ABCA1 is involved in the progression of various cancers and pathological biological processes, including dyslipidemia, neurological disorders, glaucoma, and diabetes [8] However, the effect of ABCA1 on the heterogeneous TME of GAC has not been fully elucidated. Nevertheless, Liu et al. employed bioinformatics techniques to develop a lipid droplet metabolism model for GAC and revealed that upregulated ABCA1 might contribute to the development of unfavorable clinical characteristics in GAC [27]. Similarly, Yin et al. discovered a correlation between ABCA1 expression and drug metabolism. They also developed a predictive prognostic model for patients with GAC related to drug metabolism [28]. However, these previous studies concentrated primarily on bulk RNA-seq data without delving into the pathogenic processes of ABCA1 using scRNA-seq. Furthermore, these studies failed to verify the effect of ABCA1 in GAC through in vitro assays. In contrast, the current study combined bulk RNA-seq, scRNA-seq, and in vitro studies to investigate the putative oncogenic pathways of ABCA1 in GAC. As such, the findings of this study contribute to the establishment of a theoretical foundation for future targeted therapies utilizing ABCA1 in the treatment of GAC.

First, we analyzed the GAC bulk-seq data in TCGA-STAD and five GEO cohorts. The results revealed that ABCA1 was highly expressed in GAC than in normal gastric tissue, while ROC curves confirmed that it had good diagnostic value, suggesting that ABCA1 may function as an oncogene in GAC. This differential expression in ABCA1 was further verified by in vitro experiments. High ABCA1 expression was generally associated with adverse clinical features and poor OS in GAC. Additionally, multivariate Cox analysis suggested that ABCA1 was an independent risk factor. Therefore, to enhance the application value of ABCA1 in the screening of high-risk patients with GAC, we constructed an ABCA1-nomogram for the accurate and reliable prediction of OS among patients with GAC.

In addition, GSEA was performed to elucidate the potential mechanisms underlying the oncogenic functions of ABCA1. Previous studies have demonstrated that EMT activation, angiogenesis, and IL-6/JAK/STAT3 pathways increase the risk of tumor invasion, metastasis, and drug resistance [29,30,31]. For instance, Prijic et al. revealed that EMT regulates ABCA1 expression in breast cancer cells [32]. Meanwhile, Fang et al. found that ABCA1 knockdown leads to dysregulated angiogenesis in zebrafish [33]. In another study, the overactivation of the IL-6/JAK/STAT3 pathway was found to promote tumor proliferation and invasion while suppressing anti-tumor immune responses [30]. Frisdal et al. observed that IL-6 can stimulate ABCA1 expression by activating the JAK-STAT3 pathway, regulating macrophage cholesterol homeostasis [34]. Our findings also indicate that the complement, inflammatory response, KRAS signaling, and TNF-⍺ signaling via NF-κB pathways were activated in patients with GAC and high ABCA1 expression. Magrini et al. reported that complement activation not only induces the formation of an inflammatory TME and suppresses anti-tumor immunity, but also promotes various processes, such as tumor angiogenesis and EMT, accelerating the invasion and migration of tumor cells [35]. The important role of the KRAS signaling and TNF⍺ signaling via NF-κB pathways in tumor progression has garnered significant attention [36, 37]. However, further investigation is needed with a focus on the relationship between ABCA1 and these pathways in GAC.

The TME is a complex and heterogenous environment characterized by hypoxia, low pH, and numerous growth factors and proteolytic enzymes. It also contains myriad tumor cells, immune cells, and interstitial tissues [38]. Our study demonstrated that ABCA1 upregulation impacted the infiltration of TICs in the GAC TME, especially with respect to the increased recruitment of M2 macrophages. M2 macrophages are associated with the elevated expression of factors that promote angiogenesis and stimulate tumor cell proliferation in vivo, including matrix metalloproteinases, IL-10, and vascular endothelial growth factor [39]. Indeed, Guo et al. reported that M2 macrophages facilitate cancer cell invasion by promoting EMT and αB-crystallin expression [40]. Furthermore, ABCA1 expression negatively correlated with plasma cells and T helper cells; plasma cells have a positive role in anti-tumor immunity [41]. Meanwhile, T helper cells secrete various cytokines, including IL-2, TNF-⍺, and IL-12, which enhance natural killer cell-mediated immune responses to exert anti-tumor effects [42]. This aligns with apparent immune response suppression elicited by ABCA1 overexpression in GAC.

