The remaining resected brain tissue was fixed in 10% buffered formalin for histological and immunohistochemical studies. After being fixed, the paraffin‐embedded tissue was sectioned (6‐µm thick) and then mounted on polylysine‐coated slides for immunohistochemical analysis. One section of every specimen was processed for Nissl staining. The same preparation protocols were performed for analysis of tissue obtained from the control.

Reverse transcriptase‐polymerase chain reaction (RT‐PCR)

Total RNA from each frozen tissue was isolated using Trizol Reagent following the instructions of the supplier (Invitrogen, La Jolla, CA, USA). The RNA concentration and quality were evaluated by spectrophotometric measurements at 260 nm. Single‐stranded cDNA was synthesized from 1 µg of total RNA with the use of an Oligo (dT)₂₀ primer (Toyobo, Osaka, Japan) and a ReverTra Ace‐α first‐strand cDNA synthesis kit (Toyobo). For PCR, the following primers were used (Table 3).

Table 3.

Primer characteristics. Abbreviations: GAPDH = glyceraldehyde‐3‐phosphate dehydrogenase; VEGF = vascular endothelial growth factor.

	Sense primer (5′→3′)	Antisense primer (5′→3′)
VEGF‐C	AGCAAAGATCTGGAGGAGCAG	TTATGTTGCCAGCCTCCTTTC
VEGFR‐2	AAAGTGATCGGAAATGACAC	GGAATCACCACAGTTTTGTT
VEGFR‐3	TTACAACTGGGTGTCCTTTC	TTCTTGTCTATGCCTGCTCT
GAPDH	ACGGATTTGGTCGTATTGGG	TGATTTTGGAGGGATCTCGC

After reverse transcription, the PCR parameters included 5 minutes at 95°C for 1 cycle, 30 s at 95°C, 30 s at 55°C and 30 s at 72°C for 30 cycles, with a final extension for 10 minutes at 72°C. The RT–PCR products were separated by agarose gel electrophoresis and identified by ethidium bromide staining. Human glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) gene was used as an internal reference in order to normalize the data for RNA quantity and quality.

Antibody characterization

The primary antibodies used in this study were listed in Supporting Information Table S1.

Western blot analysis

Western blotting was performed to quantify the amount of VEGF‐C, VEGFR‐2/3 and downstream moleculars protein in TSC cortical tubers compared with control cortex. For each blot, cortical samples from two individual patients with TSC were homogenized respectively, and compared with a cortical sample obtained from a control subject. In brief, tissue lysates were prepared by homogenization in RIPA buffer [150 mM sodium chloride, 50 mM Tris‐HCl, pH 7.5, 1% NP‐40, 0.1% sodium dodecylsulfate, 2 µg/mL aprotinin, 2 µg/mL leupeptin and 1 mM phenylmethyl‐sulfonylfluoride (PMSF)]. Tissue debris was removed by centrifugation, and the supernatants were stored at −80°C until assayed. Protein concentrations were determined using the Bradford protein assay (Bio‐Rad, Munich, Germany). Equal amounts of protein lysates (50 mg) were resolved in 10% SDS‐polyacrylamide gels and transferred to nitrocellulose membranes in glycine transfer buffer. For immunoblotting, membranes were blocked in 5% dry milk for 1 h and incubated overnight with primary antibody. After washing and incubating with peroxidase‐conjugated secondary antibody (goat anti‐rabbit or goat anti‐mouse, 1:10 000; ZYMED, South San Francisco, CA, USA), the positive signal was visualized with chemiluminescent substrates. GAPDH protein was applied as an internal control to normalize protein loading.

In situ hybridization and immunostaining

In situ hybridization was carried out using commercially available kits (Boster, Wuhan, China), with high‐performance liquid chromatography (HPLC)‐purified oligonucleotide probes specific for VEGF‐C (Cat: MK2145), VEGFR‐2 (Cat: MK1438) and VEGFR‐3 (Cat: MK1439) mRNA. The manufacturer‐recommended procedure was employed. Paraffin sections (6 µm) were dewaxed and rehydrated, incubated in 3% H₂O₂ for 10 minutes at room temperature, then pretreated with pepsin to expose mRNA (100 µg/mL in 3% citric acid, 30 minutes, 37°C). Prehybridization was performed at 38°C for 3 h and then hybridization with the DIG‐UTP labeled oligonucleotide probe was performed at 38°C overnight in a humidified chamber with 20% glycerin. After rinsing, sections were treated with confining liquid at 37°C for 30 minutes. The staining was performed using a ready‐for‐use SABC peroxidase system (Boster) with 3,3′‐diaminobenzidine as the chromogen. Sections were counterstained with hematoxylin, dehydrated and placed on a coverslip. As a control, sections were incubated with a nonsense probe or phosphate‐buffered saline (PBS) (omitting oligonucleotide probes), which resulted in no detectable signal.

