Lipid droplets are fundamental in the storage and release of fatty acids. Lipid droplets are ubiquitously present organelles in many cells with a hydrophobic core surrounded by a single layer of phospholipids decorated with various sets of proteins. The hydrophobic core contains neutral lipids, including triacylglycerols, sterol esters, retinyl esters and 1-O-acylceramides (54–57). Emerging evidence indicated lipid droplet accumulation in several cancers (58, 59), especially in hypoxic cancer cells, and several studies have linked lipid droplet abundance with more aggressive tumor phenotype (34, 36, 58, 60). It has also shown lipid metabolism-dependent proliferation and survival of glioblastoma following radiation (61, 62), antiangiogenic (63) and ketone diet therapy (64). As such, they have recently been proposed to play direct roles in many cancers and considered a hallmark of fundamental tumor processes (58, 60, 65). The following sections discuss the possible roles of lipid droplet metabolism, including (1) fatty acid storage as neutral lipids, (2) lipolysis of neutral lipids, and (3) lipophagy in limiting or triggering ferroptosis.

Lipogenesis (lipid acquisition) is an energy-expensive process of fatty acid synthesis and subsequent esterification to neutral lipids into lipid droplets. The primary process of synthesizing fatty acids from acetyl‐CoA subunits that are produced most commonly from carbohydrate catabolism is called de novo lipogenesis (DNL) (66). Acetyl-coenzyme A (acetyl-CoA) is a key metabolite precursor in DNL, and its abundance is closely monitored by cellular adaptive mechanisms. While glucose is the most common supplier of carbon units for DNL ( Figure 1A ), fructose is metabolized to glycerol and subsequently converted to a lipogenic substrate glycerol-3-phosphate (G3P) that can drive DNL.

Glutamine is another non-lipid metabolite that contributes carbon to lipogenic acetyl-CoA ( Figure 1A ) through two distinct pathways: Glutamine metabolism can produce citrate either through the tricarboxylic acid (TCA) cycle in mitochondria or by the reductive carboxylation of cytoplasmic α-ketoglutarate (αKG) to citrate by IDH1 ( Figure 2 ) (67, 68). Interestingly, many cancer cells, including glioblastoma, generate 10–25% of their lipogenic acetyl-CoA from glutamine through reductively carboxylation (69), highlighting the general use of reductive carboxylation as the primary route to convert non-lipid carbon to lipids in cancer cells. In agreement with this, blood-borne glucose contribute minimal fraction to the acetyl-CoA pool of glioblastoma and brain metastases that originate from other tissues (70). In addition to a direct contribution of supplying a carbon unit, glucose and glutamine uptake promotes lipogenesis through transcriptional regulation in a sterol regulatory element-binding protein (SREBP) dependent manner in glioblastoma (71, 72).

Acetate, often dispensable for cells, has also been shown as a source of acetyl-CoA for DNL in cancer cells, especially primary glioblastomas and other cancer metastasis to brain (73, 74). In this case, three important processes need to be considered: acetate production, its intercellular import across the cell membrane, and the carbon fluxes from acetate to lipids. Imported acetate is converted to acetyl-CoA in a reaction catalyzed by acetyl-CoA synthetase 2 (ACSS2) ( Figure 1A ) (75). Thus, hypoxia-associated deficits in acetyl-CoA can be supplemented by increasing the use of free extracellular acetate (including plasma, interstitial fluid, and neighboring cells) and the intercellular pool of mainly acetate released from the deacetylation of histones and other cellular proteins (76). In line with acetate metabolism, apparent ACSS2 expression is associated with poor progenesis in glioblastoma patients (73, 74). Despite the fundamental importance of acetate and its vital roles to several intracellular and extracellular metabolism, little is known about the mechanisms and regulatory processes in acetate metabolism.

Cytoplasmic acetyl-CoA is then carboxylated by acetyl-CoA carboxylase (ACC) to produce malonyl-CoA ( Figure 1B ), which is further used to synthesizes medium-and long-chain fatty acids. The key three proteins involved in the initial fatty acyl chain biosynthesis, ACLY, ACC, and FASN, are commonly upregulated and activated in many cancers (36, 60, 65), and they are in the frontline of pre-clinical target evaluations, drug discovery pipelines, and clinical trials. For instance, the therapeutic targeting of FASN is currently being explored in a phase 2 clinical trial in patients with glioblastoma using the FASN inhibitor TVB-2640 in combination with bevacizumab (NCT03032484) (28).

De novo synthesized fatty acids are then further esterified to neutral lipids and stored in their nontoxic (inert, neutral) form in lipid droplets ( Figure 1B ). Esterification starts with the acylation of glycerol-3-phosphate with two fatty acids to produce phosphatidic acid (PA), followed by dephosphorylation to yields diacylglycerol (DAG). DAG then serves as a precursor for the synthesis of both phospholipids and acylation to yield triacylglycerols (TAGs), which is catalyzed by diacylglycerol acyltransferase 1 and 2 (DGAT1 and DGAT2) (77, 78). TAG is stored in lipid droplets as energy reservoir, supplying the cell with energy when required.

