Methionine restriction and mimetics to ameliorate human aging and disease

L-methionine α-deamino-γ-mercaptomethane-lyase (Methionine-γ-lyase or methioninase)

Methioninase is a bacterial enzyme originally isolated from Clostridium sporogenes that degrades methionine to ammonia, α-ketobutyrate, and methanethiol [43]. Other sulfur-containing amino acids (i.e., cysteine, homocysteine, cystathionine, etc.) are also likely sensitive to methioninase [44, 45]. Reprogramming methionine metabolism with such non-mammalian enzymes represents a 'xenotopic approach' [46]. Genetic encoding of methioninase in yeast [47] and Drosophila [48] extends lifespan. In mice, recombinant methioninase prevents diet-induced obesity [49]. Such examples provide conceptual support to translate methioninase as a therapeutic strategy for humans.

Production of recombinant methioninase in E. coli [34], when coupled with additional engineered modifications such as conjugation with a polyethylene glycol (PEG), confers reduced immunogenicity [50] for human administration [51]. Building upon the observation that multiple cancer cell types exhibit methionine dependency, coupled with a signal of efficacy from preclinical studies in murine cancer models, recombinant methioninase has been tested in human clinical trials [33, 52, 53]. Indeed, recombinant methioninase stabilized prostate-specific antigen (PSA) levels in patients with advanced prostate cancer [37].

Caveats of xenotopic proteins include their immunogenicity and the potential generation of harmful metabolic byproducts, which for methioninase includes ammonia, α-ketobutyrate, and methanethiol. Nonetheless, early promising results provide rationale for ongoing efforts to mitigate off-target effects. For example, a thermophilic form of methioninase has been developed to deplete methionine in food, while extracting degradative byproducts (methanethiol and its derivatives) via heating [54]. In mice, such an engineered diet reduced serum, liver, and intestinal methionine concentrations and inhibited prostate tumor growth. Re-engineering the human cystathionine-γ-lyase enzyme—which normally has no methioninase activity—to selectively degrade methionine circumvents the immunogenicity problem and effectively reduces serum methionine levels and inhibits growth of prostate tumors without causing weight loss or toxicity [55].

Degradation of dietary methionine by mild oxidation

The low palatability of chemically-defined MetR foods is addressable by instead degrading the methionine component of casein by mild oxidation [56]; however, additional amino acids such as tryptophan and lysine are also degraded and therefore must be reintegrated in the diet. Administration of such a diet to mice resulted in metabolic benefits comparable to a standard chemically defined MetR diet, including increased energy expenditure, decreased body weight (despite increased food and water intake), increased serum FGF-21, and improved metrics of glucose homeostasis. The oxidized casein diet also produced similar effects on intermediates of methionine metabolism and liver transcriptional programs [56]. However, it remains unknown whether oxidation of protein sources other than casein would also effectively degrade methionine without the generation of harmful byproducts.

Methionine adenosyltransferase 2A (MAT2A) inhibitors

Methionine adenosyltransferase (MAT) is conserved from E. coli to humans and catalyzes the first step in methionine metabolism that produces S-adenosyl-L-methionine (SAM) from methionine and ATP. SAM is the principal methyl donor and the second most widely used enzyme substrate after ATP [57]. Driven by methyltransferases, SAM donates its methyl group to acceptor molecules (e.g., DNA, RNA, proteins, and other cellular metabolites), generating S-adenosyl-L-homocysteine (SAH) as a byproduct. Methyltransferases regulate the epigenomic landscape, gene expression, lipid biosynthesis, protein repair, protein translation, autophagy, and cell differentiation, amongst other processes that may link MetR to downstream health benefits [58]. MAT is encoded by three different genes in mammals, MAT1A, MAT2A, and MAT2B, with the former two being the catalytic subunits [59]. MAT1A is predominantly expressed by hepatocytes, while MAT2A is expressed by non-hepatocytes in the liver and broadly in extrahepatic tissues [60]. Limiting MAT1A-activity in the liver is detrimental, as reduced SAM is associated with chronic liver disease. By contrast, inhibition of MAT2A may phenocopy some benefits of MetR without detrimental effects on liver health. Of recently developed MAT2A inhibitors, at least two (IDE-397 and AG-270/S095033) have been tested in clinical trials for treatment of solid tumors and lymphomas [61]. AG-270/S095033 administration to patients with advanced malignancies (NCT03435250) reduced plasma SAM concentrations by 54% to 70% [62]. Moreover, novel MAT2A inhibitors penetrate the blood–brain barrier, thereby modulating SAM levels in the brain [63]. Whether MAT2A inhibitors can functionally mimic MetR to delay aging remains to be established.