Immunohistochemical staining was performed by the avidin‐biotin peroxidase staining technique, using 3,3‐diaminobenzidine as a substrate. In brief, every paraffin‐embedded sample was sectioned (6 µm thick) and spread on polylysine‐coated slides. Sections were deparaffinized, rehydrated and incubated in 0.3% H₂O₂ diluted in methanol to quench endogenous peroxidase activity. The sections were heated in a microwave oven for 20 minutes at 98°C in citrate buffer (0.01 M, pH 6.0) for antigen retrieval, washed with PBS, incubated for 45 minutes in 10% normal goat serum (Boster) and then incubated with the primary antibodies overnight at 4°C. Thereafter, sections were washed in PBS and then used in the ready‐for‐use SABC peroxidase system (Boster) with 3,3′‐diaminobenzidine as the chromogen. Sections were counterstained with hematoxylin, dehydrated and placed on a coverslip. No immunoreactive cells were detected in negative control experiments including secondary antibody alone, pre‐absorption with a 10‐fold excess of specific blocking antigen or incubation with an isotype‐matched rabbit polyclonal antibody (see Supporting Information Figure S1). A Leica DMIRB microscope (Nussloch, Germany) was used to acquire the image.

For the double labeling studies, the snap‐frozen tissues were sectioned in a cryostat (−18°C), flattened on slides and then fixed in cold 4% paraformaldehyde for 15 minutes. Double staining was performed on sections with intact morphology, consisting of TSC cases and normal human cortex. Following overnight incubation at 4°C with the primary antibodies, sections were incubated for 2 h at room temperature with the following secondary antibodies: FITC‐conjugated anti‐rabbit antibody (1:200, Zhongshan Co., Beijing, China) and Cy5‐conjugated anti‐mouse antibody (1:500, Jackson ImmunoResearch, West Grove, PA, USA). For the controls, the primary antibody was omitted. Counterstaining of the cell nuclei was carried out by incubating the sections with Hoechst 33258 (Biotime; dilution 1:1000) for 25 minutes. Slides were viewed with a confocal microscope (TCS‐TIV; Leica). Images were converted to TIFF format, and the contrast levels were adjusted using Adobe Photoshop 10.0.

Evaluation of RT‐PCR and Western blot analysis

For both Western blot and RT‐PCR analysis, densitometric analysis was performed using UVP LABWORKS Imaging Analysis software (UVP LABWORKS, Upland, CA, USA). The optical densities (OD value) of each cDNA/protein band were calculated and normalized to the OD value of the internal reference cDNA/protein (GAPDH).

Evaluation of immunostaining

To evaluate the immunohistochemical experiments, a semi‐quantitative analysis was performed as previously reported (3). A Leica DMIRB microscope was used to examine each slice, and 200 high‐power non‐overlapping fields (of 0.0625 mm × 0.0625 mm width; a total microscopical area of 781 250 µm²) were defined in the center of the tubers using a square grid inserted into the eyepiece. DNs and GCs can be clearly distinguished from other cell components because of their characteristic cytological properties. Only cells that could be clearly identified as DNs or GCs were included in the analysis. The staining intensity was assessed on a semi‐quantitative three‐point scale where immunoreactivity (IR) was defined as −, absent (0); +, weak (1); ++, moderate (2); +++, strong staining (3).This score represents the overall staining intensity found in each section and was calculated as the average of the selected fields (Table 4). In addition, we also calculated the labeling index (LI) for GCs and DNs performed as previously described (45). The LI was defined as the ratio of immunolabeled cells related to the entire specific cell population. The LI for a specific antibody was expressed as a percentage for all TSC specimens (Table 5).

Table 4.

VEGF‐C, VEGFR‐2 and VEGFR‐3 distribution in TSC specimens (% of cases with immunoreactive cells). Abbreviations: TSC = tuberous sclerosis complex; VEGF = vascular endothelial growth factor.

Antibody	TSC*
	Dysplastic neurons				Giant cells
	−	+	++	+++	−	+	++	+++
VEGF‐C	0	0	8%	92%	0	0	8%	92%
VEGFR‐2	0	7%	21%	72%	7%	7%	57%	29%
VEGFR‐3	0	7%	14%	79%	14%	21%	36%	29%

^* n = 13 for VEGF‐C; n = 14 for VEGFR‐2 and VEGFR‐3; −, absent; +, weak, ++, moderate; +++, strong staining.

Table 5.

Labeling index of DNs and GCs in TSC specimens. Abbreviations: TSC = tuberous sclerosis complex; VEGF = vascular endothelial growth factor.

Antibody	TSC (cortical tubers)*
	Dysplastic neurons (DNs) (%)	Giant cells (GCs) (%)
VEGF‐C	96.3 ± 0.9	95.5 ± 0.5
VEGFR‐2	87.0 ± 6.3	63.4 ± 7.0
VEGFR‐3	92.3 ± 2.1	64.0 ± 7.8

^* n = 13 for VEGF‐C; n = 14 for VEGFR‐2 and VEGFR‐3.

Data analysis and statistics

Data are represented as means ± SEM, and analysis of variance (ANOVA) was performed using SPSS 10.0 software (SPSS Inc., Chicago, IL, USA). Statistical values of P < 0.05 were considered as significant.

RESULTS

RT‐PCR and Western blot analysis of VEGF‐C, VEGFR‐2 and VEGFR‐3

The initial purpose of this study was to perform a screen for the expression of the VEGF‐C signaling system in cortical tubers of TSC. We then used semi‐quantitative RT‐PCR to analyze the mRNA expression of the VEGF‐C system. As shown in Figure 1A, bands corresponding to the VEGF‐C (122 bp), VEGFR‐2 (173 bp) and VEGFR‐3 (193 bp) were detected in both cortical tubers and control brain tissue. Surprisingly, there was an obvious increase in VEGF‐C and VEGFR‐2/3 mRNA expression in tubers compared with the control cortex (CTX) samples.