Lipid droplets also store 1-O-acylceramides and cholesteryl ester. Although some studies indicated a minimal abundance of1-O-acylceramides in tissue (0.25-1.3 per mg of mice tissue), the indication of its level during Western diet consumption (79) warrants special attention, mainly its correlation with cancer progression in obese glioblastoma patients. Cholesteryl ester is another neutral lipid stored in lipid droplets. Acyl-CoA acyltransferase, also known as acyl-Coenzyme A or cholesterol acyltransferase (SOAT1 and SOAT2), catalyzes the formation of cholesteryl esters, using cholesterol and long-chain fatty acyl-CoA as substrates. Both isoforms of acyl-CoA acyltransferase (ACAT1 and ACAT2) are highly expressed and post-translationally regulated in glioblastomas and other cancers, and their expression and activation levels correlate with patient survival (79–81). In vivo studies showed that both genetic silencing of ACAT1 or blocking its activity using inhibitors suppresses tumor growth (80, 82, 83). These suggest that targeting ACAT1 and cholesteryl ester synthesis may be a promising anticancer strategy.

The de novo synthesis of fatty acids can be extended to the formation of either saturated or unsaturated fatty acids. Unsaturated fatty acids can be further classified into monounsaturated fatty acids [MUFAs, with only one double bond, e.g., oleic acid (18:1)] and polyunsaturated fatty acids [PUFAs, with at least two double bonds, e.g., linoleic acid (18:2 n-6) and α-linolenic (18:3 n-3)]. Unsaturated fatty acids can bind membrane phospholipids in differential degrees that subject tumor cells to oxidative stress. Studies have revealed that an increase in the MUFA/PUFA ratio results in fewer peroxidation-susceptible targets, ultimately reducing the susceptibility of cancer cells to ferroptosis (84). On the other hand, an increase in the synthesis of PUFAs can promote subsequent lipid peroxidation in cancer cells under oxidative stress conditions, suggesting that the formation of acetyl-CoA derivatives (de novo synthesis) of these PUFA is necessary for producing ferroptosis death signals (43, 45). Similarly, the esterification of fatty acids into neutral lipids has been shown to deplete the substrates for lipid peroxidation and increase the resistance of cancer cells to ferroptosis-mediated cell death (34, 36, 60, 85).

Lipid peroxidation normally occurs at a slow rate because PUFAs are required to react with the free radicals generated by iron acyl-CoA synthase long-chain family member 4 (ACSL4). Using genetic ablation and pharmaceutical inhibition, it has been shown that ACSL4 is a rate-limiting enzyme to catalyze the conversion of long-chain PUFAs, such as arachidonic acid and eicosapentaenoic acid to PUFA-CoA, and ultimately this increases lipid peroxidation and decreases resistance to ferroptosis (86–88). Under oxidative or energetic stress, PUFAs, particularly arachidonoyl (AA, 20:4 n-6) and adrenic acid (AdA), are oxygenated by different classes of enzymes, including ACSL4, lysophosphatidylcholine acyltransferase (LPCAT), and 15-lipoxygenase (15-LOX/ALOX15), to generate PUFA-containing phospholipids and various bioactive lipids that modulate sensitivity to ferroptosis. In this regard, genetic or pharmacological inhibition of ACSL4 suppresses ferroptosis in glioblastoma (86). On the other hand, overexpression of ACSL4 was shown to increase the levels of ferroptosis markers, including 5-hydroxyeicosatetraenoic acid (HETE), 12-HETE, and 15-HETE (86), indicating a key role of ACSL4 in regulating ferroptosis and proliferation of glioma cells (89). Independent experimental evidence showed that suppressing miR-670-3p, which targets ACSL4, also modulates the sensitivity of glioblastoma cells to ferroptosis-mediated cell death (87, 90).

Blocking the esterification of fatty acyl-CoA prevents not only the formation of triglycerides but also modifies cell resistance to ferroptosis. It has been shown that in 3D tumor spheroids and in vivo, DGAT inhibition induces significant cytotoxic effects, especially when combined with dietary long-chain PUFAs (LC-PUFAs), implicating the combination of diet (LC-PUFAs) and DGAT inhibitor (DFATi) administration as a highly relevant therapeutic combination to induce ferroptosis (85). Mechanistically, DGATi administration inhibits lipid droplet formation resulting in the availability of more LC-PUFAs for peroxidation and ferroptotic cell death (85). In contrast, DGAT1 inhibition was recently reported to drive fatty acid-dependent oxidation in mitochondria generating high levels of reactive oxygen species, leading to apoptosis and significant growth inhibitory effects in glioblastoma (91). These contradictory reports are likely due to the difference in cell lineage, or a potential secondary effect associated with DGAT1 inhibition. Therefore, it will be relevant in the future to dissect the relative contributions of DGAT in different cancers, including glioblastoma, and the potential overlapping effects of DGAT inhibition on ferroptosis and apoptosis.

