Eight-week-old male C57BL/6J mice (Envigo, United States) were used and maintained in conditions of constant temperature (21 °C± 2°C) and a 12-h reverse light cycle. For this study, we used a total of 136 male mice. Ninety-six mice were used for the behavioral tests: sixteen for each test (8 Control and 8 CIE). For the RNA sequencing and analysis, eight mice were used (4 Control and 4 CIE). Finally, we used 32 mice to measure KYN, TRP and 5-HT levels (24 and 72 h after the last EtOH exposure).

Mice were initially housed in groups of eight with ad libitum access to food and water for 5 days. Next, they were individually housed, randomly assigned to either the Control or CIE group, and habituated to drinking water ad libitum from a drinking pipette for 7 days.

The CIE paradigm, a validated EtOH dependence and relapse drinking model, was conducted as previously described (Lopez et al., 2017; Gil de Biedma-Elduayen et al., 2022; Olivencia et al., 2023). First, mice underwent a two-bottle choice limited access (2BC) procedure to establish baseline EtOH and water intake. For 2 weeks, they had 2-h access to two drinking pipettes containing either 15% v/v EtOH or water from Monday to Friday. Water and EtOH intake were measured daily, and EtOH preference was calculated as follows: 15% EtOH intake/total fluid intake x 100. The Control group had no access to EtOH at any time. Subsequently, the CIE group was exposed to EtOH vapor in inhalation chambers (CIE: 16 h/day for 4 days), whereas Control group inhaled air alone. Air flux and ethanol concentration in the inhalation chambers were adjusted daily to maintain a plasma EtOH concentration between 150–250 mg/dL (Lopez et al., 2017; Gil de Biedma-Elduayen et al., 2022).

Before being introduced in the chambers, mice were injected i. p. with 1.6 g/kg EtOH and 1 mmol/kg pyrazole, an alcohol dehydrogenase inhibitor (Lopez et al., 2017; Gil de Biedma-Elduayen et al., 2022). Control animals received saline and pyrazole. Following the fourth inhalation session, mice underwent a 72-h abstinence period, followed by 5 days of the same 2BC protocol used during baseline. This CIE + 2BC regimen was repeated for three cycles, with a fourth cycle consisting of only four inhalation sessions. Twenty-four or 72 hours after the last inhalation session, mice were subjected to various behavioral tests or euthanized by decapitation (Figure 1A).

Consistent with previous studies (Lopez et al., 2017; Gil de Biedma-Elduayen et al., 2022; Olivencia et al., 2023), the CIE paradigm produced an escalation in voluntary EtOH consumption (Figure 1B; t (14) = 13.43, p < 0.0001) and preference for EtOH (Figure 1C; t (14) = 2.45, p = 0.028), without altering water consumption (Figure 1D; t (14) = 0.25, p = 0.805). Weekly weight measurements showed no difference in weight gain between animals subjected to the CIE paradigm and the Control group (Figure 1E). Two-way ANOVA analysis revealed a significant time effect (F (8, 126) = 4.42, p < 0.001), but no effect of treatment (F (1, 126) = 3.67, p = 0.058) and no interaction between factors (F (8, 126) = 0.81, p = 0.597).

All behavioral tests were conducted 24 h after the last exposure to EtOH vapors (Figure 1A), following established protocols (Espejo-Porras et al., 2013; Ledesma et al., 2017; Giménez-Gómez et al., 2019). In addition, elevated plus maze (EPM) and novel object recognition (NOR) were also performed at 72 h post-EtOH exposure to explore if the behavioral changes persisted over time. Prior to beginning any behavioral testing, mice were transported to a dimly lit test room and allowed a 1-h habituation period. Different sets of animals were used for each behavioral test to avoid interference between them. All the behavioral tests were analyzed by a blinded observer.

To assess locomotor activity, the Open Field test (OFT) was conducted using a computer-assisted actimeter (Actitrack, Panlab, Spain). This apparatus consists of a 45 × 45 cm enclosure surrounded by infrared sensors positioned 1 cm from the ground and 2.5 cm apart. These are connected to a control unit that analyzes various locomotor parameters. The horizontal activity (distance travelled in cm) and the average speed of the mice were recorded over a 10-min period.