Figure 1.

RT‐PCR and Western blot analysis showing VEGF‐C and its receptors mRNA and protein expression in surgical samples from patients with cortical tubers of tuberous sclerosis complex (TSC), in comparison with normal cortex (CTX) samples. (A) Representative band patterns of VEGF‐C, VEGFR‐2 and VEGFR‐3 PCR products in TSC and CTX. The right panel shows densitometric analysis of the RT‐PCR. GAPDH was used as an internal control. There is a statistically significant increase of VEGF‐C, VEGFR‐2 and VEGFR‐3 mRNA levels in cortical tubers of TSC, in comparison with CTX. (B) Representative immunoblot of VEGF‐C (∼80 kDa), VEGFR‐2 (∼200 kDa) and VEGFR‐3 (∼150 kDa) in total homogenates from TSC and CTX. The expression of internal control protein GAPDH (37 kDa) was shown in the same protein extracts. Statistical data show that both VEGF‐C and its receptors signals clearly increase in TSC. The density of the target (T) band is divided by the density of the control (C) band of β‐actin mRNA or GAPDH protein, representing the normalization factor. The CTX tissues used in this experiment were normal‐appearing cortex surrounding cortical tubers Relative optical densities (OD) of the bands were obtained from three independent experiments. The data are expressed as means ± SEM. *P < 0.05; **P < 0.01 compared with control. GAPDH = glyceraldehyde‐3‐phosphate dehydrogenase; RT‐PCR = reverse transcriptase‐polymerase chain reaction; VEGF = vascular endothelial growth factor.

We next examined the expression of the VEGF‐C system at the protein level by Western blot analysis. Figure 1B shows that VEGF‐C was present as an 80 kDa band, with higher expression levels in cortical tubers. The receptor proteins VEGFR‐2 and VEGFR‐3 were visualized as bands of approximately 200 kDa and 150 kDa, respectively. Consistent with our results obtained by RT‐PCR analysis, the VEGFR‐2 and VEGFR‐3 protein levels in cortical tubers were significantly higher than those in CTX samples. The CTX tissues presented in this figure were normal‐appearing cortex surrounding cortical tubers because it is exposed to the same seizure activity, drugs and fixation time, and the age and gender are the same.

The mRNA expression patterns of VEGF‐C, VEGFR‐2 and VEGFR‐3 in cortical tubers

Tubers are characterized histologically by a loss of the normal six‐layered structure of the cortex and the presence of a spectrum of cells with aberrant morphologies, such as DNs, reactive glial cells, bizarrely shaped astrocytes and a unique cell type known as GCs. DNs in cortical tubers exhibit disorganized neuronal orientation in the cortex and abnormal dendritic arborization. GCs are distributed throughout the cortex and subcortical white matter without clear radial or laminar orientation. GCs are usually ovoid to elliptical or polygonal in shape and also have irregular short dendritic arbors.

Histologically normal cortex displayed a weak VEGF‐C mRNA expression, and the signal was restricted to pyramidal neurons and endothelial cells of blood vessels. Glial cells did not show VEGF‐C mRNA expression (Figure 2A). In cortical tubers, there was a strong VEGF‐C mRNA expression in DNs, GCs and astrocytes; weak to moderate vascular endothelial expression was observed in all cases (Figure 2B). As shown in Figure 2C, weak to modest VEGFR‐2 mRNA expression was detected in pyramidal neurons and endothelial cells of blood vessels. Figure 2D showed strong VEGFR‐2 mRNA expression in cortical tubers. High magnification view showed strong VEGFR‐2 expression in GCs (Figure 2E) and DNs (Figure 2F), and moderate VEGFR‐2 signal in endothelial cells (Figure 2G). As shown in Figure 2H, weak VEGFR‐3 mRNA signal was observed in CTX, and the signal was restricted to pyramidal neurons; no notable VEGFR‐3 mRNA expression was observed in endothelial cells. Figure 2I showed strong VEGFR‐3 mRNA expression in DNs, GCs and astrocytes in cortical tubers, whereas endothelial cells of blood vessels were negative for the VEGFR‐3 probe. The CTX tissues presented in this figure were normal‐appearing cortex surrounding cortical tubers.

Figure 2.