Drug resistance is a major constraint in the application of chemotherapy. Therefore, exploring biomarkers that can predict the sensitivity to anti-tumor drugs has significant implications for improving the outcome of precision treatment and prolonging the OS of patients with GAC [43]. We found that patients with high ABCA1 expression were more likely to exhibit resistance to cisplatin, etoposide, mitomycin C, and sorafenib, and were more sensitive to imatinib, sunitinib, shikonin, and paclitaxel. Prior research has elucidated the underlying mechanism by which ABCA1 modulates the sensitivity of malignancies to cisplatin. For instance, Chen et al. revealed that reduced ABCA1 expression can augment the susceptibility of lung cancer cells toward cisplatin through a mechanism governed primarily by Valproic acid [44]. Furthermore, Wang et al. found that ANXA2 overexpression facilitates the emergence of resistance to etoposide in neuroblastoma by activating the NF-κB signaling pathway [45]. Indeed, a positive association has been reported between mitomycin C resistance in bladder cancer and activation of the EMT and NF-κB signaling pathways [46]. In conjunction with the enrichment analysis conducted in the current study, it was observed that the EMT and NF-κB pathways exhibited frequent enrichment in GAC patients with high ABCA1 expression. This finding suggests that ABCA1 overexpression may serve as an underlying mechanism contributing to the development of resistance against etoposide and mitomycin C. In addition, ABCA1 overexpression could facilitate the activation of the IL6/JAK/STAT3 signaling pathway. Similarly, Jiang et al. reported that IL-6 activation in liver cancer leads to an increase in STAT3 expression, resulting in heightened resistance to sorafenib treatment. Moreover, Leporini et al. evaluated the use of imatinib as an anti-angiogenic modulator in patients with GAC and bone metastases [47]. Meanwhile, Hojo et al. reported that combined sunitinib and pterostilbene therapy exerts significant anti-tumor effects in GAC [48]. The combination of sunitinib with S-1 and cisplatin has also been applied in treating GAC [49]. Additionally, shikonin reportedly suppresses the invasion and activity of GAC cells; its underlying mechanism may be related to the Toll-like receptor 2/NF-κB-mediated pathway, and reactive oxygen species [50, 51]. In contrast, paclitaxel acts on tubulin in cancer cells, promoting microtubule polymerization and leading to cell division suppression [52]. It has also exhibited good therapeutic efficacy in GAC in several clinical trials [53,54,55]. However, further research is needed to elucidate the drug resistance mechanism of ABCA1 in GAC.

Recently, immunotherapy has fundamentally altered the direction of tumor treatment; PDCD1/PDCD1 ligand 1 (PDCD1LG1) drugs have been applied in first-line clinical trials for GAC [56, 57]. However, the application of immunotherapy is predicated on selecting clinically beneficial groups. Our study revealed that patients with GAC and high ABCA1 expression exhibited upregulation of immune checkpoints, such as PDCD1, CTLA4, LAG3, HAVCR2, and PDCD1LG2, suggesting that adjustments may be needed to the immunotherapy strategies of patients with high ABCA1 expression to achieve beneficial therapeutic outcomes.

To our knowledge, this study is the first to identify the ABCA1 features of GAC at the single-cell level. Based on scRNA-seq data, we confirmed ABCA1 expression in the T cell, plasma cell, endothelial cell, mesenchymal stromal cell, B cell, macrophage, and epithelial cell subpopulations, among which, macrophages exhibited the highest expression. However, ABCA1 expression was negligible in the cells of normal gastric tissues. Similarly, Hoppstädter et al. observed that, compared with normal lung tissues, the elevated ABCA1 expression in tumor-associated macrophages of lung adenocarcinoma promotes cholesterol efflux from macrophages [58]. A similar phenomenon was observed in a mouse model of ovarian cancer, wherein the cholesterol efflux mediated by ABCA1 upregulation promoted the transition of macrophages to M2 macrophages [59]. The exported cholesterol may support tumor cell growth [59, 60]. In the current study, cell trajectory analysis demonstrated that ABCA1(+) macrophages appeared primarily toward the end of the differentiation trajectory, consistent with the differentiation direction of GAC cells from the superficial to deep layers. Moreover, ABCA1 expression increased with the progression of GAC infiltration, implying that ABCA1 upregulation may promote the onset and development of GAC. Furthermore, the expression pattern of genes in Cluster 4 was consistent with that of ABCA1, and enrichment analysis revealed that they were involved primarily in TNF-⍺ signaling via NF-κB, complement, inflammatory response, KRAS signaling, and complement and coagulation cascade pathways. These findings agree with the bulk-seq data-based GSEA, suggesting that these pathways represent the key biological mechanisms for promoting GAC invasion and metastasis by ABCA1(+) macrophages.

