Introduction
Pancreatic cancer (PC) is one of the most malignant cancers. According to GLOBOCAN, nearly 466,000 people die from PC annually, which is almost equal to the number of diagnosed cases (496,000). The 5-year survival rate is < 8%, which makes PC the seventh most important cause of cancer-related death worldwide [
1]. The deep anatomical position of the pancreas and the lack of typical symptoms and effective diagnostic methods in the early stage cause nearly 80%–90% of the patients to miss the optimal surgical treatment period [
2,
3]. In addition, the benefits derived by patients with PC from comprehensive treatment strategies, such as systemic chemotherapy and targeted therapy, are extremely poor [
4]. Therefore, the molecular mechanism of PC needs to be further elucidated.
Mitochondria are biological organelles with self-replicable and transcribable DNA, and mitochondrial DNA (mtDNA) lesions are an important cause of disease [
5]. Furthermore, transport of metabolites through mitochondrial membrane is a critical phase in the process of mitochondria partaking in cellular oxidation and energy output, and this step is heavily regulated by inner membrane [
6]. The solute carrier family is chiefly located on the membranes of cells and organelles and is the second largest group of membrane transporters after G protein-coupled receptors [
7]. SLC25A members, the largest solute carrier group of the solute carrier family, contains a highly specific sequence of six counter-clockwise-aligned transmembrane α-helices and three repetitive short α-helices [
8]. This sequence links several metabolic processes (e.g., oxidative phosphorylation and citric acid cycle) among different cellular compartments (cytoplasmic matrix and mitochondrial matrix) by mediating solute translocation across the membrane [
9]. Recent studies have shown that SLC25A members are involved in multiple processes of tumorigenesis, including metabolic reprogramming, mitochondrial apoptosis, maintenance of cellular redox homeostasis, the transformation of metastatic phenotypes, and promotion of cancer stem cell stemness and chemoresistance [
10‐
18]. However, the biological role of SLC25A members and its mechanism of action in PC have not been entirely elucidated.
In the present study, differences in the expression and function of SLC25A members in PC and its correlation with tumor immune infiltration were analyzed. Furthermore, a prognostic risk model was constructed using public databases. In addition, the functions of the identified SLC25A members (SLC25A11, SLC25A29, and SLC25A44) in PC were explored, and transcription factor (TF) networks were created based on them. Considering the heterogeneity of the tumor cells, the role of SLC25A members in the tumor microenvironment (TME) was also analyzed using single-cell RNA-seq (scRNA) data.
Discussion
In this study, based on the differential expression of SLC25A members in PC, its key role and potential mechanism in carcinogenesis and PC progression were analyzed using multi-omics techniques. A total of 38 SLC25A genes were found to be differentially expressed in PC, and some of them showed prognostic value in the disease. Subsequently, a prognostic risk model was constructed, which showed good predictive ability in both the training set and the validation cohort. Finally, the roles of SC25A11, SLC25A29, and SLC25A44 in PC were explored, and all three genes were found to be up-regulated in PC tissues. Furthermore, their co-expressed genes were observed to be involved in the metabolic pathway of tumor tissues. The results of the immune infiltration analysis showed the presence of a significant correlation between the three factors and the abundance of various immune cell infiltrations. It is noteworthy that SC25A11, SLC25A29, and SLC25A44 were observed to mainly exist in the malignant PC cells, stromal cells, and immune cells in the public scRNA database and were closely related to the formation of TME. Moreover, SCX and NKX2-1 were speculated to be the most likely related TFs among the three. Finally, the sensitivity to cisplatin and SNS-134 was found to increase with the up-regulation of SC25A11, SLC25A29, and SLC25A44.
SLC25A members are crucial in controlling how normally occurring pancreatic cells operate physiologically. In addition to producing ATP, mitochondria in pancreatic cells also produce metabolic substrates for glucose stimulated insulin secretion (GSIS) [
19]. In this process, the phase by which pyruvate carboxylase transforms pyruvate into oxalacetic acid to promote the generation of crucial intermediates for TCA cycle function and insulin secretion is the limiting step [
20]. Mitochondria citrate carrier (
SLC25A1, CIC) contributes significantly to the above phases in two different mechanisms [
21]. On the one hand, it makes citrate transport to the cytoplasm, and then through the cascade reaction to produce malonyl coenzyme A, to inhibit carnitine acyltransferase 1, resulting in an accumulation of acyl coenzyme A in the cytoplasm, thereby stimulating pancreatic GSIS. Isocitrate dehydrogenase activation, on the other hand, can encourage isocitrate breakdown to create -ketoglutaric acid and nicotinamide adenine dinucleotide phosphate (NADPH), which further stimulate pancreatic GSIS. Additionally, itaconate, which is synthesized from cytoplasmic citrate, has the potential to suppress the production of inflammatory mediators, controlling the metabolic pathway in inflammation, according to recent findings [
15]. Therefore, we hope to reveal whether SLC25A members also play a potentially important role in pancreatic carcinogenesis and tumor immunity to some extent.
