Blog

By Yao-Hui Wang, Tian-Lan Huang, Xin Chen, Si-Xun Yu, Wei Li, Tao Chen, Yang Li, Yong-Qin Kuang, and Hai-Feng Shu

Excerpt from the article published in Journal of Neuropathology & Experimental Neurology, Volume 80, Issue 2, February 2021, Pages 137–149, https://doi.org/10.1093/jnen/nlaa149

Editor’s Highlights

• Seizures are common in patients with brain tumors.  The risk of epilepsy development is increased in low-grade glioma (LGG) cases.
• The formation of new recurrent excitatory circuits after brain injury contributes to the epileptogenesis of posttraumatic epilepsy.
• Thrombospondins (TSPs) play an important role in the development of the central nervous system (CNS) by promoting synaptogenesis and repairing synaptic circuits after insult.
• TSP2 may be involved in peritumoral structural remodeling, contributing to the epileptogenesis of glioma-related epilepsy.
• TSP binding to α2δ1 plays an epileptogenic role in posttraumatic epilepsy
• TSP2 is partly derived from reactive astrocytes. C6 cell-implanted models demonstrated that TSP2 upregulation occurs in the tumor center.  
• The binding of astrocytic TSPs to postsynaptic α2δ1 triggers the activation of Rac1 and then promotes synapse formation and maturation in the developing cortex.
• Glioma-derived TSP2 diffuses into the peritumoral region and promotes plasticity of excitatory synapses via the α2δ1/Rac1 pathway, resulting in cortical hyperexcitability, which contributes to the epileptogenesis of glioma-related epilepsy. 

Abstract

Seizures are common in patients with glioma, especially low-grade glioma (LGG). However, the epileptogenic mechanisms are poorly understood. Recent evidence has indicated that abnormal excitatory synaptogenesis plays an important role in epileptogenesis. The thrombospondin (TSP) family is a key regulator of synaptogenesis. Thus, this study aimed to elucidate the role of TSP2 in epileptogenesis in glioma-related epilepsy. The expression of TSP2 was increased in tumor tissue specimens from LGG patients, and this increase may have contributed to an increase in the density of spines and excitatory synapses in the peritumoral area. A glioma cell-implanted rat model was established by stereotactic implantation of wild-type TSP2-expressing, TSP2-overexpressing or TSP2-knockout C6 cells into the neocortex. Similarly, an increase in the density of excitatory synapses was also observed in the peritumoral area of the implanted tumor. In addition, epileptiform discharges occurred in the peritumoral cortex and were positively correlated with the TSP2 level in glioma tissues. Moreover, α2δ1/Rac1 signaling was enhanced in the peritumoral region, and treatment with the α2δ1 antagonist gabapentin inhibited epileptiform discharges in the peritumoral cortex. In conclusion, glioma-derived TSP2 promotes excitatory synapse formation, probably via the α2δ1/Rac1 signaling pathway, resulting in hyperexcitability in the peritumoral cortical networks, which may provide new insight into the epileptogenic mechanisms underlying glioma-related epilepsy.

INTRODUCTION

Seizures are common in patients with brain tumors. Approximately 30%–50% of patients with brain tumors seek medical treatment for epilepsy as the initial symptom (1, 2). Moreover, the risk of epilepsy development is increased in low-grade glioma (LGG) cases, accounting for >75% of the risk associated with the most common primary brain tumors (2, 3). It is widely accepted that there is generally an epileptogenic process before seizure onset in glioma-related epilepsy (4–7); however, the mechanisms underlying the formation of the epileptic network are still poorly understood.

Increasing evidence has shown that plasticity of the cortical network, including the synaptogenesis of excitatory synapses, is essential for epileptogenesis (8–10). For example, the formation of new recurrent excitatory circuits after brain injury contributes to the epileptogenesis of posttraumatic epilepsy (9). Recent studies have indicated that synaptic communication between glioma cells and neurons in the peritumoral cortex is facilitated by the new formation of neurogliomal synapses or gap junctions (11, 12). Moreover, peritumoral synaptic network activity is disrupted by tumor masses, resulting in network excitability (13), which may contribute to the occurrence of hyperexcitability in the peritumoral cortex. Therefore, intervening in the potential molecular cascade that results in synaptogenesis in the peritumoral cortex may be helpful for understanding the epileptogenesis of glioma-related epilepsy.

