Cortical zeta-inhibitory peptide injection reduces local sleep need

Abstract

Local sleep need within cortical circuits exhibits extensive interregional variability and appears to increase following learning during preceding waking. Although the biological mechanisms responsible for generating sleep need are unclear, this local variability could arise as a consequence of wake-dependent synaptic plasticity. To test whether cortical synaptic strength is a proximate driver of sleep homeostasis, we developed a novel experimental approach to alter local sleep need. One hour prior to light onset, we injected zeta-inhibitory peptide (ZIP), a pharmacological antagonist of protein kinase Mζ, which can produce pronounced synaptic depotentiation, into the right motor cortex of freely behaving rats. When compared with saline control, ZIP selectively reduced slow-wave activity (SWA; the best electrophysiological marker of sleep need) within the injected motor cortex without affecting SWA in a distal cortical site. This local reduction in SWA was associated with a significant reduction in the slope and amplitude of individual slow waves. Local ZIP injection did not significantly alter the amount of time spent in each behavioral state, locomotor activity, or EEG/LFP power during waking or REM sleep. Thus, local ZIP injection selectively produced a local reduction in sleep need; synaptic strength, therefore, may play a causal role in generating local homeostatic sleep need within the cortex.

Statement of Significance

The need for sleep arises because of waking activity, yet the specific biological changes within the brain that necessitate sleep are unclear. We developed a novel experimental approach (local injection of zeta-inhibitory peptide) to produce a local reduction in sleep need without affecting sleep need throughout other brain regions or behavioral state. These data support a potential role for increased synaptic strength as a proximate driver of local sleep need. Continued elucidation of the biological mechanisms responsible for sleep homeostasis is critical for addressing the multitude of sleep disorders characterized by insufficient sleep and/or dysregulated sleep homeostasis.

Introduction

It is both intuitive and experimentally validated that sleep is homeostatically regulated; sleep need increases as a consequence of waking duration and/or activity [1]. Despite the existence of a well-established electrophysiological measure of sleep need (slow-wave activity [SWA], the electroencephalographic [EEG] power between 0.5 and 4.0 Hz during NREM sleep), the waking-induced neurobiological alterations that necessitate sleep remain unclear. Identifying these neurobiological substrates is essential for understanding the function of this enigmatic behavior.

Many hypotheses propose that regulating synaptic strength is central to the function of sleep. The synaptic homeostasis hypothesis [2, 3] suggests that sleep functions to produce global downscaling of synaptic strength to alleviate metabolic and/or signaling burdens imposed by waking-induced learning and associated synaptic potentiation. By contrast, memory consolidation hypotheses [4–8] suggest that a primary function of sleep is to strengthen synapses to enhance and/or preserve waking-induced learning. These prominent hypotheses differ concerning the consequences of sleep itself for synaptic strength, yet each typically presupposes that increased synaptic strength during waking serves as a proximate driver of sleep need. Such proposals are consistent with widespread correlational evidence between molecular [9–12], electrophysiological [9, 13–15], and ultrastructural [11, 16, 17] indices of waking-induced synaptic potentiation and sleep need.

The need for sleep is not unitary throughout the brain as evident by extensive regional variation in SWA. Cortical SWA typically decreases along the anterior–posterior axis [18, 19], increases locally within brain regions that experienced enhanced waking activity and/or plasticity [20–22], and decreases locally following inactivity during waking [23, 24]. If synaptic strength truly underlies sleep need, this regional variability could arise as a consequence of learning-induced synaptic plasticity; learning also appears to be an inherently local process wherein synaptic strength of specific neuronal circuits is persistently modified to produce a novel engram [25–27]. However, it remains unclear whether counteracting waking-induced plasticity can functionally alleviate sleep need.

Experimentally counteracting widespread waking-induced plasticity is difficult as the vast majority of pharmacological interventions that prevent the induction and expression of plasticity must be administered during the initial learning or subsequent memory reactivation [28–31]. By contrast, zeta-inhibitory peptide (ZIP), an antagonist of protein kinase Mζ (PKMζ), can produce pronounced depotentiation of previously strengthened synapses without the need for prior memory reactivation [32–34]. PKMζ is an atypical isoform of protein kinase C that is rapidly translated (10–40 min) within local dendritic processes following learning and/or the induction of long-term potentiation [35, 36]. By increasing clustering of the postsynaptic anchor protein, PSD-95 [37], and augmenting N-ethylmaleimide-sensitive factor (NSF) trafficking of GLUR2-containing α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPAR) [38], PKMζ ultimately enhances synaptic strength by persistently increasing postsynaptic AMPAR receptor expression [32, 34, 39, 40]. Ongoing PKMζ activity appears critical for the continued maintenance and expression of LTP and long-term memory [41, 42]. Consequently, ZIP-induced inhibition of PKMζ may reduce synaptic strength and thereby serve as a powerful tool to test whether potentiated synapses are responsible for sleep need.

