Age-dependent cognitive deficits and neuronal apoptosis in cyclooxygenase-2 transgenic mice

Abstract

The cyclooxygenases catalyze the rate-limiting step in the formation of prostaglandins from arachidonic acid and are the pharmacological targets of (NSAIDs). In brain, cyclooxygenase-2 (COX-2), the inducible isoform of cyclooxygenase, is selectively expressed in neurons of the cerebral cortex, hippocampus, and amygdala. As an immediate-early gene, COX-2 is dramatically and transiently induced in these neurons in response to NMDA receptor activation. In models of acute excitotoxic neuronal injury, elevated and sustained levels of COX-2 have been shown to promote neuronal apoptosis, indicating that upregulated COX-2 activity is injurious to neurons. COX-2 may also contribute to the development of Alzheimer's disease, for which early administration of NSAIDs is protective against development of the disease. To test the effect of constitutively elevated neuronal COX-2, transgenic mice were generated that overexpressed COX-2 in neurons and produced elevated levels of prostaglandins in brain. In cross-sectional behavioral studies, COX-2 transgenic mice developed an age-dependent deficit in spatial memory at 12 and 20 months but not at 7 months and a deficit in aversive behavior at 20 months of age. These behavioral changes were associated with a parallel age-dependent increase in neuronal apoptosis occurring at 14 and 22 months but not at 8 months of age and astrocytic activation at 24 months of age. These findings suggest that neuronal COX-2 may contribute to the pathophysiology of age-related diseases such as Alzheimer's disease by promoting memory dysfunction, neuronal apoptosis, and astrocytic activation in an age-dependent manner.

Fig. 1.

Analysis of hCOX-2 expression in brains of transgenic and nontransgenic C57B6/J mice. A, Western blot analysis of murine COX-2 (mCOX-2) using species-specific monoclonal antibody 12A2, which detects human COX-2 and not murine COX-2. Human recombinant COX-2 is detected but not mouse COX-2 or mouse COX-1 (left panel). In 3-month-old hCOX-2 brains, a 70 kDa band (arrow) representing transgenic hCOX-2 is detected that is not present in nontransgenic brain. B, Western blot analysis of 20- to 24-month-old hCOX-2 mice using polyclonal anti-COX-2 antibody detects both human and murine COX-2 expression and demonstrates continued high expression of transgenic hCOX-2 in aged animals. Note the faint 70 kDa band consisting of endogenous murine COX-2 in nontransgenic (NTg) 20- to 24-month-old animals. C–E, Immunocytochemistry of hCOX-2 protein on paraffin sections of brain from nontransgenic and hCOX-2 mice stained with 12A2. C, Low-power (50×) magnification of hippocampus of a 14 month hCOX-2 brain (top panel) and a nontransgenic age-matched control (bottom panel) demonstrating the presence of transgenic hCOX-2 protein in pyramidal neurons of the hippocampus in transgenic hCOX-2 but not nontransgenic brain. Note the presence of hCOX-2 protein in the CA1 region (arrows) and CA3 region (asterisks) and the absence of hCOX-2 protein in dentate gyrus (d). D, Higher magnification (100×) of frontal cortex demonstrating hCOX protein in layers II/III and V. The hCOX-2 protein is present in the same subcellular patterns and anatomic patterns of expression as endogenous murine COX-2, which is present in a somatodendritic distribution in neurons and hippocampus and layers V and II/III of cortex. E, Higher magnification (400×) of pyramidal neurons of CA1 (top panel) and CA3 (bottom panel).Arrows point to hCOX-2 protein in apical dendritic processes in CA1 and CA3 pyramidal neurons. F, G, Determination of brain PGE2 levels in transgenic and nontransgenic mice. Data are mean ± SEM. F, PGE2 levels were significantly increased in transgenic (hCOX-2) versus nontransgenic (NTg) mice at 4, 6, and 9 months of age and were elevated on average ∼10- to 12-fold over endogenous PGE2 levels. G, PGE2 levels in hCOX-2 and nontransgenic mice can be rapidly reduced with administration of the COX-2 inhibitor celecoxib.