In this study, we hoped to clearly distinguish patients with PC via hierarchical clustering of non-negative matrix factorization (NMF). However, contrary to our expectations, although the new classification showed good prognostic significance in the TCGA-PAAD cohort, PCA revealed that it was not able to differentiate patients with PC based on mRNA expression levels. Therefore, a prognostic risk model for SLC25A members was established based on Cox multivariate analysis and lasso regression analysis. Validation was further performed on the PAAD-CA queue, and a rosette was constructed to calibrate the curve recognition model distinction. The rate of glycolysis has been reported to be higher in activated T cells as well as in differentiated Th1, Th2, and Th17 cells than in OxPhos [
22]. In this study, a higher infiltration abundance of activated T-cells, Th1, Th2, and Th17 immune cells was observed in the high-risk group and, interestingly, lower OxPhos and TCA cycle activity were seen. This observation suggests that the associated immune cells influence the metabolic pattern of tumors in the high-risk group. Recent studies have shown that glutamine-derived aspartate is translocated to the cytoplasm via the oncogenic activation pathway of
KRAS in pancreatic ductal adenocarcinoma cells. Subsequently, aspartate is converted to pyruvate and reduced NADPH via a sequential reaction, thus maintaining redox homeostasis in cancer cells [
17]. Notably, alanine_aspartate_glutamate metabolism was found to be more active in the low-risk group. This result implies that tumors in the low-risk group are more involved in the maintenance of redox homeostasis than those in the high-risk group, but further validation is needed.
Since Warburg first proposed the hypothesis that cancer cells produce lactate from glucose because of mitochondrial destruction in 1956, ATP in cancer cells mainly comes from glycolysis, cancer metabolism research has entered a new era [
23]. However, cancer energy metabolism is not as simple as assumed by Warburg. Studies have reported that inhibition of glycolytic ATP production by knocking down pyruvate kinase (PKM2) does not prevent tumorigenesis. This finding shows that mitochondrial OxPhos is still the main ATP supply pathway in various cancer cells despite the increase in glycolysis rate [
24,
25]. Based on this observation, OxPhos has been proposed as a therapeutic target for cancers [
26]. It is well known that in OxPhos, the functioning of proton pump and the production of ATP requires electron transport [
27]. Kang Jinhua showed that cytoplasmic Nicotinamide Adenine Dinucleotide (NADH) is the main source of electrons in cancer cells, in which electrons are transferred from cytoplasmic NADH to mitochondrial NAD transport via the malate–aspartic acid shuttle, and finally, enter OxPhos [
28].
SLC25A11 is mitochondrial malate–α-ketoglutarate carrier protein, which is one of the two reverse transporters of the malate–aspartic acid shuttle. Our study found that
SLC25A11 and its positively related genes are mainly involved in the mitochondrial OxPhos pathway in PC. This finding suggests that
SLC25A11 may enhance the rate of OxPhos and promote ATP synthesis in PC cells by participating in the transfer of NADH-reducing equivalents. Furthermore, its expression was found to be negatively correlated with focal adhesion pathway and cell-substrate adhesion. Buffet et al. observed that deletion of
SLC25A11 in a mouse model of metastatic paraganglioma mediated the acquisition of pseudohypoxic features, hypermethylated phenotype, and metastatic properties. They further demonstrated that
SLC25A11 could serve as a novel tumor suppressor gene [
29]. Moreover, the expression of ScRNA in the endothelial cells and fibroblasts of PC tissues was found to be considerably different from that in normal tissues. This finding implies that
SLC25A11 may potentially inhibit the transformation of tumors to invasive and metastatic phenotypes in PC.
Additionally, the results of immune infiltration analysis showed that the expression of
SLC25A11 was positively correlated with the recruitment of immune cells, such as CD8 + T, CD4 + T, and DC, which alludes that
SLC25A11 may also play a role in anti-tumor immunity. The carcinogenic effect of
SLC25A11 has been confirmed in lung adenocarcinoma, melanoma, and hepatocellular carcinoma [
30,
31]. Our study observed that the OS of patients with a high expression of
SLC25A11 was longer in PC. Similarly, Pan [
30] also showed that patients with hepatocellular carcinoma with a high expression of
SLC25A11 had a better prognosis. Therefore, combined with the literature findings,
SLC25A11 can be used as a favorable prognostic marker for PC.
