PAK inactivation impairs social recognition in 3xTg-AD Mice without increasing brain deposition of tau and Aβ

Abstract

Defects in p21-activated kinase (PAK) are suspected to play a role in cognitive symptoms of Alzheimer's disease (AD). Dysfunction in PAK leads to cofilin activation, drebrin displacement from its actin-binding site, actin depolymerization/severing, and, ultimately, defects in spine dynamics and cognitive impairment in mice. To determine the role of PAK in AD, we first quantified PAK by immunoblotting in homogenates from the parietal neocortex of subjects with a clinical diagnosis of no cognitive impairment (n = 12), mild cognitive impairment (n = 12), or AD (n = 12). A loss of total PAK, detected in the cortex of AD patients (-39% versus controls), was correlated with cognitive impairment (r(2) = 0.148, p = 0.027) and deposition of total and phosphorylated tau (r(2) = 0.235 and r(2) = 0.206, respectively), but not with Aβ42 (r(2) = 0.056). Accordingly, we found a decrease of total PAK in the cortex of 12- and 20-month-old 3xTg-AD mice, an animal model of AD-like Aβ and tau neuropathologies. To determine whether PAK dysfunction aggravates AD phenotype, 3xTg-AD mice were crossed with dominant-negative PAK mice. PAK inactivation led to obliteration of social recognition in old 3xTg-AD mice, which was associated with a decrease in cortical drebrin (-25%), but without enhancement of Aβ/tau pathology or any clear electrophysiological signature. Overall, our data suggest that PAK decrease is a consequence of AD neuropathology and that therapeutic activation of PAK may exert symptomatic benefits on high brain function.

Figure 1.

PAK loss in AD correlates with disease severity in human. A, Actin-normalized PAK was decreased in AD patients compared with controls, based on clinical diagnoses. B, C, Decreases in actin-normalized PAK (B) and pPAK (C) were observed in individuals with definitive neuropathological diagnosis of AD (CERAD 1) compared with those with a “probable” diagnosis of AD (CERAD 2) or low AD probability (possible or none, CERAD 3 or 4). D–G, Linear regression analyses showed that actin-normalized PAK (in the soluble fraction) was correlated negatively with the ratio of insoluble tau fraction on the soluble tau fraction (D) and insoluble PHF_tau (E), but not with the accumulation of insoluble Aβ42 (F), but was positively associated with global antemortem cognitive score (G). H, Examples of Western blot results obtained for PAK and pPAK in control, MCI, and AD patients. All samples were run simultaneously on the same SDS-PAGE gels. Statistical analyses: one-way ANOVA/Dunnett's multiple-comparison test (A), Kruskal-Wallis/Dunn's multiple-comparison test (B,C), and linear regression (D–G).

Figure 2.

Characterization of PAK, membrane cofilin, and drebrin impairments in homozygous 3xTg-AD mice according to age. A, Soluble PAK_1/2/3 was decreased in 12- and 20-month-old 3xTg-AD mice compared with NonTg mice of the same age. No statistical differences were found between the two genotypes in young animals (4 months). B, When comparing both genotypes at different ages, we observed no alteration of soluble phospho-PAK in 3xTg-AD mice at 4, 12, or 20 months. C, Loss of soluble PAK_1/2/3 was observed in both isoforms 1 and 3 at all ages. Despite an age effect for drebrin level, we did not found any change of membrane cofilin (D) or drebrin (E) levels in 3xTg-AD compared with NonTg mice. Values are expressed as mean ± SEM. Statistical comparisons were performed using a one-way ANOVA followed by a Tukey-Kramer post hoc test (age comparison, A–C) and an unpaired t test (D,E). *p < 0.5, **p < 0.01, ***p < 0.001.

Figure 3.

Consequences of PAK inactivation in 12- or 20-month-old heterozygous 3xTg-AD mice. A, Myc-Tag immunofluoresence (green) in the cortex confirmed that the dnPAK transgene was only expressed in neurons (NeuN in red) from 3xTg-AD × dnPAK mice compared with 3xTg-AD mice. pPAK immunostaining (green) in the cortex highlighted a neuronal localization (NeuN in red) of pPAK (s141) and an apparent decrease in 20-month-old 3xTg-AD mice with genetically PAK inactivation compared with 3xTg-AD mice (n = 3 per group). Western blot analyses revealed that dnPAK transgene expression reduced pPAK (B) and increased the level of PAK₁ in 12-month-old mice, but was inefficient to modulate PAK₃ at both ages (C) in the soluble fraction. dnPAK transgene expression did not induce any effect on cofilin (D) or phospho-cofilin (E) levels in the cortex of 3xTg-AD mice. Interestingly, drebrin level was reduced in 20-month-old mice in the membrane fraction (F). Finally, no significant effect was observed on detergent-soluble synaptophysin level whatever the age of mice (G). Square bar equals 100 μm. Values are expressed as mean ± SEM. Statistical analyses were performed using an unpaired t test. For PAK3 analysis (C), only the upper band was used. *p < 0.05, **p < 0.01.

