Background
Patients with chronic liver diseases (cirrhosis, hepatitis...) may present hepatic encephalopathy with cognitive and motor alterations including attention deficits, mild cognitive impairment, and reduced spatial memory [
1‐
4]. Hyperammonemia and inflammation are the main contributors to the neurological alterations in hepatic encephalopathy [
5,
6]. Rats with chronic hyperammonemia similar to that present in patients with liver cirrhosis also show cognitive alterations, including impaired spatial learning and memory [
7,
8]. Chronic hyperammonemia per se induces neuroinflammation, which mediates the cognitive alterations in rats [
8,
9]. Neuroinflammation is also a main contributor to cognitive deficits in many chronic (e.g., cirrhosis, diabetes), mental (e.g., schizophrenia), and neurodegenerative (e.g., Alzheimer’s) diseases and in situations such as post-operative cognitive dysfunction or aging [
10‐
15]. Neuroinflammation-induced cognitive impairment is therefore a highly and increasingly prevalent situation with serious health, social, and economic consequences. Neuroinflammation alters cognitive function by altering neurotransmission [
15,
16]. The mechanisms by which neuroinflammation alters neurotransmission may share common aspects in different pathologies. Unveiling these mechanisms may provide therefore the bases to design new therapeutic approaches which could be applied in different highly prevalent pathologies.
Neuroinflammation impairs spatial learning by mechanisms which are beginning to be unveiled.
Sustained expression of IL-1β in hippocampus impairs spatial learning and memory [
17,
18].
Spatial learning is mainly modulated by NMDA and AMPA receptors for glutamate in hippocampus [
19,
20]. A main mechanism modulating glutamatergic neurotransmission and synaptic plasticity in hippocampus is the modulation of membrane expression of AMPA receptors, which is mainly mediated by changes in phosphorylation of the GluA1 subunit in Ser831 and Ser845 and of the GluA2 in Ser880 [
21‐
26].
Neuroinflammation alters membrane expression of glutamate (AMPA, NMDA) and GABA receptors in hippocampus and impairs spatial learning [
17,
27,
28]. Lai et al. [
29] showed that exposure to IL-1β reduces phosphorylation in Ser831 and membrane expression of GluA1 in hippocampal neurons and this was prevented by IL-1Ra, an antagonist of IL-1 receptors. Machado et al. [
30] also found that IL-1β reduced phosphorylation of GluA1 subunit at Ser831 and Ser845 60 min after contextual fear memory reactivation and that intra-hippocampal administration of IL-1β after memory reactivation also induced a decrease in surface expression and total expression of GluA1.
We have recently proposed that impaired spatial learning and memory in rats with hepatic encephalopathy due to portacaval shunts would be due to the increased levels of IL-1β in hippocampus [
28]. We have also shown that both rats with hepatic encephalopathy and rats with hyperammonemia without liver failure show neuroinflammation, with increased levels of IL-1β and other pro-inflammatory markers, and altered membrane expression of AMPA receptor subunits GluA1 and GluA2 in hippocampus [
7,
28,
31].
We hypothesize that hyperammonemia-induced increase in IL-1β in hippocampus would be responsible for the altered membrane expression of GluA1 and GluA2 subunits of AMPA receptors. The aims of this work were to (1) assess if increased IL-1β levels and activation of its receptor (IL-1R) are responsible for the changes in GluA1 and/or GluA2 membrane expression in hyperammonemia and (2) identify the mechanisms by which activation of IL-1R leads to altered membrane expression of GluA1 and GluA2.
The model of chronic hyperammonemia in rats used consisted in administering them an ammonium-containing diet as described in [
32]. Membrane expression and phosphorylation of the GluA1 and GluA2 were analyzed in freshly isolated hippocampal slices from control and hyperammonemic rats. To assess the role of IL-1β in the changes in GluA1 and GluA2, we tested whether blocking the IL-1β receptor with the endogenous antagonist IL-1Ra reverses these changes. We also analyzed the intracellular pathways mediating the effects of IL-1β by assessing the effects of modulating different steps on GluA1 and GluA2 phosphorylation and membrane expression.
