Tumors and the organization of the metabolic microenvironment

As has emerged so far, insulin can help create a biological environment favorable to neoplastic growth; the next step, now, is to understand what happens when a growth signal reaches the inside of the cell. This is where one of the most important communication networks in cell biology comes into play: the PI3K/AKT/mTOR pathway. Under physiological conditions, as seen in depth in the dedicated chapter 4, this cascade functions as a sophisticated control system that allows the cell to adapt growth, proliferation, metabolism, and protein synthesis to the availability of nutrients and signals from the environment. The problem arises when this system stops responding correctly to its normal regulatory mechanisms and remains persistently active. In tumor cells, in fact, genetic or functional alterations can transform a physiological pathway essential for cellular life into a circuit capable of sustaining uncontrolled growth, survival, motility, and disease progression (Wu et al., 2022) [12].

We can imagine PI3K, AKT, and mTOR as three stations placed along the same transmission line. When PI3K is activated, the signal is progressively transferred toward AKT and then toward a series of intracellular effectors, among which mTOR occupies a fundamental position. In a normal cell, this line does not remain constantly on: it is activated when needed and braked when the stimulus wanes. In a tumor, however, the hyperactivity of PI3K, the loss of function of negative regulators like PTEN, or activating alterations of AKT can contribute to keeping the circuit persistently operational. The result is that the cell continues to receive signals favorable to growth and survival even when, physiologically, it should slow down or stop (Jiang et al., 2024) [13].

mTOR represents one of the main convergence points of this system and operates through two functionally distinct complexes, mTORC1 and mTORC2. mTORC1 controls processes strictly linked to growth and protein synthesis; mTORC2, on the other hand, participates in the full activation of AKT, thus contributing to the maintenance of the signal. The relationship should therefore not be imagined simply as a row of dominoes falling one after the other. It is more like a network in which some elements located downstream can contribute to keeping those that precede them active. This organization helps to understand why the deregulation of the axis can become particularly effective in supporting the tumor cell (Ye et al., 2020) [14].

The clinical relevance of this hyperactivation emerges clearly in nasopharyngeal carcinoma. By analyzing 119 patients and comparing them with 30 healthy subjects, some researchers found greater expression of PI3K, AKT, and mTOR in tumor tissues. But the most interesting data concerned the behavior of the disease: higher levels of the three proteins were associated with more advanced tumor stages, greater lymph node involvement, recurrence, and distant metastases. Furthermore, PI3K correlated positively with AKT and AKT with mTOR, showing in human tissue the same functional continuity predicted by the molecular cascade. Patients characterized by greater expression of these components also presented lower survival and, in multivariate analysis, the overexpression of PI3K, AKT, and mTOR emerged as an independent prognostic factor (Yu et al., 2019) [15].

A similar picture was observed in oral squamous cell carcinoma. In tumor tissues, positivity for PI3K, AKT, and mTOR reached 58.8%, 62.5%, and 51.3% respectively, while in adjacent normal tissues it was only 10%, 15%, and 10%. The activation of the cascade was associated with advanced stage and lymph node metastases; furthermore, PI3K and AKT were associated with lower tumor differentiation. Particularly significant was the position of AKT: AKT positivity remained independently associated with worse survival, with a hazard ratio (an indicator that compares the risk of an event between two groups over time) of 2.156 (Gu, 2026) [16].

These observations help to understand an essential concept: the activation of PI3K/AKT/mTOR does not seem to be limited to making a tumor mass grow more rapidly. It can also contribute to the acquisition of characteristics that allow cells to become more aggressive and to disseminate. In lung carcinoma, for example, this cascade participates in the regulation of the epithelial-mesenchymal transition, or EMT (transformation of epithelial cells into more mobile cells, capable of moving and invading other tissues). During this process, the cell progressively loses some characteristics typical of epithelial tissue, including part of its adhesive properties, and acquires a more mobile and invasive phenotype. It is as if a cell normally bound to its place in the tissue progressively acquired the tools necessary to detach, move, and colonize other environments. Numerous molecular regulators converge on the PI3K/AKT/mTOR pathway, influencing this very behavior; its persistent activation has been linked not only to proliferation but also to invasion, metastasis, angiogenesis, and therapeutic resistance (Moghbeli, 2024) [17].

A particularly intuitive experimental example comes from thyroid carcinoma. Midkine, a growth protein indicated by the acronym MDK, is overexpressed in tumor cells and its expression increases with the progression of the disease. When researchers experimentally reduced MDK, the ability of tumor cells to migrate and invade decreased along with the levels of PI3K, AKT, and mTOR. But the most interesting step happened subsequently: by pharmacologically reactivating the cascade via IGF-1, migration and invasiveness were restored. In short, turning off the signal reduced the aggressive behavior of the cell, while turning it back on allowed the cell to recover it (Yuan et al., 2024) [18].

A similar principle emerges in triple-negative breast carcinoma. In these cells, microRNA (tiny single-stranded RNA molecules that do not produce proteins) is reduced, while the proto-oncogene MYB (a normal gene that helps cells grow and divide in a healthy way) is more highly expressed. Experimentally increasing the microRNA resulted in a reduction of proliferation, migration, and invasion; conversely, its silencing favored these phenomena. MYB proved to be a direct target of this microRNA and its overexpression was able to reactivate PI3K/AKT/mTOR and restore cellular proliferation. The experiment therefore shows how different molecular signals can converge on the same cascade and use it as a sort of common node through which to sustain malignant behavior (Tang et al., 2025) [19].

But a cell that proliferates does not only need growth signals. It must also procure energy and, above all, material with which to build new cells. For this reason, the PI3K/AKT pathway takes on an even broader significance: it connects the proliferative signal to the reorganization of metabolism. The evidence collected by some researchers shows how AKT can increase glucose entry into the cell through the regulation of GLUT1 and favor the activity of glycolytic enzymes. The tumor cell, therefore, does not use PI3K/AKT only as a command that tells it to grow: it uses the same circuit to reorganize its metabolic resources in function of growth (Fontana et al., 2024) [20].

Making the picture even more complex is the fact that this pathway can also be influenced by the cells surrounding the tumor. In osteosarcoma, cancer-associated fibroblasts were found to be capable of modifying the metabolism of neoplastic cells by increasing the expression of phosphoserine aminotransferase 1, PSAT1, an enzyme involved in serine biosynthesis. The increase in PSAT1 was accompanied by the activation of the PI3K/mTOR/S6K1 axis and greater migration and epithelial-mesenchymal transition. The silencing of PSAT1 reduced these phenomena, while the inhibition of mTOR via rapamycin limited both cellular migration and the formation of metastases in animal models (Feng et al., 2025) [21]. This demonstrates that PI3K/mTOR can also represent one of the points through which the surrounding microenvironment communicates with the neoplastic cell and modifies its behavior.

The centrality of this cascade becomes even more evident by observing what happens when it is experimentally blocked. In renal carcinoma, the dual inhibitor GNE-477 reduced the growth, viability, proliferation, migration, and invasion of tumor cells, simultaneously inducing apoptosis. The compound blocked the phosphorylation of various components of the cascade, including AKT, simultaneously interfering with mTORC1 and mTORC2. Above all, when researchers artificially restored a constitutively active AKT, the antitumor effect of the compound was substantially canceled. Conversely, genetically eliminating AKT produced effects similar to those produced by the drug (Ye et al., 2020) [14].

However, the very network structure of the cascade explains why blocking it pharmacologically is more difficult than it might seem. The inhibition of a single point can remove some physiological feedback and allow the cell to reactivate AKT or use alternative circuits. For this reason, inhibitors capable of acting simultaneously on PI3K and mTOR have been developed: the goal is to close multiple escape routes at the same time and obtain a more complete suppression of the signal. The available data show that these compounds can reduce proliferation and favor apoptosis, but pharmacological resistance, toxicity, and molecular differences between tumors continue to limit their clinical efficacy (Wu et al., 2022) [12]. Even in head and neck carcinoma, in fact, PI3K, AKT, and mTOR inhibitors have produced heterogeneous results, and the blocking of a single node can be followed by the compensatory reactivation of the cascade (Jiang et al., 2024) [13].

Taken together, this evidence allows us to see PI3K/AKT/mTOR not as a simple sequence of proteins, but as a true biological infrastructure of the tumor cell. Its activation can link extracellular signals, growth, survival, protein synthesis, metabolism, migration, invasion, and metastatic capacity. Studies conducted in profoundly different tumors, from nasopharyngeal to oral carcinoma, from lung to breast, from thyroid to kidney and osteosarcoma, therefore converge on the same principle: when this network remains persistently active, the cell has a system capable of coordinating many of the functions necessary for neoplastic progression.

But among these functions, there is one that deserves to be followed separately. A cell that grows continuously must find a way to sustain exceptional energy and biosynthetic demands, even when the environment surrounding it becomes poor in oxygen and nutrients. It is precisely at this point that mTOR meets one of the most important regulators of cellular adaptation, HIF-1α, and that the metabolism of the tumor cell begins to take on characteristics profoundly different from those we would expect in a normal cell. This will be the necessary step to understand, in the next paragraph, the relationship between mTOR, HIF-1α, and the Warburg effect.

In light of what has been observed, the activation of the PI3K/AKT/mTOR pathway is capable of providing the tumor cell with continuous growth and survival signals. But a rapidly proliferating cell does not only need the order to grow: it must also obtain energy and, above all, a large quantity of material with which to build new membranes, proteins, and nucleic acids. It is precisely in this step that mTOR takes on even greater significance, because it links the proliferative signal to the reorganization of cellular metabolism. One of the most characteristic phenomena of this transformation is the so-called Warburg effect, that is, the tendency of many tumor cells to strongly increase the use of glycolysis (the process by which cells break down glucose to obtain energy) and the production of lactate (a substance produced by cells during glucose metabolism) even when oxygen is available. This does not necessarily mean that the mitochondria have stopped working: in numerous tumors, aerobic glycolysis and oxidative phosphorylation can coexist. AKT can in fact simultaneously increase glucose uptake, glycolytic flux, and, in some contexts, even mitochondrial energy production, so much so that it has been proposed as a possible “Warburg kinase” (Robey & Hay, 2009) [22].

To understand the advantage of this, one must abandon the idea that the cell chooses its metabolism solely on the basis of the maximum amount of ATP obtainable from a glucose molecule. A tumor cell must duplicate rapidly and therefore also requires metabolic intermediates that can be used to synthesize nucleotides (small molecules that constitute the fundamental building blocks of DNA and RNA), lipids, and other macromolecules. Highly active glycolysis allows part of the glucose carbon to be diverted toward these biosynthetic pathways. It is as if a factory, instead of using all its raw material exclusively to produce energy, retained a portion along the production chain to build new machinery and new structures. The neoplastic cell can thus transform its metabolism from a simple energy system into a true infrastructure for proliferation (Robey & Hay, 2009) [22].

One of the most important molecular links between the proliferative signal and this metabolic transformation passes through mTOR and HIF-1α. Normally, HIF-1α is known primarily as a component of the cellular response to reduced oxygen. However, in cells characterized by oncogenic activation of mTOR, the program normally associated with hypoxia can be sustained even when oxygen is available. In this context, mTOR promotes the increase of HIF-1α, which in turn contributes to the expression of proteins involved in glucose utilization and glycolysis. The cell can therefore behave metabolically as if it had to adapt to an oxygen deficiency even while in normoxic conditions: a sort of “pseudohypoxia” generated by intracellular signaling (Sun et al., 2011) [23].

Some researchers have shown this phenomenon using cells in which the loss of TSC1, TSC2, or PTEN (a protein that functions as a natural brake on the PI3K/AKT/mTOR pathway) resulted in strong activation of mTOR. These cells developed intense aerobic glycolysis even in the presence of oxygen; when mTOR was inhibited with rapamycin, the glycolytic phenotype decreased significantly. In parallel, there was an increase in the expression of pyruvate kinase M2, PKM2, an isoform associated with proliferating and tumor cells. When PKM2 was experimentally suppressed, both aerobic glycolysis and tumor growth decreased. The Warburg effect thus appeared not as an isolated metabolic phenomenon, but as a functional consequence of the hyperactivation of upstream oncogenic signaling (Sun et al., 2011) [23].

Further demonstration comes from pancreatic carcinoma. In these cells, ARHGAP4 (a human gene that produces an enzymatic protein) is reduced, while PKM2 is more highly expressed. When ARHGAP4 was experimentally restored, mTOR phosphorylation, HIF-1α and PKM2 levels decreased, along with glucose uptake and lactate production. Conversely, silencing ARHGAP4 increased all these parameters. Above all, the metabolic effects produced by the loss of ARHGAP4 were countered by both the inhibition of mTOR with rapamycin and the inhibition of HIF-1α. The experiment thus outlines a functional sequence in which mTOR activation sustains HIF-1α, HIF-1α contributes to the glycolytic program, and PKM2 represents one of the effectors through which this program is concretely realized (Shen et al., 2019) [24].

The same principle emerges in colorectal carcinoma. The long non-coding RNA CRNDE (an RNA molecule that does not produce proteins but regulates gene activity) was found to be capable of sustaining the Warburg effect through the AKT/mTORC1 pathway. Its silencing reduced glucose uptake, ATP and lactate production, and GLUT1 expression, accompanied by lower proliferation, greater apoptosis, and increased sensitivity to cisplatin (a life-saving platinum-based chemotherapy drug). Conversely, the overexpression of CRNDE intensified glycolytic metabolism, but these effects were abolished by AKT and mTOR inhibitors (Yang et al., 2022) [25]. The important consequence is that the Warburg effect does not appear simply as the result of tumor growth: in these models, it actively participates in maintaining cellular proliferation and survival.

In gemcitabine-resistant pancreatic carcinoma (an antitumor drug that blocks DNA replication and limits tumor cell proliferation), mTOR overexpression indeed produced an increase in glucose consumption, lactate production, and ATP, along with an increase in GLUT1 and PKM2. Inhibition of mTOR using everolimus (a drug that inhibits mTORC1, helping to reduce cell growth) reversed this metabolic program and simultaneously reduced cell growth and migration, promoting apoptosis; in animal models, it also decreased the growth of gemcitabine-resistant tumors (Cui et al., 2021) [26]. Also in non-small cell lung carcinoma, the experimental reduction of PI3K/AKT/mTOR activity through treatment with 1,25-dihydroxy-vitamin D3 was accompanied by a decrease in GLUT1 and PKM2 and the suppression of the glycolytic phenotype, observed both in vitro and in experimental in vivo models (Yiyan et al., 2022) [27].

