Figure 1.
Schematic of our physiologically based computational model.
Retinal light input is projected to the suprachiasmatic nucleus (SCN) and ventrolateral preoptic area (VLPO). Mutual inhibition between the sleep-promoting VLPO and wake-promoting monoaminergic (MA) nuclei forms the basis of the sleep/wake switch (outlined). During wake, MA firing rates are high and VLPO firing rates are low. SCN output is relayed to the sleep/wake switch via the subparaventricular zone (SPZ), dorsomedial hypothalamus (DMH), and orexinergic neurons in the lateral hypothalamus (LHA). The MA also receives excitatory cholinergic input (ACh). A homeostatic sleep drive (H) to the VLPO is modeled. Wake/sleep state and rest/activity are defined by MA firing rate, , which provides behavioral feedback to H and gates environmental light.
Figure 2.
SPZ modulation of SCN output yields a spectrum of nocturnal to diurnal phenotypes.
Model simulations of rodent sleep/wake patterns are shown for different values of the modulation parameter , which takes values from −1 (nocturnal) to 1 (diurnal) going from left to right. The simulated rodent was under a 24-h light/dark cycle with 12 h of 100 lux followed by 12 h of 0 lux. Panels (A)–(E) show the MA firing rate,
, across a 24-h period (blue) and averaged across this period in 10-min non-overlapping windows (black). Wakefulness is defined as
s−1. Panels (F)–(J) show arousal state across a 24-h period (red), with high values corresponding to wake and low values corresponding to sleep, as well as average percentage wakefulness (black), averaged across 30 days in 10-min non-overlapping windows.
Figure 3.
Cooperation and competition between DMH/VLPO and DMH/LHA relays result in unimodal and bimodal activity patterns, respectively.
Simulated MA firing rate, , for a primate under a LD cycle, with 500 lux for 6–18 h and 0 lux for 18-6 h (shaded). The DMH/LHA pathway is (A) excitatory (i.e., cooperative), or (B) inhibitory (i.e., competitive). (C) The averaged activity patterns of a spider monkey living in the wild; the data are adapted from [12] and replotted manually here. Simulations and data are averaged in 5 min non-overlapping windows over 60 days and double-plotted.
Figure 4.
Circadian and homeostatic processes in cooperative versus competitive networks.
The average waveforms (over 60 days) are shown for (A) cooperative and (B) competitive DMH/VLPO and DMH/LHA relays. Parameter values correspond to (A) and (B) in Figure 3, respectively. In each panel, we show the average waveform of the homeostatic drive for sleep (gray line), as well as the average waveforms of the circadian drives for wakefulness to both the VLPO via the DMH (black solid line) and the MA via the DMH/LHA (black dashed line). More positive values of the circadian drives for wakefulness correspond to greater inhibition of VLPO and greater excitation of MA for the DMH/VLPO and DMH/LHA relays, respectively. Panel (C) shows a double-plotted raster diagram of sleep (dark bars) over 60 days for the competitive case in panel (B).
Figure 5.
Experiments and simulations of temporal niche switching in degus.
Temporal niche switching can be elicited in degus by the presence of a running wheel (days 1–22 and 68–108) in either LD (30 lux from clock time 8–20; days 1–88) or DD conditions (days 89-end). Panel (A) shows data from a single animal, with dark bars representing periods of above average core body temperature. Panels (B)–(D) show raster plots for single runs of the model using the same protocol, with dark bars representing periods of high activity ( s−1), averaged in 10 min sliding windows. All raster plots are double-plotted. The effect of the running wheel is modeled by: (B) masking and circadian signal inversions, (C) circadian signal inversion, and (D) masking inversion. The data in panel (A) are adapted from [9] and manually replotted here.
Figure 6.
Simulations and data for SCN lesions in the squirrel monkey.
Double-plotted raster plots of sleep/wake patterns are shown for (A) simulated intact animals and (B) simulated SCN-lesioned (SCNx) animals, with dark bars representing sleep ( s−1). Percentage of time spent awake as a function of circadian time is shown for (C) data (mean ± SEM), and (D) simulation, for both intact (open circles) and lesioned (filled squares) animals. Circadian time is defined with respect to the 25.0-period of the intact animal, with 24 h of circadian time corresponding to one full circadian cycle. Circadian time zero is defined as the first bout with >50% average wake in the intact animal. Spectral amplitudes (normalized to an area of 1) are plotted versus period for (E) Drinking patterns in the intact animal, (F)
in the simulated intact animal, (G) Drinking patterns in the SCN-lesioned animal, (H)
in the simulated SCN-lesioned animal. Note that the model uses a higher sampling rate than data. The data in panels (C), (E), and (G) are adapted from [43] and manually replotted here.