Background
Mesial temporal lobe epilepsy (MTLE) is the most common cause of intractable epilepsy in adults and is characterized by hippocampal sclerosis, neuronal loss, gliosis, and mossy fiber sprouting [
1-
4]. Psychiatric comorbidities are frequent in MTLE patients, and in population-based studies, epilepsy has been consistently associated with increased risk of schizophrenia [
5]. However, the exact biological substrate behind the association of MTLE and psychiatric comorbidities is unknown [
6,
7]. We have recently shown neuropathological alterations in the hippocampus of patients with epilepsy and the history of major depression or interictal psychosis, which may indicate that structural changes and neurochemical dysfunctions may underlie psychiatric symptoms in MTLE [
8-
10].
Neuroinflammation-related abnormalities such as glial pathology, glutamate dysregulation, and blood-brain-barrier dysfunction are found not only in epilepsy, but also in schizophrenia and major depression [
11]. Glial proteins, such as metallothionein I and II (MT-I/II), are able to quench free zinc and modulate glutamatergic neurotransmission [
12], and aquaporin 4 (AQP4), found in astrocytic endfeets, is a regulator of water homeostasis that majorly controls edema formation and tissue excitability [
13,
14]. In schizophrenia, upregulation of MT-I/II and of astrocyte and microglia markers have been documented in several brain regions [
15-
17]. By contrast, neuropathological studies in specimens from major depression patients indicate reduction in hippocampal glial fibrillary acidic protein (GFAP)-positive astrocytes and of AQP4 and MT-I/II in the frontal cortex [
18,
19]. Protein expression and neuropathological features in MTLE with psychiatric comorbidities may resemble what is found in the pure form of the correspondent psychiatric illness [
20]. Therefore, we hypothesized that expression of reactive astrocytes, activated microglia, glial MT-I/II, and AQP4 would be altered in the hippocampal formation of MTLE patients with major depression and interictal psychosis.
Discussion
In the present study, we investigated the expression of glial proteins GFAP, HLA-DR, MT-I/II, and perivascular AQP4 in the hippocampal formation of MTLE with and without psychiatric comorbidities and in non-epileptic controls. Comparing the MTLE groups, we found in specific hippocampal subfields an increased immunoreactive area of GFAP and HLA-DR and decreased MT-I/II and AQP4 in specimens from the MTLE patients with psychosis; while in specimens from patients with MTLE and major depression, GFAP and MT-I/II were decreased. Differences between MTLE and controls, in astrogliosis, microgliosis, increased MT-I/II, and decreased perivascular AQP4 in the epileptogenic hippocampus, were similar to what is currently described in the literature [
4,
31]. Given that differences between epileptogenic and control hippocampi are already well established in the literature, our discussion will focus mainly on psychiatric subgroups and their differences when compared to MTLE without psychiatric comorbidities, unless otherwise specified.
Studies in humans and animal models of epilepsy have shown upregulation of several inflammatory molecules [
32,
33]. However, only a few experimental studies have focused on inflammatory changes in correlates of major depression comorbid with epilepsy. For example, rats injected with pilocarpine exhibit behavioral equivalents of anhedonia and despair and alterations in inflammatory molecules as found in human major depression [
34,
35]. No information regarding neuroinflammatory mechanisms in psychosis of epilepsy is available to date.
Astrogliosis and microgliosis are part of a common response to injury. Although the reactive astrocyte expression profile may depend on the type of inducing injury, increased inflammatory-related molecules are always found in reactive astrocytes [
36]. Interleukins 1b (IL-1beta) and 6 (IL-6) are among the molecules released after injury that can lead to glial reaction [
37]. In fact, the crosstalk between activated microglia and reactive astrocytes seems crucial to the maintenance of chronic gliosis [
38]. Increased astrocytic GFAP expression, a marker of reactive astrogliosis, is a common finding in the hippocampus of MTLE patients [
3,
4]. Likewise, we detected an increased GFAP immunoreactive area in all hippocampal subfields of MTLE patients. In patients with MTLE + D, GFAP expression levels were intermediary between the controls and MTLE
W or MTLE + P. Since studies have shown that patients with major depression have reduced GFAP expression [
18,
39], it is possible that the mechanisms underlying the decreased GFAP expression observed in major depression counterbalance the increase found in epilepsy, resulting in the intermediary values observed in our MTLE + D cases. In schizophrenia, a study showed that increased GFAP expression in the prefrontal cortex of patients with schizophrenia is increased [
16], but other cortical areas and the hippocampus have shown inconclusive results [
17,
40,
41]. In MTLE patients with interictal psychosis, we found increased GFAP expression, especially when compared to the MTLE + D cases, in agreement with a recent hypothesis that astrocyte pathology may be associated with psychotic symptoms, although the exact nature of this change remains unclear [
16]. In particular, increased GFAP in schizophrenia/psychotic symptoms could be closely related to increased neuroinflammatory markers [
42], as well as to increased IL-1beta and IL-6 serum levels [
43]. A recent study comparing MTLE hippocampi from patients with and without
de novo psychosis (postoperative psychosis) analyzed GFAP expression and found no qualitative differences between groups [
44]. In our present series, quantitative differences in GFAP between MTLE
W and MTLE + P were also subtle, and major differences were seen in respect to the MTLE + D group.