In the present study, we locally injected ZIP into the motor cortex of freely behaving rats immediately preceding their primary sleep phase. We hypothesized that ZIP-mediated PKMζ inhibition within this anterior cortical region that typically generates high levels of SWA would produce a pronounced reduction in sleep need as evident by reduced motor cortex SWA and/or NREM sleep duration.

Methods

Surgery

Three- to 4-month-old, male Sprague-Dawley rats (n = 8, Charles River; Wilmington, MA) were housed individually under standard laboratory conditions (12 hr light/dark cycle, access to food and water ad libitum). Stereotactic surgery was performed under isoflurane anesthesia (3.5% induction, 2%–3% maintenance). Once anesthetized, rats were given a subcutaneous preoperative analgesic (Meloxicam; 2 mg/kg; MWI, Boise, ID) and an intramuscular preoperative antibiotic (Penicillin; 100 000 units/kg). During surgery, the rat’s skull was exposed to enable implantation of electrodes, anchor screws, and a cannula. Specifically, to record the local field potential (LFP), we affixed a Teflon-coated stainless steel wire (0.005″ bare diameter, A-M systems, Sequm, WA) to the outside of an injection cannula (C313GRL-SPC, 22 gauge; Plastics One, Roanoke, VA) that was implanted into the right motor cortex (anterior–posterior [AP]: +2.0 mm, medial–lateral (ML): +3.0 mm, dorsal–ventral [DV]: −1.5 mm). A second wire was attached to a screw affixed to the skull above the left parietal cortex (approximate AP: −4.0 mm, ML: −4.3 mm) and served as an electroencephalogram (EEG). Two additional stainless steel wires were affixed to screws above the cerebellum and served as a reference for our LFP/EEG leads and as a ground. Lastly, we implanted two braided stainless steel wires under the nuchal muscles to serve as an electromyogram (EMG). All wires were connected to a commercially available headmount (8239 2EEG/1EMG Rat Headmount, Pinnacle Technologies; Lawrence, KS), and all wires, anchor screws, and the headmount were affixed in place with dental acrylic (Lang Dental; Wheeling, IL). Twelve to 24 hr postsurgery, all rats received a postoperative analgesic (Meloxicam; 2 mg/kg). Rats were allowed a minimum of seven complete days to recover following surgery before experiments commenced. These methods and those below were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by Middlebury College’s Institutional Animal Care and Use Committee.

Local cortical injections

Each rat received two local cortical injections (vehicle and zeta-inhibitory peptide [ZIP]) through the indwelling cannula in the right motor cortex. Vehicle solution consisted of an 80% saline/20% ethanol mix. ZIP solution was prepared by dissolving 1 mg of ZIP (Tocris, Minneapolis, MN) into 100 µL of vehicle (final concentration: 5.04 mM). A syringe pump (11 Plus, Harvard Apparatus, Holliston, MA) was used to deliver each injection (0.5 µL/min; final volume of 12.5 nmol ZIP in 2.5 µL) to the rat in its home cage in the hour preceding light onset. Saline and ZIP injections were separated by 24 hr, with saline injections always preceding ZIP injections. All rats were additionally recorded during a post-ZIP recovery day during which time no injections were administered (see Figure 1A for complete experimental timeline).

Figure 1.

Local ZIP injection does not alter the sleep/wake cycle or locomotor activity. (A) Experimental timeline. (B) Typical electrophysiological recordings depict how neuronal activity (RMLFP/LPEEG) and muscle tone (EMG) change across behavioral state. Behavioral state was manually scored in 4 s epochs throughout the experiment. (C) Average time spent in each behavioral state, bout duration, and locomotor activity (across all states, or waking epochs only) did not significantly differ following saline and ZIP injections. RMS = root-mean squared. N = 8 (except EMG activity where N = 7).

Electrophysiological recordings and behavioral state determination

To begin our chronic recordings of EEG/LFP/EMG signals, a flexible preamplifier (100× amplification, EEG/LFP high pass filter: 0.5 Hz, EMG high pass filter: 10 Hz; Pinnacle Technologies) was attached to the rat’s headmount. The preamplifier was connected to a commutator (SL6C, Plastics One, Roanoke, VA) to enable unobstructed movement throughout the rat’s home cage. Electrophysiological signals were passed into the data acquisition system (8401 DACS, Pinnacle Technologies) and were continuously recorded at 250 Hz (Sirenia Acquisition, Pinnacle Technologies) for the duration of the experiment. To characterize behavioral state, each day’s recorded electrophysiological signals were manually scored in 4 s epochs by visual inspection. Waking was classified as low-voltage, high-frequency EEG/LFP activity along with elevated EMG activity. NREM sleep was classified as high-voltage, low-frequency EEG/LFP activity and an absence of EMG activity. REM sleep was classified as low-voltage, high-frequency EEG/LFP activity and an absence of EMG activity. Vigilance state could be resolved for all epochs. Epochs that contained movement artifact in either EEG/LFP channel (<3% of all scored epochs) were assigned behavioral state as described above but were not included in any time-series analyses (see descriptions below). Due to instability in the signal across the entire recording period, subsequent analyses also excluded the left parietal EEG from one rat and the EMG from another rat.