In situ hybridization showing mRNA expression profiles of VEGF‐C, VEGFR‐2 and VEGFR‐3 in cortical tubers of tuberous sclerosis complex (TSC). (A) Weak VEGF‐C mRNA expression is observed in normal control cortex (CTX), and the signal is restricted to pyramidal neurons (arrows) and endothelial cells of blood vessels (inset). (B) Strong VEGF‐C mRNA expression in giant cells (arrowheads), dysplastic neurons (arrows) and astrocytes (double arrows) in cortical tubers of TSC, the insets showing high magnification views of endothelial cells (a) and giant cells (b). (C) In CTX, weak to modest VEGFR‐2 mRNA expression is detected in pyramidal neurons (arrows) and the endothelial cells of blood vessels (inset). (D) Strong VEGFR‐2 mRNA expression is present in giant cells (arrowheads), dysplastic neurons (arrows) and astrocytes (double arrows) in tubers. High magnification view showing strong VEGFR‐2 expression in giant cells (E) and dysplastic neurons (F). (G) Moderate VEGFR‐2 mRNA signal in endothelial cells. (H) Weak VEGFR‐3 mRNA signal was observed in CTX, and the signal is restricted to pyramidal neurons (arrows), VEGFR‐3 mRNA expression in blood vessels was undetectable (double arrows and inset). (I) Strong VEGFR‐3 mRNA expression in giant cells (arrowheads), dysplastic neurons (arrows) and astrocytes (double arrows) in cortical tubers, the insets showing giant cells (a), endothelial cells (b) and dysplastic neurons (c). Diaminobenzidine (DAB) is applied as the chromogen, and sections are counterstained with hematoxylin. The CTX tissues used in this experiment were normal‐appearing cortex surrounding cortical tubers Scale bars: 100 µm for A, B, C, D, H, I; 20 µm for E, F, G. VEGF = vascular endothelial growth factor.

Characterization of cell types expressing VEGF‐C, VEGFR‐2 and VEGFR‐3 in normal cortex and cortical tuber

Cellular distribution of VEGF‐C

VEGF‐C immunolabeling was detected in control cortical specimens (Figure 3C), but only weak staining was observed in pyramidal neurons (Figure 3D). Neuropil staining was undetectable to weak, and no immunostaining was observed in glial cells. Occasionally, weak staining could be detected in the endothelial cells of blood vessels (Figure 3E).

Figure 3.

Cell‐type distribution of VEGF‐C in cortical tubers of tuberous sclerosis complex (TSC). (A) Hematoxylin and eosin (H&E) staining of control cortex (CTX, obtained from autopsy). (B) Representative photomicrographs of TSC showing areas of cortical disorganization containing different cell types, including dysplastic neurons (DNs; arrowheads) with high Nissl substance staining and diverse directionality, giant cells (GCs) with eosinophilic cytoplasm (arrows in B). (C) Low magnification view of the control cortex (obtained from autopsy) with weak VEGF‐C immunoreactivity (arrows: pyramidal neurons). (D) Weak immunostaining is restricted to pyramidal neurons (arrows), and no glial labeling is observed in control normal cortex. (E) Occasionally, weak staining could be detected in the endothelial cells of blood vessels (double arrow). (F) VEGF‐C in cortical tubers showing strong immunostaining within the dysplastic cortex (inset and double arrow: vascular staining). (G) Strong VEGF‐C immunostaining was observed in dysplastic neurons (arrow) and giant cells (arrowheads). (H–I) Co‐localization of VEGF‐C (green, H) with NF‐200 (red, I) in dysplastic neurons (arrows) and giant cells (inset in H, I). Note that reactive astrocytes (double arrows) and endothelial cells of blood vessels (arrowheads) were moderately labeled with VEGF‐C. (J–K) Double labeling of VEGF‐C (green, J) with NeuN (red, K) showed no neuronal or glial (double arrows) co‐localization in cortical tubers (arrowhead: giant cells; arrow: dysplastic neurons). (L) The merged image shows no co‐localization of VEGF‐C (green) with GFAP (red) in dysplastic neurons (arrows) and giant cells (arrowhead), but colocalization of VEGF‐C with GFAP in reactive astrocytes (double arrows; yellow), while distinctive gliosis in cortical tuber was observed via GFAP immunoreactivity. (M) Merged image showing no colocalization of VEGF‐C (green) with CD68 (red) in cortical tubers (arrowhead: giant cells; arrow: dysplastic neurons; long‐arrow: macrophage‐like cells). Scale bars: 100 µm for A,B; 80 µm for C,F; 50 µm for D,E; 20 µm for G; 30 µm for H–M. GFAP = glial fibrillary acidic protein; VEGF = vascular endothelial growth factor.

In the large majority of TSC cases (Table 4), VEGF‐C was intensely expressed in 96.3% ± 0.9% of DNs (n = 653) distributed from the pial surface to the subcortical white matter (Figure 3F, Table 5). Strong immunostaining was also detected in 95.5% ± 0.5% of GCs (n = 488) (Figure 3F–G, Table 5). In addition, weak to moderate staining was observed in the endothelial cells of blood vessels (inset in Figure 3F). Double‐labeling experiments confirmed the co‐localization of VEGF‐C immunostaining with the neuronal marker NF‐200 in DNs and GCs (Figure 3H–I). Intriguingly, we could not detect any co‐immunostaining of VEGF‐C with another neuronal marker, NeuN, which is usually recognized as a marker of mature neurons, in both DNs and GCs in all cases (Figure 3J–K). Gliosis is a distinctive feature of cortical tubers; Figure 3L shows that intense glial fibrillary acidic protein (GFAP) staining was especially prominent and spread across the cortical tubers, whereas GFAP immunostaining was not detected in any VEGF‐C‐positive DNs and GCs. It has been reported that an important source of VEGF‐C in peripheral tissues is CD68‐positive macrophages in some peripheral tissues (32), but we observed no CD68‐positive cells (with macrophage‐like morphology) in the TSC tubers that expressed VEGF‐C (Figure 2M).

