Dopaminergic cells mainly located in the midbrain differentially innervate the dorsal striatum (nigrostriatal pathway), the nucleus accumbens (mesolimbic pathway), and prefrontal regions (mesocortical pathway) which we will focus on in the present review (for recent reviews on the anatomical organization of dopamine pathways see Björklund and Dunnett, 2007; Lammel et al., 2014; Morales and Margolis, 2017). The mesocortical pathway originates from the medial part of the ventral tegmental area (VTA) and is comprised of dopamine neurons (30–40%) but also an important fraction of GABAergic and glutamate neurons (Morales and Margolis, 2017). In primates, dopamine neurons innervate prefrontal and sensorimotor cortical regions whereas in rodents this is limited to the mPFC, the agranular insular and the perirhinal cortices (Berger et al., 1991). Dopamine innervation is stronger in ventral parts of the mPFC (prelimbic cortex or Brodmann area 32 / infralimbic cortex or Brodmann area 25; Laubach et al., 2018), and is especially observed in deep cortical layers (layers V–VI; Van Eden et al., 1987; Heidbreder and Groenewegen, 2003). In contrast to the mPFC, rodent orbitofrontal regions are weakly innervated by dopamine neurons. Dopamine neurons rarely send collateral projections to other regions, meaning that mesocortical dopamine neurons represent a distinct population compared to mesolimbic neurons (Fallon, 1988; Beier et al., 2015). Dopamine neurons display two characteristic activity patterns: (1) a “tonic” low frequency mode (2–5 Hz) maintaining basal levels of dopamine in downstream targets and (2) a “phasic” mode with short high frequency (10–20 Hz) bursts transiently increasing dopamine release (Grace and Bunney, 1984a,b; Floresco et al., 2003; Lapish et al., 2007; Rice et al., 2011). This burst firing is especially observed in response to relevant environmental stimuli (Schultz, 2007; Bromberg-Martin et al., 2010). The exact functions of these different modes of dopamine signaling are still debated but through its actions at downstream targets and within local microcircuits, dopamine guides, invigorates and updates complex behaviors (Niv, 2007; Schultz, 2007; Bromberg-Martin et al., 2010; Salamone and Correa, 2012; Berke, 2018).

Mesocortical dopamine neurons respond to both appetitive (Hernandez and Hoebel, 1990; Ahn and Phillips, 1999; Bassareo et al., 2002; St Onge et al., 2012; Ellwood et al., 2017) and aversive stimuli (Abercrombie et al., 1989; Mantz et al., 1989; Vander Weele et al., 2018). Moreover, they have unique properties compared to other dopamine projections including low levels of dopamine transporter DAT and D2 autoreceptors associated with higher basal frequency (Sesack et al., 1998; Lammel et al., 2008). These properties result in different kinetics of dopamine signaling compared to subcortical regions (Garris and Wightman, 1994; Cass and Gerhardt, 1995; Vander Weele et al., 2018). Prefrontal neurons also directly regulate other dopamine circuits by exciting VTA mesocortical and mesolimbic neurons (Carr and Sesack, 2000; Heidbreder and Groenewegen, 2003; Gabbott et al., 2005; Beier et al., 2015) and by modulating dopamine release within subcortical regions including the nucleus accumbens (Taber et al., 1995; Karreman and Moghaddam, 1996; Rice et al., 2011; Nolan et al., 2020).

With its unique characteristics, the mesocortical dopamine pathway plays a key role in prefrontal function (Fuster, 2001; Miller and Cohen, 2001). Current theories postulate a “dual-state” function of prefrontal dopamine signaling. Tonic dopamine levels may act to maintain goal representation and adapted behavioral strategies by stabilizing the activity of specific prefrontal neuronal ensembles. Phasic dopamine release occurs in response to relevant stimuli and would gate new afferent signals from cortical and subcortical regions to update goal representations and behavioral strategies (Montague et al., 2004; Seamans and Yang, 2004; Daw et al., 2006; Durstewitz and Seamans, 2008; Ellwood et al., 2017).

