Altered regulation of Akt signaling with murine cerebral malaria, effects on long-term neuro-cognitive function, restoration with lithium treatment

Abstract

Neurological and cognitive impairment persist in more than 20% of cerebral malaria (CM) patients long after successful anti-parasitic treatment. We recently reported that long term memory and motor coordination deficits are also present in our experimental cerebral malaria model (ECM). We also documented, in a murine model, a lack of obvious pathology or inflammation after parasite elimination, suggesting that the long-term negative neurological outcomes result from potentially reversible biochemical and physiological changes in brains of ECM mice, subsequent to acute ischemic and inflammatory processes. Here, we demonstrate for the first time that acute ECM results in significantly reduced activation of protein kinase B (PKB or Akt) leading to decreased Akt phosphorylation and inhibition of the glycogen kinase synthase (GSK3β) in the brains of mice infected with Plasmodium berghei ANKA (PbA) compared to uninfected controls and to mice infected with the non-neurotrophic P. berghei NK65 (PbN). Though Akt activation improved to control levels after chloroquine treatment in PbA-infected mice, the addition of lithium chloride, a compound which inhibits GSK3β activity and stimulates Akt activation, induced a modest, but significant activation of Akt in the brains of infected mice when compared to uninfected controls treated with chloroquine with and without lithium. In addition, lithium significantly reversed the long-term spatial and visual memory impairment as well as the motor coordination deficits which persisted after successful anti-parasitic treatment. GSK3β inhibition was significantly increased after chloroquine treatment, both in lithium and non-lithium treated PbA-infected mice. These data indicate that acute ECM is associated with abnormalities in cell survival pathways that result in neuronal damage. Regulation of Akt/GSK3β with lithium reduces neuronal degeneration and may have neuroprotective effects in ECM. Aberrant regulation of Akt/GSK3β signaling likely underlies long-term neurological sequelae observed in ECM and may yield adjunctive therapeutic targets for the management of CM.

Figure 1. Pathology observed during ECM.

A–C: While control mice and PbN-infected mice demonstrated no hemorrhage in the brain, PbA infection resulted in significant degrees of hemorrhage in the cerebellum (F(_{2, 11)} = 7.84; p<0.01) (B- arrows). D–F: PbA infection also resulted in significantly higher hemorrhage (E- arrow) in the areas associated with cognitive function, e.g. cerebral cortex (D–F), corpus callosum, hippocampus, fornix and thalamus (F_{(2, 12)} = 5.29; p<0.05). Semi-quantitative determination of hemorrhage as previously published . n = 4 Con, 7 PbA, 3 PbN. Con = control, PbA = P. berghei ANKA infected mice, PbN = P. berghei NK65 infected mice.

Figure 2. Representative immunoblots of GSK3β and Akt expression.

(A) Brain lysates were probed with antibodies to phospho-GSK3β at Ser9 (pGSK3β (S9)), total GSK3β, phospho-Akt at Ser473 (pAkt (S473)), and total Akt. (B) Densitometry measurements of immunoblots demonstrated that PbA infection resulted in significantly lower expression of pGSK3β (S9) (F_{(2, 18)} = 19.87; p<0.001) when compared to uninfected control mice or to mice infected with PbN. PbA-infected mice demonstrated a 32% decrease in phospho-GSK3β (S9) expression compared to controls (n = 8 Con, 7 PbA) and a 23% decrease compared to mice infected with PbN (n = 7 PbA, 6 PbN). This was associated with general effects of infection condition on the mean protein expression of total GSK3β (F_{(2, 18)} = 3.63; p<0.05); however, post-hoc Tukey's multiple comparison did not demonstrate any specific effects of individual conditions when comparing the different treatment groups. (C) There was a corresponding 42% decrease in the expression of pAkt (S473) in PbA-infected mice compared to control, and a 48% decrease compared to PbN-infected mice, though this did not reach statistical significance (F_{(2, 17)} = 2.79; p = 0.09). Interestingly, there were significant effects of PbA infection on the expression of total Akt (F_{(2, 18)} = 5.84; p<0.05), with Tukey's post-hoc analysis demonstrating significant reduction due to PbA when compared to uninfected controls. No differences were observed between PbA and PbN infected mice or between control and PbN infected mice. Total Akt and GSK3β protein expression levels were normalized to GDI, and absolute total protein levels were used to normalize respective phosphorylated protein levels. All densitometry measurements are illustrated as a percentage of corresponding measurements in uninfected controls. Values plotted as mean ± standard error (SEM). *p<0.05, **p<0.001. Con = control, PbA = P. berghei ANKA infected mice, PbN = P. berghei NK65 infected mice. GDI was used as loading control.

