Lysine succinylation and SIRT5 couple nutritional status to glutamine catabolism

ABSTRACT

The metabolic microenvironment of tumors is characterized by fluctuating and limited nutrient availability. To survive these conditions, cancer cell-intrinsic mechanisms sense and signal nutritional status. We describe how glutaminase (GLS) is destabilized by lysine succinylation and stabilized by the NAD⁺-dependent desuccinylase sirtuin 5 (SIRT5), coupling nutrient levels to metabolic flux.

Proliferative signals in mammalian cells drive biosynthetic programs that support cell growth and replication. In healthy cells this process is tightly regulated by growth factors, but in cancer cells oncogenic lesions result in continuous signaling to the metabolic machinery. Oncogene-driven metabolic reprogramming is supportive of tumorigenesis but renders cancer cells hypersensitive to metabolic stress, a phenomenon that is exploited for cancer therapy. Compounding this metabolic sensitivity, the tumor vasculature is typically disorganized with a diminished ability to supply nutrients and remove waste products, and its constituent cell types compete for limiting metabolites. To survive in this harsh environment, intrinsic mechanisms are required for cancer cells to sense their nutritional status and coordinate the response to fluctuating supplies.

Glutamine is the most abundant amino acid in blood serum (~500 μM) and is often avidly consumed by tumors, such that radioactive glutamine analogues are being evaluated for clinical imaging. In mitochondria, glutamine is hydrolyzed into glutamate and ammonium by glutaminase (GLS), and subsequent deamination of glutamate generates the tricarboxylic acid (TCA) cycle intermediate α-ketoglutarate (α-KG). The use of glutamine as a carbon source for TCA cycle anaplerosis – and as a nitrogen source for biosynthesis – causes diverse cancers to become addicted to GLS activity. Consequently, efforts have been made to target GLS for cancer therapy, and the GLS-selective inhibitor CB-839 is currently in phase I/II clinical trials.

Expression of GLS is upregulated at both the transcriptional and translational level by numerous oncogenic signals.^1,2 In a recent study, we uncovered a post-translational mechanism involving acylation of a specific lysine residue, which modulates the stability of GLS protein.³ Lysine acylation is a class of post-translational modification (PTM) regulated by the sirtuins (SIRT1-7), a family of nicotinamide adenine dinucleotide (NAD⁺)-dependent deacylases. Lysine acylations are generated by reactive acyl-coenzyme A (CoA) metabolites, including the TCA cycle intermediate succinyl-CoA.⁴ Thus, there is a sirtuin-mediated connection between nutrient levels, redox status, and the protein PTM landscape.

The lysine acylation/deacylation cycle changes the characteristics of substrate proteins and thereby influences diverse processes including gene expression, DNA repair, and metabolism. Altered expression of sirtuins and corresponding aberrant lysine acylation has been observed during aging and in a variety of pathological states including metabolic diseases, neurodegeneration, and cancer.⁵ Three sirtuins, SIRT3-5, localize primarily to mitochondria (although specific isoforms are also present in the cytosol), and modulate potentially hundreds of substrate proteins to support metabolic homeostasis.⁵

Previous proteomics screens for PTMs revealed that lysine (K) residues K164 and K311 on GLS are subjected to succinylation.^6,7 SIRT5 has robust desuccinylase activity, whereas SIRT3 is the major mitochondrial deacetylase and SIRT4 catalyzes removal of acyl moieties derived from leucine oxidation.⁸ Consistent with these data, we observed elevated levels of succinyllysine on GLS following knockdown of SIRT5 in cancer cells with high endogenous expression of GLS.³ Notably, increased lysine succinylation was accompanied by ubiquitination of GLS and its subsequent degradation. Thus, depletion of SIRT5 is sufficient to eliminate much of the GLS protein from breast cancer and lung cancer cells, resulting in sharply suppressed cellular glutamine consumption. Using a mutagenesis approach, we identified residue K164 as the key succinylation site on GLS, and we further found that succinylation of this residue primes GLS for ubiquitination at residue K158. This mechanism suggests a homeostatic feedback loop, whereby accumulation of succinyl-CoA – which is immediately downstream of μ-KG in the TCA cycle – signals that the cycle is replete by succinylating GLS and marking it for degradation. Conversely, depletion of succinyl-CoA, along with the NAD⁺-dependent activation of SIRT5, results in stabilization of GLS and thus promotes glutamine-mediated anaplerosis (Figure 1).

