Hyperammonemia alters membrane expression of GluA1 and GluA2 subunits of AMPA receptors in hippocampus by enhancing activation of the IL-1 receptor: underlying mechanisms

Male Wistar rats (120–140 g at the beginning of the diet, Charles River Laboratories, Barcelona, Spain) were made hyperammonemic by feeding them a diet containing standard diet supplemented with ammonium acetate as in [32]. In the present work, the diet contained 25% of ammonium acetate instead of 20% as in [32]. This change was made because now we house the rats in ventilated racks which were not used in [32]. These racks extract some ammonia, and the increase in ammonium acetate in the diet was necessary to reproduce the ammonia levels in blood obtained previously in un-ventilated cages using 20% ammonium acetate. The ammonium-containing diet was prepared as described in [33] (but using 25% ammonium acetate) and increased blood ammonia levels around threefold, an increase similar to that found in patients with liver cirrhosis. Rats remain hyperammonemic for long periods of time and reproduce many cognitive and motor alterations present in cirrhotic patients with hepatic encephalopathy and has allowed identifying some underlying mechanisms [7‐9, 34‐36]. Experiments were performed after 4–5 weeks in the ammonium diet, when the rats are 10–11 weeks old. The experiments were approved by the the Comite de Etica y Bienestar en Experimentacion Animal, Prince Felipe Research Center-Consellería de Agricultura, Generalitat Valenciana, and carried out in accordance with the European Communities Council Directive (86/609/EEC).

In each experiment, four rats (two control and two hyperammonemic rats) were decapitated at 4–5 weeks of hyperammonemia and their brains quickly transferred to a plate where they were carefully dissected, separating the two hemispheres by cutting through the midline and extracting both hippocampi using thin spatulas. Once dissected, hippocampi were immersed immediately into ice-cold Krebs buffer (in mmol/L): NaCl 119, KCl 2.5, KH₂PO₄ 1, NaHCO₃ 26.2, CaCl₂ 2.5, and glucose 11, aerated with 95% O₂ and 5% CO₂ at pH 7.4. After that, hippocampi were placed longitudinally on a manual chopper and cut to obtain transverse slices (400 μm). Slices were transferred to incubation wells (15 slices in each well, from the two rats per group) without any distinction between dorsal or ventral hippocampus and incubated for 20 min at 35.5 °C in Krebs buffer for stabilization. One hundred nanogram per millileter of IL1Ra (R&D Systems cat# 1545-RA-025, Minneapolis, USA), 100 μM ifenprodil (Abcam cat# ab120111, Cambridge, MA), 20 μM SB239063 (Tocris cat# 1962, Bristol, UK), 10 μM rottlerin (Sigma cat# R5648, Darmstadt, Germany), or 10 μM PP2 (Sigma cat# P0042, Darmstadt, Germany) was added to the incubation buffer during 30 min to specifically inhibit IL-1 receptor, GluN2B, p38, PKCδ, and Src activities, respectively. After the treatments, slices were collected and homogenized by sonication for 20 s in a buffer (Tris-HCl 66 mM pH 7.4, SDS 1%, EGTA 1 mM, glycerol 10%, leupeptin 0.2 mg/mL, NaF 1 mM, Na orto-vanadate 1 mM). Samples were subjected to immunoblotting as in Felipo et al. [37], using antibodies against IL-1β (1:500, cat# AF-501-NA) from R&D Systems; Src (1:1000, cat# ab47405), Src phosphorylated at Tyr416 (1:1000, cat# ab40660), GluN2B phosphorylated at Ser1303 (1:1000, cat# ab81271), GluA2 phosphorylated at Ser880 (1:2000, cat# ab52180), and PKCζ phosphorylated at Thr560 (1:1000, cat# ab59412) from Abcam; GluN2B (1:1000, cat# 06-600), GluN2B phosphorylated at Tyr1472 (1:1000, cat# AB5403), GluA1 (1:1000, cat# 04-855), GluA1 phosphorylated at Ser831 (1:1000, cat# 04-823), and GluA2 (1:2000, cat# AB1768) from Millipore (Darmstadt, Germany); p38 (1:1000, cat# 9212) and p38 phosphorylated at Thr180/Tyr182 (1:500, cat# 9211) from Cell Signaling (Leiden, Netherlands); and PKCζ (1:2000, cat# sc-17,781) from Santa Cruz (Dallas, TX). As a control for protein loading, the same membranes used to quantify the amount of proteins were incubated with an antibody against Actin (1:5000, cat# ab6276) from Abcam. Secondary antibodies were anti-rabbit (cat# A8025), anti-goat (cat# A7650), or anti-mouse (cat# A3562) IgG, 1:4000 dilution conjugated with alkaline phosphatase from Sigma (St. Louis, MO). The images were captured using the ScanJet 5300C (Hewlett-Packard, Amsterdam, Netherlands), and band intensities quantified using the Alpha Imager 2200, version 3.1.2 (AlphaInnotech Corporation, San Francisco). Phosphorylation levels were normalized to the total amount of the respective proteins.