In rodents, the first mesencephalic dopamine neurons appear during embryonic life (E12) and immediately start to extend projections to the developing telencephalic regions (Riddle and Pollock, 2003; Van den Heuvel and Pasterkamp, 2008; Islam et al., 2021). Dopamine fibers are detected in the striatum and the nucleus accumbens at E15-E17 (Voorn et al., 1988). The nigrostriatal and mesolimbic dopamine pathways end their development during the first postnatal week (Kalsbeek et al., 1988; Voorn et al., 1988). Dopamine fibers appear in the mPFC shortly after from E19 and continue to develop during postnatal life (Van Eden et al., 1987; Kalsbeek et al., 1988). In contrast to other neuromodulatory systems in the mPFC, such as noradrenaline or serotonin (Levitt and Moore, 1979; Lidov et al., 1980; Lambe et al., 2000; Naneix et al., 2012), the mesocortical dopamine pathway has a uniquely protracted maturation (Figure 1B). The density of dopamine fibers continues to increase until adulthood in rodents and primates in all mPFC subregions (Kalsbeek et al., 1988; Rosenberg and Lewis, 1994; Benes et al., 1996; Naneix et al., 2012; Hoops and Flores, 2017; Willing et al., 2017; Reynolds and Flores, 2021). This late development parallels increased basal dopamine levels in the mPFC during adolescence and decreased levels of dopamine metabolites (Nomura et al., 1976; Naneix et al., 2012) suggesting a broad increase in dopamine availability and action in prefrontal regions from childhood to adulthood. This is associated with an increase in the proportion of prefrontal cells, especially GABA neurons, in contact with dopamine fibers but also in the absolute number of dopamine varicosities per neuron (Benes et al., 1996), suggesting an increased dopaminergic modulation of these cells.

The continuous growth of dopamine fibers in prefrontal regions likely involves a combination of continued long-range axon growth from the VTA and increased sprouting from dopamine axons already present in the mPFC. There is now considerable evidence that these processes are under the control of several extracellular guidance cues including ephrins, semaphorins, netrins or slits during embryonic and postnatal stages (Van den Heuvel and Pasterkamp, 2008; Brignani and Pasterkamp, 2017; Hoops and Flores, 2017; Islam et al., 2021). The development of the mesocortical dopamine pathway during adolescence seems to be especially dependent of the Netrin-1/DCC (Hoops and Flores, 2017). Dopamine neurons express high levels of the Netrin-1 receptor during embryonic and early post-natal life (pre-weaning) which decrease toward adulthood (Manitt et al., 2010). A proportion of dopamine axons innervating the mPFC at adulthood are still growing through the nucleus accumbens during adolescence (Reynolds et al., 2015, 2018; Reynolds and Flores, 2021). This process is under the control of the Netrin-1 receptor and DCC. Reduced DCC expression in dopamine neurons increases dopamine innervation and dopamine levels in the mPFC (Manitt et al., 2011, 2013; Reynolds et al., 2018). These results suggest that high DCC-expressing dopamine neurons are attracted by brain regions expressing low levels of Netrin-1 such as the nucleus accumbens thus forming the mesolimbic pathway early in development (Manitt et al., 2011). In contrast, prefrontal neurons express high levels of Netrin-1 which attract a specific population of dopamine neurons expressing low levels of DCC during adolescence (Flores et al., 2005; Grant et al., 2007, 2009; Manitt et al., 2011; Reynolds et al., 2018). In addition to guidance cues the refinement of the dopamine system during adolescence may be under the influence of glial cells which contribute to synaptic development and pruning (for review see Béchade et al., 2013; Kettenmann et al., 2013; Brenhouse and Schwarz, 2016). Recent findings show that the downregulation of dopamine receptors in the nucleus accumbens during early adolescence is dependent of microglial activation, especially in males (Kopec et al., 2018). A similar mechanism within the mPFC remains to be demonstrated but microglia cells are already known to regulate the pruning of mPFC glutamatergic synapses during adolescence (Mallya et al., 2019; Blagburn-Blanco et al., 2022).

In addition to the growth of dopamine fibers the activity of dopamine cells also undergoes important changes during adolescence. In vivo and ex vivo recordings have revealed that VTA dopamine neurons in rats have higher tonic firing rates during adolescence compared to adulthood, peaking during late adolescence (PND40-50). In addition, although adolescent rats have a similar proportion of phasic activity episodes, these are consistently longer compared to those of adults (McCutcheon and Marinelli, 2009; McCutcheon et al., 2012; Marinelli and McCutcheon, 2014; but see Kim et al., 2016; McCane et al., 2021). It is currently unknown whether these age-related changes are specific to different defined neuronal populations or pathways. However, in vivo study in mice has demonstrated that the increase in dopamine neuron activity during adolescence may participate specifically to the development of the mesocortical pathway as the phasic activation of dopamine neurons in adolescents, but not in adults, promotes the formation of new synaptic boutons on mesocortical neurons (Mastwal et al., 2014). This is consistent with the role of dopamine in synaptogenesis also demonstrated in vitro (Parish et al., 2001; Fasano et al., 2008, 2010).