Anhedonia was assessed using the Sucrose Preference Test, based on a 2-h free-choice paradigm with two pipettes available for 8 h. One pipette contained a 1% sucrose solution while the other contained water. The positions of the pipettes were alternated between the cages to prevent potential place preferences. No food or water deprivation was carried out before conducting the test. Sucrose preference was calculated as the ratio of sucrose intake to total liquid intake. A decrease in sucrose preference is widely accepted in the literature as an indicator of anhedonia (Giménez-Gómez et al., 2019).

The EPM paradigm was performed to assess the anxiety levels of the animals. Mice were placed in the center of the EPM consisting of four arms (30 × 5 cm), two open and two enclosed, forming a central platform (5 × 5 cm) at the intersection point, elevated 45 cm above the floor. The animals were allowed 5 minutes to freely explore the maze. The number of entries into the open arms and the percentage of time spent in the open arms were used as indicators of anxiety (Ledesma et al., 2017).

Memory function was evaluated using the NOR test. Mice were placed in an open box (24 × 24 × 15 cm) initially containing two identical objects and allowed 3 minutes of free exploration. After this, they were returned to their home cage for a 1-min interval during which one object was replaced with a new and distinct object. The animals were then reintroduced to the box for another 3 minutes of exploration. The position of the objects in the cage was alternated between animals and experimental groups to minimize any potential place preference. Object exploration was defined as intentional contact with the object by the mouse’s snout or front paws within 2 cm of the object. The time spent exploring the old (familiar) and new (novel) objects during the second exposure was recorded, and the discrimination index (DI) was calculated using the formula: DI = (tnovel - tfamiliar)/(tnovel + tfamiliar) × 100 (Ledesma et al., 2017).

Twenty-four hours after the final exposure to EtOH vapors, the mice were euthanized by decapitation (Figure 1A). Trunk blood was collected in 10 mL K₂-EDTA tubes (BD, United States) and immediately centrifuged twice at 1300 x g (4°C) for 10 min to obtain plasma, which was then stored at −80°C. Brains were extracted and dissected on ice using a mouse brain matrix for coronal slices (World Precision Instruments, United States). After removing the olfactory bulbs, brain tissue anterior to the optic chiasm was collected and referred to as “limbic forebrain” (Giménez-Gómez et al., 2018; Giménez-Gómez et al., 2021; Gil de Biedma-Elduayen et al., 2022). The cortex of the remaining brain portion, as well as the cerebellum, were also collected. The transcriptomic analysis was conducted on the NAc. Using a matrix, 1 mm thick sections were made, and the NAc was dissected with a punch (Stoelting Co., United States), between bregma 1.98–0.62 mm. The distinct brain regions were either stored at −80°C or preserved in the appropriate buffer for subsequent use.

RNA extraction, sequencing and analysis were performed by Macrogen (Korea). RNA was extracted using the RNeasy Mini Kit from Qiagen following the manufacturer’s protocol. DNase treatment was applied to remove DNA contamination. The RNA integrity was verified using Agilent Technologies 2,100 bioanalyzer, with a RIN value ≥7. Paired-end libraries were then generated using the TruSeq Stranded mRNA Sample Prep Kit (Illumina^®, United States). After PCR amplification, fragments with insert sizes between 200–400 base pairs were selected. The cDNA libraries were sequenced using sequencing by synthesis technology (Illumina^®, United States) and the quality of reads was evaluated using FastQC v.0.11.7 (Babraham Bioinformatics, United Kingdom). To minimize biases, Trimmomatic 0.38 was used to remove low-quality reads, adapter sequences, contaminating DNA, and PCR duplicates (Usadel Lab, Germany).

cDNA fragments were mapped to the Mus musculus (mm10) genome using HISAT2 v.2.1.0 (Kim Lab, United States) and the Bowtie2 2.3.4.1 aligner (SourceForge, United States). Known genes and transcripts were assembled with StringTie v.2.1.3b (Johns Hopkins University, United States) based on the reference genome. The abundance of genes/transcripts was quantified in terms of read counts, FPKM (fragments per kilobase of transcript per million mapped reads) and TPM (transcripts per million per kilobase). The read count values of known genes obtained using the -e option of StringTie were used as the original raw data (45,777 genes). The analysis involved filtering low-quality transcripts, normalizing data, and removing samples with multiple zero read count. To minimize systematic bias, size factors were estimated using the “estimate Size Factors” method, and normalization was carried out using the Relative Log Expression (RLE) method in the DESeq2 R library. Log2-transformed values (read count+1) and regularized log(rlog) transformations were used for data visualization. Using the normalized values, similarity tests between samples were conducted, including correlation analysis, multidimensional scaling, and hierarchical clustering.

