Permanent Suppression of Cortical Oscillations in Mice After Adolescent Exposure to Cannabinoids: Receptor Mechanisms

2.1. Animals

All experiments were performed according to the University of Maryland School of Medicine Institutional Animal Use and Care Committee protocols, and in compliance with the National Institute of Health guide for the care and use of laboratory animals. All efforts were made to minimize animal suffering and to reduce the number of animals used. Male CD-1 mice (Harlan Laboratories, Inc. Frederick, MD, USA) were obtained at postnatal day 30 (P30) and were treated with a variety of cannabinoid compounds, or a vehicle solution, via intraperitoneal (i.p.) injection. Treatment was administered for 20 days during adolescence (from P35–P55), or for 6 days during early (P35–P40) or late (P47–P52) adolescence (Figure 1A). After the last injection, animals were left undisturbed until adulthood (> P80). Animals assigned to groups in which the effects of antagonist + agonist administration were tested were given 2 separate injections, spaced 20–30 minutes apart, with the antagonist administered first. In antagonist alone groups, these injections were a) antagonist + vehicle; in agonist alone groups, they were b) vehicle + agonist; and in antagonist + agonist groups, animals received injections of c) antagonist + agonist. All injections were titrated such that total injection volume was ≤ 1 mL/kg.

2.2. Drugs

The CB1R/CB2R agonist WIN55,212–2 (1 or 2 mg/kg; Sigma Aldrich, St. Louis, MO, USA), the CB1R inverse agonist/antagonist AM251 (0.3, 0.5, 1, or 2 mg/kg), the CB1R/CB2R agonist Δ⁹tetrahydrocannabinol (THC; 5 mg/kg; National Institute on Drug Abuse Drug Supply Program, Bethesda, MD, USA), and the putative CB1R-inactive enantiomer WIN55,212-3 were dissolved in 100% ethanol and administered in a 1:1:18 solution of ethanol : Emulphor (Alkamuls EL-620, Rhodia Chemicals, USA) : 0.9% saline at a final volume of 1 mL/kg. Control animals were injected with vehicle (1:1:18 ethanol: Emulphor: saline). The CB1R neutral antagonist AM4113 (1 mg/kg; a generous gift from Dr. Alexandros Makriyannis, Northeastern University, Boston, MA) and the CB2R inverse agonist/antagonist AM630 (1 mg/kg; Tocris Bioscience, Bristol, UK) were dissolved in 1:1 dimethyl sulfoxide (DMSO) : 100% ethanol with sonication, and were injected in a 0.5:0.5:1:18 solution of DMSO solution: 100% ethanol: Emulphor: 0.9% saline. A separate cohort of control animals were injected with the DMSO-containing vehicle (0.5:0.5:1:18 DMSO: ethanol: Emulphor: saline) and were compared to the 1:1:18 vehicle animals to ensure that DMSO had no effect on its own. Pharmacologically-evoked cortical oscillations recorded in vitro did not differ significantly between DMSO-containing vehicle-administered mice and those exposed to the non-DMSO containing vehicle (data not shown). Therefore, we used only 1:1:18 (ethanol: Emulphor: saline) administered animals for our vehicle control population.

Doses of antagonist were chosen from the literature (Guindon et al., 2007; McMahon and Koek, 2007; Jarbe et al., 2008) and from preliminary studies in our lab that indicate antagonism of behavioral effects of 5 mg/kg WIN — a dose 2.5 and 5 times higher than what we repeatedly administered to adolescent mice — by 1 mg/kg AM251. Throughout the Methods and Results sections, AM251 (a CB1R inverse agonist/antagonist), AM4113 (a neutral CB1R antagonist), and AM630 (a CB2R inverse agonist/antagonist) will be classified as “antagonists,” and their respective mechanisms of antagonism will be discussed later. Kainic acid (KA) was dissolved in normal artificial cerebrospinal fluid (ACSF; recipe described below) at a concentration of 1 mM. The dissolved KA solution was brought to pH 7.4 by adding 1M NaOH, 1 μL at a time. Carbachol (CCh) was dissolved in MQ pure water at a concentration of 10 mM.