The link with HIF-1 has also been observed in the cells of a type of triple-negative breast carcinoma. An experimental extract of Holothuria scabra reduced the activity of the AKT/mTOR/HIF-1 axis, accompanied by a decrease in lactate production and an increase in apoptosis. Although this was exclusively a cellular model and not clinical therapeutic evidence, the experiment provides further confirmation of the functional link between AKT/mTOR signaling, HIF-1, and the maintenance of tumor glycolysis (Yurasakpong et al., 2021) [28].

Other experiments show that it is also possible to modify metabolism by intervening upstream in the circuit. In specific tumor cells, berberine (a plant substance that can influence energy metabolism) reduced AKT/mTOR activity, GLUT1 availability, glucose entry, and ATP production; by experimentally reactivating AKT, most of these effects were attenuated, reinforcing the hypothesis of a functional relationship between the signaling cascade and glucose utilization (Guo et al., 2021) [29]. In docetaxel-resistant prostate carcinoma (a chemotherapy drug), however, the reduction of PI3K, AKT, mTOR, and HIF-1α was accompanied by a decrease in lactate, ATP, and PKM2 and an increase in pyruvate dehydrogenase, suggesting a partial reversal of the glycolytic programming associated with drug resistance (Xu et al., 2023) [30].

Further mechanistic confirmation, although coming from non-tumor cells, shows how closely proliferative signaling and metabolism can be linked. In pulmonary artery smooth muscle cells, PDGF (a platelet-derived growth factor) simultaneously increases proliferation, activation of the PI3K/AKT/mTOR pathway, HIF-1α, and aerobic glycolysis; blocking the cascade reduces both the Warburg effect and proliferation. This model shows that a growth signal can be transformed directly into a metabolic program intended to support cell multiplication (Xiao et al., 2017) [31].

Taken together, these findings allow us to interpret the Warburg effect as part of a coordinated metabolic strategy. The activation of AKT and mTOR does not just tell the cell to grow: it helps prepare the metabolism so that such growth is materially possible. HIF-1α represents one of the main points of connection between these two worlds, because it translates the signal coming from mTOR into a program capable of increasing glucose utilization and glycolysis even when oxygen is available. Fully understanding this step, however, requires looking at HIF-1α more closely: not just as a metabolic regulator, but as one of the main systems through which the tumor cell modifies its behavior and adapts to a progressively more hostile environment.

In this context, HIF-1α represents one of the main meeting points between the proliferative signals of the tumor cell and its metabolic reprogramming. To fully understand its significance, however, it is necessary to observe what happens inside a solid tumor as the neoplastic mass grows. A cell population that is rapidly increasing in size requires ever-increasing amounts of oxygen and nutrients, but the vascular network that should supply it does not necessarily develop at the same speed. Furthermore, tumor vessels are often structurally and functionally abnormal. This leads to the formation of regions where the oxygen supply becomes insufficient relative to the cells' needs. It is important, however, to clarify what it really means to talk about hypoxia. The 20-21% oxygen present in the atmosphere, frequently used as a “normoxia” condition in cell experiments, does not correspond to the normal oxygenation of our tissues. In healthy tissues, physiological concentrations, more accurately defined as physioxia, generally range between approximately 3% and 7.4%, with an average around 5%. In solid tumors, values can drop further, to approximately between 0.3% and 4.2%, and in most neoplasms, median values of less than 2% have been observed, with particularly low levels in some prostate and pancreatic cancers (McKeown, 2014) [32].

This oxygen deficiency is not just a passive consequence of tumor growth. As the disease progresses, it becomes a true selective pressure: cells unable to adapt to the more difficult conditions are penalized, while those that possess or acquire systems capable of tolerating hypoxia gain an advantage. A greater presence of hypoxic cells has indeed been associated with a worse prognosis, and hypoxic regions can favor cell populations endowed with particularly resistant and aggressive characteristics (McKeown, 2014) [32]. In this scenario, HIF-1α can be imagined as an emergency switch that allows the cell to understand that the environment has changed and to rapidly modify its behavior.

HIF-1, an acronym for Hypoxia-Inducible Factor 1, is a transcription factor consisting of two components, HIF-1α and HIF-1β. The decisive component in the response to oxygen is HIF-1α. When oxygen is sufficiently available, specific enzymes called PHDs modify HIF-1α. This modification allows it to be directed toward the system through which it is rapidly degraded. In other words, in the presence of oxygen, the cell produces HIF-1α but, at the same time, has a system that prevents its accumulation. When oxygen decreases, the activity of PHDs is limited; HIF-1α escapes normal degradation, stabilizes, accumulates, and reaches the nucleus. Here, it associates with HIF-1β to form functional HIF-1, capable of recognizing specific DNA sequences defined as Hypoxia Response Elements, or HREs, and modifying the expression of numerous genes (Basheeruddin and Qausain, 2024) [33].

This mechanism originated as a physiological adaptation system. The problem in a tumor is that it is exploited to survive in an environment that might otherwise hinder the growth of the neoplastic mass. Once stabilized, HIF-1α coordinates a response much broader than simple energy conservation: it intervenes in metabolism, cell survival, angiogenesis (the formation of new blood vessels), invasiveness, and the cell's ability to interact with the microenvironment. Furthermore, its activation does not depend exclusively on oxygen deficiency. Oncogenic signals and pathways such as PI3K/AKT/mTOR can increase its expression or activity even when oxygen is not drastically reduced (Qannita et al., 2024) [34]. This connection is particularly important because it directly links what was described in the previous paragraphs to the hypoxic response: in a tumor, HIF-1α can be sustained both by the physical characteristics of the microenvironment and by molecular alterations within the cell.

One of the most profound consequences of its activation concerns the ability to reorganize the entire cellular metabolism. HIF-1 does not only modify the fate of glucose, but also intervenes in lipid and amino acid metabolism and mitochondrial function. Under hypoxic conditions, it limits the entry of pyruvate into oxidative metabolism through the induction of PDK1 (an enzyme that contributes to activating AKT), modifies the functioning of the respiratory chain, reduces oxygen consumption, and helps control the production of reactive oxygen species. It can also promote, through two proteins, mitophagy (a biological process in which the cell selectively eliminates its old, damaged, or no longer useful mitochondria) and reduce mitochondrial biogenesis. The consequence is therefore not simply the “switching off” of the mitochondrion, but its reorganization aimed at better utilizing the little oxygen available and preventing an excessive imbalance of reactive oxygen species from compromising cell survival (Infantino et al., 2021) [35].

Lipid and amino acid metabolism are also adapted to the new needs. Cells can increase the accumulation of lipid droplets, modify the synthesis and utilization of fatty acids, and rely more on glutamine and serine to support biosynthesis and the control of redox balance (Infantino et al., 2021) [35]. HIF-1α must therefore be interpreted not as a simple oxygen sensor, but as a coordinator capable of redesigning the metabolic priorities of the tumor cell.

This adaptability extends well beyond metabolism. HIF-1α promotes processes involved in proliferation and survival, contributes to the epithelial-mesenchymal transition, and can increase the expression of molecules that facilitate the remodeling of the extracellular matrix, migration, and invasion. In this way, hypoxia can transform from an obstacle to growth into a stimulus capable of selecting more adaptable and invasive phenotypes (Ghosh et al., 2022) [36].

The hypoxic (oxygen-poor) environment modifies the relationships between tumor cells, fibroblasts, and immune cells. These changes favor tumor survival and growth, reducing the effectiveness of the immune response. Hypoxia also alters the metabolism of tumor cells, increases oxidative stress, and hinders apoptosis (programmed cell death). A central role is played by HIF-1α, which allows cells to adapt to oxygen deficiency and supports various processes linked to tumor progression. Understanding these mechanisms can favor the development of new therapies, including glycolysis and HIF inhibitors, anti-angiogenic drugs, and hypoxia-activated treatments (Chen et al., 2022) [37].

A particularly important aspect concerns cancer stem cells, small cell populations capable of self-renewal and considered relevant in the persistence and recurrence of the disease. HIF-1α contributes to the maintenance of stemness and, under hypoxic conditions, has been associated with the expansion of populations characterized by specific markers and the increase of factors linked to self-renewal. This phenomenon has been described in various neoplastic contexts, including glioblastoma, breast carcinoma, hepatocellular carcinoma, lung, colorectal, and bladder carcinoma, as well as some hematological neoplasms (Zhao et al., 2023) [38]. It is as if the less oxygenated areas of the tumor could become true protective niches in which cells particularly capable of surviving, regenerating the tumor population, and resisting environmental pressures are favored.

The role of HIF-1α, in fact, does not seem limited to solid tumors. Its high activity has also been observed in leukemias, lymphomas, and multiple myeloma. In diffuse large B-cell lymphoma, high levels of HIF-1α have been associated with a worse prognosis in patients treated with immunochemotherapy; in multiple myeloma, HIF-1α has been linked to the production of pro-angiogenic signals, proliferation, invasion, and drug resistance (Zhao et al., 2023) [38]. The phenomenon cannot therefore be considered exclusively the product of poorly vascularized regions of a solid mass: HIF-1α is inserted into broader oncogenic networks that can sustain its activity in different neoplastic contexts.

In breast carcinoma, this network appears particularly evident. The persistent activation of HIFs is associated with metabolic reprogramming, invasive capacity, the maintenance of cancer stem cells, and chemoresistance. In particular, HIF-1α can increase the expression of transporters responsible for drug efflux, contributing to the reduction of their intracellular concentration (Luo et al., 2022) [39]. This helps to understand why hypoxia does not represent only a metabolic problem, but can also concretely influence the therapeutic response.

The relationship between HIF-1α and treatment resistance is further strengthened by its ability to simultaneously coordinate metabolism, extracellular matrix remodeling, invasion, and microenvironment modification. Furthermore, the reactive oxygen species produced at the mitochondrial level can contribute to the stabilization of HIF-1α, generating reinforcement circuits through which the adaptive response tends to be maintained (Magar et al., 2024) [40]. This results in an extremely dynamic system in which hypoxia, metabolism, and intracellular signaling feed each other.

HIF-1α thus assumes the role of a true coordination center for tumor adaptation. The decrease in oxygen, which might initially seem like a limit to growth, is translated into a biological program that allows the cell to change its metabolism, preserve its survival, modify the microenvironment, acquire invasive characteristics, and increase its ability to resist therapies. It is therefore not surprising that its overexpression is frequently associated with more aggressive phenotypes and unfavorable prognosis. But HIF-1α does not directly carry out all these transformations: it does so by activating specific target genes. In the next sections, it will therefore be necessary to follow separately some of the main tools through which this program is concretely realized, starting from one of the most important for the energy supply of the tumor cell: GLUT1.

Based on what has been described, HIF-1α allows the tumor cell to adapt to an environment where oxygen becomes progressively harder to find. But adapting to hypoxia does not just mean managing to survive with less oxygen: it also means finding an alternative way to ensure the cell has a sufficient flow of nutrients to support proliferation, biosynthesis, and survival. In this process, one of the most important targets of HIF-1α is GLUT1, the glucose transporter. If we imagine the cell membrane as the border of a city, GLUT1 can be compared to one of the main gates through which glucose manages to enter. Increasing the number and activity of these gates means allowing the tumor cell to intercept a greater quantity of glucose from the surrounding environment and thus fuel the glycolytic metabolism upon which many neoplasms particularly rely. This relationship between HIF-1α and GLUT1 therefore represents one of the points through which the response to hypoxia is transformed into a true functional modification of tumor metabolism.

A first, particularly intuitive demonstration of this relationship comes from studies that have linked HIF-1α and GLUT1 to the uptake of the 18F-FDG radiopharmaceutical used in PET scans. The principle is also important for clinically understanding what happens in a tumor: 18F-FDG is a glucose analog, and its increased uptake can therefore reflect, at least in part, the increased capacity of neoplastic cells to acquire glucose. In an experimental rabbit lung tumor model, researchers observed that tumor uptake of the radiopharmaceutical was positively correlated with both GLUT1 and HIF-1α. Furthermore, GLUT1 expression was directly correlated with that of HIF-1α, while tracer uptake also increased in relation to the size of the tumor mass. These results allow us to visualize, almost from the outside, what is happening inside the cell: as the tumor grows and develops hypoxic regions, HIF-1α increases, GLUT1 increases, and the capacity to intercept glucose grows in parallel (Xu et al., 2013) [41].

The strength of the HIF-1α/GLUT1 axis emerges even more clearly when observed during the progressive transformation of normal tissue into neoplastic tissue. In cervical carcinoma, 158 samples were analyzed, including 28 cases of normal cervical epithelium, 32 cervical intraepithelial neoplasias, and 98 invasive cervical carcinomas. The expression of HIF-1α and GLUT1 increased progressively along this sequence, moving from normal tissue to the precancerous lesion and finally to invasive carcinoma. HIF-1α was associated with the degree of pathological differentiation, clinical stage, extent of lesions, and patient age, while GLUT1 showed a significant association with lymph node metastases. Above all, a strong positive correlation emerged between HIF-1α and GLUT1 (Cheng et al., 2013) [42].

When specific cervical carcinoma cells were experimentally placed in hypoxic conditions, the phenomenon became even more evident. Proliferation and invasiveness increased, apoptosis decreased, and glycolytic activity grew significantly. Simultaneously, both the gene and protein expression of GLUT1 and hexokinase II (the enzyme that initiates glucose utilization within cells) increased. HIF-1α, on the other hand, showed a different behavior: its mRNA did not increase significantly, while the amount of protein increased markedly (Cheng et al., 2013) [42].

This detail connects directly to what was described in the previous paragraph: during hypoxia, HIF-1α can increase not because the cell necessarily needs to produce much more mRNA, but because the protein is stabilized and protected from normal degradation mechanisms. Once accumulated, it can act on the cell's transcriptional program and increase the expression of genes necessary for glycolysis, including GLUT1. In this way, the decrease in oxygen is translated into an increase in the capacity to acquire glucose.

The functional importance of GLUT1 emerges with particular clarity in hepatocellular carcinoma. In this context, analysis of tumor tissues showed that 68.2% of samples exhibited an increase in GLUT1 mRNA compared to the corresponding non-neoplastic liver tissue. Protein expression detectable by immunohistochemistry was present in 13.2% of tumors and was absent in non-tumor liver tissues. In positive samples, GLUT1 was associated with a higher proliferative index, more advanced tumor stages, and lower histological differentiation (Amann et al., 2009) [43]. The transporter, therefore, did not seem to simply accompany neoplastic transformation, but rather characterized cells endowed with more aggressive biological behavior.