Discussion
There is increasing evidence that many pathological situations, including neurodegenerative and chronic diseases, lead to neuroinflammation which, in turn, is a main contributor to cognitive impairment in these situations. Neuroinflammation would impair cognitive function by altering neurotransmission. Unveiling the mechanisms by which neuroinflammation alters neurotransmission would allow identifying pathways and therapeutic targets to try to reverse the alterations in neurotransmission and therefore in cognitive function in different pathological situations.
In this study, we have identified two pathways by which neuroinflammation alters membrane expression of the GluA2 and GluA1 subunits of AMPA receptors, respectively, in hippocampus of hyperammonemic rats. It is shown that hyperammonemia increases IL-1β and activation of its receptor, leading to activation of Src which increases phosphorylation at Tyr1472 and membrane expression of GluN2B, which leads to activation of p38. Activated p38 binds to and reduces phosphorylation at Thr560 and activity of PKCζ. This is associated with reduced phosphorylation of GluA2 at Ser880 and enhanced membrane expression of GluA2. On the other hand, activated Src also increases phosphorylation of PKCδ which enhances phosphorylation of GluN2B at Ser1303, reducing membrane expression of CaMKII and phosphorylation at Ser831 and membrane expression of GluA1.
Modulation of the membrane expression of AMPA receptors plays a main role in synaptic plasticity in hippocampus [
52]. Long-term potentiation (LTP) is mediated by enhanced and long-term depression (LTD) by reduced membrane expression of AMPA receptors [
53,
54]. LTP in hippocampus is considered the bases for spatial learning and memory [
55]. Both LTP in hippocampus [
56] and spatial learning and memory [
7] are impaired in hyperammonemic rats. Altered modulation of GluA1 and GluA2 membrane expression in hyperammonemic rats would be a main contributor to the impairment of LTP and spatial learning.
We show here that the alterations in membrane expression of GluA1 and GluA2 in hyperammonemic rats are a consequence of neuroinflammation and of enhanced activation of IL-1 receptor by increased levels of IL-1β. It has been already shown that high levels of IL-1β impair LTP [
57]. Altogether, these data support that in hyperammonemic rats, (and likely in other pathological situations) increased levels of IL-1β in hippocampus alter membrane expression of GluA1 and GluA2 subunits of AMPA receptors, which would lead to impairment of LTP and of spatial learning and memory.
We also identify the intracellular signal transduction pathways by which activation of IL-1 receptor leads to changes in phosphorylation and to opposite effects on membrane expression of GluA1 and GluA2. These pathways are presented in Fig.
6 and in the graphical abstract.
Hyperammonemia increases IL-1β, enhancing activation of IL-1 receptor. This leads to activation of Src, reflected in increased phosphorylation of Tyr416. These steps are common to the pathways leading to altered membrane expression of the GluA1 and GluA2 subunits. The changes in membrane expression of both GluA1 and GluA2 are reversed by blocking the IL-1 receptor with IL-1Ra or by inhibiting Src with PP2, thus confirming the contribution of these steps to the changes in membrane expression.
However, after Src activation, the pathways diverge. The enhanced activity of Src in hyperammonemic rats results in increased phosphorylation of GluN2B at Tyr1472 and membrane expression of GluN2B, which leads to activation of p38. Activated p38 binds to and reduces phosphorylation at Thr560 and activity of PKCζ, thus resulting in reduced phosphorylation at Ser880 and enhanced membrane expression of GluA2 (Fig.
6a). The changes in membrane expression of GluA2 are reversed by blocking the GluN2B-containing NMDA receptors with ifenprodil or inhibiting p38 with SB239063, thus confirming the contribution of these steps to the changes in membrane expression of GluA2 (Fig.
