Receptor and metabolic insights on the ability of caffeine to prevent alcohol-induced stimulation of mesolimbic dopamine transmission

The consumption of alcohol and caffeine affects the lives of billions of individuals worldwide. Although recent evidence indicates that caffeine impairs the reinforcing properties of alcohol, a characterization of its effects on alcohol-stimulated mesolimbic dopamine (DA) function was lacking. Acting as the pro-drug of salsolinol, alcohol excites DA neurons in the posterior ventral tegmental area (pVTA) and increases DA release in the nucleus accumbens shell (AcbSh). Here we show that caffeine, via antagonistic activity on A2A adenosine receptors (A2AR), prevents alcohol-dependent activation of mesolimbic DA function as assessed, in-vivo, by brain microdialysis of AcbSh DA and, in-vitro, by electrophysiological recordings of pVTA DA neuronal firing. Accordingly, while the A1R antagonist DPCPX fails to prevent the effects of alcohol on DA function, both caffeine and the A2AR antagonist SCH 58261 prevent alcohol-dependent pVTA generation of salsolinol and increase in AcbSh DA in-vivo, as well as alcohol-dependent excitation of pVTA DA neurons in-vitro. However, caffeine also prevents direct salsolinol- and morphine-stimulated DA function, suggesting that it can exert these inhibitory effects also independently from affecting alcohol-induced salsolinol formation or bioavailability. Finally, untargeted metabolomics of the pVTA showcases that caffeine antagonizes alcohol-mediated effects on molecules (e.g. phosphatidylcholines, fatty amides, carnitines) involved in lipid signaling and energy metabolism, which could represent an additional salsolinol-independent mechanism of caffeine in impairing alcohol-mediated stimulation of mesolimbic DA transmission. In conclusion, the outcomes of this study strengthen the potential of caffeine, as well as of A2AR antagonists, for future development of preventive/therapeutic strategies for alcohol use disorder.


Introduction
Caffeine and ethyl alcohol (alcohol) are the two most consumed psychopharmacologically active substances in the world (1,2).The pharmacological consequences of their combined use have been extensively investigated, but different studies have produced con icting data that diverge based on species, strains, dosages, routes, and schedules of administration.In this fragmented scenario, it is di cult to capture a unique pattern of the in uence of caffeine on alcohol effects but, as far as the effects on alcohol consumption are concerned, it seems that caffeine may exert bi-directional in uences depending on several experimental parameters (3)(4)(5)(6)(7)(8)(9).
To further characterize caffeine-alcohol interaction, we focused our research on the potential ability of caffeine to affect the neurophysiological and neurochemical processes underlying the reinforcing properties of alcohol.In particular, we previously showcased that caffeine, at doses borderline for eliciting arousal (10,11) and locomotor activity (7,12), can functionally antagonize alcohol reinforcement by demonstrating that it could prevent the acquisition of alcohol-elicited conditioned place preference and aversion (13), pointing to the dopamine (DA)-dependent underlying associative learning process (14) as the possible mechanism targeted by caffeine to gain this behavioral outcome.This hypothesis was grounded in the observations that alcohol-elicited place conditioning is prevented by DA receptor antagonists (15,16), that caffeine exerts, through an antagonistic action on A 2A adenosine receptors (A 2A R) (17), a direct negative control on neuronal ring of DA cells in the posterior ventral tegmental area (pVTA) (18), and that caffeine administration prior to alcohol also prevents its DAdependent (19) ability to increase the expression of phosphorylated Extracellular signal Regulated Kinase (pERK) in the shell of the nucleus accumbens (AcbSh) (13,20).Notably, increased pERK expression is a DA receptor-dependent marker of activation of mesolimbic DA transmission by alcohol (19,21) and other addictive substances (22,23), but not by caffeine (22,24), as well as a DA-dependent mechanism at the basis of associative learning (14,(25)(26)(27).