The main physiological role of
SLC25A29 is to transport arginine and lysine to the mitochondria for the synthesis of mitochondrial proteins and the degradation of amino acids. Zhang found that the gene was significantly up-regulated in various cancer cells, especially in cancers with enhanced glycolysis [
32]. The knockdown of
SLC25A29 in cancer cells led to a reduction in mitochondrial-derived NO, which resulted in enhanced mitochondrial respiration and reduced glycolysis, thereby reversing the metabolic process. Our study too showed that
SLC25A29 was significantly up-regulated in PC. In addition, enrichment analysis showed that the co-expressed genes in PC were involved in the biological processes of mitochondrial respiratory chain complex assembly, protein extension, and oxidative phosphorylation. PC is mainly characterized by aerobic glycolysis [
33], which suggests that
SLC25A29 may cause metabolic remodeling in PC cells by participating in the transport of substrates required for NO synthesis. The results of the co-expression analysis indicated that
SLC25A29 may play a role in suppressing the activation of various immune cells. Analysis of immune infiltration further revealed that the gene may prevent the immune cells from being recruited to confer pro-tumor immunity.
Recent studies have shown that
SLC25A44 is associated with cerebrovascular diseases [
34,
35]. The role of this gene in cancer has not been entirely investigated. This study showed that
SLC25A44 is significantly overexpressed in PC and has prognostic value. The results of co-expression analysis explained its role in PC. Genes negatively correlated with
SLC25A44 were found to be mainly involved in adhesion binding, PC, cell cycle, HIF-1, p53, and Hippo pathway in PC. As a member of
SLC25A,
SLC25A44 has only recently been proven to transport cytoplasmic branched-chain amino acids (BCAAs) to the mitochondrial matrix and be chiefly involved in the catabolism of mitochondrial BCAAs [
36,
37]. According to previous literature, in the early stage of pancreatic ductal adenocarcinoma, tissue protein decomposition increases, and systemic circulating BCAAs increase [
38]. In a metabolic report on early pancreatic ductal adenocarcinoma, the researchers stated that leucine, a BCAA in acinar cells, is the main source of acetyl CoA [
39].
SLC25A44 facilitates the transport of cytoplasmic BCAAs to the mitochondrial matrix and eventually decomposes them into acetyl-CoA and succinyl-CoA, which are transferred to the TCA cycle. However, the oxidative utilization of BCAAs occurs in a small proportion of PCs, and other shunt pathways exist for the transport of catabolic derivatives of BCAAs to the mitochondria [
40]. More detailed metabolic flux analysis is needed to explain it.
Drug sensitivity analysis of SLC25A members showed that the drug sensitivity to SNS-314 increased with the up-regulation of
SLC25A11. SNS-314 is an aurora enzyme inhibitor that can inhibit tumor growth and enhance chemotherapeutic sensitivity by mediating aurora enzyme activity [
41]. A study has reported that AURKA may be a downstream target of the MEK/ERK pathway in PC [
42]. This finding suggests that
SLC25A11 may be involved in the MEK/ERK pathway. Cisplatin is one of the standard chemotherapeutic regimens for PC [
43]. Our study too found that sensitivity to this drug was up-regulated with the increase in
SLC25A29 expression. SLC25A29 may be involved in the movement of platinum drugs on the cell membrane. In addition, OSI-027 is a selective inhibitor of motor targets [
44]. In a xenograft mouse model of PDAC, OSI-027 significantly blocked the G0/G1 phase of the cell cycle and inhibited tumor cell proliferation; furthermore, OSI-027 synergistically enhanced gemcitabine cytotoxicity in vitro and in vivo [
45]. Interestingly, in our study,
SLC25A44 was found to enhance the sensitivity to OSI-027. This discovery may form the basis for new clinical combination therapies for PC.
Our study has certain limitations. This study was based on a comprehensive bioinformatics analysis of multiple public databases, which has provided only preliminary insight into the key role of SLC25A members in the progression of PC. However, more research is required to understand how it affects the OxPhos of pancreatic cancer cells and the interactions between OxPhos and the immune microenvironment. Secondly, we focused on the roles of SLC2511, SLC25A29, and SLC25A44 in PC. However, considering the presence of tumor heterogeneity, their potential mechanisms in other cancers must be further analyzed.