Beyond energy and growth: the role of metabolism in developmental signaling, cell behavior and diapause

Embryogenesis is a complex series of processes that requires coordination between developmental progression and the biochemical reactions that comprise the metabolic landscape in cells, tissues and the organism. Cells must sense diverse nutrients and use metabolic pathways to produce the fundamental building blocks that fuel energy production and growth (Box 1). Moreover, products of metabolic pathways are vital factors supporting epigenetic modifications and the regulation of transcription and translation: key regulatory nodes that control embryonic development (Boxes 2 and 3). Differentiation often coincides with a shift in the metabolic profile of a cell as it becomes specialized in a subset of metabolic processes that support new cellular functions. Some cells exhibit metabolic flexibility upon differentiation, allowing them to adapt to genetic mutations or nutrient limitations. This concept is apparent when observing the clinical manifestation of mutations in metabolic enzymes in humans, collectively referred to as inborn errors of metabolism (IEMs).

IEMs are a class of monogenic diseases that, if not lethal, typically manifest in the first days and years of life, and have a broad spectrum of phenotypes (Alba, 2021). Some IEMs produce congenital malformations, such as shortening of limbs or craniofacial dysmorphia (Alba, 2021), an indication that metabolism is crucial for the development of embryonic structures in utero. Other metabolic deficiencies can lead to developmental delay and/or impaired neuronal or muscular development that manifest in the neonatal period (e.g. glycogen deficiencies; Ellingwood and Cheng, 2018). Importantly, these conditions can be a guide to the necessity of specific metabolic pathways during development, which is often conserved in model systems (Solmonson et al., 2022). What is particularly interesting in IEMs is the tissue-specific pathologies that manifest despite ubiquitous activity of some metabolic enzymes. For example, most cell types oxidize pyruvate in the mitochondria via pyruvate dehydrogenase (PDH) but individuals with PDH deficiency primarily display neuronal dysfunction (Pavlu-Pereira et al., 2020; Tanner et al., 2023). This suggests that other tissues can compensate for PDH deficiency by enacting mechanisms of metabolic flexibility. Which cell types are capable of metabolic flexibility and the mechanisms by which these cells compensate is an underexplored facet of developmental metabolism.

There are also crucial gaps in defining how aberrant developmental metabolism results in lethality, malformations or dysfunction in tissue-specific patterns. As we discuss here, there is considerable evidence for metabolic processes supporting development beyond energy production and growth. The goal of this Review is to provide a framework for understanding the interplay between metabolism and developmental processes. First, we examine how metabolic states evolve during development and are linked with cell growth and identity. Next, we discuss how metabolites regulate developmental signaling pathways. Finally, we examine how developing organisms monitor metabolic status and control developmental progression via diapause under periods of nutrient stress.

Metabolism is a dynamic process that adapts cellular and organismal nutrient availability to meet growth demands and promote developmental events. Embryonic development requires exquisite control of metabolic activities to support sustained biomass accretion, changing energetic demands and precise regulation of various cellular processes. Although gene expression changes are sufficient to adapt metabolic profiles, transcription is often not necessary for metabolic shifts to occur as the expression of a metabolic gene does not directly correlate to the level of its activity (Hicks et al., 2023; Ingolia et al., 2012). Transcriptional signatures are commonly used to determine embryogenic stages from early to late development, as unsupervised transcriptomic clustering identifies distinct developmental stages (e.g. 2C, 4C, etc.). In the same vein, metabolic profiles can also indicate distinct stages of embryonic development (Solmonson et al., 2022; Tennessen et al., 2014; Wadsworth and Riddle, 1989; Zhao et al., 2021). Metabolic transitions are likely linked to cells that require distinct nutrients to advance through developmental events, which could be considered metabolic checkpoints (Box 4; Fig. 1).

In vitro studies of preimplantation mouse embryos using limited nutrient conditions or stable isotopes (e.g. [¹³C]glucose) illustrate the need for specific metabolic activities at precise developmental times (Fig. 1). Preimplantation embryos require pyruvate to progress through the zygotic genome activation (ZGA) during the 2-cell (2C) stage, pyruvate or lactate (but not glucose) to advance from 2C to 8C stage, and glucose for blastocyst formation (Aoki et al., 1997; Brown and Whittingham, 1991; Lane and Gardner, 2000; Sharpley et al., 2021). These studies crucially demonstrate that metabolic requirements are intrinsic to cell stage and occur before the first differentiation event of embryogenesis (reviewed by Zhao et al., 2023). It is currently unclear why these specific nutrients are required when there is substantial overlap in their use by cells. Given that metabolism of glucose, pyruvate and lactate require co-factors such as NAD⁺ or NADH, one explanation could be to ensure redox balance, but this has not been clearly established.