The most important proof, however, came from experiments in which GLUT1 was directly silenced. Reducing its expression via siRNA (short double-stranded RNA molecules) significantly decreased both the proliferation and migratory capacity of hepatocellular carcinoma cells. At the same time, glucose uptake and lactate secretion were reduced (Amann et al., 2009) [43]. This step is fundamental: GLUT1 does not appear only as an indicator of a metabolically active cell, but as a functional element that contributes to the maintenance of the tumor phenotype. If the gate through which glucose enters is narrowed, the cell loses part of its capacity to support both its own metabolism and certain behaviors associated with neoplastic progression.

The functional dependence between HIF-1α and GLUT1 was further tested in laryngeal carcinoma using genetic engineering. Under hypoxic conditions, tumor cells developed greater radioresistance, increased 18F-FDG uptake, and simultaneously showed activation of the PI3K/AKT/mTOR pathway and greater expression of HIF-1α and GLUT1. When HIF-1α or GLUT1 were eliminated, cell viability and glucose uptake decreased; when they were eliminated simultaneously, the effects were even more marked (Bao et al., 2022) [44].

This study also reveals something particularly important for understanding the complexity of tumor signaling: the relationship does not seem to be simply a unidirectional line in which PI3K/AKT/mTOR activates HIF-1α, HIF-1α increases GLUT1, and the process ends there. PI3K/AKT/mTOR activation increased HIF-1α and GLUT1, but genetic inactivation of HIF-1α or GLUT1 in turn reduced the activity of the PI3K/AKT/mTOR cascade. The authors therefore proposed the existence of a reciprocal regulation circuit, without being able to definitively establish which component occupied the position hierarchically further upstream (Bao et al., 2022) [44]. The tumor cell thus appears less like a linear chain of switches and more like a network of mutually reinforcing circuits: the proliferative signal supports metabolism, and metabolism, in turn, contributes to keeping survival signals active.

The importance of the HIF-1α/GLUT1 axis is further supported by experiments in which its interruption was associated with a reduction in tumor growth. In hypopharyngeal carcinoma, treatment of tumor cells with pseudoginsenoside DQ (a species of medicinal plant similar to true ginseng) reduced both HIF-1α and GLUT1 at the mRNA and protein levels, simultaneously decreasing glucose uptake. Microscopy analysis also showed something that simple quantification of expression could not have highlighted: not only did the total amount of GLUT1 decrease, but the molecules present on the plasma membrane were also reduced and their organization into clusters was altered (Wang et al., 2021) [45]. The availability of a transporter, in fact, does not depend only on how much is produced, but also on how much manages to reach and organize itself on the membrane, which is exactly the point where it must perform its function.

Further confirmation comes from breast carcinoma. In different experimental models, honokiol (a natural substance found in the bark of the magnolia plant) reduced the amount of HIF-1α and simultaneously the expression of GLUT1. The metabolic consequence was a decrease in glucose uptake, lactate production, and ATP generation, accompanied by a reduction in extracellular acidification and an increase in oxygen consumption. From a biological point of view, these modifications were associated with lower proliferation, greater apoptosis, and reduced tumor growth in vivo (Yi et al., 2022) [46].

Here too, HIF-1α was regulated primarily at the level of protein stability. Honokiol did not significantly modify its mRNA, but promoted its degradation through a mechanism involving specific ligases. By reducing HIF-1α, its downstream glycolytic targets, including GLUT1, were therefore decreased (Yi et al., 2022) [46]. This experiment reinforces a principle that already emerged in cervical carcinoma: to profoundly modify tumor metabolism, it is not always necessary to change the transcription of HIF-1α; it may be sufficient to modify its protein stability and, through this, the entire metabolic program it controls.

A conceptually analogous mechanism was observed in colorectal carcinoma through the study of sodium butyrate. In cells of this tumor line, treatment increased SIRT4 (a protein found in mitochondria), promoting the autophagic degradation of HIF-1α. The decrease in HIF-1α was accompanied by a reduction in GLUT1, lower glucose uptake, lower lactate production, and lower ATP generation. In parallel, the survival and clonogenic capacity (the ability of a cell to reproduce and generate other cells) of the tumor cells decreased, while in a xenograft model, butyric acid inhibited tumor growth (Zhang et al., 2024) [47].

The data is particularly interesting because it links the availability of HIF-1α to the amount of GLUT1 and, through GLUT1, to the very possibility of the cell effectively fueling its glycolytic metabolism. When SIRT4 promoted the autophagic elimination of HIF-1α, the entire downstream metabolic flux decreased: less GLUT1 meant less available glucose, and this was associated with lower levels of lactate and ATP (Zhang et al., 2024) [47].

Further, very recent evidence expands the biological significance of the HIF-1α/GLUT1 axis even more. In highly metastatic osteosarcoma, a link was identified between lysosomal zinc metabolism (zinc contained in lysosomes) and glucose metabolism through the HIF-1α/GLUT1 axis. In experimental models, the activation of TRPML1 (a protein that forms an ion channel located on the lysosomal membrane) modified the intracellular flux of the zinc ion and signaling mediated by reactive oxygen species, leading to the inhibition of HIF-1α, the reduction of GLUT1, and a consequent decrease in glucose uptake (Wang et al., 2026) [48].

In this study, the impairment of glucose metabolism produced consequences that went beyond the simple reduction of energy availability. The lower utilization of glucose compromised the regeneration of antioxidant systems, reducing the cell's ability to control oxidative stress. The experimental platform used also induced immunogenic programmed cell death and, in in vivo models, reduced both osteosarcoma growth and lung metastases; the combination with an oncological immunotherapy drug further increased the antitumor response (Wang et al., 2026) [48]. It is still a preclinical result, but it suggests how deeply the control of glucose entry can be intertwined not only with energy production but also with redox homeostasis and the tumor immune microenvironment.

Putting these pieces of evidence together, a coherent picture emerges. HIF-1α allows the tumor cell to perceive and cope with reduced oxygen availability; GLUT1 represents one of the tools through which this response is concretely transformed into a metabolic advantage. When HIF-1α stabilizes, GLUT1 expression can increase and the cell membrane becomes more efficient at intercepting glucose from the environment. The more glucose that enters the cell, the greater the amount of substrate available to support glycolytic metabolism. Studies on cervical, hepatocellular, laryngeal, hypopharyngeal, breast, and colorectal carcinoma, together with experimental models of specific rabbit tumors and osteosarcoma, converge in showing that interfering with HIF-1α or GLUT1 can reduce glucose uptake and be accompanied by a decrease in proliferation, survival, or tumor growth, albeit with important differences between the various experimental models.

GLUT1, therefore, should not be considered just a passive gate. In the context of tumor metabolic reprogramming, it becomes one of the points through which the cell adjusts its glucose supply to its own needs. HIF-1α helps increase this capacity especially when oxygen becomes scarce, allowing the cell to continue fueling a metabolism that is heavily dependent on glycolysis. But once it has entered through GLUT1, the glucose must be metabolized, and this process produces consequences that do not remain confined within the cell. One of the most important is the increasing production of lactate, an element that will progressively assume a central role not only in the metabolism of the neoplastic cell but also in the construction of the tumor microenvironment.

HIF-1α plays a fundamental role in the adaptation of tumor cells to oxygen deficiency. When oxygen levels decrease, this factor promotes cell survival through two important mechanisms. On one hand, it stimulates the production of VEGF, a protein that promotes the formation of new blood vessels. On the other, it modifies energy metabolism, increasing the reliance on glycolysis (a process that allows energy to be derived from glucose). This metabolic shift leads the tumor cell to use more glucose and to convert a greater portion of pyruvate into lactate. In this way, even when oxygen is scarce, the cell can continue to produce energy and sustain its activities.

However, the increased production of lactate creates a new requirement. To avoid excessive accumulation, the cell must have systems capable of transporting it across the membrane. This is where monocarboxylate transporters (MCTs) come into play—proteins that facilitate the passage of lactate. Among these, MCT4 assumes particular importance in tumor cells characterized by high glycolytic activity.

The general link between hypoxia, increased glycolysis, and lactate production can also be observed outside of tumors. In human adipocytes exposed to 1% oxygen, researchers observed that lactate release increased 2.3-fold after just 24 hours. After 48 hours, MCT1 mRNA increased 8.5-fold and MCT4 mRNA by as much as 14.3-fold. Returning to normoxia also reversed the observed changes, indicating that this was a dynamic response to oxygen availability (Pérez de Heredia et al., 2010) [49]. Although these are not tumor cells, this experiment is useful for understanding the physiological principle: when oxygen is scarce, increased lactate production and the remodeling of its transport systems can proceed together.

In tumors, this adaptation takes on an even more complex significance. In breast carcinoma, other researchers have shown that the constitutive activation of HIF-1α in fibroblasts (cells that produce the structural support for tissues) induces a shift toward aerobic glycolysis, with greater production and secretion of L-lactate, an increase in MCT4, and a simultaneous reduction in mitochondrial activity. When these fibroblasts were co-injected with specific breast carcinoma cells, they significantly promoted tumor growth (Chiavarina et al., 2012) [50]. This finding introduces a fundamental concept: the lactate produced in response to HIF-1α should not be considered simply a "waste product" of glycolysis. Its production and transport can become part of the metabolic organization of the tumor and the microenvironment that surrounds it.

The distribution of MCT4 in human neoplastic tissue makes this concept even more intuitive. By analyzing advanced head and neck carcinomas, researchers observed that GLUT1 and MCT4 exhibited a very similar spatial pattern. As the distance from blood vessels increased—and therefore in regions where oxygen tends to become progressively less available—the expression of both increased. At a distance of approximately 150–200 μm from the vessels, the positive fraction was two to four times higher. Furthermore, GLUT1 and MCT4 showed high colocalization (Rademakers et al., 2011) [51]. It is almost possible to imagine a metabolic map of the tumor: as one moves away from the vessels, GLUT1 increases, supporting the entry of glucose necessary for glycolysis, and simultaneously MCT4 increases, consistent with the need for glycolytic cells to manage and export the produced lactate.

A similar response has been observed directly in brain and breast tumor cells. Researchers compared specific glioblastoma cells and breast carcinoma cells and found that hypoxia increased MCT1 and MCT4 in both lines. MCT2, however, showed a response dependent on the cell type; they interpreted the increase in MCT4 as consistent with its role in the efflux of lactate produced by highly glycolytic cells (Cheng et al., 2012) [52]. This step is essential: if glycolysis can be imagined as an accelerated production line, MCT4 represents one of the main pathways through which the cell prevents the final product of that line from accumulating uncontrollably.

However, available data show that not all tumors rely primarily on MCT4 for this function. In glioblastoma, researchers observed that hypoxia primarily increased the expression of MCT1 on the plasma membrane. This increase was accompanied by greater glucose consumption, lactate production, and an increase in glycolytic markers such as GLUT1 and HKII (an enzyme that initiates the use of glucose by cells). Reducing MCT1 decreased lactate production and compromised proliferation and invasiveness, while in human tissues, MCT1 was associated with hypoxic regions and specific markers (Miranda-Gonçalves et al., 2016) [53]. The significance of this observation is important: MCT4 represents a central element of hypoxic glycolytic adaptation, but lactate management cannot be described as a system entrusted to a single transporter.

The same principle emerges in bladder carcinoma. High expression of MCT1 was associated with lymph node metastases, distant metastases, and lower overall survival. When MCT1 was experimentally reduced, proliferation, migration, and invasion decreased, along with the lactate present in the extracellular medium and the expression of HK2, GLUT1, and LDHB (a component of an enzyme that allows cells to transform lactate into pyruvate). Above all, hypoxia simultaneously increased HIF-1α and MCT1, while silencing HIF-1α resulted in a reduction of MCT1 (Zhang et al., 2018) [54]. These results therefore link the hypoxic response, glycolytic reprogramming, and lactate metabolism to biological characteristics associated with tumor aggressiveness.

In colon cancer, other researchers provided further evidence of how closely HIF-1α can integrate with this metabolic reorganization. The depletion of TRAP1 (a protein located in the mitochondria) compromised mitochondrial function, increased ROS, and pushed cells toward a higher glycolytic flux, with an increase in glucose consumption and lactate secretion. Simultaneously, HIF-1α showed greater nuclear localization and higher transcriptional activity of hypoxia-response elements, while GLUT1 and MCT1 were upregulated (Tsai et al., 2025) [55]. In other words, the system does not act only by increasing glycolysis: it can simultaneously limit the entry of pyruvate into oxidative metabolism, favoring a metabolic configuration in which the production and management of lactate acquire increasing importance.

An even more direct relationship between HIF-1α and a lactate transporter emerges from the study of other researchers on hexavalent chromium. In lung tumor cells, the metal increased HIF-1α, MCT1, and MCT4 and favored the use of lactate. Experimental induction of HIF-1α increased MCT1, while inhibition of HIF-1α or MCT1 reduced lactate use and cell growth. The most mechanistically relevant aspect is that HIF-1α was found to be capable of binding directly to the promoter region of MCT1, promoting its transcription (Li et al., 2025) [56]. This study therefore shows how HIF-1α can intervene directly in the apparatus necessary to use lactate, even if in this specific model the functional protagonist is MCT1 and not MCT4.

Finally, a non-tumor model offers further confirmation of the functional relationship between HIF-1α and lactate metabolism. In bovine mammary epithelial cells, stimulation with LPS (a laboratory technique used in immunology) increased glycolysis, intracellular lactate, HIF-1α, and MCT1. Induction of HIF-1α further increased lactate accumulation, while its inhibition produced the opposite effect and modified MCT1 in parallel (Cheng et al., 2025) [57].

Overall, an adaptive system emerges in which the reduction of oxygen and the activation of HIF-1α can favor the shift of metabolism toward glycolysis, consequently increasing lactate production. At that point, the cell must necessarily manage this metabolite through MCT transporters. MCT4 appears particularly associated with hypoxic and glycolytic tumor regions and the function of lactate export, while MCT1 shows greater versatility and can assume a significant role in the efflux or use of lactate depending on the metabolic context. The tumor thus builds a sort of monocarboxylate management network that allows it to adapt to the profound differences in oxygenation present within it. But the lactate that is produced and exported does not disappear. It accumulates in the microenvironment, can be used by other cell populations, and can profoundly modify their behavior. It is precisely this transformation of lactate from a product always considered a waste of glycolysis into a biologically active component of the tumor microenvironment that we will address in the next paragraph.

As has emerged so far, the activation of HIF-1α promotes a metabolism more oriented toward glycolysis, increasing lactate production and necessitating the strengthening of the systems responsible for its transport. At this point, however, it is necessary to definitively abandon the idea that lactate represents only the final product of particularly intense glycolysis. For a long time, it was described almost as a waste material that the cell simply had to remove. Evidence accumulated over the last few decades has profoundly changed this interpretation. Lactate can be used as fuel, transferred from one cell to another, participate in cell signaling, and influence the behavior of the cells that make up the tumor microenvironment. More recently, the possibility has even emerged that it can directly link metabolism to the regulation of gene expression through lactylation. In a tumor, therefore, what initially appears as the end point of glycolysis can become the beginning of numerous other biological processes.

