Caffeine Regulates GABA Transport Homeostasis in the Adolescent Mouse Frontal Cortex via Adenosine A1 Receptor and PKC‐Dependent Pathways

ABSTRACT

Adolescence is a period of several brain changes, making it especially vulnerable to external influences. Abuse of psychoactive drugs, such as caffeine (CAFF), generates changes in cognitive functions. The main pharmacological targets of CAFF are mainly the A₁R and A_2AR adenosine receptors, which can regulate GABA homeostasis. We therefore evaluated the influence of CAFF intake upon GABA uptake and release in the frontal cortex (FC) of adolescent Swiss mice and addressed the underlying mechanism. Mice were treated for 5 days with a subcutaneous injection of CAFF (10, 20, and 40 mg/kg) every 24 h, and the FC dissected out at 1 h after the last injection for measurement of [³H]‐GABA uptake and release, cAMP accumulation, and density levels of GAT‐1, A₁R, A_2AR, PKA, and PKC. Calcium (Ca²⁺) imaging was performed on primary neuronal cultures treated with CAFF (200 μM). CAFF increased [³H]‐GABA uptake at all doses studied, an effect reversed by incubation with the selective GAT‐1 uptake inhibitor, NO‐711 (10 μM). At 20 and 40 mg/kg, CAFF also increased [³H]‐GABA release. CAFF also increased A₁R, but not A_2AR levels. The influence of CAFF involves pPKC activity since CAFF enhanced the pPKC/PKC ratio, while the PKC‐inhibitor Gö 6983 (100 nM) reversed the facilitatory action of CAFF upon GABA transport and prevented the CAFF‐induced increase in the frequency of Ca²⁺ transients in neuronal cell cultures. We conclude that CAFF alters GABAergic homeostasis in the FC, increasing GABA transport through PKC‐activity modulation.

Caffeine alters inhibitory neurotransmission by blocking adenosine A1 receptors and activating a non‐canonical intracellular signaling pathway. This mechanism increases calcium signaling and PKC activity, leading to enhanced GABA uptake and stimulated GABA release.

Abbreviations

[³H]‐GABA

tritiated gamma‐aminobutyric acid

A₁R

adenosine receptor type 1

A_2AR

adenosine receptor type 2A

AC

adenylyl Cyclase

ANOVA

analysis of variance

APMC

(likely typo for AC/PKA cascade, but kept as in text)

AR

adenosine receptor

BDNF

brain‐derived neurotrophic factor

Ca²⁺

calcium ion

CAFF

caffeine

cAMP

cyclic adenosine monophosphate

CAPES

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior

CHA

N6‐Cyclohexyladenosine (A₁R agonist)

CNPq

Conselho Nacional de Desenvolvimento Científico e Tecnológico

DAG

diacylglycerol

ECL

enhanced chemiluminescence

EDTA

ethylenediaminetetraacetic acid

EFSA

European Food Safety Authority

FBS

fetal bovine serum

FC

frontal cortex

FDA

Food and Drug Administration

GABA

γ‐aminobutyric acid

GAT‐1

GABA transporter type 1

Gö 6983

protein kinase C (PKC) inhibitor

GPCR

G protein‐coupled receptor

HBSS

Hank's balanced salt solution

HEPES

4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid

HPLC

high‐performance liquid chromatography

IP3

inositol 1,4,5‐trisphosphate

KCl

potassium chloride

MEM

minimum essential medium

MEPSC

Miniature Excitatory Postsynaptic Current

Na⁺

sodium ion

NO‐711

selective GAT‐1 inhibitor

PDL

poly‐D‐lysine

PKA

protein kinase A

PKC

protein kinase C

PLC

phospholipase C

PN

postnatal day

pPKC

phosphorylated protein kinase C

proBDNF

pro brain‐derived neurotrophic factor

PV

parvalbumin‐positive interneurons

RGS

Regulators of G protein Signaling

RIPA buffer

Radioimmunoprecipitation assay buffer

ROI(s)

region(s) of interest

RRID

Research Resource Identifier

SC

subcutaneous

SDS‐PAGE

sodium dodecyl sulfate‐polyacrylamide gel electrophoresis

TBS‐T

tris‐buffered saline with Tween 20

μM

micromolar

1. Introduction

Caffeine (1,3,7‐trimethylxanthine, CAFF) is the most consumed drug in the world, reaching 80% of the world's population (Gurley et al. 2015; Temple et al. 2017; Willson 2018). CAFF is a methylxanthine alkaloid that has several endogenous molecular targets, whose activation and effects depend on a dose–response relationship. At low doses, like those in a few cups of coffee, CAFF can act as a non‐selective and competitive antagonist of adenosine receptors (mainly A₁ and A_2A receptors) (Daly and Fredholm 1998; Fredholm 1995; Fredholm et al. 2017). At higher and highly toxic doses, CAFF can promote inhibition of the cyclic nucleotide phosphodiesterase enzyme, and subsequent accumulation of cyclic adenosine monophosphate (cAMP) (Gurley et al. 2015; Ribeiro and Sebastiao 2010), block GABA receptor type A, and mobilize intracellular calcium (Ca²⁺) levels (Cappelletti et al. 2015; Williams and Jarvis 1988; Willson 2018). Recently it has been described that around 75% of children and adolescents aged 6–19 have consumed CAFF daily above the doses recommended by the Food and Drug Administration (FDA) and the European Food Safety Authority (EFSA) which are 2.5–3 mg/kg per day respectively (Mitchell et al. 2015; Poole et al. 2016; Temple et al. 2017; Verster and Koenig 2018; Wikoff et al. 2017). The increment in CAFF abuse can be explained by the expansion of CAFF‐based products and the marketing designed to attract young consumers (Ahluwalia and Herrick 2015). Importantly, children and adolescents may be especially vulnerable to the effects of CAFF due to lower body mass and the ongoing maturation of the central nervous system, resulting in greater neurobiological impact for a given dose when compared to adults (Temple et al. 2017).

Beyond central effects, increased consumption of CAFF, particularly through energy drinks, has been associated with elevated blood pressure, changes in heart rate, and severe cardiac alterations in adolescents and young adults (Temple et al. 2017). Moreover, CAFF intake has been linked to risk‐taking behaviors, irritability, aggression, violence, and increased susceptibility to substance abuse (Arria et al. 2014; Harris and Munsell 2015; Temple et al. 2017). Animal studies further demonstrate that early‐life CAFF exposure increases vulnerability to anxiety‐ and depression‐like behaviors, sleep disturbances, and impairments in mineral absorption and bone health, effects that may persist into adulthood even after CAFF withdrawal (Ahluwalia and Herrick 2015; Harris and Munsell 2015; O'Neill et al. 2016). Notably, CAFF exerts a biphasic and dose‐dependent effect on the central nervous system, with low‐to‐moderate doses producing beneficial effects on reaction time, motor performance, vigilance, attention, mood, and information processing, while higher doses lead to deleterious outcomes (Ahluwalia and Herrick 2015; Lorist and Tops 2003; Mednick et al. 2008).

Several studies indicate that human adolescents consume CAFF primarily to counteract sleepiness during nighttime leisure activities, such as video gaming and the use of electronic devices, as CAFF increases sleep latency and disrupts sleep architecture (James et al. 2011; Temple et al. 2017). In humans, CAFF consumption during adolescence has been associated with impairments in attention, learning, and memory, as well as alterations in sleep quality (Aepli et al. 2015; Harris and Munsell 2015; Poole et al. 2016). Sleep disturbances may also serve as behavioral markers of CAFF‐related effects in children and adolescents, including withdrawal‐associated somnolence and deficits in academic performance (Ahluwalia and Herrick 2015; James et al. 2011; Nehlig 2016).

