Intact memory in TGF-β1 transgenic mice featuring chronic cerebrovascular deficit: recovery with pioglitazone

Abstract

The roles of chronic brain hypoperfusion and transforming growth factor-beta 1 (TGF-β1) in Alzheimer's disease (AD) are unresolved. We investigated the interplay between TGF-β1, cerebrovascular function, and cognition using transgenic TGF mice featuring astrocytic TGF-β1 overexpression. We further assessed the impact of short, late therapy in elderly animals with the antioxidant N-acetyl-L-cysteine (NAC) or the peroxisome proliferator-activated receptor-γ agonist pioglitazone. The latter was also administered to pups as a prophylactic 1-year treatment. Elderly TGF mice featured cerebrovascular dysfunction that was not remedied with NAC. In contrast, pioglitazone prevented or reversed this deficit, and rescued the impaired neurovascular coupling response to whisker stimulation, although it failed to normalize the vascular structure. In aged TGF mice, neuronal and cognitive indices--the stimulus-evoked neurometabolic response, cortical cholinergic innervation, and spatial memory in the Morris water maze--were intact. Our findings show that impaired brain hemodynamics and cerebrovascular function are not accompanied by memory impairment in this model. Conceivably in AD, they constitute aggravating factors against a background of aging and underlying pathology. Our data further highlight the ability of pioglitazone to protect the cerebrovasculature marked by TGF-β1 increase, aging, fibrosis, and antioxidant resistance, thus of high relevance for AD patients.

Figures

Figure 1

Effects of NAC and pioglitazone on TGF-β1-induced deficits in cerebrovascular reactivity. Early 1-year pioglitazone (pio) administration prevented the development of cerebrovascular dysfunction in adult TGF mice (data shown in Supplementary Figure 1), whereas short treatment completely reversed the established deficits of ∼18-month-old TGF mice relative to their wild-type (WT) littermates (★P<0.05, ★★P<0.01, ★★★P<0.001) in response to ACh, CGRP, ET-1, and NOS inhibition with -NNA (10−5 mol/L) (untreated versus treated TGF mice, *P<0.05, **P<0.01, ***P<0.001). Pioglitazone had no significant effect on responses of treated WT mice, except for a small enhancement of the ET-1-induced constriction (+P<0.05). In contrast, NAC was ineffective on TGF-β1-induced cerebrovascular impairments. Dilatations to the nitric oxide donor SNP were comparable among groups (n=3 to 7 mice per group). Errors bars represent s.e.m.

Figure 2

Pioglitazone failed to normalize vascular structure. (A) Short pioglitazone (pio) treatment in 18-month-old TGF mice did not reverse upregulation (★★P<0.01, ★★★P<0.001) of markers related to vascular fibrosis, as measured by western blot in pial vessels. However, early 1-year treatment significantly prevented (*P<0.05) the increase of the TGF-β1 effector molecule CTGF. Actin was used to normalize loading variation (n=4 to 6 mice per group). (B) Pioglitazone did not counter total collagen accumulation in the pia and intraparenchymal vessels of 18-month-old TGF mice relative to their wild-type (WT) littermates (★★★P<0.001), as measured by Sirius red staining intensity in 5-μm-thick paraffin sections (n=3 to 5 mice per group). Bar=25 μm. (C) Similar to the unabated collagen upregulation in elderly TGF mice, structural alterations were still apparent in thick (25 μm) sections from adult transgenic animals (★P<0.05) that had received early, long (1-year) pioglitazone treatment (n=4 mice per group). Bars: main images=50 μm, insets=25 μm. Errors bars represent s.e.m.

Figure 3

Pioglitazone rescued functional hyperemia and, after long treatment, normalized astrogliosis. (A) Short pioglitazone (pio) treatment rescued (***P<0.001) the impaired neurovascular coupling response of 18-month-old TGF mice (★★★P<0.001) to whisker stimulation (indicated by arrows in right panel) as measured by laser Doppler flowmetry (n=5 mice per group) (traces, green: wild-type (WT); black: treated WT; blue: TGF; red: treated TGF mice). (B) The astrocyte activation in adult TGF mice (★★P<0.01) was significantly attenuated in TGF animals treated early with pioglitazone (pio) for 1 year (*P<0.05) (n=4 to 5 mice per group). In contrast, short therapy in 18-month-old TGF mice did not reverse the increased percentage of cortex occupied by GFAP-positive astrocytes (★P<0.05), which were readily distinguished at higher magnification (inset) (n=3 to 5 mice per group). Bars: main image=400 μm, large inset=200 μm, small inset=25 μm. Error bars represent s.e.m.

Figure 4

The cerebral glucose uptake (CGU) elicited by sensory stimulation was intact in elderly TGF mice compared with wild-type (WT) littermates, and there was no effect of pioglitazone (pio). Ratios of standard uptake value (SUV) in the activated somatosensory cortex (contralateral to whisker stimulation) relative to the analogous ipsilateral cortex were comparable among all groups. There was a tendency for reduced striatal uptake associated with TGF-β1 overexpression and pio treatment in WT mice, respectively, confirming the basal hypometabolism reported in this model (Galea et al, 2006) and induced by pio (n=4 to 7 mice per group). Corresponding anatomic levels across groups were selected on the metabolic PET images using average WT and TGF MRI templates as guides. The enlarged ventricles seen in TGF brains have been reported in this model (Wyss-Coray et al, 1995). SUV scales were slightly adjusted to highlight activation ratios (arrows) in individual groups rather than the global differences across groups. Scans from representative mice are shown. Error bars represent s.e.m.

Figure 5

Total and perivascular cholinergic innervation of 18-month-old TGF mice. (A) The number of ChAT-positive fibers and terminals was comparable between elderly TGF and wild-type (WT) mice, and remained unaffected by short pioglitazone (pio) therapy (n=3 to 5 mice per group). Bar=25 μm. (B) There was no difference in the number of cortical ChAT terminals (dark brown, indicated by arrows) apposed to collagen IV-stained vessels (blue) between WT and TGF mice, and no effect of pioglitazone (n=3 to 4 mice per group). Bar=10 μm. Errors bars represent s.e.m.

Figure 6

Despite cerebrovascular dysfunction, elderly TGF mice (▪) performed as well as their wild-type (WT) counterparts (•) in two variations of the Morris water maze involving either (A) lack of pretraining (n=7 to 11 mice per group) or (B) 3 days of visible-platform training (n=5 to 9 mice per group). TGF-β1 expression did not affect daily escape latency, or time and traveling distance in the target quadrant during the probe trial. However, it induced a significant decrease in the number of platform crossings in cohort 2 (★P<0.05). Pioglitazone (pio) had no effect on treated WT (○) or TGF (□) mice, except for a small increase in the latency of WT mice (cohort 2) on training day 3 (+P<0.05). Error bars represent s.e.m.