Previous studies have indicated that thrombospondins (TSPs) play an important role in the development of the central nervous system (CNS) by promoting synaptogenesis and repairing synaptic circuits after insult (14–16). Thrombospondin-2 (TSP2), which is one of the most studied members of the TSP family, plays an important role in synaptogenesis (17, 18). Astrocyte-derived TSP1 and TSP2 promote excitatory synaptogenesis in purified retinal ganglion cells in vitro (14). In ischemic injury animal models, TSP1 and TSP2 are upregulated in the affected cortex and contribute to synaptic reorganization and axon sprouting, which are required for postischemic motor recovery (15). Accordingly, we hypothesized that TSP2 may be involved in peritumoral structural remodeling, contributing to the epileptogenesis of glioma-related epilepsy.

In this study, we examined the expression of TSP2 in specimens from patients with LGG and HGG, and in the cortices of C6 glioma cell-implanted rat models. We explored the contribution of TSP2 to synaptogenesis and excitability in the peritumoral cortex. We further observed the effect of treatment with gabapentin (GBP), which is a TSP2 receptor antagonist, that may have antiepileptogenic properties and may provide a new strategy for preventing glioma-related epilepsy.

…

RESULTS

Hyperexcitability Occurs in the Peritumoral Area of LGG

In a representative case that underwent cortical electroencephalography monitoring during the craniotomy (Supplementary Data Fig. S1A, left), epileptiform activity was observed in the neighboring area of the tumor, whereas few epileptiform discharges were found in the center of the tumor (Supplementary Data Fig. S1A, right). These findings were consistent with a previous report demonstrating that the epileptic zone is located in the peritumoral area (24). In C6 cell-implanted rats, the transplants showed an invasive growth pattern (Supplementary Data Fig. S1B). A series of pathological changes, including neural edema, neural polarity disorder, and neuronal vacuolation, was found in the adjacent cortex (Supplementary Data Fig. S1B). In this study, we defined the cortex adjacent to tumors as the peritumoral area, which contained a very small amount of glioma cells and some reactive astrocytes. By implanting epidural electrodes in the tumor-bearing cortex of model rats (Supplementary Data Fig. S1C, n = 12) and recording the EEG, we detected epileptiform discharges in both the peritumoral and contralateral cortexes, in which the frequency of EEG events tended to increase significantly compared with that in the glioma center (Supplementary Data Fig. S1D, E). During recording of the EEG, ∼8%–17% of C6 wild-type cell-implanted rats showed epileptic behavior (Supplementary Data Table S6). Taken together, these data indicate that hyperexcitability occurs in the peritumoral cortex but not the center of the tumor in C6 cell-implanted rats, which was consistent with the results of the LGG clinical surgical case described above.

TSP2 Expression Is Increased in the Glioma Center

Western blot analysis showed that the TSP2 protein level was increased in the tumor tissues from LGG/HGG patients with/without epilepsy compared with the control cortex (Fig. 1A). Interestingly, the upregulation of the TSP2 protein in LGG with epilepsy was greater than that in LGG without epilepsy (Fig. 1A, p < 0.01). However, there was no difference in the protein expression of TSP2 between HGG patients with epilepsy and HGG patients without epilepsy (Fig. 1A, p = 0.328). Similarly, compared with the naïve or sham group, the protein level of TSP2 in the cortex of C6 cell-implanted rats was increased in the glioma center (Fig. 1B, p < 0.05). The IHC data further indicated that the TSP2 immunoreactivity (TSP2-IR) was mainly distributed in the glioma center (Fig. 1F) of the C6-implanted group compared with the sham and contralateral groups (Fig. 1C, D, and G). In the adjacent area between the transplant and the neighboring cortex, moderate TSP2-IR clustering with a high localized concentration was also observed (Fig. 1E).