Adolescence is defined by the World Health Organization as the period between 10 and 19 years of age and is characterized by profound hormonal, physiological, neural, and behavioral changes (Ahmed et al. 2015; Spear 2013; Zimmermann et al. 2019). This developmental stage is associated with increased novelty seeking, emotional reactivity, social reorientation, and heightened vulnerability to psychosocial stressors, which may contribute to a greater risk of substance use and abuse (Arria et al. 2014; Spear 2013). These behavioral traits are evolutionarily conserved and observed across several mammalian species (Doremus‐Fitzwater et al. 2010). In rodents, adolescence is typically defined as the postnatal period between postnatal day (PN) 21 and PN60, while pubertal maturation of the hypothalamic–pituitary‐gonadal axis occurs between PN28–42 (or PN30–39) in females and PN42–49 (or PN40–45) in males (Hueston et al. 2017; Koss and Frick 2017; Silveri 2014).

During adolescence, cognitive functions such as working memory, attention, and executive control are still developing, which may result in poorer performance on cognitive tasks compared to adulthood (Anderson et al. 2001; Gur et al. 2012; Koss and Frick 2017). This cognitive immaturity reflects ongoing structural and functional maturation of key brain regions, including the frontal cortex (FC), which has an important role in the control and coordination of activities through cortical and subcortical connections, as well as attention, working memory, and decision‐making (Chai et al. 2018; Datta and Arnsten 2019; Dembrow et al. 2010), which undergo significant neurocognitive development during adolescence (Perica and Luna 2023). Mouse adolescence is established from 21 until 60 postnatal days (PN) (Doremus‐Fitzwater et al. 2010; Silveri 2014). Some studies show that teenagers have problems taking some cognitive tests, an issue attributed to the fact that key brain regions are not yet fully developed during adolescence in numerous mammalian species (Koss and Frick 2017). For example, in rodents, the FC undergoes anatomical and functional remodeling processes during cortical development, occurring later and in “waves” involving synaptic remodeling and increased connectivity with subcortical structures (Ahmed et al. 2015; Caballero and Tseng 2016). Consequently, FC‐dependent functions such as cognitive control, attentional regulation, learning, memory, and inhibitory behavior continue to mature throughout adolescence (Ahmed et al. 2015; Spear 2013; Zimmermann et al. 2019).

In addition, adolescence is also a period of refinement of the circuitry of different neurotransmitter systems. Among these systems, the activity of the GABAergic system is included. In rodents, GABAergic receptors reach maturity and are more responsive to GABA in stressful situations in this phase (Kilb 2012; Luhmann et al. 2014). In the prefrontal cortex, an initial increase in GABAergic interneurons during preadolescence is followed by a reduction in GABAergic inputs and neuronal size, particularly in projections from the nucleus accumbens to the FC in both humans and rodents (Schepis et al. 2008). While vesicular GABA transporters (VGATs), responsible for intracellular GABA storage, show relatively stable expression during cortical development. Membrane transporters, responsible for GABA reuptake (GAT‐1,2,3), located both in neurons and glial cells, increase their expression in adolescence and tend to decline in adulthood (Kilb 2012; Silveri 2014). GABA transporters can also undergo changes in activity, as a consequence of the modulatory action of phosphatases and kinases, as well due to changes in second messengers, hormones, pH and Na⁺ gradient (do Nascimento et al. 1998; Hu and Quick 2008; Whitworth and Quick 2001). Studies from our group have also shown that the activation of other neurotransmitter systems may be associated with changes in GABA transport in the adolescent mice FC (Martins et al. 2018) and in the avian retina (Borges‐Martins et al. 2019; Ferreira et al. 2014; Kubrusly et al. 2018). In addition, recent work has demonstrated that chronic exposure to CAFF was able to alter the GAT‐1 transporter in the avian retina, via activation of A₁R and cAMP/PKA signaling (Borges‐Martins et al. 2019). Despite the growing body of evidence linking CAFF consumption to behavioral and cognitive alterations during adolescence, the underlying neurobiological mechanisms, particularly those involving inhibitory neurotransmission in the FC, remain poorly understood. Therefore, in the present study, we investigate the mechanisms by which CAFF modulates GABA transport in the FC during adolescence.

2. Material and Methods

2.1. Ethics Statement

All experiments were carried out under institutional approval of the Animal Care and Use Committee of Fluminense Federal University (CEUA/968/2017), following Brazilian Law n° 11.794/2008 and with the Guide for the Care and Use of Laboratory Animals, as adopted and promulgated by the National Institutes of Health. Experiments in Portugal complied with European Rules and Guidelines (2012/707/EU) and the Portuguese legislative action (DL 113/2013) for the protection of animals used for scientific purposes.

Mice and rats were obtained from institutional animal facilities that perform routine welfare monitoring, and rodents presenting signs of suffering were humanely euthanized by facility staff in accordance with ethical guidelines before inclusion in experimental procedures. At the investigators' request, rats originating from very small litters (< 3 pups per litter) were not allocated to experimental protocols, in order to minimize potential developmental variability.

Mice were anesthetized with isoflurane by inhalation in a chamber (induction at approximately 3%–4% isoflurane in oxygen, maintained until loss of reflexes) and euthanized by decapitation. Rat pups used for primary cell culture preparation were euthanized by decapitation without prior anesthesia, in accordance with institutional and European guidelines for neonatal rodents, as this method provides rapid death and minimizes tissue exposure to anesthetic agents that could interfere with subsequent cellular analyses.

In all cases, all efforts were made to minimize the number of mice and rats used and their suffering. No formal randomization procedure was employed for allocation of mice or rats to experimental groups; mice were assigned based on cage allocation performed by the animal handling facility staff, as described in the Methods section. In addition, no blinding procedures were applied during the experimental procedures or during data analysis.

2.2. Material

[³H]‐GABA (specific activity 35 Ci/mmol) was obtained from Perkin Elmer (Waltham, MA, USA). All other reagents, including caffeine (C0750), A₁ adenosine receptor agonist N(6)‐cyclohexyladenosine (CHA—C9901), PKC inhibitor Gö 6983 (G1918), and PKA inhibitor H‐89 dihydrochloride hydrate (H‐89—B1427) were obtained from Sigma‐Aldrich (St Louis, MO, USA).

2.3. Animals and Caffeine Treatment

Swiss mice (RRID: MGI:3795101) were obtained locally from the Laboratory Animal Center (NAL) at the Federal Fluminense University (UFF). NAL is responsible for the breeding of laboratory animals at UFF and is currently linked to CPE/PROPPI. The mice were housed in groups of 4–5 in clear polycarbonate cages 48 cm × 27 × 20, in an acclimatized room at a temperature between 18°C and 20°C in a 12 h light/dark cycle, with water and feed ad libitum. The mice were transported from the animal facility to the experimental room 3 h before each experimental procedure.

For experiments involving cell culture, pregnant Sprague–Dawley (RRID: MGI:5651135) rats were obtained from Charles River Laboratories and housed at the animal maintenance facility of the Instituto de Medicina Molecular (IMM), University of Lisbon. A total of four pregnant dams were used. Litters were collected at postnatal Day 18–19, with an average litter size of 14 pups. Pups from one litter were pooled prior to cell isolation, and each pooled preparation was considered one independent biological replicate.

On PN 21, the mice were weaned and separated by sex. Treatment was performed in vivo in male and female Swiss mice between PN 31 and PN 40 (190 mice) weighing approximately 30–40 g; at least 4 different animals were used in each experimental group. This treatment period corresponds to the onset of puberty, which is based on brain development and drug reactivity patterns, representing the beginning and middle of adolescence in rodents (Ribeiro‐Carvalho et al. 2008). At PN31, mice received subcutaneous injections of CAFF (10, 20, or 40 mg/kg) or saline once daily during the light cycle. Experimental groups were defined as follows (Figure 1): group A received CAFF or vehicle for 5 days (PN31‐PN35); group B received the same treatment followed by a 5‐day withdrawal period (PN36‐PN40); group C received a single acute CAFF administration 1 h before experiments at PN40; and group D underwent the same protocol as group B with an additional CAFF re‐exposure 1 h before experiments at PN40. The vehicle and CAFF groups received an equal volume (100 μL) of saline or saline + CAFF. In addition to being treated, the mice were weighed and had their snout/broth length measured on Days 1, 3, and 5 of treatment. Except for group B, the experiments were performed 1 h after the last CAFF administration.

