Aerobic Glycolysis Is Required for Spatial Memory Acquisition But Not Memory Retrieval in Mice

Abstract

The consolidation of newly formed memories and their retrieval are energetically demanding processes. Aerobic glycolysis (AG), also known as the Warburg effect, consists of the production of lactate from glucose in the presence of oxygen. The astrocyte neuron lactate shuttle hypothesis posits that astrocytes process glucose by AG to generate lactate, which is used as a fuel source within neurons to maintain synaptic activity. Studies in mice have demonstrated that lactate transport between astrocytes and neurons is required for long-term memory formation, yet the role of lactate production in memory acquisition and retrieval has not previously been explored. Here, we examined the effect of dichloroacetate (DCA), a chemical inhibitor of lactate production, on spatial learning and memory in mice using the Morris water maze (MWM). In vivo hyperpolarized ¹³C-pyruvate magnetic resonance spectroscopy revealed decreased conversion of pyruvate to lactate in the mouse brain following DCA administration, concomitant with a reduction in the phosphorylation of pyruvate dehydrogenase. DCA exposure before each training session in the MWM impaired learning, which subsequently resulted in impaired memory during the probe trial. In contrast, mice that underwent training without DCA exposure, but received a single DCA injection before the probe trial exhibited normal memory. Our findings indicate that AG plays a key role during memory acquisition but is less important for the retrieval of established memories. Thus, the activation of AG may be important for learning-dependent synaptic plasticity rather than the activation of signaling cascades required for memory retrieval.

Keywords: aerobic glycolysis; lactate; magnetic resonance spectroscopy; memory; metabolism; synaptic plasticity.

Figure 1.

Hyperpolarized ¹³C-pyruvate magnetic resonance spectroscopic imaging reveals a decline in lactate following DCA administration. A, Injection and imaging regime of 9-month-old mice using hyperpolarized MRSI of [1-¹³C] pyruvate to measure the conversion of pyruvate to lactate. Mice were imaged after the first ¹³C-pyruvate injection (Before) and 30 min later were then injected with DCA (200 mg/kg). Following a 30 min recovery time, another injection of ¹³C-pyruvate and imaging were performed (After). B, ¹H MRI of the brain in the coronal field overlayed with MRSI voxels containing spectra of ¹³C-labeled pyruvate and lactate (yellow). C, Conversion of pyruvate to lactate was measured as a ratio of the observed lactate peak to pyruvate peak from before DCA injection (blue line) and after (red dashed line). A pyruvate hydrate peak was also recorded. D, DCA injection reduces the ratio of lactate to pyruvate in the mouse brain (p = 0.04). Data shown are the mean ± SEM. n = 4.

Figure 2.

DCA injection reduces the phosphorylation of PDH in the frontal cortex and hippocampus. A, B, Western blot analysis (left) was performed on extracts from the frontal cortex (A) and the hippocampus (B) of mice intraperitoneally injected with either saline or DCA (200 mg/kg) 30 min before being killed. Densitometric analysis of Western blots (right) revealed significantly lower PDH phosphorylation in DCA-treated mice relative to saline-injected mice (*p < 0.05, ***p < 0.001; n = 6 and 6, respectively, for extracts from the frontal cortex of saline- and DCA-injected mice; and n = 7 and 7, respectively, for extracts from the hippocampus of saline- and DCA-injected mice). Band densities were standardized to β-actin controls. Data shown are the mean ± SEM.

Figure 3.

DCA administration causes impairment in spatial learning. Mice were intraperitoneally injected with saline or DCA (200 mg/kg) 30 min before each day of training and allowed to find the location of a hidden platform in the northwest quadrant. A–C, The latency to find the platform (A), the total path length (B), and the mean speed (C) were recorded on each training day. D–G, On day 5, a probe trial was performed without DCA injection, and mice were allowed to swim for 60 s. D, The swim path for each group of mice was compiled into heat map representations. E–G, Measurements were taken for the total distance traveled (E), the percentage of time spent in the correct quadrant (F), and the number of times the boundary of the platform was crossed (G). H, A week after the first probe trial, mice were again tested with a second probe trial, and measurements were taken for the number of times the boundary of the platform was crossed. Data shown are the mean ± SEM. *p < 0.05, **p < 0.01. n = 7 and 7 for saline and DCA treatments, respectively.

Figure 4.

DCA administration does not interfere with memory retrieval. MWM performed by mice intraperitoneally injected with saline or DCA (200 mg/kg) on the probe trial. A, Mice were trained for 4 consecutive days (4 trials/d) to find the location of a hidden platform in the southwest quadrant, and the latency to escape was recorded. B–G, On day 5, a probe trial was performed in which the platform was removed and mice were allowed to swim for 60 s. B, The swim path for each group of mice was recorded and compiled into heat map representations. C–E, Measurements were taken for the total distance traveled (C), the percentage of time spent in the correct quadrant (D), and the number of times the boundary of the platform was crossed (E). F, Immediately after the probe trial, a flag trial was performed and the latency to find the flag was recorded. Data shown are the mean ± SEM. n = 9 and 10 for saline and DCA treatments, respectively. No significant differences were observed between saline-treated (sham) and DCA-injected mice.