Membrane surface expression of the GluA1 and GluA2 subunits of AMPA receptors, GluN2B subunit of NMDA receptors, and CaMKII (1:1000 primary antibody from Thermo Fisher cat# MA1-048, Waltham, MA) in whole hippocampal slices was analyzed at 4–5 weeks of hyperammonemia as described by Boudreau and Wolf [38], by cross-linking with BS3 (bis(sulfosuccinimidyl) suberate, Pierce cat# 21580, Rockford, IL). After the treatments (see above), slices were added to tubes containing ice-cold Krebs buffer with or without 2 mM BS3 and incubated for 30 min at 4 °C with gentle shacking. Cross-linking was terminated by quenching the reaction with 100 mM glycine (10 min, 4 °C). The slices were homogenized by sonication for 20 s. Samples treated or not with BS3 were analyzed by western blot as describe above. The surface expression of each subunit was calculated as the difference between the intensity of the bands without BS3 (total protein) and with BS3 (non-membrane protein) as described by Cabrera-Pastor et al. [39].

Results are expressed as mean ± standard error. All statistical analyses were performed using the software program GraphPad Prism. Normality was assessed using the D’Agostino and Pearson Omnibus test and the Shapiro-Wilk normality tests. Differences in variances of normally distributed data were assessed using Bartlett’s test. Data were analyzed by a parametric two-way analysis of variance (ANOVA) followed by Bonferroni’s post hoc test to determine the individual and interaction effects between hyperammonemia and/or treatments on membrane expression and phosphorylation levels of proteins [40]. A confidence level of 95% was accepted as significant.

Membrane expression of the GluA1 and GluA2 subunits of AMPA receptors was altered in opposite ways in hippocampus of hyperammonemic rats. Membrane expression of GluA1 was reduced (p < 0.05) to 86 ± 6% in control rats (Fig. 1a), while membrane expression of GluA2 was increased (p < 0.001) to 136 ± 7% in control rats (Fig. 1b).

Changes in membrane expression were associated with changes in phosphorylation. Phosphorylation of GluA1 in Ser831 was reduced (p < 0.001) in hyperammonemic rats to 73 ± 7% in control rats (Fig. 1c), and phosphorylation of GluA2 in Ser880 was reduced (p < 0.05) to 84 ± 4% in control rats (Fig. 1d).

As it has been shown that IL-1β modulates membrane expression of AMPA receptors, we assessed its possible contribution to the effects of hyperammonemia. In hyperammonemic rats, the content of IL-1β in hippocampus was increased to 129 ± 8% in control rats (p < 0.05). To assess if changes in membrane expression of GluA1 and GluA2 are due to enhanced activation of IL-1 receptor, we tested if blocking this receptor with the endogenous antagonist IL-1Ra reverses these changes. Perfusion of IL-1Ra increased (p < 0.001) membrane expression of GluA1 in hippocampal slices of hyperammonemic rats, reaching 124 ± 7% in control rats (Fig. 1a), in parallel with an increase (p < 0.001) of phosphorylation of GluA1 in Ser831 to 140 ± 16% in control rats (Fig. 1c).

IL-1Ra reduced (p < 0.001) membrane expression of GluA2 to the same levels in control rats (Fig. 1b) in parallel with a normalization of phosphorylation of GluA2 at Ser880 (Fig. 1d).