The mechanism by which alcohol activates mesolimbic DA transmission has been the subject of intense research for decades, as this pathway is critical in mediating the reinforcing effects of alcohol, as well as other drugs of abuse (28,29).Even in humans, positron emission tomography studies have shown that alcohol induces a release of DA in the ventral striatum (30), and that this fast release of DA is associated with alcohol-induced reinforcing effects and acquisition of conditioned responses (31).In this regard, the metabolic conversion of alcohol into acetaldehyde has been recognized as critically involved (32,33), and this suggestion was further extended by the observation that another molecule, 1-methyl-6,7dihydroxy-1,2,3,4-tetrahydroisoquinoline (salsolinol), obtainable by Pictet-Spengler condensation of acetaldehyde and DA, could be responsible for the reinforcing properties of alcohol and for its addictive potential (34,35).This hypothesis was recently substantiated by multiple robust lines of evidence.The rst line of evidence refers to experiments showing that systemic (36) or local (37) salsolinol administration elicits conditioned place preference, exerts alcohol-like motivational/sensitization effects (38,39) and leads to excessive alcohol intake (39).The second line of evidence refers to an invitro electrophysiological study, in which the ability of alcohol to stimulate the ring rate of DA neurons of the pVTA critically depended on the availability of DA, as well as on the metabolic conversion of alcohol into acetaldehyde (40).Finally, direct evidence was provided also by Bassareo et al. (2021), in which the systemic administration of alcohol resulted in the in-vivo formation of salsolinol in the pVTA which was connected, in a mechanistic-and time-locked manner, to increased DA transmission in the AcbSh via m opioid receptor (41).In addition, this study also demonstrated that the inhibition of brain catalase, the enzyme responsible for alcohol oxidation into acetaldehyde, prevents both the formation of salsolinol in the pVTA and the increase of DA transmission in the AcbSh (41).
Hence, in order to understand the mechanistic in uence of caffeine on DA-mediated alcohol effects (13,20), we veri ed whether caffeine can affect the ability of alcohol, administered at a dose that results in mild behavioral activation (42)(43)(44), to activate DA transmission in the AcbSh, and if this in uence also involves the alcohol-dependent generation and availability of salsolinol in the pVTA (41).Moreover, since caffeine is an A 1 R and A 2A R antagonist (45), we also veri ed if the effects of caffeine could be attributable to an action onto adenosine receptors.To this end, the effects of caffeine, and of the selective A 1 R and A 2A R antagonists, DPCPX and SCH 58261, on alcohol-stimulated DA transmission in the AcbSh and newly formed salsolinol in the pVTA (41) were simultaneously investigated through invivo dual probe brain microdialysis.Additionally, a catalase-mediated synthesis of salsolinol was set up, in-vitro, to verify whether caffeine, similarly to the non-competitive catalase inhibitor 3-amino-1,2,4triazole (3AT), could prevent the formation of salsolinol by directly inhibiting the enzyme.Moreover, to further characterize the mechanism of action of caffeine on alcohol-mediated stimulation of mesolimbic DA neurons, in-vitro patch-clamp recordings from pVTA slices were performed.Additionally, to verify whether caffeine could also show effects unrelated to salsolinol generation, we veri ed its activity on the enhancement of mesolimbic DA transmission mediated by exogenous salsolinol, as well as by another m receptor agonist, morphine.Finally, we also performed region speci c untargeted metabolomics of the pVTA in alcohol-treated rats, with and without caffeine pre-treatment, to detect changes in the biochemical pro les that might also be related to the stimulatory effects of alcohol on mesolimbic DA function.measure and analyze the ring rate and other membrane kinetic parameters of pVTA neurons and the occurrence of HCN-mediated I h currents (see below).The cell-attached con guration was used to monitor the spontaneous and pharmacologically conditioned ring rate of DA neurons.After obtaining a pipette-membrane seal with a GΩ resistance, at least 10 min were allowed before recording to obtain a stable and regular spontaneous ring rate.In addition, the whole-cell con guration was obtained at the end of each recording to determine the presence of I h currents, to con rm the identity of pVTA DA neurons (55).Accordingly, in our experimental conditions, identi ed pVTA DA neurons showed both a robust I h (mean amplitude: -134.In-vitro synthesis of salsolinol

