PPARγ recruitment to active ERK during memory consolidation is required for Alzheimer's disease-related cognitive enhancement

Abstract

Cognitive impairment is a quintessential feature of Alzheimer's disease (AD) and AD mouse models. The peroxisome proliferator-activated receptor-γ (PPARγ) agonist rosiglitazone improves hippocampus-dependent cognitive deficits in some AD patients and ameliorates deficits in the Tg2576 mouse model for AD amyloidosis. Tg2576 cognitive enhancement occurs through the induction of a gene and protein expression profile reflecting convergence of the PPARγ signaling axis and the extracellular signal-regulated protein kinase (ERK) cascade, a critical mediator of memory consolidation. We therefore tested whether PPARγ and ERK associated in protein complexes that subserve cognitive enhancement through PPARγ agonism. Coimmunoprecipitation of hippocampal extracts revealed that PPARγ and activated, phosphorylated ERK (pERK) associated in Tg2576 in vivo, and that PPARγ agonism facilitated recruitment of PPARγ to pERK during memory consolidation. Furthermore, the amount of PPARγ recruited to pERK correlated with the cognitive reserve in humans with AD and in Tg2576. Our findings implicate a previously unidentified PPARγ-pERK complex that provides a molecular mechanism for the convergence of these pathways during cognitive enhancement, thereby offering new targets for therapeutic development in AD.

Keywords: Alzheimer's; hippocampus; in vitro reconstitution; protein complex; transgenic.

Figure 1.

PPARγ associates with pERK in vivo in Tg2576 hippocampal multiprotein complexes. A, B, Western immunoblots (IBs) for pERK and PPARγ in pERK IPs from Tg2576 using anti-pERK-conjugated Sepharose beads with increasing input of hippocampal nuclear extract. C, Input–output IP linear relationship for pERK IPs (r² = 0.991 up to 750 μg of input). Densitized Western blot values were normalized to the loading control described in Materials and Methods and Fig. 2A. Dotted lines represent the 95% confidence intervals. IgHC, Ig heavy chain.

Figure 2.

Quantification method to determine PPARγ/pERK2 ratios. A, Shown is an example Western blot for PPARγ and pERK in pERK IPs from four individual mice. For quantification across multiple immunoblots of IP material, a homogenate prepared from pooled brains from C57BL/6J mice was used as a LC and was resolved in triplicate (lanes 1, 4, and 7) on each SDS-PAGE gel. pERK IPs from four individual mice (lanes 2, 3, 5, and 6) are depicted. For the data described herein, immunoblots for PPARγ or pERK2 from the IPs were normalized relative to the LC. PPARγ in the LC lanes was chosen as the normalization protein because it tracked in the linear range with immunoprecipitated PPARγ and pERK2 for their respective exposures. After acquiring normalized values for immunoprecipitated PPARγ and pERK2 proteins for each individual animal's hippocampal extract, the amount of PPARγ that coimmunoprecipitates with pERK was calculated by taking the ratio of normalized PPARγ to normalized pERK2. In the example above, Mouse 2 has a hippocampal PPARγ/pERK2 ratio of 0.658. B, PPARγ/pERK2 ratios are highly reproducible. Western blots of PPARγ and pERK in four independent pERK IPs from six individual animals (lanes 2, 3, 4, 6, 7, and 8) and the triplicate LC (lanes 1, 5, 9) resolved by four separate gels. The PPARγ/pERK2 ratios were calculated as in Figure 2A, and the coefficient of variation for each individual animal was determined. All replicate IPs yielded a coefficient of variation of ≤4.8%.

Figure 3.

PPARγ/pERK2 ratios in human AD brains and Tg2576 mouse hippocampi correlate with cognitive performance. A, Correlation between AD human brain PPARγ/pERK2 ratios and MMSE, a measure of cognitive reserve in humans (n = 7, r² = 0.87, p = 0.003, power > 80%; Cohen, 1992). No correlation was found between complex ratios and cognitive reserve in control human brains (n = 7). B, Western blots for PPARγ and pERK as a function of MMSE score. C, Correlation between Tg2576 mouse hippocampal PPARγ/pERK2 ratios and contextual freezing, a measure of cognitive reserve in mice (n = 7, r² = 0.59, p = 0.043). No correlation was found between complex ratios and cognitive reserve in control, WT hippocampi (n = 9). D, Hippocampal PPARγ/pERK2 ratios in WT mice (Tg2576) and Tg2576 mice treated with (+) or without (−) RSG. No significant interaction between genotype or treatment on PPARγ/pERK2 ratios. Two-way ANOVA, n = 7–12/group, p = 0.565, F_(1,34) = 0.3375. Densitometric analysis of the Western blots are presented as the mean ± SEM. ns, Nonsignificant.

Figure 4.

PPARγ agonism increases the recruitment of PPARγ to pERK during memory consolidation in Tg2576 mice. A, Experimental paradigm: Tg2576 mice fed control (−) or RSG (+) diet were either naive (FC−) or trained in the FC task (FC+), then were killed 4 h post-training during consolidation to determine hippocampal PPARγ/pERK2 ratios. For PPARγ antagonism studies, 4 h before training vehicle (GW−) or GW9662 (GW+) were intracerebroventricularly administered, and ratios were determined 4 h after training. B, C, Effects of RSG and fear conditioning on nuclear ratios (two-way ANOVA, n = 7–8/group, F_(1,26) = 11.28, p = 0.002, 0.025, 0.002 for interaction, treatment, and training, respectively; B) and non-nuclear ratios (two-way ANOVA, n = 7/group, F_(1,24) = 8.155, p = 0.009, 0.064, 0.015 for interaction, treatment, and training, respectively; C). D, E, Effects of PPARγ antagonism on nuclear ratios (two-way ANOVA, F_(1,40) = 5.705, p = 0.022, 0.121, 0.559 for interaction, treatment, and intracerebroventricular injection, respectively; D) and non-nuclear ratios (two-way ANOVA, ns, F_(1,31) = 1.016; E). F, G, Neither RSG treatment nor GW9662 antagonism had any effect on WT PPAR/pERK2 ratios in nuclear (two-way ANOVA, p = 0.41, ns, F_(1,37) = 0.694; F) or non-nuclear (two-way ANOVA, p = 0.78, ns, F_(1,24) = 0.074; G) fractions. H, I, pERK association MEK (n = 5/group, representative blot; H) or p90RSK (n = 2/group; I). *p < 0.05; **p ≤ 0.01.

Figure 5.

PPARγ and pERK recombinant proteins associate in vitro. A, Western blot for pERK (top) and PPARγ (bottom) following incubation of recombinant human GST-pERK2 with increasing amounts of human PPARγ followed by glutathione bead affinity isolation. B, Input–output relationship for PPARγ pulldown from GST-pERK2 IP. Dotted lines represent 95% confidence intervals. C, Western blot for nonphosphorylated ERK (top) and PPARγ (bottom) following incubation of recombinant GST-nonphosphorylated ERK2 with increasing amounts of human PPARγ followed by glutathione bead affinity isolation.

Figure 6.

Working model for PPARγ-mediated enhancement of memory consolidation in AD. In the cognitively impaired Tg2576 AD model mice, ligand-bound PPARγ is recruited to activated ERK following a learning event. The complex recruits a number of other transcriptional regulatory proteins, ultimately increasing ERK downstream efficiency, including Cre-mediated gene transcription, as well as activation of p90RSK and Elk-1.