Esophageal cancer is the eighth most common malignancy and the sixth cause of cancer death globally, which accounts for more than 570,000 new cases and 500,000 deaths annually (Bray et al., 2018; Wang et al., 2018). Esophageal adenocarcinoma (EAC) is the predominant pathological type in western countries, with an increasing proportion from 35 to 61% over the past 30 years (Alsop and Sharma, 2016). The global incidence rate of EAC is approximately 0.7/100,000 person-years, and the 5-year survival rate is merely less than 20%, although multidisciplinary treatments have been applied, including esophagectomy, radiation, and chemotherapy (Arnold et al., 2015; Markar et al., 2017; Smyth et al., 2017; Zhao et al., 2019). Considering the chemotherapeutic resistance, several targeted agents have been applied in patients with EAC, such as imatinib (Mayr et al., 2012). Unfortunately, the efficacy is still not satisfactory. Recently, immunotherapy-targeting PD-1 has revolutionized the therapy in cancer patients. However, not all patients with EAC respond to immunotherapy (Däster et al., 2020). Therefore, there is an urgent necessity to better understand the molecular characteristics and genetic features that could help predict accurate survival and identify suitable patients who will benefit from immunotherapy.

The tumorigenesis and development of EAC is a highly complex biology, involving the tumor cell-intrinsic and cell-extrinsic factors (Quante et al., 2018; Talukdar et al., 2018). Genetic alterations are the primary mechanisms that drive the initiation and progression of EAC, not only conferring tumor cells infinite proliferative abilities but also reprogramming metabolic pathways to adapt to the hostile environment, such as aerobic glycolysis (Hochwald and Zhang, 2017; Talukdar et al., 2018). The seminal discovery of tumor glycolysis has been considered a hallmark of cancer, proposed by Otto Warburg in 1923 (Warburg and Minami, 1923). The glycolytic phenotype renders cancer cells selective advantages by unlimited growth and attenuated apoptosis (Xu et al., 2017). In addition, it is gradually evident that elevated glycolysis is closely related to the immune escape by changing the microenvironment and inhibiting the functions of immune cells (Jiang et al., 2019a). Mounting evidence from cell-based assays has linked glycolysis to TME, and preclinical investigations have demonstrated the effectiveness targeting glycolysis in some cancers (Lim et al., 2017; Kornberg et al., 2018; Jiang et al., 2019a; Jiang et al., 2019b; Kang et al., 2020).

Genetic mutation results in the rewire of the glucose metabolism decreased cancer cell apoptosis and immortal growth. Consequently, these events bring about the component reconstruction in the tumor microenvironment (TME), thus changing the purity of the tumor. Reciprocally, the intimate interactions between glycolytic cells and the extracellular matrix further exacerbate the remodeling of TME, including the stromal and immune cells. It is well accepted that the tumor is highly dependent on TME, which is preponderant on prognosis and impacts the therapeutic efficacy profoundly, such as the immune checkpoint therapy (Wu and Dai, 2017; Alsina et al., 2018; Taube et al., 2018; Hinshaw and Shevde, 2019; Li et al., 2020). However, there are few studies exploring the associations between glycolysis and TME in EAC, and far less is known about how genetic mutations orchestrate the glycolysis under these aberrant TME conditions. Herein, we investigated the effects of glycolysis on immune cells and revealed the genetic mutation diversity. Our study unraveled that glycolysis could influence TME under the driver of genetic mutation and could serve as prognostic biomarkers. Moreover, we constructed the risk score system to predict drug sensitivity and immunotherapeutic response. The results hold great promise in targeting glycolysis and utilizing TME to improve the treatment in patients with EAC.

The abilities of cancer cells to switch metabolisms and evade the immunity system in TME are well-documented characteristics in tumors. Elevated glycolysis is commonly observed in cancer progression and is associated with significant disruptions of a previous finely tuned microenvironment (Chang et al., 2015; Jiang et al., 2019a). As the tumors develop, they constantly interact with neighboring cells, such as stromal cells and immune cells, under the driver of genetic mutations, thus altering their phenotypes and functions (Butturini et al., 2019). In the context of these intricate crosstalks between malignant cells and non-malignant cells, the impacts of increased glycolysis and a dysregulated TME on immune response and effective therapy are of vital importance (Roma-Rodrigues et al., 2019). Although the research studies focusing on the tumor glycolysis and TME have exploded exponentially in recent years, the underlying mechanisms of how they act both independently and synergistically still remain elusive. In this study, we explored systematically the links between glycolysis and TME in patients with EAC as well as the relations with genetic mutations. In the present study, we found that the glycolytic level was higher in the ESTIMATE ^high group, reasonably reflecting the fact that glycolysis may change the tumor purity. By constructing the predictive survival model based on IG signatures, we discovered that three genes (NUP88, SEH1L, and NUP37) may serve as independent prognostic biomarkers for OS in EAC. In addition, we revealed that gene mutation types and frequencies were distinct between the different ESTIMATE score groups, lending us a hypothesis that genetic alteration may drive TME changes. In addition, our data suggested that glycolysis could influence drug sensitivity and immunotherapeutic response.