Materials And Methods
The protocol followed to synthesize salsolinol was an adaptation of Akbayeva et al., 2023 (56) to obtain a catalase-mediated oxidation of alcohol into acetaldehyde and the production of salsolinol in presence of DA via Pictet-Spenlger reaction.The blank consisted of bovine catalase (Sigma Aldrich, Italy) at 0.33 mg/mL (666.67-1666.67units/mL) and DA hydrochloride (Sigma Aldrich, CAS No. 62-31-7) at 1.5 mM dissolved in PBS.Triplicates of blank, blank + 0.05 M catalase inhibitor 3AT (Sigma Aldrich, Italy), and blank + 0.05 M caffeine (Sigma Aldrich, Italy) were kept in an agitator at 37°C for 20 min.After that, PBS or 1 mM alcohol in PBS + 0.06 M hydrogen peroxide (Sigma Aldrich, Italy) in PBS were added to the solutions and the samples were placed back in an agitator at 37°C for additional 30 min.The reactions were then quenched with formic acid (FA, nal concentration 1% v/v).The same steps were also followed using a more diluted catalase solution (0.0033 mg/mL or 6.67-16.67units/mL).Samples were centrifuged at 4 °C for 15 min at 14 000 g, and the supernatant was collected and diluted 1:1000 in LC graded H 2 O before untargeted metabolomics analysis.
pVTA harvesting and sample preparation Rats were randomly divided into 4 experimental groups: saline-water, saline-alcohol, caffeine-water, caffeine-alcohol.Subjects received pre-treatment with saline or caffeine i.p. (15 mg/kg).Twenty minutes after pre-treatment, rats were treated with water or alcohol (1 g/kg) i.g. and returned to their home cages.After 30 min from alcohol treatment, rats were decapitated under 5% iso urane deep anesthesia, brains were removed and pVTA from both hemispheres harvested, weighted, and immediately frozen in dry ice.Pre-chilled LC graded 50% MeOH:H 2 O containing 1 µM sulfadimethoxine, as an internal standard, was added to each pVTA sample to obtain a nal 1:20 w/v ratio.One 5 mm stainless steel bead was added to each sample before homogenization at 25 Hz for 5 min (TissueLyser II, Qiagen).Samples were left to incubate for 1 h at 4 °C before centrifugation at 14 000 g for 15 min at 4 °C.In separate Eppendorf's tubes, 900 µL of supernatant was collected and added to 180 µL of formic acid (500 nM).Samples were then centrifuged again for 10 min at 14 000 g and 4 °C.The collected supernatant (1 mL) was then dried overnight in a speed vacuum concentrator.Samples were stored at -80°C and on the day of the untargeted metabolomics experiments were reconstituted in 200 µL of 50% acetonitrile (ACN) and vortexed.

Untargeted metabolomics and in-vitro synthesis of salsolinol experiments
For the metabolomics experiments, a Vanquish ultra-high performance liquid chromatography (UHPLC) system coupled to a Q Exactive quadrupole orbitrap mass spectrometer (Thermo Fisher Scienti c, Waltham, MA, USA) was used.Samples ( Before ordination, data was robust center log ratio transformed using v 2.6.Batch effects were corrected using the removeBatchEffect function of v 3.54.Supervised multivariate partial least square discriminant analysis (PLS-DA) models were generated using and performance (classi cation error rate) was assessed using a 4-folds cross validation.

Statistical analysis
Statistical analysis was carried out either via Statistica 8.0 (StatsSoft Inc., Tulsa, OK, USA) or PRISM, GraphPad 8 Software (San Diego, CA, USA) with signi cance set for all the experiments at p < 0.05.For microdialysis experiments, basal dialysate salsolinol and DA were calculated as the average ± SEM of the last three consecutive samples differing by no more than 10%, collected during the time preceding each treatment.Changes in dialysate salsolinol and DA were expressed as fmol/10 μl of dialysate and were analyzed by two-or three-way Analysis of Variance (ANOVA) with repeated measures over time.For electrophysiology experiments, all data are reported as mean ± SEM.Before ANOVA analyses, the normal distribution of data was evaluated by skewness and kurtosis, and homoscedasticity via the Bartlett test.Comparisons among experimental conditions were obtained using at least n=4 rats/group and was performed by one-way ANOVA followed by Tukey's post-hoc test.Detailed statistical analysis for microdialysis and electrophysiology experiments is available in Supplementary Tables 1 and 2.