Similar to lipid fluxes into lipid droplets that act as buffers, the release of fatty acids from storage controls the cancer cells’ fate, including ferroptosis. Lipid droplet breakdown occurs via two major mechanisms-lipolysis, and lipophagy ( Figure 1C ), which provide a substrate for lipid peroxidation during ferroptosis. Unlike lipogenesis, the lipolysis process enables a highly regulated release of fatty acids from TAGs and is mediated by three essential lipases: adipocyte triglyceride lipase (ATGL), hormone-sensitive lipase (HSL), and monoacylglycerol lipase (MGL) (92–94). These three enzymes work together to promote the lipolysis of triglycerides and produce fatty acids and glycerol. ATGL catalyzes the first rate-limiting reaction whereby triglycerides are hydrolyzed to DAG. HSL then transforms DAG to monoacylglycerol (MAG), and MGL finally converts MAG to fatty acids and glycerol (93–96). However, the molecular regulation of the lipolysis of lipid droplet-containing triglycerides is complex. It involves a combination of subcellular localization, posttranslational modification (particularly phosphorylation), and protein-protein interactions (92, 97). For instance, hypoxia-induced lipid droplet-associated protein (HILPDA), also called hypoxia-induced gene-2 (HIG2), inhibits ATGL activity (98). Experimental evidence showed the correlation of the HIG2 expression with the glioma tumor grade and glioblastoma patient prognosis (34, 99).

Catabolism of lipid droplets also occur via lipophagy, a form of selective (macro)autophagy. Lipophagy is a process that tags specific lipid droplets and traffic whole lipid droplets to lysosomes for bulk degradation ( Figure 1C ). Lysosomal localized hydrolytic enzymes, such as triglycerides and cholesteryl ester hydrolase lysosomal acid lipase (LAL) (100), liberate neutral lipids stored in lipid droplet and generate free fatty acids and cholesterol in the cell. Although mechanistically lipolysis (the step-wise release of stored neutral lipids from lipid droplets) and lipophagy (a complete breakdown of lipid droplets) are distinct metabolic processes, it is not yet clear whether there is considerable crosstalk between the two processes and this may serve a distinct purpose in the cell (100, 101). It has been shown that perilipins (PLINs), lipid droplet surface proteins, act as gatekeepers to the lipophagy process and their degradation is a prerequisite for lipophagy. During starvation, perilipin 2 (PLIN2) and perilipin 3 (PLIN3) are proteolytically removed in parallel with the translocation of cytosolic ATGL and macroautophagy proteins onto lipid droplets (102). There is also evidence that PLIN2-mediated lipophagy regulation emerges as a key nodal point in modulating cellular sensitivity to ferroptosis. PLIN2 affects the proliferation of gastric carcinoma cells by modulating ferroptosis-related genes, including acyl-CoA synthetase long-chain family member 3 (ACL3), arachidonate 15-lipoxygenase (ALOXs), microtubule-associated protein 1 light chain 3 alpha (LC3), and the transcription factors pr/set domain 11 and importin 7 (IPO7) (103). Likewise, PLIN2 expression contributed to a decreased arachidonate 15-lipoxygenase (ALOX15) expression and arrested the occurrence of ferroptosis in gastric cancer. Conversely, PLIN2 knockdown facilitated a higher ALOX15 expression and accelerated ferroptosis (103). In contrast, increased lipid storage by tumor protein D52 (TPD52), a PLIN2 interactor protein, diminished lipid peroxidation to trigger ferroptosis (104). Recently, PLIN2 was also shown to play a key role in glioblastoma pathology, although its role in ferroptosis has not been directly investigated (105).

It has also been shown that the knockdown of RAB7A (member RAS oncogene family 7A) and autophagy-associated gene 5 (ATG5), a cargo receptor of lipid droplets, can prevent induced lipid peroxidation and subsequent ferroptosis (104, 106). Mechanistically, RAB7A is an indispensable factor for docking lysosomes to the lipid droplet surface during lipophagy under nutrient deprivation (107). Moreover, RAB7A has been demonstrated to act as a tumor suppressor in glioblastoma and prostate cancer (108), unlike other cancers.

Although emerging evidence reinforces the idea that the amount of lipid and the localization of lipids in lipid droplets potentially affect the ability to induce ferroptosis in cancer cells, studies on the effect of lipolysis and lipophagy on sensitizing glioblastoma cells for ferroptosis are limited. There is limited information on how microenvironment and disease progression affect the biogenesis and breakdown of lipid droplets (101). There is a need for a better understanding of specific trafficking itinerary of lipophagy which may lead to new anticancer approaches, particularly against “lipid addicted tumors” like glioblastoma. Finally, the molecular regulators of 1-O-acylceramide hydrolysis and whether this process impacts tumorigenesis are unknown.