Fig. 2.

Performance of hCOX-2 and nontransgenic control littermates during probe trials in the Morris water maze.A, Order of platform and probe trials during training in the place discrimination task. Daily sessions consisted of 10 platform (open circles) and 2 probe trials. During the first day, the training started with five platform trials followed by a probe trial with short delay (5–7 min, ringed circle). The short-delay probe trial that was conducted at the end of the day's sessions was used as an index of memory for the platform location with a minimum retention interval. After the first day of training, probe trials with 24 hr of delay (filled circles) were included to assess the strength of long-term memory of the platform location and were performed on days 2–5. B, C, Preference for the platform location during short-delay (B) and long-delay (C) probe trials. Open bars, Nontransgenic (NTg) mice; filled bars, hCOX-2 transgenic (Tg) mice. At the age of 7 months, mice did not differ in any of the measures in both types of probe trials. At the age of 12 months, Tg mice performed similarly to NTg mice in an easy task, the short-delay probe, except for Annulus 40, and were significantly impaired in all measures in the more challenging task, the 24-hr-delay probe trial. Introduction of the 24 hr delay led to a more pronounced impairment of spatial memory in middle-aged Tg mice that was expressed both as a significant difference from NTg littermates of the same age as well as in a significant aging effect compared with younger Tg mice. At 20 months, the performance deficit of Tg mice had already occurred in short-delay probe trials. However, both Tg and NTg 20-month-old mice were significantly impaired in the more difficult 24-hr-delay probe trials. *Significant effect of genotype (simple main effect, p < 0.05). Data are mean ± SEM.

Fig. 3.

Average latency (A) and swim speed (B) to reach the hidden platform during platform trials in the place discrimination task. Open bars, Nontransgenic (NTg) mice; filled bars, hCOX-2 transgenic (Tg) mice.A, Young Tg mice (7 months) did not differ from their age-matched littermates, whereas middle aged Tg mice (12 months) and old Tg mice (20 months) needed significantly more time to find the platform than their NTg counterparts. The latency increased with age in both groups of mice. However, the onset of this increase occurred earlier in Tg mice (at the age of 12 months) compared with NTg mice, in which it occurred later at age 20 months. B, Tg and NTg mice did not differ in swim speed at any age, suggesting that the longer latencies to the platform in Tg mice were not attributable to poor swimming abilities but rather to compromised cognition. *Significant effect of genotype (simple main effect,p < 0.05); ^#significant decline with age (post hoc test, p < 0.5). Data are mean ± SEM.

Fig. 4.

Performance of transgenic hCOX-2 (TG, open bars) and nontransgenic (NTG, filled bars) mice in straight alley (A) and visual discrimination (B) tasks. A, Average latency to reach the platform across sessions in the straight alley test. TG and NTG groups of mice did not differ at any of the ages tested. An age-related decline in the ability to swim was observed in both groups of mice at 12 months of age, with additional deterioration at 20 months of age. B, Average distance to reach a visible platform across all visual discrimination trials. Both groups of mice showed similar performance in all ages tested. Data are mean ± SEM.

Fig. 5.

Performance of transgenic hCOX-2 (Tg, shaded bars) and nontransgenic (NTg, open bars) mice in aversively motivated tasks. A, The learning in the T maze active avoidance task is represented as the number of trials to the first avoidance during the acquisition session. Higher scores indicate a poorer acquisition of avoidance reaction. Age-related alterations occurred only in 20-month-old Tg mice but not in 20-month-old NTg mice compared with their younger counterparts. *Bracketsindicate a significant effect of age in Tg mice (p < 0.01) as a result of simple main effect ANOVA. Data are mean ± SEM. B, Inhibitory avoidance task. Inhibitory avoidance reaction was measured by the latency to enter the previously shocked compartment. In this case, the higher latency reflects a better retention after a 24 hr delay. The age-related impairment occurred only in Tg mice, which showed a decreased latency at age of 20 months. This effect was not present in NTg mice. *Brackets indicate a significant effect of age in Tg mice (p < 0.01) as a result of χ² tests. The latency for every group is presented as median ± interquartile range.