Metabolomics and stable isotope infusions in pregnant mice have demonstrated that a metabolic transition occurs in the placenta and embryo from E10.5 to E11.5 (Solmonson et al., 2022), demonstrating increasing mitochondrial nutrient oxidation in the embryo throughout midgestation. Surprisingly, fetal tissues display disparate mitochondrial fuel preferences at this developmental stage (Fig. 1). Consistent with previous reports (Johnson et al., 2003), impairment in mitochondrial oxidation in E10.5 mouse embryos negatively impacts developmental progression. It is unclear whether enhanced mitochondrial metabolism during midgestation supports energy production, biosynthesis and/or aspects of cell signaling. This metabolic requirement may support distinct mechanisms in specific cell types, and the capacity for some cells but not others to survive mitochondrial dysfunction may be the basis for abnormal development. Identification of metabolic patterns that define embryonic stages, and specifically those that occur independently of transcription, is needed to understand new functions of metabolism during development. The availability of conditional alleles and cell type-specific gene editing provides many opportunities to test specific hypotheses and expand our understanding of the metabolic requirements of embryogenesis at a more granular level.

What general principles underly these cell-specific metabolic requirements? Waves of proliferation and differentiation drive embryonic development. As a result, cells must dynamically shift their metabolic programs to support the biosynthetic, energetic and signaling demands of each developmental stage (Box 3). During proliferation, metabolism is geared toward biosynthesis to double the amount of lipids, nucleotides and proteins required to produce a viable daughter cell. As cells differentiate, metabolism supports the transition to and demands of the new cell fate. Metabolic control of cell proliferation has been assessed in many different model systems that have solidified the essentiality of metabolism during development; over one century ago, measurements in sea urchins demonstrated an increase in oxygen consumption after fertilization (Warburg, 1928). Subsequent assessments of metabolic pathways in several model organisms, including developing chickens, rats and lambs, and in other models support metabolic control of cell proliferation and organismal growth (Hay et al., 1983; Hayashi et al., 2011; Shambaugh et al., 1977; Shenstone, 1968; Tennessen et al., 2011; Wadsworth and Riddle, 1989). Here, we discuss some key examples of how metabolism supports proliferating cells in meeting their biosynthetic and energetic demands. This is by no means a comprehensive inventory because we feel this topic has been discussed in detail elsewhere and refer the reader to many excellent reviews (Faubert et al., 2020; Sieber and Spradling, 2017; Vander Heiden et al., 2009).

One overarching theme to emerge from these disparate systems is that proliferating cells tend to show higher levels of glycolysis relative to non-proliferating cells. The rationale for why dividing cells increase glycolysis is multifactorial, including the higher demand of proliferating cells for biomass than for energy production (Vander Heiden et al., 2009). A non-proliferative cell can meet its energy demands by oxidizing glucose through glycolysis or oxidative phosphorylation in the mitochondria. Conversion of glucose to pyruvate in glycolysis generates a net of two adenosine triphosphate (ATP) molecules, whereas pyruvate oxidation by oxidative phosphorylation generates 36 ATP, making mitochondrial oxidation of pyruvate more energetically efficient (Lehninger et al., 2013). A cell that has been stimulated by growth factors has biosynthetic demands and must shift its metabolism to produce all the lipids, proteins and nucleotides for cell division. Thus, proliferating cells will take up more glucose, but rather than using it solely for ATP synthesis, some glucose will feed anabolic pathways, such as the pentose phosphate pathway, which supports nucleotide and lipid synthesis (Vander Heiden et al., 2009) (Fig. 2).

The oxidation of pyruvate in the mitochondria is a heavily regulated metabolic node with oxygen as a major regulatory factor because the electron transport chain (ETC) requires oxygen to maintain mitochondrial membrane potential and synthesize ATP. In many non-proliferative cells, hypoxia blocks pyruvate oxidation in the mitochondria and promotes pyruvate conversion to lactate – a pathway known as anaerobic glycolysis. In some cases, proliferating cells convert pyruvate to lactate (rather than oxidizing pyruvate in the mitochondria) even when oxygen is abundant – a pathway known as aerobic glycolysis or the ‘The Warburg Effect’ (Fig. 2) (DeBerardinis and Chandel, 2020; Tennessen et al., 2014). This metabolic phenotype has confounded researchers for many years because it is less energetically efficient, but it seems to support proliferating cells, in part, by maintaining a redox balance for continued glycolytic flux. Glycolysis requires NAD⁺ for the activity of glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and the conversion of pyruvate to lactate generates NAD⁺ from NADH (Fig. 1). In this way, lactate dehydrogenase (LDH), which converts pyruvate to lactate, is coupled to glycolysis through the regeneration of NAD⁺, and the secretion of lactate is required to maintain NAD⁺ production by LDH (Luengo et al., 2021). Importantly, the Warburg effect is often observed in vitro in cultured homogeneous cell types under conditions of surplus nutrients and sufficient oxygen, but in vivo cells often show different aspects of metabolism induced by their extrinsic environment (Faubert et al., 2020; Tennessen et al., 2011).