Figure 3. Abnormal tau expression in ECM.

Aberrantly phosphorylated tau was evident in PbA infected mice when compared to control and PbN infected mice. (A) Brain lysates were probed with antibodies to PHF-1 which stains for phosphorylated tau at Ser396/404, and MC-1 which recognized misfolded tau protein. (B) PbA infected mice demonstrated a 48% higher in PHF-1 immunoreactivity compared to controls and a 134% more compared to PbN. Significant group effects on the means was demonstrated by one-way ANOVA (F_{(2, 8)} = 14.85; p<0.01) with significant mean differences between PbA mice and both control and PbN mice using post-hoc Tukey's multiple comparison test. Tukey's test did not demonstrate a significant effect of PbN infection on the mean PHF-1 expression when compared to control mice. (C) PbA-infected mice displayed a 131% more MC-1 expression compared to controls and 137% higher expression compared to PbN-infected mice. One way ANOVA demonstrated significant group effects in the expression of MC-1 (F_{(2, 8)} = 51.29; p<0.001) with post-hoc Tukey's test demonstrating significant effect of PbA infection on MC-1 expression when compared to either control or PbN mice, but no effect of PbN infection when compared to controls. Densitometry measurements are illustrated as a percentage of corresponding measurements in uninfected controls. Values plotted as mean ± SEM. *p<0.05, **p<0.001. Con = control, PbA = P. berghei ANKA infected mice, PbN = P. berghei NK65 infected mice. Ponceau was used as loading control.

Figure 4. Immunofluorescence staining for phospho-tau (Red) and phospho-GSK3β (S9) (Green) in the brainstem.

PbA-infected mice displayed an increase in tau-positive neurons when compared to control mice or mice infected with PbN and an overall decrease or absence of intra-nuclear staining of pGSK3β (S9) in those neurons.

Figure 5. Expression of pAkt (S473) after chloroquine and lithium treatment.

(A–B) expression of Akt phosphorylated at Ser473 reached control levels in ECM mice treated with NaCl. Tukey's analysis demonstrated a modest but significant increase in the expression of pAkt (S473) due to lithium treatment of ECM mice when compared to control (p<0.05), although a one-way ANOVA did not show significant overall group effects on the mean protein expression (F_{(2, 18)} = 3.361; p = 0.058). (A, C) There were no differences in total Akt expression between the groups. For the purposes of analysis, NaCl and LiCl treated control groups were combined as there were no significant differences in the means and no effect of treatment on mean protein expression. Total Akt protein expression levels were normalized to GDI. Absolute total Akt total protein levels were used to normalize respective pAkt (S473) levels. All densitometry measurements are illustrated as a percentage of corresponding measurements in uninfected controls. Values plotted as mean ± SEM. n = 4 Na Con, 6 Na PbA, 6 Li PbA, 5 Li Con. τ = significant Tukey's multiple comparison analysis (p<0.05). Na Con = NaCl treated control, Na PbA = NaCl treated P. berghei ANKA infected mice, Li PbA = LiCl treated P. berghei ANKA infected mice, Li Con = LiCl treated control. GDI was used as loading control.