FIGURE 1.

Experimental timeline of mice treatments. The green line at the top of the image represents the developmental period of the mice used in this study, spanning preadolescence (up to PN35) and middle adolescence (PN35 onward). Groups A, B, C, and D are represented in distinct colors, with a black dot indicating the time of euthanasia for ex vivo assays. Group A (blue): Mice received subcutaneous injections of caffeine (10, 20, or 40 mg/kg) or vehicle from PN31 to PN35, administered once daily during daylight hours at the same time each day, followed by euthanasia on PN35. Based on results, a dose of 20 mg/kg was selected for subsequent groups. Group B (yellow): Mice received caffeine (20 mg/kg) or vehicle from PN31 to PN35, followed by a withdrawal period (PN36‐PN40), and then euthanasia for experiments on PN40. Group C (orange): Mice received an acute administration of caffeine (20 mg/kg) 1 h before euthanasia on PN40. Group D (pink): Mice received subcutaneous caffeine (20 mg/kg) or vehicle, following the protocols of group B, but in addition on PN40 1 h before the experiments were re‐exposed to caffeine (20 mg/kg).

2.4. Isolation and Dissection of the Frontal Cortex

The brains of mice were quickly extracted from the cranial cavity and, after removing the meninges and the olfactory bulb, the third anterior to the corpus callosum was isolated, this region corresponding to the FC. The FC hemispheres were separated and distributed in duplicates or triplicates for each experimental condition: in the [³H]‐GABA uptake biochemical assays, cAMP assays, and western blot assays. Slices were also made by hand from the anterior third of the corpus callosum with the aid of a scalpel. Then, each slice was distributed in one well (triplicate of one animal) for each experimental condition in the assay of [³H]‐GABA release.

2.5. [ ³H]‐GABA Uptake

Mice previously treated subcutaneously with injected CAFF as detailed in 2.3 above were anesthetized with isoflurane, decapitated, and the brain quickly removed. FC was dissected and placed in wells containing 1 mL 4 mM glucose Hank's Balanced Salt Solution (4 mM HBSS) buffered at pH 7.4 at 37°C. The tissue was incubated with 1 μCi of [³H]‐GABA and 20 μM of non‐radioactive GABA for 60 min. In other experimental protocols, the FC hemispheres were incubated in 4 mM HBSS, to which NO‐711 (10 μM), CHA (100 nM), or Gö 6983 (100 nM) were added for 15 min before the incubation period with [³H]‐GABA and non‐radioactive GABA. At the end of the incubation, the medium was removed, and the tissue was washed three times with 3 mL of cold 4 mM HBSS. This procedure was sufficient to wash out the free radioactivity (not taken up by the tissue). Ultrapure deionized water (Milli‐Q) was then added to burst the cells. Following successive freeze–thaw cycles, cellular radioactivity was assayed using a liquid scintillation counter. The scintillation count was normalized by the protein content, which was assayed by the Lowry protein method (Lowry et al. 1951).

2.6. [ ³H]‐GABA Release

For [³H]‐GABA release assays, the FC slices were incubated with [³H]‐GABA and 20 μM of non‐radioactive GABA for 60 min as described above (see 2.5). After 1 h of reaction, the slices were washed with 12 mM glucose Hank's Balanced Salt Solution (12 mM HBSS) at 37°C to remove radioactive material that was not incorporated by the cells. The CF slices were incubated for successive periods of 5 min with 12 mM HBSS as follows: sample 1 (discarded, transition period); sample 2–3 (collected); sample 4 (discarded, transition period); sample 5–6 (collected). The concentration of KCl was increased to 80 mM immediately after collecting sample 3 and onwards. The incubations were performed statically in a water bath at 37°C. By the end of the incubations, the amount of tritium in each of the collected samples was counted by liquid scintillation. Counts from samples collected under similar conditions were taken as duplicates and were then pooled together. Samples 2 and 3 correspond to basal release under non‐depolarizing conditions, whereas samples 5 and 6 correspond to depolarization‐evoked release.

2.7. Western Blot Assay

Samples of FC were homogenized with 1 mL of RIPA buffer with a protease inhibitor cocktail. Protein concentration was estimated by the Bradford protein assay (Bradford 1976). Briefly, samples were diluted in buffer composed of 10% glycerol (v/v), 1% ß‐mercaptoethanol, 3% SDS, and 62.5 mM Tris base and boiled for 5 min. Approximately 45 μg of protein from each sample was separated by 10% SDS‐PAGE electrophoresis and transferred to nitrocellulose membranes (ECL‐Hybond). Membranes were washed with Tween 20 Tris‐buffered saline (TBS‐T) and blocked for 1.5 h with 5% non‐fat milk in TBS‐T. Membranes were then incubated with anti‐A₁ receptor (1:200 in TBS‐T; Millipore Cat# 119117‐100UL, RRID: AB_211412), anti‐A_2A receptor (1:1000 in TBS‐T; Santa Cruz Biotechnology Cat# sc‐70 321, RRID: AB_2226516), anti‐GAT‐1, (1:500 in TBS‐T, Millipore Cat# AB1570W, RRID: AB_90791), anti‐PKA alpha + beta phospho Thr197 (1:750 in TTBD; GeneTex Cat# GTX25815, RRID: AB_380704), anti‐PKC (1:1000 in TBS‐T; Millipore Cat# PC20, RRID: AB_2252828) and (anti‐pPKC alpha 1:1000; Huabio Cat# ET1702‐17, RRID: AB_3070280) overnight at 4°C. The next day, membranes were rinsed in TBS‐T and incubated with anti‐rabbit peroxidase‐conjugated secondary antibody (1:4000 in TBS‐T; Millipore Cat# AP132P, RRID: AB_90264) for 2 h at room temperature followed by three washes in TBS‐T (10 min. each). Immunoreactive bands were detected by chemiluminescence using a ChemiDoc imaging system (Bio‐Rad) following protein separation on hand‐cast (in‐house prepared) Mini‐PROTEAN gels and detected with an ECL kit (Amersham). Blots were re‐probed with anti‐β‐tubulin antibody (1:20.000 in TBS‐T, Sigma‐Aldrich Cat# T8660, RRID: AB_477590) for 1 h at room temperature, rinsed in TBS‐T buffer, and incubated with anti‐mouse peroxidase‐conjugated secondary antibody (Sigma‐Aldrich Cat# 401215, RRID: AB_10682749) for 45 min. at room temperature. Following three TBS‐T washes (10 min. each). Image acquisition was initially performed using the automatic exposure setting; when signal saturation was indicated by the system, manual exposure times were applied and selected at a time point before saturation for quantitative analysis.

2.8. cAMP Assay

Samples of FC from treated and control mice were obtained, cut into < 2 mm² pieces, and incubated for 10 min. at 37°C in Minimum Essential Medium (MEM) buffered with 20 mM HEPES, pH 7.3, containing 100 μM ascorbic acid, 100 μM pargyline, and 0.5 mM RO‐20 (a phosphodiesterase inhibitor). The reaction was stopped by adding 10% TCA (final concentration). The cAMP was purified and assayed by previously described methods (Gilman 1970; Matsuzawa and Nirenberg 1975), with cAMP levels normalized by protein content obtained utilizing the Lowry protein assay (Lowry et al. 1951).

