Caffeine Functions by Inhibiting Dorsal and Ventral Hippocampal Adenosine 2A Receptors to Modulate Memory and Anxiety, Respectively

As a nonspecific antagonist of the adenosine A2A receptor (A2AR), caffeine enhances learning and improves memory impairment. Simultaneously, the consumption of caffeine correlates with a feeling of anxiety. The hippocampus is functionally differentiated along its dorsal/ventral axis and plays a crucial role both in memory and anxiety. Whether caffeine exerts its regulation by inhibiting A2ARs in different subregions of the hippocampus is still unknown. In the present study, we found that after chronic intake of drinking water containing caffeine (1 g/L, 3 weeks), mice exhibited aggravated anxiety-like behavior and enhanced memory function. Tissue-specific, functional disruption of dorsal hippocampal A2ARs by the CRE-LoxP system prevented the memory-enhancing effect of caffeine, while selective disruption of ventral hippocampal A2ARs blocked the impact of caffeine on anxiety. These results, together with the enhanced memory of dorsal hippocampus A2AR knockout mice and greater anxiety-like behavior of ventral hippocampus A2AR knockout mice without caffeine, indicates a dissociation between the roles of ventral and dorsal hippocampal A2A receptors in caffeine’s effects on anxiety-like and memory-related behavioral measures, respectively. Furthermore, optogenetic activation of dorsal or ventral hippocampal A2ARs reversed the behavioral alterations caused by drinking caffeine, leading to impaired memory or decreased anxiety-like behaviors, respectively. Taken together, our findings suggest that the memory- and anxiety-enhancing effects of caffeine are related to the differential effects of inhibiting A2ARs in the dorsal and ventral hippocampus, respectively.


INTRODUCTION
As the most widely consumed psychotropic substance and a component of the most popular beverages, caffeine is used to counteract performance impairments associated with sleep loss (Irwin et al., 2020).The consumption of caffeine attenuates memory impairments associated with aging and Alzheimer's disease (AD) (Solfrizzi et al., 2015;Jacobson et al., 2020;Londzin, et al., 2021)and enhances memory in healthy humans (Borota et al., 2014;Irwin et al., 2020). However, controlled studies have confirmed that caffeine produces negative effects such as increased anxiety (Garcia and Salloum, 2015;Khurana and Bansal, 2019). Similarly, mouse studies have also shown that caffeine ingestion improves memory and induces anxiety (Xu and Reichelt, 2018). The main molecular targets of caffeine in the brain are adenosine receptors, including the inhibitory A 1 receptor (A 1 R) and facilitatory A 2A receptor (A 2A R) (Ikram et al., 2020). From results obtained in animal models, it was further concluded that the effect of caffeine on both anxiety and memory performance was mimicked by the selective blockade of A2A, but not of A1 receptor (Kaster et al., 2015;Machado et al., 2017).
The hippocampus belongs to the limbic system and plays an important role in memory, spatial navigation and emotion. The hippocampus can be divided into two segments along its longitudinal axis: the dorsal and ventral subregions. Early anatomical studies demonstrated differences between the afferent and efferent nerve projections of the dorsal hippocampus (dHPC) and ventral hippocampus (vHPC) (Swanson and Cowan, 1977). According to a current consensus, the role played by the most dorsally located hippocampal segment is on cognitive operations like spatial navigation, while internally monitoring functions related to emotionality are taken on by the ventral segment of the hippocampus (Trompoukis and Papatheodoropoulos, 2020). Moreover, some studies have found that caffeine reverts memory impairment in a depression-prone mouse strain with upregulation of adenosine A 2A receptors in the hippocampus (Machado et al., 2017), while selective A 2A R knockout in the forebrain region (striatum, hippocampus, and cortex) induces anxiety-like behavior (Wei et al., 2014). All the evidence suggests that the hippocampal adenosine A 2A receptor is likely to emerge as an important receptor in the regulation of caffeine on memory and anxiety. However, whether the different effects of caffeine on overall function are derived from its action on adenosine A 2A receptors in different subregions of the hippocampus is still unclear.
To explore the above question, we first verified that overall A 2A R knockout reduced the effect of caffeine on mice. Furthermore, we found that localized knockout of dHPC A 2A Rs reduced the regulation of caffeine on memory but did not affect anxiety, whereas localized knockout of vHPC A 2A Rs reduced the regulation of caffeine on anxiety without affecting memory. To confirm the memory-enhancing effects of dHPC A2AR knockout, we assessed the level of SNAP-25, a synaptic marker reflecting synapse formation and remodeling (Batista et al., 2017). To confirm the anxiogenic effects of vHPC A2AR knockout, we assessed the level of SNAP-25, a glutamatergic-selective marker and labels excitatory glutamatergic neurons (Zheng et al., 2015;Martineau et al., 2017

