Mevalonate metabolism governs cancer immune surveillance

ABSTRACT

The metabolic reprogramming that drives immunity engages the mevalonate pathway for cholesterol biosynthesis and protein prenylation. The importance of tight regulation of this metabolic route is reflected by the fact that too low activity impairs cellular function and survival, whereas hyperactivity can lead to malignant transformation. Here, we first address how mevalonate metabolism drives immunity and then highlight ways of the immune system to respond to both, limited and uncontrolled flux through the mevalonate pathway. Immune responses elicited by mevalonate pathway dysregulation may be harnessed to increase the clinical efficacy of current cancer therapy regimens.

Abbreviations

ABCA1

ATP-binding cassette transporter A1

ACAT-1

acetyl-CoA acetyltransferase 1

ATP

adenosine triphosphate

ACLY

ATP citrate lyase

BTN

butyrophilin

CRM

caloric restriction mimetic

DMAPP

dimethylallyl diphosphate

FA

fatty acid

FPP

farnesyl diphosphate

GM-CSF

granulocyte/macrophage-colony-stimulating factor

GGPP

geranylgeranyl diphosphate

GPP

geranyl diphosphate

GTP

guanosine triphosphate

HC

hydroxycitrate

IFN

interferon

Ig

immunoglobulin

IL

interleukin

IPP

isopentenyl diphosphate

mTOR

mechanistic target of rapamycin

N-BP

nitrogen-containing bisphosphonate

OXPHOS

oxidative phosphorylation

PI3K

phosphoinositide 3-kinase

ROCK

Rho-associated protein kinase

SCAP

SREBP cleavage-activating protein

SREBP

sterol regulatory element-binding protein

TCA

tricarboxylic acid

TCR

T cell receptor

Th

T helper

TLR

toll-like receptor

Although important observations made by Otto Warburg in cancer cells date back several decades, only now the general requirement of metabolic reprogramming for the shift of a cell from quiescence to an activated state, has become obvious. Also in immune cells, dramatic metabolic changes are required to effectively perform the diverse cellular functions that constitute a productive immune response against invading pathogens and tumor cells. These changes also include the enhanced engagement of the mevalonate pathway, best known for the biosynthesis of cholesterol. It has been the seminal work of Michael Brown and Joseph Goldstein, awarded by the Nobel Prize in Physiology or Medicine in 1985,¹ which represented the foundation for further intense investigation of mevalonate metabolism. Although immunometabolism has become a distinct category of immunological research in the past few years, mevalonate metabolism still receives little attention.^2,3 The purpose of this review is to provide a general overview of mevalonate metabolism in immune cells and to highlight ways of the immune system to respond to mevalonate pathway dysregulation.

Mevalonate metabolism

T cells

The rapid increase in glycolysis is a fundamental feature of T cell activation.^4-6 The increase in aerobic glycolysis ensures the availability of biosynthetic building blocks required for cell growth, proliferation, differentiation and effector function. In metabolic reprogramming, the tricarboxylic acid (TCA) cycle, is a center of metabolic activity. To initiate lipogenic pathways, some of the mitochondrial citrate is exported to the cytosol and cleaved by ATP citrate lyase (ACLY) to generate cytosolic acetyl-CoA, which then serves not only as a key metabolite for fatty acid (FA) biosynthesis but also directly fuels the growth-promoting mevalonate pathway (Fig. 1).⁷ In a recent study, a nuclear phosphoproteomic screen revealed that ACLY is a key phosphoprotein effector of interleukin (IL)-2-mediated T-cell responses.⁸ ACLY becomes phosphorylated on serine 455 in T lymphocytes upon IL-2-driven activation.⁸ Conversely, depletion or inactivation of ACLY compromised IL-2-promoted T cell growth. ACLY has previously been shown to be a substrate of AKT,⁹ which has several downstream effects including activation of the key metabolic sensor mTOR (mechanistic target of rapamycin). mTOR, which critically regulates lymphocyte metabolism, consists of 2 complexes, mTORC1 and mTORC2.¹⁰ mTORC1 increases protein translation via phosphorylation of 4E-BP1 and p70S6 kinase¹¹ and controls the expression of numerous genes involved in mevalonate metabolism mainly through the sterol regulatory element-binding protein (SREBP) transcription factors (Fig. 2).^{12, 13} The entire pathway of SREBP2 target genes has been shown to be induced after T cell receptor triggering,¹⁴ emphasizing the importance of mevalonate metabolism for T cell activation.

Figure 1.