Differentially expressed genes between the Control and CIE groups were determined using Fold Change (FC) and the binomial Wald Test, performed via DESeq2 for pairwise comparisons. Genes with |FC| ≥ 2 and P < 0.05 were considered significant. This analysis provided a list of differentially expressed genes, which was then used for hierarchical clustering analysis and graphical representation via a heatmap and volcano plot. Additionally, a gene set enrichment analysis based on gene ontology was conducted using the g:Profiler tool (http://geneontology.org/) to identify prevalent molecular functions or signaling pathways within the set of differentially expressed genes. All data files are available upon request.

Samples were processed and normalized to tissue weight as previously described (Giménez-Gómez et al., 2018; Giménez-Gómez et al., 2019; Giménez-Gómez et al., 2021; Abuin-Martínez et al., 2021; Gil de Biedma-Elduayen et al., 2022). Plasma samples were deproteinized by adding 75 μL of 6% perchloric acid to the mixture of 50 μL plasma and 125 μL of water. Limbic forebrain, cortex and cerebellum samples were homogenized in five volumes of deionized water by sonication (Labsonic 2000U, B. Braun Melsungen AG, Germany) at 30% amplitude for 15 s. The homogenates were deproteinized by adding 25 μL of 6% perchloric acid per 100 µL of sample. In both cases, the acidified samples were vortexed and kept at room temperature for 10 min before being centrifuged at 16,000 g for 15 min at 4°C to collect the supernatants.

For KYN measurement, 20 μL (plasma) or 60 μL (limbic forebrain, cortex and cerebellum) of supernatant were applied onto a reversed-phase column (80 mm × 4.6 mm, 3 mm; HR-80; Thermo Fisher Scientific, United States). KYN was isocratically eluted using a mobile phase containing 0.1 M sodium acetate and 4% acetonitrile, pH 4.6, at a flow rate of 1 mL/min. KYN was measured by UV detection at 360 nm using a 2487 UV detector (Waters, United States). Under these conditions, the retention time for KYN was 4.2 min. For TRP and 5-HT measurement, 20 μL of supernatant were applied to the same column, and TRP and 5-HT were isocratically eluted using a mobile phase containing 0.1 M ammonium acetate and 8% methanol, pH 3.8 (adjusted with glacial acetic acid), at a flow rate of 1 mL/min. TRP and 5-HT were detected fluorometrically at excitation/emission wavelengths of 270/360 nm and 290/398 nm, respectively using a 2,475, Multi Fluorescence Detector (Waters, United States). Under these conditions, retention times were 7 min for TRP and 4.5 min for 5-HT.

Initial group size (n = 8) was selected based on our previous experience to ensure a statistical significance considering the possible loss of mice due to the application of the criteria of humane endpoints if physical signs other than ethanol intoxication were observed, loss of consumption data due to malfunction of the pipette tip or outlier values (determined by the extreme studentized deviate method [ROUT test] with significance level of Q = 1%). These three conditions were the only exclusion criteria employed and account for differences in group sizes. For the RNA sequencing and analysis, the initial group size was n = 4, considering the high sensitivity and analytical power of this technique.

Data are presented as mean ± SEM. The threshold for statistical significance was set at P < 0.05. Normality was assessed using the Shapiro–Wilk test (Supplementary Table S1). Comparisons between two groups were conducted using an unpaired Student’s t-test when the data met the assumption of normality, and the Mann-Whitney U test when they did not. For analyses involving two independent variables, a two-way ANOVA was used. As the only two-way ANOVA performed (Figure 1E) did not show statistically significant main effects or interaction between factors, no post hoc multiple comparisons tests were conducted. All statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software Inc., United States). Sample extraction, processing, and data collection and analyses were carried out by a researcher blinded to the treatment group of each animal.