Animals
Adult male C57BL/6 mice (weighing 25-30 g, 11-13 weeks old) were purchased and used in our study. Global A 2A R knockout mice were established on a C57BL/6 background as our previously described (Zeng et al., 2020), and their littermates were used as wild-type (WT) mice in this experiment. Mice with a 'floxed' adenosine A 2A R gene (A 2A flox/flox mice) created by insertion of loxP sequences into introns flanking an exon of the A 2A R gene (Bastia et al., 2005;Yu et al., 2009) were provided by Dr. Chen. The experimental procedures were performed in accordance with the guidelines of the Animal Ethical and Welfare Committee of the Army Medical University.
For optogenetic manipulations of neuronal A 2A R, a chimeric rhodopsin-A 2A R protein (optoA 2A R) was developed by replacing the intracellular domain of rhodopsin with that of A 2A R as described in our previous research (Li et al., 2015). Extracellular adenosine or caffeine cannot react with optoA 2A R. Similarly, pAAV-CaMKIIa-optoA 2A R-mCherry virus and pAAV-CaMKIIa-MCS-mCherry-3FLAG virus (control virus) were constructed and supplied by OBiO Technology (Shanghai) Corp. Ltd. All viruses were used at titers of~4-8*10 12 vg/ml.

Drug Treatments
In the first part of our experiment, mice (knockout or wild-type) were randomly allocated to two groups: a control group provided with drinking water without drug and a treatment group provided with drinking water containing caffeine (1 g/L, Sigma) starting 3 weeks before behavioral tests. This dose and schedule of administration of caffeine was chosen since it was sufficient to improve memory and increase anxiety according to a previous study (Kaster et al., 2015).
In the second part of the study, A 2A flox/flox mice were injected with CRE virus or EGFP virus. Half the mice in each group were then provided with caffeine-containing drinking water starting 3 weeks before behavioral tests, and the other half were provided with normal drinking water.
In the last part, wild-type mice were randomized into two groups: an experimental group injected with the optoA 2A R virus and a control group were injected with the mCherry virus. Then, all mice underwent optical fiber implantation and were treated with caffeine starting 3 weeks before optogenetic manipulation.
In all experiments, the drinking water with or without caffeine were provided every day, until the mice were killed.
To express optoA 2A R in hippocampal neurons, we injected 1.5 μL pAAV-CaMKIIa-OptoA 2A (A400S)-mCherry virus (or control virus) into the left dorsal or ventral subregion (viral injection coordinates are described above). Four weeks after injection, a 200 μm optical fiber (Shanghai Fiblaser Technology Co., Ltd.) was implanted in the same coordinates. Via a patch cable, the optical fiber was connected to a 473 nm DPSS laser (100 mW; Shanghai Laser and Optics Century). The power density at the fiber tip was approximately 5 mW/mm 2 , and light was delivered with a 50 m pulse width (10Hz). For the optogenetic experiment, mice were habituated with an optical fiber connected to the optical patch cable without laser stimulation for 30 min before the behavior tests, and optical stimulation was delivered specifically according to the different behavioral tests (see below). For immunofluorescence analyses, mice were euthanized by cervical dislocation following 10 min of optical stimulation.