The SREBP-driven mevalonate pathway. The transcriptional activity of sterol regulatory element-binding proteins (SREBPs) governs expression of mevalonate pathway enzymes. ATP citrate lyase (ACLY, EC 2.3.3.8), which can be inhibited by hydroxycitrate (HC), generates cytosolic acetyl-coenzyme A (acetyl-CoA). 3-hydroxy-3-methylglutaryl CoA (HMG-CoA)-synthase (HMGCS1, EC 2.3.3.10) condenses 3 molcules of acetyl-CoA to form HMG-CoA. HMG-CoA reductase (HMGCR, EC 1.1.1.34), which can be inhibited by statins, in turn generates mevalonate (also known as mevalonic acid) in the first committed step of the pathway. Sequential phosphorylation reactions by mevalonate kinase (MVK, EC 2.7.1.36), which is mutated in MVK deficiency, and phosphomevalonate kinase (PMVK, EC 2.7.4.2), respectively, followed by a decarboxylation step catalyzed by mevalonate diphosphate decarboxylase (MVD, EC 4.1.1.33) produce isopentenyl diphosphate (IPP). Farnesyl diphosphate (FPP) synthase (FDPS, EC 2.5.1.10), which is the target of nitrogen-containing bisphosphonates (N-BPs) condenses IPP (C₅) and its isomer dimethylallyl diphosphate (DMAPP, C₅) to form geranyl diphosphate (GPP, C₁₀) and subsequently GPP with another IPP unit to generate FPP (C₁₅). FPP is the common precursor for cholesterol and steroids in the sterol branch as well as for products of the nonsterol branch including geranylgeranyl diphosphate (GPPP, C₂₀), dolichol, heme A and ubiquinone (coenzyme Q). GGPP is formed by GGPP synthase (GGPS, EC 2.5.1.29) through condensation of FPP with yet another IPP unit. In protein prenylation, FPP and GGPP, serve as prenyl group donors in posttranslational modifications of multiple Ras protein family members and G protein-coupled receptors. Farnesylation (using FPP) or geranylgeranylation (using GGPP) are required for membrane attachment and function of these proteins. Prenylated proteins constitute approximately 0.5% to 2% of proteins in mammalian cells. Prenyldiphosphate synthase-1 (PDSS1, EC 2.5.1.91) catalyzes the elongation of GPP or FPP with several IPP moieties to form the polyisoprenoid chain of ubiquinone (coenzyme Q). ACAT1 catalyzes cholesterol esterification for storage purposes.

Figure 2.

The metabolic reprogramming associated with immune cell activation engages the mevalonate pathway. Immune cell stimulation activates the PI3K-AKT-mTOR pathway and promotes uptake of glucose via glucose transporter (GLUT) proteins followed by glycolysis and oxidation of glucose-derived pyruvate in the mitochondrial tricarboxylic acid cycle (TCA) cycle, which drives oxidative phosphorylation (OXPHOS). Activated immune cells also export citrate to the cytosol, where it is converted back to acetyl-coenzyme A (acetyl-CoA) by ATP citrate lyase (ACLY). Cytosolic acetyl-CoA serves as a metabolic precursor not only for fatty acid (FA) synthesis and protein acetylation but also for mevalonate metabolism (the lipogenic pathways). mTOR signaling also leads to the activation of sterol regulatory element-binding proteins (SREBP), which are the main transcription factor for mevalonate pathway-associated genes in immune cells. In a feed-forward loop, enhanced mevalonate metabolism facilitates prenylation of Ras proteins and thus further enhances pathway activity. In an apparently futile cycle of concurrent synthesis and oxidation, de novo synthesized fatty acids (FA) are stored in the lysosome. Lysosomal acid lipase (LAL) catalyzes the release of these FA, which are shuttled into the mitochondria for ATP generation through ß-oxidation.

Two major branches of mevalonate metabolism (Fig. 1) have emerged as important regulators of T lymphocyte biology. The sterol branch for cholesterol biosynthesis critically regulates T cell cycle progression and effector function. Activated CD8 T cells therefore rapidly reprogram their metabolism through the actions of the SREBP and liver X receptor (LXR) transcription factors to ensure cholesterol availability by promoting cholesterol biosynthesis, while concomitantly decreasing cholesterol efflux.¹⁴ Activated CD8⁺ T cells can further increase plasma membrane levels of free cholesterol by preventing cholesterol esterification for storage.^{15, 16} Specific inhibition of the cholesterol esterification enzyme ACAT1 (Fig. 2) improved immunological synapse formation and TCR signaling, resulting in enhanced production of cytokines, degranulation and proliferation of CD8⁺ T cells. ACAT-1 might also be an attractive therapeutic target in tumor therapy (Fig. 3), since ACAT1 inhibition has already been shown to improve the function of antitumor CD8⁺ T cells reactivated by immune checkpoint blockade to treat melanoma in mice.^{16, 17}

Figure 3.