One of the most radical interpretations of this phenomenon was proposed by researchers San-Millán and Brooks. The authors of the study suggest shifting the focus from the high glucose consumption characteristic of the Warburg effect to the resulting lactate production, introducing the concept of “lactogenesis.” According to this hypothesis, persistent lactate production would not simply represent a consequence of tumor metabolic reprogramming, but could constitute one of its functional goals. Oncogenes, loss of tumor suppressor gene activity, increased GLUT1, activation of glycolysis and LDHA (a gene that encodes a subunit of the enzyme L-lactate dehydrogenase), inhibition of pyruvate dehydrogenase (an enzyme complex that allows pyruvate to be used for energy production through the Krebs cycle), and an increase in MCT transporters would all converge toward a system capable of maintaining high lactate production and distribution. The accumulated lactate could then participate in feedback circuits capable of further supporting glycolysis, angiogenesis, migration, and tumor adaptation (San-Millán and Brooks, 2017) [58].

What emerges most solidly is that in tumors, lactate can reach very high concentrations and its presence is associated with unfavorable biological characteristics. Some researchers report tumor concentrations up to forty times higher than in normal tissues and a relationship between high lactate levels, lymph node or distant metastases, and reduced survival. Lactate can also contribute to the maintenance of the glycolytic phenotype, migration, and angiogenesis, and can profoundly modify the tumor microenvironment (Zhang et al., 2023) [59]. This means that lactate production does not end the moment the molecule is expelled from the cell that generated it. From that moment on, the lactate enters a shared space and can become a resource or a signal for other cells.

A review by other researchers describes this very conceptual transformation, defining lactate as a true oncometabolite. Intense tumor glycolysis can cause significant acidification of the extracellular microenvironment, while MCT1 and MCT4 allow for a continuous exchange of lactate between different cell populations. Increased expression of these systems has been observed in numerous neoplasms and is frequently associated with greater invasiveness, metastasis, and unfavorable prognosis (de la Cruz-López et al., 2019) [60]. In other words, the tumor should not be imagined as a collection of metabolically identical cells, but as an ecosystem in which different populations can produce, export, import, and use metabolites in different ways.

One of the most elegant experimental demonstrations of this cooperation was provided by other researchers. In solid tumors, oxygen distribution is not uniform: some cells are located in hypoxic regions and depend more on glycolysis, while others are located near blood vessels and have more oxygen available. The authors demonstrated that the lactate produced by hypoxic cells can be taken up through MCT1 by oxygenated tumor cells and converted into pyruvate to fuel oxidative metabolism. In this way, the better-oxygenated cells can use the lactate and save glucose, leaving it available to the hypoxic cells that need it more. When MCT1 was pharmacologically inhibited or silenced, the oxygenated cells were forced to return to glucose; this increased competition for this resource, and the hypoxic cells died from energy deprivation. In mouse models and human xenografts, the disruption of this system also slowed tumor growth and increased sensitivity to radiotherapy (Sonveaux et al., 2008) [61].

Lactate can therefore be imagined as a sort of “energy currency” that circulates between different compartments of the tumor. Cells that produce it in abundance can release it, while other cells can recover and use it. Some researchers have described this organization as a true metabolic symbiosis. Hypoxic and glycolytic cells tend to be characterized by MCT4 and release lactate; those that are more oxygenated can instead express MCT1, import it, convert it into pyruvate through lactate dehydrogenase, and use it to fuel the Krebs cycle and oxidative phosphorylation. The availability of oxygen, nutrients, oncogenic alterations, and the microenvironment therefore determine the appearance of metabolically specialized populations within the same tumor (Nakajima and Van Houten, 2013) [62].

A particularly intuitive example of this plasticity comes from glioblastoma (a malignant brain tumor). Some researchers subjected specific cells to glucose deprivation and observed that the addition of lactic acid allowed them to maintain higher intracellular ATP levels and reduced cell death. At the same time, MCT1 and MCT4 increased, HIF-1α decreased, and proteins associated with oxidative phosphorylation increased. In the presence of lactate, therefore, cells subjected to glucose scarcity were able to modify their energy program and orient themselves more toward mitochondrial respiration (Duan et al., 2018) [63]. This is a fundamental conceptual shift: a tumor cell that produces lactate does not necessarily have to be the same one that subsequently uses it. Within the tumor mass, there can be actual production and consumption circuits.

But lactate does not act only as fuel. Its accumulation also modifies the physical-chemical conditions of the surrounding environment. Other researchers report tumor microenvironment pH values between approximately 6.3 and 6.9 in relation to lactate accumulation, resulting in the activation of metalloproteinases, cathepsins, and hyaluronidases (enzymes that break down tissue components) that can facilitate the degradation of the extracellular matrix. In parallel, hyaluronic acid can increase, creating conditions favorable to motility and invasion (Zhang et al., 2023) [59]. A 2019 review similarly describes how the acidic environment contributes to tumor progression, angiogenesis, therapeutic resistance, and the selection of more aggressive phenotypes (de la Cruz-López et al., 2019) [60].

This ability of lactate to modify the behavior of surrounding cells extends to the immune system. Some researchers describe lactate as a regulatory molecule capable of acting through MCT transporters. In the tumor microenvironment, its accumulation can reduce the cytotoxic function of NK cells (the ability to recognize and destroy infected or tumor cells) and CD8⁺ T lymphocytes and promote characteristics associated with M2 macrophage polarization. Lactate can also participate in angiogenesis through its entry into endothelial cells mediated by MCT1, with stabilization of HIF-1α and an increase in VEGF (Liu et al., 2024) [64]. These aspects will be explored in the following paragraphs; what is of interest here is to understand that a molecule produced by tumor metabolism can become a means through which the tumor influences cells that do not genetically belong to the neoplasm.

In recent years, this interpretation has further expanded with the discovery of lactylation, a post-translational modification through which groups derived from lactate metabolism can be attached to lysine residues of histone and non-histone proteins (proteins that perform various cellular functions, including regulating and protecting DNA). Other researchers describe lactylation as a possible bridge between metabolism, epigenetics (the set of mechanisms that can activate or reduce gene activity without changing the DNA), and cell signaling. Through modifications of chromatin and non-histone proteins, this process can influence gene transcription, DNA repair, the cell cycle, and therapeutic response. The analyzed evidence links lactate and lactylation to proliferation, invasion, metastasis, recurrence, and resistance to treatments (Fang et al., 2026) [65]. The concept is fascinating because it implies that a metabolic variation can leave a functional trace on the apparatus that controls which genes are used by the cell.

The same relationship emerges from the review by other researchers. The lactylation of histone and non-histone proteins has been associated with the regulation of genes involved in proliferation, migration, autophagy, epithelial-mesenchymal transition, and cell survival. In several tumors, high levels of lactylation are associated with more aggressive biological characteristics; furthermore, specific modifications have been linked to resistance to bevacizumab (a drug that hinders the formation of new blood vessels), DNA repair, chemoresistance, and radioresistance (Li et al., 2026) [66]. Even in this case, however, most therapeutic strategies directed against these mechanisms remain in the experimental phase and require clinical confirmation.

Thus, a radical transformation of the way we must observe lactate is taking shape. It is not simply the substance that remains at the end of glycolysis. It can be produced by one cell population and consumed by another; it can allow oxygenated cells to save glucose for hypoxic ones; it can contribute to the acidification of the microenvironment; it can influence angiogenesis, invasion, and immune cells; and, through lactylation, it can even link metabolic status to the regulation of gene expression. The most recent reviews bring these aspects together, describing lactate and lactylation as integrated components of the tumor's metabolic and epigenetic plasticity (Fang et al., 2026) [65].

The tumor therefore appears less and less as a mass of cells that indiscriminately consume glucose and more and more as an ecosystem capable of distributing its resources. Some cells produce lactate, others use it; some live predominantly on glycolysis, others maintain significant oxidative activity. Lactate thus becomes one of the languages through which these populations communicate metabolically. And it is precisely this ability to transform a metabolic product into a signal and an energy resource that opens the next step: understanding through which receptors and transporters lactate can be perceived and used by the tumor cell and how this communication can reach AKT and mTOR again.

In light of these observations, lactate is not merely the end product of glycolysis; it can become a fuel, a signal, and a tool through which different cell populations communicate within a tumor. For this to happen, however, the cell must be able to "perceive" the presence of lactate or transport it across its membrane. Two systems are therefore of particular importance: GPR81, a lactate-sensitive membrane receptor, and monocarboxylate transporters, including MCT1. Through these two entry points—one primarily for signaling and the other for metabolism—lactate can reach intracellular networks fundamental to growth and survival, including AKT and mTOR.

The importance of GPR81 in tumors emerged clearly from the work of researcher Roland in 2014. The receptor was found to be expressed in cell lines belonging to numerous types of neoplasia and, above all, was strongly present in human pancreatic carcinoma: 148 of the 158 samples analyzed, or about 94%, showed high expression of GPR81. Silencing the receptor had relatively modest effects when glucose was abundant, but it became rapidly lethal when glucose availability decreased and lactate served as an alternative energy substrate. Furthermore, lactate increased MCT1 only in cells that retained GPR81. Conversely, the loss of the receptor reduced lactate uptake and mitochondrial activity and, in in vivo models, resulted in a marked reduction in tumor growth and metastasis (Roland et al., 2014) [67].

GPR81 can therefore be imagined as a sensor placed on the cell surface: it detects that lactate is available in the microenvironment and helps activate the adaptations necessary to utilize it. This means that signaling and metabolism can proceed together. This concept appears with particular clarity in non-small cell lung carcinoma resistant to EGFR inhibitors. In the presence of erlotinib (a targeted molecular anticancer drug), the addition of lactate increased the viability of specific cells and reduced their apoptosis. Simultaneously, GPR81 and AKT increased. When GPR81 was silenced, AKT activation decreased and the cells regained sensitivity to erlotinib (a drug that blocks EGFR, reducing growth-promoting signals). At the same time, lactate increased MCT1, supporting cellular metabolic capacity; in clinical samples, GPR81 and MCT1 were more highly expressed in resistant tumors compared to sensitive ones (Ma et al., 2022) [68].

Lactate therefore seems to be able to act through two complementary arms: through GPR81, it can transmit a survival-promoting signal via AKT, while through MCT1, it can enter the cell and participate directly in metabolism. This distinction is important because it avoids viewing GPR81 and MCT1 as two components necessarily placed on the same line. In some contexts, lactate may indeed act primarily through the receptor; in others, what matters most is its entry through the transporters.

A demonstration of the first scenario comes from osteoblasts (specialized cells of bone tissue): in specific cells, lactate enhanced parathyroid hormone-induced differentiation through GPR81 and a cascade involving AKT. Inhibition of MCT1 did not abolish this effect, while interference with GPR81 or AKT significantly reduced it (Wu et al., 2018) [69]. Although not an oncological model, the study provides an important mechanistic demonstration: lactate can be recognized extracellularly and transformed into an intracellular signal capable of reaching AKT without its transport via MCT1 necessarily representing the determining step.

The opposite was observed in adult hippocampal progenitor cells. In this model, lactate increased proliferation, but the effect depended on entry through MCT1 and MCT2 transporters and not on the presence of GPR81. Parallel to this, AKT phosphorylation increased transiently. The deletion of GPR81 did not prevent the proliferative response, whereas blocking MCTs abolished it (Pötzsch et al., 2021) [70]. This is also a non-tumor model, but it demonstrates how lactate can use different pathways to reach the same signaling nodes.

The connection can extend further to mTOR. In pancreatic carcinoma treated with radiotherapy, lactate produced by surviving tumor cells increased persistently and modified the behavior of myeloid-derived suppressor cells (MDSCs). In these cells, lactate activated a sequence involving GPR81, mTOR, and HIF-1α. Inhibition of PI3K or mTOR reduced the lactate-induced accumulation of HIF-1α; furthermore, interference with HIF-1α in turn modified AKT, suggesting the existence of reinforcement circuits between these pathways. By blocking tumor lactate production or eliminating HIF-1α in MDSCs, the antitumor activity of T cells was restored and disease control after radiotherapy was improved (Yang et al., 2020) [71].

Further evidence of lactate's ability to activate AKT comes from endothelial cells. Some researchers have shown that lactate enters through MCT1 and promotes the activation of specific receptor tyrosine kinases, which converge on the PI3K/AKT pathway. This activation was necessary for the formation of tubular structures and new blood vessels. Blocking the receptors or their respective ligands substantially reduced the angiogenic response (Ruan and Kazlauskas, 2013) [72]. In this case, too, lactate acts as a metabolic messenger capable of reaching one of the main growth pathways already encountered in the previous paragraphs.

The relationship with mTOR finds further confirmation in physiological systems. In rat skeletal muscle, acute oral administration of lactate increased the expression of AKT and mTOR and resulted in a significant increase in mTOR phosphorylation after just thirty minutes, even though AKT phosphorylation did not vary significantly (Kyun et al., 2020) [73]. In a subsequent experiment lasting five weeks in mice, lactate administration increased AKT and mTOR mRNA, indicating a modification of signals associated with protein synthesis and degradation (Kyun et al., 2024) [74]. These studies do not naturally prove that the same mechanism operates automatically in tumors, but they confirm that lactate can influence the AKT/mTOR network even in non-neoplastic tissues.

MCT1 can also directly link lactate to the regulation of mTOR through different mechanisms. In human periodontal ligament stem cells, high concentrations of lactate interfered with autophagy and osteogenic differentiation. Blocking MCT1 partially recovered autophagic markers, while rapamycin restored both autophagy and several osteogenic markers, indicating the functional involvement of an MCT1-mTOR axis (Luo et al., 2022) [75]. The data once again underline that lactate entry can modify mTOR in a context-dependent manner and that the biological consequence is not necessarily always proliferative.

Taken together, these findings show that lactate can re-enter the PI3K/AKT/mTOR network through multiple doors. It can be recognized by GPR81, it can cross the membrane through MCT1, and it can indirectly modify the activity of receptor tyrosine kinases or other metabolic circuits. The result depends on the cell type and the context: in some models, GPR81-AKT prevails, in others MCT1 takes the lead role, while in the tumor immune microenvironment, a GPR81/mTOR/HIF-1α axis emerges directly. Lactate, therefore, should not be imagined as a simple linear signal that invariably leads to AKT and mTOR, but as a metabolite capable of entering different molecular networks and using these central nodes to modify survival, metabolism, and response to the microenvironment. It is precisely the activation of AKT that now opens a new step in our sequence: the relationship between AKT, MDM2, and p53 (a protein that protects cells from DNA damage).