Figure 1.

Lysine succinylation and sirtuin 5 (SIRT5) connect nutrient availability to metabolic flux. Glutaminase (GLS) initiates the conversion of glutamine to α-ketoglutarate (α-KG) for tricarboxylic acid (TCA) cycle anaplerosis. Reaction of the TCA cycle intermediate succinyl-CoA with lysine K164 on GLS tags it for ubiquitination (at residue K158) and subsequent degradation. SIRT5 stabilizes GLS by catalyzing the desuccinylation of K164, a reaction which requires nicotinamide adenine dinucleotide (NAD⁺) as a co-substrate.

Based on our findings, and the observation that elevated SIRT5 expression corresponds with poor prognosis in breast cancer patients, we tested the impact of SIRT5 knockdown on breast cancer growth. This revealed that SIRT5 supports cell proliferation and anchorage-independent growth ex vivo, and breast cancer tumorigenesis in vivo. Expression of SIRT5 is widely upregulated in tumors relative to surrounding tissue, and SIRT5 supports tumorigenesis and/or drug resistance in diverse cancer types.^3,9 In contrast SIRT3 and SIRT4 are downregulated across a spectrum of human cancers, and Sirt3- or Sirt4-knockout mice spontaneously develop mammary or lung tumors respectively, indicating that these sirtuins possess context-specific tumor suppressor activity.¹⁰ To understand the opposing roles of the mitochondrial sirtuins in cancer, it is informative to examine the substrate proteins of each enzyme. A primary function of SIRT3 is to support oxidative phosphorylation, which it achieves by activating the pyruvate dehydrogenase (PDH) complex and upregulating components of the respiratory electron transport chain (ETC) along with antioxidant defense mechanisms. By promoting aerobic respiration and mitigating reactive oxygen species (ROS), SIRT3 opposes key metabolic reprogramming events known to support tumorigenesis. The precise mechanisms underlying the tumor suppressor activity of SIRT4 have not been established, but one possible contributor is SIRT4-mediated suppression of glutamine catabolism.¹⁰

In several respects, SIRT5 antagonizes the actions of SIRT3 and SIRT4. Mitochondrial SIRT5 inactivates the PDH complex and components of the ETC to suppress respiration, and a cytosolic isoform of SIRT5 activates glycolytic enzymes.⁹ Collectively, these changes support the Warburg effect, a defining feature of proliferative metabolism. Our recent work shows that SIRT5 also supports mitochondrial glutamine catabolism, a metabolic hallmark of cancer that can be targeted for therapeutic benefit. Our current understanding of the functions of SIRT5 raises a number of questions. How does succinyllysine act as a tag to recruit ubiquitin ligases, which ligases are recruited, and does this mechanism operate on other proteins? Moreover, why does a mitochondrial deacetylase (SIRT3) oppose metabolic reprogramming and a desuccinylase (SIRT5) favor it? To frame this question differently, why does an increased mitochondrial succinylome hinder tumorigenesis and why does an increased acetylome favor it? Finally, can SIRT5 be targeted using small-molecule inhibitors in vivo, and are these inhibitors effective at blocking tumor growth?

Funding Statement

This work was supported by the Breast Cancer Coalition of Rochester; National Cancer Institute [R01 CA223534]; National Cancer Institute [R01 CA201402]; National Cancer Institute [U54 CA210184]; National Institute of General Medical Sciences [R35 GM122575].

Acknowledgments

We thank Cindy Westmiller for her assistance with manuscript preparation. The work was supported by NIH grants R35 GM122575, R01 CA201402, R01 CA223534, and U54 CA210184. M.J.L. gratefully acknowledges a research award from the Breast Cancer Coalition of Rochester.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.