Proliferating cells can and do oxidize pyruvate in the mitochondria when oxygen is available, meaning not all proliferative cells exhibit the Warburg effect. Thus, it is important not to conflate high rates of glycolysis with the Warburg effect, as these phenotypes have disparate implications for cellular metabolism. Extracellular lactate accumulation can influence various aspects of metabolism and cell behavior (Gatie et al., 2022; Lane and Gardner, 2000; Oginuma et al., 2017; Tauffenberger et al., 2019). Therefore, in vivo, cells exhibiting the Warburg effect can couple lactate secretion in one cell type to lactate uptake by nearby cells, secrete lactate as a signaling mechanism for developmental events (Bulusu et al., 2017; Oginuma et al., 2017) or excrete lactate as a waste product for removal (Fig. 2). However, proliferative cells with increased rates of glycolysis concomitant with stable or increased mitochondrial respiration are not exhibiting the Warburg effect (DeBerardinis and Chandel, 2020). Cells under both conditions may demonstrate increased expression of glucose transporters and glycolytic enzymes, but the expression of specific isoforms of monocarboxylate transporters (Bosshart et al., 2019), activity of lactate dehydrogenase (Valvona et al., 2016) and oxygen consumption rates can distinguish between cells undergoing the Warburg effect and those simply exhibiting higher rates of glycolysis.

Proliferating cells can also use alternative fuels to meet their energy or growth demands. Fatty acids and amino acids can be oxidized in the mitochondria to support ATP synthesis in addition to glucose-derived pyruvate. In dividing cells, the synthesis of nucleotides and lipids is supported by intermediate metabolites linked to the tricarboxylic acid (TCA) cycle, including citrate and aspartate (Fig. 2). The export of these metabolites from the cycle (a process called cataplerosis) necessitates that the cycle be refueled in a process called anaplerosis (Owen et al., 2002; Vander Heiden et al., 2009). This is a conserved process that has been demonstrated in mammals, worms and flies (Cetinbas et al., 2016; Feng et al., 2023; Saari et al., 2019; Yang et al., 2014). Glutamine is an abundant amino acid that serves as an anaplerotic source in many proliferating cells. Another carbon source that can be used to meet growth and energetic demands is lactate. In adult mammals, lactate secreted from muscle and other tissues is taken up by the liver as part of the Cori cycle and secreted as glucose to fuel tissues with high glycolytic demand (Cori and Cori, 1929; Cori et al., 1939). However, many tissues, including the gut, adipose tissue, liver, muscle and tumors, use circulating lactate directly as an energetic fuel source for mitochondrial respiration or for anaplerotic reactions. (Faubert et al., 2017; Hui et al., 2017; Tasdogan et al., 2020). Lactate is also direct embryonic fuel in sheep, rodents and humans, with some studies demonstrating rat placenta, fetal liver and fetal brain can both excrete lactate and oxidize lactate to CO₂ (Bartelds et al., 1999; Carter et al., 1993; Hay et al., 1983; Marconi et al., 1990; Shambaugh et al., 1977). Glucose and lactate are the most abundant circulating carbon metabolites in mammals and the uptake of each nutrient has implications on the metabolic network beyond energy production, so it is reasonable that some developing tissues will alter lactate uptake or secretion based on redox balance (as discussed above) or to support cell-specific activities such as signaling. Early developmental growth can also be met by acquisition of intermediates from maternal stores, as is the case with deoxynucleotides in early Drosophila embryos (Song et al., 2017) or with amino acids acquired from breakdown of albumin or vitellogenin in the human and zebrafish yolk (Burton et al., 2001; Sant and Timme-Laragy, 2018). These are various examples by which growth and proliferation requirements in vivo can be achieved by various metabolic pathways using different nutrients.