Fig. 6.

Performance of transgenic hCOX-2 (TG, shaded bars) and nontransgenic (NTG, open bars) mice in control tasks.A, Sensitivity to shock assessed as latency of the first escape reaction during the T maze active avoidance training did not differ between TG and NTG mice at any of the ages tested. No significant age-related changes were revealed for this index of sensitivity to shock (see Results for the results of burst activity measures). B, Total motor activity is expressed as the number of cells in the open field crossed in 5 min of testing. Mice of both genotypes did not differ for all ages tested. An age-related decline in exploratory activity was observed to the same extent in both TG and NTG mice at the age of 12 months. C, Performance in the sensorimotor tasks is shown as the average Zscore from a set of different tasks (turning in an alley, traversing bridges, wire hanging, and falling from a wire screen). There were no significant differences in sensorimotor skills between TG and NTG mice. A significant effect of age was observed in both groups of mice and consisted of a decline in performance at 20 months of age.

Fig. 7.

Histopathological analysis of hCOX-2 and nontransgenic age-matched littermates. A, Levels of GFAP were compared with an assay for astrocytic activation, proliferation, or both in aged hCOX-2 mice. Quantitative Western blot analysis (50 μg of protein/lane) of GFAP and actin was performed at 20–24 months (n = 7 for each COX-2 and nontransgenic group). A representative set of four pairs of COX-2 and nontransgenic (NTg) controls demonstrates increases in GFAP in aged hCOX-2 brain compared with nontransgenic controls in three of four pairs, with the fourth pair displaying an equal amount of GFAP staining. B, Standardized GFAP of 20- to 24-month-old hCOX-2 and nontransgenic mice demonstrates a significant increase in GFAP in hCOX-2 mice compared with nontransgenic mice at 20–24 months of age. *Significant between-genotype difference as a result of ANOVA. Data are mean ± SEM. C, TUNEL analysis of 8-, 14-, and 22-month-old hCOX-2 and age-matched control littermates. Two-way (age × genotype) ANOVA with the square root transformation of the number of apoptotic cells in cerebral cortex per section showed significance for both effects (age,F_(2,335) = 16.34; p< 0.0001; genotype, F_(1,335) = 5.87;p < 0.02). Both groups of mice demonstrated a higher number of apoptotic cells at 14 and 22 months. However, the age-associated increase in apoptosis was more pronounced in hCOX-2 mice compared with nontransgenic littermates. ^#,##Significant age-related increase in the number of stained cells as a result ofpost hoc tests; ^#p < 0.01; ^##p < 0.0001. *p < 0.01 for post hoc tests applied to the effect of genotype within each age group. Data are mean ± SEM. D–F, Determination of cellular phenotype of TUNEL-positive cells using double immunofluorescent staining. After TUNEL labeling, selected sections were doubly stained with Neu N monoclonal and anti-GFAP polyclonal antibodies to identify astrocytes and neurons, respectively. Apoptotic cells uniformly stained for Neu N and did not stain for GFAP, indicating that the TUNEL-positive cells are neurons. D, Light micrograph 630× magnification of a TUNEL-positive cell; note nuclear condensation and margination. E, Immunofluorescent staining of the same region with anti-Neu N antibody demonstrating staining of the nuclear compartment of a TUNEL-positive cell (vertical arrow); horizontal arrows in all panels demonstrate nonapoptotic neurons also stained with Neu N.F, Immunofluroscent staining with anti-GFAP antibody demonstrating the absence of GFAP staining of the TUNEL-positive cell; note GFAP-stained astrocyte (asterisks in all three panels).