Before starting the study, approval was obtained from the Ethics Committee of the Affiliated Hospital of Guangdong Medical University in accordance with the Declaration of Helsinki (Grant No. YJYS2021025KT) for all research involving human participants. In diagnosing TLE, a combination of clinical symptoms (such as autonomic, psychic, and sensory disturbances), abnormal electroencephalography patterns (like focal spikes, sharp waves, or spike-and-wave complexes originating in the temporal lobes), and neuroimaging findings (such as mesial temporal sclerosis, tumors, vascular malformations, and other structural lesions in the temporal lobes) are typically considered.

A total of 76 human volunteers, consisting of 38 TLE patients and 38 healthy controls, were recruited from the Affiliated Hospital of Guangdong Medical University. All patients or their families/legal guardians were of Han Chinese descent and provided informed consent upon enrollment. Along with basic information like age and sex, specific details such as age at TLE onset and drug response were documented for the TLE patients during the field investigation. In 2010, the International League Against Epilepsy proposed a classification for drug-resistant epilepsy in patients with TLE [17]. Patients were categorized based on their response to drug treatments. Those who did not show significant improvement, had a reduction in seizure frequency of less than 60%, or experienced an increase in seizure frequency after one year of treatment with at least two suitable and tolerated medications were classified as drug-resistant. Patients who did not meet these criteria were considered drug-sensitive.

Firstly, the CpG islands surrounding the promoter regions of the four SLCs (SLC2A1, SLC6A1, SLC9A64, and SLC35A2) were assessed, ranging from –2 kb to 1 kb downstream of the first exon. The criteria included a length greater than 200 bp, a percentage of alleles C+G greater than 50%, and an observed/expected ratio exceeding 0.60. SLC2A1, SLC6A1, SLC9A64, and SLC35A2 each exhibited three, two, two, and one CpG islands, respectively, containing a total of 77, 38, 34, and 10 CpG sites within them. Polymerase chain reaction (PCR) primers were designed based on the DNA sequences of CpG islands (from Homo sapiens build CBI37/hg19) using Methylation Primer software 1.0 from Tiangen Biotech, Beijing. The primers were specifically designed for the multiplex amplification of bisulfite-converted DNA. A total of four, two, two, and one pair of primers were individually constructed to cover the DNA sequences of CpG islands in SLC2A1, SLC6A1, SLC9A64, and SLC35A2. The primer sequences can be found in Table 1.

Peripheral blood (3 mL) was collected from each participant, and 1 mL was used to extract genomic DNA by an automatic nucleic acid extractor (Tiangen Biotech, Beijing, China). The genomic DNA was then subjected to bisulfite treatment with the EZ DNA Methylation-Gold kit (Zymo Research, Orange County, CA, USA), in accordance with the manufacturer’s protocol, to transform the unmethylated allele C to allele U. The bisulfite-treated genomic DNA was amplified with the optimized panel of primer mixtures using a HotStarTaq polymerase kit (Takara, Tokyo, Japan). Biotinylated primers were utilized in the construction of libraries, and the resulting biotinylated products were subjected to paired-end sequencing (2 $\times$ 150 bp) following the manufacturer’s protocol (Illumina, San Diego, CA, USA). Subsequent to various quality control measures and data analyses, the methylation levels of the epilepsy-associated SLCs were determined based on the reads of allele C and alleles C+U within their CpG islands (Genesky Biotechnologies Inc., Shanghai, China).