2.9. Primary Culture of Neurons

Primary neuronal cultures were prepared from Sprague–Dawley fetuses (prenatal day 18/19) that were euthanized by decapitation and using a standard protocol (Fonseca‐Gomes et al. 2024). Cerebral cortex and hippocampus tissues were isolated, meninges removed, and the tissue digested in 10% (v/v) trypsin‐ethylenediaminetetraacetic acid (EDTA) in HBSS for 15 min at 37°C. Trypsin activity was stopped with 30% (v/v) fetal bovine serum (FBS) in HBSS, followed by 3× centrifugation cycles at 200 g for 10 min (Eppendorf, 5810R, Hamburg, Germany, RRID: SCR_019855). Cells were resuspended in supplemented Neurobasal medium (0.5 mM l‐glutamine, 25 μM glutamic acid, 2% (v/v) B‐27, and 12 μg/mL gentamycin), strained, and plated at 5 × 10⁴ cells/well plates in 9.4 × 10.7 × 6.8 (mm) 8‐well glass‐bottom plates (ibidi GmbH, Martinsried, Germany, RRID: SCR_027508) previously coated with Poly‐D‐lysine (PDL). Cultures were maintained at 37°C, 5% CO₂, with medium replenishment on the 7th day in culture.

2.10. Ca²⁺ Imaging and Analysis

Ca²⁺ signaling recordings from cultured neurons of rat were performed as described previously (Ferreira et al. 2017). We used rats instead of mice since the yield is much higher in fetuses from rats than from mice. The use of mice and rat in different parts of the study follows established literature (Ferreira et al. 2017; Fonseca‐Gomes et al. 2024; Ghosh et al. 2024; Gonçalves‐Ribeiro et al. 2024; Martins et al. 2020) and reflects the conserved biological processes targeted. Also, while performing experiments in cell cultures we moved from adolescent mice where caffeine was administered in vivo to fetal cultured cells where caffeine was administered ex vivo. We acknowledge that species, age and administration procedure differences should be considered as a potential confounding factor when interpreting the results, but their consistency throughout the study somehow precludes major misinterpretations. To prepare neuronal cultures, brain tissue was pooled by litter prior to homogenization and plating. Neurons were cultured in Ibidi plates and mounted on an inverted microscope (Axiovert 135TV, Zeiss) equipped with a xenon lamp and 340‐nm and 380‐nm bandpass filters. Cells were maintained at 37°C in a humidified atmosphere during the experiment. Test drugs were added directly to the medium, 5 min after starting recordings. Each neuronal culture preparation was derived from a single litter, and the experimental sample size corresponds to the number of independent culture preparations (i.e., litters). Time zero in the figures indicates the starting of recording. Image pairs were taken every 5–10 s using excitation wavelengths of 340 and 380 nm, and ratio images were generated. The excitation wavelengths were switched via a Lambda DG‐4 high‐speed wavelength changer (Sutter Instrument), and emission was detected at 510 nm. Data were recorded with a CCD camera (Photometrics CoolSNAP fx) and analyzed using MetaFluor software (Universal Imaging, MetaFluor Fluorescence Ratio Imaging Software) (RRID: SCR_014294). Regions of interest (ROIs) were defined by profiling cells, and fluorescence intensity values were converted into 340/380 nm ratios. These values were normalized to the first recorded ratio for each cell. For the analysis of Ca²⁺ transients, the frequency of events was extracted using MATLAB software (RRID: SCR_001622) as previous described (Lopes et al. 2024). A baseline was defined in the initial 5 min, where the mean ± SEM values were obtained for each cell. Transients were validated based on criteria adapted from (Horvat et al. 2016): a transient was considered valid only if the 340/380 nm ratio exceeded the baseline mean plus three times the SEM, remained above this threshold for 10–700 s, and was not observed during the baseline period. For each region of interest, transient peaks and occurrences were recorded for statistical analysis. Cells with spontaneous Ca²⁺ transients during baseline were excluded from further analysis.

2.11. Statistics

All statistical analyses were performed using GraphPad Prism version 10.3.0 (GraphPad Software Inc., RRID: SCR_002798). In each experiment, mice from at least three independent litters were used per condition, and mice were arbitrarily assigned to saline or caffeine (CAFF) treatment groups.

Data normality was assessed using the Shapiro–Wilk test (Table S1). Outlier detection was performed using Grubbs' test with a significance level of 0.05; no outliers were identified.

Regarding sex as a biological variable, preliminary analyses were conducted comparing male and female mice within each experimental condition (Table S2). These analyses revealed neither a significant main effect of sex. Consequently, data from males and females were pooled for subsequent analyses, and statistical comparisons were performed considering treatment as the primary factor. Experiments included mice of both sexes; when minor imbalances in sex distribution occurred, the absence of sex‐dependent effects justified consolidation of the data.

For body mass assessment, linear regression analysis was performed with evaluation of slope and y‐intercept to assess changes in body mass over time. Comparisons between two independent groups were performed using two‐tailed Student's t‐tests. Comparisons involving more than two groups were analyzed using one‐way or two‐way ANOVA, as appropriate. For two‐way ANOVA, ordinary two‐way ANOVA was followed by Fisher's least significant difference (LSD) post hoc tests, applied to compare cell means within rows and columns in a planned 2 × 2 factorial design. For one‐way ANOVA, Bonferroni's post hoc test was applied for multiple comparisons, with all groups compared against the first column, which served as the reference condition. Nonparametric data were analyzed using the Kruskal‐Wallis test, followed by Dunn's multiple‐comparisons test. Data are presented as Mean ± SEM with the corresponding test statistics (t, degrees of freedom, or F values with numerator and denominator degrees of freedom, as applicable). For all analyses, p < 0.05 was considered statistically significant.

Sample size was estimated a priori using G*Power (α = 0.05, power = 0.80). Due to ethical constraints associated with animal use, the calculated sample size could not always be achieved and is acknowledged as a limitation of the study.

3. Results

3.1. Role of Caffeine in Growth Measures: Body Mass and Length (Snout/Tail)

The consumption of CAFF has been associated with changes in body mass both in men and in women, which is linked to CAFF thermogenic action and an increase in the metabolism and oxidation of lipids, promoting fat burning (Greenberg et al. 2006; Lopez‐Garcia et al. 2006). Such effects may impact mice development. Therefore, we first evaluated whether different doses of CAFF affected mass gain and length over 5 days of treatment, with three weighing points being performed: Day 1 corresponding to the weighing immediately after CAFF administration, and Days 3 and 5 being the weighing points after CAFF had already been administered.

Body mass was monitored across the experimental period to assess potential effects of repeated CAFF administration on general development. Although body mass increased over time in all groups, no differences were observed between CAFF‐treated and control mice at any time point, and body mass remained comparable across treatments throughout the experimental period (Figure 2a). These changes are consistent with normal growth rather than treatment‐related effects.

FIGURE 2.

Effects of caffeine on body mass and body length. (a) Body mass was measured at PN31, PN33, and PN35 in control mice and animals treated with caffeine (10, 20, or 40 mg/kg). Body mass increased over time in all groups, with no differences between caffeine‐treated and control mice at any time point. CTRL (PN31: 26.97 ± 0.74; PN33: 28.49 ± 0.88; PN35: 29.21 ± 0.87; n = 41), CAFF 10 mg/kg (PN31: 26.45 ± 0.58; PN33: 28.01 ± 0.53; PN35: 28.84 ± 0.57; n = 41), CAFF 20 mg/kg (PN31: 24.82 ± 0.90; PN33: 26.22 ± 0.94; PN35: 26.62 ± 0.92; n = 27), CAFF 40 mg/kg (PN31: 24.65 ± 1.03; PN33: 25.47 ± 0.89; PN35: 26.05 ± 0.85; n = 23). Linear regression slope analysis: CTRL F(1,121) = 3.662, p = 0.0580; CAFF 10 mg/kg F(1,121) = 9.127, p = 0.0031; CAFF 20 mg/kg F(1,79) = 1.949, p = 0.1666; CAFF 40 mg/kg F(1,67) = 1.152, p = 0.2869. (b) Body length (snout‐to‐tail) was measured at PN31 and PN35 in control and CAFF 20 mg/kg‐treated mice. No differences were observed between treatments or across time points. CTRL (PN31: 17.50 ± 0.36; PN35: 17.64 ± 0.28; n = 7), CAFF 20 mg/kg (PN31: 17.38 ± 0.32; PN35: 18.13 ± 0.25; n = 8). Two‐way ANOVA: Interaction F(1,26) = 0.9880, p = 0.3294; time F(1,26) = 2.137, p = 0.1558; treatment F(1,26) = 0.3419, p = 0.5638. Data are presented as mean ± SEM. Statistical analyses were performed using linear regression with slope analysis (a) and two‐way ANOVA followed by Fisher's LSD multiple‐comparisons test was used to compare cell means within rows and columns (b). Each dot in b represents one individual mice. No statistically significant differences were observed between groups. Median (center line), mean (+), SD (box), and min–max (whiskers).