As previously highlighted, lactate can reach the PI3K/AKT network and, in certain contexts, contribute to its activation. At this point, it becomes necessary to follow the signal even further downstream, because AKT does not only control metabolism and cell growth: it can also interfere with one of the most important protection systems against neoplastic transformation, the one governed by p53. The connection passes through MDM2, a protein that functions as the main negative regulator of p53. To understand this mechanism, we can imagine p53 as an inspector who continuously checks the genomic integrity of the cell. If it detects sufficiently significant damage, it can slow down the cell cycle, promote repair, or prevent that cell from continuing to proliferate. MDM2, on the other hand, represents one of the systems through which the activity of this inspector is physiologically limited. When AKT enhances MDM2, the balance can shift further toward the reduction of p53.

One of the fundamental molecular demonstrations of this relationship was provided by some researchers in 2002: the activation of AKT reduced the amount of the p53 protein without decreasing the relative mRNA, indicating that the phenomenon did not derive from a lower production of the genetic messenger, but from a greater degradation of the already synthesized protein. The intermediate step was MDM2: AKT promoted its phosphorylation and increased the ability of MDM2 to ubiquitinate p53, that is, to apply a molecular signal that favors its degradation (Ogawara et al., 2002) [76].

This mechanism clarifies why the AKT–MDM2 axis can become particularly relevant in tumors. AKT provides survival and proliferation signals; p53, conversely, can impose an arrest when the cell accumulates dangerous alterations. Enhancing MDM2 therefore means reducing one of the main brakes that could counteract the survival of a damaged cell. In colorectal cancer, for example, involucrasin A (a natural flavonoid) reduced the phosphorylation of AKT and MDM2, while increasing p53. This was associated with a cell cycle arrest and an increase in apoptosis. When a constitutively active form of AKT was introduced or p53 was eliminated, the antiproliferative and pro-apoptotic effects of the compound were significantly reduced (Wei et al., 2023) [77].

A similar relationship has been observed in glioma (the primary tumor of the central nervous system). FOXN3 (a human gene that regulates the cell cycle) is reduced in tumor tissues, and low levels of the protein are associated with lower overall survival. Its overexpression reduces proliferation, colony formation, migration, and invasion, while simultaneously increasing apoptosis. From a molecular point of view, the increase in FOXN3 reduces AKT and MDM2 and increases p53; silencing it, on the other hand, produces the opposite profile. In xenografts, FOXN3 overexpression also slows tumor growth (Wang et al., 2021) [78]. Once again, when the pressure exerted by AKT and MDM2 decreases, p53 returns to accumulate and the tumor behavior is attenuated.

An experiment conducted in a non-tumor context helps to understand how powerful this circuit can be in controlling cell survival. In doxorubicin-induced cardiotoxicity (a chemotherapy drug), licochalcone A (a natural flavonoid extracted from licorice roots) increases the phosphorylation of PI3K, AKT, and MDM2 and reduces p53, protecting cells from ferroptosis (a form of programmed cell death). The inhibition of PI3K/AKT or the overexpression of p53 eliminates much of this protective effect (Chen et al., 2024) [79]. The significance for our journey is not therapeutic, but mechanistic: the same AKT–MDM2–p53 axis can be used physiologically to favor the survival of a cell subjected to stress. In a tumor, however, this capacity can become advantageous for cells that would otherwise be better to eliminate.

In esophageal cancer, genistein (a natural substance of plant origin with a structure similar to human estrogens) offers confirmation in the opposite direction. Treatment reduces EGFR and the phosphorylation of AKT and MDM2, while simultaneously increasing p53. This modification is accompanied by greater apoptosis and cell cycle arrest. In xenograft models, a dose-dependent reduction in tumor growth is also observed (Gao et al., 2020) [80]. The axis therefore appears as a point where proliferative signals coming from membrane receptors can be translated into a reduction of the control exerted by p53.

Signals coming from the extracellular matrix can also use this mechanism. In stem cells derived from adipose tissue, fibronectin and vitronectin increase AKT and MDM2 and reduce p53, attenuating cellular senescence. When MDM2 is inhibited, p53 increases and the share of senescent cells increases simultaneously (Tragoonlugkana et al., 2024) [81]. In this case, too, it is not an oncological model, but the result clearly shows that the AKT/MDM2/p53 axis constitutes a general system through which extracellular signals can influence the cell's decision to continue to proliferate or to arrest.

The oncological importance of MDM2 phosphorylation has also been demonstrated directly in vivo. Some researchers studied a specific residue of MDM2, a target of AKT, observing that preventing its phosphorylation does not substantially change the p53 response to ionizing radiation, but profoundly changes the response to oxidative stress. The mutation reduces the nuclear stability of MDM2, increases p53 activity, and promotes senescence. In murine models of liver cancer, lung cancer, and skin carcinogenesis, the inability to phosphorylate this residue is accompanied by a significant reduction in tumorigenesis (Chibaya et al., 2021) [82]. The study therefore demonstrates that the AKT–MDM2 axis is not limited to modifying the amount of p53 in a test tube, but can concretely influence the ability of a tissue subjected to oncogenic stress to develop a neoplasm.

At this point, we can also reconnect the mechanism to the lactate metabolism described in the previous paragraphs. In hepatocellular carcinoma, lung adenocarcinoma, and ovarian cancer cells, lactate increases the expression of ESM1 (a specific gene that regulates the formation of new blood vessels). ESM1 subsequently activates AKT and MDM2, determining the functional inhibition of p53 and attenuating the response to DNA damage caused by cisplatin (an antitumor drug that damages cell DNA) (Tan et al., 2026) [83]. A particularly interesting connection thus emerges: lactate → ESM1 → AKT → MDM2 → p53, through which a metabolic change can directly reach a central system of the DNA damage response.

The same architecture also contributes to chemoresistance. In oral squamous cell carcinoma, MINK1 (a specific gene that regulates cell movements, cell shape, and the response to biological stress) promotes the activation of AKT and the phosphorylation of MDM2, contributing to the degradation of p53. The knockout of MINK1 reduces AKT and MDM2 and increases phosphorylated forms of p53 associated with its stabilization. The same effect was obtained experimentally through pharmacological inhibition of MINK1 (Mohanty et al., 2022) [84].

In bladder cancer, the mechanism appears even more direct. TPI1, known mainly as a glycolytic enzyme, acquires an oncogenic function independent of its traditional metabolic activity: it forms a complex with AKT and MDM2, facilitating their encounter. In this way, it increases the phosphorylation of MDM2 and promotes ubiquitination (a process by which the cell marks certain proteins to regulate or eliminate them) and degradation of p53. The inhibition of AKT counteracts this effect and preserves p53. High levels of TPI1 are also associated with a worse prognosis, while its silencing reduces proliferation, migration, invasion, and tumor growth (Wang et al., 2025) [85].

All these pieces of evidence therefore show how AKT, MDM2, and p53 constitute a fundamental node in which metabolic, receptor, and oncogenic signals can converge. AKT can enhance MDM2; MDM2 can ubiquitinate p53 and favor its degradation; the reduction of p53 can finally allow the cell to avoid some of the main programs that would normally limit the proliferation of a damaged cell. To fully understand why the functional loss of p53 represents such an important advantage for the tumor, however, it is necessary to observe separately everything that this protein is capable of doing. It is therefore from p53 and its multiple tumor-suppressive actions that our journey must continue.

So far, we have mainly observed what happens inside the tumor cell: the activation of proliferative pathways, the intervention of HIF-1α, the increase in glycolysis, the production of lactate, and the progressive loss of the control systems exerted by p53. But a tumor never grows in isolation. Neoplastic cells are immersed in a complex environment with which they communicate continuously and which, over time, is profoundly transformed by the very presence of the neoplasm. This set of elements is called the tumor microenvironment, or TME, and includes stromal and immune cells, fibroblasts, the extracellular matrix, blood and lymphatic vessels, as well as physical and metabolic conditions such as hypoxia, acidity, interstitial pressure, and nutrient availability. The TME must therefore be imagined not as the passive background in which the tumor grows, but as a true dynamic ecosystem in which the different components influence each other (Wang et al., 2025) [115].

Tumor cells actively participate in the construction of this ecosystem. Through cytokines, chemokines, and numerous other mediators, they can modify the behavior of surrounding stromal cells (cells that support and protect tissues and organs) and immune cells, progressively inducing them to adopt characteristics favorable to proliferation, angiogenesis, invasion, and metastasis. Fibroblasts (cells that produce collagen and other substances necessary to support tissues), the extracellular matrix, and the vascular compartment also participate in maintaining these conditions, creating a network of signals in which clearly distinguishing what belongs to the tumor from what belongs to the surrounding tissue becomes increasingly difficult (Hinshaw & Shevde, 2019) [116].

This cooperation also concerns metabolism. The representation that all tumor cells use glucose in the same way through very intense aerobic glycolysis is now insufficient to describe the heterogeneity observed within neoplasms. In the same tumor, predominantly glycolytic populations and populations with a more oxidative metabolism can coexist. This diversity allows for the development of true metabolic cooperation, in which the products generated by one compartment can become energy substrates for another (Wilde et al., 2017) [113].

It is within this perspective that the concept of the reverse Warburg effect arises. In the classic Warburg effect, it is the tumor cell that strongly increases glycolysis and produces lactate even in the presence of oxygen. In the reverse model, however, part of this glycolytic behavior is assumed by stromal cells, particularly cancer-associated fibroblasts, or CAFs. These fibroblasts can acquire a predominantly glycolytic metabolism and produce lactate, pyruvate, and other high-energy metabolites that are then used by tumor cells, in which oxidative mitochondrial metabolism can remain very active (Fu et al., 2017) [118].

A sort of division of labor is thus created. The stromal compartment becomes more catabolic: it consumes substrates and produces metabolites. The neoplastic compartment, on the other hand, can use those metabolites to support energy production, biosynthesis, and growth. In this organization, what one cell eliminates can become the fuel for another.

One of the experimental mechanisms proposed to explain how this transformation of the stroma can begin involves oxidative stress. Some researchers, using a co-culture of specific tumor cells and fibroblasts, have observed that neoplastic cells release hydrogen peroxide and increase oxidative stress in the surrounding fibroblasts. In fibroblasts, mitochondrial activity decreases and glucose uptake increases, while in tumor cells the opposite phenomenon occurs: mitochondrial activity increases and glucose uptake and GLUT1 expression decrease (Martinez-Outschoorn et al., 2011) [117].

Oxidative stress can therefore push fibroblasts toward autophagy, mitophagy, and progressive impairment of mitochondrial function, favoring the shift to glycolysis. This process has also been associated with the activation of HIF-1α and NF-κB. Thus, glycolytic enzymes and systems dedicated to the production and transport of metabolites increase, creating a circuit through which the stromal compartment becomes progressively more suitable for producing substrates usable by the tumor (Fu et al., 2017) [118].

Lactate naturally occupies a central position in this system. Glycolytic CAFs can export it primarily through MCT4, while tumor cells can acquire it via MCT1 and subsequently use it in oxidative metabolism. This metabolic shuttle allows the two compartments to specialize: one produces and exports, the other imports and uses (Wilde et al., 2017) [113].

The importance of this model does not derive only from cellular experiments. In prostate cancer, some researchers analyzed samples from 480 patients and observed a marked metabolic difference between the stroma and the neoplastic compartment. CAFs strongly expressed MCT4, while in tumor cells, intense membranous expression of MCT1 was evident. Furthermore, proteins associated with fatty acid oxidation were predominantly localized in tumor cells, consistent with a more oxidative metabolism (Pértega-Gomes et al., 2014) [119].

A particularly interesting aspect was the early onset of stromal modifications: MCT4 showed alterations already during the first phases of malignant transformation. Furthermore, the combination of high stromal expression of MCT4 and strong tumor expression of MCT1 was associated with the pT3 stage (a tumor that has exceeded the boundaries of the area in which it formed) and other unfavorable prognostic characteristics (Pértega-Gomes et al., 2014) [119]. The stroma, therefore, can be metabolically reprogrammed very early and become an integral part of the evolution of the neoplasm.

Clinical-pathological confirmation comes from a subsequent analysis of 535 patients with prostate cancer. Some researchers observed that neither MCT1 nor MCT4, considered in isolation, were significantly associated with biochemical recurrence-free survival. The data changed, however, when they were analyzed together. The profile characterized by high MCT1 expression in tumor cells and high stromal MCT4 expression was associated with a significantly worse prognosis and maintained its value even after adjustment for other prognostic factors, with a hazard ratio (a value that indicates how much the risk of an event occurring increases or decreases) of 1.99 (Andersen et al., 2015) [114]. The significance is important: what matters is not necessarily the presence of a single transporter, but the combination of the two compartments that makes metabolic exchange possible.

An even more marked result was observed in triple-negative breast cancer. Some researchers studied over 180 patients, finding that high stromal expression of MCT4 was associated with a ten-year overall survival of less than 18%, while in patients without stromal MCT4 expression, survival reached approximately 97%. Conversely, MCT4 expressed by epithelial tumor cells did not show the same prognostic value (Witkiewicz et al., 2012) [120].

This data makes it particularly evident how reductive it can be to observe exclusively the neoplastic cell. In that specific cohort, a metabolic characteristic of the stroma provided much more relevant prognostic information than the same marker expressed directly by the tumor.

The TME, however, does not only concern metabolism and fibroblasts. Its physical, metabolic, and immune compartments are deeply interconnected. Hypoxia and acidity favor tumor adaptation and resistance to treatments; HIF-1α increases glycolytic and angiogenic programs; the accumulation of lactate contributes to the acidification of the medium and modifies the behavior of immune cells. In parallel, CAFs, the extracellular matrix, and altered vascularization participate in invasion, migration, and the difficulty with which drugs and immune cells manage to penetrate the tumor mass (Wang et al., 2025) [115].

However, it is fundamental not to replace one overly simple model with another that is equally rigid. The reverse Warburg effect does not imply that all tumor cells use stromal lactate nor that the classic Warburg effect disappears. The two configurations can coexist in the same tumor. Some neoplastic populations can remain strongly glycolytic, while others can use oxidative phosphorylation more and exploit metabolites produced by the stromal compartment (Fu et al., 2017) [118].

The tumor thus appears as a metabolically organized and plastic ecosystem, in which resources are not simply consumed, but redistributed. Lactate can be produced by one cell, exported into the extracellular space, and subsequently recovered by another population. Tumor-associated fibroblasts assume, in this system, a particularly important function: they are no longer just structural cells of the stroma, but can become true metabolic partners of the neoplasm. It is therefore precisely on the relationship between lactate and CAFs that the next step must now focus.