Data processing, analyses, and statistical approaches

All electrophysiological data were processed and analyzed using custom scripts in Mathworks MATLAB (Natick, MA). Additional statistics including dependent-samples t-tests and repeated measures ANOVAs were performed with SPSS (IBM, Armonk, NY). All data are presented as mean ± the standard error of the mean.

To provide a broad measure of locomotor activity, we calculated the root-mean square of EMG activity for each 4 s behavioral state epoch [43]. EEG/LFP power spectra were calculated for each 4 s epoch via Welch’s method [44–46] (hamming window). State-dependent power spectra were obtained by averaging all artifact-free epochs of a given behavioral state (as determined above). Throughout this report, all raw EEG/LFP data and power spectral data are presented as µV and µV², respectively. Recorded voltages were not calibrated separately for each individual electrode; recorded signals may therefore exhibit interelectrode variability in the absolute voltages recorded. However, as all analyses presented in this report are conducted both within-subject and within-electrode, any variability arising from electrode characteristics will not affect the statistical approaches used herein.

To calculate SWA, EEG/LFP power between 0.5 and 4.0 Hz was summed for all artifact-free NREM epochs. Average SWA was calculated to provide a measure of sleep pressure [1]. To provide a measure of the cumulative amount of SWA expressed throughout the light period, we calculated slow-wave energy (SWE). This measure accounts for both the average amount of SWA expressed as well as the total duration of NREM sleep, thereby providing an index of the total efficacy of preceding sleep [47, 48]. SWE is calculated as a cumulative measure by summing SWA from NREM epochs across the light period.

Individual slow-wave properties were also characterized using methods similar to those previously reported [13]. Briefly, EEG/LFP signals were band-pass filtered in the delta frequency range (0.5 to 4.0 Hz) with a zero-phase Chebyshev Type II filter. We then extracted all EEG/LFP activity that occurred during epochs previously identified as NREM sleep. Individual slow waves were identified from local maxima. Maxima within 200 ms were considered part of the same individual slow wave (i.e. multipeak waves), whereas maxima separated by greater than 200 ms were considered unique individual slow waves. For each individual slow wave identified, we calculated the amplitude, ascending slope (a measure of transition towards the down-state of the slow oscillation [13]), descending slope (transition towards the up-state [13]), and the number of peaks. These values were then used to assess whether sleep pressure and/or local injections affected the characteristics of individual slow waves to better understand how these factors affect synchronization and desynchronization of neurons during the NREM slow oscillation.

Results

Local ZIP injections do not alter global behavioral state or locomotor activity

We first assessed whether ZIP-induced PKMζ inhibition within the right motor cortex affected behavioral state and/or locomotor activity (see Figure 1A for complete experimental timeline). Behavioral state was manually classified in 4 s epochs across the entire recording period using recorded LFP, EEG, and EMG signals (see Figure 1B for example traces). Local ZIP injection did not significantly affect time spent in each behavioral state across the 24 hr day [treatment x state: F (2, 14) = 0.92, p = 0.42; Figure 1C]; waking (saline: 59.65 ± 4.10%, zip: 55.59 ± 4.93%), NREM sleep (saline: 33.97 ± 2.99%, zip: 37.49 ± 4.09%), and REM sleep (saline: 6.39 ± 1.23%, zip: 6.92 ± 1.25%) durations were similar following saline and ZIP injections.

Given the polyphasic nature of rodent sleep, we examined whether ZIP injection instead produced fragmentation or consolidation of sleep/wake bouts. Local ZIP injection, however, did not significantly affect the average bout duration of any behavioral state [treatment: F (1, 7) = 0.74, p = 0.42; treatment x state: F (2, 14) = 1.60, p = 0.24; Figure 1C]; similar waking (saline: 5.43 ± 0.61 min, zip: 4.71 ± 0.58 min), NREM sleep (saline: 2.83 ± 0.16 min, zip: 2.97 ± 0.22 min), and REM sleep bout durations (saline: 1.71 ± 0.08 min, zip: 1.82 ± 0.10 min) were observed. Lastly, we quantified EMG activity recorded from the nuchal muscles to determine whether local ZIP injection into the motor cortex produced any gross abnormalities in motor activity (Figure 1C). Across both the entire 24 hr day [t (6) = 0.23, p = 0.82] and during waking specifically [t (6) = 1.18, p = 0.28], we did not observe any significant difference in EMG activity between saline and ZIP injection days. Thus, local ZIP injections do not appear to significantly alter global behavioral state or locomotor activity.