Behavioral Experiments
To eliminate the acute effects of caffeine while preserving its longterm chronic effects, all behavioral tests were performed within 24-48 h after 3-week caffeine treatment.
Open Field Test: Mice were placed in a square chamber (40 cm × 40 cm × 40 cm length-width-height) with a dimly lit (approximately 65 lux). The center zone was a 20 × 20 cm square. Mice were allowed to freely explore the environment for 5 min, while activities were recorded and analyzed by EthoVision XT behavioral tracking software (Noldus Information Technology Inc.). For optogenetic manipulation, light stimulation was maintained for the entire 5-min duration of the open-field test.
Elevated Plus Maze (EPM) Test: The elevated plus maze had open and closed arms (30 cm × 8.5 cm length-width, 20 cm tall closed arms) that extended from a central platform (8.5 × 8.5 cm). The maze was placed at a height of 58 cm and the room light was about 65 lux. The mouse was placed in the central platform of the maze facing an open arm and were allowed to freely explore the EPM for 5 min, and their activities were recorded and analyzed by EthoVision software (Noldus Information Technology Inc.). For optogenetic experiments, light stimulation was used for a total period of 5 min while mice explored the EPM.
Novel Object Recognition (NOR) Test: NOR test was carried out in a home-cage arena (45 cm × 45 cm × 50 cm length-widthheight) which they were allowed to freely explore for 10 min on the previous day. In the training trial on the next day, mice were presented with two same objects (a red cube, 2.8 cm × 2.8 cm × 2.8 cm) placed in opposite corners for 10 min. The exploration of the objects, defined by mice showing investigative behaviors (head orientation or sniffing) or playing within 1 cm around the object, was measured. In the testing trial (24 h later), one of the identical objects was changed for a novel object (a yellow pyramid, height 3.0 cm, base 2.8 cm × 2.8 cm), and the animals were left in the cage for 10 min. The exploration time for the familiar and the novel object during the test phase was recorded with EthoVision software. For the optogenetic experiment, light stimuli were only delivered during the testing trial.
Y-Maze Test: Y-maze test was carried out in a gray maze formed by three arms (28 cm × 8.5 cm × 20 cm length-widthheight, 58 cm above the ground) so as to form a Y shape. Each mouse was first allowed to explore the maze for 5 min while one arm was blocked (acquisition phase). After 2 h, mice had access to all three arms for a 5 min period (retrieval phase). During the second period, the time spent in each arm was measured by a video-tracking system (Noldus Information Technology Inc.). For the optogenetic experiment, light stimuli were only presented during the retrieval trial.

Statistical Analysis
Results are expressed as the means ± SEM. All semi-quantitative assessments of histological staining were made by a single investigator blinded to the genotype and treatment of the experimental animals. Sample size was chosen according to previous reports and our pre-experiments. Two-way analyses of variance (ANOVAs) were used to assess the effects of caffeine, gene manipulation and the caffeine × gene manipulation interaction in the A 2A R total (or region-specific) knockout mice. Two-way ANOVAs were used to assess the effects of optoA 2A R virus, light stimulation and the optoA 2A R virus × light stimulation interaction in the optoA 2A R mice. A value of p < .05 was considered statistically significant.