As we have seen, the tumor microenvironment (TME) can organize itself as a true metabolic ecosystem, in which neoplastic cells and the stromal compartment continuously exchange nutrients, metabolites, and signals. Within this network, cancer-associated fibroblasts (CAFs) play a particularly important role. They are not simply structural cells located around the tumor mass: they can be progressively reprogrammed to acquire properties capable of supporting proliferation, invasion, fibrosis, immunosuppression, and therapeutic resistance. Lactate represents one of the main tools through which the tumor cell can influence this transformation. The relationship, however, is more complex than a simple transfer of fuel from one cell to another: lactate can be used metabolically by CAFs, can modify their redox state, can reprogram their epigenome (the set of chemical signals that regulate gene activity), and can orient their secretory profile toward phenotypes capable of altering the immune compartment as well.

A particularly clear demonstration comes from pancreatic ductal adenocarcinoma. In this tumor, neoplastic cells, which are highly glycolytic, compete effectively for glucose and divert a significant portion of it away from the surrounding stromal cells. CAFs, therefore forced to live in conditions of relative glucose restriction, can adapt by using the very lactate produced by the tumor cells. Through MCT1, they import it, incorporate it into oxidative metabolism, and use it to fuel the Krebs cycle, thus supporting their own proliferation even when glucose is scarce (Kitamura et al., 2023) [121].

This behavior profoundly changes the significance of lactate in the microenvironment. The same tumor cell that monopolizes glucose simultaneously produces a metabolite that allows the surrounding fibroblasts to survive. But the effect is not only energetic. Exposure to lactate increases the production of immunosuppressive mediators in CAFs, including IL-6, and fibroblasts stimulated in this way reduce the activity of cytotoxic immune cells. Furthermore, in PDAC models (the most common and aggressive type of pancreatic cancer), high tumor LDHA (an enzyme that helps cells produce energy by converting pyruvate to lactate) is associated with a greater presence of CAFs and lower infiltration of CD8⁺ T cells (immune system cells that recognize and destroy tumor cells), while experimental inhibition of LDHA reduces both tumor growth and the fibroblastic component and increases the infiltration and cytotoxic activity of CD8⁺ cells (Kitamura et al., 2023) [121].

However, lactate can act even earlier, contributing to the actual formation of the CAF phenotype. In breast carcinoma, it directly increases the transcription of LDHA and pushes tumor cells toward greater production and secretion of lactate. The latter not only increases the proliferation, migration, and invasiveness of neoplastic cells but also induces their conversion in fibroblasts toward a CAF-associated phenotype. When LDHA is reduced, tumor growth and the metabolic activity of xenografts decrease in parallel (Martínez-Ordoñez et al., 2021) [122].

The general principle that emerges is therefore one of bidirectional communication. Tumor cells can produce lactate and use it to “educate” the fibroblast; once transformed, the CAF can in turn produce metabolites, cytokines, and signals capable of supporting the neoplasm. The most recent evidence describes lactate as a central node in this dialogue, with effects on stromal remodeling, epithelial-mesenchymal transition, CAF heterogeneity, and therapeutic resistance. In addition to this metabolic and signaling function, there is also lactylation, through which lactate can modify proteins and chromatin and directly link metabolism to gene regulation (Tan et al., 2025) [123].

A particularly significant example of this link between metabolism and epigenetics has been observed in pancreatic carcinoma: a study shows that lactate produced by tumor cells can promote tumor growth and spread. Lactate modifies the behavior of neighboring cells, transforming them into cells capable of supporting tumor development. This process involves TET enzymes, which modify the activity of certain genes, also favoring the increase of a receptor involved in tumor invasion; lactate can therefore modify the molecular memory of the stromal cell. It does not just induce a temporary signal, but can participate in an epigenetic reorganization that helps stabilize new functional characteristics (Bhagat et al., 2019) [124].

Gastric carcinoma offers a complementary demonstration. Tumor cells produce lactate both in normoxia and hypoxia, but production increases when oxygen decreases. Lactate is transported into bone marrow-derived mesenchymal stem cells and promotes their acquisition of a pro-tumor phenotype similar to iCAFs. Blocking transport or reducing lactate production attenuates this transformation. Cells activated in this way show greater migration and invasion of tumor cells (Tao et al., 2023) [125].

Even in hepatocellular carcinoma, the link is strictly dependent on glycolytic reprogramming: a study shows that lactate produced by liver tumor cells promotes the formation of an environment that facilitates cancer spread. In particular, lactate contributes to transforming neighboring cells into cancer-associated fibroblasts (CAFs), capable of supporting their growth and spread. Researchers have identified a specific protein as an important regulator of this process. Blocking this protein reduced lactate production, CAF activation, and metastasis formation in experimental models (Mai et al., 2025) [126].

Overall, this evidence shows that lactate can support CAFs on at least three levels. It can become a metabolic fuel, it can function as a signal capable of modifying their redox and transcriptional state, and it can contribute to relatively stable epigenetic reprogramming. However, the fibroblast transformed in this way does not remain confined to its own compartment: it produces IL-6, IL-8, and other mediators that modify tumor cells and immune populations. Lactate thus becomes the starting point of a multi-step communication: tumor cell → CAF → immune cells. Among the latter, tumor-associated macrophages play a particularly relevant role, because some of the signals produced by CAFs promote their recruitment and polarization toward M2 phenotypes (immune system cells that help reduce inflammation). It is therefore from the relationship between lactate, TAMs, and M2 polarization that our journey must continue.

In light of these observations, the lactate produced by the tumor cell profoundly modifies the behavior of cancer-associated fibroblasts, transforming a product of glycolytic metabolism into a true signal capable of remodeling the stroma (the supporting tissue surrounding the cells of an organ or tumor). The same principle also extends to the immune compartment. Among the cells most influenced by the metabolic microenvironment are tumor-associated macrophages, or TAMs. These cells possess remarkable plasticity and can assume different functional states in response to signals they receive from the surrounding tissue. The distinction between M1 and M2 macrophages represents a useful simplification to describe two extremes of this spectrum: the former are generally associated with inflammatory and antitumor functions, while the latter exhibit more immunoregulatory, pro-angiogenic, and reparative characteristics, and, in the neoplastic context, are frequently favorable to tumor progression. Lactate is one of the metabolic signals capable of shifting this balance toward M2 characteristics.

One of the fundamental demonstrations of this phenomenon was provided by a 2014 scientific study: in mouse models of Lewis lung carcinoma and B16 melanoma, lactic acid produced by tumor cells induced the expression of markers associated with the M2 phenotype in macrophages. Tumor lines that produced more lactate induced a greater macrophage response, while reducing tumor lactate production by silencing PKM2 reduced tumor growth. The elimination of a specific gene in macrophages also significantly limited the development of the neoplasm (Colegio et al., 2014) [127].

The significance is important: tumor metabolism can directly modify the functional identity of an immune cell. The neoplastic cell produces lactate to support its own metabolism, but this metabolite diffuses into the microenvironment and reaches the macrophages, where it can activate tumor-promoting transcriptional programs. In gastric carcinoma, this relationship has been further clarified. High glycolytic activity was associated with a greater presence of M2 macrophages. Reducing LDHA in tumor cells decreased their ability to induce this polarization. Even more importantly, blocking monocarboxylate transporters (MCTs) almost completely abolished the M2 response, as did the silencing of HIF-1α in macrophages (Zhang and Li, 2020) [128].

A particularly linear sequence thus emerges: the tumor increases glycolysis, produces lactate, the lactate enters the macrophage through MCTs and, via HIF-1α, modifies the expression of genes that characterize the M2 phenotype. It is therefore not only the acidification of the microenvironment that is relevant: it is the metabolite itself, once transported into the cell, that actively participates in the reprogramming of the macrophage.

In esophageal carcinoma, however, a second pathway emerges. Exposure to lactate increases the phosphorylation of AKT. The intensity of the response increases as the lactate concentration rises, while reducing the production of this metabolite using oxalate attenuates both M2 polarization and AKT activation. These reprogrammed macrophages, in turn, increase the proliferation and migration of tumor cells; when AKT signaling is attenuated, these protumoral effects decrease (Zhang et al., 2023) [129].

This introduces a principle we will encounter repeatedly: the communication between tumor and TAMs is bidirectional. The tumor cell modifies the macrophage through its metabolism, but the modified macrophage returns signals to the neoplasm capable of increasing its aggressiveness.

Pituitary adenoma provides a particularly clear example of this circuit. In invasive tumors, lactate production is higher and is associated with higher infiltration of M2 TAMs. Lactate induces polarization through mTORC2; subsequently, M2 macrophages produce CCL17: this chemokine acts on tumor cells and activates mTORC1, increasing invasive behavior. High expression of CCL17 is also associated with larger tumor size, invasiveness, and greater susceptibility to postoperative recurrence (Zhang et al., 2021) [130].

In melanoma, an additional metabolic level has been identified. Tumor-conditioned medium induces an increase in key markers associated with the M2 phenotype in macrophages; hypoxia further amplifies the phenomenon. Lactate is one of the main mediators, and blocking its transport attenuates M2 polarization. In this model, however, the process depends primarily on oxidative metabolism: inhibition of the Krebs cycle reduces key markers, while blocking glycolysis does not produce equivalent effects. Macrophages polarized in this way increase the proliferation, clonogenicity (the ability of a single cell to multiply and form a colony of new cells), migration, and invasion of melanoma cells, and reduce their apoptosis (Wang et al., 2025) [131].

Lactate can also modify the macrophage through epigenetics. In melanoma, some researchers have shown that interfering with the MCT1 complex using crizotinib (an antitumor drug) reduces both tumor secretion and macrophage uptake of lactate. This reduction is associated with less M2 polarization and a better response to immune checkpoint blockade (control mechanisms that regulate the activity of the immune system). In macrophages, lactate increases another specific chemokine, and this phenomenon is accompanied by greater lactylation in the gene promoter. Reducing lactate transport decreases both the genetic modification and the chemokine (Zhou et al., 2025) [132].

This step is particularly significant because it adds a new level to the sequence: lactate does not necessarily have to be limited to activating a kinase or stabilizing a transcription factor. It can contribute to modifying the chromatin and, therefore, the genetic program with which the macrophage responds to the tumor.

The most recent reviews confirm this integration between metabolism and epigenetics. In macrophages, lactate is predominantly associated with the promotion of M2 phenotypes, and the lactylation of other proteins can support immunosuppressive transcriptional programs, favoring invasion, metastasis, chemoresistance, and reduced phagocytosis (Dong et al., 2025) [133].

Intratumoral Tregs represent an example of cells particularly adapted to this environment. Unlike effector T lymphocytes, which suffer more from glucose scarcity and the accumulation of metabolites, Tregs can maintain significant immunosuppressive activity thanks to their greater dependence on oxidative phosphorylation and their flexibility in using different energy substrates (Rao et al., 2021) [134]. This study does not directly demonstrate M2 polarization mediated by lactate, but it helps to understand why a metabolically hostile TME can simultaneously disadvantage effector cells and select for immunoregulatory populations.

Finally, an experimental demonstration in the opposite direction further reinforces the role of lactate on TAMs. In small-cell lung carcinoma, ACT001 (an experimental drug of natural origin) reduces glycolytic activity and leads to a decrease in lactate production. This metabolic change corresponds to a reduction in M2 macrophage infiltration and an increase in M1 characteristics, along with reduced tumor growth, invasion, and metastasis (Ding et al., 2025) [135].

Overall, therefore, lactate represents one of the main points of communication between tumor metabolism and the macrophage compartment. Through MCT, HIF-1α, AKT, mTOR, mitochondrial metabolism, and lactylation, it can orient TAMs toward M2-like functional states. These macrophages then produce other mediators capable of supporting angiogenesis, invasion, growth, and immune evasion. Lactate does not only modify what the tumor cell does: it also modifies those surrounding it, progressively creating an immune ecosystem that is more tolerant toward the neoplasm.

And it is precisely this ability to transform the immune microenvironment that leads us to the next step: the activation of HIF-1α and CD73, through which tumor metabolism will begin to link directly to the production of adenosine (a substance produced naturally by our body, which helps regulate sleep, brain activity, and blood circulation).

In the journey followed so far, we have seen how hypoxia is not merely a consequence of disordered tumor growth, but a condition capable of profoundly modifying the behavior of neoplastic cells and the microenvironment surrounding them. When oxygen becomes insufficient, the stabilization of HIF-1α allows the cell to adapt through extensive transcriptional reprogramming. Part of this adaptation concerns energy metabolism, another concerns angiogenesis, and yet another modifies the relationship between the tumor and the immune system. It is precisely in this latter context that a new protagonist appears: adenosine. In tumor tissues, in fact, this molecule can reach higher extracellular concentrations compared to normal tissues, and its production is strictly linked to the activity of two ectoenzymes, CD39 and CD73 (proteins present on the outer surface of cells to transform or degrade substances present in the surrounding environment) (Arab et al., 2019) [136].

To understand the mechanism, we can imagine a small assembly line placed on the surface of the cells. When ATP and ADP (molecules used by cells to provide energy for their activities) accumulate in the extracellular space following stress, ischemia, or cell damage, CD39 converts them into AMP; subsequently, CD73 transforms AMP (a molecule present in cells that helps signal a reduction in available energy) into adenosine. Under physiological conditions, this system helps contain excessive inflammation. In a tumor, however, the same pathway can be exploited within a chronically altered environment. Hypoxia promotes this process through HIF-1α, which increases the expression of CD73, thereby helping to create favorable conditions for the persistent production of adenosine (Arab et al., 2019) [136].

An important experimental confirmation of this connection comes from hepatocellular carcinoma. Some researchers have studied a specific microRNA (microRNA-223) (small RNA molecules that regulate gene activity) that is particularly expressed in healthy cells and reduced in HCC tumor tissues. Its absence aggravated tumorigenesis and was accompanied by greater angiogenesis, hypoxia, and immunosuppression. The fundamental step was represented precisely by HIF-1α: this microRNA acted directly on this factor, and in models lacking it, CD39, CD73, and adenosine were increased. Conversely, increasing microRNA-223 in tumor cells subjected to hypoxia decreased the ectoenzymes and the concentration of adenosine (Fu et al., 2023) [137].

These results allow us to read the phenomenon as a physiological sequence: hypoxia favors HIF-1α; HIF-1α supports the CD39/CD73 pathway; the activity of these ectonucleotidases (enzymes present on the cell surface that transform molecules such as ATP and ADP) increases the microenvironment's ability to produce adenosine. In the same model, in fact, the inhibition of CD73 or the adenosine A2A receptor reduced both certain immunosuppressive signals and tumor size, reinforcing the functional relevance of the entire circuit (Fu et al., 2023) [137].

