Twenty male C57BL/6 mice with no significant difference in learning ability were screened using the water maze pre-test, and divided into control and model groups, with 10 mice in each group, according to the random number table method. We used ≥99.5% anhydrous ethanol (Aladdin, China) and distilled water to get a 10% (V/V) alcohol solution. The model group had free access to 10% (V/V) alcohol solution. After continuous access to 10% alcohol solution for 60 days, in accordance with previous studies, the water maze test and other experiments were used to verify the learning and memory impairments and determine whether the mouse model of chronic alcohol exposure was established (Salling et al., 2016). The control group had free access to the same volume of distilled water as the alcohol solution. The two groups of mice were bred in the same environment and conditions, with free drinking intake, but were separated into different cages with general access to 10% alcohol solution. Animal experiments in this study were approved by the Animal Ethics Committee of the Xinxiang Medical University, and all experiments strictly followed the 3R principles.

The main instruments used in this study include a qTOWER2.2 fluorescent qPCR instrument (Analytik Jena, Germany), centrifuge D3024R (Scilogex, United States), ordinary PCR instrument EasyCycler (Analytik Jena, Germany), UV-visible spectrophotometer UV-1800 DS2 (Mapada, China), ScanDrop100 (Analytik Jena, Germany), high-speed refrigerated centrifuge (Heraeus, Germany), and ELx800 Automatic Microplate Tester (BioTek, United States of America).

TruScript 1st Strand cDNA Synthesis kit (Ai Delai); Ultrafine total ATPase determination, protein quantitative test, mitochondrial extraction (Nanjing Jiancheng Institute of Biological Engineering, China) kits; 2 × SYBR^® Green premix (China DF); 10 μL tip (GCS, United States of America), 200 μL tip (GCS, United States of America), 1 mL tip (United States of America, GCS), 1.5 mL RNA-free EP tube (GCS, United States of America), low white PCR reaction tube (Bio-RAD, United States of America), optical sealing film (Bio-RAD, United States of America) (The gun head and EP tube were sterilized and dried before use), and ethanol (Aladdin, China).

The Morris water maze experiment: experimental animals were allowed to swim and learn to find a platform hidden in the water (Tian et al., 2019). A place navigation experiment was used to measure the learning ability of mice in the water maze. The experiment lasted for 4 days. The route map and time required for the mice to find and climb the platform were observed and recorded, and the incubation period and swimming speed were recorded. The mice were placed in the water with their heads facing the pool wall. One of the four starting positions (East, West, South, and North) was randomly selected, and the time (s) when the mice found the underwater platform was recorded. In the previous training session, if the time exceeds 60 s, the mice were guided to the platform. The mice were allowed to remain on the platform for 10 s, and the incubation period was recorded at 60 s. After each training session, the mice were dried and placed under a 150 W incandescent lamp for 5 min when necessary, and then returned to the cage. Each mouse was trained four times a day with an interval of 15–20 min between two training sessions for four consecutive days. The incubation period was then calculated for each group.

All mice were euthanized and the part of the hippocampus were taken from eight mice in each group. Mitochondria were isolated using a mechanical method involving low-speed differential centrifugation to remove debris and large organelles, followed by high-speed differential centrifugation according to the instructions of the mitochondrial extraction kit. After the protein concentration of the sample was measured according to the instructions of the protein quantitative test kit, the samples were successively pre-treated, and enzymatic and phosphorus-fixing reactions were performed, according to the instructions of the ultrafine total ATPase assay kit. Absorbance was determined at a wavelength of 636 nm with a spectrophotometer, according to formula U = [∆A/(0.01 × t)] × D (where U is the enzyme activity, △A is the change of absorbance value during the reaction time, t is the reaction time [min], and D is the dilution multiple) to calculate the ATPase activity.

Samples of the part of hippocampus were taken from eight mice in each group. The blood on the surface of hippocampus tissue was washed with 50 mmol/L of potassium phosphate buffer pre-cooled by ice, then dried with filter paper, weighed, and stored −70°C. A homogenate solution (mass volume ratio: 0.12) was prepared with 50 mmol/L potassium phosphate buffer (pH 8.0) at 0°C–4°C. The homogenate solution was centrifuged at 12,000 rpm at 4°C for 10 min. The supernatant (75 μL) was placed into another centrifuge tube. Approximately 0.25 mL of zinc acetate solution was added to volume fractions of 0.01 and 0.45 mL distilled water at room temperature and incubated for 10 min. Approximately 0.25 mL trichloroacetic acid was added to a volume fraction of 0.1 mL of distilled water and centrifuged for 10 min at 7,000 r at 4°C. The supernatant was collected and 133 μL of N, N-dimethyl-p-phenylenediamine hydrochloride (20 mmol/L)-hydrochloric acid (7.2 mol/L) buffer and 133 μL of ferric chloride (30 mmol/L)-hydrochloric acid (1.2 mol/L) buffer solution were added fully mixed. After 20 min, absorbance was measured at 670 nm wavelength using an automatic microplate reader. Brain tissue H₂S was calculated according to the H₂S standard curve, and the level was expressed as the amount of H₂S per unit mass of tissue (nmol/g).