At the postsynaptic level, both PFC pyramidal glutamate neurons and GABA interneurons express dopamine D1-like (D1, D5) and D2-like (D2, D3, D4) receptors (Berger et al., 1991; Gaspar et al., 1995; Le Moine and Gaspar, 1998). Dopamine receptor expression during adolescence seems to present a complex reorganization in the cortex. However, data exploring this has produced somewhat conflicting and inconclusive results. For example, autoradiography studies in rats have shown either a steady increase in the expression of D1- and D2-like receptors (Tarazi and Baldessarini, 2000) or a binding peak at weaning followed by a decrease afterwards (Leslie et al., 1991). However immunohistochemistry (Andersen et al., 2000; Brenhouse et al., 2008; Brenhouse and Andersen, 2011) or qPCR studies (Naneix et al., 2012) have shown that the expression of dopamine receptors peaks during mid-adolescence, suggesting a short window of overexpression followed by pruning of dopamine synapses during late adolescence/early adulthood which may be related to the increased activity of dopamine neurons during the same period (McCutcheon and Marinelli, 2009; McCutcheon et al., 2012; Figure 1B). Recent advances in single cell RNA sequencing may clarify this picture by potentially showing different patterns of maturation within mPFC neuronal populations (Bhattacherjee et al., 2019; Tiklová et al., 2019).

Together with the dopamine system, the endocannabinoid (eCB) system is another central neuromodulator of executive functions (Egerton et al., 2006; Pattij et al., 2008) in adults and shares a similar late maturation during adolescence with the dopamine system (Caballero et al., 2016). In the following sections we will discuss how endocannabinoids and their endogenous receptors change across adolescence and how this influences the maturation of the PFC.

The two main endocannabinoids are 2-arachidonoylglycerol (2-AG) and N-arachidonoylethanolamine (AEA or anandamide; Mechoulam et al., 1995; Sugiura et al., 1995). They both bind to two key G-protein coupled receptors (GPCRs): cannabinoid receptor 1 (CB1)—expressed throughout the central nervous system (Herkenham et al., 1990; Mechoulam et al., 1995)—and cannabinoid receptor 2 (CB2) expressed predominantly in immune cells but also found in some neurons and glia (Atwood and Mackie, 2010; Zhang et al., 2014). We will focus here on the CB1 receptor due to its ubiquitous expression throughout the brain including the prefrontal cortex (Herkenham et al., 1990; Marsicano and Lutz, 1999) and its role in PFC development.

CB1 receptors are coupled to inhibitory G-proteins (G_i/o proteins) and their activation reduces cAMP levels, decreasing cAMP-dependent protein kinase activity (Howlett and Mukhopadhyay, 2000; Howlett, 2002). This induces the activation of A-type potassium channels, the inhibition of voltage-gated calcium channels and the disruption of neurotransmitter vesicles (Lovinger, 2008), which decreases the probability of neurotransmitter release. Unlike classical modulatory neurotransmitters which are stored in vesicles, endocannabinoids are synthesized and released de novo in times of sustained neuronal activity (Howlett and Mukhopadhyay, 2000; Freund et al., 2003) acting on presynaptic terminals (Wilson and Nicoll, 2001; Kreitzer and Regehr, 2002). This retrograde action allows neurons to filter and select afferent inputs via negative feedback. The principle psychoactive ingredient in cannabis, Δ9- tetrahydrocannabinol (THC), is a potent CB1 agonist (Gaoni and Mechoulam, 1964; Huestis et al., 2001).

Both 2-AG and AEA are derived from poly-unsaturated fatty acids but have distinct synthesis and metabolic pathways (Ueda et al., 2010; Ueda and Tsuboi, 2012). 2-AG is a major product of the phospholipase Cβ - diacylglycerol lipase pathway synthesized post-synaptically following sustained depolarization and by the subsequent increased calcium influx following activation of voltage-gated Ca²⁺ channels (Kano et al., 2009). 2-AG is metabolized by monoacylglycerol lipase (MAGL) localized mainly with pre-synaptic CB1 receptors (Häring et al., 2012). AEA synthesis on the other hand is a relatively minor biproduct of fatty acylethanolamine (N-acylethanolamines) species production triggered by increased intracellular calcium or cAMP levels in postsynaptic neurons (Bara et al., 2021). AEA is metabolized by fatty acid amide hydrolase (FAAH) localized post-synaptically to provide feedback inhibition (Häring et al., 2012).