2.3. In vitro Slice Preparation and LFP Recordings

Experiments were performed by experimenters blind to the adolescent treatment condition of the animals. Adult male mice (> P80) were deeply anesthetized with ketamine (100 mg/kg) and rapidly decapitated. Two or three 400 μm-thick coronal sections were cut from each cortical region: medial prefrontal cortex (mPFC; ~ 1.3 – 2.3 mm anterior to bregma) and primary somatosensory cortex (SCx; ~1.0 – 2.1 mm posterior to bregma) from either the left or the right hemisphere, using a microtome (Integraslice 7550MM, Campden Instruments, IN, USA). During brain extraction and cutting, tissue was immersed in ice-cold or 4°C thermoregulated ACSF, respectively, containing (in mM): 26 NaHCO₃, 5 BES, 15 glucose, 200 sucrose, 3 KCL, 1.5 MgSO₄, 2 CaCl₂ (sucrose ACSF). Slices were then incubated for 30 minutes at 36°C, followed by 30 minutes at 22°C in ACSF containing (in mM): 120 NaCl, 25 NaHCO₃, 5 BES, 15 glucose, 3 KCl, 1.3 MgSO₄, 2 CaCl₂ (normal ACSF). Throughout cutting, recovery, and recording, slices were continually saturated with 95% O₂/5% CO₂. Slices were maintained at 36°C in an interface-type recording chamber and were perfused at 0.7 mL/min with ACSF containing (in mM): 120 NaCl, 25 NaHCO₃, 5 BES, 15 glucose, 5 KCl, 1.3 MgSO₄, 2 CaCl₂ (High K⁺ ACSF). Kainic acid (KA; 400 nM) and carbachol (CCh; 20 μM) were added to to reliably evoke robust beta (13–29 Hz) and gamma (30–80 Hz) oscillations (Buhl et al., 1998; Oke et al., 2010; Raver et al., 2013) that strongly resemble network activity in the intact neocortex (Steriade et al., 1996). We employed this in vitro pharmacological approach to provide a uniform stimulus to all samples. Local field potential (LFP) recordings were made through a glass pipette filled with normal ACSF (impedance ~0.5–1 MΩ) with an ER-1 amplifier (Cygnus Technology, USA), sampled at 5 kHz, filtered between 0.1 and 1 KHz, and stored on an Apple computer using Igor Pro (Version 6.1, Wavemetrics, Portland, OR, USA). LFPs were recorded from cortical layers II/III of the prelimbic area (PL) of mPFC, and layers II/III in the barrel field of SCx after at least 45 minutes of KA + CCh perfusion. Example KA + CCh evoked LFP oscillations are presented in Figure 1. These LFP recordings from adult animals treated in adolescence with a vehicle solution show that, in baseline conditions (ACSF alone), there is little oscillatory activity in either mPFC or SCx (Figures 1B&C). Robust robust oscillatory activity is evoked in slices containing mPFC (Figure 1B) or SCx (Figure 1C) during perfusion of KA + CCh.

2.4. In vitro LFP Data Analysis

Data were analyzed by experimenters blind to the treatment condition of the animals. We analyzed in vitro LFP data with custom-written Igor Pro scripts. Discrete fast Fourier transforms (FFTs) were performed on 10 seconds of LFP data and oscillation power (area under the FFT curve) was integrated at different frequencies [theta (θ 4–7 Hz); alpha (α 8–12 Hz); beta (β; 13–29 Hz); gamma (γ 30–80 Hz)]. Designation of frequency bandwidths is based on previous reports (Buhl et al., 1998; Uhlhaas and Singer, 2010). Example FFTs of LFPs from the mPFC and SCx of adolescent vehicle-treated animals are presented in Figure 1B and C, respectively. KA + CCh evokes robust oscillations in vitro in both cortical areas, as indicated by higher power at all frequencies in the presence of KA + CCh relative to baseline conditions, and peaks of oscillation power in the beta and gamma bandwidth in mPFC (Figure 1B), and in the gamma frequency range in SCx (Figure 1C). Average power in each frequency bandwidth was calculated for vehicle-administered mice and was used to normalize all other oscillation power values in the different treatment conditions. Two or three slices of mPFC and SCx were collected for each animal, and oscillation power in each bandwidth was normalized to the average oscillation power in that frequency bandwidth for vehicle-treated mice. Four normalized values (theta, alpha, beta, gamma) were obtained from each LFP (2–3 LFPs per animal, one per slice) and were combined for each treatment condition. This allowed us to reassemble the entire oscillation frequency spectrum, to test whether antagonists altered normalized oscillation power compared to vehicle or WIN/THC. Statistical analyses were performed with STATA (Version 12, StataCorp, College Station, TX, USA). Data were tested for normality, normalized oscillation power values were plotted with cumulative probability distributions and were analyzed with non-parametric Kolmogorov-Smirnoff (KS) tests (significant p < 0.05) to examine the effect of adolescent cannabinoid exposure on cortical oscillations in adulthood. The KS test statistic represents the distance between distributions of normalized oscillation power in animals treated with different cannabinoid compounds, or a vehicle solution. A KS test statistic that fails to reach statistical significance (p > 0.05) indicates that the distributions of normalized oscillation power overlap, and, therefore, that the cannabinoid treatments do not significantly alter the power of pharmacologically-evoked cortical oscillations relative to each other, or relative to oscillations in vehicle-treated animals. A KS test statistic that reaches statistical significance (p < 0.05) indicates that the distributions of normalized oscillation power from two different treatments are distinct. A significant leftward shift of one distribution relative to another (Figure 2B, for example) represents an attenuation of oscillation power, as a greater proportion of oscillations in the leftward-shifted treatment occur with lower power.