As embryonic development progresses, cells transition from pluripotency to terminally differentiated cells with distinct functions. This process is accompanied by changes in both gene expression and metabolic programs. As we discuss above, many studies suggest that proliferating cells rely heavily on glycolysis but, upon differentiation, they increase the rate of mitochondrial oxidative metabolism (Chakrabarty and Chandel, 2021; Chung et al., 2007; Owusu-Ansah and Banerjee, 2009; Zhang et al., 2013; Zheng et al., 2016). Studies in humans, mice and flies have shown that the increase in mitochondrial oxidative metabolism is required for differentiation of cardiac cells, neurons, embryonic stem cells and hematopoietic cells (Chakrabarty and Chandel, 2021; Chung et al., 2007; Owusu-Ansah and Banerjee, 2009; Solmonson et al., 2022; Zhang et al., 2013; Zheng et al., 2016). Required changes in metabolism support the idea that metabolic transitions (reviewed by Mahmoud, 2023) are a component of differentiation programs and as cells differentiate, their metabolic activities shift to support the function of the new cell identity.

Cellular functions and the extracellular niche dictate the metabolism that is required during differentiation. Any given metabolic pathway may be essential for one cell type and dispensable for another. Metabolism of the non-essential amino acid serine clearly reveals this concept. Serine is taken up by cells and/or synthesized de novo using carbon derived from glucose and nitrogen derived from glutamine. Serine is essential for cells because it supports synthesis of proteins, nucleotides and phospholipids, yet the mechanisms employed to maintain serine levels are distinct among different cell types (Fig. 1).

The first step in de novo serine synthesis derives from glycolysis, where 3-phosphglycerate dehydrogenase (PHGDH) converts 3-phosphoglycerate to 3-phosphohydroxypyruvate. Dietary serine is not sufficient to support embryonic development because mice lacking PHGDH undergo embryonic lethality after E13.5 due to defects in central nervous system (CNS) development (Yoshida et al., 2004). Similarly, human deficiencies in PHGDH manifest primarily as CNS defects because serine is essential for synthesis of sphingomyelins and cerebrosides for neuronal development and for neuronal function, via its enantiomer D-serine (Tabatabaie et al., 2010). In mouse neural tissue, the expression of PHGDH is elevated in radial glia and astrocyte lineages but is largely absent from neurons, indicating that, during differentiation, neurons become exclusively dependent upon extracellular serine secreted by other cells (Furuya et al., 2000; Mitoma et al., 1998; Yamasaki et al., 2001). Some cell types display metabolic flexibility and resistance to PHGDH deficiency by taking up circulating serine, but the selective permeability of the blood-brain barrier requires that most serine in the CNS be synthesized locally (Lefauconnier and Trouve, 1983). Thus, PHGDH deficiency limits serine synthesis by astrocytes and glial cells, reducing the local serine concentrations in the CNS, causing aberrant neuronal development and function (Jaeken et al., 1996). The dependence upon de novo serine synthesis within the CNS could reflect an evolutionary mechanism to support brain function under conditions of sparse dietary protein but whether the expression of PHGDH is incompatible with neuronal survival and function is not known.

Some cell types use the serine biosynthetic pathway even under replete serine conditions, suggesting that biosynthetic efficiency is not always a driving factor for metabolic activities. During bone development, chondrocytes meet 50% of their demand for serine via the de novo serine synthesis pathway, even with sufficient availability of extracellular serine to meet 100% of their demand (Stegen et al., 2022). With genetic loss or pharmacological inhibition of PHGDH, cultured chondrocytes can adapt by increasing uptake of extracellular serine to maintain intracellular levels. Conversely, when extracellular serine is depleted (in culture or through dietary restriction), chondrocytes respond by increasing glucose catabolism and serine biosynthesis with both adaptations driven by an ATF4-dependent mechanism (Stegen et al., 2022). Although chondrocytes show metabolic flexibility in culture, genetic ablation of Phgdh in mouse growth plate chondrocytes in vivo leads to reduced nucleotide synthesis, decreased proliferation and reduced long bone length (Stegen et al., 2022). This demonstrates that although chondrocytes have the capacity for metabolic flexibility, there are in vivo limitations that may be related to serine availability in the avascular environment of the growth plate.

Cells can fail to differentiate and/or die due to a lack of metabolic flexibility, particularly when challenged by metabolic toxins or genetic deficiencies. Endothelial cells (ECs) require the serine biosynthetic pathway to maintain viability despite replete extracellular serine (Vandekeere et al., 2018). Mouse ECs rely on serine to support heme biosynthesis and glutathione production in addition to nucleotide and protein synthesis. Phgdh loss in mouse ECs depletes intracellular serine, reducing heme production, which causes electron transport chain dysfunction and increased oxidative stress (Vandekeere et al., 2018). With reduced glutathione synthesis, these cells cannot combat higher levels of oxidative stress and ultimately undergo apoptosis even with abundant extracellular serine (Vandekeere et al., 2018). This metabolic constraint is unique to ECs, and it is not understood why these cells cannot activate the same ATF4-dependent mechanism seen with Phgdh loss in chondrocytes. This example highlights the idea that cellular identity is intimately linked to the capacity for metabolic flexibility. Metabolic requirements are context dependent, thus as the environment or identity of a cell changes, so do nutrient demands and their fates.