TSP2 Promotes Excitatory Synapse Formation in the Peritumoral Cortex

Previous studies have indicated that TSPs promote excitatory synapse formation in the CNS (18). Treatment with pure TSP2 leads to an increase in the number of excitatory synapses formed by retinal ganglion cells in culture (14). Accordingly, we further explored the expression of synapses in the peritumoral tissues in the surgical specimens from LGG patients with/without epilepsy. Sections of peritumoral tissue were double-immunolabeled for presynaptic and postsynaptic markers of excitatory synapses, VGLUT1 and PSD95. The number of colocalized VGLUT1 and PSD95 puncta (Fig. 2A-a1, arrows), which represent synaptic contacts, was counted as described previously (25). Excitatory synapse markers were more strongly expressed in the peritumoral cortices of LGG patients with epilepsy than in those of the control group and LGG patients without epilepsy (Fig. 2A, B). In the wild-type C6 cell-implanted rats (C6^WT group), the total number of synapses in the peritumoral area, contralateral cortex and sham-operated cortex was determined by IF labeling synaptophysin (Supplementary Data Fig. S2A–C). The statistical data indicated that the density of the synapses was significantly increased in the peritumoral area relative to that in the contralateral cortex and sham cortex (Supplementary Data Fig. S2D). In addition, to explore whether the increase in the synaptic density was associated with the glioma cell-derived TSP2 level and to identify the contribution of the excitatory component to the increase in total synapses, we further analyzed the changes of excitatory synapse numbers in the peritumoral cortex using colabeling with VGLUT1 and PSD95. The data showed that the density of excitatory synapses in the peritumoral region of the C6^WT rats was significantly increased compared with that in the corresponding cortex of sham rats (Fig. 2C, D, p < 0.001). Interestingly, the overexpression of TSP2 in the transplanted C6 cells increased the upregulation of excitatory synapses, whereas the depletion of TSP2 attenuated this effect (Fig. 2C, D).

TSP2 Modulates the Spine Density in the Peritumoral Cortex

TSPs are involved in the development of dendritic spines in physiological and pathological conditions (26–28). To investigate the effect of TSP2 in the dendritic spines of the peritumoral cortex, we analyzed the morphology of the neurons within layers II and III of the adjacent cortex using Golgi staining (Fig. 3A). The arrangement of neurons around the tumor seemed to be disordered in the peritumoral cortex of C6^TSP2+/+ rats (Fig. 3A). In the Sholl analysis, pyramidal neurons in the adjacent area of the tumor-bearing cortex had an overall marginally significantly greater number of dendrite intersections compared with those in the sham group (Fig. 3B, p < 0.001). The number of neuron branches in the tumor-bearing cortex was greater than that in the corresponding cortex in the sham group. The mean number of neuron branches in the peritumoral area of C6^TSP2+/+ rats was higher than those in the C6^WT and C6^TSP2–/– rats (Fig. 3B, p < 0.05). However, there was no significant difference between the C6^WT and C6^TSP2–/– groups (Fig. 3B). As dendritic spines receive the most excitatory connections in spiny neurons to regulate intersynaptic signaling (29), it is important to explore the changes in peritumoral neuron synaptic spines. We found that the spine density of the peritumoral neurons was significantly increased in C6^WT rats compared with sham group rats (Fig. 3C, D, p < 0.001). The overexpression of TSP2 in C6 cells enhanced the increase in the spine density in the peritumoral neurons, whereas spine density was decreased in TSP2-depleted C6 cells compared with the C6^WT group (Fig. 3C, D).

TSP2 Affects the Ultrastructure of Synapses in the Peritumoral Cortex

We further examined the effect of TSP2 on the ultrastructure of the synapses in the peritumoral cortex by transmission EM. Compared with that in the sham group, the synapse density in the peritumoral area was significantly increased in the C6^WTgroup, which is consistent with the IF staining data above (Fig. 4A, B, p = 0.0070). Moreover, the synapse density was significantly increased in the peritumoral cortex of the C6^TSP2+/+ group but decreased in the C6^TSP2–/– group compared with the C6^WTgroup (Fig. 4A, B). In addition, the mean number of synaptic vesicles per terminal was increased in the C6^WT group compared with the sham group (p = 0.0274), whereas there was no significant difference in the number of synaptic vesicles among the C6^WT, C6^TSP2–/– and C6^TSP2+/+ groups (Fig. 4C). The analysis of the width of the synaptic cleft showed that the synaptic cleft was significantly widened in C6^TSP2+/+ rats compared with sham rats, whereas there was no significant difference among the other 3 groups (sham, C6^WT and C6^TSP2–/–; Fig. 4D). We found that there was no significant difference in the width of the PSD among the 4 groups (Fig. 4E). However, the PSD was longer in the C6^WT group than the sham group (Fig. 4F). Moreover, the overexpression of TSP2 in C6 glioma cells (C6^TSP2+/+) enhanced the increase in the length of the PSD in the peritumoral neurons, whereas the depletion of TSP2 in C6 glioma cells (C6^TSP2–/–) decreased the length compared with that in the C6^WT group (Fig. 4F). Taken together, these data indicate that glioma-derived TSP2 may induce changes in synaptic plasticity by alternating the ultrastructure of synapses in the peritumoral area.