Results
Effects of caffeine and adenosine receptor antagonists on alcohol-elicited pVTA salsolinol formation and AcbSh DA increase in-vivo.
Simultaneous dual probe in-vivo brain microdialysis was used to verify the effects of caffeine and adenosine receptors antagonists on alcohol-dependent salsolinol generation, in the pVTA, and DA transmission, in the AcbSh (Figure 1A).Alcohol elicited the formation of salsolinol in the pVTA and stimulated DA transmission in the AcbSh and caffeine signi cantly prevented these effects (Figure 1B and 1C, Three-way ANOVA followed by Tukey's post hoc test).No production of salsolinol was observed after alcohol administration also when DPCPX or SCH 58261 were used as pre-treatment (Figure 1D, Three-way ANOVA, p > 0.05).In addition, caffeine and SCH 58261, but not DPCPX, prevented the increase of DA after alcohol administration (Figure 1B-D, Three-way ANOVA followed by Tukey's post hoc test).Invivo brain microdialysis was also used to verify the effect of intra-pVTA caffeine on alcoholdependent salsolinol generation in the pVTA and DA transmission in the AcbSh.Salsolinol and DA concentrations during reverse dialysis application of caffeine in the pVTA failed to reveal any effect of alcohol (Figure 1E, Two-way ANOVA p > 0.05).Given that local application of caffeine by reverse dialysis prevented both systemic alcohol-dependent salsolinol formation in the pVTA and DA increase in the mixOmics vegan lim ma mixOmics AcbSh, these results suggest that the systemic effects of caffeine might be mediated by a direct action on the pVTA.
Additionally, an in-vitro synthesis of salsolinol was set up to verify whether caffeine could prevent salsolinol formation acting as a catalase inhibitor.Speci cally, the ability of the catalase-inhibitor 3AT and caffeine to prevent catalase-mediated alcohol oxidation to acetaldehyde and, consequently, its condensation with DA to generate salsolinol were investigated (Supplementary Figure 1A, B).As expected, salsolinol formation was catalase-dependent, as lowering the units/mL of the enzyme also reduced salsolinol abundance (Supplementary Figure 1A, One-way ANOVA followed by Tukey's post hoc test).Caffeine, differently from 3AT, did not prevent the formation of salsolinol (Supplementary Figure 1A, One-way ANOVA followed by Tukey's post hoc test), ruling out the possibility of a direct inhibitory activity on catalase.
In-vivo brain microdialysis was also used to verify the effects of caffeine on salsolinol bioavailability in the pVTA and on salsolinol-induced DA transmission in the AcbSh.The systemic administration of caffeine failed to signi cantly affect pVTA salsolinol concentrations during pVTA perfusion with exogenous salsolinol (Figure 1F, Two-way ANOVA p > 0.05), pointing out that caffeine does not affect salsolinol bioavailability.However, caffeine pre-treatment signi cantly reduced the increase of AcbSh DA induced by reverse dialysis of exogenous salsolinol in the ipsilateral pVTA (Figure 1F, Two-way ANOVA followed by Tukey's post hoc test).These last results suggest that other mechanisms of action in the pVTA, in addition to the prevention of alcohol-induced salsolinol formation, should be envisioned to explain caffeine inhibitory effects on alcohol-induced increase of mesolimbic DA transmission.