Global A 2A R Knockout Blocked the Effect of Caffeine on Anxiety and Memory
The open-field test and EPM test were used to evaluate anxiety-like behaviors. Chronic caffeine consumption (3 weeks) significantly reduced the time ( Figure 1A, for caffeine, F (1,28) = 5.741, p < .01; for knockout, F (1,28) = 16.341, p < .01; for caffeine × knockout interaction, F (3,28) = 4.124, p < .01) and distance ( Figure 1B, for caffeine, F (1,28) = 56.114, p < .001; for knockout, F (1,28) = 23.554, p < .01; for caffeine × knockout interaction, F (3,28) = 4.451, p < .01) in the center of the open field, and had no significant effect on the total distance (Supplementary Figure S1A, for caffeine, F (1,28) = 0.372, p = .547; for knockout, F (1,28) = .001, p = .975; for caffeine × knockout interaction, F (3,28) = 0.985, p = 0.331). In the EPM, caffeine treatment decreased the percentage of time ( Figure 1D, for caffeine, F (1,28) = 9.125, p < .01; for knockout, F (1,28) = 21.934, p < .01; for caffeine × knockout interaction, F (3,28) = 23.344, p < .01) and number of entries ( Figure 1C, for caffeine, F (1,28) = 14.624, p < .01; for knockout, F (1,28) = 21.679, p < .01; for caffeine × knockout interaction F (3,28) = 8.824, p < .01) in the open arms which confirmed the open-field test results. We next examined object recognition to evaluate the role of caffeine in memory formation. The caffeine-treated mice showed greater exploration of the novel object over the familiar object ( Figure 1E, for caffeine, F (1,28) = 12.347, p < .01; for knockout, F (1,28) = 15.472, p < .01; for caffeine × knockout interaction, F (3,28) = 14.147, p < .01). The spatial recognition memory of mice was probed using a two-trial Y-maze paradigm. Likewise, caffeine-treated mice showed increased time in the novel arm compared with control mice ( Figure 1F, for caffeine, F (1,28) = 22.712, p < .01; for knockout, F (1,28) = 15.755, p < .01; for caffeine × knockout interaction, F (3,28) = 21.457, p < .01). This finding suggests that chronic caffeine consumption for 3 weeks enhanced memory performance in mice. However, chronic caffeine consumption did not alter anxiety-like behaviors in global A 2A R knockout mice. There were no significant differences between caffeine-treated mice (caffeine + knockout) and control mice (water + knockout) in either the open-field test or the EPM test ( Figure 1C and Figure 1D). At the same time, A 2A R knockout blocked the effect of caffeine in the NOR test and Y-maze test ( Figures  1E,F), suggesting that the effect of caffeine on memory was prohibited by A 2A R inactivation. It is worth noting that in global A 2A R knockout mice, there were baseline differences in behaviors that are independent of caffeine ( Figures 1A-F). But the costimulation with caffeine and A 2A R knockout did not produce additive effects which reflected the interaction between caffeine-A 2A R function.  dHPC A 2A R knockout did not affect the caffeine-induced anxiety-like behavior in mice. However, after inactivation of dorsal hippocampal A 2A Rs, there was no significant difference in behavior in the NOR test ( Figure 2E, for caffeine, F (1,28) = 1.413, p < .01; for CRE, F (1,28) = 19.413, p < .01; for caffeine × CRE interaction, F (3,28) = 15.439, p < .01) or Y-maze test ( Figure 2F, for caffeine, F (1,28) = 17.435, p < .01; for CRE, F (1,28) = 10.235, p < 0.01; for caffeine × CRE interaction, F (3,28) = 5.636, p < .01) between the caffeinetreated group and the control group, indicating that dHPC A 2A R knockout eliminated the effect of caffeine on memory. Immunofluorescence results showed that the level of SNAP-25 in the pAAV-EGFP group given caffeine (EGFP + caffeine) was significantly higher than that in the group given water (EGFP + water). However, caffeine did not alter the expression of SNAP-25 in the pAAV-CRE group (Figure 3). Interestingly, the NOR test and Y-maze test showed that dorsal hippocampal A 2A R knockout alone improved memory similar to chronic caffeine consumption ( Figures 2E,F), and these behavioral changes were in accordance with the trend observed in SNAP-25 expression (Figure 3). This evidence suggests that the effect of caffeine on memory in mice is derived from its action on dorsal hippocampal A 2A Rs.

Brain Region-specific Activation of dHPC/ vHPC A 2A Rs Reversed the Regulation of Caffeine on Memory and Anxiety
Six weeks after dHPC/vHPC injection of pAAV-CaMKIIa-optoA 2A R-mCherry ( Figures 6A,7A), mice were exposed to caffeine in drinking water for 3 weeks. Light stimulation for 5 min significantly increased the levels of c-Fos in pAAV-optoA 2A R mice but not in pAAV-mCherry mice ( Figure 6D), indicating that illumination was sufficient to activate the optoA 2A R signaling pathway.

