The present study demonstrates that the response to cocaine and sucrose of many genes in dopamine projection areas and the hypothalamus is regulated in a time-dependent manner. While interactions between circadian rhythms and drugs of addiction within the mesocorticolimbic reward system have been reported previously [59, 26, 60, 61], our findings show that cocaine and sucrose, respectively a drug of abuse and a strong natural reinforcer in rodents, trigger very different changes in gene and protein expression in these brain areas. The genes and proteins investigated part of mechanisms controlling DNA methylation, appetite and satiety and circadian rhythms, underlining that depending on the reinforcer type, alterations in critical biological functions and their behavioral consequences may be very different. The current data further support initial fMRI and electrophysiological studies [1–3] and provide new insights into molecular and cellular mechanisms showing that the neural circuits activated by both reinforcers do not overlap.
Drugs of abuse and natural reinforcers can lead to addiction or dependence characterized as complex relapsing disorders occurring overtime in which compulsive seeking behavior can persist despite aversive consequences. Dopamine is known to play an essential role in the motivational aspect of reward and the dopamine theory of addiction highlights its major and common role in the mesolimbic system [62, 63]. That is why we have looked at dopamine projection areas (PFCx and CPu), although drugs of abuse also sensitize noradrenergic and serotonergic neurons via non-dopaminergic mechanisms [64, 6, 60, 65] and rewarding drugs such as opiates or psychostimulants induce different behavioral and neurobiological responses [66, 67]. Developing pharmacological intervention specifically targeting dopamine receptors or its transporter system has been also reported to be a daunting task for the treatment of drugs of abuse [68]. Thus, the unitary account of addiction involving dopamine [67] has been challenged by several studies.
We have compared chronic passive cocaine intake with that of sucrose, since the latter is the main source of energy in the brain and a major component in highly palatable food. While sweetness has been reported to surpass cocaine reward in rodents [46], sugar addiction has been subjected to controversial issues [69, 70]. Contrary to cocaine, little evidence has been provided for human sugar addiction or for a sugar-addiction model of overweight to support clinical DSM criteria, suggesting that sugary food does not directly promote excessive weight or obesity, but rather contributes minimally to 'food dependence' and increased risk of weight gain by overeating [71, 72]. By analyzing body weight development, no significant differences were found between cocaine and sucrose treated rats and their respective control groups (supplementary Fig. S2).
Overlap between drugs of abuse and natural reinforcers has been well illustrated by considering orexin, NPY and ghrelin. Indeed, these peptides regulate appetite and satiety and their receptor antagonists have been characterized as promising therapeutic targets for addictive behaviors, drug abuse disorders or metabolic diseases [36–38, 56, 73]. Orexinergic neurons widely project throughout the brain including key structures of the limbic system which are "multi-tasking" neurons regulating functions like arousal, sleep/wake states, feeding behavior, energy homeostasis, anxiety and addictive behaviors. Orexin mRNA was induced by both cocaine and sucrose in the CPu although with a different timing (Fig. 6c). In the LH, i.e. its site of transcription, the number of orexin positive cells was increased by cocaine, but decreased by sucrose 15 hours after the last administration (Fig. 7, ZT16). This increase elicited by cocaine is in line with the upregulation of the number of LH orexin neurons after exposure to various drugs of abuse [74] and with the critical role of orexins in controlling arousal [75, 76] and in the motivation for cocaine notably illustrated in Orx and Orx R1 knockdowns [35, 77]. Moreover, it is noteworthy here that the Orx R1 gene has been shown to be regulated by cocaine through DNA methylation [5].
Variations in Dnmt and Tet gene expression in response to reinforcers have been documented in various experimental models. Dnmts display biphasic time-course regulation following cocaine chronic passive treatment or in cocaine self-administering rats [12, 8, 9, 5]. Tissue-specific biological functions and specific target sequences have been reported for both gene families [78–80]. Our data concerning cocaine treatment are consistent with these studies by showing a heterogeneous pattern of expression depending not only on the brain structures, but also on the time of the day considered (Figs. 2 and 3). We also found a significant increase in global DNA methylation following cocaine administration and a decrease in response to sucrose in the PFCx and the CPu (Fig. 4), in contrast to other global DNA methylation analyses that have reported very little changes in response to cocaine [31, 81, 82, 32, 83, 34]. To some extent, chronic passive and voluntary drug intake cause similar behavioral and molecular effects. We therefore looked for differentially methylated regions (DMRs) associated with the 15 genes analyzed here from a genome-wide study performed in the PFCx of cocaine self-administering rats [34]. DMRs were found for 13 genes studied here, suggesting that they are strongly regulated by DNA methylation. In agreement with previous global analyses, most DMRs were found within gene bodies for which no obvious correlation was established between 5mC hypermethylation and gene repression. Nevertheless, a single hypermethylated DMR was identified upstream at -2 419 bp relative to the transcription start site of the Clock gene with a methylation ratio [cocaine]/[saline] of 1.44, that was repressed by 40% at ZT16. In addition, an upstream DMR at -2 719 bp relative to the transcription start site of the Dbp1 gene was repressed by 52% at ZT11, while it was found to be hypermethylated with a 5mC ratio [cocaine]/[saline] of 1.18. The Cry2 gene that is repressed at ZT11 and ZT16 by 36 and 26%, respectively, was found to be hypermethylated in a downstream sequence with a 5mC ratio [cocaine]/[saline] of 1.75. Considering that cocaine and sucrose regulate differentially the expression of Dnmt and Tet genes (Fig. 2, 3) as well as global DNA methylation (Fig. 4) in the PFCx and the CPu, one may speculate that both agents consequently differ in modulating the rhythmicity of day/night cycles.