Cellular distribution of VEGFR2

In CTX specimens, weak to modest VEGFR‐2 immunostaining was demonstrated in pyramidal neurons and the endothelial cells of blood vessels (Figure 4A). The detailed expression pattern of VEGFR‐2 in control samples was consistent with the previous result described by Boer et al (3).

Figure 4.

Cell‐type distribution of VEGFR‐2 in cortical tubers of tuberous sclerosis complex (TSC). Panels A–E: representative photomicrographs of immunocytochemical staining for VEGF‐C. (A) A histologically normal control cortex (CTX, obtained from autopsy), showing the neuronal distribution of VEGFR‐2 with weak immunoreactivity (IR) in pyramidal neurons (arrows) (inset: vascular staining). (B) VEGFR‐2 immunostaining in dysplastic neurons (arrows) and giant cells (arrowheads, inset) of different morphology and intensity. (C) High magnification view showing moderate VEGFR‐2 IR in dysplastic neurons (arrows) and giant cells (arrowheads). (D) Strong VEGFR‐2 expression in dysplastic neurons. (E) Modest immunostaining in blood vessels (double arrow). (F) Confocal images showing co‐localization of VEGFR‐2 (green) with NF‐200 (red) in a giant cell (arrow), but another VEGFR‐2‐positive giant cell was not co‐labeled with NF‐200 (arrowhead). (G–I) GFAP‐positive astrocytic gliosis (red) surrounding a VEGFR‐2‐positive giant cell (green, arrow). VEGFR‐2 expression was co‐localized with GFAP in reactive astrocytes (arrowheads in I). Scale bars: 100 µm for A,B; 30 µm for C–E; 50 µm for F–I. GFAP = glial fibrillary acidic protein; VEGF = vascular endothelial growth factor.

In the majority of TSC specimens (Table 4), approximately 87.0% ± 6.3% of DNs (n = 374) exhibited moderate‐to‐prominent staining, but only a smaller proportion (63.4% ± 7.0%) of GCs (n = 461) displayed strong staining for VEGFR‐2 throughout the cortex and white matter (Figure 4B–D, Table 5). Weak vascular endothelial expression was observed in all cases (Figure 4E). Double‐labeling immunostaining demonstrated co‐localization of VEGFR‐2 with NF‐200 in the majority of DNs, suggesting that most of DNs expressing VEGFR‐2 were of neuronal lineage, whereas only a small percentage of GCs were verified by co‐localization of VEGFR‐2 with NF‐200 (Figure 4F). VEGFR‐2 and GFAP co‐immunolabeling was not detected in any DNs and GCs, although prominent co‐labeling of reactive astrocytes was noted in all of the TSC cases examined (Figure 4G–I).

Cellular distribution of VEGFR‐3

Barely detectable to weak VEGFR‐3 labeling in CTX was observed by immunocytochemistry, and the staining was restricted to pyramidal neurons. Resting glial cells and the endothelial cells of blood vessels in both cortex and white matter were negative for the VEGFR‐3 antibody (Figure 5A).

Figure 5.

Cell‐type distribution of VEGFR‐3 in cortical tubers of tuberous sclerosis complex (TSC). Panels A–E: representative photomicrographs of immunocytochemical staining for VEGFR‐3. (A) Weak neuronal labeling was observed in normal control cortex (CTX, obtained from autopsy), and the staining was restricted to pyramidal neurons (arrows); inset and double arrows show undetectable VEGFR‐3 IR in blood vessels. (B) VEGFR‐3 in tuber tissue showing strong immunostaining within dysplastic neurons (arrows) and giant cells (arrowheads). (C) VEGFR‐3 labeling was observed in giant cells of different intensity within subcortical white matter (arrowheads); the double arrow indicates undetectable expression of VEGFR‐3 in endothelial cells of blood vessels. (D) Higher magnification views of the dysplastic neuron within subcortical white matter (arrow). (E) Higher magnification views of weak VEGFR‐3 immunostaining in giant cells (arrowheads). (F–H) Co‐localization of VEGFR‐3 (green) with GFAP (red) in bizarre multinucleated astroglial cell (arrowheads). (I) Merged image showing expression of VEGFR‐3 in NF‐200‐positive dysplastic neurons. (J–L) GFAP‐positive astrocytic gliosis (red) surrounding a VEGFR‐3‐positive dysplastic neurons (arrows) and giant cells (arrowheads). VEGFR‐3 expression co‐localized with GFAP in reactive astrocytes (double arrows). Scale bars: 40 µm for A–C; 30 µm for D–E; 20 µm for F–H; 50 µm for I–L. IR = immunoreactivity; GFAP = glial fibrillary acidic protein; VEGF = vascular endothelial growth factor.

In the majority of cortical tubers analyzed (Table 4), VEGFR‐3 staining was encountered in both DNs and GCs (Figure 5B). Moderate to strong immunolabeling was observed in 92.3% ± 2.1% of DNs (n = 586), but only 64.0% ± 7.8% of GCs (n = 610), within the areas analyzed. Interestingly, no notable immunostaining was detected in endothelial cells in any of the specimens examined (Figure 5C, Table 5). Double‐labeling experiments confirmed that VEGFR‐3 IR co‐localized with NF‐200 in the majority of DNs. The frequency of VEGFR‐3 and NF‐200 co‐expression in GCs was much less than that in DNs (Figure 5I). In addition, the expression of VEGFR‐3 IR in astrocytes of different size and shape (co‐localization with GFAP, Figure 5F–H), but VEGFR‐3 and GFAP co‐immunolabeling, was not observed in any DNs or GCs (Figure 5J–L). This expression pattern was similar to that observed for VEGFR‐2/GFAP.