Methionine aminopeptidase 2 (MetAP2) inhibitors

MetAP enzymes remove methionine from the amino-terminus of newly synthesized proteins, crucial for protein stability while allowing for amino terminal modifications such as acetylation by N-alpha-acetyltransferase and myristoylation of glycine by N-myristoyltransferase. MetAP2 is one of two cytoplasmic methionine aminopeptidases expressed in mammalian cells [64]. In addition to removing N-terminal methionine, MetAP2 binds eukaryotic initiation factor 2α, protecting it from inhibitory phosphorylation, and promoting general protein synthesis [65]. Originally developed as anti-angiogenic agents for cancer, several highly potent MetAP2 inhibitors have been tested in human clinical trials, though none are approved, in part due to sustained weight loss [64]. These properties form the basis for ongoing development for obesity and metabolic diseases [64, 66]. At low doses that do not affect angiogenesis, MetAP2 inhibitors reduce weight in mice and humans suggesting alternative mechanisms underlying their anti-obesity effects [67, 68]. One possibility is that the intracellular methionine pool is reduced by inhibiting cleavage of methionine from the amino terminus of newly synthesized proteins, thereby creating a state of MetR. Indeed, the MetAP2 inhibitor beloranib increased two circulating biomarkers of MetR efficacy (adiponectin and FGF-21) in clinical trials [66]. A network meta-analysis of 96 randomized controlled trials for weight loss identified beloranib as the top-ranked efficacious treatment in women [69]. However, it remains to be established whether MetAP2 inhibitors effectively reduce intracellular methionine levels and whether apparent beneficial effects are attributable to a MetR-dependent mechanism.

Live synthetic biotics

Engineering of live bacterial cells to enzymatically degrade methionine may allow local control of methionine concentrations or systemic levels by targeting gastrointestinal methionine and its absorption [70–72]. One approach [71] generated an attenuated strain of Salmonella typhimurium, overexpressing methioninase, to deprive cancer cells of methionine. This attenuated Salmonella typhimurium strain, which does not produce functional endotoxin, exhibited a well-tolerated safety profile in phase I clinical trials [73]. A single intratumoral injection significantly reduced tumor methionine concentrations and growth of PC3 prostate and MDA-MB-231 triple-negative breast cancer xenograft tumors without affecting methionine blood levels or body weight [71]. The same approach reduced growth and metastasis of orthotopic transplants of murine breast cancer cells in immune-competent BALB/c mice [71].

Oral gavage of E. coli that produces recombinant methioninase (E. coli JM109-rMETase) effectively reduced body weight in aged mice compared to control bacteria, albeit to a lesser degree than dietary MetR [72]. Both MetR and the probiotic significantly reduced plasma methionine after 15 days of treatment, however, the effect was sustained at 29 days only with dietary MetR. One caveat is that the mice were gavaged twice daily, which presents a practical challenge for a potential long-term treatment and longevity study [72]. Another engineered methioninase-producing bacterium (Salmonella typhimurium, A1-R-rMETase) also reduced intratumor methionine levels and delayed xenograft tumor growth [74].

Synthetic probiotics are live biotherapeutics introduced to the gastrointestinal tract, which can be engineered to replicate the benefits of dietary restriction. A non-pathogenic ‘Nissle 1917’ strain of Escherichia coli, already used in the treatment of gastrointestinal conditions, was engineered to express methionine decarboxylase (MetDC), an enzyme that converts methionine into 3-methylthiopropylamine (3-MTP) [70]. In mice with liver-specific downregulation of the cystathionine β-synthase (CBS) gene (which models classic homocystinuria), this novel probiotic (SYNB1353) reduced plasma methionine and homocysteine levels. Its administration to male cynomolgus monkeys following an oral methionine load blunted the resultant plasma surge of methionine and homocysteine. In a first-in-human randomized, placebo-controlled, double-blinded phase 1 study in healthy volunteers [70], SYNB1353 was generally well tolerated with mild to moderate, predominantly gastrointestinal, adverse events that were transient. SYNB1353 decreased plasma methionine levels after an oral methionine load—and in contrast to the monkey study—did so without affecting homocysteine [70]. The main caveats of synthetic probiotics are the potential for compensation by dietary cysteine (discussed in ‘compensatory mechanisms’ below) and the outstanding question of whether the modest degree of methionine depletion is sufficient to achieve clinical benefits.

Cysteine.