To further evaluate somatic development, snout‐to‐tail length was measured in mice receiving the 20 mg/kg dose of CAFF. Snout‐to‐tail length did not differ between vehicle‐ and CAFF‐treated mice, nor did it change over the course of the 5‐day treatment period within the same group (Figure 2b). Together, these results indicate that 5 days of CAFF administration did not produce detectable alterations in developmental parameters, including body mass and body length.

3.2. The Dose–Response and Temporal Curve of Caffeine Treatment

Our next objective was to investigate how different doses of CAFF (10, 20, or 40 mg/kg) could affect the GABA uptake and release system. In our study model, mice in experimental group A received CAFF (10–40 mg/kg) or vehicle treatment for 5 days according to the timeline presented in the methodology section. CAFF treatment increased [³H]‐GABA uptake in the FC across all doses tested, with higher uptake levels observed in CAFF‐treated mice compared to controls (Figure 3a).

FIGURE 3.

Effects of caffeine on GABA uptake and release. (a) [³H]‐GABA uptake was measured in FC slices from mice pretreated with caffeine and evaluated at PN35. Caffeine increased GABA uptake at all doses tested compared with control: CTRL (99.20 ± 3.98; n = 15), CAFF 10 mg/kg (149.4 ± 9.92; n = 8), CAFF 20 mg/kg (159.0 ± 12.89; n = 15), CAFF 40 mg/kg (175.0 ± 21.89; n = 4). One‐way ANOVA revealed a significant treatment effect (F(3,38) = 9.135; p = 0.0001), with multiple‐comparisons analysis indicating differences between CTRL and all caffeine‐treated groups. (b) [³H]‐GABA release of basal and CAFF‐treated animals was assessed under non‐depolarizing and depolarizing conditions. Increased GABA release was observed at 20 and 40 mg/kg caffeine doses (CTRL: 4.93 ± 0.46; n = 10; 10 mg/kg: 6.82 ± 0.88; n = 9; 20 mg/kg: 7.56 ± 0.62; n = 17; 40 mg/kg: 7.86 ± 0.95; n = 11), depolarization with KCl increased [³H]‐GABA release when compared to basal but not CAFF‐treated animals (Control: 8.34 ± 0.42; n = 12; 10 mg/kg: 8.51 ± 0.49; n = 12; 20 mg/kg: 8.29 ± 0.53; n = 11; 40 mg/kg: 8.73 ± 0.81; n = 12). Two‐way ANOVA showed a significant effect of KCl (F(1,86) = 12.16; p = 0.0008), with no main effect of treatment (F(3,86) = 2.16; p = 0.0983) and no interaction (F(3,86) = 1.65; p = 0.1848). (c) [³H]‐GABA uptake was further evaluated following distinct caffeine exposure paradigms. Increased uptake was observed in Group A (PN35; 142.6 ± 10.45; n = 9), Group C (PN40, acute caffeine re‐exposure after abstinence; 127.6 ± 7.02; n = 14), and Group D (PN40, prolonged caffeine exposure; 128.4 ± 9.22; n = 16), compared with CTRL (99.20 ± 3.98; n = 15). Group B (PN40, caffeine abstinence) showed uptake levels comparable to control (83.17 ± 7.47; n = 6). One‐way ANOVA indicated a significant effect of treatment (F(4,55) = 6.834; p = 0.0002), with multiple‐comparisons analysis identifying differences between CTRL and Groups A, C, and D, but not Group B. Data are presented as mean ± SEM. Statistical analyses were performed using one‐way ANOVA followed by Bonferroni multiple‐comparisons tests to compare cell means to control column. Each dot represents one individual mice. Asterisks indicate statistical significance as follows: *p < 0.05; **p < 0.01; ***p < 0.001. If not shown, the comparison was not significant (ns). Median (center line), mean (+), SD (box), and min–max (whiskers).

To determine whether CAFF also modulates GABA release in the FC of adolescent mice, [³H]‐GABA release was evaluated under non‐depolarizing and depolarizing conditions. In FC samples taken from mice that did not receive CAFF (control mice), KCl induced an increase in the release of [³H]‐GABA (Figure 3b, gray bars). In FC samples from animals treated with CAFF, the basal release of [³H]‐GABA was already increased even under basal non‐depolarizing conditions, being statistically different (p < 0.05) from controls (samples from non‐treated mice under non‐depolarizing conditions) at 20 and 40 mg/kg of CAFF. KCl‐induced depolarization did not cause any further increase in the release of [³H]‐GABA, likely due to saturation of the release process.

Based on the dose–response curve for both GABA uptake and release modulation, the intermediate CAFF dose of 20 mg/kg was selected for subsequent experiments. We then aimed to evaluate whether the facilitatory effect of CAFF upon GABA uptake depends on its presence in the body shortly before tissue preparation for experimental analysis. For that, we performed experiments with different times of exposure to CAFF (see timeline in methodology session—Figure 1). Group A was exposed to CAFF or vehicle for 5 days, as in the experiments mentioned above; group B was exposed to CAFF or vehicle for 5 days and then a 5‐day withdrawal period before tissue preparation; group C was exposed to a single acute CAFF administration 1 h before tissue preparation; group D was exposed to CAFF for 5 days, then a 5‐day withdrawal, and then a single acute dose of CAFF or vehicle 1 h before tissue preparation. Treatment with CAFF at 20 mg/kg for 5 days resulted in increased [³H]‐GABA uptake in Groups A and D, whereas no change in uptake was observed in Group B (Figure 3c). Acute administration of CAFF (group C) also significantly increased [3H]‐GABA uptake (Figure 3c).

In conclusion, the dose–response analysis for GABA uptake indicated that a maximal effect of CAFF was already attained with a dose of 10 mg/kg, reaching a plateau for concentrations up to 40 mg/kg. In contrast, release experiments demonstrated significant effects only at doses of 20 and 40 mg/kg. Temporal assessment revealed that the facilitatory action of CAFF required administrations performed shortly before tissue preparation. Based on these observations, all subsequent experiments were conducted following the protocol applied to Group A and at a dose of caffeine of 20 mg/kg.

3.3. Caffeine Modulates [ ³H]‐GABA Transport via Enhancement of GAT‐1 Activity

Previous studies demonstrated that, during adolescence, there is an important role of the GAT‐1 transporter in maintaining GABAergic homeostasis (Kubrusly et al. 2020; Martins et al. 2018). Therefore, our next objective was to investigate whether the GAT‐1 transporter was involved in the increase in [³H]‐GABA uptake induced by CAFF treatment. To do so, we compared in the same experiments the effect of CAFF in the presence and absence of the GAT‐1 inhibitor, NO‐711 (10 μM), in the incubation medium. As before, [³H]‐GABA uptake in FC samples from mice exposed to CAFF (20 mg/kg) was significantly higher (p < 0.001, Figure 4a) than that obtained in samples from mice not exposed to CAFF (control). The presence of the GAT‐1 blocker, NO‐711, in the incubation medium, decreased [³H]‐GABA in control samples (non‐CAFF exposed, open bars, Figure 4a, p < 0.01) and abolished the enhancement of [³H]‐GABA uptake in samples from mice exposed to CAFF (20 mg/kg). Indeed, [³H]‐GABA uptake in FC samples taken from mice exposed to CAFF (filled bars, Figure 4a) was markedly lower (p < 0.001) in the presence of NO‐711 than in its absence (filled bars, Figure 4a). To determine whether the CAFF‐induced increase in GABA uptake was associated with changes in transporter abundance, GAT‐1 protein levels were assessed by western blotting. CAFF treatment did not alter GAT‐1 immunoreactivity (Figure 4b). Collectively, these findings indicate that CAFF enhances GABA uptake through modulation of GAT‐1 transporter activity rather than by increasing its protein density.