Increased cortical and hippocampal HLA-DR+ microglia has been described in schizophrenia [
17,
45], in accordance to our findings in several hippocampal subfields of MTLE patients with interictal psychosis. Of note, increased hippocampal HLA-DR was particularly associated with paranoid schizophrenia [
45], a core symptom especially represented in interictal psychosis [
20]. HLA-DR levels in MTLE specimens from patients without psychiatric comorbidities and in those with major depression were similar and higher than in the controls, although statistically significant differences were detected only in the granule cell layer and CA3-1 of MTLE
W versus control. Microglia is an important source of inflammatory molecules [
32], and a high expression of pro-inflammatory cytokines is observed in major depression [
46]. Similar apparent microglial activation in MTLE
W and MTLE + D could partially explain why patients with epilepsy frequently develop mood disorders, an association still incompletely understood [
47]. However, the levels of microglial-related inflammatory molecules such as cytokines remain to be evaluated in human epilepsy with and without major depression.
Metallothioneins are regulators of free zinc levels, an important modulator of glutamatergic neurotransmission [
12]. Besides metals, oxidative stress agents and inflammatory molecules can induce MT-I/II expression [
12]. Knockout mice for IL-6 have low microglial activation and low expression of MT-I/II, indicating a crucial role of inflammation in MT-I/II expression [
48]. In fact, mice overexpressing MT-I/II show reduced microgliosis and reduced levels of interleukins following kainic acid
status epilepticus [
49]. In psychiatric diseases, MT-I/II gene expression in the prefrontal cortex has been found increased in schizophrenia [
15] and decreased in major depression [
19]. No reports are available regarding the hippocampus, but in our series, we found decreased values in cases with psychosis and in those with major depression when compared to MTLE without psychiatric comorbidities in several hippocampal subfields. Interestingly, we have found in other series of patients decreased mossy fiber sprouting in MTLE + P and increased in MTLE + D [
8,
9]. Since mossy fibers are zinc enriched and MT-I/II chelates zinc, cadmium, and copper, it would be expected that hippocampi from MTLE + D have a deficient metal homeostasis and likely zinc excess in neurons and glial cells and in the neuropile. A possible mechanism would be through zinc overflow from serum to brain [
50] due to an inefficient blood-brain barrier in major depression [
51]. In fact, low-serum zinc is a hallmark of major depressive disorders [
52]. Our results of decreased hippocampal MT-I/II in MTLE associated to psychosis can be related to decreased hippocampal zinc levels/mossy fibers in interictal psychosis [
8,
9], as well as in schizophrenia [
53,
54]. Interestingly, MTLE + P patients taking haloperidol showed increased expression of MT-I/II in the inner molecular layer. In the amphetamine animal model of schizophrenia, zinc administration is able to revert behavioral equivalents of positive symptoms [
55], but it is unknown if systemic zinc administration alters hippocampal zinc and/or MT-I/II expression. In fact, zinc is a ligand of the haloperidol-sensitive sigma 2 receptor in the mossy fiber of rats [
56], suggesting that an increased MT-I/II expression in the mossy fiber of MTLE + P patients would facilitate zinc chelation and proper haloperidol binding. We also found a trend to increased MT-I/II in the CA2 of patients who achieved complete seizure remission after surgery, in agreement to the MT-I/II role in the control of excitability [
57]. Likewise, other recent evidences indicate that hippocampal expression of proteins used as markers of full-blown epileptogenesis might be able to predict seizure outcome [
58,
59].
AQP4 is the main water channel in the central nervous system and presents with multifaceted functions. In inflammatory conditions, microglia can release IL-1beta, which in turn induces AQP4 expression in astrocytes [
60]. AQP4 is able to regulate brain response to insults or injury, and also to influence synaptic plasticity and behavior [
61]. In the AQP4 knockout mouse, memory is impaired [
62]. In accordance, our results showed a direct correlation between perivascular AQP4 expression in the Sommer sector and IQ scores. AQP4 participation in synaptic plasticity and cognition occurs together with neurotrophin (NT) receptors [
62]. Of note, NTs and NT receptors are differentially regulated in MTLE with psychiatric comorbidities [
9,
58], which could further change how AQP4 modulates plasticity. For instance, the brain-derived neurotrophic factor (BDNF) is increased in MTLE
W but decreased in MTLE + P [
9]. In addition, tyrosine kinase receptor type 2 (TrkB) (a BDNF receptor) is increased in MTLE + P but not in MTLE
W [
58]. Given that low levels of AQP4 associated with increased BDNF and TrkB or p75 neurotrophin receptor (p75NTR) may result in increased excitability, AQP4 levels near to control levels could be more efficient in controlling excessive excitatory activation trough a BDNF-TrkB or p75NTR loop [
62,
63]. In fact, we found a trend to increased AQP4 in cases with complete seizure remission, thus reinforcing the role of AQP4 in neuron activity. In addition, AQP4 has an important role in K
+ homeostasis [
13,
64], and AQP4 knockout mice have higher seizure threshold but longer seizure duration [
13].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
Conception and design of research (LK); performed research (LK, JEPS, MRM, RCS); analyzed data (JEPS); contributed with reagents/analytic tools and/or important intellectual input (JAA, CGC, JEH); wrote the manuscript (LK, JEPS, JPL). All authors read and approved the final manuscript.