DISCUSSION
In the present study, we demonstrated that chronic caffeine consumption (1 g/L, 3 weeks) increased anxiety-like behavior, as assessed by the open-field test and EPM, and enhanced memory, as reflected in the NOR and Y-maze. Consistent with our results, caffeine has been shown to reverse cognitive impairments in aging and Alzheimer's disease (AD) (Solfrizzi et al., 2015;Londzin et al., 2021) and lead to depression and anxiety-like behaviors (Richards and Smith, 2015;Mikkelsen et al., 2017). Acute caffeine consumption is widely used to counteract mood and performance impairments associated with sleep loss (Irwin et al., 2020). To eliminate the acute effects of caffeine while preserving its long-term chronic effects, all behavioral tests were performed after 3-week caffeine treatment. Chronic but not acute treatment with caffeine is considered to induce those behavioral alterations. As caffeine is most widely consumed all over the world, our exploration of chronic caffeine consumption is more meaningful. Moreover, caffeine seemed to have no effects on anxiety and memory in our total A 2A R knockout mice. Statistical analysis also showed the strong interactions between caffeine and A 2A R function, which suggests that caffeine functions through the antagonism of A 2A Rs. The main targets for caffeine in the brain are the inhibitory A 1 R and the facilitatory A 2A R (Fredholm et al., 2005). Which is different from antiinflammatory effect of A 1 R against noxious brain conditions (Martins et al., 2015), A 2A R may play a more important role in memory and anxiety.
Many evidence suggests that pathological brain conditions associated with memory impairment are accompanied by a local increase of the extracellular levels of adenosine (Chen et al., 2013) and an up-regulation and aberrant signaling of the brain A 2A R  (Cunha and Agostinho, 2010;Chen et al., 2013). Thus, A 2A R blockade could attenuate the impairment of brain function (Cunha, 2016) and memory function in particular (Cunha and Agostinho, 2010). However, whether A 2A R blockade increases learn and memory in healthy animals remains contentious. Several groups reported that A 2A R blockade and caffeine did not increase memory performance in control rodents, whereas stressed mice displayed increased memory performance upon caffeine consumption or upon blocking A 2A R (Prediger et al., 2005;Kaster et al., 2015;Laurent et al., 2016;Carvalho et al., 2019). But genetic KO studies have shown that inactivation of A 2A R is sufficient to improve memory in healthy animals (Zhou et al., 2009;Wei et al., 2011). The mechanism by which genetic inactivation of A 2A Rs strengthens memory is not clear. loss of A 2A Rs may impact cortical function through neuronal networks such as basal ganglia loop (Zhou et al., 2009) or potentiate striatal dopaminergic signaling via D 2 Rs to produce the memory enhancement (Wei et al., 2011). While studies have shown that intraperitoneal injection of the A 2A R agonist CGS21680 produces strong anxiety-like behavior (El Yacoubi et al., 2000), in our study, the A 2A R knockout mice showed greater anxiety and enhanced memory compared to wildtype littermates. Considering that A 2A R is widely distributed in the brain, the overall effect of drinking caffeine is likely be a superposition of its antagonism towards A 2A R in multiple brain regions. Therefore, A 2A R in a particular brain subregion may be responsible for caffeine's effects on anxiety-like and memoryrelated behaviors.
Another major advance provided by this study is the dissociation of the roles of the ventral and dorsal hippocampal A 2A receptors in caffeine's effects on anxiety-like and memory-related behavioral measures, respectively, which are also consistent with the reported roles of the dorsal and ventral hippocampus more generally (Fanselow and Dong, 2010;McHugh et al., 2011;Schumacher et al., 2018). We discovered that knocking out dorsal hippocampal A 2A Rs blocked the memory-enhancing effects of caffeine without affecting its anxiogenic effects, whereas knocking out ventral hippocampal A 2A Rs did not affect the memory-enhancing effects of caffeine but blocked its anxiogenic effects. In addition, there were baseline differences in the behaviors of site-specific A 2A R knockout mice, and statistical analysis confirmed caffeine-A 2A R interactions. These results indicate that caffeine modulates memory by inhibiting dorsal hippocampal A 2A R and modulates anxiety by acting through ventral hippocampal A 2A R.
To confirm the memory-enhancing effects dHPC A 2A R knockout, we assessed the level of SNAP-25. Our study showed that inactivation of dHPC A 2A R upregulated the density of synaptic proteins, consistent with a previous study that showed that activation of hippocampal A 2A R is sufficient to attenuate synaptic plasticity and further impair memory (Li et al., 2015). In recent years, synaptic density and synaptic connections in the hippocampus have been associated with learning and memory (Benito and Barco, 2010;Latina et al., 2017), and the overexpression of adenosine receptors revealed a hippocampal LTD-to-LTP shift to impair synaptic plasticity (Temido-Ferreira et al., 2018). Thus, chronic caffeine consumption may affect synaptic function to enhance memory by inhibiting dHPC A 2A Rs. However, the specific mechanism remains to be further confirmed.