Inflammatory and immune responses to mevalonate pathway dysregulation. The responses induced by restricted flux have been studied using pharmacological inhibitors (statins and nitrogen-containing bisphosphonates, N-BPs), caloric restriction mimetics (CRM) such as hydroxycitrate (see also Fig. 1) or by genetic inactivation of geranylgeranyltransferase (GGTase), mevalonate kinase (MVK) or SREBP cleavage-activating protein (SCAP). Enhanced or uncontrolled flux can result from gain-of-function p53 mutation, sustained NFkB activation associated with chronic inflammation, ectopic expression of HMG-CoA reductase, or possibly also by futile metabolic constellations. Cell longevity resulting from sustained mevalonate metabolism and protein prenylation may physiologically be important for T cell memory establishment or pathologically manifest as malignant transformation. Among the various tools currently available for mevalonate pathway manipulation, N-BPs are unique, since they increase levels of IPP and simultaneously inhibit protein prenylation. Vγ9Vδ2 T cells, which are activated by increased levels of IPP and other mevalonate pathway intermediates, are intended to perform broad immune surveillance of enhanced mevalonate metabolism.

The nonsterol branch for protein prenylation (Fig. 1) also determines multiple aspects of T cell function, including synapse formation, migration, proliferation and cytotoxic effector responses.⁶ The prototype of small guanosine triphosphatases (GTPases) Ras is activated through prenylation in response to TCR stimulation and various cytokines. In protein prenylation, which represents one out of multiple forms of post-translational modifications, FPP (C₁₅) and GGPP (C₂₀), respectively, represent the activated forms of the farnesyl and geranylgeranyl units that are covalently attached to the cysteine residue of a distinct tetrapeptide motif (CaaX) of many members of the Ras protein superfamily.¹⁸ The prenyl side chain mediates membrane association, which is essential for Ras protein biologic activity. In addition, γ proteins of heterotrimeric G proteins (G_αβγ), which are activated by G protein-coupled receptors, are also subject to farnesylation (γ1) or geranylgeranylation (γ2).¹⁹ Ras activates not only the MAPK signaling cascade but also the phosphoinositide 3-kinase (PI3K)-AKT-mTOR pathway (Fig. 2).⁶ Signaling through this pathway is essential not only for glycolytic metabolism²⁰ but also for the lipogenic program.²¹ mTOR promotes glycolysis, which is prerequisite for the accumulation of cytosolic citrate and AKT stimulates the conversion of citrate into acetyl-CoA by phosphorylating ACLY.⁹ Abundant cytosolic acetyl-CoA then fuels mTOR/SREBP-driven mevalonate metabolism and the resulting accumulation of FPP (or GGPP) facilitates prenylation of Ras, thus also generating a feed forward loop (Fig. 2).

In experimental autoimmune encephalomyelitis, a murine model of multiple sclerosis, GGPP has been shown to be crucial for proliferation, whereas both GGPP and FPP regulated type 1 T helper (Th1) cell differentiation of myelin-reactive T cells.²² Specifically, geranylgeranylated RhoA and farnesylated Ras have been implicated in proliferative and cytokine responses of these autoreactive T cells. Likewise, inhibition of farnesylation has been shown to impair cytokine production in murine Th1 and Th2 T cell clones.²³

A distinct form of prenylation is also required to maintain ATP generation through oxidative phosphorylation (OXPHOS). Coenzyme Q (CoQ or ubiquinone) serves as a diffusible, lipid-soluble electron shuttle between large, relatively immobile macromolecular complexes in the electron transport chain at the inner mitochondrial membrane, which is the site of OXPHOS in eukaryotes. In human CoQ₁₀ (Fig. 1), the lipid membrane anchor is a decaprenyl side chain (C₅₀) consisting of 10 C₅ isoprenyl units.⁶ Human prenyl (decaprenyl) diphosphate synthase, subunit 1 (PDSS1) (Fig. 1) catalyzes the elongation of GPP or FPP with several IPP moieties to form the polyisoprenoid chain, which is then attached to a 4-hydroxybenzoate ring. Therefore, efficient CoQ₁₀ biosynthesis depends on the functionality of PDSS1 on the one hand, and on the other, on the availability of its mevalonate-derived substrates IPP, GPP and FPP. Long-term therapy with statins, which inhibit the mevalonate pathway early on, may cause CoQ₁₀ deficiency via PDSS1 substrate depletion with dramatic cardiac consequences, preferentially in elderly patients.²⁴ Conversely, PDSS1 inactivation due to mutations in the PDSS1 gene has been reported to cause clinically apparent OXPHOS disorders.²⁵ Collectively, these observations point toward the importance of mevalonate metabolism for OXPHOS.