Expression of Akt, Bad and Erk1/2 in cortical tuber of TSC

Both VEGFR‐2 and VEGFR‐3 are tyrosine kinase receptors, which directly trigger the multiple signal transduction pathways including those involving cell survival and apoptosis. Therefore, the important downstream factors in VEGF‐C pathway, such as Akt, Bad and Erk 11, 43, were also observed in the present study. Western blot analysis (Figure 6) showed that the protein levels of Akt‐1, p‐Bad and ERK1/2 were significantly increased in cortical tubers. Moreover, the protein level of two other isozymes of human Akt which we tested, Akt‐2 and AKT‐3, in cortical tubers was slightly higher than in the CTX tissues, and this difference also achieved statistical significance. However, densitometric analysis of the protein bands of Bad showed that there was no statistical difference between tubers and CTX tissues. The CTX tissues used in this experiment were normal‐appearing cortex surrounding cortical tubers.

Figure 6.

Western blot showing expression of Akt, Bad and ERK1/2 in total homogenates from tuberous sclerosis complex (TSC) tubers and normal control cortex (CTX) specimens. (A) Representative immunoblot of Akt‐1(∼62 kDa), Akt‐2 (∼56 kDa), Akt‐3 (∼60 kDa), phosphorylated‐Bad (p‐Bad, ∼25 kDa), Bad (∼25 kDa) and ERK1/2 (∼44 kDa) in total homogenates from TSC and CTX. The expression of internal control protein GAPDH (37 kDa) was shown in the same protein extracts. (B) Statistical data show that protein levels of Akt‐1, p‐Bad and ERK1/2 are significantly increased in cortical tubers (P < 0.01). Moreover, the protein level of two other isozymes of human Akt which we tested, Akt‐2 and AKT‐3, in cortical tubers is slightly higher than in the CTX tissues, and this difference also achieved statistical significance (P < 0.05). However, densitometric analysis of the protein bands of Bad showed that there was no statistical difference between tubers and CTX tissues (P > 0.05). The CTX tissues used in this experiment were normal‐appearing cortex surrounding cortical tubers. *P < 0.05; **P < 0.01; #P > 0.05, ANOVA. C = control; GAPDH = glyceraldehyde‐3‐phosphate dehydrogenase; OD = optical densities; T = target.

DISCUSSION

It is accepted that VEGF‐C and its receptors are lymphogenesis regulators. Recently, increased attention has focused on the physiological and pathological functions of the VEGF‐C signaling axis in the CNS. In the present study, we found that the expression of VEGF‐C and its receptors, VEGFR‐2 and VEGFR‐3, is upregulated in cortical tubers of TSC at both the mRNA and protein levels. Intriguingly, the in situ hybridization and immunostaining results demonstrate that the high‐level expression of mRNA and protein, including VEGF‐C and its receptors, was mainly localized within GCs, DNs and reactive astrocytes. In addition, protein levels of Akt‐1, p‐Bad and ERK1/2, the important downstream factors in VEGF‐C pathway, were significantly increased in cortical tubers, indicating involvement of VEGF‐C‐dependent prosurvival signaling in cortical tubers. These results may further extend our understanding of the VEGF‐C signaling system for the pathogenesis of cortical tubers that induces intractable epilepsy.

To our knowledge, the limited reports on the expression of VEGF‐C in the human brain are mainly focused on brain tumors, such as astrocytomas (17), primitive neuroectodermal brain tumors (PNET) (20), glioma endothelium (18), glioblastomas and hemangioblastomas (23). In these studies, there are sporadic reports of VEGF‐C expression in the normal human brain. For instance, VEGF‐C mRNA expression was detected in the normal human brain (23) and normal‐appearing cortex tissue obtained from deep‐seated, nonmalignant lesions or epilepsy surgery (18). However, no previous studies have fully described VEGF‐C protein expression or localization in the human normal cortex. In this study, we detected VEGF‐C mRNA and protein expression in histologically normal‐appearing cortex (autopsy and perilesional) by semi‐quantitative RT‐PCR and Western blotting. Furthermore, in situ hybridization and immunostaining results revealed weak VEGF‐C expression in pyramidal neurons and low levels of VEGF‐C expression in vascular endothelial cells in the normal cortex.

Consistent with previous studies 3, 40, our data demonstrated that no significant differences in the results of the morphology or homogenates were observed between histologically normal cortices from autopsy specimens and perilesional normal‐appearing cortex cortices. It is worth emphasizing that the high‐quality specimens obtained at rapid autopsy (within 0.5 h–6 h post‐mortem) is a crucial factor in our study. Within this post‐mortem interval, the mRNAs and proteins, including the members of VEGF family, are stable and well preserved 21, 22.