Effects of caffeine on the excitation of pVTA DA neurons induced by alcohol, morphine, and salsolinol: in-vitro electrophysiological experiments
To further characterize the effects of caffeine on alcohol-induced stimulation of mesolimbic DA signaling, in-vitro patch-clamp recordings from pVTA slices were performed.As expected from previous reports (47)(48)(49)65), acute perfusion of 60 mM alcohol signi cantly increased (40.1 ± 4.4 %) the ring rate of pVTA DA neurons, an effect that was promptly reversed by drug washout (Figure 2A, B, F, one-way ANOVA followed by Tukey's post hoc test).In contrast, 5 min of acute perfusion with 10 µM caffeine signi cantly decreased (-34.3 ± 7.7 %) DA neuron ring rate, (Figure 2C, D, F, one-way ANOVA followed by Tukey's post hoc test).The modulatory effect of alcohol on the ring rate of pVTA DA neurons was completely suppressed in the presence of 10 µM caffeine (Figure 2E, F, one-way ANOVA followed by Tukey's post hoc test).
In-vitro patch-clamp recordings from pVTA slices were also performed to test whether this effect of caffeine was mediated by an antagonistic component on A 1 R or A 2A R. Independent bath perfusion with either antagonist decreased the ring rate of pVTA DA neurons (Figure 2G-I, one-way ANOVA followed by Tukey's post hoc test).Interestingly, while the effect of alcohol was completely blocked by the A 2A R antagonist SCH 58261 (Figure 2G, I, one-way ANOVA followed by Tukey's post hoc test), it was indistinguishable from its effect when tested alone in presence of DPCPX (Figure 2H, I, one-way ANOVA followed by Tukey's post hoc test), suggesting that the ability of caffeine to suppress the modulatory effect of alcohol on DA ring rate is mediated by an action on A 2A R, but not A 1 R.
Finally, in-vitro patch-clamp recordings were performed to investigate whether caffeine could interfere with the positive modulatory effect of salsolinol or morphine on DA neuron ring rate.Accordingly, the acute perfusion of 10 nM salsolinol signi cantly increased (45.1 ± 5.2 %) the ring rate in pVTA DA neurons, an effect that was reversed 5 min after of drug removal (Figure 2J, K, M, one-way ANOVA followed by Tukey's post hoc test).The effect of salsolinol on DA neuron ring rate was completely suppressed in the presence of caffeine (Figure 2L, M, one-way ANOVA followed by Tukey's post hoc test), reinforcing the results obtained in-vivo (Figure 1F).Similarly, the acute perfusion of 1 μM morphine caused a strong increase (75.1 ± 15.2 %) in ring rate in pVTA DA neurons, which was easily washed out after 5 min after of drug removal (Figure 2N, O, Q, one-way ANOVA followed by Tukey's post hoc test).The modulatory effect of morphine was also completely abolished in the presence of caffeine (Figure 2P, Q, one-way ANOVA followed by Tukey's post hoc test).These last results con rm that, in addition to the prevention of alcohol-induced salsolinol formation, other mechanisms must be involved in caffeine's inhibitory effects on alcohol-induced increase of mesolimbic DA transmission.
Pre-treatment with caffeine had the biggest impact on the pVTA biochemical pro les, pairwise PLS-DA model saline-water vs caffeine-water (CER = 0.18).Caffeine increased the abundance of several amino acids, such as the indole amino acids, phenylalanine, tryptophan, and tyrosine, methionine, arginine, and gamma-glutamylglutamate, glycerophospholipids, including different PCs (heptadecanoyl-, hexadecyl-, octadecanoyl-, stearoyl-hydroxy-glycero-phosphocholine) and PEs, such as palmitoyl-hydroxy-glycerophosphoethanolamine and stearoyl-hydroxy-glycero-phosphoethanolamine, sphingolipids, like tetracosenoyl-sphingenine and erythro-sphinganine, several fatty amides, including oleoylethanolamine and predicted ones, inosine, and 6-oxopurine.On the contrary, the abundance of adenosine and adenosine monophosphate, of indole-acetyl-glutamate, and of arachidonoylthio-PC appeared to decrease in response to caffeine.Moreover, caffeine generally reduced the carnitine pool: accordingly, Lcarnitine, acetyl-carnitine, butyryl-carnitine, lauroyl-carnitine and three predicted ones all decreased (Figure 3D and Supplementary Table 5).However, comparison of caffeine or saline pretreatment under alcohol treatment, pairwise PLS-DA model saline-alcohol vs caffeine-alcohol (CER = 0.36) revealed that caffeine had a completely opposite effect on carnitines under alcohol treatment.Accordingly, L-carnitine, butyrylcarnitine, arachidonoylcarnitine, and other four predicted carnitines increased with caffeine pretreatment under alcohol treatment (Figure 3C).These last results suggest that caffeine might affect carnitines pool differentially depending on the presence of alcohol.