Macrophages and dendritic cells

Macrophages can adopt distinct phenotypes, which either participate in immune responses against microbial pathogens and tumors (M1) or contribute to immunity against parasites as well as to tissue repair (M2).²⁶ Classical activation with interferon (IFN)-γ in combination with toll-like receptor (TLR) agonists such as lipopolysaccharide (LPS) generates M1 macrophages, whereas full M2 activation, also known as alternative activation, occurs in response to IL-4 and macrophage colony stimulating factor (M-CSF).²⁷ Classical activation of M1 macrophages involves the upregulation of GM-CSF production.²⁸ Importantly, GM-CSF but not M-CSF primes monocytes for enhanced responses to LPS.²⁹ Recent work has revealed the metabolic imprinting that underlies the GM-CSF priming effect. GM-CSF promotes glycolysis and concomitantly mevalonate metabolism by increasing 3-hydroxy-3-methylglutaryl CoA (HMG-CoA) reductase in macrophages.³⁰ GM-CSF thus creates the metabolic requirements for the subsequent proinflammatory response induced by the TLR4 agonist LPS. Inhibition of both, glycolysis or HMG-CoA reductase, prevented the inflammatory effects of GM-CSF priming in macrophages. LPS has also been shown to boost glycolysis³¹ and to stimulate the lipogenic pathway, which may in part be due to the induction of endogenous GM-CSF. In macrophages, lipogenesis is regulated by sterol regulatory element–binding protein 1a, which also directly activates inflammasome function resulting in caspase-1 mediated release of bioactive IL-1ß and IL-18.^{32, 33} This finding established a strong link between the lipogenic program and inflammation. During macrophage activation, AKT can directly regulate mevalonate pathway activity by phosphorylating the isopentenyl diphosphate (IPP)-generating enzyme MVD (Fig. 1).³⁴ The resulting increase in mevalonate metabolism facilitated Rac1 activation, which requires prenylation, and thus promoted macrophage survival.³⁴ Specific evidence for the importance of mevalonate metabolism in inflammation was obtained in numerous studies showing that pharmacological mevalonate pathway inhibition can modulate a broad range of pro-inflammatory mechanisms by suppressing protein prenylation.³⁵ Collectively, these findings clearly indicate that classical activation of M1 macrophages is associated with enhanced mevalonate metabolism (Fig. 2).

Alternative (M2) activation is driven by IL-4,²⁶ an anti-inflammatory cytokine that suppresses macrophage development and M1 activation.^{36, 37} M-CSF synergizes with IL-4 to support full M2 activation in an mTORC2-mediated pathway.²⁷ In contrast to GM-CSF, M-CSF is unable to prime macrophages for enhanced responses to LPS through metabolic reprogramming.^{29, 30} Surprisingly, M2 activation has recently been found to also depend on glycolysis, however it appears to be a feature of M2 macrophages that glucose metabolism in M2 cells specifically increases FA oxidation to support OXPHOS, and these processes appear to be critical for full M2 activation.²⁷ An interesting and, at first glance, futile metabolic aspect of full M2 activation is the concurrent operation of FA synthesis and FA oxidation. However, increasing evidence supports the view that FA synthesis is required to effectively fuel FA oxidation.^{38, 39} Together, these findings indicated that glycolysis primarily fuels the lipogenic program including mevalonate metabolism in M1 macrophages (Fig. 2). In contrast, glucose metabolism increasingly fuels FA synthesis for enhanced FA oxidation to drive OXPHOS in M2 macrophage activation. Such a metabolic constellation may limit mevalonate pathway activity, which likewise depends on acetyl-CoA, and thus prevent pro-inflammatory responses.

Dendritic cells (DCs) are professional antigen-presenting cells (APCs).⁴⁰ In steady-state, DCs are immature with poor T cell stimulatory capacity and instead sample the microenvironment for the presence of pathogens. In response to pro-inflammatory signals, DCs undergo an activation program that has also been referred to as maturation, which is characterized by the downregulation of antigen uptake capacity and upregulation of the machinery related to T cell activation. DC development from monocytes under inflammatory situations depends on GM-CSF, and LPS can activate these DCs suggesting that the metabolic regulation of DC maturation might be similar to M1 macrophage activation and therefore also requires high flux through the mevalonate pathway (Fig. 2). DCs differentiated from bone marrow in the presence of GM-CSF have been used as a model for inflammatory monocyte-derived DCs.⁴¹ LPS has been shown to induce early glycolysis in such DCs within minutes, which feeds the TCA cycle, however, not for OXPHOS and ATP generation but for enhanced citrate production and export into the cytosol, where ACLY converts citrate back into acetyl-CoA for lipogenesis.⁴¹ LPS-induced de novo FA synthesis in DCs has been shown to be important for the generation of additional membranes to expand the endoplasmic reticulum (ER) and the Golgi. Since cholesterol is a vital component of these membranes, LPS-stimulated membrane biogenesis is likely to depend on mevalonate metabolism. De novo lipogenesis may also support the generation of veil-like membrane protrusions, which constitute the typical dendritic morphology of mature DCs. Whereas numerous veils provide abundant contact surface for lymphocyte activation, expanded ER and Golgi ensures the secretory capacity of mature T cell stimulatory DCs. The membrane dynamics that participate in the formation and movement of veils depend on Rho GTPase and its effector Rho-associated protein kinase (ROCK), which are important regulators of the actin cytoskeleton.⁴² Importantly, Rho must be geranylgeranylated to be able to interact with ROCK and to induce its activation. The vesicular transport that mediates secretion of inflammatory cytokines depends on Rho and Rab proteins,⁴³ which likewise require prenylation for their activation.⁶ Taken together, DC maturation depends on a glycolysis-fueled lipogenic program engaging both, FA synthesis and mevalonate metabolism (Fig. 2).