Beyond their different synthetic and metabolic pathways, 2-AG and AEA also differ in terms of their pharmacological and functional actions via CB1 receptors. Pharmacologically 2-AG is a full agonist of the CB1 receptor, whereas AEA is a partial agonist (Sugiura et al., 1995, 2006). Despite binding to CB1 receptors with high affinity, AEA induces relatively poor intracellular signal transduction (Hillard et al., 1995). The slower time course of AEA's action makes it a good candidate for a tonic signal (Hill and Tasker, 2012) and thus AEA is implicated in long-term plasticity (Mackie, 2006). 2-AG on the other hand has a relatively low binding affinity for CB1 receptors but produces a much more robust intracellular response (Hillard et al., 1995). These dynamics combined with 2-AG release following sustained depolarization show a more characteristically phasic signal seen in shorter, activity-dependent synaptic plasticity (Lu and Mackie, 2016). In addition, 2-AG levels are generally several orders of magnitude higher than AEA levels in rodent brain tissue (Bisogno et al., 1999; Ueda and Tsuboi, 2012) which may be related to their different synthesis pathways.

Whilst we focus this review on the function and dysfunction of the eCB system in shaping the PFC through actions at CB1 receptors, it is important to consider that endocannabinoids also interact with other receptors and systems. Whilst predominantly expressed in peripheral tissues CB2 receptors are expressed in some microglia within the brain and their activation may have important effects we do not fully understand (Cabral et al., 2008). In addition, eCBs can activate or facilitate the activation of other receptors. For example, AEA is a TRPV1 receptor agonist and 2-AG can potentiate GABA-A receptors independently of CB1 receptor effects (Piscitelli and Di Marzo, 2012). In fact, eCB signaling has been shown to engage as many as 12 different receptors and involve many biosynthetic and degradative enzymes, transporters and other proteins diversifying the ways eCB signaling may impact local brain circuits (for a comprehensive discussion of the complex interactions between endocannabinoids and other receptors, transporters, and enzymes see Maccarrone, 2020).

The CB1 receptor is the most abundant GPCR in the brain and the eCB system is uniquely positioned to have large scale influence on developing brain circuits. Our current understanding of the functioning and role the eCB system is still relatively new but we know this system is fundamentally important in all phases of brain development especially in controlling long-range axonal guidance and patterning of synaptic contacts (Berghuis et al., 2007; Harkany et al., 2007, 2008; Maccarrone et al., 2014; Bara et al., 2021). In cortical regions CB1 receptors are expressed at higher levels in GABAergic interneurons than in glutamatergic neurons with a higher expression in the PFC compared to motor and sensory cortices (Marsicano and Lutz, 1999). PFC CB1 expression in rodents peaks to its maximal density in early adolescence (Meyer et al., 2018), and then declines into adulthood (Figure 1B; Berrendero et al., 1999; Ellgren et al., 2008; Heng et al., 2011a; Peters et al., 2021c).

During adolescence 2AG and AEA levels also fluctuate. Within the PFC, AEA steadily increases from early to late adolescence, with concentrations reaching three times higher levels in late versus early adolescence before reaching stable levels into adulthood (Ellgren et al., 2008; Tirado-Muñoz et al., 2020). PFC 2-AG levels markedly decrease from early to late adolescence before stabilizing in adulthood (Ellgren et al., 2008; Rubino et al., 2015; Bara et al., 2021). These changes are accompanied by complementary changes in the activity of FAAH and MAGL (Long et al., 2012). Increased FAAH levels in adolescence may reflect in particular the tighter regulation of AEA during this key developmental period (Long et al., 2012).

Changes in the anatomical organization of the endocannabinoid system during adolescence are associated with important functional changes in synaptic plasticity which may also play an important role in long-range axonal pathfinding and patterning of synaptic contacts, both essential processes for prefrontal development (Berghuis et al., 2007; Maccarrone et al., 2014).