Figure 6. Expression of phospho-GSK3β and tau after chloroquine and lithium treatment.

(A–B) PbA infected mice treated with NaCl demonstrated 220% more phospho-GSK3β expression than controls, and PbA infected mice treated with LiCl had 203% higher expression of phospho-GSK3β. One-way ANOVA demonstrated a significant effect of treatment condition (F_{(2, 18)} = 54.82; p<0.0001) with post-hoc Tukey's test confirming significant different means between uninfected control and NaCl-treated ECM mice, and between uninfected controls and LiCl treated mice, but not between the two PbA groups. (C) There was no difference in the expression of total GSK3 between any of the groups. (A, D–E) There were no differences in phosphorylated tau or total tau protein expression between any of the groups. Phosphorylated tau levels returned to normal with CQ treatment with or without adjunctive lithium, and there were no group effects of treatment condition on protein expression of phosphorylated or total tau. For analysis, NaCl and LiCl treated control groups were combined as there were no significant differences in the means and no effect of treatment on mean protein expression. Total GSK3β and tau protein expression levels were normalized to GDI, and absolute total protein levels were used to normalize respective phosphorylated protein levels. Densitometry measurements are illustrated as a percentage of corresponding measurements in uninfected controls. Values plotted as mean ± SEM. n = 4 Na Con, 6 Na PbA, 6 Li PbA, 5 Li Con. **p<0.001. Na Con = NaCl treated control, Na PbA = NaCl treated P. berghei ANKA infected mice, Li PbA = LiCl treated P. berghei ANKA infected mice, Li Con = LiCl treated control. GDI was used as loading control.

Figure 7. Immunofluorescence staining with NeuN (Red) and phospho-GSK3β (S9) (Green) in the brainstem.

PbA-infected mice displayed different patterns in the distribution of phospho-GSK3β within neuronal cells (NeuN-positive), with a lack of phospho-GSK3β in neuronal nuclei in contrast to LiCl treated ECM mice and uninfected control mice treated either with or without LiCl. Na Con = NaCl treated control, Na PbA = NaCl treated P. berghei ANKA infected mice, Li PbA = LiCl treated P. berghei ANKA infected mice, Li Con = LiCl treated control, NeuN = neuronal nuclear antibody.

Figure 8. Effects of lithium on cognitive function in chloroquine treated mice.

(A) There were significant group effects on mouse performance in the object placement test of spatial memory (F_{(3, 34)} = 13.56; p<0.001), with post-hoc Tukey's multiple comparison test confirming our previous observations that PbA infection resulted in significantly lower preference scores than uninfected controls despite CQ treatment (Na PbA: 51.8±3.05 v. Na Con: 70±1.6 v. Li Con: 74.2±3.11). Adjunctive treatment with LiCl resulted in the prevention of spatial memory impairment as displayed in the object placement test, as PbA infected mice treated with LiCl scored significantly higher in the object placement test than NaCl treated PbA mice (Li PbA: 71.44±2.48 v. Na PbA: 51.8±3.05). (B) Moreover, a significantly higher proportion of NaCl-treated ECM mice showed spatial memory deficits (Na PbA: 75% v. Na Con: 0% v. Li Con: 0%), with a significantly lower proportion of LiCl-treated ECM mice exhibiting spatial memory deficits (Li PbA: 8.3% v. Na PbA: 75%). (A–B) There were no significant differences in the preference scores between the LiCl-treated ECM group and either uninfected control group, and no differences between sodium chloride and lithium chloride treated control mice. (C) Correspondingly, one-way ANOVA (F_{(3, 34)} = 7.951; p<0.001) demonstrated significant group effects on visual memory in the mice, with the lack of adjunctive lithium therapy in CQ-treated PbA mice resulting in significantly lower mean preference score in the object recognition test after a 45 min retention interval compared to uninfected controls (Na PbA: 51.92±3.42 v. Na Con: 69.8±1.1 v. Li Con: 72.59±3.41). Lithium treatment resulted in the prevention of visual memory impairment as ECM mice treated with LiCl had significantly higher preference scores in the object recognition test than mice receiving NaCl (Na PbA: 51.92±3.42 v. Li PbA: 69.01±3.28). (D) In addition, a significantly higher proportion of NaCl-treated ECM mice had deficits in visual memory compared to LiCl-treated ECM mice (Na PbA: 67% v. Li PbA: 8.3%) or uninfected, NaCl and LiCl treated control mice (Na PbA: 67% v. Na Con: 0% v. Li Con: 0%). (C–D) Just as with the object placement test, there were no differences in the preference scores of LiCl-treated ECM mice or in the proportion of LiCl ECM mice with a preference in the object recognition tests compared to either of the control groups. In addition, both sodium and lithium treated control mice performed similarly in the object recognition tests. n = 4 Na Con; 12 Na PbA; 12 Li PbA; 10 Li Con. *p<0.05, **p<0.01, τ = significant ANOVA F-test (p<0.05). Na Con = NaCl treated control, Na PbA = NaCl treated P. berghei ANKA infected mice, Li PbA = LiCl treated P. berghei ANKA infected mice, Li Con = LiCl treated control.