Monocyte-derived DCs (MoDCs) that develop in the presence of GM-CSF and IL-4⁴⁴ have been shown to produce substantial amounts of M-CSF,⁴⁵ suggesting that these cells might display a mixed metabolic profile combining aspects of M1 and M2 macrophages (Fig. 2).²⁷ This particular metabolic constellation may be required to enable DC differentiation through IL-4 mediated inhibition of macrophage activation³⁷ on the one hand, and on the other, to concomitantly maintain the GM-CSF mediated responsiveness to pro-inflammatory stimuli such as LPS³⁰ that induce the further differentiation of these cells into fully mature immunogenic DCs.⁴⁶

Collectively, these observations suggest the view that resting conditions (homeostasis) only support core mevalonate metabolism because acetyl-CoA is preferentially used for FA synthesis to fuel FA oxidation for OXPHOS and ATP generation. In contrast, inflammation boosts aerobic glycolysis that translates into enhanced mevalonate metabolism for cholesterol synthesis and protein prenylation in DCs and macrophages. From a metabolic point of view, classical (M1) macrophage activation and DC maturation may therefore be considered pro-inflammatory effector responses comparable to that of CD8⁺ T cells, since they critically depend on mTOR-driven glycolytic capacity and high flux through the mevalonate pathway (Fig. 2).

The futile cycle and mevalonate metabolism

A distinct metabolic constellation revolving around acetyl-CoA has recently been described for various immune cells. Memory T cells appear to operate an apparently futile cycle of concurrent FA synthesis and FA oxidation (Fig. 2) to maintain long-term survival.³⁸ De novo synthesized FAs are stored in the lysosome. Lysosomal acid lipase catalyzes the release of these FAs, which are shuttled into the mitochondria for ATP generation through ß-oxidation. Recent findings by Guijas et al. suggest that foam cells,³⁹ which are considered a hallmark of the atherosclerotic lesion, engage a similar or possibly the same futile cycle to ensure their own survival, thus driving the atherogenic process. As outlined above, M2 macrophages were also found to combine FA synthesis and FA oxidation.²⁷ FA synthesis is obviously required to effectively fuel FA oxidation to drive OXPHOS. This observation made in T cells and macrophages raised the question of whether the longevity of other cells such as transformed cells including cancer-initiating cells is a consequence of this particular futile metabolism (Fig. 3).^{47, 48} Importantly, the acetyl-CoA resulting from FA breakdown becomes available not only for FA biosynthesis - to drive the futile cycle - but also for the survival-promoting mevalonate pathway.⁶ Sustained mevalonate metabolism, although at a relatively low level, may facilitate prolonged protein prenylation and may thus promote cell longevity, suggesting that FA metabolism might also be an attractive target to manipulate mevalonate metabolism in cancer therapies.

Immune responses to limited flux

Given the importance of mevalonate pathway products in multiple aspects of cell biology, it is reasonable to assume that a lack of mevalonate metabolism is hardly compatible with cell survival and tissue homeostasis.⁴⁹ Conditions of limited flux can be simulated by inhibiting critical enzymatic steps of the biosynthetic route using drugs or by inactivation or attenuation of the corresponding genes. In addition, an autosomal recessive metabolic disorder, referred to as hyper immunoglobulin D syndrome turned out to be useful for studying the consequences of insufficient mevalonate pathway activity. In these patients, mevalonate kinase (MVK) activity is reduced to 5 to 15 % of normal due to a loss-of-function mutation in the MVK gene (Fig. 1).⁵⁰ A recurrent observation in these studies was that limited flux through the mevalonate pathway did not simply result in cell death by apoptosis, which would be immunologically silent, but instead could also trigger inflammatory and immune responses.

Immunogenic cell death (ICD)-inducing cancer therapy based on anthracyclines or oxaliplatin has been demonstrated to elicit antitumor immune responses that substantially contribute to the long-term success of these chemotherapy regimens.⁵¹ Autophagy is indispensable for ICD, because autophagy ensures release of ATP into the extracellular space and recruits APCs into the close vicinity of dying tumor cells. Hydroxycitrate (HC) is a drug candidate (a caloric restriction mimetic, CRM), which mimics the biochemical effects of starvation.⁵² By competitively inhibiting ACLY (Figs. 1 and 2), HC reduces the availability of cytoplasmic acetyl-CoA and thus limits mevalonate metabolism as well as protein acetylation, both conditions that increase autophagic flux.^{53, 54} In this manner, HC has recently been shown to improve ICD-inducing chemotherapy in a T cell-dependent fashion and this effect involved the depletion of tumor-infiltrating regulatory T cells. Importantly, the stimulatory effect of HC could only be observed in the treatment of autophagy-competent tumors.⁵²