Local ZIP injections alter local electrophysiological activity during ensuing sleep

Neuronal activity during sleep is influenced by global behavioral state as well as more proximate physiological conditions and can produce differences in local sleep need at different cortical locations [13, 49, 50]. Consequently, although our local ZIP injections did not appear to affect global behavioral state, they may have instead altered local neuronal activity and/or local sleep need. To test these possibilities, we calculated power spectra at our injection site for each behavioral state during the first and last 3 hr of the light period (when global sleep pressure is highest and lowest, respectively; Figure 2). We then ran a three-way repeated measures ANOVA to assess the effects of frequency, treatment, and global sleep pressure on LFP power (see Table 1 for complete results). As expected, we observed significant main effects of frequency on LFP power within each behavioral state consistent with the well-characterized 1/f nature of typical LFP power spectra [51]. During waking and NREM sleep, we also observed a significant interaction between frequency and global sleep pressure.

Figure 2.

Average RMLFP power spectra under conditions of high- and low-sleep pressure (first and last 3 hr of the light period, respectively) following saline and ZIP injections. Complete statistical analyses of these data are presented in Table 1. Black circles depict significant differences within each frequency bin between saline and ZIP conditions (post hoc t-tests; uncorrected for multiple comparisons, p < 0.05).

Table 1.

Statistical analysis of RMLFP power spectral data presented in Figure 2 and similar analyses for LPEEG data

Behavioral state	Factor(s)	DF	F value	P	LPEEG p value
Waking	Frequency	38, 266	16.33	0.000	0.000
	Condition	1, 7	0.05	0.82	0.68
	Time	1, 7	0.25	0.64	0.35
	Frequency x Condition	38, 266	0.42	0.99	0.90
	Frequency x Time	38, 266	2.03	0.001	0.000
	Condition x Time	1, 7	3.41	0.11	0.73
	Frequency x Condition x Time	38, 266	2.27	0.000	0.94
NREM	Frequency	38, 266	22.93	0.000	0.000
	Condition	1, 7	2.06	0.20	0.51
	Time	1, 7	15.61	0.006	0.03
	Frequency x Condition	38, 266	2.03	0.001	0.98
	Frequency x Time	38, 266	16.28	0.000	0.000
	Condition x Time	1, 7	2.23	0.18	0.14
	Frequency x Condition x Time	38, 266	2.84	0.000	0.26
REM	Frequency	38, 266	8.58	0.000	0.000
	Condition	1, 7	0.24	0.64	0.50
	Time	1, 7	3.91	0.09	0.26
	Frequency x Condition	38, 266	1.38	0.08	0.94
	Frequency x Time	38, 266	0.84	0.75	0.000
	Condition x Time	1, 7	0.13	0.73	0.24
	Frequency x Condition x Time	38, 266	0.47	0.99	0.64

Results from a three-way, repeated measures, ANOVA comparing effects of frequency (between 0.5 and 20 Hz with 0.5 Hz bins), condition (saline vs. ZIP), and time (early vs. late light period) on local field potential power at the site of injection. Significant p values are bolded with trends in presented in grey. p Values obtained from identical analyses performed on LPEEG data are presented in the final column.

Strikingly, although local ZIP injections produced alterations in LFP power at the injection site during the ensuing light period (Figure 2), these ZIP injections only significantly affected LFP power during periods of subsequent sleep. Specifically, during NREM sleep, we observed a significant interaction between treatment and frequency (this interaction trended towards significant during REM sleep). Post hoc analyses (Figure 2; paired t-tests comparing ZIP and saline power within each frequency bin, uncorrected for multiple comparisons [52, 53]) revealed that ZIP primarily produced reductions in LFP power within two frequency ranges, when sleep pressure was high: (1) a low-frequency range (0.5 – 6.0 Hz) inclusive of SWA and (2) a higher frequency range (15.5 – 18.5 Hz). These effects were largely absent when sleep pressure was low (Figure 2). Consistent with these results, we also observed a significant three-way interaction (Frequency x Condition x Time), during NREM sleep (Table 1).

To determine whether the effects of ZIP injection extend beyond the site of injection, we performed the same analyses presented above on data recorded from a contralateral, parietal EEG (Table 1). We again observed a main effect of frequency during all behavior states, a main effect of time during NREM sleep, and a significant interaction between frequency and time during NREM sleep. Strikingly, however, we did not observe any main effects of condition or any significant interactions that include condition. In other words, ZIP injections did not significantly affect EEG activity away from the injection site while typical homeostatic alterations in EEG activity were maintained.

Local ZIP injection, therefore, appears to selectively modulate certain oscillatory patterns at the injection site during sleep without significantly affecting the waking power spectrum. Given that these ZIP-induced alterations appear to strongly manifest as reductions in NREM sleep SWA and that this oscillatory activity is highly associated with sleep function, we focused the remainder of our analyses on further characterizing how ZIP-induced PKMζ inhibition alters this distinct pattern of neuronal communication.