Both dopamine and endocannabinoids modulate the activity of pyramidal glutamate neurons and prefrontal glutamatergic plasticity. In adults, dopamine's action on prefrontal circuits is highly dependent of the excitability state of pyramidal neurons and interneurons. Low dopamine levels maintain prefrontal networks, mainly through actions on GABAergic interneurons and D2 receptors. Increases in dopamine levels during phasic activity recruits D1 receptors (Montague et al., 2004; Seamans and Yang, 2004; Durstewitz and Seamans, 2008) required for prefrontal functions such as executive functions (Floresco and Magyar, 2006; Floresco and Jentsch, 2011) and appetitive learning (Baldwin et al., 2002). Dopamine D1 signaling interacts with glutamatergic N-methyl-D-aspartate (NMDA) receptor signaling through D1-dependent mechanisms to induce long-lasting plateau depolarizations (up-state) of pyramidal cells (Tseng and O'Donnell, 2005). Similarly, CB1-dependent eCB signaling regulates prefrontal glutamatergic activity through depolarization-induced suppression of excitation process (Auclair et al., 2000; Kreitzer and Regehr, 2002; Fortin and Levine, 2007; Lovinger, 2008) and long-term depression in deep cortical layers (V–VI; Lafourcade et al., 2007).

As prefrontal D1 expression levels change during adolescence (Leslie et al., 1991; Tarazi and Baldessarini, 2000; Brenhouse et al., 2008; Naneix et al., 2012), ex vivo studies also showed that D1-NMDA interactions emerge during adolescence (Tseng and O'Donnell, 2005; Heng et al., 2011b; Flores-Barrera et al., 2014; Caballero et al., 2016). Despite important anatomical changes on both glutamatergic projections from the amygdala and the hippocampus to the PFC (Cunningham et al., 2002; Johnson et al., 2016; Calabro et al., 2020), only the latter presents a delayed functional maturation during adolescence (Caballero et al., 2014b; Flores-Barrera et al., 2014). Long-term potentiation (LTP) of hippocampal-prefrontal synapses on pyramidal only appear after PND50 in relationship with changes in the expression of GluN2B subunit composition of NMDA receptors. This process parallels the decrease of CB1-related inhibition of glutamatergic transmission (Heng et al., 2011a; Caballero et al., 2016) which may facilitate excitatory transmission from the PFC to subcortical circuits. Interestingly, anatomical studies have shown important changes in D1 expression levels in prefronto-accumbens neurons which may underlie increased incentive salience in response to environmental cues seen in adolescents (Brenhouse et al., 2008; Sonntag et al., 2014).

As for glutamatergic signaling, prefrontal GABAergic transmission is also still undergoing important developmental changes during adolescence. Previous studies demonstrated a progressive increase in the expression levels of calcium binding proteins parvalbumin and a parallel decrease in the expression of calretinin (Caballero et al., 2014a), increasing the proportion of fast-spiking interneurons. This process is dependent on NMDA glutamatergic transmission onto GABAergic interneurons (Flores-Barrera et al., 2014; Caballero et al., 2016; Bogart and O'Donnell, 2018) which may be related to the continuous development of projections from the hippocampus and the amygdala during this period (Cunningham et al., 2002; Johnson et al., 2016; Calabro et al., 2020).

Before adolescence, dopamine facilitates GABAergic transmission by increasing GABA interneurons excitability only through D1 receptor signaling (Gorelova et al., 2003; Tseng et al., 2006). Intriguingly, during late adolescence (PND50), dopamine increases GABA neurons excitability through both D1 and D2 signaling (Tseng et al., 2006; Tseng and O'Donnell, 2007; Caballero et al., 2016). This process is still poorly understood but may involve changes in the expression of D2 receptors and their signaling through β-arrestins which are able to enhance firing of parvalbumin interneurons in the cortex (Urs et al., 2016).

Less is known about the developmental changes of eCB-dependent regulation of GABAergic transmission. In adults, CB1 signaling suppresses GABA release onto glutamatergic cells (Chiu et al., 2010; Hill et al., 2011) and drives depolarization-induced suppression of inhibition at these synapses (Kiritoshi et al., 2013). Moreover, CB1 receptors also colocalize with dopamine D2 receptors on GABA cells, participating in the establishment of long-term depression of GABA synapses (Chiu et al., 2010). It is currently unknown if the changes in prefrontal CB1 receptor expression during adolescence follow the same pattern in glutamatergic and GABAergic neurons (Berrendero et al., 1999; Ellgren et al., 2008; Meyer et al., 2018; Peters et al., 2021c). As for glutamatergic pyramidal neurons, potential changes in CB1 and dopamine modulation of prefrontal GABAergic transmission may contribute to the establishment of excitation/inhibition balance, increasing the selection of specific inputs, decreasing the signal-to-noise ratio and refining prefrontal processing (Montague et al., 2004; Seamans and Yang, 2004; Durstewitz and Seamans, 2008; O'Donnell, 2011; Caballero et al., 2016).

