NMDA receptors subserve persistent neuronal firing during working memory in dorsolateral prefrontal cortex

Abstract

Neurons in the primate dorsolateral prefrontal cortex (dlPFC) generate persistent firing in the absence of sensory stimulation, the foundation of mental representation. Persistent firing arises from recurrent excitation within a network of pyramidal Delay cells. Here, we examined glutamate receptor influences underlying persistent firing in primate dlPFC during a spatial working memory task. Computational models predicted dependence on NMDA receptor (NMDAR) NR2B stimulation, and Delay cell persistent firing was abolished by local NR2B NMDAR blockade or by systemic ketamine administration. AMPA receptors (AMPARs) contributed background depolarization to sustain network firing. In contrast, many Response cells were sensitive to AMPAR blockade and increased firing after systemic ketamine, indicating that models of ketamine actions should be refined to reflect neuronal heterogeneity. The reliance of Delay cells on NMDAR may explain why insults to NMDARs in schizophrenia or Alzheimer's disease profoundly impair cognition.

Figure 1

The experimental paradigm and dlPFC neural circuitry underlying spatial working memory. A. The ODR spatial working memory task. Trials began when the monkey fixated on a central point for 0.5 sec. A cue was present in 1 of 8 possible locations for 0.5 sec and was followed by a delay period of 2.5 sec. When the fixation point was extinguished, the monkey made a saccade to the location of the remembered cue. The position of the cue changed on each trial in a quasi-random manner, thus requiring the constant updating of working memory stores. B. The region of monkey dlPFC where recordings occurred. PS=principal sulcus; AS=arcuate sulcus. C. An example of a Delay cell with spatially tuned, persistent firing during the delay period. Rasters and histograms are arranged to indicate the location of the corresponding cue. The neuron’s preferred direction and the opposing, nonpreferred direction are indicated; subsequent figures will show neuronal responses to only these two directions. This cell exhibited significant delay-related activity for the 180° location but not other directions. D. An illustration of the deep layer III microcircuits subserving spatially tuned, persistent firing during the delay period, based on Goldman-Rakic, 1995. B = GABAergic Basket cell. E. Working model of glutamate actions at NMDAR and AMPARs on long, thin dendritic spines of layer III pyramidal cells in monkey dlPFC.

Figure 2

NMDAR in primate dlPFC: immunoEM labeling and computational theory. A–D. Localization of NMDA NR2B subunits using immunogold labeling in layer III of the rhesus monkey dlPFC. Four typical synapses are shown, including a perforated synapse in D: Fig. 1A shows pNR2B labeling, while Figs. 1B–D show total NR2B label. Both pNR2B and NR2B labeling was found exclusively within the post-synaptic density; no labeling was observed outside the synapse. Black arrowheads indicate pNR2B or NR2B labeling; white arrows delineate the synapse. E. The effects of iontophoretic NMDA blockade on spatial working memory activity in a computational model of dlPFC neuronal persistent firing. Under control conditions, a stimulus cue selectively activates a group of neurons, leading to persistent activity sustained by NMDAR dependent recurrent excitation. NMDA conductance is reduced from control (i) to 90% (ii), 80% (iii) and 70% (iv) of a reference level, in 10 pyramidal neurons in the network model. Stimulus-selective persistent activity gradually decreases with more NMDAR blockade, and eventually disappears in these affected cells; model based on Brunel and Wang (2001) and Wang (2002). See text for more details.

Figure 3

The effects of intra-PFC iontophoresis of the NMDA antagonists MK801 or Ro25-6981 on the task-related firing of Delay cells in the primate dlPFC. A. An example of an individual dlPFC Delay cell under control conditions and following iontophoresis of MK801 (25nA). The rasters and histograms show firing patterns for the neuron’s preferred direction and the nonpreferred direction opposite to the preferred direction. Iontophoresis of MK801 markedly reduced task-related firing; firing returned towards control levels when delivery of MK801 was stopped (Recovery; p<0.05). B. Average response showing the mean±SEM firing patterns of 15 dlPFC Delay cells for their preferred vs. nonpreferred directions under control conditions (blue) and following iontophoresis of MK801 (red). MK801 markedly decreased task-related firing, especially for the neurons’ preferred direction. C. The spatial Tuning Index (TI) comparing each neuron’s firing for its preferred vs. nonpreferred directions to examine the neuron’s spatial tuning. Iontophoresis of MK801 significantly weakened spatial tuning by reducing TI. D. An example of an individual dlPFC Delay cell under control conditions and following iontophoresis of Ro25-6981 (15–25nA). Iontophoresis of Ro25-6981 markedly reduced task-related firing in a dose-dependent manner; firing returned towards control levels when delivery of Ro25-6981 was stopped (Recovery; p<0.05). E. Average response showing the mean±SEM firing patterns of 31 dlPFC Delay cells for their preferred vs. nonpreferred directions under control conditions (blue) and following iontophoresis of Ro25-6981 (red). Ro25-6981 markedly decreased task-related firing, especially for the neurons’ preferred direction. F. Iontophoresis of Ro25-6981 significantly weakened spatial tuning by reducing TI.