To assess the effect of CIE model exposure on the selected behavioral and cognitive parameters, the behavioral tests were performed 24 h after the last exposure to EtOH vapors. Analysis of locomotor activity via the OFT revealed no significant differences in the distance traveled (Figure 2A; t (13) = 1.79, p = 0.097) or average speed (Figure 2B; t (13) = 0.77, p = 0.457) between the Control and CIE groups. Similarly, no differences were found in sucrose preference, used as an indicator of anhedonia, between the two groups (Figure 2C; U = 24.5, p = 0.458, exact). Anxiety levels were assessed using the EPM, measuring the percentage of time spent in and the number of entries into the open arms as indicators. A reduction in the percentage of time spent in the open arms was observed in the CIE animals (Figure 2D; t (11) = 2.28, p = 0.043), while the number of entries did not change (Figure 2E; U = 21.5, p = 0.730, exact). This suggests that CIE animals entered the open arms as frequently as the Control animals but spent less time exploring them, indicating a higher level of anxiety in the CIE group. Finally, the NOR test was used to evaluate memory by assessing the DI between the new and old objects. The Student’s t-test revealed a decrease in the discrimination index within the CIE group (Figure 2F; t (12) = 4.04, p = 0.002), indicating memory impairment 24 h after vapor exposure.

To determine whether the observed anxiety- and memory-related changes persisted over time, we conducted EPM and NOR tests again at 72 h post-withdrawal. At this later time point, no significant differences were found between Control and CIE groups in the time spent in (Supplementary Figure S1A; t (14) = 0.726, p = 0.480) or entries into the open arms (Supplementary Figure S1B; t (14) = 0.214, p = 0.834) of the EPM. Similarly, no differences in the DI were observed in the NOR test (Supplementary Figure S1C; t (12) = 0.413, p = 0.687). These results suggest that the behavioral alterations identified at 24 h post-withdrawal did not persist at 72 h.

After describing the alterations in anxiety and memory produced by the CIE model, we performed RNA sequencing to examine gene expression profiles in the NAc of mice 24 h after the last vapor exposure. Post-sequencing, a total of 45,777 known genes were identified as raw, unprocessed data. Filtering out low-quality transcripts and normalizing the data resulted in 18,857 genes retained for statistical analysis, while 26,920 were excluded (Supplementary Figure S2A). Quality assessments demonstrated high sample similarity, as shown by correlation analysis based on Pearson coefficients (Supplementary Figure S1B), multidimensional scaling (Supplementary Figure S1C), and hierarchical clustering (Supplementary Figure S1D), with no outlier samples identified.

Statistical analysis identified 81 differentially expressed genes between the Control and CIE groups based on fold change (|FC| ≥ 2) and a significance threshold of P < 0.05 (Figures 3A,B; Supplementary Table S2). Among these genes, 52 were upregulated, and 29 were downregulated in the CIE group (Figure 3A). Hierarchical clustering affirmed strong correlations among samples within the same group (Figure 3C).

The differential expression analysis revealed notable changes in genes primarily associated with immune response and inflammation in mice exposed to the CIE model. For instance, the gene Cd74, which exhibited the highest overexpression (FC = 6.446) in the CIE group, is known for its role in activated microglia and proinflammatory signaling (Hwang et al., 2017). Several genes related to MHCII (H2-Aa, H2-Eb1, H2-Ab1, H2-Q6, H2-D1, H2-Q7, H2-Q4, H2-K1) were also upregulated. MHCII expression is known to increase in activated microglia in response to proinflammatory cytokines (Panek and Benveniste, 1995). Similarly, numerous genes implicated in the immune responses, pro-inflammatory signaling, macrophage function and leukocyte trafficking, including Igtp, Ifi209, Iigp1, Ifi27l2a, the Gbp family, Bst2, Siglec1, Usp18, Nlrc5, Dusp2, Clec4a2, Fut7, Treml2, and Cd274 or Pd-L1, showed increased expression (Ford and McVicar, 2009; Wei et al., 2013; Schneider et al., 2017; Basters et al., 2018; Siddiqui et al., 2019; Tiwari et al., 2019; Mishra et al., 2020; Nasu et al., 2020; Haque et al., 2021; Zhang et al., 2021).