Changes in the prefrontal endocannabinoid system during adolescence may also impact the dopamine system and dopamine signaling in prefrontal subcortical targets, modulating dopamine-related behaviors such as reward prediction error, learning, motivation and event salience (Floresco and Magyar, 2006; Berridge, 2007; Schultz, 2007; Wickens et al., 2007; Bromberg-Martin et al., 2010; Berke, 2018).

In the VTA, dopamine neurons synthesize and release eCBs but do not express CB1 receptors (Julian et al., 2003), VTA eCBs act on local GABA interneurons as well as on GABA and glutamate terminals (Szabo et al., 2002; Melis et al., 2004; Riegel and Lupica, 2004; Wang et al., 2015). This results in the indirect disinhibition of dopamine neurons and the increase of dopamine release in terminal regions including the nucleus accumbens (Cheer et al., 2000, 2007; Oleson et al., 2012; Zlebnik and Cheer, 2016; Covey et al., 2017; Peters et al., 2021a,b) and the prefrontal cortex (Pistis et al., 2002). As in cortical regions, CB1 receptor expression in the VTA increases during early adolescence (PND 40) before decreasing to reach adult levels (de Fonseca et al., 1993) but the neuronal specificity of such changes remains unknown. This may be related to changes in CB1 levels on prefrontal glutamatergic terminals in the VTA controlling the activity of dopamine neurons (Carr and Sesack, 2000; Heidbreder and Groenewegen, 2003; Gabbott et al., 2005; Beier et al., 2015). These complex changes may represent an important mechanism underlying developmental changes in VTA responses to rewarding and aversive environmental stimuli (Kim et al., 2016; McCane et al., 2021) and dopamine-related behaviors (Spear, 2000).

Prefrontal glutamatergic neurons also strongly project to striatal regions (Vertes, 2004; Gabbott et al., 2005) to modulate action control processes (Alexander et al., 1986; Haber, 2003; Voorn et al., 2004). In the nucleus accumbens, cortical glutamatergic terminals control the activity of GABAergic medium spiny neurons (Surmeier et al., 2011) but also modulate local dopamine release through indirect actions on cholinergic interneurons (Cachope et al., 2012; Mateo et al., 2017). In adults, this cortical control of mesolimbic dopamine signal is dependent on CB1 receptors expressed on glutamatergic terminals. Combined with changes in the expression of dopamine D1 receptors by prefronto-accumbens neurons (Brenhouse et al., 2008; Sonntag et al., 2014), changes in CB1 expression in the same neurons may strongly change dopamine release in subcortical regions between adolescents and adults, impacting the control of motivated behaviors. Accordingly, ex vivo and in vivo studies have reported age-related changes in dopamine release in different striatal regions (Matthews et al., 2013; Pitts et al., 2020) and in striatal neuronal responses during instrumental and Pavlovian tasks (Galiñanes et al., 2009; Sturman and Moghaddam, 2012) which may also underlie enhanced reward sensitivity and executive functions deficits during adolescence.

Cannabis is one of the most commonly used drugs of abuse, especially amongst adolescents (Miech et al., 2020). THC, the principal psychoactive constituent of cannabis, is a potent CB1 agonist (Gaoni and Mechoulam, 1964; Huestis et al., 2001). We do not fully understand how adolescent use in particular impacts the developing brain, however there is emerging evidence that altering the eCB system through such external insults may have important consequences especially at the prefrontal level (Schneider, 2008; Bara et al., 2021). Accordingly, numerous studies in humans and animal models report associations between adolescent cannabis exposure and increased risk of psychiatric conditions which a emerge in adolescence including psychosis, mood disorders and substance use disorders (Laviolette and Grace, 2006; Rubino et al., 2012; Renard et al., 2016).

Chronic CB1 agonist treatment in adolescent animals causes alterations in multiple PFC dependent behaviors, including anxiety (O'Shea et al., 2004), short- and long-term spatial and working memory (Rubino et al., 2009; Renard et al., 2013), motivational and emotional processes (Schneider and Koch, 2003), social behaviors (Schneider et al., 2008), and drug-seeking responses (Ellgren et al., 2007; Higuera-Matas et al., 2008). These alterations are specific to adolescence and are not observed when agonists are administered in adulthood. Importantly, the functioning and vulnerability of the eCB system presents important sex-differences that we do not discuss here (see Ginder et al., 2022 for a complete review of this topic) but which may play a key role in normal and pathological development during adolescence and puberty.

