Bearing a five-carbon ribose moiety, inosine has the potential of being an excellent carbon source, evidenced by the ribose 5-phosphate (R5P) in the well-known pentose phosphate pathway (PPP). How inosine is incorporated into the PPP and the subsequent glycolysis pathway remains to be determined. This phenomenon is depicted by Wang et al. (8) using immune cells from both mouse and human. The phenomenon is based on the tumor microenvironment (TME), which showcases cancer cells’ intrusion within the host cell habitat, resulting in an assembly of diverse cell types (11). The TME is characterized by hypoxia and a dearth of nutrients due to the enhanced glucose consumption of cancer cells (12). This limitation in glucose availability can impair the function and survival of immune cells, which requires the rewiring of immune cells to adapt to this environment (13). One of them is the CD8⁺ T cell, which plays a crucial role in anti-tumor immune responses. Given the critical environment, CD8⁺ T cell relies on inosine as a compensatory energy substrate (8).

The inosine enters the CD8⁺ T cell though a protein transporter. However, both studies (8, 14) skip specifying the name of the transporter. For inosine to accomplish its role as a carbon source, the ribose moiety should be isolated before entering the PPP. The route of the inosine in the carbohydrate metabolism is tracked by a stable-isotope-based metabolomics approach. The ribose moiety of inosine is labeled with the ¹³C isotope ([1′,2′,3′,4′,5′-¹³C₅]inosine) (8). It is shown that the ribose isolation step in CD8⁺ T cells is handled by the enzyme PNP, which converts inosine to hypoxanthine and R1P ( Figure 1 ). As an interconvertible substrate with the R5P molecule, R1P is readily converted to R5P catalyzed by phosphopentomutase (7). R5P, an intermediate of PPP, undergoes the PPP along with xylulose 5-phosphate (Xu5P), providing an end product fructose 6-phosphate (F6P), which is one the important intermediates of the glycolysis pathway. Following the glycolysis pathway and subsequent TCA cycle, the ribose moiety of inosine achieves its role as a carbon source ( Figure 2 ). In the same study (8), it is also reasoned that the PNP is an essential enzyme in the aforementioned process. Upon the use of a PNP inhibitor, forodesine (foro), a reduction in the bioenergetic activity, proliferation, viability, and tumor-killing activity is obvious in both mouse and human T effector cells cultured in inosine-containing medium. Furthermore, PNP is knocked down with PNP-targeted short-interfering RNA (siRNA), and the results follow the previous experiment. Both experiments suggest PNP as a crucial enzyme in inosine metabolism as a carbon source, which suggests a promising therapeutic strategy to manipulate T-cell metabolism and to enhance anti-tumor immune responses. There is another interesting point in this metabolic route. The hypoxanthine obtained from the breakdown of inosine by PNP, as mentioned before, is not only a nondescript by-product of the whole process as many may think. The uric acid derived from hypoxanthine has an anti-tumor effect by acting directly on CD8⁺ T cells (15, 16).

As inosine has this imperative role as an alternative carbon source in CD8⁺ T cells to fight against cancer, some microbiota that can produce inosine as metabolites show great enhancement in immune checkpoint blockade treatment (9, 17, 18).

In addition to its role in glycolysis (as a novel carbon source), inosine significantly influences mitochondrial respiratory capacity by integrating key metabolic pathways and signaling networks.