Several studies have confirmed the close relationships between glycolysis and EAC (Lynam-Lennon et al., 2014; Harada et al., 2020; Kang et al., 2020). Consistent with these findings, our functional enrichment analysis results showed that IGs were strongly enriched in the glycolytic processes, ATP generation, the HIF−1 signal pathway, RNA activities, and so on. The presence of aerobic glycolysis under normal conditions efficiently promotes tumor cell growth by the following mechanisms: 1) the rate of glycolysis is accelerated at 100 times compared to oxidative phosphorylation to compensate for the mathematical disadvantage in terms of net ATP production (glucose oxidative phosphorylation although mitochondria generate 18 times ATP compared to glycolysis) (Pfeiffer et al., 2001); 2) glycolysis provides sufficient and essential intermediates, such as NADPH and ribose-5-phosphate, which are indispensable for biosynthesis to meet avid proliferative requirements (Boroughs and DeBerardinis, 2015); 3) lactate, the obligatory product derived from glycolysis, could activate the HIF−1 signal pathway to induce vascular endothelial growth factor (VEGF) expression to stimulate angiogenesis in the microenvironment (Sonveaux et al., 2012). In addition, lactate is crucial to reorganize the tumor physical architectures in TME, and the accumulation of extracellular lactate is detrimental for normal healthy cells, such as immune cells (Romero-Garcia et al., 2016; Cassim and Pouyssegur, 2019). The study about how RNA interacts precisely with glycolysis is still in its infancy. However, Hua Q et al. gave us a hint that long non-coding RNA may promote glycolysis by sponging miRNA (Hua et al., 2019). Hence, it is reasonable to speculate that certain mutational genes regulated and modified by RNA may be involved in the glycolysis in EAC (Hochwald and Zhang, 2017).

The prognostic signature of glycolysis in EAC was established based on three genes (NUP88, SEH1L, and NUP37). NUP88, located at chromosome 17p13, encodes Nup88 protein (Zhao et al., 2012). Nup88 is a nucleoporin comprising nuclear pore complexes (NPCs) and plays critical roles in maintaining the spindle stability and preventing aneuploidy formation during mitosis (Hashizume et al., 2010). Unanimously, the GO results in our study also illustrated that glycolytic genes had intimacy with the nuclear pore, emphasizing the importance of nuclear proteins in glycolysis. The Nup88 overexpression is highly associated with tumor development and decreased survival, suggesting that NUP88 acts as an oncogene (Martínez et al., 1999; Naylor et al., 2016). In line with these studies, our results also proved that NUP88 was a risk factor for OS (HR > 1). Another prognostic gene is NUP37, which shares similarities with NUP88 and is also a member of NPC. It exerts primary functions of sustaining NPC integrity and modulates the cell cycle (Chen et al., 2019). Previous studies showed that the elevated expression of NUP37 is associated with worsened survival rates in liver cancer (Uhlen et al., 2017). In addition, a latest study by Huang L et al. demonstrates that NUP37 silencing induces inhibition of lung cancer cell proliferation (Huang et al., 2020). These findings were in agreement with our results, pointing out that NUP37 played an oncogenic role in OS (HR > 1). Nonetheless, the function of NUP37 in EAC has never been explored and needs further experiments to confirm in vitro and in vivo. SEH1L, also known as Seh1, is a part of NPC as well. However, the field is still in its infancy, and only a handful of animal models have been developed to investigate the role of SEH1L. Studies have shown that Seh1 could promote oligodendrocyte differentiation (Liu et al., 2019; Raices and D’Angelo, 2019). However, little is known about how it works in EAC. Undeniably, more research studies are warranted to understand the specific effects of SEH1L on EAC.