TSP2 Contributes to Hyperexcitability in the Peritumoral Cortex

Increasing evidence has indicated that aberrant and excessive excitatory connections initiate epileptogenesis (9, 30). Therefore, we further investigated whether the TSP2-induced increase in excitatory synapses is accompanied by hyperexcitability in the peritumoral cortex using long-term cortical EEG recording. The representative traces showed that the sham-operated rats had no epileptiform discharges or associated behavioral seizures (Fig. 5A). However, 3–4 days postresuscitation after implantation, epileptiform EEG events were present in the recordings from the electrodes implanted in the peritumoral cortexes of the C6^WT, C6^TSP2+/+ and C6^TSP2–/–groups (Fig. 5B–D). The corresponding time-frequency analysis results showed that the energy of the epileptic discharge in the C6 cell-implantation rats was higher than that in the sham-group rats (Fig. 5A–D upper, p < 0.001). In addition, the mean epileptiform discharge frequency within 38 bins (5 minutes/bin) and the amplitude of the EEG events were significantly increased in the glioma-bearing rats compared with the sham-operated rats (Fig. 5E, F). Interestingly, the frequencies were markedly increased in the C6^TSP2+/+ group and reduced in the C6^TSP2–/– group compared with the C6^WT group (Fig. 5E). However, the amplitudes in the C6^WT, C6^TSP2+/+ and C6^TSP2–/– groups were not significantly different from each other (Fig. 5F).

α2δ1/Rac1 Signaling Pathway May Be Involved in Peritumoral Synaptogenesis

α2δ is a voltage-gated calcium channel subunit on the cell membrane, and it functions independently from the calcium channel (31). The EGF-like repeats of TSPs are known to bind to the VWF-A domain of α2δ and induce synaptogenesis (32, 33). Therefore, we further explored the expression of α2δ1 and its downstream molecule Rac1 in the peritumoral area. The Western blot data showed that the protein level of α2δ1 was increased in the peritumor cortex (p = 0.0429), but not the center of the implanted tumor, compared with naïve control tissue (Fig. 6A, B). The IF staining data showed that α2δ1 was mainly localized in the neighboring cortex of the tumor (Fig. 6C). The mean IF density of α2δ1 in the peritumoral cortex was significantly higher than that in the contralateral cortex (Fig. 6D, p < 0.001). Due to Rac1 being involved in various stages of synaptic development (34) and potentially acting as a downstream molecular signal of TSP2/α2δ1 interactions controlling synaptogenesis (27), we next investigated the expression of Rac1 in the cortex of C6 cell-implanted rats. As expected, the expression of Rac1 was significantly increased in the peritumoral cortex compared with the center of the glioma and naïve cortical tissue (Fig. 6E, F), suggesting that the activation of the TSP2/α2δ1/Rac1 pathway in the peritumoral area may be involved in peritumoral synaptogenesis.

GBP Treatment Reduces the Hyperexcitability of the Peritumoral Cortex

As an α2δ1 ligand, GBP inhibits pain and seizures when combined with α2δ1 (35–38). To explore whether GBP also inhibits the progression of peritumoral synaptogenesis, which may be the underlying mechanism of TSP2/α2δ1/Rac1 pathway activation, we started GBP intervention (300 mg/kg/day intragastrically for 10 days) immediately after implantation in the C6^WT-group rats. Representative EEGs in the peritumoral area and corresponding time-frequency analyses of the saline and GBP treatment groups are shown in Figure 6G. Compared with saline treatment, chronic treatment with GBP resulted in a significant reduction in the frequency and duration of EEG events in the peritumoral area (Fig. 6H, I, p < 0.001). However, there was no significant difference in amplitude between the saline and GBP treatment groups (Fig. 6J). Taken together, these data suggest that GBP may inhibit the progression of glioma-related hyperexcitability via the competitive inhibition of the TSP2/α2δ1/Rac1 pathway.

