In this paper, statistical analysis and visualization were accomplished in R package (Version 3.6.3). “Glmnet” (4.1-2) and “survival” (3.2-10) were used for LASSO analysis, “rms” was applied for nomogram and calibration, and ROC analysis was run by “timeROC”. Furthermore, the results with p < 0.05 were considered to be of statistical significance and ggplot2 package was used for graph plotting. The area under the curve (AUC) is used to measure the model’s discriminative ability, AUC always falls between 0 and 1, AUC > 0.8 indicates high accuracy of the model.

Utilizing TCGA data, Cox analysis was conducted to assess the correlation between SLC22A1 expression and the survival rates of pan-cancer patients. The results demonstrated that SLC22A1 RNA level was related to both poor and improved prognosis (in terms of OS and DSS) in specific cancers. Table 1 outlines the outcomes of the OS analysis, revealing that elevated SLC22A1 expression correlated with poor prognoses in ACC, HNSC, KICH, KIRC, KIRP, LAML, LGG, MESO, and SARC (HR > 1, p < 0.05). Conversely, higher SLC22A1 expression in BLCA, LIHC, and PCPG was associated with better clinical outcomes (HR < 1, p < 0.05). DSS analysis further indicated that increased SLC22A1 levels were significantly correlated with longer disease-specific survival time in LIHC patients (HR < 1, p < 0.05), while in ACC, HNSC, KICH, KIRC, LGG, MESO, and SARC, higher SLC22A1 expression was associated with shorter DSS (HR > 1, p < 0.05).

Next, we intended to assess the influence of SLC22A1 on immune cell recruitment. Using the TIMER tool, we evaluated the relation between SLC22A1 RNA and immune cell infiltration levels, including neutrophils, helper T cells (T cell CD4+), macrophages, myeloid dendritic cells, B cells, and cytotoxic T cells (T cell CD8+) in various cancers. Fig. 1A illustrates the most significant correlations observed in LGG and LIHC. In LIHC, SLC22A1 expression positively correlated with the infiltration of five immune cells, while in LGG, the mRNA level of SLC22A1 was remarkable negatively correlated with infiltration levels of six immune cells. Fig. 1B indicates that higher immune infiltration levels significantly correlated with worse prognosis for LGG patients (p < 0.01), while the survival time of LIHC patients was not affected by immune infiltration levels.

It is widely thought that the interaction of immune checkpoints can hinder immune cell recognition and the killing of tumor cells, enabling cancer cells to evade immune surveillance [32–34]. Screening 49 immune checkpoints expressed in tumor cells [35], we found that the number of checkpoints positively and negatively related to SLC22A1 expression was highest in LGG and LIHC, respectively (Fig. 1C). Notably, cancers of immune organs (DLBC and TGCT) exhibited the highest coefficients. These findings suggest that SLC22A1 might influence LGG tumor cell development by enhancing immune cell recruitment and immune escape through immune checkpoints.

Genetic and epigenetic alterations are common features of cancer cells, and abnormal gene expression can impact the immune landscape within tumors. To investigate the role of SLC22A1 expression in carcinogenesis, we utilized the TIMER database to evaluate the relation between SLC22A1 RNA and the transcription levels of mismatch repair (MMR) enzymes (MutL homolog 1(MLH1), MutS homolog 2 (MSH2), MutS homolog 6 (MSH6), and Postmeiotic segregation increased 2 (PMS2)) as well as methyltransferase enzymes (DNA (cytosine-5-)- methyltransferase 1(DNMT1), DNA methyltransferase3alpha (DNMT3A), DNA methyltransferase3beta (DNMT3B), and DNA (cytosine-5-)-methyltransferase 3-like (DNMT3L)). As shown in Fig. 1D, SLC22A1 expression was associated with at least one MMR gene in many cancers. Notably, in LIHC, the transcription level of all four MMR genes was negatively related with SLC22A1 levels. Fig. 1E indicates a relationship between the expression of methyltransferase enzymes and SLC22A1 expression across various cancers. Especially in KIRC and LGG, SLC22A1 and four methyltransferases exhibited a significantly positive co-expression, while an inverse relationship was observed with PRAD. In summary, SLC22A1 was associated with the genomic stability of various tumors. Given the strongest correlation observed in LGG between SLC22A1 and tumor immunity, it is highly conceivable that SLC22A1 influences the tumor immune environment by affecting DNA methylation levels in LGG.