Local ZIP injections reduce local SWA and SWE without affecting activity at a distal cortical location

SWA is the best electrophysiological marker of sleep need, exhibiting the highest values at the onset of the major sleep phase and dissipating as the sleep period progresses [1]. At both our injection site (the right motor cortex) and at a distal cortical location (left parietal cortex), we observe this canonical decrease in SWA across the light period [Figure 3, A and B; F (11, 77) = 15.94, p < 0.001 and F (11, 66) = 5.01, p < 0.001, respectively]. ZIP injection significantly reduced SWA at the site of injection [F (1, 7) = 8.63, p < 0.05], but did not alter SWA recorded from the left parietal cortex [F (1, 6) = 0.12, p = 0.74]. Post hoc tests (paired t-tests, p < 0.05; Figure 3) show that the ZIP’s effects were only significant during the beginning of the sleep period (hours 2–8 after light onset). We did not observe any significant interaction between ZIP injection and time of day in either the motor cortex [F (11, 77) = 0.51, p = 0.89] or the parietal cortex [F (11, 66) = 1.18, p = 0.32]. Thus, a local ZIP injection 1 hr prior to light onset produced a selective reduction in SWA at the injection site without significantly altering SWA away from the injection site. These effects were statistical significant early in the sleep period, when sleep pressure is typically highest; we did not, however, observe significant differences in SWA in the last 4 hr of the light period.

Figure 3.

Local ZIP injection selectively reduces SWA and SWE at injection site. (A and B) Average hourly SWA (the summed EEG/LFP power between 0.5 and 4.0 Hz) during the light period following injection of saline or ZIP. SWA significantly decreases across the light period in both locations following both injections, whereas ZIP injection significantly reduces SWA only at the injection site (p < 0.05). A black circle indicates a significant hourly difference between ZIP and Saline (post hoc paired t-tests; p < 0.05). (C and D) Cumulative SWE during the light period following injection of saline or ZIP. ZIP injection significantly reduces SWE at the injection site only (p < 0.05). SWE is depicted for each channel as a percentage of the total SWE observed during the light period following saline injection. N = 8 (right motor); N = 7 (left parietal).

To further quantify the extent to which local ZIP injection reduced local sleep need, we calculated SWE. SWE is a cumulative measure of SWA that takes into account both the magnitude of SWA and the duration of sleep to provide an index of the cumulative SWA that has been produced throughout the subjective night [47, 48]. In the motor cortex, ZIP injection significantly reduced SWE [Figure 3C; F (1, 7) = 7.14, p < 0.05] with average reduction of 14.65 ± 4.96% by the end of the light period when compared with saline control. By contrast, parietal SWE was not significantly altered following ZIP injection in the motor cortex [Figure 3D; ZIP SWE: 110.37 ± 16.97% of saline control at the end of the light period, F (1, 6) = 1.14, p = 0.33]. The interaction between time and condition was statistically significant within the motor cortex [F (11, 77) = 2.57, p < 0.01] but not within the parietal cortex [F (11, 66) = 0.83, p = 0.61]. These data collectively demonstrate that the accumulation of SWE in the motor cortex is significantly slower following a ZIP injection than a saline injection and that distal SWE is unaffected by local ZIP injection.

Reduced slope and amplitude of individual slow waves underlies ZIP-induced reductions in SWA/SWE

Although SWA is currently the best electrophysiological marker of sleep need, this measure is sensitive to a variety of changes in the characteristics of the individual slow waves that collectively constitute SWA. Consequently, to better understand potential mechanisms of ZIP-induced reductions in SWA, we identified all individual slow waves present during NREM sleep (see Methods) and characterized how they were affected by local ZIP injection (see Figure 4, A and B for average waveforms). Consistent with previous reports [13, 54], across the light period, we observed that individual slow waves exhibited: significantly decreased amplitude, significantly decreased ascending and descending slopes, and a significant increase in the likelihood of containing multiple peaks (Figure 4, C and D; see Table 2 for complete stats). These main effects were independent of injection type and reflect typical homeostatic alterations in individual slow-wave characteristics.

Figure 4.

Homeostatic sleep pressure and local ZIP injections affect characteristics of individual slow waves. (A and B) Average waveforms of individual slow waves in the motor and parietal cortices under high global sleep pressure (first 3 hr of light period; solid lines) and low global sleep pressure (last 3 hr of light period; dashed lines) following saline or ZIP injection. (C and D) Average amplitude, ascending/descending slope, and multipeak wave incidence in the motor and parietal cortices following saline or ZIP injection during high global sleep pressure (first 3 hr of light period), medium sleep pressure (hours 4.5–7.5 of light period), and low sleep pressure (last 3 hr of light period). Each characteristic is plotted as a percentage of its corresponding value observed under high sleep pressure following saline injections. ZIP injections significantly reduce slow-wave amplitude and ascending/descending slope (see Table 2 for complete stats for C/D). N = 8 (right motor); N = 7 (left parietal).