Enhanced expression of VEGF‐C in cortical tubers

Previous studies observed that VEGF‐C mRNA was highly expressed in other neurological diseases such as brain tumors 17, 18, 20, 23. Grau et al (18) found that VEGF‐C was mainly expressed in endothelial cells and other regions tightly associated with blood vessels in brain tumor tissues. In the present study, we detected high levels of VEGF‐C mRNA and protein in cortical tuber tissue homogenate. In addition, our in situ hybridization and immunostaining results demonstrate a characteristically high level of expression of VEGF‐C in DNs and GCs in cortical tubers and faint staining in vascular endothelium, as well as mild or strong staining in reactive astrocytes. Furthermore, double labeling immunostaining provided evidence for the co‐expression of the neuronal marker NF‐200, but not the astrocytic marker GFAP, in VEGF‐C‐positive DNs and GCs, suggesting a neuronal lineage. Interestingly, none of the VEGF‐C‐positive DNs and GCs co‐expressed the mature neuron marker NeuN. It is well established that NF‐200 is a specific marker of A‐fiber neurons (myelinated neurons). Thus, VEGF‐C‐positive DNs and GNs may be a homologous population with characteristics of immature myelinated A‐fiber neurons (ie, pyramidal neurons) in the cortex.

The molecular mechanism of high VEGF‐C expression in TSC cortex tubers is not clear. It has been well established that mTOR is the integrating site of multiple growth factors and nutriment‐mediated signaling pathways. Mutations in TSC1 and TSC2 result in a loss of inhibition of the mTOR signaling pathway. Previous studies reported that VEGF‐A expression was directly modulated by TSC1 and TSC2 via mTOR‐dependent pathways in vitro 6, 13. A recent study also showed upregulated VEGF‐A expression in human tubers and SEGAs (37). Several lines of evidence indicate that HIF‐1α plays a crucial role in enhanced VEGF‐A expression mediated by mTOR. More importantly, HIF‐1α not only regulates VEGF‐A expression, but also significantly promotes VEGF‐C expression (49). Thus, we speculate that the TSC‐mTOR‐HIF‐1α axis may participate in the upregulation of VEGF‐C in cortical tubers of TSC. Additionally, seizure activity is the major clinical symptom of TSC. In this study, all samples were obtained from epilepsy patients. It is now generally believed that hypoxia takes place during seizures. When oxygen concentrations decrease, HIF‐1a expression increases, which may then upregulate VEGF‐C gene expression. Therefore, there is a possibility that seizure activity also contributes to the upregulation of VEGF‐C in cortical tubers.

Expression and cellular distribution of VEGF‐C receptors in cortical tubers

Because of high affinity for VEGF‐A, the roles of VEGFR‐2 in neurological diseases including epilepsy 3, 41 have gained increasing amounts of attention 5, 19, 27, 38, 39. However, different studies have revealed inconsistent expression patterns for VEGFR‐2 in various epilepsy samples. Rigau et al found that VEGFR‐2 upregulation occurs in microvascular endothelial cells but not in neurons or glial cells in specimens from patients with temporal lobe epilepsy (41). In contrast, Boer et al (3) demonstrated that VEGFR‐2 is expressed strongly in dysplastic neurons and balloon cells and weakly in the vascular endothelium in epilepsy‐associated focal cortical dysplasia (FCD). In the present study, we showed that DNs and GCs in cortical tubers moderately, but sometimes strongly, expressed VEGFR‐2 and also co‐localized with NF‐200, suggesting that VEGFR‐2 is expressed in neuronal lineage cells.

The high level of VEGFR‐2 expression in cortical tubers may mediate multiple effects. The neuroprotective effect mediated by VEGFR‐2 is well recognized (50). In tubers, the high expression of VEGFR‐2 in DNs and GCs may promote their survival. Furthermore, VEGFR‐2 can increase the permeability of the blood–brain barrier (BBB) 41, 51 and enhance inflammation. These effects are positively correlated with epileptic episodes 3, 52, 53, 54. Therefore, the VEGFR‐2 signaling pathway might be involved in epilepsy activity through modulation of the permeability of the BBB and inflammation response in cortical tubers.

Until now, few studies have examined the expression patterns of VEGFR‐3 in normal cortex. Previous reports found that VEGFR‐3 mRNA is expressed at a relatively high level in the cortex plate of human embryos (36), but not in the adult brain (1), suggesting that VEGFR‐3 expression correlates closely with developmental status (28). Our results of the cellular distribution revealed that VEGFR‐3 is mainly (but weakly) found in pyramidal neurons. In agreement with current views (23), we observed no VEGFR‐3 mRNA and protein expression in the endothelial cells of blood vessels in the control brain.

Compared with VEGFR‐2, there are far fewer current reports about the roles of VEGFR‐3 in the CNS. Initially, Maurer et al (33) found increased VEGFR‐3 mRNA during hypoxia in stem cells of the rat hippocampus, accompanied by the highest survival rate of these cells. A recent study showed that VEGFR‐3 was expressed by neuronal progenitor cells of mouse brains. The proliferation of neuronal progenitor cells that expressed VEGFR3 was blocked after VEGF‐C knockout (28). These results clearly demonstrate that VEGF‐C/VEGFR‐3 exerts the effects of a nerve growth factor. Most recently, Shin et al (46) found that after a short period of brain ischemia, the innermost layer of granular cells in the hippocampal dental gyrus displayed increased levels of VEGFR‐3 expression. This might be related to neurogenesis in the hippocampus after cerebral ischemia. In the present study, we found that VEGFR‐3 exhibits characteristic high levels of expression in GCs and DNs. Given the above findings, we propose that the high expression of VEGFR‐3 in cortical tubers may play a role in neuroprotection, thereby increasing the survival capacity of GCs and DNs.