Figure 4.

PAK inactivation impaired social recognition in 20-month-old 3xTg-AD mice. (A) NonTg mice showed higher general activity than 3xTg-AD mice, as reflected in a vertical activity test over a 10 min observation span. B, Time spent in the illuminated compartment of the black and white box was not significantly different in NonTg, 3xTg-AD, and 3xTg-AD × dnPAK mice. C, 3xTg-AD and 3xTg-AD × dnPAK mice showed a similar number of alternances, but both were lower than NonTg, suggesting that the level of anxiety was not influenced by PAK inhibition. D, Exploratory behavior was similarly decreased in 3xTg-AD and 3xTg-AD × dnPAK mice compared with NonTg animals. E, Social events in which a mouse interacts with its dyad partner were recorded (sniffing, following, grooming the partner, crawling over or under) at day 1 and at day 2. F, 3xTg-AD mice with normal PAK activity had a 40% reduction of interaction time with their dyad mate at day 2 compared with day 1. In contrast, PAK-deficient 3xTg-AD mice displayed the same interaction time with their partner at day 1 and day 2, consistent with a complete absence of social recognition. Values are expressed as mean ± SEM. Statistical comparisons were performed using an unpaired Student's t test. *p < 0.05, ***p < 0.001.

Figure 5.

PAK inactivation reduced Aβ and tau depositions in 12- and 20-month-old heterozygous 3xTg-AD mice. PAK inactivation did not alter concentrations of sAβ40 (A) and sAβ42 (B) in soluble fractions at any age, but decreased iAβ40 (C) and iAβ42 (D) levels in insoluble fractions from 12-month-old 3xTg-AD mice. Whereas the inactivation of PAK induced a decrease of the insoluble/soluble tau ratio in 12-month-old mice (E), it did not induced any effect on the soluble phospho-tau/total tau ratio in mice whatever their age (F). Finally, no effect was observed on phospho-tau levels in the insoluble fraction (G). Values are expressed as mean ± SEM. Statistical comparisons were performed using an unpaired t test. ROD indicates relative optical density. Comparison between genotype: *p < 0.05, **p < 0.01; comparison between age: ^#p < 0.05, ^##p < 0.01, ^###p < 0.001.

Figure 6.

Intrinsic properties of entorhinal cortex neurons in 12-month-old homozygous 3xTg-AD mice and in 20-month-old 3xTg-AD mice with genetically induced PAK inactivation. A–F, Expression of transgenes in homozygous 3xTg-AD mice did not change resting potential (A) but decreased CC (B) of entorhinal cortex deep layer neurons. Interestingly, we found a positive correlation between soluble PAK_1/2/3 and the CC (C). Events of sEPSC occurred more frequently in 3xTg-AD neurons (D), but event amplitudes were lower (E). F, Examples of spontaneous EPSC recordings made in NonTg and 3xTg-AD neurons. NonTg mice are represented in white and 3xTg-AD mice are represented by stripes in graphs. G–K, To investigate the physiological function of PAK in AD, we studied effects of PAK inactivation in intrinsic and synaptic properties of deep-layer lateral entorhinal cortex neurons of NonTg (white), NonTg with a deficiency in PAK activity (white with black stripes), heterozygous 3xTg-AD (gray), and heterozygous 3xTg-AD mice with a deficiency of PAK activity (black). Genetically, PAK inactivation in NonTg or 3xTg-AD mice did not influence impairments of passive properties of entorhinal cortex observed in 3xTg-AD mice, as reflected by resting potential (G) and CC (H) analyses. Similarly, frequency (I) and amplitude (J) of sEPSCs observed in 3xTg-AD mice were not influenced by PAK activity. K, Examples of sEPSC recordings made in NonTg, 3xTg-AD, and PAK-deficient 3xTg-AD neurons. Values are expressed as mean ± SEM of the numbers of cells analyzed. Statistical analyses were performed using unpaired t test. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 7.

Counterbalancing the loss of PAK might be of therapeutic value in AD. Late during AD progression, and not in individuals with mild cognitive impairment, the levels of PAK decrease in the brain in close relationship with accumulation of tau neuropathology and declining cognitive performance. In the 3xTg-AD mouse model, expression of transgenes leading to Aβ and tau pathologies induced a global decrease of PAK. Conversely, genetic deactivation of PAK in the 3xTg-AD mouse led to a transient decrease in insoluble Aβ and tau. This suggests that PAK decrease is a consequence rather than a cause of Aβ and tau neuropathologies. However, inactivating PAK in the brains of 3xTg-AD mice generated an animal model closer to AD by decreasing drebrin levels and social recognition capacities. Overall, our study supports the hypothesis that a therapeutic intervention aimed at increasing PAK activity may improve AD symptoms without correcting Aβ and tau pathologies.