The notion that glycolysis has profound impacts on immune cells in TME is well recognized. The investigation linking metabolic demands and immune cells was first documented in neutrophils and macrophages (Alonso and Nungester, 1956; Newsholme et al., 1986). Immune cells hold a resemblance with tumor cells to engage glycolysis, which require rapid energy sources to produce immune-mediators for migration and phagocytosis. Our study demonstrated that NUP88 had significant correlation with B-cells and mast cells (all p < 0.05) and NUP37 had positive correlation with T-cells and negative correlation with dendritic cells (DCs) and macrophages (all p < 0.05). These findings enforced the concept that glycolysis contributes to dramatic alterations of immune cells in TME, hence influencing the immune response and immune-based treatments. Glycolytic tumor cells compete with T-cells for glucose and impair T-cell activation (Peng et al., 2016). In addition, IFN-γ secreted by Th1 cells is sensitive and is unstable to lactate. Noteworthily, this could polarize the T-cells toward differentiating into the Th2 subpopulation that favors tumor progression by inhibiting the antitumor effect of M1 macrophages (Kareva and Hahnfeldt, 2013). Paradoxically, direct evidence from in vivo experiments supports that glycolysis could promote Th1 cell differentiation through an epigenetic mechanism (Peng et al., 2016). This calls for further studies aiming to shed light on the characteristics of glycolysis on T-cells. The divergent effects of glycolysis on DC are also observed. Glycolysis produces excessive lactate and lowers the pH in TME. Lactate together with decreased pH suppresses DC differentiation and consequently abolishes the antigen-presenting functions to T-cells (Harmon et al., 2020). However, inhibition of glycolysis also blocks DC maturation through HIF−1α. This highlights the multiple roles of DC in glycolysis and needs to be carefully interpreted in a context-dependent manner.

Gene mutations facilitate tumor cell metabolic plasticity to create favorable microenvironments beneficial to uncontrolled proliferation. Deepening our knowledge about the differences of gene alterations between different TMEs may allow for the development of therapeutic strategies. Spurred by this promising target, we explored upon this issue and showed that ESTIMATE ^high and ESTIMATE ^low groups manifested different gene mutation profiles, in which TP53 and TTN were the most prevalent mutant genes. Moreover, the most common type is missense mutation. Mutant TP53 endows tumor cells with adaptabilities to cope with the harsh microenvironment by providing adequate nutrients, thereby escaping from antitumor immune attack. Experimental mice models in the study by Basu S et al. testified that mutant TP53 could rewire the tumor glycolytic metabolism and enhance metastasis in TME (Basu et al., 2018). Mutant TP53 has additional impacts on TME beyond changing tumor metabolic phenotypes. Myriad studies have indicated that mutant TP53 can remodel TME by several mechanisms. First, mutant TP53 could induce neo-angiogenesis by stimulating VEGF secretion (Kieser et al., 1994). Second, mutant TP53 could regulate chemokines that are involved in the homeostatic microenvironment (Yeudall et al., 2012). Last but not the least, mutant TP53 could reprogram the infiltrating immune cells and reshape the microenvironment (Cooks et al., 2018). TTN is located on chromosome 2q31, consisting of 364 exons, and it is the longest described coding gene (Chauveau et al., 2014). TTN ranks the third most abundant protein in both cardiac and skeletal muscle tissues, followed by actin and myosin (Chauveau et al., 2014). However, its mutation is not a rare entity in various cancers. Consistent with prior investigation by Cheng X et al., we also found that TTN missense mutation was a rather frequent type (Cheng et al., 2019). In addition, TTN and TP53 co-mutation is often accompanied during tumorigenesis and may serve as a prognostic biomarker either alone or in combination ^[61-63].

Current guidelines recommend adjuvant chemotherapy for patients at an advanced stage. However, how to select suitable patients who will benefit more from the chemotherapeutic regime is the prior concern. Our data demonstrated that patients with a high-risk score were more sensitive to camptothecin, gemcitabine, methotrexate, and vorinostat, suggesting that targeting glycolysis may alleviate the chemotherapeutic resistance. Despite immunotherapy bringing about a breakthrough for cancer patients, only a minority of patients could reap survival benefits actually. The TMB analysis in our study will accurately and effectively identify which patients will respond to immunotherapy in patients with EAC. Collectively, the proposed risk score system in our study has potency to help clinicians devise an individualized treatment strategy.

The strength of the present study is such that we performed a systematic analysis about glycolysis and TME in EAC for the first time based on the National Public Database, which provides robust data and statistical support. This study draws a close link between tumor glycolysis and the microenvironment and tentatively explains the mechanisms from the viewpoint of genetic alteration. Meanwhile, there are several limitations. First of all, TME is a complex mixture of parenchymal cells, the extracellular matrix, and numerous cytokines except for tumor and immune cells. These are not available from the public database and may greatly affect the analysis. Second, the results are not validated in in vitro and in vivo experiments. Last, the methods proposed in this study may not be applicable to all tumors as a result of heterogeneity. Notwithstanding its limitations, our study does provide an overview of glycolysis and TME in EAC, and this lays the foundation for further basic research in the area of metabolism and the microenvironment.

In summary, we found that glycolysis could change the microenvironment under the driver of genetic mutation and influence the immunotherapy in EAC. New efforts target that EAC should incorporate the idea that the glycolytic metabolism could reshape TME. Further studies are necessary to confirm our conclusion.