These behavioral phenotypes are associated with profound anatomical and functional alterations of prefrontal circuits (Rubino et al., 2015). These include alterations in mPFC CB1 receptor density and impaired eCB-mediated long-term depression (LTD) of prefrontal excitatory transmission (Rubino et al., 2015; Cuccurazzu et al., 2018). These changes are associated with an increased expression of GluA1 and GluN2B subunits in AMPA and NMDA receptors respectively (Rubino et al., 2015), altering the normal development of PFC glutamatergic transmission (Lovelace et al., 2015; Rubino et al., 2015). Interestingly, these effects are reversed by enhancing AEA levels at adulthood (Cuccurazzu et al., 2018), highlighting the importance of eCB signaling in normative prefrontal development and function.

In addition to its action on prefrontal glutamatergic transmission, there is emerging evidence that prefrontal GABA interneurons may be especially vulnerable to adolescent THC (or synthetic agonist) exposure (Cass et al., 2014; Renard et al., 2018; Peters et al., 2021c). Adolescent, but not adult, exposure to the synthetic CB1 agonist WIN55,212-2 decreases prefrontal GABA signaling resulting in dysregulation of cortical oscillatory activity (Cass et al., 2014; Renard et al., 2017). Conversely, activating prefrontal GABA-A receptors restores behavioral and dopaminergic abnormalities observed following adolescent THC exposure (Renard et al., 2017). Reduced prefrontal GABAergic transmission is associated with hyperdopaminergic state and alterations of prefrontal cortex-related behaviors (Yee et al., 2005; Enomoto et al., 2011), mimicking key behavioral alterations observed for instance in schizophrenia (Carlsson et al., 1997; O'Carroll, 2000; Nakazawa et al., 2012).

Adolescent cannabis exposure also directly impacts the activity of dopamine neurons in the VTA with potential alterations of eCB-mediated inhibition of VTA GABA neurons and glutamate terminals, resulting in a decrease of eCB-dependent disinhibition of dopamine neuron firing and dopamine release in the nucleus accumbens (Scherma et al., 2016), suggesting a broader impact of cannabis during adolescence on dopamine functions. The specific impact of adolescent THC exposure on mesocortical dopamine neurons or on dopamine release and signaling within the prefrontal cortex still needs to be investigated. Similarly, other drugs of abuse seem to have a critical influence on corticolimbic circuits and brain functions if they are used during adolescence compared to adulthood (for a complete view on the impact of drugs on adolescent brains see these recent books edited by Bell and Rahman, 2021a,b).

In parallel to cannabis use by teenagers, in recent decades there has been a dramatic rise of obesity amongst children and adolescents, due to the increased availability of energy-dense and highly palatable foods and drinks (Crews et al., 2007; Wang et al., 2008; Ogden et al., 2016). Moreover, animal studies have shown that adolescents are more sensitive to palatable foods such as high-fat and high-sugar foods (Wilmouth and Spear, 2009; Friemel et al., 2010; Marshall et al., 2017, 2020) which may increase food-seeking behaviors, lead to overconsumption, unbalanced dietary habits and obesity.

The dopamine system is well-known to respond directly to food rewards and modulate food-seeking behaviors by acting in prefrontal and subcortical regions (Hajnal and Norgren, 2001; Bassareo et al., 2002; Hajnal et al., 2004; Berridge, 2007; Avena et al., 2008; de Araujo et al., 2010; Kenny, 2011; McCutcheon, 2015; Alhadeff et al., 2019; Zimmerman and Knight, 2020; Grove et al., 2022). Similarly, eCBs are tightly linked to feeding as they are derived from poly-unsaturated fatty acids and are able to modulate homeostatic and non-homeostatic/hedonic feeding (Mahler et al., 2007; Maccarrone et al., 2010; Lau et al., 2017; Coccurello and Maccarrone, 2018). In humans and rodents, obesity in adults is associated with alteration of both dopamine and eCB circuits (Berthoud and Morrison, 2008; South and Huang, 2008; Volkow et al., 2011; Bello et al., 2012) but also with dysfunction of prefrontal functions (Reichelt, 2016; Lowe et al., 2019). In adolescents, the delayed development of both dopamine and eCB systems may simultaneously support this adolescent sensitivity to overnutrition, but also provides a vulnerability window for the long-term effects of unbalanced dietary habits.

