Searching for cognitive enhancement in the Morris water maze: better and worse performance in D-amino acid oxidase knockout (Dao(-/-)) mice

Abstract

A common strategy when searching for cognitive-enhancing drugs has been to target the N-methyl-d-aspartate receptor (NMDAR), given its putative role in synaptic plasticity and learning. Evidence in favour of this approach has come primarily from studies with rodents using behavioural assays like the Morris water maze. D-amino acid oxidase (DAO) degrades neutral D-amino acids such as D-serine, the primary endogenous co-agonist acting at the glycine site of the synaptic NMDAR. Inhibiting DAO could therefore provide an effective and viable means of enhancing cognition, particularly in disorders like schizophrenia, in which NMDAR hypofunction is implicated. Indirect support for this notion comes from the enhanced hippocampal long-term potentiation and facilitated water maze acquisition of ddY/Dao(-) mice, which lack DAO activity due to a point mutation in the gene. Here, in Dao knockout (Dao(-/-) ) mice, we report both better and worse water maze performance, depending on the radial distance of the hidden platform from the side wall of the pool. Dao(-/-) mice displayed an increased innate preference for swimming in the periphery of the maze (possibly due to heightened anxiety), which facilitated the discovery of a peripherally located platform, but delayed the discovery of a centrally located platform. By contrast, Dao(-/-) mice exhibited normal performance in two alternative assays of long-term spatial memory: the appetitive and aversive Y-maze reference memory tasks. Taken together, these results question the proposed relationship between DAO inactivation and enhanced long-term associative spatial memory. They also have generic implications for how Morris water maze studies are performed and interpreted.

Keywords: D-serine; N-methyl-d-aspartate receptor; glutamate; pre-clinical model; schizophrenia.

Figure 1

Morris water maze platform locations and swim analysis zones. (A) In Experiment 1, the hidden platform was located 50 cm from the side wall of the pool, in either the NW or SE quadrant. (B) In Experiment 2, the hidden platform was located either 75 or 25 cm from the side wall (henceforth referred to as the central and peripheral platform locations, respectively), in either the NW or SE quadrant. (C) Using the software package ANY‐maze 4.5 (Stoelting, Wood Dale, IL, USA), the pool was divided into nominal central and peripheral zones, encompassing the central and peripheral platform locations. Percent time spent in each of these zones was computed for each mouse for each trial of Experiment 2.

Figure 2

Enhanced Morris water maze performance in Dao ^−/− mice. A total of 24 mice (n = 6 males and n = 6 females per genotype) participated in Experiment 1. (A) With data collapsed across the four trials of each training session, genotype had no apparent impact on water maze performance (path length to platform) during the acquisition phase or reversal learning phase. (B) Across the first 4 days of training, however, Dao ^−/− mice significantly outperformed WT mice on the first trial of each day (path length to platform), but not on the subsequent three trials, yielding a significant interaction between genotype and trial number (P = 0.002). The graph shows path length to the platform in the four trials of a session (Day), collapsed across days 1–4. (C) The same pattern of performance was evident on the very first day of training, again yielding a significant genotype × trial number interaction (P = 0.020). The graph shows path length to the platform in the four trials of day 1. (D) Genotype had no effect on probe trial 1 performance. The graph shows percent time spent in each quadrant of the pool (platform indicates the quadrant in which the platform was located during training; other quadrants are defined relative to this quadrant). *P‐value < 0.05. Error bars depict the SEM. Dashed line, chance performance (25%).

Figure 3

Water maze performance in Dao ^−/− mice depends on the radial distance of the hidden platform from the side wall. A total of 46 mice (n = 11 males and n = 12 females per genotype) participated in Experiment 2. (A) With data collapsed across the four trials of each training session, and across the two radial platform distances, genotype had no apparent impact on water maze performance (path length to platform) during the 4‐day acquisition phase. (B) The performance (mean path length to platform across all 16 acquisition trials) of Dao ^−/− mice was numerically superior to that of WT mice in the peripheral platform group, but numerically inferior in the central platform group, yielding a borderline‐significant interaction between genotype and radial platform distance (P = 0.05). (C) Dao ^−/− mice spent significantly more time than WT mice in the peripheral zone (see Fig. 1C) of the pool. (D) Genotype had no effect on probe trial performance. The graph shows percent time spent in each quadrant of the pool (platform indicates the quadrant in which the platform was located during training; other quadrants are defined relative to this quadrant). (E) Across the 4 days of training, WT mice in the central platform condition significantly outperformed Dao ^−/− mice on trial 1 (path length to platform), but not on the subsequent three trials, yielding a significant interaction between genotype and trial number (P = 0.022). (F) By contrast, there was no genotype × trial number interaction for path length to platform in the peripheral platform condition. **P‐value < 0.01. Error bars depict the SEM. Dashed line, chance performance (25%).

Figure 4

Increased anxiety in Dao ^−/− mice in the anxiogenic open field test. A total of 28 male mice (n = 14 per genotype) participated in Experiment 3. Relative to WT mice, Dao ^−/− mice (A) spent less time in the central zone of the arena, and (B) took longer to make their first centre zone entry. (C) Distance travelled did not differ between Dao ^−/− and WT mice. *P‐value < 0.05. Error bars depict the SEM.

Figure 5

Normal spatial memory in Dao ^−/− mice in the aversively motivated Y‐maze swim‐escape task. A total of 28 male mice (n = 14 per genotype) participated in Experiment 4. (A) Genotype had no effect on choice accuracy (% correct trials) in this task. (B) Dao ^−/− mice took longer than WT mice to reach the hidden escape platform, particularly on the first block of training. (C) Genotype had no effect on probe trial performance (% time spent in the target arm of the Y‐maze, the arm in which the platform was located during training). *P‐value < 0.05. Error bars depict the SEM. Dashed line, chance performance (50% for acquisition phase and 33.3% for probe trial).

Figure 6

Normal spatial memory in Dao ^−/− mice in the appetitively motivated Y‐maze task. A total of 24 male mice (n = 12 per genotype) participated in Experiment 5. (A) Genotype had no effect on choice accuracy (% correct trials) in this task. (B) Reward latencies were numerically, but not significantly, greater in Dao ^−/− mice than WT mice. Error bars depict the SEM. Dashed line, chance performance (50%).