The need for metabolic flexibility is not limited to IEMs or nutrient limitations, as cells may also demonstrate metabolic flexibility in the face of aberrant signaling. In mouse cardiac development, cardiac-specific knockouts of hypoxia inducible factor 1 α (HIF1α) are protected against embryonic lethality by the stimulation of amino acid oxidation through compensatory activity of HIF2α and ATF4, which supports developmental progression. In contrast, von Hippel Lindau (VHL) loss in the fetal heart (i.e. HIF1α activation) leads to embryonic demise (Papandreou et al., 2006; Seagroves et al., 2001). Specifically, stabilization of HIF1α in VHL-null cardiomyocytes forces these cells to produce ATP via glycolysis because PDH activity is limited by pyruvate dehydrogenase kinase (PDK) activation (Papandreou et al., 2006; Seagroves et al., 2001). Thus, the induced persistent glycolytic phenotype is incompatible with late gestation. These results suggest that compensatory metabolism can sometimes overcome signaling alterations to achieve developmental progress. What remains unclear is whether metabolic flexibility during development comes at a physiological cost at later points in the life of the organism.

The state and function of mitochondria can clearly influence gene expression through signaling by reactive oxygen species (ROS), redox balance, free calcium, epigenetic regulatory metabolites (Baksh and Finley, 2021; Chakrabarty and Chandel, 2021) and mitochondrial morphology (Bahat and Gross, 2019). However, a new report suggests that the way carbon is metabolized through the TCA cycle can impact differentiation. A non-canonical TCA cycle that uses ATP citrate lyase (ACLY) and the mitochondrial citrate/malate antiporter is required for embryonic stem cells to exit pluripotency and for myoblasts to differentiate into myotubes (Arnold et al., 2022). This finding suggests that changes to metabolic activities pre-empt some changes to cellular identity and raises the possibility that metabolism can direct cell fate. These new concepts need to be investigated further, but they support the prevailing notion that metabolism is context dependent and crucial for proper cell identity during embryonic development.

In addition to meeting the needs for cell proliferation and differentiation, metabolism participates in developmental signaling and embryonic patterning, which is achieved by secreted ligands that form gradients of signaling pathway activation. Retinoic acid (RA) is a classic example of a metabolite that functions as a morphogen during development. RA is a metabolite of vitamin A, synthesized in the limb bud, heart, eye and somatic mesoderm, forming a gradient of signaling activity (Berenguer and Duester, 2022; Chen et al., 1994; Maden et al., 1998). The synthesis of RA is driven by expression of retinol dehydrogenases and retinaldehyde dehydrogenases, and its degradation is mediated by cytochrome P450 family 26 (CYP26) enzymes, creating a boundary for the RA gradient (Hernandez et al., 2007; Sandell et al., 2007). Deficiencies in RA synthesis impact neuronal differentiation, eye and heart development, and spermatogenesis, demonstrating its crucial function for organogenesis in several tissues (Berenguer and Duester, 2022). The mechanism of small molecules as signaling mediators is not limited to RA and includes steroids, which activate steroid receptors (e.g. estrogen receptor), and lipids, which activate peroxisomal proliferator-activated receptors (Derosa et al., 2018; Fuentes and Silveyra, 2019; Peng et al., 2021). Thus, metabolism has an essential and broad role in regulating the availability of ligands for nuclear receptors to modify the gene expression.