Discussion
Alcohol consumption is one of the leading risk factors for premature death and disability, contributing to approximately 2.5 million deaths each year worldwide (66).The ability of alcohol to stimulate mesolimbic DA function (67), as a requirement to exert its reinforcing effects (14,41,46,65,(68)(69)(70), has critical implications for the development of alcohol use disorder (AUD) (71,72).Recent studies have shown that alcohol excites DA neurons in the pVTA (40) and stimulates DA transmission in the AcbSh acting as the pro-drug of salsolinol (41).Caffeine is a psychopharmacological agent devoid of addictive potential (10,14,73,74) and equally consumed worldwide as alcohol.The widespread diffusion of these two substances in the last decades has raised several questions on the clinical impact of their simultaneous consumption.The present study was aimed at characterizing the consequences of the interaction of a behaviorally relevant acute dose of each of these substances on DA function.The results reveal that the administration of caffeine prior to alcohol prevents its ability to generate salsolinol in the pVTA and, accordingly, to increase AcbSh DA transmission.This outcome suggests that caffeine might be preventing the ability of alcohol to increase AcbSh DA by interfering with the generation and/or with the bioavailability of salsolinol in the pVTA.However, as far as the reduction of the bioavailability is concerned, this possibility can be ruled out since salsolinol detection does not signi cantly differ, with and without systemic administration of caffeine, during pVTA perfusion with salsolinol.Moreover, differently from the catalase inhibitor 3AT, caffeine does not inhibit catalasedependent formation of salsolinol in-vitro.Consequently, the possibility that caffeine affects salsolinol generation directly inhibiting the enzyme catalase, whose activity is necessary to salsolinol formation (40,41), was also ruled out.Therefore, we hypothesized that caffeine could prevent alcohol stimulation on pVTA DA neurons, as well as alcohol-dependent generation of pVTA salsolinol and AcbSh DA transmission, via an adenosine receptor-mediated mechanism.Notably, as far as the generation of salsolinol is concerned, this was the case, since both the A 1 R and A 2A R antagonists, DPCPX and SCH 58261, prevented the generation (and detection) of salsolinol in pVTA after alcohol administration.However, the administration of A 1 R and A 2A R antagonists prior to alcohol revealed that these receptors differentially affect alcohol-elicited increases of AcbSh DA.Accordingly, SCH 58261, but not DPCPX, prevents the stimulation of AcbSh DA transmission by alcohol.This latter observation appears fully in agreement with the electrophysiological recordings with A 1 R and A 2A R antagonists.Therefore, both invivo and in-vitro evidence indicates that the action of systemic caffeine on alcohol-stimulated AcbSh DA is mediated by the prevention of salsolinol formation in the pVTA, via A 1 R-and A 2A R-mediated mechanism, although the nding that A 1 R blockade fails to affect alcohol-mediated increase of DA function, in-vivo and in-vitro, strongly points out a mechanism of alcohol-stimulated AcbSh DA and pVTA neuronal ring independent from salsolinol generation.Moreover, the observation that caffeine reduces the stimulation in-vivo (AcbSh DA, by reverse dialysis) and in-vitro (pVTA DA neuronal ring) by exogenous salsolinol, as well as the stimulation of pVTA DA neuronal ring in-vitro by morphine, con rms that salsolinol generation-independent mechanisms should be envisioned.In this regard it appears reasonable to hypothesize that under these conditions blockade of A 1 R, i.e. lack of an adenosine tone on A 1 R, frees a mechanism that, in spite of the unavailability of salsolinol in the pVTA, still results in DA neurons excitation and increased AcbSh DA release.Thus, since the local application of caffeine in the pVTA results in the same effects of the systemic one, it was reasonable to look for these additional biological mechanisms in the same region.