DISCUSSION

In this study, the expression of TSP2 was increased in the tumor center in surgical specimens from LGG and HGG patients, especially those from LGG patients with epilepsy and those from HGG patients. Our previous study found that TSP2 is partly derived from reactive astrocytes. C6 cell-implanted models demonstrated that TSP2 upregulation occurs in the tumor center. Further morphological data showed that the glioma cell-derived TSP2 may promote excitatory synaptogenesis in the peritumoral area, probably via the activation of the TSP2/α2δ1/Rac1 pathway. In addition, we found that the hyperexcitability in the peritumoral cortex was positively correlated with the expression level of TSP2 in the glioma cells. Treatment with GBP, which is an antagonist of TSP2/α2δ-1, had an inhibitory effect on the hyperexcitability of the peritumoral cortex. Taken together, our data demonstrate that glioma-derived TSP2 may promote excitatory synapse formation and subsequent hyperexcitability in the peritumoral cortex via the α2δ1/Rac1 pathway, suggesting the potential mechanism underlying the epileptogenesis of glioma-related epilepsy.

Increased Excitatory Synapse Formation May Contribute to the Epileptogenesis of LGG-Related Epilepsy

LGG is the most epileptogenic primary brain tumor (24). It is widely accepted that the LGG-related epileptogenesis mechanisms are multifactorial (10) and depend on the specific tumor-related characteristics and types of pathophysiological changes occurring in the peritumoral area, such as neuronal excitability, disrupted transmitter release, neurotransmitter imbalances, and cystine/glutamate antiporter system xc- and blood-brain barrier destruction (13, 24). Increasing evidence indicates that recurrent excitatory circuits and abnormal recurrent excitatory inputs play an essential role in the epileptogenesis of various types of epilepsy (8–10). The formation of new recurrent excitatory circuits after brain injury is a major factor contributing to the epileptogenesis of posttraumatic epilepsy (9). Although it is generally known that LGGs are more prone to epilepsy, in our glioma cell-implanted rat model, significant signs of epileptic discharge were found in the EEG recordings. In this study, the density of excitatory synapses was significantly increased in the peritumoral region as the EEG recording indicated hyperexcitability, suggesting that excitatory epileptogenesis in the peritumoral cortex may participate in the epileptogenesis of LGG-related epilepsy. Furthermore, the arrested maturation of excitatory synapses also plays an important role in the epileptogenesis of autosomal dominant lateral temporal lobe epilepsy (8). The spine density and ultrastructural changes in the synapses were observed in this study by Golgi staining and EM, with the results suggesting the immaturity of newly formed synapses in the peritumoral cortex. We speculate that the increased and aberrant excitatory synapses that form in the peritumoral cortex may contribute to the epileptogenesis of LGG-related epilepsy.

TSP2 Released From Glioma Cells May Increase the Number of Excitatory Synapses in the Peritumoral Cortex