Multiple groups have reported that gradients can also be formed by metabolites involved in glycolysis during presomitic mesoderm (PSM) development in mouse and chick embryos (Bulusu et al., 2017; Miyazawa et al., 2022; Oginuma et al., 2020, 2017). Genetically encoded reporters have demonstrated that cells of the PSM in midgestation mice (i.e. E8.5-E12.5) produce a gradient of glycolytic activity with higher rates of glycolysis near the tail bud (Bulusu et al., 2017) (Fig. 3A). Mechanistically, the expression of rate-limiting glycolytic enzymes are downstream of FGF8 signaling, with increased activity of glycolysis coupled to increased lactate production (Oginuma et al., 2017). Lactate is secreted through monocarboxylate transporters that co-transport a proton, so as lactate is secreted the extracellular pH decreases. Thus, the glycolytic gradient formed in the tail bud is connected to a pH gradient where the posterior region exhibits a lower pH (Oginuma et al., 2020). Pharmacological inhibition of glycolysis using the hexokinase inhibitor 2-deoxyglucose (2-DG) limits axis elongation (Oginuma et al., 2017). In tumors, acidic conditions can activate matrix metalloproteases that modify the extracellular matrix to promote cell migration (Gioia et al., 2010; Kato et al., 2005) and it is suggested that this mechanism also promotes migration of cells in PSM development because pharmacological inhibition of hexokinase or increasing the pH leads to reduced axis elongation and cell motility (Oginuma et al., 2017). Additionally, inhibition of hexokinase by 2-DG downregulates expression of WNT activators, such as AXIN2, suggesting crosstalk between Fgf signaling and Wnt signaling can occur through glycolytic activity. Alternatively, transgenic embryos with increased glycolytic flux of fructose 1,6 bisphosphate compared with wild-type levels also have impaired mesodermal segmentation, indicating that reduced or increased glycolysis beyond normal levels can result in developmental abnormalities (Miyazawa et al., 2022). Thus, a proper balance of glycolytic activity regulated in a spatiotemporal manner is essential for proper axis elongation and segmentation in mouse embryos.

A similar glycolytic signaling mechanism has also been demonstrated in mouse neural crest migration by YAP/TEAD1 action (Bhattacharya et al. 2020). The relationship between metabolism and cellular migration is a mechanism used by various cell types in different biological conditions and environments. Similar gradients that control cell migration have been reported with phospholipids and somatic gonadal precursors in Drosophila and zebrafish (Paksa et al., 2016; Sano et al., 2005; Starz-Gaiano et al., 2001). These examples provide multiple mechanisms whereby metabolic intermediates influence embryonic patterning and cell migration. Unlike other morphogens that are hydrophobic and require modifications and binding factors to establish gradients, polar metabolites may function more effectively as long-range morphogens and represent an underexplored avenue to regulate developmental signaling events.

In addition to metabolites functioning directly as morphogens, small molecules can covalently modify and alter the activity of signaling proteins (Fig. 3). This conserved and diverse mechanism can influence gene expression, protein distribution, localization, stability and enzyme activity. A classic example of this during embryogenesis is the post-translational modification of Hedgehog (Hh) by palmitate and cholesterol (Pepinsky et al., 1998; Porter et al., 1996). During auto-processing of Hh, a cholesterol moiety is added to the C terminus and palmitate is transferred to the N terminus through the acyl-transferase skinny hedgehog (Ski in Drosophila) (Chamoun et al., 2001). These modifications make the signal peptide highly hydrophobic and promote its tight association with lipid membranes, necessitating that Hh-producing cells require Dispatched (DISP) and a cognate factor called SCUBE2 as an additional regulatory step to release soluble Hh for long-range signaling (Burke et al., 1999; Creanga et al., 2012; Li et al., 2021a; Ma et al., 2002; Tukachinsky et al., 2012; Wang et al., 2021). Similar lipid modifications have been observed with WNT signaling (Galli et al., 2007; Liu et al., 2022; Rios-Esteves and Resh, 2013; Willert et al., 2003), suggesting that lipid metabolism is key to mechanisms of embryonic patterning and although it does not seem that lipids can be limiting for these reactions, it is clear that blocking these modifications results in abnormal development and disease (Resh, 2016). In addition to lipid modifications, acylation and glycosylation, as well as reversible and irreversible oxidation, all represent possible metabolite-dependent post-translational modifications that may have important implications for developmental biology. However, the role of metabolites in post-translational modification of proteins is vastly understudied and is likely underestimated as a mechanism of regulation during embryogenesis.

What happens when the nutrient environment cannot systemically provide the necessary resources for development? In addition to metabolic checkpoints (Box 3), which provide a means for coordinating local changes in tissue metabolism with developmental progression, organisms have evolved the ability to sense their systemic physiological state and arrest their development in the face of metabolic challenges. This arrest is commonly called diapause. During embryogenesis, many species, ranging from insects to large mammals, use embryonic diapause to delay embryonic development due to nutrient deprivation and other stresses (Murphy, 2012; Renfree and Fenelon, 2017). For example, some months are highly conducive for the birth and growth of progeny, whereas other months lack the resources to support offspring survival. Diapause allows progeny to be born when nutrients are abundant, and the weather favors efficient growth. There are two primary forms of embryonic diapause: obligate diapause, which occurs according to the time of year; and facultative diapause, which is induced by nutrient deprivation or stress.