3.1. WIN exposure during early, but not late adolescence permanently suppresses cortical oscillations through CB1Rs

We tested the hypothesis that adolescent cannabinoid exposure suppresses pharmacologically-evoked oscillations in the adult neocortex by acting through CB1Rs. We predicted that WIN exposure in early adolescence (P35–P40) would attenuate oscillations in adult mice, and that this suppression would be mediated by the CB1R, and therefore antagonized by co-administration of the CB1R antagonist AM251. We also predicted that these suppressive effects would be more pronounced after WIN exposure early in adolescence than after similar exposure late in adolescence (P47–P52) (Figure 1A), due to more active development of the cortical eCB system earlier in adolescence (Heng et al., 2011; Cass et al., 2014).

Male CD-1 mice were treated with vehicle, WIN (2 mg/kg), AM251 (2 mg/kg) or AM251 + WIN for six days, from P35–P40 (Figure 2A), and oscillations were pharmacologically evoked in vitro with KA + CCh in adulthood (Figure 1B&C). In support of our prediction, WIN exposure during early adolescence significantly suppressed oscillations in adult mPFC, relative to vehicle control levels (Figure 2B, Table 1A). This is apparent in the significant leftward shift of the normalized distribution of oscillation power by WIN from that of vehicle (Figure 2B). We found no significant effect of AM251 administration alone (Figure 2C, Table 1A). Oscillation power in adult mice exposed to AM251 + WIN in early adolescence was significantly higher than in those exposed to WIN alone, and equal to oscillation power in animals treated with vehicle (Figure 2D, Table 1A), indicating that AM251 prevented the effects of early adolescent WIN exposure in mPFC. These data suggest that early adolescent WIN exposure suppresses oscillations in mPFC by acting on CB1Rs. SCx is less sensitive to the effects of early adolescent cannabinoid exposure, as neither WIN nor AM251 altered pharmacologically-evoked cortical oscillations when administered from P35–P40 (Figures 2E&F, Table 1B).

We further tested the temporal window of cortical oscillation sensitivity to adolescent cannabinoid exposure by restricting WIN exposure to 6 days late in adolescence, from P47–P52 (Figure 1A). In mPFC, neither WIN nor AM251 exposure in the late adolescent period significantly altered the power of pharmacologically-evoked oscillations, relative to vehicle levels (Table 1A). Similarly, late adolescent WIN or AM251 exposure had no effect on oscillations in SCx (Table 1B). These data further support a gradient of adolescent sensitivity to exogenous cannabinoid exposure that parallels the development of the adolescent cortical eCB system (Eggan et al., 2010; Heng et al., 2011; Lee et al., 2013), and complement a previous report from our laboratory indicating that cortical regions that are less mature at the time of adolescent cannabinoid exposure display the greatest suppressive effects (Raver et al., 2013).