Among the downregulated genes, Glra1 and Kcne2 stand out. Glra1 encodes the α1 subunit of the glycine receptor (GlyR), which is broadly expressed in the CNS and has been suggested to play a role in modulating the reward system in response to EtOH exposure (Förstera et al., 2017; Gallegos et al., 2019). On the other hand, KCNE2 encodes a transmembrane protein that is part of various potassium channels. This protein is prominently expressed on the apical membrane of the choroid plexus, where it influences the composition of cerebrospinal fluid and indirectly affects neuronal excitability (Abbott et al., 2014).

The top 20 enriched genes, identified through an ontology-based gene enrichment analysis, were represented in a dot plot. This analysis confirmed that the CIE group shows neuroinflammation and an altered immune response compared with the Control group (Figure 3D).

The CIE model produced alterations in anxiety and memory along with an increase in transcripts related to the immune response. The KYN pathway is known to be activated under inflammatory conditions and has also been associated with elevated anxiety and cognitive impairments. Based on this information, we explored the impact of the CIE model on the activation of the KYN pathway, a possible link between EtOH consumption, neuroinflammation and behavioral alterations (Jiang et al., 2020; dos Santos et al., 2021). To this end, we examined the levels of KYN, TRP, and 5-HT, and calculated the KYN/TRP and 5-HT/TRP ratios as indicators of TDO + IDO and TPH enzyme activities, respectively, 24 h post-vapor exposure.

First, we analyzed the activation of the KYN pathway in plasma. Our data demonstrated that KYN levels were elevated in plasma (Figure 4A; t (13) = 5.09, p = 0.0002), with no changes in TRP (Figure 4B; U = 26, p = 0.574, exact) or 5-HT levels (Figure 4C; t (12) = 1.24, p = 0.239). As a result of this increase in circulating KYN levels, the KYN/TRP ratio (Figure 4D; t (13) = 4.65, p = 0.0005) was elevated, while the 5-HT/TRP ratio (Figure 4E; t (12) = 0.67, p = 0.514) remained unchanged. Next, we focused on the limbic forebrain, which includes the NAc, where RNAseq analysis has been performed. In this area, we observed an increase in KYN levels (Figure 5A; t (12) = 12.13, p < 0.0001), with no changes in TRP (Figure 5B; U = 14, p = 0.121, exact) or 5-HT levels (Figure 5C; t (14) = 1.13, p = 0.276). Similarly, the KYN/TRP ratio was elevated in the CIE group (Figure 5D; t (11) = 11.80, p < 0.0001), while the 5-HT/TRP ratio remained unchanged (Figure 5E; t (13) = 0.15, p = 0.883).

Finally, given that the cortex and cerebellum have been implicated in previously described alterations (Miquel et al., 2015; Koob, 2014), we aimed to investigate whether changes in KYN levels were also observable in these areas. Our results showed that KYN levels were elevated in cortex (Figure 6A; t (12) = 7.35, p < 0.0001), with no changes in TRP (Figure 6B; t (13) = 0.49, p = 0.632) or 5-HT levels (Figure 6C; t (13) = 2.02, p = 0.065). The KYN/TRP ratio was increased (Figure 6D; t (12) = 5.18, p = 0.0002), while the 5-HT/TRP ratio remained unchanged (Figure 6E; U = 15, p = 0.152, exact). In the cerebellum (Figures 7A–E), we observed the same pattern as in the previous regions, with increased KYN levels (U = 0, p = 0.0002, exact) and KYN/TRP ratio (t (12) = 8.14, p < 0.0001), and no changes in TRP (t (12) = 2.01, p = 0.068), 5-HT (t (14) = 0.81, p = 0.432) or 5-HT/TRP ratio (t (12) = 0.052, p = 0.960).

To determine whether the increase in KYN levels persisted over time, we analyzed the levels of KYN, TRP, and 5-HT in the limbic forebrain 72 h post-withdrawal and found no significant differences in any of the metabolites analyzed: KYN (Supplementary Figure S1D; t (11) = 1.621, p = 0.133), TRP (Supplementary Figure S1E; t (12) = 0.2480, p = 0.808), or 5-HT (Supplementary Figure S1F; t (12) = 0.2459, p = 0.810).