FIGURE 4.

Evaluation of GABA uptake mechanism and GAT‐1 protein levels. (a) [³H]‐GABA uptake was measured in Group A (PN35) under basal conditions and in the presence of the selective GAT‐1 inhibitor NO‐711 (10 μM). Under basal conditions, CAFF increased GABA uptake compared with control (Control: 100.87 ± 4.76; n = 15; CAFF: 163.08 ± 13.44; n = 13). In the presence of NO‐711, uptake was decreased relative to basal conditions in both groups (Control + NO‐711: 65.50 ± 5.14; n = 16; CAFF + NO‐711: 99.75 ± 7.65; n = 4). In the caffeine treated groups, GABA uptake was significantly decreased in the presence of NO‐711. Two‐way ANOVA revealed significant main effects of treatment (F(1,44) = 23.51; p < 0.0001) and condition (F(1,44) = 22.45; p < 0.0001), with no significant interaction (F(1,44) = 1.89; p = 0.1765). (b) Western blot analysis of total GAT‐1 protein levels showed comparable expression between control and caffeine‐treated mice (CTRL: 1.002 ± 0.353; n = 6; CAFF 20 mg/kg: 1.343 ± 0.310; n = 6), with no significant difference detected (unpaired two‐tailed t‐test: T(10) = 0.73; p = 0.4835). Data are expressed as mean ± SEM. Statistical analyses were performed using ordinary two‐way ANOVA followed by Fisher's LSD multiple‐comparisons test to compare cell means within rows and columns (a), and unpaired two‐tailed Student's t‐test (b). Each dot represents one individual mice. Asterisks indicate statistical significance as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. If not shown, the comparison was ns. Median (center line), mean (+), SD (box), and min–max (whiskers).

3.4. The Effect of Caffeine on [ ³H]‐GABA Transport Involves the Adenosine A₁ Receptor

Antagonists likely alter the levels of their receptors. To evaluate the impact of CAFF upon A₁R and A_2AR, we analyzed the expression of these receptors in the FC of adolescent mice treated with saline or CAFF (20 mg/kg). Five days of CAFF treatment resulted in increased A₁R density in the FC (Figure 5a). In contrast, A_2AR density in the FC was not altered by CAFF treatment when compared with control mice (Figure 5b).

FIGURE 5.

Effects of caffeine on adenosine receptor expression and its role in GABA uptake. (a) Western blot analysis of total A₁R protein levels in FC from Group A (PN35) showed higher A₁R density in caffeine‐treated mice compared with controls (CTRL: 0.75 ± 0.23; n = 4; CAFF 20 mg/kg: 2.21 ± 0.40; n = 4; t(6) = 3.155; p = 0.0197). (b) Total A_2AR protein levels were similar between control and caffeine‐treated groups (CTRL: 1.00 ± 0.16; n = 4; CAFF 20 mg/kg: 0.91 ± 0.15; n = 4; t(6) = 0.3891; p = 0.7106). (c) [³H]‐GABA uptake was evaluated under basal conditions and in the presence of the selective A₁R agonist CHA (100 nM). Under basal conditions, caffeine increased GABA uptake relative to control (Control: 100.81 ± 4.45; n = 16; CAFF: 159.00 ± 12.89; n = 15). CHA decreased uptake in caffeine‐treated (CAFF: 95.11 ± 10.86; n = 9) but not in control slices (Control: 75.63 ± 8.11; n = 8). Two‐way ANOVA revealed significant main effects of treatment (F(1,44) = 18.44; p < 0.0001) and caffeine (F(1,44) = 14.02; p = 0.0005), with no significant interaction (F(1,44) = 3.48; p = 0.0688). Data are expressed as mean ± SEM. Statistical analyses were performed using unpaired two‐tailed Student's t tests (a and b), and ordinary two‐way ANOVA followed by Fisher's LSD multiple‐comparisons test was used to compare cell means within rows and columns (c). Each dot represents one individual mice. Asterisks indicate statistical significance as follows: *p < 0.05; ****p < 0.0001. If not shown, the comparison was ns. Median (center line), mean (+), SD (box), and min–max (whiskers).

Having detected an increase in A₁R density, our next objective was to test whether this receptor would be involved in the CAFF‐mediated increase of [³H]‐GABA uptake. To do so, we designed experiments comparing GABA uptake in tissues from mice treated/non treated with CAFF, but where the selective adenosine A₁R agonist CHA at a supramaximal concentration (100 nM) was added/not added to the assays. As before, [³H]‐GABA uptake was increased in tissues obtained from CAFF‐treated mice (Figure 5c, basal condition). The A₁R agonist CHA did not alter (p > 0.05) [³H]‐GABA uptake in control (non‐CAFF exposed) tissues. However, in tissues from CAFF‐treated mice (Figure 5c, filled bars), CHA significantly decreased (p < 0.0001) [³H]‐GABA uptake, blunting the CAFF‐induced increase in [³H]‐GABA uptake (Figure 5c). These results point towards the involvement of the A₁R upon the modulatory effect of the in vivo administration of CAFF upon GABA uptake.

3.5. Modulation of A₁R/AC/cAMP/PKA and A₁R/Ca²⁺/PKC Pathways After Caffeine Administration

Canonically, the A₁R is coupled to Gi protein and, in this way, activates the AC/cAMP/PKA pathway (Liu et al. 2019), however, it can also be coupled to Gq protein, which would thus lead to activation of the PLC/IP3, DAG/PKC pathway (Effendi et al. 2020). Considering that both pathways can modulate the expression of GATs and their ability to affect Ca²⁺ signaling (Hu and Quick 2008), our next objective was to assess whether CAFF administration was altering the second messengers associated with the A₁R.

To examine the involvement of the AC/cAMP/PKA pathway, cAMP accumulation was quantified and found to be unchanged by CAFF treatment (Figure 6a). Consistently, PKA protein levels were not altered in CAFF‐treated mice compared with controls (Figure 6b). In contrast, analysis of PKC signaling revealed an increase in phosphorylated PKC relative to total PKC following CAFF treatment (Figure 6c).

FIGURE 6.

Effects of caffeine on cAMP levels, PKA and PKC pathways, and Ca²⁺ signaling. (a) cAMP levels measured in FC from Group A (PN35) were similar between control and caffeine‐treated mice (CTRL: 243.1 ± 29.0; n = 4; CAFF 20 mg/kg: 219.5 ± 15.5; n = 5; t(7) = 0.7599; p = 0.4721). (b) Total PKA protein levels were unchanged by caffeine treatment (CTRL: 1.00 ± 0.06; n = 6; CAFF 20 mg/kg: 1.02 ± 0.20; n = 6; t(10) = 0.0868; p = 0.9325). (c) The ratio of phosphorylated PKC (pPKC) to total PKC was increased in caffeine‐treated mice compared to controls (CTRL: 0.91 ± 0.04; n = 6; CAFF 20 mg/kg: 1.20 ± 0.08; n = 6; t(10) = 3.158; p = 0.0102). (d) [³H]‐GABA uptake was evaluated under basal conditions and in the presence of the PKC inhibitor Gö 6983 (100 nM), comparing control and caffeine‐treated tissues. Under basal conditions, caffeine increased GABA uptake relative to control (Control: 100.88 ± 5.22; n = 16; CAFF: 165.06 ± 10.88; n = 17). Gö 6983 didn't modify uptake in control (Control: 135.25 ± 8.60; n = 4), but in its presence the facilitation by caffeine was no longer evident (CAFF: 136.50 ± 5.63; n = 4). Two‐way ANOVA showed significant effects of caffeine (F(1,37) = 6.374; p = 0.0160) and a significant interaction (F(1,37) = 5.896; p = 0.0202), with no main effect of treatment (F(1,37) = 0.0504; p = 0.8237). (e, f) Analysis of Ca²⁺ transient frequency revealed higher event frequency in caffeine‐treated cells compared with Gö 6983 alone and with caffeine plus Gö 6983 (Caffeine: 0.524 ± 0.057; n = 76; Gö 6983: 0.171 ± 0.049; n = 42; CAFF + Gö 6983: 0.208 ± 0.046; n = 49; Kruskal–Wallis statistic = 37.69; p < 0.0001). Statistical analyses were performed using unpaired two‐tailed Student's t‐tests (a–c), ordinary two‐way ANOVA followed by Fisher's LSD multiple‐comparisons test was used to compare cell means within rows and columns (d), and Kruskal–Wallis tests followed by Dunn's multiple‐comparisons tests (e). Each dot represents one individual mice (a–d) or one analyzed cell/event (e). Asterisks indicate statistical significance as follows: *p < 0.05; ***p < 0.001; ****p < 0.0001. If not shown, the comparison was ns. Median (center line), mean (+), SD (box), and min–max (whiskers).