This induced form of diapause is commonly seen in rodents and marsupials, and is controlled by the nutrient status and overall health of the female. Tammar wallabies can conceive throughout the year; however, depending on developmental and environmental conditions, Tammar wallaby embryos may enter and remain in diapause for up to 11 months (Murphy, 2012; Renfree and Fenelon, 2017). After a decrease in photoperiod (i.e. after the summer solstice), these embryos activate and resume development (Murphy, 2012; Renfree and Fenelon, 2017). As a result, most Tammar wallabies are born between January and March. In response to maternal stress or hormonal cues, diapausing embryos arrest development at the blastocyst stage of embryogenesis (Deng et al., 2018). Embryos in diapause display little to no growth; in some cases, e.g. bats, the embryo implants but does not differentiate (Murphy, 2012; Renfree and Fenelon, 2017). Studies in wallabies have shown that glucose and pyruvate uptake during embryonic diapause is relatively low but increases dramatically during embryonic reactivation (Spindler et al., 1995). Interestingly, reactivated embryos display only a modest increase in lactate production (Spindler et al., 1995), suggesting utilized glucose was consumed for oxidative metabolism or used by the pentose phosphate pathway. These studies highlight that glycolytic activity is suppressed during embryonic diapause and may play an essential role in embryonic activation after dormancy.

The suppression of insulin signaling is a crucial factor in embryonic diapause. Studies in killifish have shown that glycolysis is suppressed during embryonic diapause because treating embryos with exogenous glucose or glucagon can shorten the duration of embryogenesis, whereas treating cells with insulin or 2-DG can more than double the duration of embryogenesis (Gao et al., 2022). Gene expression studies in killifish have shown that cellular redox transcriptional programs are upregulated during diapause, suggesting that the role of glycolysis in NAD⁺/NADH redox balance may be vital in the duration of embryonic diapause (Gao et al., 2022). At the same time, lipid metabolism is thought to be suppressed by the Polycomb complex during embryonic diapause (Hu et al., 2020a). Although interesting, more in-depth metabolomic analysis is required to accurately define the shift in metabolic program observed during killifish diapause.

Similar to cellular quiescence (reviewed by Du et al., 2023), several studies in marsupials, fish and mammals have shown that embryos in diapause display significantly reduced levels of mitochondrial oxidative metabolism and an altered mitochondrial morphology (Hussein et al., 2020; Naeslund et al., 1980; Sousa et al., 2020). Interestingly, starvation and amino acid availability control this state of metabolic dormancy. Starvation suppresses TOR signaling via LBK/AMPK activation, whereas glutamine directly regulates TOR signaling (Hussein et al., 2020). Increased amino acid levels reactivate embryos in diapause, in part, through regulating TOR signaling (van der Weijden et al., 2021). Recent work has suggested that fatty acid oxidation is vital in sustaining cells in dormant states, such as cellular quiescence and embryonic diapause (van der Weijden and Bulut-Karslioglu, 2021). Consistent with this idea, impairing fatty acid oxidation reduces the viability of mouse embryos (Hussein et al., 2020; Yan et al., 2021). Given that glucose oxidation produces higher levels of ROS through activities of PDH and complex I of the ETC, reliance on fatty acid oxidation likely protects cells from oxidative damage during long periods of dormancy (Kushnareva et al., 2002). Overall, the metabolic state of embryonic diapause protects the embryo from oxidative damage while maintaining developmental competence to ensure progeny viability.

During worm development, physiological stresses such as heat, crowding and nutrient deprivation trigger a developmental checkpoint at the L1/L2 larval transition (Fielenbach and Antebi, 2008). This stress directs the worm into an alternative developmental program called dauer, in which the larvae remain relatively motionless but still respond to touch. Dauer animals are highly resistant to osmotic stress, redox stress, nutrient deprivation and temperature changes. Their lifespan is also increased from 2 weeks to 5 months (a 10-fold increase). To support this long-lived form, dauer animals display a dramatic shift in metabolism (Burnell et al., 2005; Fielenbach and Antebi, 2008; Madi et al., 2008; Penkov et al., 2015; Xie and Roy, 2012). In the face of metabolic stress, an integrated shift in trehalose/glucose use and NADPH redox balance drives the onset of dauer formation. This shift involves a negative-feedback loop where reduced NADPH levels impair dauer formation. Moreover, NADPH levels can be reduced by shunting glucose away from the pentose phosphate pathway into trehalose biosynthesis, which, in turn, suppresses dauer induction (Penkov et al., 2015). Dauer animals display elevated levels of anabolic pathways that enhance cellular maintenance and increase levels of polyunsaturated fatty acids (PUFAs), which have been proposed to protect cellular and organismal health (Burnell et al., 2005; Houthoofd et al., 2005). Interestingly, Dauer induction relies on NADPH production by the pentose phosphate pathway, suggesting redox balance of hormone-producing cells is a key regulator of dauer. Changes in NADPH levels may regulate Cyp450-reductases that alter the activity of cytochrome P450s, which control the production of the essential dauer hormones (i.e. dafachronic acids). This could provide a direct link between cytosolic NADPH levels and the regulation of dauer formation. Therefore, metabolism plays an essential role in the cue, onset and maintenance of Dauer formation.