HIF-1α can also coordinate CD73 with other tumor adaptation mechanisms. In triple-negative breast cancer, exposure to different chemotherapeutic agents—carboplatin, doxorubicin, gemcitabine, and paclitaxel—results in a coordinated increase of CD47 and CD73 in surviving cells. When HIF-1α activity is inhibited genetically or pharmacologically, this enrichment is prevented; conversely, hypoxia is sufficient to promote a similar phenotype even in the absence of chemotherapy (Samanta et al., 2018) [138].

This data suggests that CD73 should not be considered an isolated element. It can be part of a broader program through which HIF-1α simultaneously modifies different characteristics of the tumor cell. A complementary example is represented by CD47. In hypoxic breast cancer, HIF-1α directly increases its transcription; reducing HIF-1α or CD47 instead increases the phagocytosis (a defense process through which certain immune system cells recognize and destroy tumor cells) of tumor cells by macrophages. CD47 is also more highly expressed in tumor stem cells, and its reduction significantly decreases their population (Zhang et al., 2015) [139]. Even though CD47 belongs to a different pathway than the adenosinergic one, this observation shows how extensive the adaptive program governed by hypoxia can be.

However, there is an even more interesting aspect: HIF-1α can contribute to the increase of CD39 and CD73 even when oxygen is not necessarily the initial signal. In myeloid-derived suppressor cells of patients with non-small cell lung cancer, TGF-β (a protein that regulates cell activity, contributing to the control of inflammation) present in the tumor microenvironment induces the phosphorylation of mTOR; downstream of this response, HIF-1α increases, which is necessary for the induction of CD39 and CD73 even under normoxic conditions (Li et al., 2017) [140].

A connection is thus created with mechanisms encountered much earlier in our journey: TGF-β → mTOR → HIF-1α → CD39/CD73. CD39⁺CD73⁺ myeloid cells are particularly immunosuppressive, and their frequency is associated with lymph node involvement, metastasis, tumor stage, and a less favorable response to chemotherapy (Li et al., 2017) [140]. HIF-1α therefore emerges not only as a sensor of hypoxia but as a convergence point through which different microenvironment signals can fuel the enzymatic machinery responsible for adenosine production.

One of the most intuitive proofs of the relationship between hypoxia, HIF-1α, and adenosine comes from experiments in which the pathway is modified in the opposite direction. Some researchers used a breathing mixture containing 60% oxygen in mouse models. By reducing tumor hypoxia, HIF-1α and its dependent proteins decreased; simultaneously, the ectoenzymes involved in adenosine production and the extracellular adenosine concentration in the microenvironment decreased (Hatfield et al., 2014) [141].

It is almost like intervening on the switch placed at the beginning of the cascade: by increasing oxygen availability, the signal that maintains HIF-1α is reduced, and by decreasing HIF-1α, a part of the network that supports adenosine accumulation is weakened. In parallel, the recognition of tumor cells by cytotoxic T lymphocytes improved (Hatfield et al., 2014) [141].

The principle was subsequently developed by considering the entire hypoxia–adenosine axis. The evidence collected by Steingold and Hatfield describes hypoxia, HIF-1α, CD39/CD73, adenosine, and the A2A receptor as functionally linked components of an immunosuppressive barrier. CD39 and CD73 control the extracellular generation of adenosine, while the correction of hypoxia can reduce HIF-1α, adenosine, and several elements of the same pathway (Steingold and Hatfield, 2020) [142].

In the same year, other researchers proposed interpreting hypoxia and adenosinergic signaling as different ends of the same circuit: hypoxia and HIF-1α occupy the upstream part, while adenosine and its receptors represent the downstream part. Experimental reversal of hypoxia through respiratory hyperoxia weakens various stages of the axis and makes the microenvironment more permissive to the immune response (Halpin-Veszeleiova and Hatfield, 2020) [143].

More recent experimental approaches have sought to overcome one of the problems of solid tumors: delivering oxygen into an irregular and disorganized vasculature. A stable mixture of synthetic chemical compounds formed exclusively by carbon and fluorine atoms, associated with respiratory hyperoxia, reduced microenvironment hypoxia and weakened the HIF-1α–adenosine axis. In parallel, intratumoral infiltration and the activity of T and NK cells increased (Halpin-Veszeleiova et al., 2025) [144].

Further confirmation comes from a nanoreactor designed to intervene simultaneously on oxygen production and consumption. By reducing hypoxia, this system caused a significant decrease in HIF-1α, CD39, and CD73; in the analyzed cells, CD39 and CD73 decreased by 50.7% and 31.5%, respectively. Extracellular ATP metabolism slowed down, and adenosine formation was reduced (Gu et al., 2024) [145].

Overall, a coherent relationship emerges: the hypoxic tumor microenvironment stabilizes HIF-1α and favors the activation of a network that includes CD39 and especially CD73, progressively shifting the metabolism of extracellular nucleotides from ATP toward adenosine. Even signals independent of hypoxia, through mTOR, can converge on HIF-1α and fuel the same system. The opposite experiment completes the picture: when hypoxia is attenuated, HIF-1α decreases, CD39/CD73 are reduced, and adenosine accumulation is limited.

At this point, however, the most important question remains: why does adenosine represent an advantage for the tumor? Having understood how it is produced does not yet fully tell us what happens to immune cells when they are immersed in a microenvironment rich in this metabolite. This is precisely the step we will address next, by analyzing CD73, adenosine, and immune evasion.

To fully understand how a tumor manages to evade immune system surveillance, it is necessary to momentarily abandon the idea that immune evasion depends solely on the inability of defense cells to recognize the neoplasm. In many tumors, immune cells actually reach the tissue, recognize danger signals, and, at least initially, possess the necessary tools to react. The problem is that the tumor microenvironment can progressively transform into a place where these cells are restrained, metabolically weakened, and functionally silenced. In this process, extracellular adenosine represents one of the most important mediators. Under conditions of stress, cellular damage, and hypoxia, ATP is released into the extracellular space; here, CD39 promotes its degradation into AMP, while CD73 completes the transition by transforming AMP into adenosine. A sort of change in the biological meaning of the signal thus occurs: an extracellular molecule potentially capable of supporting inflammatory and immunostimulatory signals is progressively converted into a metabolite endowed with potent immunosuppressive properties (Kaplinsky et al., 2024) [146].

Adenosine, therefore, does not simply represent a terminal product of nucleotide metabolism, but can become a true tool through which the tumor modifies the behavior of the cells surrounding it. In the tumor microenvironment, it can reach high concentrations and act through specific receptors expressed on immune cells, particularly A2A and A2B. The result does not concern a single cell population: CD4⁺ and CD8⁺ T lymphocytes, natural killer cells, dendritic cells, regulatory T cells, tumor-associated macrophages, and myeloid-derived suppressor cells are all involved. Adenosine thus ends up functioning as an environmental signal capable of communicating to the immune defenses that this territory must be tolerated rather than attacked (Xing et al., 2023) [147].

This effect becomes particularly evident when observing T lymphocytes, one of the main weapons of antitumor immunity. The T cell may be present in the tumor and may even recognize what it should attack, but the signals necessary to transform that recognition into an effective response are weakened. In fact, in the presence of adenosine, effector differentiation, proliferation, adhesion, migration, and cytotoxicity decrease, and the production of fundamental mediators is reduced. Parallel to this, the same environment favors the differentiation and activity of regulatory T cells, further shifting the balance from an antitumor response toward tolerance (Vigano et al., 2019) [148].

A particularly clear example of this mechanism comes from gastric carcinoma. In human tumor samples, greater infiltration of FoxP3+ regulatory T cells and a greater presence of CD8⁺ T cells have been observed, but the abundance of Tregs was associated with the stage of the disease and the presence of metastases. Functional experiments showed an even more important step: Tregs isolated from patients were capable of using CD39 and CD73 to degrade extracellular ATP and generate adenosine. When these cells were placed together with CD8⁺ T lymphocytes, they promoted their apoptosis, limited their proliferation, and reduced their activity. By blocking adenosine receptors, a significant portion of this suppression was attenuated (Shi et al., 2019) [149].

This finding is fundamental because it allows for the experimental tracking of almost the entire sequence: extracellular ATP, CD39 and CD73 activity, adenosine formation, increase in Tregs, and progressive loss of CD8⁺ T cell efficiency. In other words, immune evasion can be at least partially constructed metabolically. The tumor cell does not necessarily have to become invisible: it can make those who should eliminate it less efficient.

Furthermore, regulatory T cells occupy a particularly interesting position in this circuit because they can simultaneously be producers and beneficiaries of the adenosinergic environment. The co-expression of CD39 and CD73 allows them to contribute to the local production of adenosine, while stimulation of the A2A receptor can further favor their suppressive function and their recruitment into the tumor. This creates a potentially self-reinforcing mechanism: while effector cells are restrained, those tasked with immune regulation are supported (Xing et al., 2023) [147].

The phenomenon, however, is not limited to Tregs. A particularly significant step emerges from studies on ovarian carcinoma, in which it was demonstrated that the neoplastic cells themselves can possess an efficient apparatus for producing adenosine. CD39 was detected in 29 of the 36 tumor samples analyzed and CD73 in 31 of the 34 carcinomas; in tumor cell cultures, both ectonucleotidases were functionally active. Even more interestingly, ovarian carcinoma cells showed a capacity to consume extracellular ATP and generate adenosine superior to that of activated Tregs (Häusler et al., 2011) [150].

This profoundly changes the perspective. It is not only the immunoregolatory cells recruited by the tumor that build the immunosuppressive environment: the neoplastic cell itself can participate directly in its creation. It is as if the tumor were capable of building a metabolic shield around itself. And that this shield has a functional significance is suggested by the fact that the silencing or inhibition of CD39 and CD73 increased the proliferation of CD4⁺ T cells (a type of white blood cell that coordinates immune defenses), the activation of cytotoxic T cells, and the ability of NK cells to eliminate tumor cells. Similar results were obtained by blocking the A2A gene that produces the adenosine receptor, thus linking tumor production of adenosine to the effective suppression of immune cells (Häusler et al., 2011) [150].

Immune impairment, in fact, does not end with T lymphocytes. Adenosine can also reduce the maturation, migration, and cytotoxic activity of natural killer cells and compromise the ability of dendritic cells (immune system cells that recognize and capture foreign substances, such as viruses and bacteria) to support an effective immune response. Simultaneously, it can favor more tolerogenic dendritic phenotypes and support immunosuppressive populations such as Tregs, myeloid-derived suppressor cells, and tumor-associated macrophages (Xing et al., 2023) [147]. The overall result is therefore deeper than the simple inhibition of a single cell: the entire immune ecosystem is progressively reorganized.

This network also appears particularly evident in non-small cell lung carcinoma. CD73 is expressed not only by neoplastic cells but by different immune and stromal populations. The locally generated adenosine can suppress CD4⁺ and CD8⁺ T cells and interfere with dendritic cells; in parallel, macrophages, fibroblasts, and endothelial cells participate in the adenosinergic network, contributing respectively to immunosuppression, the stromal component, and angiogenic processes (Saigí et al., 2023) [151]. In this way, CD73 should not be imagined as a switch present exclusively on the surface of the tumor cell, but as one of the nodes of a metabolic network distributed among multiple components of the microenvironment.

The same principle emerges in pancreatic ductal adenocarcinoma, where the CD39-CD73-adenosine axis simultaneously involves immune cells, cancer-associated fibroblasts, and neoplastic cells. A2A and A2B contribute to the impairment of CD8⁺ T cells and NK cells and to the support of immunoregolatory populations; tumor-associated fibroblasts can express CD39 and CD73 and contribute directly to the local production of adenosine (Liu et al., 2025) [152]. The adenosinergic pathway thus takes on the characteristics of a multicellular circuit in which the tumor, stromal, and immune compartments cooperate, albeit in different ways, in the construction of an environment less favorable to immune surveillance.

It is particularly interesting to observe that this mechanism could become relevant as early as the initial phases of neoplastic transformation. In a genetically modified mouse model characterized by the spontaneous development of pancreatic intraepithelial lesions, the inhibition of CD73 reduced pancreatitis, early neoplasms, and advanced neoplasms. Simultaneously, the immune microenvironment changed: M1 macrophages increased, while M2 macrophages decreased, and the infiltration of CD4⁺ T lymphocytes, CD8⁺ T lymphocytes, and mature B cells (immune system cells that have completed their development and are ready to recognize specific foreign substances) increased (Strickland et al., 2024) [153].

The concept of a “metabolic checkpoint” becomes even more evident when observing hematological neoplasms. In this context, CD39 and CD73 participate in the transformation of extracellular ATP into adenosine, and the consequent activation of adenosinergic signaling interferes with important factors involved in T cell activation. CD8⁺ T cells that express CD39 can exhibit characteristics of immunological exhaustion, with lower production of specific proteins that allow immune system cells to communicate with each other and higher expression of co-inhibitory molecules (Jiang et al., 2023) [154].

However, the balance between extracellular ATP and adenosine can support both pro-inflammatory and immunostimulatory signals, while its degradation through CD39 and CD73 shifts the environment toward the accumulation of adenosine and immunosuppression. Nevertheless, extracellular ATP does not possess exclusively antitumor properties and, through different receptors, can in certain contexts also participate in proliferation, invasion, and metastasis (Kaplinsky et al., 2024) [146]. Tumor physiology, once again, does not function through simple on or off switches, but through dynamic balances.

Overall, what emerges is a progressive reversal of the metabolic meaning of the extracellular environment. Damage and stress can release ATP; CD39 and CD73 progressively transform it until adenosine is produced; adenosine meets its receptors on immune cells and modifies their behavior. On one hand, the proliferation, cytokine production, and cytotoxic capacity of effector cells decrease; on the other, populations and cellular programs with immunoregolatory characteristics are supported. Adenosine thus becomes a connecting point between the tumor's altered metabolism and immune evasion. The growing relevance of this axis is also evidenced by a bibliometric analysis of 1,721 publications, in which immunotherapy, immunosuppression, T cells, Tregs, immune checkpoints, and immune infiltration emerge among the main nuclei of contemporary research on CD39-CD73 (Huang et al., 2024) [155].

The most interesting proof of the biological relevance of this pathway derives from the experimental attempt to interrupt it. In preclinical models, the blockade of CD73 can reduce tumor growth and metastasis and increase the intratumoral accumulation of CD8⁺ T cells, while the antagonism of A2A and A2B receptors can favor a greater presence of effector immune cells (Vigano et al., 2019) [148].