Cell–cell communication analysis revealed significant differences between the cell communication networks of ABCA1(+) and ABCA1(−) macrophages. That is, when the macrophages acted as signal transmitters, the CALCR (ADM-CALCRL) pathway was only enriched in the ABCA1(+) macrophage subpopulation. However, when the macrophages acted as signal receivers, significant differences were observed between the communication networks of ABCA1(+) and ABCA1(−) macrophages. More specifically, the SPP1 (SPP1–[ITGAV + ITGB1], SPP1–[ITGA5 + ITGB1]), PTN (PTN–SDC4), PROS (PROS1–AXL), visfatin (NAMPT–[ITGA5 + ITGB1]), MIF (MIF–[CD74 + CXCR4]), MK (MDK–SDC4), galectin (LGALS9–HAVCR2), GAS (GAS6–AXL), complement (C3–C3AR1), and ANGPTL (ANGPTL2–[ITGA5 + ITGB1]) pathways were significantly enriched in the ABCA1(+) macrophage subpopulation. The ligand–receptor pairs in the CALCR, SPP1, PROS, MIF, MK, galectin, GAS, and complement pathways are closely associated with tumorigenesis, while ADM can promote the angiogenic effect of epithelial cells by binding to the CALCRL receptor [61]. Larrue et al. found that activation of the ADM-CALCRL axis in acute myeloid leukemia predicts an unfavorable prognosis and is associated with chemoresistance [62]. Additionally, SPP1 and ITGB1 participate in GAC chemoresistance via the ITGB1/YBX1/SPP1/NF-κB pathway [63]. Meanwhile, in a fibrotic microenvironment, the binding of SPP1 to ITGAV inhibits the apoptosis of lung cancer cells [64]. Further, Zhu et al. demonstrated that ITGA5 overexpression promotes poor prognosis in GAC [65]. In the PROS pathway, the PROS1 ligand and its receptor AXL are overexpressed in thyroid cancer, while AXL blockade suppresses the invasion of thyroid cancer cells [66]. Within the TME, the tumor homing of mesenchymal stromal cells is regulated by the MIF-(CD74 + CXCR4) ligand–receptor pair in the MIF pathway [67]. The MDK-SDC4 ligand–receptor pair mediates the interaction between fibroblasts and ovarian cancer cells, and is closely associated with the survival prognosis of ovarian cancer [68]. In addition, the LGALS9-HAVCR2 ligand–receptor pair is an important target for mediating immune escape in various cancers, including leukemia, breast cancer, and melanoma [69, 70]. Extensive research has also been conducted on the GAS signaling pathway in cancer. For example, Zhu et al. reported that the GAS6-AXL axis promotes tumor cell invasion, drug resistance, and mitosis by activating downstream pathways, such as the PI3K-AKT-mTOR and NF-κB pathways [71]. Furthermore, the complement system is often dysregulated in the TME. Lawal et al. found that C3-C3AR1 is involved in immune escape associated with dysfunctional T cell phenotypes [72]. However, little is known regarding the role of the PTN (PTN–SDC4), visfatin (NAMPT–[ITGA5 + ITGB1]), MK (MDK–SDC4), and ANGPTL (ANGPTL2–[ITGA5 + ITGB1]) ligand–receptor pairs in tumors; this requires further investigation. In summary, the ABCA1(+) macrophage subpopulation may accelerate the progression of GAC via signaling pathways mediated by these ligand–receptor pairs.

The dysregulation of TFs greatly increases the risk of tumor progression. Therefore, drug targets tailored to TFs have been widely employed in cancer therapy [73]. In this study, the SCENIC algorithm was employed to identify the five main TFs specific to the ABCA1(+) macrophage subpopulation (NR1H3, MITF, TFEC, MAFB, and CEBPB). Wang et al. reported that microRNA 585, CAMP-responsive element binding protein 1, and mitogen-activated protein kinase 1 inhibit GAC cell proliferation and invasion by downregulating MITF [74]. In contrast, NR1H3 upregulation activates hypoxia-induced EMT, reduces the survival rate of GAC patients [75], and promotes the metastasis of renal cell carcinoma by regulating NOD-, LRR-, and pyrin domain-containing protein 3 inflammasomes [76]. TFEC dysregulation is also associated with the progression of various tumors [77]. For example, Samir et al. demonstrated that MAFB serves as a biomarker of poor prognosis in lung adenocarcinoma [78]. In GAC, CEBPB upregulation enhances the proliferative activity of cancer cells [79]. Moreover, using the TIMER database, we confirmed that these five TFs were significantly positively correlated with ABCA1 expression. Taken together, these findings provide new insights into the molecular mechanisms underlying the ABCA1 promotion of GAC progression.