These data show that blocking IL-1 receptor with IL-1Ra reverses the changes induced by hyperammonemia on GluA1 and GluA2 phosphorylation and membrane expression, suggesting that enhanced activation of IL-1 receptor is responsible for these alterations in hippocampus of hyperammonemic rats.

We then looked for the intracellular pathways mediating the changes in membrane expression induced by enhanced activation of IL-1Ra. As activation of IL-1 receptor leads often to increased phosphorylation and activity of Src kinase [41], we tested whether it is increased in hyperammonemic rats. Phosphorylation of Src at Tyr416 is increased (p < 0.05) in hyperammonemic rats to 118 ± 8% in control rats, and blocking IL-1 receptor with IL-1Ra normalizes it, returning to 77 ± 13% in control rats (Fig. 2a).

To assess if enhanced activation of Src mediates the changes in membrane expression of GluA1 and GluA2 in hyperammonemic rats, we tested if inhibiting Src with PP2 reverses these changes.

Perfusion of PP2 increased (p < 0.001) membrane expression of GluA1 in hippocampal slices of hyperammonemic rats, reaching 134 ± 20% in control rats (Fig. 2b), in parallel with an increase (p < 0.05) of phosphorylation of GluA1 in Ser831 to 103 ± 13% in control rats (Fig. 2c).

PP2 reduced (p < 0.001) membrane expression of GluA2 to the same levels in control rats (Fig. 2d) in parallel with an increase of phosphorylation of GluA2 at Ser880 (Fig. 2e).

These data suggest that enhanced activation of Src contributes to altered phosphorylation and membrane expression of GluA1 and GluA2 in hippocampus of hyperammonemic rats.

On the bases of the knowledge in the literature (see the “Discussion” section), we hypothesized that Src would be altering GluA1 and GluA2 membrane expression by modulating the GluN2B subunit of NMDA receptors. Src phosphorylates GluN2B at Tyr1472 and this increases its membrane expression [42].

As shown in Fig. 3, hyperammonemia increases phosphorylation (p < 0.05) of GluN2B at Tyr1472 to 122 ± 10% in control rats (Fig. 3a) and membrane expression (p < 0.001) of GluN2B to 151 ± 10% in control rats (Fig. 3b).

Blocking IL-1 receptor with IL-1Ra normalized both phosphorylation (Fig. 3a) and membrane expression (Fig. 3b) of GluN2B in hyperammonemic rats.

Similar effects were obtained in experiments using the Src inhibitor PP2, which normalized both phosphorylation (Fig. 3c) and membrane expression (Fig. 3d) of GluN2B in hyperammonemic rats.

These data support that enhanced activation of IL-1 receptor in hyperammonemia leads to increased activity of Src which phosphorylates and enhances membrane expression of GluN2B.

It has been reported that enhanced membrane expression of GluN2B leads to enhanced phosphorylation and activity of the MAP kinase p38 [43]. We therefore assessed whether phosphorylation of p38 was increased in hippocampus of hyperammonemic rats. Phosphorylation of p38 was increased (p < 0.05) to 124 ± 10% in control rats (Fig. 4) and was normalized by treatment with IL-1Ra (Fig. 4a), the Src inhibitor PP2 (Fig. 4b), or with ifenprodil (Fig. 4c), a selective antagonist of NMDA receptors containing the GluN2B subunit [44].

To assess if enhanced activation of p38 mediates the changes in membrane expression of GluA1 and GluA2 in hyperammonemic rats, we tested if inhibiting p38 with SB239063 reverses these changes.

Perfusion of SB239063 reduced (p < 0.01) membrane expression of GluA2 to the same levels in control rats (Fig. 4d) in parallel with an increase of phosphorylation of GluA2 at Ser880 (Fig. 4e). However, SB239063 did not normalized phosphorylation (Fig. 4g) or membrane expression (Fig. 4f) of the GluA1 subunit.

These data suggest that enhanced activation of p38 contributes to altered phosphorylation and membrane expression of GluA2 but not of GluA1in hippocampus of hyperammonemic rats.