To explore the potential mechanism by which SLC22A1 affects the tumor microenvironment, a PPI network involving SLC22A1 was created using the GeneMANIA website. As depicted in Fig. 1F, SLC22A1 exhibited significant physical interactions with ATP-binding cassette sub-family G member 2 (ABCG2), ATP-binding cassette sub-family B member 1 (ABCB1), and Alcohol dehydrogenase 1A (ADH1A). Both ABCG2 and ABCB1 are members of the ATP-binding cassette (ABC) transporter family, crucial for drug transport in the human body. These proteins use ATP hydrolysis to expel various drug molecules from cells, influencing tumor drug resistance. ABCG2 is also a biomarker for various tumor stem cells, impacting the proliferation, invasion, and prognosis of many cancers. ADH1A plays an important role in the redox detoxification metabolic function of the liver and holds potential clinical prognostic value for LIHC patients. Accordingly, our findings suggest that SLC22A1 may interact with drug transporters and tumor stem cell markers, especially ABCG2, affecting cell metabolism and altering the relation between cancer cells and TME.

The aforementioned analyses suggested that SLC22A1 might promote development of LGG by affecting the TME. To explore whether changes in SLC22A1 expression affect the prognosis of cancer patients, we assessed the prognostic value of the methylation levels of 12 CpG island sites of SLC22A1 across various cancers using the Methsurv tool. The top 20 most significantly correlated sites, based on p values, are presented in Fig. 2A. The results indicated a significant correlation between SLC22A1 methylation and a favorable prognosis in LGG patients. It is well-documented that DNA methylation, as an epigenetic modification affecting gene expression, predominantly reduces gene expression. Therefore, reducing SLC22A1 expression may potentially prolong the survival of LGG patients.

To further elucidate the impact of SLC22A1 on the tumor microenvironment, we investigated whether it influences the sensitivity of tumor cells to chemotherapy drugs and explored its underlying mechanisms. Analyzing data from the Parenteral Drug Association (PDA) approved and clinical trial anticancer drugs in the CellMiner database, we identified 17 drugs whose IC50 were correlated with SLC22A1 expression in NCI-60 cell lines. The action pathways and targets of these drugs are illustrated in Fig. 2B. Notably, SGI-1027 and CEP-9722, exhibiting the strongest correlations, were effective in inhibiting cancer through epigenetic modification. However, they demonstrated opposite sensitivities to SLC22A1 expression levels, possibly due to their corresponding targets (DNMT and PARP1) playing opposing roles in genome stability. The analysis revealed that the mRNA level of SLC22A1 in cancer patients could impact the sensitivity of anticancer drugs, thereby influencing clinical outcomes, suggesting that SLC22A1 could affect the sensitivity of tumor cells to anticancer drugs through various pathways, especially via epigenetic modification, specifically methylation (DNMT), which has a notable effect on inhibiting tumor cell proliferation, consistent with the findings in Fig. 1E.

Given the comprehensive impact of SLC22A1 on various tumor-related indicators in pan-cancer, with the strongest association observed in LGG, we sought to validate the function of SLC22A1 in LGG using data from TCGA and CGGA databases. Gliomas were classified into LGG and glioblastoma multiforme (GBM) in both databases (Table 2).

Univariate COX analysis revealed that tumor grade, age, isocitrate dehydrogenase (IDH) status, 1p/19q codeletion, biological type, and SLC22A1 expression were associated with OS in LGG patients (Fig. 2C). Multivariate analysis identified grade, age, IDH status, and SLC22A1 as independent risk factors for LGG (p < 0.05, HR > 1, Fig. 2D). SLC22A1 expression varied across clinical subgroups, including grade, 1p/19q codeletion, and biological type (Figs. 2E–2I). Kaplan-Meier analysis of CGGA data confirmed a significant association between increased SLC22A1 expression and worse prognosis in LGG patients (p = 0.009, Fig. 2J). Taken together, these results underscored the potential of SLC22A1 as a prognostic marker for LGG.