3.2. Long-term adolescent THC exposure permanently suppresses pharmacologically-evoked cortical oscillations in vitro in SCx through CB1Rs

We have recently reported that a long-term, 20 day exposure to THC and WIN in adolescence significantly suppresses pharmacologically-evoked oscillations in adult mPFC and SCx (Raver et al., 2013). We tested the hypothesis that these effects of long-duration cannabinoid administration are mediated by CB1R activation. We predicted that this suppression could be antagonized by long-term exposure to the CB1R antagonist AM251. Adolescent mice were exposed to THC (5 mg/kg), AM251 (0.3 or 0.5 mg/kg), AM251 + THC, or vehicle from P35–P55, and cortical oscillations were pharmacologically evoked in mPFC and SCx once animals reached adulthood (Figure 3A). Statistical summaries of results from mPFC are presented in Table 2, and summaries of results from SCx are presented in Table 3. Long-term adolescent THC exposure significantly suppressed the power of pharmacologically-evoked oscillations in both adult mPFC (Figure 3B, Table 2) and SCx (Figure 3D, Table 3). To test whether the CB1R antagonist AM251 can prevent these suppressive effects, we first analyzed oscillations in animals administered AM251 alone, as our data and that from other groups predict no effect on oscillations of either acute (Robbe et al., 2006; Holderith et al., 2011; Sales-Carbonell et al., 2013) or short-term CB1R antagonist exposures (Figure 2C&F, Table 1) (Cass et al., 2014). Surprisingly, in mPFC, long-term adolescent treatment with both 0.3 and 0.5 mg/kg AM251 significantly suppressed the power of oscillations relative to vehicle levels (Figure 3B, Table 2). We therefore reasoned that we cannot use these doses of AM251 to try to prevent the chronic effects of THC in mPFC, as AM251 produced the same phenotype as THC exposure.

Pharmacologically-evoked oscillations in SCx were insensitive to long-term exposure to either 0.3 or 0.5 mg/kg AM251 during adolescence (Figure 3E, Table 3), and this allowed us to test the prediction that these doses of AM251 could attenuate the suppression of oscillation power by THC. AM251 (0.3 mg/kg) paired with THC prevented oscillation suppression by adolescent THC exposure (Figure 3F, Table 3). Oscillation power in animals treated with 0.3 mg/kg AM251 + 5 mg/kg THC was significantly higher than in mice exposed to THC alone, and was equal to vehicle power (Figure 3F, Table 3), indicating a full attenuation of oscillation suppression by CB1R antagonism. AM251 (0.5 mg/kg) + THC (5 mg/kg) partially attenuated the effects of THC. Oscillation power in SCx of animals treated with 0.5 mg/kg AM251 + THC fell between that of animals exposed to THC or vehicle, and was therefore statistically indistinguishable from either THC or vehicle power (Figure 3G, Table 3). The full attenuation of THC’s oscillation suppression by 0.3 mg/kg AM251, and the partial reversal by 0.5 mg/kg AM251, indicate that chronic adolescent THC suppresses oscillations in SCx by acting on CB1Rs.

3.3. Long-term adolescent CB1R antagonist (AM251) exposure permanently suppresses pharmacologically-evoked cortical oscillations in vitro in mPFC and SCx

We tested the prediction that long-term adolescent WIN exposure suppresses pharmacologically-evoked oscillations in adult neocortex by acting on CB1Rs. Adolescent mice were exposed to WIN (5 mg/kg), AM251 (1 mg/kg), AM251 + WIN, or vehicle from P35–P55, and cortical oscillations were evoked in adult animals, as described above (Figure 4A). Long-term adolescent exposure to WIN (1 mg/kg) significantly suppressed the power of pharmacologically-evoked oscillations in both mPFC (Figure 4B, Table 2) and SCx (Figure 4D, Table 3). We analyzed oscillations in animals exposed to AM251 (1 mg/kg) (Figure 4A) to test the prediction that AM251 has no effect on its own in mPFC. Animals were exposed to a slightly higher dose of AM251 in this study than what was used in the THC study (1 mg/kg vs. 0.3 or 0.5 mg/kg) because WIN is a full CB1R agonist, whereas THC displays partial agonist activity at the CB1R (Pertwee, 2008; Pertwee, 2010). Therefore, we predicted that a higher dose of AM251 would be necessary to antagonize WIN’s effects. Oscillations in mice chronically exposed to 1 mg/kg AM251 during adolescence were significantly suppressed in both mPFC (Figure 4B, Table 2) and SCx (Figure 4D, Table 3). We therefore reasoned that we could not use AM251 at this dose to try to antagonize the effects of chronic WIN exposure, as exposure to the antagonist alone mimicked the effects seen with WIN.