As mentioned earlier, the TME is characterized by hypoxia and nutrient scarcity, making alternative energy sources crucial. In such conditions, inosine helps buffer the energy deficit by modulating mitochondrial metabolism via the mTORC1 signaling pathway. In brief, mTORC1, one of the mTOR complexes, detects growth factors and abundance of nutrients and relays the signals to promote cell growth through stimulating biosynthesis (19). Li et al. (10) first find that inosine tends to accumulate in the tumor core region of certain tumors. Later experiments prove that the elevation of inosine promotes breast cancer cells’ (MDA-MB-231) survival in glucose- and glutamine-depleted conditions. The result is corroborated by a previous study, that inosine is capable of stimulating cancer cell proliferation (20). Different pathways are involved, which will be discussed shortly. In Li et al. (10), the source of inosine is endogenous, obtained from the breakdown of IMP by 5′-nucleotidase, cytosolic II (NT5C2) (21, 22). Cytosolic inosine enhances the level of the transcription factor specificity protein 1 (SP1) (the detailed mechanism has not been discovered yet); SP1 directly binds Ras-related GTP-binding proteins (Rag) GTPase gene promoters and enhances their transcription. There are four types of Rag GTPases: RagA, RagB, RagC, and RagD. The active Rag heterodimer (RagA/B^GTP–RagC/D^GDP) is able to bind to raptor, a subunit of mTORC1, and recruit it to the surface of the lysosome (23). At the lysosome, mTORC1 is activated by Rheb (a small GTPase) (24), leading to the upregulation of citrate synthase, aconitase 1, and other enzymes essential for the TCA cycle (10) ( Figure 3 ). This enhancement of TCA cycle flux increases mitochondrial ATP production, supporting the energy demands of rapidly proliferating cancer cells, promoting the progression of cancer. However, in another study (20), it is suggested that inosine takes effect via binding to adenosine receptors A2B and A3 only (antagonists targeting the receptors A1 and A2A do not abolish the effect of inosine), possibly triggering the cyclic adenosine 5′-monophosphate–protein kinase A (cAMP–PKA) pathway.

Another question that may be asked is as follows: Given that glucose remains scarce despite the fact that TCA cycle enzymes are replete, what can the cancer cells “burn” for energy? In fact, less than 50% of the carbons in the acetyl coenzyme A (acetyl-CoA) pool are contributed by glucose (25). Cancer cells are able to utilize other carbon sources such as acetate, ketone bodies (25, 26), and inosine.

Rown adipose tissue (BAT) is a key site of non-shivering thermogenesis, which makes it an important therapeutic target for obesity and type 2 diabetes (28). Multiple factors can affect the browning/beiging of white adipose tissue (WAT) (into BAT) and the thermogenesis of BAT such as cold exposure (29), obesity (30), cancer cachexia (31), and diet (32). Niemann et al. (33) have identified inosine as a key regulator of BAT activation and thermogenesis in both humans and rodents. Apoptotic brown adipocytes release purine metabolites, including AMP, inosine, and hypoxanthine, among which only inosine has significant effects on the BAT, which is supported by another peer study (34). By binding to adenosine receptors, especially A2A and A2B, inosine increases intracellular cAMP levels, the main second messenger that boosts the differentiation process (WAT to BAT) and the thermogenesis within brown adipocytes. To be specific, PKA activated by cAMP further activates downstream targets such as p38 mitogen-activated protein kinase (p38 MAPK). It further reduces the phosphorylation of Salt-inducible kinase 2 (Sik2) (Ser576) and causes a subsequent decrease in phosphorylation of cAMP-regulated transcriptional coactivator 3 (Crtc3) (Ser72, Ser162, Ser329, and Ser370). This dephosphorylation has been proven to induce Crtc3 nuclear import and to instigate the association and activation of cAMP-responsive element binding protein 1 (Creb1) (35). These factors enhance the browning process of WAT and upregulate thermogenic gene, e.g., uncoupling protein 1 (UCP1) expression (36, 37) ( Figure 4 ). Furthermore, inosine stimulates the mTORC1 and the MAPK extracellular signal-regulated kinase pathway, which are recognized for their roles in promoting brown adipocyte differentiation and the browning of WAT (33, 38). Given its role as a para- and autocrine signaling molecule driving energy expenditure in brown/beige adipocytes, inosine represents a promising target for anti-obesity interventions (33). Further efforts should focus on the efficacy and safety of this approach in preclinical models.

The relationship between adenosine receptors and inflammation has been well discovered. A large number of both in vitro and in vivo studies have shown the anti-inflammatory effects of adenosine receptors on multiple kinds of inflammatory cells and in different inflammatory diseases (39). However, inosine also has an effect on adenosine receptors, triggering similar anti-inflammation processes. Among the four subtypes of adenosine receptors, which receptors does inosine mainly bind to during the anti-inflammatory process remain to be determined. Haskó et al. (40) suggest that A1 and A2 receptors are mostly involved, while another study by Gomez and Sitkovsky (41) mentioned A2A and A3 receptors. The activation of these receptors suppresses pro-inflammatory cytokine production, mitigating inflammation-induced tissue damage and providing protection against endotoxin-induced shock (40, 41).