To understand the role of SLC22A1 in LGG, pathway enrichment analysis was conducted. As shown in Fig. 3A, the top 20 pathways from Gene Set Enrichment Analysis (GSEA) demonstrated that the high-expression SLC22A1 group was significantly associated with enhanced immune reactions, extracellular material recognition, and cell proliferation. Pathways included neutrophil degranulation, signaling by interleukins, interferon signaling, IL18 signaling pathway, extracellular matrix organization, cell surface interaction at the vascular, cytokine receptor interaction, extracellular matrix (ECM) regulators, mitotic metaphase and anaphase, mitotic prometaphase, cell cycle checkpoints, and separation of sister chromatids, among others.

The top 500 genes associated with SLC22A1 were annotated in biological processes and pathways related to immunoregulation and immune response. These included regulation of innate immune response, nuclear factor (NF-kappaB) signaling, response to interferon-gamma, alpha-beta T cell activation, positive regulation of response to cytokine stimulus, positive regulation of cytokine-mediated signaling pathway, NOD-like receptor signaling pathway, and TNF signaling pathway (Figs. 3B, 3C). Collectively, these results indicated that SLC22A1 might influence the prognosis of LGG through immunization routes.

Univariate COX regression analysis identified 26 genes among 33 immune checkpoints associated with survival time (Table 3). Utilizing LASSO and the COX proportional hazards model, we identified three immune regulatory factors with independent predictive value and established a risk score prognosis model (Figs. 4A–4D, with CD276 included due to its p value being close to 0.05). The risk score was calculated by summing the product of the coefficient and gene expression, resulting in the scoring formula: risk score = 0.5960 * TNFRSF14 + 0.7947 * CEACAM1 + 0.4087 * CD276. A higher number of LGG patients died in the high-risk score group, with the three risk factors showing increased expression in the high-risk group (Fig. 4E). Kaplan-Meier curves further validated that LGG patients with a low risk score had significantly longer survival times than those with a high risk score (p < 0.001, Fig. 4F), consistent with overall survival curves based on the CGGA database (p < 0.001, Fig. 4G). Additionally, the risk score of recurrent LGG patients in the CGGA database was significantly higher than that of primary LGG patients (Fig. 4H, p < 0.001), suggesting that a high risk score might indicate a higher probability of recurrence. Hence, the risk score holds clinical significance in evaluating the likelihood of relapse in LGG patients. Consequently, we established an immune checkpoint risk score model related to SLC22A1, and the survival outcomes and curves demonstrated that a higher risk score implied less favorable survival of LGG patients.

The hazard ratios of different patient features indicated that age, grade, and risk score were independent risk factors for the prognosis of LGG patients (Fig. 5A). We next exploited a nomogram model to predict the survival time of LGG patients using risk score and age, as illustrated in Fig. 5B. Calibration curves verified the reliability of the nomogram in predicting the 1-year, 3-year, and 5-year survival rates of LGG patients (Figs. 5C–5E). ROC curves were employed to assess the prediction accuracy of the nomogram, revealing AUC values of 0.890, 0.878, and 0.808 for 1 year, 3 years, and 5 years, respectively (Fig. 5G). In comparison to the AUC values of the risk score alone (Fig. 5F), this demonstrated that the nomogram had better accuracy. Furthermore, the prognostic model was tested with CGGA data, yielding AUC values of 0.807, 0.813, and 0.811 for 1 year, 3 years, and 5 years, respectively (Fig. 5H), indicating good accuracy in practical application. The results substantiated that the nomogram, established based on risk scores and age, exhibited a strong ability to predict the survival rate of LGG patients.

To verify the effect of silencing SLC22A1 on proliferation gliomas U251 cells, we reduced the expression of SLC22A1 using siRNA and detected the cell proliferation. In Fig. 6A, fluorescent labeled siRNA showed that 48 h incubation time was considered to exhibited high transfection efficiency. And RT-qPCR assay confirmed that mRNA level of SLC22A1 in gliomas U251 cells was successfully silenced by SLC22A1 siRNA (Fig. 6B). As shown in Fig. 6C, the proliferation rate of U251 cells treated with SLC22A1 siRNA was significantly lower than that treated with NC siRNA. And equally the cloning efficiency of U251 cells decreased after inhibition of SLC22A1 expression (Figs. 6D and 6E). As our prediction, the decrease of SLC22A1 RNA levels depressed the growth of gliomas U251 cells.