Table 2.

Statistical analysis of individual slow-wave data presented in Figure 4

Recording site	Slow-wave characteristic	Factor(s)	DF	F value	P
Motor cortex	Amplitude	Sleep Pressure	2, 14	45.28	0.000
		Condition	1, 7	7.65	0.03
	Sleep Pressure x Condition	2, 14	0.86	0.45
	Ascending slope	Sleep Pressure	2, 14	42.87	0.000
		Condition	1, 7	9.78	0.02
		Sleep Pressure x Condition	2, 14	0.82	0.46
	Descending slope	Sleep Pressure	2, 14	43.29	0.000
		Condition	1, 7	9.95	0.02
		Sleep Pressure x Condition	2, 14	1.01	0.39
	Multipeak wave incidence	Sleep Pressure	2, 14	25.00	0.000
		Condition	1, 7	1.94	0.21
		Sleep Pressure x Condition	2, 14	1.67	0.22
Parietal cortex	Amplitude	Sleep Pressure	2, 12	35.09	0.000
		Condition	1, 6	0.004	0.95
		Sleep Pressure x Condition	2, 12	1.51	0.26
	Ascending slope	Sleep Pressure	2, 12	32.29	0.000
		Condition	1, 6	0.03	0.86
		Sleep Pressure x Condition	2, 12	1.20	0.34
	Descending slope	Sleep Pressure	2, 12	28.97	0.000
		Condition	1, 6	0.01	0.91
		Sleep Pressure x Condition	2, 12	1.45	0.27
	Multipeak wave incidence	Sleep Pressure	2, 12	15.17	0.001
		Condition	1, 6	0.94	0.37
		Sleep Pressure x Condition	2, 12	0.92	0.43

Results from two-way repeated measures ANOVAs comparing effects of global sleep pressure (high, medium, and low) and condition (saline vs. ZIP) on characteristics of individual slow waves at the injection site. Significant p values are bolded.

We then examined whether ZIP injection itself altered any characteristics of individual slow waves. ZIP injection did not alter the average number of individual slow waves observed per hour of the light period in either the motor [Saline: 3,973.60 ± 264.64, ZIP: 3,876.20 ± 239.64; t (7) = 0.38, p = 0.72] or parietal [Saline: 3,765.00 ± 323.08, ZIP: 4,143.60 ± 213.86; t (6) = −1.42, p = 0.20] cortices. Likewise, average wave duration in the motor cortex [Saline: 0.382 ± 0.003 s, ZIP: 0.383 ± 0.003 s; t (7) = −0.65, p = 0.53] and parietal cortex [Saline: 0.371 ± 0.005 s, ZIP: 0.369 ± 0.006 s; t (6) = 0.68, p = 0.52] was unaffected by ZIP injection. Instead, ZIP injection appeared to alter the average waveform of slow waves produced near the injection site without altering individual slow waves distally (Figure 4, A and B). Similar to the typical homeostatic alterations described above, local ZIP injection significantly reduced the amplitude and ascending/descending slope of individual slow waves in the motor cortex (Table 2). This effect was not observed in slow waves recorded outside of the injection site. Unlike the typical homeostatic effects described above, ZIP injection did not significantly alter the proportion of multipeak waves observed in either the motor or parietal cortex. Our observed changes in SWA/SWE, therefore, may arise from ZIP-induced alterations to specific characteristics of individual slow waves.

ZIP-induced reductions in slow waves are not compensated for during other behavioral states, the subjective day, or the following post-ZIP recovery day

Diminished SWA during NREM sleep may reflect a reduction in local sleep need. Alternatively, this result could arise from an inability to fully generate this homeostatically regulated activity. In this latter case, we would expect to see homeostatic compensation to ameliorate the lack of sufficient NREM SWA observed following local ZIP injections. We therefore assessed whether local ZIP injections altered SWA in the motor cortex during times other than the light period, NREM sleep data presented above. When compared with saline control, SWA following local ZIP injection was not significantly different during light period wake or REM sleep episodes [t (7) = 0.09, p = 0.93; t (7) = 0.27, p = 0.80, respectively. See Figure 2]. Ensuing dark period wake [t (7) = −0.02, p = 0.98] and REM sleep [t (7) = −0.10, p = 0.92] episodes also exhibited comparable SWA to the corresponding period following saline injection. Thus, there does not appear to be a redistribution of SWA from NREM sleep to other behavioral states following local ZIP injection.

