Working memory span capacity improved by a D2 but not D1 receptor family agonist

Abstract

Patients with schizophrenia exhibit poor working memory (WM). Although several subcomponents of WM can be measured, evidence suggests the primary subcomponent affected in schizophrenia is span capacity (WMC). Indeed, the NIMH-funded MATRICS initiative recommended assaying the WMC when assessing the efficacy of a putative therapeutic for FDA approval. Although dopamine D1 receptor agonists improve delay-dependent memory in animals, evidence for improvements in WMC due to dopamine D1 receptor activation is limited. In contrast, the dopamine D2-family agonist bromocriptine improves WMC in humans. The radial arm maze (RAM) can be used to assess WMC, although complications due to ceiling effects or strategy confounds have limited its use. We describe a 12-arm RAM protocol designed to assess whether the dopamine D1-family agonist SKF 38393 (0, 1, 3, and 10 mg/kg) or bromocriptine (0, 1, 3, and 10 mg/kg) could improve WMC in C57BL/6N mice (n=12) in cross-over designs. WMC increased and strategy usage decreased with training. The dopamine D1 agonist SKF 38393 had no effect on WMC or long-term memory. Bromocriptine decreased WMC errors, without affecting long-term memory, consistent with human studies. These data confirm that WMC can be measured in mice and reveal drug effects that are consistent with reported effects in humans. Future research is warranted to identify the subtype of the D2-family of receptors responsible for the observed improvement in WMC. Finally, this RAM procedure may prove useful in developing animal models of deficient WMC to further assess putative treatments for the cognitive deficits in schizophrenia.

Figure 1. 12-arm Radial Arm Maze

The fully automated radial arm maze has 12 arms radiating from a central hub. Guillotine doors separate the arms and the hub. Arm entries and food reward area entries (located at the distal end of the arms) are monitored by infrared beams. Ten arms are designated working memory arms (baited once per session) and two arms are reference memory arms (never baited).

Figure 2. Spatial coefficient of variation (CV) and frequency distribution for low and high strategy use performers

The spatial CV performances of C57BL/6N mice over time in the RAM were compared. Performance was measured by the relative transitions from one arm to another. If the subject employed a varied relative arm choice strategy (A), the total number of arms skipped was more evenly distributed resulting in a low StrategyCV score. If however, a predictable strategy is employed, the distribution of the total number of arms skipped was uneven (B) resulting in a high StrategyCV. Thus, the degree of strategy use was quantified by calculating the CV of the totals for each subject over a period of training sessions. During training, mice initially adopted a strategy and had a high StrategyCV score, their ordered strategy significantly declined with training however (C). As training progressed and the use of simple strategies declined, working memory span improved toward a stable performance (C). Data presented as mean ± s.e.m.

Figure 3. Bromocriptine decreases working memory span errors without affecting long term memory

The effects of bromocriptine administration on C57BL/6N mice performing the RAM were assessed. Bromocriptine appeared to increase working memory span capacity in mice, although this effect was not significant (A). Bromocriptine significantly reduced working memory span errors (B) however. This effect on working memory span errors was not evident across different forms of memory since reference memory errors were unaffected (C). Bromocriptine reduced the total number of arms entered (D). When measured as a percentage of total arms entered, bromocriptine still significantly reduced working memory span errors (E) with no concomitant effect on reference memory errors (F). Finally, bromocriptine did not alter the predictability of strategy use (I) suggesting that its effects on working memory span errors were not due to altered learning. Data presented as mean+s.e.m., and * denotes p<0.05 when compared to vehicle control.

Figure 4. SKF 38393 did not affect working memory span capacity or long term memory

The effects of SKF 38393 administration on C57BL/6N mice performing the RAM were assessed. SKF 38393 did not affect any aspect of working memory including working memory span capacity (A) and working memory errors (B) in mice. SKF 38393 did not affect reference memory errors (C) but did reduce the total number of arms entered (D) for the highest dose only. When measured as a percentage of total arms entered, SKF 38393 had no effect on working memory span errors (E) or on reference memory errors (F). Finally, SKF 38393 did not alter the predictability of strategy use (G) suggesting that its effect on total arms entered was not due to altered learning. Data presented as mean+s.e.m., and * denotes p<0.05 when compared to vehicle control.