One of the most important hypotheses of anxiety disorder is inhibition/excitation imbalance (Colic et al., 2018). As vGluT1 is FIGURE 5 | Selective deletion of vHPC A 2A Rs increases vGluT1 expression. Caffeine markedly induced vGluT1 expression in mice transfected with pAAV-EGFP but not in mice transfected with pAAV-CRE. pAAV-CRE transfection increased vGluT1 levels in mice exposed to water alone as well. Representative images (left) and Data (right) are presented as the mean ± SEM; n = 27 fields per group (3 fields per section, three sections per mouse, three mice per group). Scale bar = 30 μm.
Frontiers in Pharmacology | www.frontiersin.org February 2022 | Volume 13 | Article 807330 8 a glutamatergic-selective marker and labels excitatory glutamatergic neurons (Zheng et al., 2015;Martineau et al., 2017), upregulation of vGluT1 reflects excessive excitation and inhibition/excitation imbalance in vHPC to some extent. It is also consistent with the greater anxiety-like behavior induced by caffeine and dHPC adenosine A 2A receptor knockout. It is unclear whether caffeine regulates anxiety through the alteration of vGluT1, but the activation of adenosine A 2A receptor has been reported to reduce GluT and glutamate uptake in cultured astrocytes and gliosomes (Matos et al., 2012). Therefore, the upregulation of vGluT1 may underlie the effect of caffeine on anxiety via inhibition of vHPC adenosine A 2A receptor.
As a supplement to the brain region-specific knockout of adenosine A 2A receptor, an optoA 2A R approach to mimic endogenous A 2A R signaling was used. Due to the specific construct, overexpression of optoA 2A R did not generate baseline effects and caffeine could not react with optoA 2A R too. After 3-weeks antagonism by caffeine, endogenous A 2A R signaling was inhibited, optogenetic activation of optoA 2A R captured the physiological function of the native A 2A R. Light activation of dHPC adenosine A 2A receptors reversed the behavioral alterations caused by caffeine, leading to impaired memory, while light activation of vHPC adenosine A 2A receptors decreased caffeine-induced anxiety, further confirming the dissociation between the roles of the ventral and dorsal hippocampal A 2A receptors in caffeine's effects. Considering that we examined the behavioral effects of withdrawal from a chronic regime of caffeine administration, caffeine may induce compensatory effects in response to long-term drug exposure. The instant but significant effect of optoA 2A R suggests the specificity of A 2A R signaling rather than the compensatory effects.
Since adeno-associated viral vectors driven by either the synapsin-(syn-) or CaMKIIa promoter were employed, our results suggested that the effects of caffeine may result from specific inhibition of neuronal adenosine A 2A receptors, consistent with previous studies on the role of neuronal adenosine A 2A receptor in cognition (Kaster et al., 2015; Viana Recognition (NOR) test (n = 7 mice per group). C Behavior of the four groups of mice in the Y-maze test (n = 7 mice per group). Light stimulation of opto-A 2A R in the dHPC increased the time spent exploring a novel object during the testing trial (B; *p < .05 using a two-way ANOVA followed by Bonferroni post hoc t-test) and the time spent in the novel arm during the retrieval phase (C; *p < .05 using a two-way ANOVA followed by Bonferroni post hoc t-test). (D) Light stimulation for 5 min induced c-Fos expression in optoA 2A R-positive cells but not in cells transfected with AAV-mCherry. Representative images (left) and data (right) are presented. n = 27 fields per group (3 fields per section, three sections per mouse, three mice per group); Scale bar = 50 μm; **p < .001, Student's t-test comparing optoA2AR with mCherry. Data are presented as the mean ± SEM. da Silva et al., 2016;Temido-Ferreira et al., 2018). The prominent cortical connectivity of the dorsal hippocampus and the projection to the anterior cingulated cortices are involved in memory processing (Jones and Wilson, 2005;Lavenex et al., 2006). The CA1 and subiculum of the ventral hippocampus share massive bidirectional connectivity with amygdala nuclei, which plays a key role in control (Pitkanen et al., 2000;Liu and Carter, 2018). The neuronal adenosine A 2A receptors in those projections may underlie the distinct regulation of adenosine A 2A receptors in the dHPC and vHPC. No method was designed in this research to confirm the mechanism, and further research is needed to clarify the mechanisms of neuronal A 2A R-mediated regulation and modulation of caffeine.

CONCLUSION
In the present study, we blocked and reversed the effects of caffeine by specifically inactivating and activating dHPC/vHPC adenosine A 2A receptors. For the first time, caffeine was demonstrated to affect memory by inhibiting dorsal hippocampal adenosine A 2A receptors and affect anxiety by inhibiting ventral hippocampal adenosine A 2A receptors, explaining how caffeine triggers anxiety while enhancing memory. Our results may help to understand the mechanisms of anxiety and memory and provide an experimental basis for making better use of caffeine while avoiding its side effects.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The animal study was reviewed and approved by The experimental procedures were provided by the Welfare Committee of the Army Medical University.

AUTHOR CONTRIBUTIONS
YX performed experiments, and wrote the manuscript. PL and YZ designed the study and wrote the manuscript. YN, YZ, YP, and FL performed experiments. YX, YN, YZ, and PL analyzed data. PL and YZ reviewed the manuscript. PL and YZ designed the study and worked on the final approval of the manuscript and financial support.