Six mice were randomly selected from the control and model groups. The hippocampal tissue was removed and preserved in liquid nitrogen. If the sample quality inspection results were qualified, downstream experiments could be conducted.

Total RNA was extracted directly from cells or tissues using Trizol reagent and RNA was released after cell lysis. RNA and DNA were separated under acidic conditions. After the addition of chloroform, the sample was divided into aqueous and organic layers, with RNA present in the aqueous layer. After collecting the upper aqueous sample layer, the RNA was reduced by isopropanol precipitation.

The optical density (OD) and purity of the RNA were measured using a NanoDrop spectrophotometer. Few samples were subjected to agarose gel electrophoresis to determine whether the RNA was degraded.

The RNA samples were selected and sequenced using HiSeq2500 (American Illumina platform) to build libraries of mRNAs (mRNA library) and miRNAs (microRNA library). Comprehensive sequencing was performed after fragment removal, chain synthesis, end repair, A-tail and adapter connection, PCR amplification, cluster generation, and quality control.

Analysis of differentially expressed genes is an important part of comparative transcriptome sequencing, as well as the basis for excavating important regulatory miRNAs and searching for key genes to explain their biological molecular regulatory mechanisms. We analyzed the differentially expressed miRNAs and mRNAs in the chronic alcohol exposure and control groups. Furthermore, we predicted and annotated the target genes of differentially expressed miRNAs using bioinformatic methods to analyze the expression patterns of miRNAs and target genes and conducted a joint analysis with mRNA sequencing data to provide more information for exploring the molecular mechanism of miRNAs regulating chronic alcohol exposure.

We used Fragments Per Kilobase Million (FPKM) to calculate the abundance of known genes in different samples. For mRNA, |log2 fold change| ≥1 and p < 0.05 can be considered as differentially expressed. It was visualized by volcano and cluster diagrams.

Target genes with differential expression of miRNAs were predicted using the TargetScan, miRTarBase, and miRDB databases. The target genes were subsequently intersected to construct a differentially expressed miRNA-mRNA regulation correlation network, and the regulatory network was visualized using Cytoscape_v3.9.0.

Gene ontology (GO) and KEGG enrichment analyses are the two most widely used gene function analysis strategies. GO can annotate and cluster gene functions from three levels: molecular function (MF), cell composition, and biological processes (BP), while KEGG collects a large amount of gene pathway information, which contains information such as gene metabolic pathways, products, and enzymes, which can help researchers explore the specific functions of genes from a deeper level. The basic unit of GO is a term corresponding to an attribute, whereas the basic unit of KEGG is a pathway.

The STRING database covers a large amount of protein information, including protein interactions, three-dimensional structures, and related functional enrichment results. Multiple differentially expressed genes can be built using this database. A protein interacting with N proteins can assume a degree of “n.” The higher the degree score is, the more likely the protein is at a key node in the network structure. In this study, the top 30 key genes in degree rankings are in the Table 3.

Each group was randomly selected from the hippocampus of eight mice, and total RNA was extracted from the hippocampus. First, the RNA OD value (A260/A280) was detected with the ultra-trace nucleic acid protein analyzer (scandrop100), and then the Aidlab reverse transcription kit (TUREscript 1st Stand cDNA SYNTHESIS Kit) was used for reverse transcription operation using a 20 μL reaction system comprising 1,000 ng of total RNA, 4 μL of 5×RT reaction mix, 1 μL of Rondam primer/oligodT, 1 μL of TUREscript H- RTase/RI mix, and the reaction mix was raised to 20 μL with RNase Free dH2O. The reverse transcription reaction conditions used were as follows: after 40 min at 42°C and 10 min at 65°C, the cDNA was obtained and stored at −80°C. The acquisition of Ct value data in the reaction was set by a corrected threshold. The real-time qPCR method used actin as the internal reference gene, and the 2^-△△Ct method was used for relative quantification. The amplification conditions were as follows: 95°C for 3 min, 95°C for 10 s, and 60°C for 30 s, with 40 cycles of the reaction and primers for key mRNA (Table 1).

The ClusterProfiler (v4.0.3) package of the R language was used to analyze the target genes in the miRNA-mRNA regulatory network for GO and KEGG pathway enrichment to search for terms or pathways with significantly enriched differential genes. When a term or pathway had a p-value < 0.05, the entry was considered significantly enriched. GraphPad Prism 8.0.2 was used to visualize the qPCR results. The t-test was used for inter-group comparison, and p < 0.05 was deemed statistically significant.