Statins, which are a widely prescribed class of drugs for the treatment of cardiovascular disease, inhibit the pathway early on (Figs. 1 and 3) and thus cause general depletion of mevalonate and its downstream metabolites.⁵³ In contrast, nitrogen-containing bisphosphonates (N-BPs), which are used to prevent bone resorption in osteoporosis and malignant bone disease, inhibit mevalonate metabolism at the level of FPP synthase and generate a twofold effect: upstream accumulation of IPP as well as downstream depletion of FPP and GGPP (Figs. 1 and 3).⁶ The inhibitory effect of N-BPs on protein prenylation in osteoclasts is the major mechanism of N-BP mediated prevention of bone degradation. However, the metabolic perturbation resulting from statin or N-BP treatment can also lead to inflammasome and caspase-1 activation in DCs.^{55, 56} Caspase-1 proteolytically processes the inactive proforms of IL-1ß and IL-18, which can then be secreted as mature cytokines with strong costimulatory activity for innate lymphocytes (Fig. 3).⁵⁷ In the presence of IL-2, both statins and N-BPs could induce innate lymphocyte activation via caspase-1 mediated cytokine maturation (Fig. 3).^{55, 56} Zoledronic acid, the most potent N-BP currently available for clinical use, has also been shown to restore doxorubicin chemosensitivity and immunogenic cell death in otherwise multidrug-resistant human cancer cells.⁵⁸

Macrophages that fail to perform protein geranylgeranylation due to a deficiency of geranylgeranyl transferase-I (GGTase-I) have been shown to be hyper-activated by lipopolysaccharide and mice with conditional deficiency in GGTase-I in myeloid cells develop inflammatory arthritis spontaneously.⁵⁹ Mechanistic studies performed with GGTase-I deficient mice revealed that limited flux through the mevalonate pathway decreased the production of GGPP and attenuated protein geranylgeranylation. Since protein geranylgeranylation positively regulates the interaction of the Kras GTPase with the PI3K catalytic subunit p110δ (Figs. 2 and 3), impaired geranylgeranylation prevented PI3K activation. Surprisingly, disturbed interaction of Kras with p110δ resulted in hyper-inflammatory conditions, including constitutive IL-1ß release from macrophages, elevated expression of pyrin protein and spontaneous activation of the pyrin inflammasome (Fig. 3).⁵⁹

In another recent study, the immunological consequences of limited mevalonate pathway activity have been investigated in mice with macrophage-specific deletion of SCAP. Upon genetic deletion of SCAP, SREBP2 transcriptional activities are significantly attenuated, resulting in markedly reduced expression of mevalonate pathway genes (Figs. 1-3).⁶⁰ The authors reported that a reduced cholesterol pool size resulting from limited flux through the mevalonate pathway could spontaneously induce a type I IFN response in a stimulator of interferon genes (STING, TMEM173)-dependent fashion.⁶⁰ Cholesterol supplementation was able to normalize type I IFN levels by preventing STING signaling in SREBP2-deficient macrophages. Intriguingly, this lipid metabolic-inflammatory circuit facilitated the development of anti-viral immunity (Fig. 3).

Similar to the observations made with statins, lack of MVK activity (Figs. 1 and 3) in hyper immunoglobulin D syndrome (MVK deficiency) also resulted in downstream depletion of GGPP and reduced geranylgeranylation of small GTPases (Rho, Rac, Rap) in patient peripheral blood mononuclear cells followed by inflammasome-dependent caspase-1 activation and release of mature IL-1ß,⁶¹ which is known to be pivotal to the pathogenesis of most of the periodic fever syndromes.⁵⁰

Immune responses to enhanced flux

Enhanced flux through the mevalonate pathway can lead to malignant transformation.⁶² In breast cancer cells, protein geranylgeranylation has been shown to be required for the malignant, invasive phenotype caused by mutant p53, which enhances mevalonate metabolism instead of suppressing it (Fig. 3).⁶³ Moreover, ectopic expression of HMG-CoA reductase, which catalyzes the first committed step of mevalonate metabolism, enhances growth of transformed and nontransformed breast cells under anchorage-independent conditions or as xenografts in immunocompromised mice and, importantly, cooperates with prenylated RAS (Fig. 2) to drive the transformation of primary mouse embryonic fibroblasts cells.⁶⁴ The epidemiologic observation that statins can lower the risk of certain cancers⁶⁵ can be best explained by statin-mediated attenuation of protein prenylation.