The identity of the SH-SY5Y cell line (Cellverse Bioscience Technology, Shanghai, China) was confirmed through short tandem repeat (STR) profiling by the manufacturer. PCR and electrophoresis were used to test for mycoplasma contamination, and four samples consistently yielded negative results in comparison to the positive controls. The SH-SY5Y cells were cultured with high-glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) in a culture flask and then transferred to two 6-well plates. Once the cells had attached to the wall and proliferated to 2/3 of the bottom plate area, the culture medium was removed. One well was designated as the blank control group, with medium that did not contain the DNA methylation inhibitor decitabine (InvivoChem, Libertyville, IL, USA). The remaining 2–6 wells were treated with a precast solution of decitabine and culture medium, resulting in decitabine concentrations of 25 µM/L, 50 µM/L, 75 µM/L, and 100 µM/L in each well. The plates were then placed in a constant temperature incubator for 48 hours. Afterwards, genomic DNA was extracted from cells in the first 6-well plate using the DNA extraction kit (Beyotime Biotechnology, Beijing, China) for bisulfite sequencing PCR, and total RNA was extracted from cells in the second 6-well plate using the RNA extraction kit (Beyotime Biotechnology, Beijing, China) for real-time quantitative PCR.

The DNA methylation data revealed that the methylation rates of the two CpG islands near the promoter of the SLC6A1 gene were 3.8% for the first island (SLC6A1_1) and 8.1% for the second island (SLC6A1_2) in healthy controls. This indicates that the second island was highly methylated compared to the first one. Hence, the selection of the second island was made in order to evaluate how DNA methylation impacts the expression of the SLC6A1 gene. In order to assess the level of DNA methylation, we amplified a 270-bp sequence that covers the highly methylated CpG island (Homo sapiens build CBI37/hg19, NM_001348250, 11034911..11035180) using the primers F 5^′-GAAAGTYGGGGTTGGGAGAG-3^′ and R 5^′-CCCCRCAAATTATACTAAATTCTAC-3^′ for bisulfite sequencing PCR.

The genomic DNA underwent bisulfite treatment utilizing the EZ DNA Methylation-Gold kit (Zymo Research, Orange County, CA, USA) following the manufacturer’s protocol. Subsequently, the bisulfite-treated genomic DNA was amplified with a reaction mixture containing 1x HotStarTaq buffer, 2.0 mM Mg²⁺, 0.2 mM dNTP, 0.2 µM of each primer, 1 U of HotStarTaq polymerase (Takara, Tokyo, Japan), and 1 µL of template DNA. The cycling program started with an initial step of heating the sample to 95 °C for 2 minutes. This was followed by a series of 11 cycles, where the sample was heated to 94 °C for 20 seconds, then cooled to 62 °C at a rate of 0.5 °C per cycle for 40 seconds, and maintained at 72 °C for 1 minute. Subsequently, there were 24 cycles of heating to 94 °C for 20 seconds, cooling to 56 °C for 30 seconds, and maintaining the temperature at 72 °C for 1 minute. The program then concluded with a final step of heating to 72 °C for 2 minutes before cooling to 4 °C indefinitely. Finally, 0.5 U of Shrimp Alkaline Phosphatase (SAP) and 4 U of Exo I were added to 8 µL of the PCR product. The mixture was incubated at 37 °C for 60 minutes, followed by incubation at 75 °C for 15 minutes. The purified products were then sequenced using the paired-end sequencing protocol (2 $\times$ 150 bp) according to the manufacturer’s instructions (Illumina, San Diego, CA, USA).

The quantification of SLC6A1 mRNA levels was carried out through real-time quantitative PCR, utilizing the primers F 5^′-GTGGAAACAGTGCGACAACC-3^′ and R 5^′-GACGAATCCTGGAACATGC-3^′. The procedure included reverse transcription using the RevertAid First Strand cDNA Synthesis Kit by Thermo Fisher Scientific (Waltham, MA, USA). The reaction mixture underwent incubation at 42 °C for 1 hour, then was inactivated at 70 °C for 5 minutes. Subsequently, the reaction was assembled in a total volume of 20 µL, comprising 0.4 µL of cDNA, 10 µL of 2x SYBR Green Master Mix, 0.4 µL of each primer, ROX Reference Dye (50$\times$), and nuclease-free water to reach the final volume (Genstar, Beijing, China). The reactions were carried out using a LightCycler 480 sequence detector system (Roche Applied Science, Penzberg, Germany) under the following cycling conditions: an initial denaturation at 95 °C for 2 minutes, followed by 40 cycles of denaturation at 95 °C for 15 seconds and annealing/extension at 60 °C for 30 seconds. The cycle threshold (CT) values were determined, and the relative levels of SLC6A1 mRNA were calculated using the 2^{-Δ⁢Δ⁢CT} method.