The Scissor algorithm was used to identify cell subpopulations associated with known phenotypes [26]. Combined with the clinical phenotypes in TCGA-STAD cohort, we found that Scissor + cells associated with the “Dead,” “Tumor,” “Grade 3,” and “N1–3” phenotypes exhibited a higher distribution of ABCA1(+) macrophages. Conversely, ABCA1(−) macrophages were primarily distributed in Scissor- cells associated with the “Alive,” “Normal,” “Grade 1–2,” and “N0” phenotypes. This suggests that infiltration of the ABCA1(+) macrophage subpopulation in the TME may exacerbate GAC progression and poor prognosis.

ABCA1 silencing significantly inhibits the migration and invasion of ovarian cancer cells [13]. Similarly, our study revealed that ABCA1 expression was significantly higher in GAC cells than in normal gastric cells, and was highest in the HGC-27 cell line. Furthermore, ABCA1 knockdown significantly impeded the proliferation, invasion, and migration of HGC-27 cells.

The clinical importance of anti-cancer medications that specifically target the metabolic processes of cholesterol has been demonstrated. In particular, ABCA1 plays a significant role in the context of cellular cholesterol excretion processes. The regulation of intracellular cholesterol levels by ABCA1 has an indirect impact on the proliferation, migration, and invasion of cancer cells [80,81,82]. In particular, lipid rafts, which are abundant in cholesterol, have a significant impact on the adhesion and migration of cancer cells [83, 84]. Moreover, the activity of phosphoinositide 3-kinase exhibits a regulatory effect on the expression of surface proteins, including ABCA1. This regulatory mechanism has been associated with a notable increase in the likelihood of cancer cells entering the circulation and subsequent metastases [85]. Moreover, as a result of the intricately orchestrated sequence of oncogenic gene mutations in cancer, which give rise to diverse metabolic modifications, ABCA1 has been recognized as a gene that exhibits a synergistic response [86]. That is, molecules in traditional signaling pathways are associated with the action of ABCA1. In their study, Wang et al. discovered that activation of the ERK/c-Jun pathway by nicotinamide N-methyltransferase increases ABCA1 expression. This, in turn, reduces cholesterol levels within cancer cells, facilitating invasion and migration in the context of triple-negative breast cancer. Moreover, ABCA1 overexpression induced by nicotinamide N-methyltransferase is a notable factor that drives EMT [87]. Similarly, the pathway involving ERK/Fra-1/ZEB1 is responsible for promoting thyroid cancer cell invasion and progression through EMT, facilitated by ABCA1 [88]. Furthermore, Prijic et al. revealed that the upregulation of ABCA1 expression in breast cancer cell line EMT is mediated by MYC by eliminating its proximal E-box elements [30]. The current study showed an enrichment of EMT pathways in patients with GAC and high ABCA1 expression. This aligns with the outcomes of prior research studies. Indeed, the close relationship between the EMT pathway and the unfavorable evolution of malignant tumors involving ABCA1 is evident. Accordingly, we hypothesize that the modulation of ABCA1 could represent a significant advancement in treating GAC by regulating EMT or pathways associated with cholesterol metabolism. Nevertheless, the precise mechanisms underlying the effects of ABCA1 remain uncertain, necessitating additional investigations into its role in the onset, progression, infiltration, and dissemination of GAC. Furthermore, the existing body of research pertaining to the therapeutic targeting of ABCA1 in tumor treatment is limited, with a dearth of pertinent clinical trials that substantiate the potential benefits of ABCA1 targeting, specifically in terms of progression-free survival and OS. Accordingly, clinical trials are underway to assess the efficacy of ABCA1-targeted treatment in malignant tumors across different cancer types.

The present study has certain limitations. First, we primarily investigated the biological impacts of ABCA1 by utilizing publicly accessible data and in vitro cell studies. Hence, further in vivo trials are needed, as well as theoretical studies on the upstream and downstream signaling pathways that mediate ABCA1. Second, it is imperative to validate the prognostic significance of ABCA1 in GAC through extensive clinical investigations involving a large cohort of patients with GAC.

This work aims to provide a comprehensive understanding of the association between ABCA1 and macrophages in GAC, shedding light on the significance of ABCA1(+) macrophages within the GAC TME. Moreover, we contend that investigating the involvement of ABCA1(+) macrophages through scRNA-seq technology across various subtypes and stages of GAC will serve as a prospective avenue for future research.