To better understand how the Ca²⁺/PKC pathway could be involved in the CAFF‐induced increase of [³H]‐GABA uptake, we compared GABA uptake in tissues from mice that had been exposed to CAFF, but when the PKC inhibitor Gö 6983 (100 nM) was added/not added to the assay media. As observed previously, CAFF increased [³H]‐GABA uptake relative to basal conditions (Figure 6d). Gö 6983 alone did not alter GABA uptake; however, its presence prevented the increase in [³H]‐GABA uptake induced by CAFF (Figure 6d). These results indicate that the CAFF‐induced enhancement of GABA uptake depends on PKC signaling.

Previous work from our lab showed that in rat cortical neuronal cultures, caffeine increases the frequency of Ca²⁺ transients, an effect mediated by A₁R (Martins et al. 2020). In avian retina (Souto et al. 2023), and rat astrocytic cultures (Jacob et al. 2014), GABA uptake can be affected by Ca²⁺ signaling. We therefore hypothesized that in cortical neuronal cultures, the caffeine‐induced enhancement of GABA uptake could also be related to its ability to modulate Ca²⁺ signaling. If that hypothesis holds true, the transducing pathways for both effects should be identical; that is, Gö 6983 should prevent CAFF‐induced increases in Ca²⁺ transients as it did for GABA uptake. To test that hypothesis, we performed Ca²⁺ signaling assays in cultured neurons and tested if the effect of caffeine (200 μM in the assay medium) was affected by Gö 6983 (100 nM). In the absence of stimulation, calcium signaling in neurons was virtually absent (Figure 6a). Addition of CAFF to the incubation medium clearly induced Ca²⁺ transients in the culture (Figure 6e,f). In the presence of Gö 6983 alone, Ca²⁺ signaling was also virtually absent (Figure 6f). Importantly, under such condition, CAFF was unable to trigger calcium transients. Indeed, the frequency of events induced by CAFF in the absence of Gö 6983 was markedly different (p < 0.0001) from that in the presence of the PKC inhibitor, thus indicating that PKC activity contributes to the CAFF‐induced modulation of neuronal Ca²⁺ dynamics.

Overall, these results suggest that caffeine affects GABA uptake and intracellular Ca²⁺ dynamics through a similar mechanism, which requires PKC activity, likely the PLC/IP3, DAG/PKC pathway.

4. Discussion

The data herein reported demonstrate that CAFF intake modulates GABA homeostasis without affecting growth parameters such as body mass and snout/tail length. Caffeine's effects likely reflect A₁R antagonism, since it increased A₁R expression and thereby unmasked an inhibitory effect of CHA on GABA uptake. CAFF modulation of GABA homeostasis via A₁R is linked to the PKC pathway rather than the AC/PKA, as the treatment with CAFF increased PKC phosphorylation, and inhibition of PKC with Gö 6983 prevented the CAFF‐induced increase in GABA uptake. The PKC pathway is also necessary for the CAFF‐induced increase in Ca²⁺ transients in neuronal cells, highlighting similarity between both effects of caffeine.

4.1. Caffeine Alters GABAergic Transmission: A Critical Role for Adenosine A₁ Receptor Overexpression

We performed a dose–response analysis by administering daily injections of 10, 20, or 40 mg/kg/day of CAFF over a 5‐day period. From a concentration of 20 mg/kg, CAFF modulated both the release and uptake of GABA. According to the literature, a daily subcutaneous dose of 20 mg/kg of CAFF in mice leads to ≈35 μM CAFF in the serum and ≈25 μM CAFF in the brain parenchyma, which is within the range of values detected in the serum (5–70 μM) in human moderate coffee consumers (Costenla et al. 2010; Fredholm 1995; Lopes et al. 2019; Ongini et al. 1999; Serapiao‐Moraes et al. 2013). Based on this information and our results, a dose of 20 mg/kg was selected for most of the experiments. The 5‐day exposure period was chosen because it fits within the range (between 3 and 10) classified as subchronic treatment (Dall'Igna et al. 2007; De Oliveira et al. 2005; Esmaili et al. 2021), whereas longer exposures (≥ 10 days) have been considered chronic (Rezvani et al. 2013). Thus, a 5‐day regimen represents an intermediate subchronic window that allows for the assessment of sustained neurochemical adaptations without progressing to chronic exposure, while also coinciding with the critical developmental window that corresponds to the onset of adolescence in rodents (Hueston et al. 2017; Koss and Frick 2017; Silveri 2014). We deliberately avoided extending the study into adulthood, since overlapping exposure across different developmental time windows may produce opposing effects (Abreu et al. 2011; Pires et al. 2010), potentially blurring data interpretation.

Tissue extraction in our study was done at 1 h after the last administration of CAFF, thus likely coincident with peak plasma concentrations of CAFF (1 h), and well before its half‐life (3–7 h) (Arnaud 2011; Gurley et al. 2015; Jabbar and Hanly 2013; McLellan et al. 2016). Previous work in rodent retinas revealed that the effect of CAFF on [³H]‐D‐aspartate uptake was reversible after elimination (de Freitas et al. 2016). Reversibility of the effect of CAFF could also be concluded from our data since the effect vanished after 5 days' withdrawal. Remarkably, a single exposure to CAFF 1 h before the experiments was sufficient to reinstall the increase [³H]‐GABA uptake. Together, this highlights a quick onset and rapid offset of the ability of CAFF to modulate GABA homeostasis.

GABA uptake is mediated by four GAT subtypes in mice (Eskandari et al. 2017; Guastella et al. 1990), with GAT‐1 highly expressed in the rat cortex (Conti et al. 2004; Scimemi 2014a) and localized in neurons and glial cells (Fattorini, Catalano, et al. 2020; Fattorini, Melone, and Conti 2020; Melone et al. 2015). In neurons, GAT‐1 is found in presynaptic areas and axonal terminals of asymmetric synapses in the cortex (Durkin et al. 1995; Scimemi 2014b; Zhou and Danbolt 2013). Previous studies in adolescent mice revealed the expression and functionality of GAT‐1 in the FC (Martins et al. 2018). We now expanded that finding towards their neuromodulation by CAFF and showed that CAFF intake modulates the activity (but not the expression) of GAT‐1 at the FC.