Next, we investigated the molecular mechanisms underlying synapse formation within the peritumoral cortex. Multifactorial mechanisms may lead to the formation of excitatory synapses in the peritumoral cortex of LGG patients. Emerging evidence has proven that TSPs promote synapse formation in several physiological and pathological conditions (14, 15, 26–28). TSP2 is a unique member of the TSP family in that it is secreted from astrocytes and is sufficient to induce CNS synaptogenesis (14). We found that the expression of TSP2 was upregulated in the tumor center in LGG/HGG patients and may diffuse into the peritumoral region. In addition, the density of synapses in the peritumoral region was positively correlated with the level of TSP2 in glioma cells from LGG patients, suggesting that glioma-derived TSP2 may contribute to the formation of excitatory synapses in the peritumoral cortex. Moreover, there was no difference in the protein expression level of TSP2 in HGG patients regardless of whether HGG was accompanied by epilepsy, which may indicate that TSP2 plays other physiological and pathological roles in HGG as well. However, we acknowledge that there might be multiple sources of the TSP2 in addition to glioma cells in tissues from LGG/HGG patients. Under physiological conditions, TSPs begin to function in the embryonic phase, affecting the development of many organs, such as bones, muscles, the heart, and the brain (39). TSPs are mainly secreted by astrocytes in the nervous system, regulating the formation of neural circuits and promoting the development of the nervous system (14, 40). In the present study, we also found the accumulation of reactive astrocytes around the tumors (Supplementary Data Fig. S3), implying that some TSP2 may be secreted by reactive astrocytes in neighboring areas (18). In addition, other TSP family subtypes, such as TSP1, are also secreted by glioma cells (41, 42). In the CNS, TSP1 and TSP2, which are closely related and share common functional domains, both promote synaptogenesis (14). However, a great number of studies have shown that TSP1 is mainly related to angiogenesis, and our previous preliminary experiments found that the expression of TSP1 was not statistically significant. Thus, in this study, we focused on the epileptogenic role of TSP2 in LGG, but the effects of TSP1 and other TSP subtypes should also be investigated in the future.

TSP2 Controls Synaptogenesis Via the Activation of the α2δ1/Rac1 Pathway in LGG

The voltage-gated calcium channel subunit α2δ1 is a receptor for the antiepileptic and analgesic drug GBP. It has been proven that α2δ1 is also a neuronal TSP receptor responsible for excitatory CNS synaptogenesis (43). Increasing evidence shows that TSP binding to α2δ1 plays an epileptogenic role in posttraumatic epilepsy. Li et al reported that TSPs and α2δ1 levels increase after brain injury and contribute to increasing the number of excitatory synapses in the injured cortex and abnormal epileptiform burst discharges (25). In addition, recent evidence has shown that the binding of astrocytic TSPs to postsynaptic α2δ1 triggers the activation of Rac1 and then promotes synapse formation and maturation in the developing cortex (27), suggesting that TSPs control synaptogenesis via the activation of α2δ1/Rac1 pathway. In this study, both Western blotting and IF data showed that the protein level of α2δ1 was significantly increased in the peritumoral cortex of C6 cell-implanted rats. Moreover, we also found that the protein level of Rac1 was significantly increased in the peritumoral region compared with those in the control groups. Taken together, the above data suggest that α2δ1/Rac1 pathway activation may be involved in TSP2-induced synaptogenesis in the peritumoral cortex of LGG patients. However, the detailed mechanisms underlying the process need to be further investigated.

The Potential Role of GBP as an Antiepileptogenic Agent in Glioma-Related Epilepsy

Gabapentinoid drugs, which are antagonists of α2δ, have been used for the treatment of pain, anxiety and epilepsy (31, 44, 45). Recent evidence indicates that chronic GBP treatment after brain injury inhibits synapse formation and reduces posttraumatic hyperexcitability, suggesting the potential of GBP as an antiepileptogenic agent following brain injury (25, 35, 36). In this study, GBP treatment of glioma-bearing rats had an inhibitory effect, decreasing the in vivo epileptiform discharge in the peritumoral cortex, which suggests that GBP treatment may prevent the formation of a hyperexcitable network in the peritumoral cortex of glioma-bearing rats and may have therapeutic potential for use in attenuating epileptogenesis-associated glioma.

In summary, this study provides experimental evidence that glioma-derived TSP2 diffuses into the peritumoral region and promotes plasticity of excitatory synapses via the α2δ1/Rac1 pathway, resulting in hyperexcitability of the peritumoral cortex, which probably contributes to the epileptogenesis of glioma-related epilepsy. GBP, as an α2δ agonist, inhibits glioma-induced epileptogenesis. However, whether the inhibition of epileptogenesis by GBP improves glioma outcomes requires further investigation. In addition, gliomas induce neuronal hyperexcitability and seizures through nonsynaptic glutamate secretion, the secretion of synaptogenic factors, and the reduction of inhibitory interneurons in the glioma microenvironment (11). This is another aspect that deserves our attention. Furthermore, we recognize that our animal transplantation model is a high-grade glioma xenograft model and that convincing evidence for the abovementioned role of TSP2 in LGG may be lacking; this is another direction we need to explore in the future.