3.4. Long-term adolescent CB1R antagonist (AM4113) exposure permanently suppresses pharmacologically-evoked cortical oscillations in vitro in mPFC, but not SCx, and does not reverse WIN’s effects in SCx

AM251 can antagonist CB1R mediated effects of cannabinoid agonist through its action as an inverse agonist. Acting as such, AM251 both blocks the activity of endogenous and exogenous CB1R agonists and decreases the constitutive activity of CB1Rs (Pertwee, 2005). Therefore, we tested the prediction that the neutral CB1R antagonist AM4113 would antagonize WIN’s effects without altering oscillations on its own, since a neutral antagonist should not affect constitutive CB1R activity, but only block CB1R agonist action (Jarbe et al., 2008).

However, as was the case with AM251, long-term adolescent administration of the neutral CB1R antagonist AM4113 alone (1 mg/kg) (Figure 5A) significantly suppressed oscillations in mPFC, relative to vehicle-treated animals (Figure 5B, Table 2). We again reasoned that we could not use AM4113 to try to antagonize chronic WIN’s oscillation suppression in mPFC, as it produces the same phenotype as WIN alone. In SCx, AM4113 (1 mg/kg) did not alter oscillation power when administered alone (Figure 5D, Table 3), which allowed us to test the prediction that AM4113 could attenuate WIN’s suppression. This was not the case. Chronic adolescent administration of AM4113 did not attenuate the significant oscillation suppression induced by WIN in SCx, but, rather, further suppressed power relative to both vehicle and WIN levels (Figure 5E, Table 3).

3.5. Long-term adolescent CB2R antagonist (AM630) exposure permanently suppresses pharmacologically-evoked cortical oscillations in vitro in mPFC and SCx

While WIN acts as a full agonist at the CB1R, it also has high affinity for CB2Rs (Showalter et al., 1996; Pertwee, 2010). Although CB2R expression is thought to be confined to the peripheral tissues, mounting evidence indicates the presence of neuronal and glial CB2Rs in the CNS (Gong et al., 2006; Onaivi et al., 2006; Atwood and Mackie, 2010). Therefore, we tested the prediction that the CB2R antagonist, AM630, would not alter the response to WIN, if WIN suppressed oscillations exclusively via CB1Rs (Figure 5A).

As above, the requisite experiment is predicated on AM630 alone not having lasting effects on cortical oscillations. However, long-term adolescent AM630 exposure alone significantly suppressed oscillations in adult mPFC (Figure 5C, Table 2) and SCx (Figure 5F, Table 3). We were therefore unable to test whether AM630 can antagonize WIN’s effects, as AM630 produced similar results to those seen with WIN. We were also unable to exclude the possibility that CB2Rs may contribute to the effects of WIN, as a CB2R antagonist mimicked WIN’s effects.

3.6. Long-term adolescent exposure to a compound with putative CB1R inactivity (WIN-3) persistently suppresses pharmacologically-evoked cortical oscillations in vitro in mPFC, but not SCx

Thus far, we have shown that the CB1R/CB2R agonists WIN and THC, a CB1R inverse agonist/antagonist (AM251), a CB1R neutral antagonist (AM4113), and a CB2R inverse agonist/antagonist (AM630) suppress oscillations in adult neocortex when chronically administered during adolescence (for 20 days), with milder effects in SCx than mPFC. To further test whether these suppressive effects are specific to compounds with CB1R or CB2R activity, we exposed adolescent animals to a compound that is presumed to be inactive at cannabinoid receptors (Figure 6A). The chemical enantiomer of WIN55,212–2 (WIN)—WIN55,212-3 (WIN-3)—has been described to be inactive at the CB1R at low concentrations (Pacheco et al., 1991; Felder et al., 1992). If the oscillation suppressing effects of chronic adolescent WIN are due to its action at cannabinoid receptors, we predict that exposure to WIN-3 would have no effect on cortical oscillations.