The study by Lima et al. (42) reveals the detailed mechanisms of the anti-inflammatory effect of inosine. In hypercholesterolemic rats (42), the expression of inducible nitric oxide synthase (iNOS) is high (43). The high expression of iNOS is associated with the stabilization of hypoxia-induced factor 1 (HIF-1) (43, 44), resulting in inflammation and oxidative stress. In addition, by synthesizing a large amount of nitric oxide (NO), iNOS can cause further tissue damage (43, 45). By binding particularly to the A2A and A3 receptors, inosine again promotes the cascade of cAMP–PKA. PKA phosphorylates the endothelial nitric oxide synthase (eNOS) at Ser1177. Upon phosphorylation, eNOS becomes active and produces NO in a moderate amount, which acts as a vasodilator, relaxing blood vessels and reducing blood pressure (42, 46). Additionally, NO signals the activation of the soluble guanylate cyclase (sGC), increasing cyclic guanosine 3′,5′-monophosphate (cGMP) levels, which then activates protein kinase G (PKG). PKG promotes vasodilation and protects against endothelial dysfunction (42). These findings are in accord with the effect of inosine on cAMP and PKA level in previous studies, contributing to its antiplatelet and anti-inflammatory abilities (47, 48).

Aside from the aforementioned PKA–eNOS–PKG pathway, another inosine-induced anti-inflammatory pathway [p38 MAPK/nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB)/vascular cell adhesion molecule-1 (VCAM-1)] has also been demonstrated by Lima et al. (42). p38 MAPK is a key stress-activated kinase that responds to inflammatory stimuli (49). In hypercholesterolemic rat models, activated p38 MAPK phosphorylates downstream targets, including transcription factors like NF-κB. NF-κB promotes the expression of adhesion molecules, e.g., VCAM-1 and a series of inflammatory genes, contributing to atherosclerosis progression (42, 49). After treatment with inosine, the level of the NF-κB inhibitor, inhibitor of nuclear factor kappa B, alpha (IκB-α), in the hypercholesteremic rat models shows an obvious increase. Accordingly, there is also a marked reduction in p38 MAPK, NF-κB, and VCAM-1 expression (42). Note that there is a contradiction here between this pathway and that mentioned in Niemann et al. (33), where p38 MAPK is activated by inosine. Our hypothesis is that either the hypercholesteremic state impacts it or other unexplained mechanisms are present; further studies are needed to elucidate this matter.

Emerging evidence suggests that inosine enhances brain-derived neurotrophic factor (BDNF) signaling, which plays a vital role in neuronal survival, synaptic plasticity, and cognitive function in neurodegenerative diseases (50).

BDNF exerts its neuroprotective effects by activating the tropomyosin receptor kinase B (TrkB) receptor, a key mediator of long-term potentiation (LTP), which is essential for memory formation and synaptic transmission (51). It is indicated that the activation of adenosine receptors, particularly A1 and A2A, is necessary to initiate different BDNF-related functions (52). This activation ensures the maintenance of regular BDNF levels and the enhancement of BDNF-triggered enhancement of synaptic transmission within the hippocampus (53).

Many studies have shown the imperative role of inosine in enhancing the strength of BDNF upon binding to A1 and/or A2A receptors. In experimental models of Alzheimer’s disease (AD), induced by streptozotocin (STZ), the levels of TrkB receptor and BDNF mRNA expression in the hippocampus are drastically reduced. Notably, inosine administration significantly restored TrkB and BDNF levels (50). Blocking A1 or A2A receptors inhibits the inosine-induced increase in BDNF in brain (52, 54).

DSH is a rare pigmentary genodermatosis, which is acquired by autosomal dominant inheritance with high penetrance. DSH typically presents with a mix of hyperpigmented and hypopigmented macules on the backs of the hands and feet, along with freckle-like lesions on the face. The rashes generally begin during infancy or early childhood (66).

AGS is a severe, genetically inherited neuroinflammatory disorder that mimics congenital viral infection. It typically presents in infancy with symptoms such as encephalopathy, intracranial calcifications, leukodystrophy, elevated interferon-α in the cerebrospinal fluid (CSF), and developmental delays (67).