The GluA2 subunit is phosphorylated at Ser880 by protein kinase C (PKC) [45]. The above data suggest that enhanced p38 activity would reduce the activity of some isoform of PKC, resulting in reduced phosphorylation of Ser880 and increased membrane expression of GluA2. It has been shown that activated p38 binds to PKCζ and this prevents auto-phosphorylation of PKCζ at Thr560, thus reducing its activity [46].

We therefore tested whether phosphorylation of PKCζ at Thr560 is altered in hyperammonemic rats. As shown in Fig. 5, phosphorylation of PKCζ at Thr560 was reduced (p < 0.05) to 83 ± 6% in control rats and was normalized by treatment with IL-1Ra (Fig. 5a), the Src inhibitor PP2 (Fig. 5b), or by blocking GluN2B with ifenprodil (Fig. 5c).

Altogether, the above data allow proposing the pathway shown in Fig. 6a for the mechanism by which hyperammonemia increases membrane expression of the GluA2 subunit of AMPA receptors in hippocampus. Hyperammonemia increases IL-1β, enhancing activation of IL-1 receptor. This leads to activation of Src, reflected in increased phosphorylation of Tyr416. Src in turn enhances phosphorylation of GluN2B at Tyr14721 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.

The above results show that changes in phosphorylation at Ser831 and in membrane expression of GluA1 are also mediated by enhanced activation of IL-1 receptor and of Src, but not by the p38-PKC pathway. As Ser831 may be phosphorylated by PKC or by CaMKII [47, 48], we hypothesized that changes in GluA1 would be mediated by CaMKII.

We then assessed if hyperammonemic alters the amount of CaMKII in the membrane. The amount of CaMKII in the membrane is strongly reduced (p < 0.01) in membrane of hippocampal slices from hyperammonemic rats, to 54 ± 10% in control rats (Fig. 7). Moreover, this decrease is reversed by treatment with IL-1Ra (Fig. 7a) or by inhibiting Src with PP2 (Fig. 7b).

One of the mechanisms of modulation of membrane expression of CaMKII is phosphorylation of GluN2B at Ser1303. Enhancing this phosphorylation reduces membrane expression of CaMKII [49]. Phosphorylation of GluN2B at Ser1303 is increased in hyperammonemic rats (Fig. 7c, d). Moreover, blocking IL-1 receptor with IL-1Ra (Fig. 7c) or inhibiting Src with PP2 (Fig. 7d) normalizes phosphorylation of GluN2B at Ser1303. This suggests that increased phosphorylation of GluN2B at Ser1303 in hyperammonemic rats leads to reduced membrane expression of CaMKII which, in turn, reduces phosphorylation at Ser831 and membrane expression of GluA1.

Increased phosphorylation of GluN2B at Sr1303 is mediated by IL-1 receptor and by Src. However, Src is a tyrosine kinase, and may not phosphorylate Ser residues. This suggests that a serine-threonine kinase would mediate the effects of Src on phosphorylation of GluN2B at Ser1303 [50]. It has been reported that Src phosphorylates PKCδ at Tyr311, enhancing its activity [51].

To assess if increased activity of PKCδ mediates the increase in phosphorylation of GluN2B at Ser1303 in hyperammonemic rats, we analyzed whether a specific inhibitor of PKCδ (rottlerin) reverses this increase. Treatment with rottlerin completely reversed the increase in phosphorylation of GluN2B at Ser1303 in hyperammonemic rats (Fig. 8a). Moreover, rottlerin also increased membrane expression of CaMKII (Fig. 8b) even above (178 ± 75%) in control rats. The increase of CaMKII in the membrane was associated with an increase, also above control rats, of phosphorylation of GluA1 at Ser831 (142 ± 11% in controls, Fig. 8c) and of membrane expression of GluA1 (140 ± 37% in controls, Fig. 8d).

The above data allow proposing the pathway shown in Fig. 6b for the mechanism by which hyperammonemia reduces membrane expression of the GluA1 in hippocampus. Hyperammonemia increases IL-1β, enhancing activation of IL-1 receptor. This leads to activation of Src, reflected in increased phosphorylation of Tyr416. Src in turn activates PKCδ which enhances phosphorylation of GluN2B at Ser1303, reducing membrane expression of CaMKII and phosphorylation at Ser831 and membrane expression of GluA1.