Unexpectedly, chronic exposure to WIN-3 during adolescence significantly suppressed the power of oscillations in adult mPFC, relative to those recorded in vehicle-treated adult mice (Figure 6B, Table 2). In agreement with our prediction, we found no effect of adolescent WIN-3 on oscillations recorded in adult SCx (Figure 6C, Table 3), indicating that WIN-3’s lack of effect in SCx was due to its predicted mechanism as an inactive enantiomer. However, WIN-3 may suppress oscillations in mPFC via a non-CB1R mechanism. Therefore, while short-term (6 day) adolescent and acute WIN exposure suppress oscillations through CB1Rs, it appears that chronically administered WIN in adolescence may exert these suppressive effects in part through CB2Rs, or via non-cannabinoid receptor targets.

4.1. Overall Summary

Here we test the hypothesis that repeated cannabinoid exposure during adolescence permanently suppresses oscillations in adult neocortex by acting at CB1Rs. A short period of exposure (6 days) to the potent, synthetic cannabinoid WIN during early adolescence suppresses oscillations in mPFC, but not SCx, and this effect is mediated by CB1Rs. The same administration paradigm late in adolescence does not alter oscillations in either cortical region. The early adolescent sensitivity of mPFC to CB1R activation that we describe parallels recent work demonstrating a selective impairment of LFP modulation in the adult mPFC that is induced by short-term WIN exposure during early and mid, but not late, adolescence, and is attenuated by the CB1R antagonist AM251 (Cass et al., 2014). Data from adult SCx also support the hypothesis of a CB1R-mediated effect, as oscillation suppression by long-term adolescent THC exposure (20 days) is restored to vehicle levels by low doses of AM251 that do not produce effects on their own. Multiple attempts to test the prediction that mPFC exhibits similar CB1R-mediated suppression of oscillations were inconclusive, as CB1R or CB2R antagonists, at all doses attempted, suppressed oscillations in mPFC when administered alone.

4.2. CB1R modulation of cortical oscillations

Kainic acid and carbachol can evoke robust beta (13–29 Hz) and gamma (30–80 Hz) oscillations in vitro, in isolated hippocampal or cortical preparations. These drugs act by enhancing excitatory drive (Buhl et al., 1998; Cunningham et al., 2003) and by activating cholinergic receptors, primarily on GABAergic interneurons (Fisahn et al., 1998; Gulyas et al., 2010). In certain hippocampal circuits CCh can generate theta (4–7 Hz) rhythms by acting solely on inhibitory interneurons, when excitatory transmission is blocked (Nagode et al., 2011; Nagode et al., 2014). Experiments performed in the CA3 region of the hippocampus suggest that acute cannabinoid exposure suppresses pharmacologically-evoked gamma oscillations in vitro by acting at CB1Rs expressed on glutamatergic terminals (Holderith et al., 2011). This action may preferentially attenuate the frequency and firing precision of fast-spiking GABAergic interneurons that are crucial for sculpting oscillations in the gamma bandwidth (Cardin et al., 2009), resulting in smaller and less synchronized field potential activity (Holderith et al., 2011) that manifests as reduced oscillation power. A similar mechanism may underlie the CB1R-mediated suppression of oscillations that we find in SCx after long-term adolescent THC exposure, and in mPFC after early adolescent WIN exposure. However, the mechanism by which acute cannabinoid administration suppresses high-frequency oscillations in the neocortex is less clear. Genetic ablation of CB1Rs from glutamatergic neurons does not completely prevent the suppression of cortical oscillations faster than 12 Hz by acute cannabinoid exposure (Sales-Carbonell et al., 2013), as would be expected if cannabinoid-induced suppression of neocortical oscillations is mediated by a similar excitatory-CB1R mechanism as in CA3 hippocampus (Holderith et al., 2011). Furthermore, while activation of CB1Rs on inhibitory interneurons may not contribute to oscillation suppression by CB1R agonists in CA3 hippocampus in vitro (Holderith et al., 2011), removal of CB1Rs from all GABAergic forebrain neurons increases cannabinoid agonist-mediated suppression of oscillation power in cortical oscillations in vivo (Sales-Carbonell et al., 2013), suggesting a potential bi-directional regulation of oscillatory activity by CB1Rs at glutamatergic and GABAergic neurons. Interestingly, in the CA1 region of the hippocampus, CB1R-positive interneurons can generate theta oscillations (~4–7 Hz) when driven by CCh when excitatory neurotransmission is blocked, and these rhythms are completely abolished by exogenous or endogenous CB1R agonist activity (Nagode et al., 2011; Nagode et al., 2014), revealing a role for inhibitory-localized CB1Rs in pharmacologically-evoked oscillation suppression. Further tests of the contribution of CB1Rs expressed by excitatory or inhibitory neuronal populations to adolescent cannabinoid-induced suppression of cortical oscillations are necessary, and could be achieved by using transgenic mouse strains lacking CB1Rs in specific neuronal populations. Such experiments may help to clarify whether cannabinoid activity attenuates cortical oscillations through similar mechanisms as in hippocampal networks, or whether there are alternative mechanisms at work in the neocortex.