Diapause also contributes to coordinating oogenesis and embryogenesis with the nutrient environment. During germline development, Drosophila eggs transition through 14 morphologically distinct stages. Interestingly, as egg chambers progress through development they undergo stage-specific stepwise accumulation of nutrients. At the beginning of vitellogenesis, the egg chamber accumulates yolk proteins. During stages 9-10 of oogenesis, oocytes begin to store massive quantities of triglycerides and cholesterol esters. Finally, during late stages 12-14 of oogenesis, oocytes store large amounts of glycogen. During these stages, high levels of insulin signaling promotes lipid storage but must be downregulated to stimulate glycogen synthesis. However, depriving female flies of dietary amino acids induces reproductive diapause by suppressing insulin signaling (Drummond-Barbosa and Spradling, 2004; Hsu and Drummond-Barbosa, 2011; LaFever and Drummond-Barbosa, 2005; Sieber et al., 2016). Reducing germline insulin/Akt signaling blocks stem cell activity and induces mid-oogenesis arrest (Drummond-Barbosa and Spradling, 2004; Hsu and Drummond-Barbosa, 2011; LaFever and Drummond-Barbosa, 2005). These studies show that the germline monitors feeding status and nutrient levels by sensing circulating insulin-like peptides (ILPs) in the hemolymph (Drummond-Barbosa and Spradling, 2004; Hsu and Drummond-Barbosa, 2011; LaFever and Drummond-Barbosa, 2005).

Interestingly, during stages 9-10 of Drosophila oogenesis, the sterol response element binding protein (SREBP) senses lipid levels in developing egg chambers and controls LpR2-mediated (LDLR ortholog) lipid uptake in the developing egg chambers (Sieber and Spradling, 2015). Depriving females of dietary lipids or mutating LpR2 in the germline triggers an arrest at stage 8 of oogenesis and a subsequent turnover of developing oocytes that display suboptimal triglyceride levels (Parra-Peralbo and Culi, 2011). Consistent with this lipid storage checkpoint in the Drosophila germline, inactivating diacylglyceride acyl transferase (DGAT) in germ cells, which reduces triacylglycerol storage in lipid droplets, also triggers a precise stage 8 arrest (Buszczak et al., 2002). Feeding female flies a lipid-free diet does not affect oocyte triglyceride levels but significantly reduces egg laying (Sieber and Spradling, 2015). Interestingly mammalian oocytes also store large amounts of neutral lipids that are required for early embryogenesis (Dunning et al., 2014a,b, 2010). This fly stage 8 checkpoint monitors nutrient levels and oocyte quality, and ensures oocytes have sufficient lipids to survive embryogenesis. Under low nutrient conditions, the stage 8 checkpoint, arrest of germline stem cell activity and observed female reproductive diapause function are a crucial mechanism that couples the reproductive output to nutrient availability. As evidence of the conservation of this system, women suffering from anorexia nervosa display infertility, an arrest of oocyte development and defects in menstruation, suggesting a similar checkpoint exists in human oogenesis (Bulik et al., 1999; Letranchant et al., 2022; Starkey and Lee, 1969). This demonstrates that the role of nutrition in development occurs before fertilization as it has important implications for the health and longevity of the offspring.

Metabolism is crucial for accumulating biomass during cell proliferation and energy production, but its role in development expands beyond these functions. In this Review, we have highlighted many alternative roles that metabolites play in supporting embryogenesis. The metabolism that sustains embryogenesis facilitates biosynthesis, energy production and regulation in cells, tissues and organisms, providing a scale for investigation beyond many other topics. Advances in technologies have put developmental metabolism on the precipice of a vast expansion of knowledge, making this an exciting time for the field. Investigating metabolic phenotypes in well-characterized models may open new avenues for understanding metabolic mechanisms that drive developmental processes. The degree to which metabolism is flexible during development and the ability of metabolic changes to drive cell fate are new frontiers that can invigorate a new type of investigation for developmental biology.

We thank Margaret Cervantes for helpful comments and editing.