A 2024 review also highlights how the enzymes CD39 and CD73 favor the ability of tumors to escape immune defenses. These enzymes transform ATP into adenosine, and its accumulation therefore contributes to creating an environment favorable to tumor growth. CD39 could also help predict the response to immunotherapy. Blocking CD39 and CD73 represents a possible therapeutic strategy, currently being studied through experimental drugs, including in combination with other immunotherapies (Kaplinsky et al., 2024) [146].

Tumor immune evasion therefore does not appear as the result of a single signal, but as the consequence of a network in which metabolism, hypoxia, neoplastic cells, stroma, and the immune system communicate continuously. Within this network, CD73 occupies a strategic position because it participates in the transformation of AMP into adenosine and therefore in the construction of an environment capable of restraining the cells that should eliminate the tumor. Adenosine, in turn, is not limited to turning off a single response: it can simultaneously weaken multiple arms of antitumor immunity and strengthen those tasked with regulation (Xing et al., 2023) [147].

A fundamental principle is thus outlined: to survive, a neoplasm does not necessarily have to completely escape the gaze of the immune system. It can also modify the environment in which that immune system is forced to operate. Through CD39, CD73, and the accumulation of adenosine, a metabolic signal can be transformed into an immunological signal; and what was initially a territory marked by cellular damage and ATP release can progressively become an environment in which the antitumor response is contained. It is precisely in this transformation that metabolism and immune evasion end up becoming two sides of the same process.

Based on what has been described, we have seen how adenosine can contribute to immune evasion by directly weakening T cells, NK cells, and other components of the antitumor response. However, there is a second, equally important effect: adenosine does not merely shut down the cells that should be attacking the tumor, but can actively modify the behavior of the macrophages present in the microenvironment. These cells possess remarkable plasticity and can assume different functional states in relation to the metabolic, inflammatory, and tissue signals to which they are exposed. The traditional distinction between M1 and M2 macrophages remains useful for describing the extremes of this spectrum, but it should not be interpreted as a rigid classification. Adenosine represents precisely one of the signals capable of orienting macrophages toward anti-inflammatory, pro-angiogenic, and immunoregulatory characteristics that, in the tumor context, can become favorable to the progression of the neoplasm.

This step is fundamental because it prevents us from considering “M2” as a single cellular identity. The macrophage can achieve protumoral functions through different molecular programs. In the case of adenosine, the increase in IL-10 (molecules produced mainly by immune system cells, with an anti-inflammatory function) and VEGF is particularly significant: the former contributes to reducing inflammatory tone and immune activity, while the latter promotes angiogenesis and tissue adaptation. Adenosine can therefore transform a cell born to react to damage into a cell that participates in the resolution and repair phase; in a tumor, however, a program physiologically useful for healing can be exploited to protect and vascularize the neoplastic tissue.

The A2A receptor represents one of the main mediators of this response. Mechanistic evidence comes from a model of chronic cerebral hypoperfusion, in which the activation of A2AR under conditions of hypoxia and low glucose availability favored an M2 phenotype in macrophages, increasing IL-10 and reducing TNF-α and IL-1β (two molecules produced mainly by immune cells, which act as chemical messengers that activate and regulate inflammation). When the nuclear receptor was silenced, the increase in IL-10 and M2 polarization were attenuated (Mou et al., 2022) [156]. Although this is not a tumor model, this study shows that the A2A signal can directly reach a transcriptional program associated with anti-inflammatory characteristics through the receptor.

In a tumor, this relationship becomes particularly evident when observing the spatial distribution of adenosine. In pancreatic ductal adenocarcinoma, some researchers have shown that extracellular adenosine is not uniformly distributed in the neoplastic mass. The highest concentrations are found primarily in hypoxic regions, exactly where myeloid cells accumulate. M2 macrophages show high levels of the human gene that produces the A2A adenosine receptor, which are particularly marked in tumors resistant to immuno-oncology therapies. When the production of adenosine via CD73 and its signaling via Adora2a (genes that produce A2A adenosine receptors) are blocked simultaneously, M2 macrophages and Tregs decrease, and tumor growth and metastasis are reduced (Graziano et al., 2023) [157].

These results allow us to connect many of the mechanisms described in the previous paragraphs into a single sequence. Tumor growth produces poorly oxygenated regions; hypoxia increases HIF-1α and promotes the CD39/CD73 system; this system increases adenosine; adenosine encounters macrophages rich in A2A receptors and helps support their M2 characteristics. We no longer have a simple succession of independent events, but a network in which hypoxia, purinergic metabolism (a communication system between cells that uses molecules such as ATP and adenosine), and myeloid polarization reinforce each other.

An equally clear example comes from chronic lymphocytic leukemia. In the lymph nodes of patients, a marked activation of HIF-1α was observed, associated with increased adenosine production. Under hypoxic conditions, monocytes (immune system cells present in the blood) are directed toward macrophages with M2 characteristics, increasing the production of IL-6. At the same time, these macrophages acquire greater expression of CD73 and the ability to generate adenosine, while A2A increases in the various cell populations of the niche. A2A antagonism attenuates the hypoxia-induced macrophage phenotype and recovers part of the compromised immune functions (Serra et al., 2016) [158].

Here, an even more interesting element appears: the macrophage is not simply the target of adenosine, but can also become a source of it. Under hypoxic conditions, it acquires CD73, increases the generation of adenosine, and simultaneously expresses more of the receptor through which it responds to that same molecule. This creates the possibility of an autocrine and paracrine circuit (in the autocrine circuit, the cell communicates with itself; in the paracrine one, it communicates with surrounding cells) in which the microenvironment induces an immunosuppressive phenotype that, once established, contributes to maintaining the same environment.

Ovarian carcinoma offers an even more direct demonstration of this self-amplifying mechanism. Tumor cells that express CD39 and CD73 produce adenosine and promote the migration of monocytes toward the neoplasm. Once recruited, these monocytes can differentiate into TAM-like macrophages with M2 characteristics, showing higher IL-10 compared to IL-12. The resulting macrophages in turn express CD39 and CD73 and produce biologically active adenosine at concentrations comparable to those generated by tumor cells (Montalbán del Barrio et al., 2016) [159].

The circuit thus becomes: tumor adenosine → myeloid recruitment → TAM/M2 differentiation → further adenosine production. The reprogrammed macrophages are also capable of suppressing the proliferation of CD4+ T cells. The tumor thus builds a system in which the myeloid compartment becomes both a consequence and an amplifier of adenosinergic metabolism.

Particularly useful evidence also comes from the reverse process. In a mouse model of pancreatic intraepithelial neoplasia, the inhibition of CD73 significantly reduces the population of M2 macrophages and increases that of M1. This immune remodeling is accompanied by a decrease in fibrosis and a reduction in the progression of PanINs (pancreatic intraepithelial neoplasias) (Strickland et al., 2024) [153]. The result takes on a specific meaning in this context: by reducing the ability to produce adenosine via CD73, the macrophage balance of the tissue changes concretely.

Furthermore, the production of adenosine does not depend solely on tumor cells and macrophages. Regulatory T cells can also contribute significantly. Some researchers have shown that the stabilization of HIF-1α strongly increases CD73 in human Tregs. When CD73 is co-expressed with CD39, these cells acquire the entire enzymatic apparatus necessary to convert extracellular ATP into adenosine. The metabolically reprogrammed Tregs thus show a particularly high suppressive capacity (Jarvis et al., 2021) [160].

This data adds another layer to the network. In the hypoxic tumor microenvironment, HIF-1α can simultaneously promote CD73 in immunoregulatory cells and the production of adenosine; adenosine can then act on macrophages and support M2 states, while Tregs and TAMs cooperate in further reducing antitumor immunity. The review on Treg metabolism reinforces this very connection between hypoxia, HIF-1α, lactate, and adenosine, describing regulatory cells as metabolically adaptable populations capable of maintaining their immunosuppressive activity even in a TME poor in oxygen and nutrients (Mortezaee, 2026) [161].

This leads to a vision very different from the one with which we began. The M2 macrophage is not simply “present” in the tumor: it can be progressively built by the microenvironment. Hypoxia modifies HIF-1α; HIF-1α promotes CD73 and adenosine; adenosine acts on the A2A and A2B receptors of the myeloid compartment; macrophages reduce inflammatory programs and increase IL-10, VEGF, and other characteristics associated with immunoregulatory and protumoral phenotypes. Once reprogrammed, these same macrophages can produce further adenosine and amplify the circuit. Several times in this chapter, we have introduced the vascular endothelial growth factor VEGF, which, as we will see in the next paragraph, also represents an important effector downstream of the PI3K/AKT/mTOR/HIF-1α cascade.

As a tumor mass grows, its demand for oxygen and nutrients increases rapidly. However, the pre-existing vascularization is not always able to support this expansion, and hypoxic areas form within the tumor. As we have seen, hypoxia leads to the stabilization of HIF-1α, which modifies the cell's transcriptional program, allowing it to adapt to unfavorable environmental conditions. One of the most significant consequences of this response is the increased expression of factors belonging to the VEGF family, particularly VEGFA, one of the main regulators of tumor angiogenesis (the process by which a tumor stimulates the formation of new blood vessels). Intuitively, a neoplastic cell experiencing oxygen deficiency does not simply modify its own metabolism: through HIF-1α, it can send a genuine "distress signal" to the surrounding vascular compartment, stimulating the formation of new blood vessels. This process may initially seem advantageous because it attempts to restore oxygen supply, but in the neoplastic context, it ends up providing the tumor mass with new resources to support growth, survival, and spread.

A particularly clear example of this connection comes from hepatocellular carcinoma. Researchers have identified a mechanism in which the loss of a specific enzyme leads to the accumulation of the metabolite succinylacetone (a highly toxic chemical byproduct): the latter promotes the stabilization of HIF-1α and increases the expression of VEGFA. In parallel, the proliferation, migration, and angiogenic capacity of endothelial cells increase. In clinical samples of hepatocellular carcinoma, the authors also observed an inverse relationship between the enzyme and HIF-1α (Luo et al., 2023) [162].

However, the relationship between HIF-1α and VEGF does not depend exclusively on hypoxia or metabolites. The proliferative pathways described in the previous paragraphs can also converge on this axis. In osteosarcoma, the overexpression of a gene activates the PI3K/AKT/mTOR cascade, subsequently increasing HIF-1α and VEGF. This phenomenon is accompanied by the maintenance of characteristics similar to those of tumor stem cells, greater ability to form spheroids (small groups of cells cultured in the laboratory that organize into three-dimensional structures), migration, invasion, chemoresistance, and lung metastases. When PI3K or HIF-1α are inhibited, the oncogenic effects of the gene overexpression are substantially nullified (Wang et al., 2026) [163]. This data reinforces the idea that VEGF is not merely a passive response to oxygen deficiency, but can represent the end of a signaling network that starts from upstream oncogenic pathways.

Experimental confirmation of this same principle comes from hepatocellular carcinoma. A derivative of epigallocatechin gallate (a natural principle found in green tea) reduced tumor proliferation, microvessel density, and xenograft growth. These effects were associated with a decrease in the PI3K/AKT/mTOR pathway and, in parallel, a reduction in HIF-1α and VEGF at both the transcriptional and protein levels (Liao et al., 2020) [164]. The importance of this experiment lies not so much in the compound used, but in the observation that interfering with upstream proliferative pathways is accompanied by a reduction in the HIF-1α/VEGF axis and decreased tumor vascular formation.

A similar phenomenon has been observed in breast cancer. Echinacoside (an active ingredient of plant origin) reduced the expression of PI3K, AKT, HIF-1α, and VEGFA in MCF-7 cells (human tumor cells derived from breast carcinoma) and in animal models, accompanied by lower proliferation, increased apoptosis, and reduced tumor growth (Liang et al., 2024) [165]. In this case, too, the cascade appears continuous: the reduction of upstream growth signals is reflected in the hypoxic response and, ultimately, in the availability of the main angiogenic stimulus.

The HIF-1α/VEGF axis also assumes particular importance in metastasis. In cervical cancer, a specific gene is more highly expressed in patients with lymph node metastases. Its overexpression increases the migration and invasion of tumor cells, induces the epithelial-mesenchymal transition, and simultaneously increases HIF-1α and VEGFA at the mRNA and protein levels. Conversely, silencing the involved gene reduces migratory and invasive capacity (Peng et al., 2024) [166]. These results suggest that the activation of the HIF-1α/VEGFA axis may be part of a broader program through which the tumor cell simultaneously acquires invasive capacity and vascular support.

In melanoma, the same relationship was observed by studying the effect of luteolin (a flavonoid, a natural substance belonging to the polyphenol family). At concentrations that were not significantly cytotoxic, the compound reduced migration, invasion, adhesion, and the formation of tubular structures by endothelial cells. At the same time, AKT, HIF-1α, and VEGF-A levels decreased (Li et al., 2019) [167]. In this model, too, angiogenesis and invasiveness do not appear to be independent processes, but parts of the same coordinated molecular response.

Hepatocellular carcinoma offers further confirmation. Experimental treatment with arsenic sulfide reduced HIF-1α and VEGF, decreased VEGF secretion, and compromised the ability of tumor cells to induce the formation of vascular structures. Simultaneously, a reduction in migration and invasion and a modification of EMT markers were observed (Lu et al., 2023) [168]. The data therefore support the existence of a functional relationship between HIF-1α, VEGF, angiogenesis, and metastatic behavior.

Finally, in glioblastoma, the relationship between HIF-1α and VEGFA is directly involved in the dialogue between tumor cells and the microenvironment. A specific gene increases proliferation, migration, invasion, and angiogenic capacity through the HIF-1α/VEGFA axis. Glioblastoma cells also release exosomes capable of modifying surrounding astrocytes, inducing them to express higher levels of HIF-1α and VEGFA, and contributing to the transmission of resistance to chemotherapy. Reducing the drug decreases the activation of the HIF-1α/VEGFA axis in both tumor cells and the microenvironment (Shen et al., 2025) [169]. This study therefore shows that VEGF should not be considered only as the product of a single hypoxic cell: it can become part of a communication network that involves stromal cells and contributes to therapeutic resistance.

Overall, this evidence shows how the HIF-1α/VEGF axis represents one of the main tools through which the tumor cell translates oxygen deficiency and oncogenic signals into a response capable of actively modifying its environment. By stabilizing HIF-1α, the cell increases the expression of VEGF factors, stimulating angiogenesis. In this way, the tumor does not simply adapt to an oxygen-poor environment but attempts to remodel it to its own advantage. This response also intertwines with EMT, invasion, stemness, and therapeutic resistance, showing once again how metabolism, signaling, and the microenvironment are inseparable aspects of neoplastic progression.