4.3. Preferential sensitivity of rostral mPFC vs. caudal SCx

In agreement with our recent report (Raver et al., 2013), we find that the adolescent mPFC is highly sensitive to manipulation of the eCB system. Long-term exposure to CB1R/CB2R agonists, CB1R antagonists, a CB2R antagonist, and a putative CB1R inactive enantiomer of WIN (WIN-3) all significantly attenuate pharmacologically-evoked oscillations in vitro in adult mPFC. We find that the SCx is less sensitive to adolescent cannabinoid exposure, in parallel with the caudal-to-rostral gradient of cortical and eCB system development (Gogtay et al., 2004; Heng et al., 2011). Oscillations in adult SCx, unlike in mPFC, are unaffected by either lower dose of the CB1R inverse agonist/antagonist AM251, or by the CB1R neutral antagonist AM4113. Furthermore, while a shorter period of cannabinoid exposure during early adolescence significantly suppresses oscillations in adult mPFC, oscillations in adult SCx are unaffected. Although this caudal-to-rostral gradient of cortical sensitivity probably reflects the prolonged adolescent and early adult maturation of prefrontal cortical circuitry (Gogtay et al., 2004) and network oscillations (Uhlhaas and Singer, 2011), it may be somewhat specific to cannabinoid exposure. The density of CB1Rs declines more dramatically from juvenile to adult ages in rostral, associative cortical areas than in caudal, primary sensory cortical regions (Heng et al., 2011), and expression of the CB1R is significantly higher in mPFC than in SCx from juvenile to adult ages (Heng et al., 2011) (Dr. K.Y. Tseng, personal communication). This gradient of CB1R expression may underlie the sensitivity of the mPFC to chronic adolescent cannabinoid exposure that we report, as well as the preferential suppression of oscillations in mPFC by short WIN treatment in early adolescence.

4.4. CB1R antagonism and cortical oscillations

We are surprised to discover that long-term adolescent exposure to the CB1R antagonists AM251 or AM4113, an inverse agonist and a neutral antagonist, respectively, significantly attenuate the power of oscillations in adult mPFC. Although several previous studies have shown that oscillation suppression by acute CB1R agonist exposure is reversed with co-administration of a CB1R antagonist (Hajos et al., 2000; Robbe et al., 2006; Hajos et al., 2008; Holderith et al., 2011; Sales-Carbonell et al., 2013), there have been few tests of whether these agents modulate oscillatory activity by themselves. Such studies often show that acute administration of CB1R inverse agonists do not affect oscillation power (Robbe et al., 2006; Holderith et al., 2011; Sales-Carbonell et al., 2013), leading to the conclusion that tonic eCB activity does not contribute to the generation of oscillations, at least those generated pharmacologically (Gulyas et al., 2010). However, there exists ample evidence to the contrary (Kim et al., 2002; Fortin et al., 2004; Nagode et al., 2014), as eCBs have been shown to be released by exposure to CCh at doses similar to what we have used, and these eCBs potently suppress pharmacologically-evoked oscillations by acting at CB1Rs (Nagode et al., 2014). Our data suggest a role for tonic eCB activity during adolescence, as long-term blockade of CB1Rs in adolescence with AM251 or AM4113 leads to a persistent suppression of oscillation power in the adult neocortex.