Figure 9. Effects of lithium treatment on motor coordination in chloroquine treated mice.

(A) Consistent with our published observations , motor coordination deficits were evident in NaCl-treated ECM mice. A one-way ANOVA demonstrated significant group effects on the number of slips on the 1.2 cm beam (F_{(3, 34)} = 11.32; p<0.01), with PbA infection resulting in a significantly higher number of slips than uninfected control mice (Na PbA: 9.33±0.8 v. Na Con: 5±0.6 v. Li Con 4.9±0.4). LiCl treatment ameliorated the impairment in motor coordination in ECM mice as they experienced a significantly lower number of slips on the 1.2 cm diameter beam (Li PbA: 7±0.4 v. Na PbA: 9.33±0.8). LiCl-treated ECM mice still had significantly more slips than uninfected controls. (B) In addition, there were significant group effects on the latency to cross the 1.2 cm diameter beam (Na PbA 9.83±0.6 sec v. Na Con: 6.45±0.3 sec v. Li Con: 7.19±0.3 sec; F_{(3, 34)} = 6.27; p<0.01). Tukey's multiple comparison demonstrated that LiCl-treatment had no effect on the latency to cross the 1.2 cm beam when compared to NaCl-treatment in PbA-infected mice (Li PbA: 8.93±0.6 sec v. Na PbA 9.83±0.6 sec). (C) Although there were no significant differences among the four groups of mice in the number of in slips to cross the 1.8 cm diameter beam (F_{(3, 34)} = 1.56; p = NS), (D) there were significant effects of PbA infection on the latency to cross the 1.8 cm diameter beam compared to uninfected controls to cross the 1.8 cm diameter beam in ECM mice treated with NaCl (Na PbA: 4.91±0.2 v. Na Con: 3.97±0.2 v. Li Con: 3.97±0.1; F_{(3, 34)} = 6.85; p<0.01). Tukey's multiple comparison demonstrated that LiCl-treatment had no effect on the latency to cross the 1.8 cm beam when compared to NaCl-treatment in PbA-infected mice (Li PbA: 4.41±0.2 sec v. Na PbA: 4.91±0.2). (A–D) there were no differences in balance beam performance between LiCl-treated ECM mice and either of the control groups. In addition, both sodium and lithium treated control mice performed similarly in the motor coordination tests. n = 4 Na Con; 12 Na PbA; 12 Li PbA; 10 Li Con. *p<0.05, **p<0.001. Na Con = NaCl treated control, Na PbA = NaCl treated P. berghei ANKA infected mice, Li PbA = LiCl treated P. berghei ANKA infected mice, Li Con = LiCl treated control.