Figure 4

The effects of NMDA vs. AMPAR blockade on the task-related firing of Cue and Response cells in the primate dlPFC. A. Example of a Cue cell under control conditions (blue) and following iontophoresis of the NMDA NR2B antagonist, Ro25-6891 (15nA; red). NMDA blockade significantly reduced task-related firing of the Cue cell. B. Example of a peri-saccadic Response cell under control conditions (blue) and following iontophoresis the AMPA antagonist, CNQX (25nA; green), and Ro25-6891 (25nA; red). Peri-saccadic-related firing of the Response cell was reduced by NMDA but not AMPAR blockade. C. Example of a Cue cell under control conditions (blue) and following iontophoresis of CNQX (25nA; green). AMPA blockade significantly reduced task-related firing of the Cue cell. D. Example of a post-saccadic Response cell under control conditions (blue) and following iontophoresis of CNQX (25nA; green). In contrast to the Response cell shown in 4B, the post-saccadic-related firing of this Response cell was reduced by AMPAR blockade.

Figure 5

The effects of AMPAR blockade on the task-related firing of Delay cells in the primate dlPFC. A. An example of an individual dlPFC Delay cell under control conditions and following iontophoresis of NBQX (40nA). Iontophoresis of NBQX reduced task-related firing as the delay period progressed. B. Average response showing the mean±SEM firing patterns of 16 dlPFC Delay cells under control conditions (blue) and following iontophoresis of CNQX orNBQX (green), with the drug effects being most prominent late in the delay period. C. Iontophoresis of CNQX/NBQX weakened spatial tuning by reducing TI.

Figure 6

A comparison of AMPA vs. NMDAR blockade on the task-related firing of Delay cells in the primate dlPFC. A. Left graph: The percentage of neurons showing significant reduction in firing rate following iontophoresis of the NMDA antagonist, MK801 compared to the AMPA antagonists CNQX or NBQX. Right graph: The maximal degree of reduction in delay-related firing induced by the NMDA antagonist MK801 compared to the AMPA antagonists CNQX or NBQX. The reduction in firing rate was measured by the following ratio: (control-drug)/control. B. An example of an individual Delay cell treated with NMDA vs. AMPA antagonists. Under control conditions, the neuron showed prominent, spatially-tuned, delay-related firing (dark blue). Subsequent iontophoresis of the NMDA NR2B antagonist, Ro25-6981 (25nA; red), led to a large reduction in task-related firing. The iontophoretic current was then turned off and the neuron recovered normal rates of firing (light blue). Following recovery, the AMPA antagonist CNQX (40nA, green) was iontophoresed onto the neuron. CNQX had little effect on firing early in the delay epoch, but reduced firing in the later portion of the delay epoch. C. Average response showing the mean±SEM firing patterns of the 8 dlPFC Delay cells under control conditions (dark blue), during iontophoresis of Ro25-6981 (25nA; red), and during iontophoresis CNQX (40nA; green). Ro25-6981 produced a marked reduction in task-related firing, CNQX had more subtle effects, reducing firing only in the later aspects of the delay epoch. D. A comparison of mean±SEM firing rates in the five successive 0.5s epochs of the 2.5s delay period under control, MK801 and CNQX conditions. * p<0.05; ** p<0.01

Figure 7

The effects of systemic ketamine administration on the working memory performance and the physiological responses of Delay cells and Response cells in the primate dlPFC. A. The systemic administration of ketamine significantly impaired the accuracy of spatial working memory performance on the ODR task. Data represent mean±SEM collapsed across all doses (0.5–1.5 mg/kg). See text for breakdown in performance between lower and higher doses. B. The effects of systemic ketamine administration on the spontaneous firing rate of Delay cells (n=6), Response cells (n=6), and nontask-related cells (n=4) when the monkeys were resting and not performing the task. Ketamine had no significant effect on the spontaneous firing of Delay cells or nontask cells, but significantly increased the spontaneous firing of Response cells. C. An example of the effects of ketamine on the task-related firing of an individual Delay cell in the dlPFC. This neuron showed pronounced task-related firing for its preferred direction under control conditions (blue), but reduced task-related firing following injection of ketamine (red). D. Systemic administration of ketamine significantly reduced the task-related firing of the 6 Delay cells found in the monkey dlPFC. Results represent mean±SEM firing rate during the delay epoch. E. An example of the effects of ketamine on the task-related firing of an individual Response cell in the dlPFC. This neuron showed increased post-saccadic firing under control conditions (blue), which was markedly increased following injection of ketamine (red). F. Systemic administration of ketamine significantly increased the task-related firing of 6 Response cells in the monkey dlPFC. All of these Response cells showed post-saccadic firing patterns. Results represent mean±SEM firing rate during the response epoch.