6a). Some reports in the literature support the existence of the steps proposed in Fig.
6a. It has been already shown that Src phosphorylates GluN2B at Tyr1472 and this increases its membrane expression [
42] that enhanced membrane expression of GluN2B leads to enhanced phosphorylation and activity of the MAP kinase p38 [
43] and that activated p38 binds to PKCζ and this prevents auto-phosphorylation of PKCζ at Thr560, thus reducing its activity [
46]. We show here that all these steps are induced sequentially in hyperammonemic rats by activation of IL-1 receptor, leading to increased membrane expression of GluA2.
Moreover, these steps would occur in neurons and not in astrocytes. This is supported by the report of Srinivasan et al. [
58], who showed that IL-1β activates the p38 signaling pathway in hippocampal neurons, in contrast to the activation of NF-kB in hippocampal astrocytes, demonstrating cell type-specific signaling responses to IL-1 in the brain and yielding distinct functional responses. However, Srinivasan et al. [
58] did not tested if activation of p38 by IL-1β is a direct effect or it is mediated by some previous steps. We show here that increased phosphorylation of p38 is reduced by blocking the IL-1 receptor, by inhibiting Src, or by blocking the NR2B subunit of NMDA receptors, supporting that activation of p38 by IL-1β is mediated by Src and NR2B.
Altered membrane expression of GluA1 is also mediated by activation of IL-1 receptor and Src, but after this step, the pathway is different than for GluA2. Increased activity of Src in hyperammonemic rats also activates PKCδ which enhances phosphorylation of GluN2B at Ser1303, reducing membrane expression of CaMKII and phosphorylation at Ser831 and membrane expression of GluA1. These changes are reversed by blocking the IL-1 receptor with IL-1Ra and by inhibiting Src with PP2 or PKCδ with rottlerin, thus supporting the contribution of the pathway depicted in Fig.
6b in the changes in membrane expression of GluA1. Some reports in the literature support the existence of the steps proposed in Fig.
6b. It has already been reported that Src phosphorylates PKCδ at Tyr311, enhancing its activity [
59], and that enhancing phosphorylation GluN2B at Ser1303 reduces membrane expression of CaMKII [
49]. We show here that all these steps are induced sequentially in hyperammonemic rats by activation of IL-1 receptor, leading to reduced membrane expression of GluA1.
We show that the steps of the pathway summarized in Fig.
6a are induced sequentially in hyperammonemic rats by activation of IL-1 receptor, leading to increased membrane expression of GluA2. Blocking IL-1β receptor prevents all subsequent steps, indicating that this activation is in the origin of the pathway activation. Inhibiting Src with PP2 prevents changes in phosphorylation and membrane expression of NR2B, in phosphorylation of p38 and of PKCζ and phosphorylation and membrane expression of GluA2, indicating that activation of Src precedes all these steps. Similarly, blocking NR2B with ifenprodil prevents changes in phosphorylation of p38 and of PKCζ and phosphorylation and membrane expression of GluA2, indicating that changes in NR2B precede these steps. Finally, inhibiting p38 with SB239063 prevents changes in phosphorylation and membrane expression of GluA2. Altogether, these results show that the hyperammonemia induces sequential activation of the pathway depicted in Fig.
6a to modulate membrane expression of GluA2. Similarly, hyperammonemia induces sequential activation of the pathway depicted in Fig.
6b to modulate membrane expression of GluA1. Although the activation of these pathways occurs sequentially, it occurs very rapidly. The phosphorylation and membrane expression of GluA1 and GluA2 AMPA receptor subunits is very dynamic and may change rapidly in response to synaptic activity or other stimuli. The data reported here show that, under the conditions used in the present work, hyperammonemia alters these pathways as summarized in Fig.
6.
The above data demonstrate that neuroinflammation in hippocampus induced by hyperammonemia alters glutamatergic neurotransmission by altering membrane expression of AMPA and NMDA receptor subunits. Moreover, this is mediated by increased levels of IL-1β and of activation of IL-1 receptor and we identify the intracellular pathways involved.