Accordingly, untargeted metabolomics analysis of pVTA lysates points out that both alcohol and caffeine in uence the abundance of various lipids, but also that caffeine prevents alcohol-induced alterations in the concentration of most of these molecules.In addition to their structural function in cellular membranes, lipids in the brain play crucial roles in regulating various physiological processes, including signal transduction (75), synaptic plasticity (76), and the release of neurotransmitters (77).The role of lipids in addiction is well known (78, 79) and alcohol (80-82), as well as other addictive substances including morphine (83) or cocaine, can alter their signaling.Lipid signaling is involved speci cally in DA mesolimbic transmission (84), by regulating reinforcing and motivational aspects of feeding ( 85), but also VTA DA neurons ring (86, 87).One of the lipids reduced by alcohol in the present study is oleamide.Oleamide is an endogenous fatty acid amide, derived from oleic acid, which can be synthesized in the mammalian nervous system and, among other effects, enhances the amplitude of currents gated by GABA A receptors (88).Notably, we recently demonstrated that GABA A agonists (89), similarly to caffeine (13), prevent alcohol-and morphine-induced conditioned place preference, as well as pERK increase in the AcbSh (90).Interestingly, recent studies revealed that intra-VTA administration of oleic acid inhibits DA tone (86), and that oleamide, acting as PPARα/CB1 receptor dual ligand, reduces alcohol intake and alcohol and oxycodone self-administration in rats (91).In the present study, caffeine prevents alcohol-induced reduction in oleamide.Moreover, caffeine also prevents alcohol-induced changes in PC and Lyso-PC which activate PPARα/γ in addition to other signaling pathways (92) , and have been suggested as potential targets for cocaine addiction (93).Interesting effects of caffeine were observed also on the carnitine pool.In fact, not only caffeine seems to prevent alcohol-induced reduction of two predicted acyl-carnitines, but it also appears to regulate carnitines abundance bidirectionally depending on the presence of alcohol.Carnitines are amino acid derivatives essential for the transportation of fatty acids into the mitochondria (94).A potentially therapeutic role of carnitines and acyl-carnitines in AUD has already been described in rodents (95,96) and humans (97).Moreover, previous studies reported that carnitine inhibits catalase activity and prevents catalase-mediated effects of alcohol in mice (98, 99).In the present study, caffeine-induced increase in carnitine and acylcarnitines, selectively under alcohol treatment, might have reduced catalase-mediated oxidation of alcohol, explaining the prevention of salsolinol formation in the pVTA and justifying the discrepancy between the effects of caffeine on catalase-mediated salsolinol generation in-vivo (preventive) and invitro (no effect).
In conclusion, the present work reveals for the rst time that caffeine prevents alcohol-induced activation of the mesolimbic DA pathway.Encouragingly, one of the few FDA-approved drugs for AUD, the m receptor antagonist naltrexone (ReVia®; Depade®), prevents the reinforcing effects of alcohol by interfering with its enhancement of the mesolimbic DA transmission (100) strengthening the potential of caffeine, and more speci cally of A2 A R antagonists, for future development of preventive/therapeutic strategies for AUD.Moreover, not only the stimulation of the mesolimbic DA pathway is the critical initiating event of the neurocircuitry of AUD, but also of addiction in general (14,28,29) and, since our results point out that caffeine can also prevent mesolimbic DA stimulation by the m receptor agonists, salsolinol and morphine, one of the future directions of this study will be to characterize further its effects on opioids-, as well as other drugs of abuse.More detailed studies will also be required to explain how A 2A R antagonism elicits its inhibitory activity on alcohol stimulation as well as the differential effects of A 1 R blockade on alcohol-mediated generation of pVTA salsolinol and stimulation of AcbSh DA, and to interpret the involvement of lipid signaling in caffeine effects on alcohol activity in the mesolimbic system.One of the limitations of this study is its exclusive focus on the mesolimbic DA pathway in alcohol naïve rats, which is only representative of the initial phase of AUD.We acknowledge the role of other brain circuits in the onset and self-perpetuating cycle of AUD, as well as the importance of other stages (i.e.withdrawal, craving, relapse) of the disease.Hence, future studies are required to explore the therapeutic potential of caffeine and adenosine receptor antagonists in both naïve and dependent rats, at different stages of the disease.Effects of caffeine and alcohol on the biochemical pro les of rats pVTA.