The count data were analyzed using either the chi-squared test or Fisher’s exact test, while the measurement data were analyzed using Student’s t-test and presented as the mean $\pm$ standard error of the mean (SEM). To account for false positive results in multiple comparisons, the Bonferroni correction was applied. Binary logistic regression was used to control for confounding factors such as sex, age, age at onset, and drug response. Statistical analyses were primarily conducted using SPSS 23.0 (IBM, Armonk, NY, USA), with a significance level set at p $\le$ 0.0500.

A total of 38 TLE patients and 38 healthy individuals were included in this case-control study. The TLE patients were stratified based on age and age at TLE onset using a dichotomous classification method. Table 2 provides information on sex, age, age at onset, and drug response. There were no significant differences in sex or age between the TLE group and the control group (p $>$ 0.0500).

The DNA methylation rate of SLC6A1 was found to be significantly lower in the TLE group compared to the control group. Conversely, the TLE group had significantly higher DNA methylation rates in SLC9A6 and SLC35A2 than the control group. Following the Bonferroni correction, only the DNA methylation rate of SLC6A1 remained different between the TLE group and the control group (Table 3). Based on the hypomethylated SLC6A1 in TLE, a predictive model was established that showed promise in distinguishing and calibrating TLE (Fig. 1), further supporting the involvement of SLC6A1 in TLE from an epigenetic perspective.

Fig. 1.

Predictive model based on hypomethylated SLC6A1 in TLE. (A) The area under the curve (AUC) of the receiver operating characteristic curve (ROC) was 0.8102 with a 95% confidence interval (CI: 0.7118–0.9087). This value is higher than the threshold of 0.7500, indicating a promising ability of the model to discriminate TLE patients. (B) The Hosmer-Lemeshow test yielded a p-value of 0.6096, exceeding the threshold of 0.0500. Additionally, the Calibration Plot displayed a linear relationship between predicted and actual values, indicating a well-calibrated predictive model.

The study also examined TLE patients and discovered that the DNA methylation rates of SLC9A6 and SLC35A2 were notably lower in male patients than in female patients. Nonetheless, following adjustments for age, age at onset, and drug response, no remarkable distinction was detected. There was a notable difference in the DNA methylation rate of SLC6A1 between the young patients and the older controls, with the former showing lower levels. Additionally, the stratified analysis of age at onset and drug response revealed no difference in methylation rates. The stratified analysis for TLE patients is comprehensively outlined in Table 4.

As mentioned above, the 270-bp sequence covering 17 CpG sites was employed to evaluate the silencing effect of DNA methylation on the expression of the SLC6A1 gene. Under incubation with decitabine, a DNA methylation inhibitor with a set dose gradient, no methylation was detected in eight of them, while the remaining nine had varying rates of DNA methylation. These rates displayed a decreasing trend with increasing concentrations of decitabine (Fig. 2A). Additionally, the relative levels of SLC6A1 mRNA showed a significant increasing trend with increasing concentrations of decitabine (Fig. 2B). The data presented here provides experimental proof for the role of DNA methylation in suppressing the transcription of the SLC6A1 gene through its promoter region.

Fig. 2.

Silencing effect of DNA methylation on SLC6A1 gene expression. (A) Among the 17 CpG sites, no methylation was detected in 8 of them. Among the remaining sites, the numbers of 1–33% methylation show an increasing trend with increasing concentration of decitabine, while the numbers of 67–100% methylation show a decreasing trend with increasing concentration of decitabine. (B) The relative levels of SLC6A1 mRNA showed a 1.29-fold increase at 50 µM/L (p = 0.0496), a 1.56-fold increase at 75 µM/L (p = 0.0177), and a 2.03-fold increase at 100 µM/L (p = 0.0009) compared to the blank control group (0 µM/L). *p 0.0500 and ***p 0.0010.