In addition to a general increase in flux through the mevalonate pathway, individual branches of the pathway may be fueled preferentially, for instance as a consequence of redistribution of isoprenoid precursors. Chronic inflammation has been shown to inhibit hepatic cholesterol conversion into bile acids, causing a redistribution of metabolites to other branches of the mevalonate pathway (Fig. 1).⁶⁶ The resulting accumulation of FPP and GGPP facilitated protein prenylation, especially geranylgeranylation of a Rho GTPase.⁶⁶ This finding may also provide an explanation - at the metabolic level – for the well-established link between inflammation and cancer development (Fig. 3).⁶⁷

Surveillance of permanently enhanced mevalonate metabolism by Vγ9Vδ2 T cells

A potentially oncogenic mevalonate pathway obviously requires surveillance. Ectopic expression of HMG-CoA reductase not only drives malignant transformation⁶⁴ but also endows transfected B lymphoblasts (Daudi) with the capacity to activate Vγ9Vδ2 T cells (Fig. 3),⁶⁸ which are the most abundant population of γδ T cells in human blood.⁶⁹ Although Vγ9Vδ2 T cells display remarkable functional plasticity, a main function of this subset is obviously to perform strong cytotoxic responses against infected and transformed cells and to produce effector cytokines with innate-like kinetics.^{69, 70} Vγ9Vδ2 T cells have recently also been identified as a major source of IL-9, a pleiotropic cytokine also contributing to antitumor immunity.⁷¹ The TCR of these unconventional T cells acts as a pattern recognition receptor and responds to the mevalonate-derived diphosphate-containing C₅ to C₂₀ isoprenoids (IPP, DMAPP, GPP, FPP, GGPP) (Fig. 1), which are also referred to as phosphoantigens (pAgs) in γδ T cell biology.^{72, 73} As compared with the frequencies of bacteria- or virus-specific αβ T cells, which range from 5 to 170 per 10⁶ naïve T cells,⁷⁴ Vγ9Vδ2 T cells with their semi-invariant, pAg-reactive TCRs represent 5.000 to 50.000 per 10⁶ T cells in peripheral blood during homeostasis and up to 60% of all T cells during infection⁶⁹ indicating that pAgs constitute strong danger signals and emphasizing the importance of immune surveillance of mevalonate metabolism.

A pharmacological strategy to induce intracellular accumulation of mevalonate-derived pAgs is the use of N-BPs, which not only inhibit protein prenylation, but also induce upstream accumulation of mevalonate metabolites (Figs. 1 and 3). The N-BP induced increase of the FPP synthase substrate IPP in antigen-presenting cells or other cells induces activation of Vγ9Vδ2 T cells.⁶⁸ This side effect of N-BPs, which was detected more incidentally, when patients with bone disease had increased frequencies of Vγ9Vδ2 T cells after their first N-BP administration,⁷⁵ indicated for the first time that accumulation of mevalonate pathway intermediates alerts Vγ9Vδ2 T cells.

Extracellular mevalonate metabolism

Numerous studies have demonstrated that exogenous (i.e. synthetic) pAgs can activate Vγ9Vδ2 T cells,^{71, 72} however, they concomitantly raised the question of whether and how pAgs may be released from cells to accumulate in the extracellular space under physiologic conditions. Castella and colleagues have recently addressed this important issue and provided evidence that the cholesterol efflux transporter ABCA1 (ATP-binding cassette transporter A1) is involved in the export of intracellular pAgs from DCs⁷⁶ (Fig. 4). ABCA1 is well known to mediate the efflux of cholesterol, which is actually made from IPP (Fig. 1).⁶ Castella et al. further demonstrated that apoA-I is required for Vγ9Vδ2 T cell activation through extracellular pAgs in a butyrophilin 3A1 (BTN3A1)-dependent manner (Fig. 4). BTN3A1 has been reported to bind pAgs and mediate activation of Vγ9Vδ2 T cells.^{77, 78} To facilitate this process, BTN3A1 appears to be physically associated with ABCA1.⁷⁶ The work by Castella et al. also represents a valuable contribution to an ongoing debate in the field regarding 2 different mechanistic models of Vγ9Vδ2 T cell activation by pAgs (Fig. 4). In the allosteric model proposed by Harly et al., the interaction between intracellular pAgs and the intracellular B30.2 domain of BTN3A1 induces conformational changes of the extracellular domain, which are sensed by Vγ9Vδ2 T cells leading to their activation.⁷⁸ In the antigen presenting model put forward by Vavassori et al., intracellular pAgs are exported to the extracellular microenvironment by an “unidentified plasma membrane-associated transporter” and presented to Vγ9Vδ2 T cells by the BTN3A1 extracellular domain.⁷⁷ According to the findings of Castella et al., ABCA1 has now been identified as a crucial pAg exporter.⁷⁶ The 2 mechanisms may indeed cooperate to increase the degree of flexibility. Thus, both intracellular and extracellular pAg accumulation can be sensed and import-export mechanisms may ensure Vγ9Vδ2 T cell activation by either the allosteric or the presentation mechanism (Fig. 4).

Figure 4.