While the role of orexins in the central and peripheral regulation of glucose homeostasis and metabolism has been reported, the effect of sugar on orexin expression or production has been less documented. Nevertheless, hypothalamic orexin neurons regulate arousal according to energy balance and a negative correlation was found between orexin expression and blood glucose and food intake [84]. Glucose was also reported to inhibit orexin neurons [85, 86] and conversely glucodeprivation was activating them [87]. In this respect, the decrease in orexin positive cells found in the LH following sucrose passive intake (Fig. 7) further illustrates a marked difference relative to passive cocaine administration. In the CPu, sucrose was found to induce orexin mRNA levels (Fig. 6c), which may sound surprising relative to the decreased orexin peptide expression in the LH (Fig. 7). However, orexins display local brain structure specific functions, so such differences may be attributed to different regulatory mechanisms between peptide production in the LH, where it is exclusively transcribed and its mRNA transport to the CPu recently documented [5].
Unlike orexins, Npy neuropeptide is widely expressed throughout the central nervous system and is secreted along with GABA and glutamate [56]. Because of its widespread distribution including the peripheral nervous system, it is implicated in multiple physiological processes like cortical excitability, stress response, food intake, energy metabolism, memory, sleep regulation and circadian rhythms [57]. While its regulation in the CNS has been studied extensively in the hippocampus and the hypothalamus, its expression in response to cocaine relative to sucrose has not yet been well investigated in the PFCx and CPu. In the PFCx, its baseline expression was not affected at any time and it was solely repressed at ZT11 by both cocaine and sucrose (Fig. 7a). This similar regulation suggests a common role of both reinforcers in NPY-mediated cortical cognitive functions such as short term/working or maintenance of memory, attention and arousal. In the CPu, Npy showed a time-of-day dependence in both control groups (Fig. 7b), indicating that in this brain structure, unlike other genes, its expression is not affected by daily NaCl injections which may be associated with stress or pain [55] that can also affect DNA methylation in male rats notably following nerve injury [88].
As expected, analyzing all genes relative to those solely regulated by circadian rhythms revealed a high number modulated by both reinforcers (supplementary Fig. S3). Indeed, most genes were induced by sucrose in the PFCx (10/14) and in the CPu (10/15), whereas most of them were repressed by cocaine in the PFCx (9/14) and the CPu (8/15). Hence, sucrose and cocaine strongly differ in the regulation of 11/14 genes in the PFCx and 10/15 in the CPu at least at one time point. The overall data further confirm earlier functional magnetic resonance imaging [1], electrophysiological [2, 3] and molecular studies [4, 5] by showing that neural circuits activated by cocaine and natural reinforcers do not overlap.
In summary, our data highlight a discrepancy in the molecular and cellular mechanisms triggered by sucrose and cocaine in dopamine projection areas and in the hypothalamus. We previously reported that various genes were differentially affected by voluntary administration of food vs. cocaine [4, 5]. By analyzing key factors involved in DNA methylation, circadian rhythms and appetite and satiety in passive intake, we provide new insights into marked differences in mechanisms occurring in the brain upon administration of a drug of abuse or a natural reinforcer independently of the mode of administration. Whether the described cocaine-induced changes are long lasting remains to be investigated, but cocaine and not natural reward was shown to produce persistent LTP in the VTA [89] and persistent variations in neuronal DNA methylation [90]. In addition, as recently reviewed, some studies have documented long-lasting changes in the expression of epigenetic enzymes and molecules that persist for weeks after the last drug exposure [91].
The regulation of DNA methylation factors by both agents raises questions regarding their common and tissue-specific target-methylated DNA sequences [92, 93, 79] and their role in the establishment and the development of drug addiction. Cocaine alters epigenome with potential consequences on future generations and de novo DNA methylation changes can be inherited in acquired or innate behaviors or diseases induced by environmental factors [94–96, 33, 97]. Understanding the mechanisms dissociating drugs of abuse from natural reinforcers is a prerequisite for the design of selective therapeutic tools to treat compulsive behaviors or addictions in the fascinating field of neuroepigenetics.