As mentioned previously, high levels of VEGFR‐3 expression were mainly localized to immature cells in both humans and animals. Notably, VEGFR‐3 was expressed in CD34‐positive endothelial stem/progenitor cells. These cells also co‐expressed the stem/progenitor cell marker CD133, indicating that VEGFR‐3 might be a potential marker of stem/progenitor cells (44). Recently, Ma et al also found that tumor stem cells isolated from astrocytoma samples exhibit positive VEGFR‐3 expression (31). In our study, we found that VEGFR‐3 was characteristically expressed in GCs and DNs in the cortical tubers; this is consistent with previous findings, such as the immature morphology of GCs and DNs (9) and the expression of the progenitor marker such as nestin, ki67 8, 25, CD34 (15), and doublecortin‐like (4). Combined with the findings of the present study, these results suggest that VEGFR‐3 may be a novel marker of immature cells, and many abnormal cells (DNs and GCs) within tubers may retain embryonic phenotypic features.

Additionally, in the present study, we observed similar expression patterns for VEGF‐C, VEGFR‐2 and VEGFR‐3 in DNs and GCs within tubers. This finding suggests that VEGF‐C may function via a paracrine‐ or autocrine‐like mechanism. Our results also showed that reactive astrocytes express VEGF‐C and its receptor, thus suggesting that glial cells may be a potential source of brain VEGF‐C in tubers, and involved in the multiple effects mediated by the VEGF‐C pathway through an autocrine/paracrine manner and neuron‐glial cell.

The activation of Akt, Bad and ERK in tubers

Exceptional biological behaviors of abnormal cells are characteristic feature of cortical tuber, as evidenced by the occurrence of disrupted cell cycle control, the expression of immature neuronal markers and the alterations in apoptotic machinery (8). Disruption of the apoptotic machinery may lead to abnormal persistence of immature cells, the arrest of neuronal migration and disordered cortical lamination, and it may play a role in the pathogenesis of tubers. It is now well established that VEGF‐C signaling pathways involves in cell anti‐apoptosis and survival via activation of VEGFR‐2/3 and subsequent signal transduction pathways including the PI3K/Akt, ERK (43) and Bcl‐2 protein family (11), these molecules are well known factors for anti‐apoptosis and prosurvival. Indeed, in the present study, a pronounced increase in Akt‐1, p‐Bad and ERK1/2 protein levels was detected in TSC tubers compared with CTX tissues. Our data indicate that activation of Akt‐1, Akt‐2, p‐Bad and ERK1/2 is potentially involved in the VEGF‐C‐dependent prosurvival signaling in tubers (see Supporting Information Figure S2, proposed model for VEGF‐C‐mediated pathways in the pathogenesis of cortical tubers in TSC). However, it should be noted that Akt/Bad/ERK transduction can also be activated by multiple growth factors including insulin‐like growth factors (IGF), platelet‐derived growth factor (PDGF) transforming growth factor (TGF), basic fibroblast growth factor (bFGF), and (perhaps) at least some of Akt/Bad/ERK transduction is triggered via these growth factors (12). Therefore, further research is still needed to investigate the direct contribution of VEGF‐C to the activation of the Akt/Bad/ERK transduction in tubers.

Although there is as yet no direct evidence for the effects of VEGF‐C on neuronal electrophysiological characteristics, previous studies still suggest a potential correlation between upregulated VEGF‐C system expression and the epileptogenesis of cortical tuber. Constitutive and complicated activation of inflammatory pathways has been accepted as a striking feature that occurs in epilepsy‐associated cortical tuber (2). It has been shown that VEGF‐C is an important inflammatory mediator that participates widely in the inflammatory system (24). Thus, the high expression of VEGF‐C in cortex tubers may promote an inflammatory response that leads to chronic and constitutive inflammation in lesions. Recent research suggests that inflammatory responses in the cortex can locally increase neuron circuit excitability and decrease the threshold of seizures, thus promoting seizures in human studies and animal models (42). Therefore, one of the important roles of VEGF‐C in cortical tubers may be to regulate epilepsy activities through an inflammatory mechanism.

In conclusion, our findings demonstrate that VEGF‐C and it receptors are expressed in epilepsy‐associated cortical tubers of TSC. The high expression of VEGF‐C signaling in DNs, GCs and reactive astrocytes indicates that different cellular components of cortical tubers are involved in VEGF‐C signaling. These data provided further insights into the molecular events governing tuber formation and the complex interactions between abnormal cells and their surrounding cells. Future studies will be needed to elucidate the biological significance of VEGF‐C signaling in tubers and will be essential for designing more effective therapies for epilepsy‐associated cortical tubers.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supporting information

Figure S1. No immunoreactive cells were detected in negative control experiments. (a–f) pre‐absorption with blocking antigen; (g–h) incubation with an isotype‐matched rabbit polyclonal antibody.

Figure S2. Proposed model for VEGF‐C‐mediated pathways in the pathogenesis of cortical tubers in TSC.

Table S1. Antibodies used for immunostaining and Western blot.

Supporting info item

Supporting info item

Supporting info item