Cysteine is a sulfur-containing amino acid and a product of methionine metabolism, the depletion of which may explain some of the beneficial effects of MetR, particularly weight loss [25, 101, 102]. Indeed, in a model of cystathionine γ-lyase deficiency, which impairs the conversion of methionine to cysteine, resultant cysteine deprivation in the absence of matched methionine lowering triggers rapid weight loss, indicating that cysteine may be a more important component than methionine in driving some benefits of SAAR [101, 102]. Importantly, the first proof-of-concept study of dietary MetR in humans with metabolic syndrome showed increased fat oxidation and reduced hepatic lipid content after 16 weeks of MetR [17]. This study may have underestimated benefits of MetR, however, as the dietary product used to achieve MetR contains supplemental cysteine. By contrast, studies in model organisms are typically conducted with diets also deplete of cysteine, underscoring that a more accurate term is sulfur amino acid restriction (SAAR), even though MetR has been more commonly used historically.

Microbiota-derived methionine.

Aside from the diet, gastrointestinal microbes are an additional potential source of methionine. MetR influences microbiome composition, and in turn, microbiota also affects host methionine metabolism via complex mechanisms beyond the scope of this review. Importantly, microbiota-derived methionine production is sufficient to compensate for extreme dietary restriction consistent with a state of methionine starvation. Ubiquitous expression of methioninase in Drosophila degrades all host cell-derived methionine and most flies exhibit a dramatically shortened lifespan in line with that expected of extreme methionine starvation; however, a small fraction are “super survivors”. This subset is lost when the Drosophila flora is eliminated by antibiotic treatment, suggesting that microbiota-derived methionine is sufficient to compensate for extreme methionine depletion [48]. It is therefore plausible that microbiota-derived methionine may neutralize MetR benefits in future human trials.

Intercellular methionine transfer.

In clear cell renal cell carcinoma, tumor-derived succinate acts as a trigger for a subset of pericytes to activate autophagy for methionine scavenging, thus promoting tumorigenesis and drug resistance by shunting methionine to cancer stem cells. [103]. It remains unclear whether MetR affects methionine levels equally across cell types or whether achieving effective reduction of methionine levels in tissues critical to therapeutic benefit can be prevented by methionine transfer from other tissues. Among nine potential methionine transporters, evidence suggests either SLC7A5 or SLC43A2 are essential for methionine transport into cancer stem cells, while also establishing potential molecular basis for heterogeneity of methionine transport and intracellular abundance. Determining which methionine transporters regulate cell-type specific methionine uptake may inform development of small molecules targeting specific transporters to achieve tissue-specific MetR mimetic benefits.

Extracellular vesicle (EVs) derived methionine enzymes.

Key enzymes involved in methionine metabolism (MATα2/MAT2β) are transferred between cells via EVs [104]. This may limit efficacy of MetR-mimetics targeting specific tissues.

Collectively, these examples illustrate the complex and interconnected nature of the metabolic pathways that dictate the effective methionine concentrations at the tissue level. Consequently, there is not always a matched relationship between absolute methionine in the diet and tissue exposure, even though methionine is an essential amino acid. This complexity will be critical to consider in the interpretation of clinical trials results as it could account for variable signals of efficacy within or across studies.

Study population.

A common early translational approach is to enrich the study population with high-risk participants most likely to benefit from the intervention, thereby gaining statistical power to detect efficacy within a reasonable amount of time. The ideal study population would therefore include participants already exhibiting early evidence of a particular pathology that may be sensitive to the experimental intervention. For MetR, the most robust and consistent signal of benefit from preclinical studies is on metrics of metabolic health, including insulin sensitivity and beta cell function. Importantly, insulin resistance underlies a plethora of cardiometabolic diseases [106, 107]. Age-associated insulin resistance manifests independently of body mass index, and extreme longevity is associated with insulin sensitivity [108]. Moreover, drugs that improve glucose metabolism and related metabolic health parameters (metformin, GLP-1 agonists, and SGLT-2 inhibitors) are among the most studied in anti-aging clinical trials [109, 110]. Not surprisingly, early MetR clinical trials were performed in participants with excess body weight or obesity and demonstrated preliminary evidence of improved metabolic health (Sup. Table 1). Population studies also demonstrate associations between sulfur amino acid intake and cardiometabolic diseases [111–113]. Collectively, this provides a rationale for the design of a study focused on metabolic aging with insulin sensitivity as a primary outcome measure that is modifiable in a relatively short time, while also establishing a framework with broader applicability to other aging-related diseases (e.g., coronary artery disease, hypertension, Alzheimer's disease). We therefore propose that an early study should focus on relatively young participants prior to the inflection point in the fifth decade when the risk of aging-related metabolic disease increases dramatically [114] but exhibit relatively modest metabolic dysfunction (e.g., prediabetes). Since lifestyle interventions, including dietary interventions, remain the cornerstone of therapy for prediabetes, such a study would not present ethical issues; participants randomized to a control arm would receive standard of care life-style counseling and any control diet with ‘high/normal’ methionine content could reflect macronutrient content in line with current clinical guidelines.