While we find it surprising that exposure to cannabinoid receptor antagonists recapitulates the effects seen with agonist administration, these effects are not unprecedented (Manzanedo et al., 2010). Indeed, chronic administration of CB1R antagonists can induce lasting changes in CB1R distribution and eCB levels (Castelli et al., 2007; Guidali et al., 2011), reminiscent of those seen after repeated CB1R agonist administration (Hunter and Burstein, 1997; Romero et al., 1997; Sim-Selley and Martin, 2002; Rubino and Parolaro, 2008).

Through its action as a CB1R inverse agonist, AM251 modulates tonic CB1R activity by both reducing the receptor’s constitutive activity in the absence of an agonist, and by occluding the action of eCBs (Pertwee, 2005). AM4113 also interferes with endogenous CB1R agonist activity, but it lacks the ability to modulate constitutive CB1R signaling (Chambers et al., 2007; Sink et al., 2008). Therefore, similar results produced by both AM251 and AM4113 in mPFC suggest that chronic interference with ongoing CB1R activity during adolescence—such as antagonism or displacement of eCBs at the CB1R—impairs the ability of the adult mPFC to generate pharmacologically-evoked oscillations. Further tests of this hypothesis could be performed by chronically administering indirect cannabinoid agonists that prevent the synthesis or breakdown of eCBs, and would therefore interfere with tonic eCB signaling in adolescence.

4.5. Non-CB1R contribution to oscillation suppression

In SCx, we find that long-term adolescent exposure to the neutral CB1R antagonist AM4113 alone does not alter oscillations in adults, but does not reverse long-term adolescent WIN-induced oscillation suppression. Surprisingly, AM4113 paired with WIN suppresses oscillation power further from that seen with WIN alone, suggesting that AM4113 and WIN may act synergistically and suppress oscillations, or act at different targets. Because WIN has potent CB2R agonist activity in addition to its action as CB1Rs (Showalter et al., 1996; Atwood et al., 2012), we posit that CB2Rs may contribute to oscillation suppression by WIN. CB2Rs are expressed in both mPFC and SCx (Gong et al., 2006; Onaivi et al., 2006), positioning them to contribute to the phenotype we observe. This hypothesis of a CB2R contribution is supported by our data showing a significant suppression of oscillations in adult mPFC and SCx by the CB2R antagonist AM630. Although centrally expressed CB2Rs are increasingly implicated in diverse functions, including cannabinoid-mediated memory impairments and drug reward and reinforcement (Atwood and Mackie, 2010; Xi et al., 2011; Garcia-Gutierrez et al., 2013), to date, little information exists regarding a possible for of the CB2R in network activity or oscillations. Interestingly, in entorhinal cortex, CB2R activation suppresses GABAergic inhibition that is not occluded by CB1R antagonism (Morgan et al., 2009), revealing the functional expression of CB2Rs at central inhibitory synapses that may contribute to oscillation generation. Additional tests of a possible contribution of CB2Rs to oscillations are warranted.

Further evidence that non-CB1Rs may contribute to oscillation suppression in mPFC by long-term adolescent cannabinoid exposure comes from experiments with the putative inactive enantiomer of WIN, WIN-3. We find that WIN-3 unexpectedly and significantly attenuates oscillations in mPFC, but has no effect in SCx. Although converging evidence of mPFC oscillation suppression by chronic adolescent exposure to CB1R/CB2R agonists, CB1R antagonists, and a CB2R antagonist suggest the involvement of CB1Rs and/or CB2Rs in this suppression, similar suppression by WIN-3 suggest that the effects of WIN may be, at least in part, cannabinoid receptor independent. Indeed, previous reports reveal that WIN-3 can produce identical effects as WIN that are not mediated by CB1Rs (Price et al., 2004; Price et al., 2007; Nemeth et al., 2008). Our data present the possibility that long-term adolescent WIN exposure may affect the development of cortical oscillations through non-CB1R/CB2R mechanisms (Pertwee, 2010). These mechanisms may involve the non-CB1R/CB2R putative cannabinoid receptor GPR55 (Ryberg et al., 2007), the TRPV-1 like receptors (Pertwee, 2006) that modulate glutamate release, as well as allosteric modulation of receptors for other neurotransmitters (Barann et al., 2002), or direct modulation of cation permeable channels (Shen and Thayer, 1998).