Adenosine receptors are differentially expressed in the CNS, with A₁R more prevalent than A_2AR in the FC (Ribeiro et al. 2002). Exposure to agonists is known to decrease A₁R expression and affinity, while antagonists such as CAFF induce A₁R upregulation in the hippocampus, cerebellum, and cortex (Hettinger‐Smith et al. 1996; Johansson et al. 1997; Sousa et al. 2011). Consistent with the literature, both A₁R and A_2AR receptors were expressed in the FC of adolescent mice, but CAFF selectively increased A₁R expression without affecting A_2AR expression. GABA release is facilitated by A_2AR activation at the hippocampus (Cunha and Ribeiro 2000; Rombo et al. 2015), but inhibited in the striatum (Kirk and Richardson 1994; Kurokawa et al. 1994). A₁R agonists applied acutely to the perfusion system inhibit [³H]‐GABA release in hippocampal slices (Saransaari and Oja 2005), but this effect cannot be detected when isolated nerve terminals were used as preparation (Cunha and Ribeiro 2000). Our data showing that in FC slices CAFF exposure in vivo increases GABA release, induces overexpression of A₁R and unmasks an inhibitory action of an A₁R agonist, is consistent with previous findings in hippocampal slices (Saransaari and Oja 2005) as well as in the avian retina, where CAFF enhances GABA release, through a mechanism that involves A₁R (Borges‐Martins et al. 2019). In the retina, the enhanced release of GABA induced by CAFF is coupled to GAT‐1 activity (Ferreira et al. 2014). Whether the now reported CAFF‐induced enhancement of GABA release occurs through reversal of GAT‐1 activity, or whether it occurs in astrocytes or neurons is unknown. However, taking into consideration that phasic GABAergic transmission at the hippocampus, an area with many similarities with the FC in what concerns adenosinergic neuromodulation, is insensitive to A₁Rs, (Lambert and Teyler 1991; Rombo et al. 2016; Yoon and Rothman 1991), we are tempted to speculate that the presently observed enhanced release results from the reversal of GAT‐1 activity rather than from an enhancement of exocytotic release of GABA from nerve terminals.

Previous studies indicate that adenosine receptors modulate GABA uptake, with A_2AR reducing GAT‐1 mediated transport in the globus pallidus (Gonzalez et al. 2006) and facilitating GAT‐1 mediated transport in cortical synaptosomes (Cristóvão‐Ferreira et al. 2009), whereas A₁R/A_2AR heteromers have a biphasic action in cortical astrocytes (Cristovao‐Ferreira et al. 2013). In our study, CHA alone did not cause a statistically significant effect, though the raw data indicated a ≈25% reduction compared to the control, thus consistent with a marginal inhibitory action. Importantly, CHA blunted the CAFF‐induced increases in GABA uptake, restoring basal levels, confirming A₁R involvement in the mechanism of action of CAFF.

4.2. A₁ Adenosine Receptor Modulating Ca²⁺ Signaling and PKC Pathway

Previous studies from our group showed a dual effect of CAFF on A₁R and A_2AR receptors, with A₁R mediating the CAFF‐induced increase in the frequency of Ca²⁺ transients via the A₁R (Martins et al. 2020). We now show that the CAFF‐induced facilitation of Ca²⁺ signaling and of GABA uptake is both abolished by the PKC blocker Gö 6983, corroborating previous findings and reinforcing the importance of the Ca²⁺/PKC pathway in modulating GABAergic homeostasis and GAT‐1 activity in the forebrain (Souto et al. 2023). We did not detect alterations in cAMP or PKA levels in FC slices from mice treated with CAFF, suggesting that the cAMP/PKA‐mediated mechanisms are not involved in the modulatory action of CAFF upon GABA homeostasis in this brain area, in clear contrast with what occurs in the striatum (Kubrusly et al. 2021) or retina (Borges‐Martins et al. 2019), where the canonical cAMP/PKA‐dependent pathway is involved in the A1R‐mediated modulation of GABA transport.

5. Conclusion

Based on the findings presented here, we propose a unified mechanistic model of caffeine action on GABAergic neurotransmission is schematized in Figure 7. As our data show, caffeine modulates GABAergic signaling primarily through upregulation of adenosine A₁R, which leads to the engagement of PKC‐dependent pathways and enhanced Ca²⁺ signaling. This is likely due to activation of phospholipase C (PLC) and consequent production of diacylglycerol (DAG) and inositol triphosphate (IP₃), followed by the release of Ca²⁺ from the endoplasmic reticulum (ER). The elevation of intracellular Ca²⁺ levels may contribute both to the increased release of GABA and modulation of GABA uptake via modulation of GAT‐1 activity. Caffeine thus exerts coordinated and bidirectional control of GABAergic neurotransmission by simultaneously influencing its release and reuptake, which suggests an overall increase in GABA turnover. Future studies employing selective manipulation of GABA transporters and release mechanisms will be required to clarify if the observed increase in uptake is causally upstream, compensatory, or secondary to enhanced release.

FIGURE 7.

Schematic summary of the proposed signaling pathway underlying the effect of caffeine. Caffeine actions herein described primarily result from antagonism of A₁R receptors, which, rather than coupling to the canonical G_i/adenylyl cyclase‐cAMP/PKA pathway, may be associated with a G_q protein—a non‐canonical coupling widely described in the literature. This alternate signaling pathway involves the hydrolysis of PIP₂ to generate inositol trisphosphate (IP₃) and diacylglycerol (DAG). IP₃ facilitates the release of Ca²⁺ from intracellular stores, while DAG activates PKC, as evidenced by the increase of PKC phosphorylation and Ca²⁺ events. The activated PKC appears to modulate GAT‐1 activity, leading to an increase in [³H]‐GABA uptake. In parallel, the elevation of intracellular Ca²⁺ and enhanced PKC signaling promote vesicular GABA release via a canonical exocytotic mechanism. A₁R, adenosine A₁ Receptor; DAG, diacylglycerol; ER, Endoplasmic reticulum; IP3, Inositol 1,4,5‐trisphosphate; PIP2, Phosphatidylinositol 4,5‐bisphosphate; PKC, Protein Kinase C; PLC, Phospholipase C.

In summary, altogether, the data herein reported support the conclusion that caffeine intake, through up‐regulation of A₁Rs, affects GABAergic tone in the FC through mechanisms dependent on PKC signaling and intracellular Ca²⁺ dynamics, rather than cAMP signaling.

Author Contributions

Vladimir Pedro Peralva Borges‐Martins: methodology, formal analysis, writing – original draft, writing – review and editing. Joana Gonçalves‐Ribeiro: methodology, investigation, writing – review and editing. Carlos Henrique de Carvalho Teixeira: methodology, investigation. Ana M. Sebastião: conceptualization, formal analysis, resources, writing – review and editing, supervision, project administration, funding acquisition. Regina Célia Cussa Kubrusly: conceptualization, formal analysis, resources, supervision, project administration, funding acquisition, writing – review and editing. Sandra H. Vaz: methodology, investigation, writing – review and editing, formal analysis, supervision, funding acquisition. Ricardo A. de Melo Reis: formal analysis, resources, writing – review and editing, supervision, funding acquisition. Robertta Silva Martins: conceptualization, methodology, formal analysis, investigation, writing – original draft, writing – review and editing, visualization.

Funding

This work was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES, Finance Code 001), by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brazil), and by the Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ, Brazil) through research scholarships. Additional support was provided by the Fundação para a Ciência e para a Tecnologia (FCT, Portugal; Grant PD/BD/150342/2019 to JG‐R) and by the International Society for Neurochemistry (Career Development Grant 2021 to Sandra H. Vaz).

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Normality assessment of datasets used in the study using the Shapiro–Wilk test.

Table S2: effect of sex on uptake and release assays were assessed using two‐way ANOVA.

Figure S1: Full, uncropped chemiluminescent western blot membrane showing GAT‐1 (~58 kDa) and β‐tubulin in control and caffeine (20 mg/kg) samples.

Figure S2: Full, uncropped chemiluminescent western blot membrane showing A1R (~39 kDa), A2AR (~43 kDa), and β‐tubulin in control and caffeine (20 mg/kg) samples.

Figure S3: Full, uncropped chemiluminescent western blot membrane showing PKA (~42 kDa) and β‐tubulin in control and caffeine (20 mg/kg) samples.

Figure S4: Full, uncropped chemiluminescent Western blot membrane showing phosphorylated PKC (pPKC; ~80 kDa) and total PKC (~76 kDa) in control and caffeine (20 mg/kg) samples.

Data Availability Statement

The data are available from the corresponding author upon reasonable request.