The differential time course in the development of cortical and subcortical dopamine pathways may be responsible of increased reactivity to palatable foods during adolescence, leading to overconsumption and increased food-seeking responses (Wilmouth and Spear, 2009; Friemel et al., 2010; Marshall et al., 2017, 2020). As eCB signaling controls the activity of mesocorticolimbic circuits, increased levels of CB1 receptors during adolescence may also participate to this process by decreasing VTA GABAergic activity and decreasing cortical control on VTA and striatal neurons, all together leading to increased dopamine activity (Lau et al., 2017; Sallam and Borgland, 2021) and food-seeking behaviors (Mateo et al., 2017).

Several studies have also investigated the impact of adolescent exposure to palatable food on related brain functions to understand how vulnerable these circuits may be to natural rewards. Adolescent overnutrition causes complex behavioral alterations which seem to be highly dependent on diet parameters including but not limited the constitution of the diet (e.g. high fat, high sugar, cafeteria diet), the duration of exposure (age at start, length of time exposed, withdrawal period between diet and testing) or other exposure parameters (e.g., continuous or limited access, combined access with control diet or not). We focus here on reward- and prefrontal-related functions but there is also an important literature on the effect of these dietary habits on learning, memory and emotional behaviors (for recent reviews on the topic see Kendig, 2014; Tsan et al., 2021). Taken together, several studies indicate that exposure to highly palatable foods during adolescence, but not at adulthood, strongly impacts reward-related functions including the preference for palatable foods (Vendruscolo et al., 2010b; Carlin et al., 2016; Naneix et al., 2016; Rabasa et al., 2016; Gueye et al., 2018), conditioned place preference for food (Privitera et al., 2011), motivation for food rewards (Frazier et al., 2008; Vendruscolo et al., 2010a; Reichelt et al., 2016; Tantot et al., 2017; Wong et al., 2017; Naneix et al., 2018; Ducrocq et al., 2019), hedonic processing (Naneix et al., 2016; but see Steele et al., 2019), and sensitivity to drugs of abuse (Blanco-Gandía et al., 2017; Naneix et al., 2017). More importantly, several adolescent dietary habits also seem to impact prefrontal-related functions including control of goal-directed behavior (Tantot et al., 2017; but see Kendig et al., 2013), memory extinction (Baker and Reichelt, 2016), and cognitive control (Reichelt et al., 2015; Labouesse et al., 2017; Robertson and Rasmussen, 2017; Steele et al., 2019).

At the neurobiological level, exposure to highly palatable foods during adolescence is associated with broad anatomical and functional alterations of the dopamine system including reduced dopamine clearance in the striatum (Baladi et al., 2015), increased sensitivity of the mesolimbic pathway (Naneix et al., 2017) and changes in the expression of dopamine markers (Teegarden et al., 2009; Carlin et al., 2016; Naneix et al., 2017, 2018) in the prefrontal cortex and the nucleus accumbens. To our knowledge, the specific effects of such adolescent diets on prefrontal dopamine release, transmission and associated plasticity has not been investigated. However, the functioning of prefrontal microcuircuits is strongly impacted by adolescent consumption of palatable foods. As with cannabis exposure, prefrontal GABA signaling seems especially vulnerable, with diet-induced decreased of GABA levels (Sandoval-Salazar et al., 2016) and numbers of parvalbumin interneurons (Reichelt et al., 2015, 2019; Baker and Reichelt, 2016). Moreover, juvenile obesity in mice also impairs prefrontal glutamatergic transmission including NMDA-dependent synaptic plasticity and AMPA postsynaptic currents (Labouesse et al., 2017). Surprisingly, given its role in feeding and appetite, there are no clear data available on the impact of such diet exposure on the eCB system. In adult animals, obesity or the exposure to high-sugar/high-fat foods alters eCB-dependent cortical synaptic plasticity and insulin effects on VTA dopamine neurons to control ingestive behaviors (Labouèbe et al., 2013; Liu and Borgland, 2015; Liu et al., 2016; Lau et al., 2017). As maternal high-fat diet dysregulates the eCB system (Almeida et al., 2022; Urbonaite et al., 2022), we might predict high-fat / high-sugar diet exposure during adolescence would have similar effects. The exact nature of these alterations is not well-understood and the mechanisms which make adolescent PFC circuits particularly vulnerable are yet to be determined. However, altered dopamine and eCB signaling may both contribute to long-lasting diet-induced dysfunctions in executive functions.