Outcome measures.

The primary outcome measure for any MetR clinical trial should be dictated by the age-related pathology that is the focus of the study. In addition, at this early stage of human translation, for which the size and duration of any proposed study is likely to be limited in scope, we suggest selection of endpoints that are sensitive and quantitative. For a study focused on insulin resistance, for example, we propose using glucose infusion rate (GIR) during hyperinsulinemic-euglycemic clamp as a primary endpoint, or alternatively indices that utilize glucose and insulin concentrations measured in fasting plasma samples (e.g. Homeostasis model assessment of insulin resistance index, HOMA-IR) or samples collected during an oral glucose tolerance test (e.g., the Matsuda Insulin Sensitivity Index [115]).

Secondary endpoints should reflect four key objectives. (1) Secondary endpoints ideally provide an orthogonal test of the primary hypothesis. With metabolic aging, for example, directionally consistent improvements in one or more related metrics of glucose and lipid metabolism—such as HbA1C, clinical lipid/cholesterol panels, and non-esterified free fatty acids— would strengthen any potential signal of efficacy for metabolic aging. (2) When feasible, secondary assessments of health parameters beyond those affected by the primary disease focus should also be assessed. Clinical trials of the most popular—albeit unproven—anti-aging interventions, such as rapamycin/sirolimus, nicotinamide riboside, metformin, spermidine, senolytics, and others, have outlined measures relevant to various healthspan domains that can be assessed in MetR trials [109, 110]. Such assessments could identify potential signals of benefit in domains for which there is already an evidence base (e.g. neuromuscular function) and/or to detect unexpected signals of benefit to guide design of subsequent trials to test novel emerging hypotheses (e.g. measures of vascular function: blood pressure, intima media thickness, flow-mediated dilation). (3) Secondary endpoints may report on the biology of aging more broadly, guided by the concept that common mechanisms underlie diverse diseases of aging. For example, an expert task force suggested a set of outcomes for geroscience trials [105], incorporating onset of age-related disease and various healthspan parameters into a composite score. Several recently developed ‘omics clocks’ may report on biological age, although surprisingly, dietary MetR did not significantly affect epigenetic clocks in mice or humans [1]. It remains to be determined whether this null result reflects a limitation of the specific ‘clocks’ deployed or insufficient intensity, duration, or efficacy of MetR to modify systemic biological age. Incorporation of such analyses in future clinical studies may clarify fidelity of the ‘clocks’ while also holding potential to report on systemic aging. (4) Lastly, exploratory secondary endpoints may provide mechanistic insight. Multi-omics analyses, for example, may reveal candidate pathways mediating effects of MetR. In addition, longitudinal analyses could reveal MetR-mediated slowing of molecular aging or overt rejuvenation to a more youthful molecular signature.

Degree of MetR.

Studies in murine diet-induced obesity demonstrate that MetR is most effective within a relatively narrow range of dietary methionine content. Efficacy requires achieving a reduction below 0.25%, by weight; however, extreme continuous reduction of dietary methionine leads to detrimental effects and a de facto state of methionine starvation [116]. The range of dietary methionine concentrations should be tested until the optimal methionine content of food is defined. It will also be critical to incorporate measurements of plasma methionine levels, urinary sulfate levels (a potentially more reliable readout of effective methionine restriction), biomarkers of the physiological response to MetR (e.g. FGF-21 and IGF-1), and functional parameters related to bone (e.g. turnover markers, bone density and microarchitecture), which is the most likely tissue to exhibit harm from MetR [91, 117].

Trials of MetR mimetics.

Considering the challenge of long-term compliance with any dietary intervention in humans, coupled with the onerous limitations imposed on food choices by MetR regimens, capturing benefits of MetR in the broader population may ultimately require development of MetR mimetics. We do not advocate for delaying translational studies of dietary MetR, some of which have been completed and which will continue to be critical for establishing proof of concept for targeting methionine metabolism. However, multiple compounds with potential MetR mimetic properties are at various stages of clinical development. Ultimately, mimetics are unlikely to recapitulate all MetR effects (whether positive or detrimental) due to tissue-specific differences in pharmacokinetic penetration and/or methionine metabolism. Therefore, it will be interesting to move beyond typical placebo-controlled phased development and test mimetic effects head-to-head with dietary MetR. We propose starting with MAT2A inhibitors because: i) they should not affect SAM levels in the liver; ii) inhibition of MAT2A is associated with improved regenerative capacity and strength of aged skeletal muscle [118], with reduced liver fibrosis [104, 119], and improved bone function [120, 121], which contrasts concerns that dietary MetR may negatively impact bone density; and iii) MAT2A inhibitors have promising anti-cancer properties in mouse studies [31, 122, 123].