A lack of sufficient SWA can be compensated for during recovery sleep in subsequent NREM periods [1, 48]. SWA during the dark period did not significantly differ between saline and ZIP injection days [t (7) = 0.69, p = 0.51]. Additionally, we examined whether a homeostatic rebound in SWA was present 24 hr following ZIP injection. During the light period of this post-ZIP recovery day, neither SWA [t (7) = 1.01, p = 0.34] nor SWE [t (7) = 0.94, p = 0.38] were significantly different from saline control. These results demonstrate that ZIP-induced reductions in SWA do not: (1) produce a permanent reduction in the ability of motor cortex to generate slow waves and (2) appear to be homeostatically compensated for at any other time point.

Discussion

When compared with vehicle injections, local injections of ZIP into the motor cortex of freely behaving rats significantly reduced subsequent light-period NREM SWA and SWE in the injected motor cortex but not within the contralateral parietal cortex. This reduction in SWA appears to be mediated by changes in the characteristics of individual slow waves including smaller amplitudes and decreased ascending and descending slopes. This local reduction in SWA did not appear to be compensated for during NREM sleep in the dark period, during the post-ZIP recovery day, or during waking or REM sleep. Moreover, ZIP injections did not appear to affect the NREM power spectrum apart from SWA nor did it significantly affect the waking or REM sleep power spectra. Collectively, these novel results demonstrate that local cortical injections of ZIP can serve as a powerful tool to produce a local reduction in homeostatic sleep need. Moreover, they appear consistent with the idea that synaptic potentiation is causally responsible for generating sleep need within the cerebral cortex.

ZIP, a pseudosubstrate inhibitor of PKMζ, affords a unique tool to test whether synaptic plasticity reduces sleep need. PKMζ is both necessary and sufficient for the continued expression of long-term potentiation [33, 35, 55] by maintaining elevated expression levels of postsynaptic AMPARs [38, 40]. ZIP blocks PKMζ activity and produces synaptic depotentiation in vitro [32, 55, 56] and in vivo [33], with in vivo application additionally producing corresponding memory impairment [33, 57]. These effects do not require prior reactivation of previous memory traces and affect both recently potentiated synapses and long-term (i.e. days to months) memory traces (reviewed in [34]). Consequently, local injection of ZIP into cortical tissue can produce extensive and robust synaptic depotentiation without altering the total amount of neuronal activity exhibited during the preceding waking hours.

Recent reports, however, have raised concerns about the specificity of ZIP’s mechanism of action [58–60]. Paradoxically, off-target ZIP effects have been reported to both decrease LFP power [58] and increase neuronal activity [59], with the latter ultimately leading to excitotoxic cell death. Instead of specific inhibition of PKMζ activity, ZIP may therefore independently alter neuronal excitability. As discussed below, our present data do not appear consistent with nonspecific ZIP-induced alterations in neuronal excitability and instead appear more consistent with ZIP-induced synaptic depotentiation. Of note, however, the current study is unable to directly identify the precise downstream effects of ZIP that produced the observed reductions in local SWA.

The half-life of injected ZIP is ~2 hr [61], and previous reports indicate that spontaneous neuronal activity is not significantly altered after this 2 hr time point [62, 63]. We observed a ZIP-induced decrease in NREM SWA that was significant between 2 and 8 hr after injection. By contrast, ZIP-induced decreases in LFP power have been reported [58] in urethane-anesthetized rats in the 120 min following an intrahippocampal ZIP injection of a similar dose to the present study (10 nmol in 1 µL vehicle and 12.6 nmol in 2.5 µL vehicle, respectively). In that report, LFP power was predominantly reduced at low frequencies (<5 Hz) and was only recorded from the anesthetized brain. In the present study, we report significant reductions in low-frequency LFP power during NREM sleep, a behavioral state that, like the anesthetized brain, can be characterized by an abundance of large slow waves [47, 64]. Critically, however, we did not observe any significant alterations in LFP power during waking or REM sleep as would be expected from a nonspecific ZIP-induced decrease in neuronal excitability. The observed reductions in low-frequency power during NREM sleep and anesthesia, therefore, may reflect impairments in neuronal recruitment, rather than decreased excitability. Computer simulations have predicted that a decrease in synaptic strength could account for reduced SWA, with weaker synapses failing to produce robust neuronal synchronization leading to reduced amplitude and slopes of individual slow waves [65, 66]. We experimentally observed these predicted changes in individual slow-wave characteristics following local ZIP injection, further supporting the likelihood that our observed effects of ZIP on local SWA are mediated by synaptic depotentiation.

ZIP-induced increases in the neuronal activity of cultured neuronal tissue have also been reported [59]. Specifically, ZIP increased intracellular calcium levels and the frequency of miniature excitatory postsynaptic potentials and this increased excitability was associated with subsequent excitotoxic cell death [59]. As spontaneous waking increases the amplitude and frequency of mEPSCs [14] along with sleep pressure, we would expect similar ZIP-induced excitation to increase, rather than decrease, SWA as observed. Should our observed reduction in SWA instead arise from excitotoxic cell death, we would not expect to have observed: (1) a SWA recovery during the following day after local ZIP injections (no significant difference in SWA was observed between baseline and post-ZIP recovery day) and (2) alterations to waking and/or REM sleep LFP spectra. Thus, although significant concerns surrounding off-target ZIP effects have been raised, the most parsimonious explanation for the totality of our results is that local ZIP injections produce local synaptic depotentiation and a reduction in homeostatic sleep need.