It is noteworthy that neuroinflammation may also alter AMPA receptors by other mechanisms. TNF-α also may alter membrane expression of AMPA receptor subunits. Moreover, the effects of IL-1β and of TNF-α on membrane expression of AMPA receptors are the opposite. IL-1β reduces membrane expression of the GluA1 subunit while TNF-α increases GluA1 but reduces GluA2 membrane expression [
29,
60‐
62]. This indicates that different types or grades of neuroinflammation may lead to different alterations in AMPA receptor membrane expression depending on the prevalence of IL-1β or TNF-α effect [
29,
60‐
62].
TNF-α selectively enhances membrane expression of GluA1 in hippocampal neurons and the proportion of GluA2-lacking receptors, resulting in AMPA receptors with different properties becoming calcium-permeable, inwardly rectifying and inhibited by polyamines [
61]. In vivo nanoinjection of TNF-α in rats also increases synaptic expression of the GluA1 subunits with a concurrent decrease in the GluA2 subunit [
62]. In contrast, IL-1β at high concentrations reduces membrane expression of GluA1 [
29]. The results reported here show that, in hyperammonemic rats, increased levels of IL-1β reduces GluA1 and enhances GluA2 membrane expression through activation of IL-1 receptor.
It is also noteworthy that, although the effects of IL-1β and TNF-α on membrane expression of AMPA receptors are the opposite, both increased levels of IL-1β and of TNF-α result in altered glutamatergic neurotransmission by altering AMPA receptor function. Sustained overexpression of either IL-1β or TNF-α impairs hippocampal LTP [
63]. Increased levels of IL-1β also impair spatial learning and memory [
17,
64]. Some reports suggest that TNF-α also impairs spatial learning [
65,
66].
A potential limitation of the present work is that the effects reported could be limited to a certain period of time in the progression of chronic hyperammonemia. The present work has been performed at 4–5 weeks of hyperammonemia. However, the type or intensity of neuroinflammation could be different at longer times of hyperammonemia, triggering other mechanisms. The above reports indicate that neuroinflammation may alter hippocampal neurotransmission and spatial learning by different mechanisms depending on the type and grade of neuroinflammation, for example on the relative contribution of the increases in IL-1β and TNF-α [
17,
29,
60‐
66]. This may explain the discrepancy between the effects induced by 4–5 weeks of hyperammonmeia reported here and those reported in [
7] showing increased GluA1 and reduced GluA2 membrane expression at 8 weeks of hyperammonemia. We have preliminary results showing that neuroinflammation in hyperammonemic rats is dynamic and changes with time, resulting in a different pattern of inflammatory markers at different types. Dynamic changes in neuroinflammation have been also reported in situations such as Parkinson’s disease [
67], stroke [
68,
69], ischemia [
70], amyotrophic lateral sclerosis, AIDS, and multiple sclerosis [
71]. These dynamic changes in neuroinflammation and its consequences on neurotransmission suggest that the treatments to reverse neuroinflammation-induced cognitive impairment should be different depending on the type and grade of neuroinflammation reached, which would be different in different pathological situations.
We show that in rats with chronic moderate hyperammonemia, similar to that present in patients with liver cirrhosis, neuroinflammation in hippocampus alters membrane expression of AMPA receptors mainly through activation of IL-1 receptor by increased levels of IL-1β. Blocking IL-1 receptor with the endogenous antagonist IL-1Ra reverses completely the alterations in AMPA receptor membrane expression. It has been shown that IL-1Ra also prevents the impairment of LTP by IL-1β [
63]. This suggests that blocking this receptor would also restore spatial learning impaired by overexpression of IL-1β in different in pathological situations associated with neuroinflammation, including hyperammonemic rats, and possibly, patients with minimal hepatic encephalopathy.