Both diseases are found to be related to ADAR1. DSH has been found to be solely controlled by ADAR1 (68), while multiple genes have been found to be responsible for AGS, e.g., three prime repair exonuclease 1 (TREX1), ribonuclease h2 subunits a, b, and c (RNASEH2A–C), SAM domain and HD domain–containing protein 1 (SAMHD1), and interferon induced with helicase c domain 1 [IFIH1 or the protein it encodes, melanoma differentiation-associated protein 5 (MDA5)], including ADAR1 (69). The exact relationship between ADAR1 and DSH is still unclear. A few studies hypothesize that dysregulated RNA editing affects the expression of genes involved in melanocyte function, pigment distribution, or skin homeostasis (70–72), which leads to the pigmentary phenotype of DSH. AGS, on the other hand, is a result of an autoimmune response, caused by the accumulation of unedited self-RNA (73). Normally, A→I editing by ADAR1 stops the immune system from mistaking self-dsRNA as foreign by preventing its detection by sensors like MDA5, which triggers the mitochondrial antiviral signaling protein (MAVS) pathway and type I interferon production. Without proper editing, the body mistakes its own RNA for viral RNA, leading to constant interferon activation and inflammation, especially in the brain—a key feature of AGS (64, 74).

A → I RNA editing mediated by ADAR1 and ADAR2 is a vital post-transcriptional modification that reshapes both coding and noncoding RNAs in cancer. ADAR1-driven editing of coding transcripts—such as antizyme inhibitor 1 (AZIN1) (serine→glycine substitution)—enhances protein oncogenic potential across colorectal, endometrial, gastric, liver, and esophageal cancers (75–80). ADAR1 also impairs tumor suppressors [e.g., filamin B in triple-negative breast cancer, bladder cancer associated protein (BLCAP) in cervical cancer, and Nei-like DNA glycosylase 1 (NEIL1) in multiple myeloma], promoting proliferation, invasion, signal transducer and activator of transcription 3 (STAT3) activation, or genomic instability (81–83). In contrast, in contexts such as breast cancer, ADAR1-mediated editing of gamma-aminobutyric acid type A receptor subunit alpha 3 (GABRA3) attenuates protein kinase B (AKT) signaling and suppresses metastasis (84). The bimodal feature makes utilizing inosine as a novel therapeutic molecule more difficult, where more future studies are warranted.

Beyond coding RNAs, ADAR1 regulates noncoding RNAs—especially miRNAs and circular RNAs (circRNAs). In hepatocellular carcinoma (HCC), editing of miR-3144-3p by ADAR1 shifts its targeting away from tumor suppressors and toward oncogenes, and editing of pri-miR-26a or miR-200b in leukemia and thyroid cancer, respectively, disrupts miRNA maturation or epithelial–mesenchymal transition (EMT) suppression (85–88). ADAR1 also influences circRNA biogenesis by editing Alu-paired sequences, suppressing tumor-suppressive circRNA ARSP91 (circARSP91) in HCC, while certain circRNAs such as circRNA derived from the NEIL3 gene (circNEIL3) provide feedback to regulate ADAR1 expression (89–91).

ADAR2 generally functions as a tumor-suppressive RNA editor and is frequently downregulated in multiple cancers, with restoration of its editing activity reversing malignant phenotypes. In glioblastoma multiforme, low ADAR2 protein levels correlate with poorer overall survival, and suppression of ADAR2 enhances proliferation, migration, and anchorage-independent growth, and upregulates invasion-associated genes, e.g., a disintegrin and metalloproteinase domain-containing protein 12 (ADAM12), pentraxin 3, and EMT pathway genes (92). In esophageal squamous cell carcinoma, ADAR2 is reduced in tumors, and its overexpression induces apoptosis via editing and stabilizing insulin-like growth factor-binding protein 7 (IGFBP7), inhibiting Akt signaling. Knockdown of ADAR2 or IGFBP7 eliminates this pro-apoptotic effect (93). In the core-binding factor acute myeloid leukemia with t(8;21) or inv(16) fusions, transcriptional repression of ADAR2 by runt-related transcription factor 1 (RUNX1-ETO) reduces editing of key targets, coatomer protein complex subunit alpha (COPA), and component of oligomeric Golgi complex 3 (COG3). Loss of ADAR2 catalytic activity enhances leukemic clonogenicity (94).