Extracellular mevalonate metabolism. Intracellular accumulation of mevalonate derivatives triggers homeostatic mechanisms leading to their export into the extracellular space. In dendritic cells (DC), the ATP-binding cassette transporter A1 (ABCA1) not only mediates efflux of cholesterol, which arises from IPP, but also exports IPP itself, which acts as an agonist of Vγ9Vδ2 T cells (“phosphoantigen,” pAg). In the extracellular environment, pAgs can bind to apoA-I, which facilitates pAg presention by butyrophilin 3A1 (BTN3A1) on the DC surface to Vγ9Vδ2 T cells. The ecto-ATPase CD39 also exhibits intrinsic isoprenoid diphosphate phosphohydrolase activity and may control strength and duration of antigen presentation through the hydrolytic inactivation of pAgs. Finally, scavenger receptors such as CD36 may internalize not only IPP-derived CoQ₁₀ but also IPP itself.

The finding that pAgs are exported into the extracellular space for BTN3A1-mediated presentation to Vγ9Vδ2 T cells⁷⁶ calls for controlling mechanisms, since active export or lytic release of isoprenoid-derived pAgs from dysregulated or lysed cells into the extracellular space may lead to apoA-I mediated binding of pAgs to BTN3A1 on healthy adjacent tissue, thus causing collateral damage. We have recently shown that the ecto-ATPase CD39 also exhibits isoprenoid diphosphate phosphohydrolase activity⁷⁹ and may hence be involved in surveillance of this process. CD39 expressed by Vγ9Vδ2 T cells, T_reg cells, or other cells may inactivate secreted pAgs by dephosphorylation (Fig. 4) to avoid such side effects and ensure the specific elimination of the dysregulated cells.⁷⁹

Transcellular lipid metabolism refers to a particular form of short distance intercellular communication, in which a lipid intermediate synthesized and released by one cell type, can be incorporated and further metabolized by another cell type. Such interaction between different cell types by shared metabolism is a well-described phenomenon during eicosanoid biosynthesis.⁸⁰ In accordance with such a concept of transcellular lipid metabolism, extracellular pAgs were previously shown to have the ability to enter cells and feed into mevalonate metabolism. In add-back experiments, drug-induced mevalonate pathway inhibition was restored by exogenous pAgs. Thus, addition of FPP, or more often of GGPP, was able to reinstate protein prenylation during statin-mediated inhibition of mevalonate metabolism (Fig. 1).^{6, 22, 63} While pAg uptake could undoubtedly be documented, the mechanisms of internalization remained less clear. CD36, originally referred to as fatty acid translocase and a defining member of the class B scavenger receptor family, has now been identified as a likely candidate. CD36 is a multifunctional receptor that also participates in a range of processes unrelated to fatty acid uptake. Recently, CD36 has been shown to also mediate cellular uptake of coenzyme Q₁₀ (CoQ₁₀),⁸¹ thus implicating CD36 in the internalization of extracellular isoprenoids (Fig. 4). In human CoQ₁₀ (Fig. 1), the lipid membrane anchor is a decaprenyl side chain (C₅₀) consisting of 10 C₅ isoprenyl units.⁶ By analogy with IPP efflux,⁷⁶ which can be mediated by a transporter of IPP-derived cholesterol, IPP uptake may be performed by a scavenger receptor known to incorporate IPP-derived CoQ₁₀.⁸¹ Given that CD36 may mediate the internalization of isoprenoids,⁸¹ dephosphorylation of pAgs by CD39,⁷⁹ which also increases pAg lipophilicity, may facilitate uptake and recycling of pAgs.⁸²

Concluding remarks and perspectives

To survive and maintain core functions during homeostasis, immune cells have to ensure basal flux through the mevalonate pathway. In contrast, high flux is transiently required to provide important products such as cholesterol or to enable protein prenylation for the realization of diverse cellular functions ranging from proliferation and migration to cytokine production and cytotoxic degranulation during an immune response. Recent work has convincingly demonstrated that both, restricted flux (below basal flux) as well as permanently increased flux through the mevalonate pathway triggers alarms and leads to distinct inflammatory and immune responses. Intriguingly, evidence has also been obtained that such immune responses resulting from metabolic dysregulation may be harnessed to improve cancer immunotherapy. Manipulation of mevalonate pathway activity may be very effective in attempts to awake latent immunosurveillance for cancer therapy. Checkpoint inhibition blockade,⁸³ ICD-inducing chemotherapy,⁵¹ or any type of tumor vaccine, for instance based on DCs,⁸⁴ could be combined with mevalonate pathway manipulation to generate synergistic effects. Promising data have already been obtained in murine cancer models showing that ACLY inhibition can enhance immunogenic chemotherapy⁵² and ACAT-1 inhibition was effective in generating better T cells in the context of checkpoint inhibitor-based immunotherapy.^{16, 17} Finally, N-BPs may play a special role, since they increase tumor cell immunogenicity and concomitantly recruit Vγ9Vδ2 T cells, which are poised to perform strong cytotoxic antitumor responses.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by the Austrian Science Fund (FWF; P 28923-B28).