Figures
Figures

Figure 1 Effects
Figure 1

Figure 2 Effects
Figure 2

Figure 3
Figure 3 The mass spectrometer was operated in data-dependent acquisition (DDA) mode, and it was used in an m/z range from 100 to 1500 Da in the pVTA experiments and 50 to 750 Da in the in-vitro synthesis of salsolinol experiment, operating in positive ionization mode.Full scan MS1 was performed at 1e6 with a resolution of 35 000 and 70 000 for the pVTA and in-vitro synthesis of salsolinol experiment respectively, with a maximum ion injection time (IT) of 100 ms.MS2 experiments were performed at a resolution of 17 500 with maximum IT of 100 ms for pVTA and 50 ms for catalase, and TopN was used for the 5 most abundant precursor ions per MS2.The MS2 precursor isolation window was set to 1 m/z with no offset.The step collision energy was set to 20 eV, 30 eV, and 40 eV.Feature list was rst cleaned though blank ltering, only features with peak area ratio > 5 compared to blanks were kept.Data quality was assessed calculating coe cient of variance of internal standard in the samples and of the 6 standards present in the quality control samples (QCmix).Principal component analysis (PCA), via v 6.22 package, was used to inspect data and visualize possible outliers.
(59)) were injected into a Kinetex C18 column (50 × 2.1 mm, 1.7 µM particle size, 100 A pore size; Phenomenex, Cat#00B-4475-AN) at 30 °C column temperature.A ow rate of 0.5 mL/min was used for both the in-vitro synthesis of salsolinol and pVTA experiments with elution carried out using LC grade H 2 O (A) and 100% ACN (B), both acidi ed with 0.1% FA.Different elution gradients were used.For the in-vitro synthesis of salsolinol experiment: 0-1 min 0.1 % B, 1-3 min Acquired .rawleswereconvertedintoopen-access .mzMLformatusingMSConvert3.0.23(57).Both .rawand.mzMLleshave been deposited and can be downloaded from public metabolomics repository GNPS/MassIVE (https://massive.ucsd.edu/)under the accession codes MSV000094216 (pVTA experiment) and MSV000094218 (in-vitro synthesis of salsolinol experiment).Feature detection and extraction was performed using MZmine 3.9 (58).Brie y, mass detection noise for MS1 and MS2 was set at 5e4 and 1e3 respectively.ADAP chromatogram builder parameters were set as 4 minimum consecutive scans, 8e4 minimum absolute height, and 10 ppm m/z tolerance.Local minimum feature resolver module was set at 85% chromatographic threshold, 0.05 minimum search range RT, and 1.70 minimum ratio of peak top/edge.The 13C isotope lter was applied with an m/z tolerance of 5 ppm and a retention time tolerance of 0.03 minutes.Features were aligned using a m/z tolerance of 5 ppm and retention time tolerance of 0.2 minutes, with weight for m/z over RT was set to 3:1.Features not present in at least two samples and without MS2 acquisition were discarded.Finally, a feature list and two .mgfles,onefor molecular networking(59)and one for SIRIUS (60), were exported for downstream analysis.set.Same parameters were set for library search.Generated annotation table was used for subsequent analysis and network were visualized using Cytoscape 3.10 (62).Compound classes were predicted using CANOPUS (63) in SIRIUS 5.8.5.For the in-vitro synthesis of salsolinol experiment, a targeted peak extraction was also performed using Skyline v23.1 (64).Feature list was imported in R 4.2.2(The R Foundation for Statistical Computing, Vienna, Austria) for univariate and multivariate analyses.