According to the water maze experiment results (Figure 1), there was no substantial difference in latency between the mice in each group on the first day (p > 0.05); however, from the second day onwards, the latency of the mice in the model group was significantly higher than that of the control group (p < 0.05), indicating that chronic alcohol exposure seemed to have some negative effects on their learning capacity. The latency of the mice in each group decreased as the number of empirical days increased (Figure 1A). The 4-day place navigation experiment data were analyzed and processed, and the results revealed that the model group had considerably higher latency than the control group (p < 0.05), indicating that the mice chronically exposed to alcohol had impaired learning functions. This indirectly demonstrated the successful initial construction of the mice model with alcohol addiction (Figure 1B).

According to Figure 2, the hippocampal tissues of the mice in the model group had a significantly lower mitochondrial ATPase activity than that of the control group (p < 0.01). This finding suggests that chronic alcohol exposure impairs the ability of mitochondria to normally convert energy from food.

As shown in Table 2, mice in the model group had significantly higher levels of H₂S in their hippocampus than mice in the control group (p < 0.01). This finding suggests that chronic alcohol exposure can activate CBS, which catalyzes the production of significant amounts of H₂S.

Differential expression analysis was performed on the miRNA sequencing data with the screening conditions of p < 0.05. In total, 208 miRNAs were differentially expressed, of which 77 were upregulated and 131 were downregulated. Figures 3A, C show volcano plots and heatmaps of the differentially expressed miRNAs. The mRNA sequencing data were subjected to differential expression analysis, with the screening condition of a |log Fold change| ≥1 and p < 0.05. A total of 377 differentially expressed mRNAs were identified, of which 185 were upregulated and 192 were downregulated, and the volcano plot and heatmap of differentially expressed mRNAs are shown in Figures 3B, D.

Downstream target prediction of differentially expressed miRNAs was performed using the Targetscan, miRTarBase, and miRDB databases, and 437 miRNA-mRNA interactions were detected by intersecting the results of the three databases. The target mRNAs were then intersected with the previously mentioned 377 differentially expressed mRNAs. The final miRNA-mRNA network contained 41 differentially expressed miRNAs and 99 differentially expressed mRNAs (Table 3). Cytoscape was used to build and show the miRNA-mRNA regulatory network (Figure 4).

The GO function and KEGG pathway enrichment analyses of target genes in the miRNA-mRNA regulatory network were performed using the cluster Profiler package in the R language. Approximately, 2,762 GO enrichment terms were obtained, including 2,190 BP terms, 248 cellular component (CC) terms, and 324 MF terms. BPs mainly comprised neuronal death, regulation of the Wnt signaling pathway, regulation of developmental growth, anion transmembrane transporter activity, and histone modification protein dephosphorylation; CC terms were mainly related to RNA polymerase II transcription regulator complex, transcription regulatory complex, and lysosomal membrane; and MFs comprised phosphoric ester hydrolase activity, DNA binding transcription activator activity, and phosphatase activity. The bubble chart displays the top ten enriched BP, CC, and MF terms (Figure 5). A chord diagram was constructed using the GO plot package in R to illustrate the top five findings, with the most significant enrichment of GO term functions to demonstrate a smaller selection of high-dimensional data (Figure 6). The KEGG pathway enrichment analysis yielded 190 signaling pathways, namely, the mitogen-activated protein kinase (MAPK) signaling pathway, pathways of neurodegeneration-multiple disease, phosphoinositide 3-kinase-protein kinase B (PI3K-Akt) signaling pathway, mitogen-activated protein kinase (Ras) signaling pathway, and dopaminergic synaptic pathway (Figure 7).

The PPI network was analyzed using the STRING database for 118 target genes in the miRNA-mRNA regulatory network (Figure 8), and the PPI protein interaction network constructed in this study consisted of 117 nodes and 96 edges (Table 4 shows the mRNAs in the top 30 of degree). Then, we delete the scattered nodes, leaving the central part of the network. Among them, the Kirsten rat sarcoma viral oncogene homolog (KRAS), cadherin 1 (CDH1), Fos proto-oncogene, transcription factor activator protein-1 subunit (FOS), SRY-box transcription factor 2 (SOX2), brain-derived neurotrophic factor (BDNF), and early growth response 1 (EGR1) were the top six nodes with the highest degree scores, and thus were the key node genes in this PPI network.

Real-time qPCR was used to detect key genes, such as FOS and EGR1. The results indicated that FOS and EGR1 mRNA expression in the hippocampus of chronic alcoholic mice was significantly decreased (p < 0.05), suggesting that chronic alcohol exposure alters the expression of several important mRNAs in the hippocampus of mice. Furthermore, GraphPad Prism software was used to display the qPCR results (Control refers to control mice, Model refers to mice in the chronic alcohol exposure group). Visualization results are shown in Figure 9).