Changes in the characteristics of individual slow waves shed further light on both the origin and consequences of ZIP-induced local reductions in sleep need. An individual slow wave can originate anywhere within the cortex and typically propagates throughout as additional neurons are recruited into a more global up state [67]. This neuronal recruitment is not always complete; local slow waves may be present in one cortical region yet fail to propagate to another [49, 50]. When multiple slow waves originate at distinct locations, their asynchronous propagation and convergence may generate multipeak slow waves within other cortical regions [54]. Thus, cortical SWA emerges from interactions between both local and global phenomena. In the present study, ZIP injections did not significantly alter the total number of slow waves or the number of multipeak waves at the injection site. Instead, local ZIP injection appeared to disrupt injection site neuronal recruitment/decruitment during individual slow waves as evident by significant reductions in individual slow-wave amplitudes and slopes [65, 66]. That local ZIP injections also failed to alter any characteristics of individual slow waves recorded in the parietal EEG further confirms that ZIP injections are capable of producing a specific and local reduction in sleep need as indexed by SWA.

Not only did our local ZIP injections fail to alter cortical activity at distal locations, they also did not significantly alter episode duration or proportion of time spent in each behavioral state. In Drosophila, a reciprocal network that includes disparate neurons responsible for generating sleep pressure and promoting behavioral sleep has been partially elucidated [68]. No analogous mechanism in vertebrates has been described; thus, whether cortical sleep pressure can directly affect communication between wake- and sleep-promoting subcortical nuclei that constitute the “sleep switch” [69–72] and thereby alter global behavioral state remains unclear. As recently reviewed [73], however, SWA and total sleep time in vertebrates can be highly divergent and therefore may be independently regulated. For example [74], systemic atropine injections produce a dissociated state characterized by a sleep-like cortex (elevated SWA and decreased c-Fos expression), a wake-like behavioral phenotype, and wake-like increases in subcortical c-Fos expression. After 6 hr in this dissociated state, rats exhibit a prominent rebound in both SWA and total sleep time. Thus, subcortical activity within nuclei that comprise the “sleep switch” may regulate behavioral state and homeostatic sleep need independently of cortical activity. Future studies that produce reductions in sleep need throughout the cortex are necessary to elucidate whether residual sleep need outside our injection site accounts for the lack of observed effects on behavioral state or whether behavioral state and cortical sleep need are independently regulated.

The distinctly local effects of PKMζ-inhibition on SWA are consistent with sleep as an emergent property of local cortical ensembles [75, 76]. Local SWA variability may facilitate the independent regulation of sleep need by local cortical circuits throughout NREM sleep. Consistent with recent proposals [4, 77–79], such independent regulation could account for both the sleep-induced synaptic downscaling central to the synaptic homeostasis hypothesis [2, 3] and sleep-induced synaptic maintenance and/or strengthening inherent to memory consolidation hypotheses [4–8]. In fact, a selective downscaling of weak synapses during NREM up-states has recently been demonstrated [80, 81]. With the ability to modulate sleep pressure selectively within cortical networks, local ZIP injections establish a promising novel methodological approach that can be used to further investigate the relationships between local sleep need, SWA, behavioral state regulation, and the function of sleep itself. Further characterization of the downstream effects of local ZIP injection will shed light on the underlying mechanisms responsible for the genesis of local sleep need. As ZIP has previously been shown to produce pronounced synaptic depotentiation, it will be particularly instructive to determine whether local ZIP injection counteracts markers of synaptic plasticity that have been previously associated with an increase in local sleep need including: (1) molecular (e.g. increased AMPAR phosphorylation and expression, calcium/calmodulin-dependent protein kinase II phosphorylation, and Homer1a expression) [9–11], (2) structural (e.g. increased number and size of dendritic spines) [16, 17], and (3) electrophysiological markers (e.g. increased slope and/or amplitude of evoked responses [9, 15] and increased frequency/amplitude of spontaneous miniature excitatory postsynaptic currents [14]). Lastly, given the relatively small N of the present study, further investigation of the effects of ZIP on EEG/LFP activity apart from SWA is warranted. It would be valuable to explore the observed statistical trend of a main effect of injection condition on REM power, especially in regards to REM theta activity (see Figure 2). Additionally, significant decreases in higher-frequency LFP power during NREM sleep (i.e. 15.5–18.5 Hz) suggest that local ZIP injections may provide a useful approach for exploring the coordination of SWA and higher-frequency oscillatory activity during NREM sleep.

Funding

This work was supported by Middlebury College Start-Up Funds.

Conflict of interest statement. None declared.