Chemical Embryology Redux: Metabolic Control of Development

Abstract

Recent studies on metabolic control of development are new additions to the models of embryogenesis, so far dominated by genes and signals. But it is not a new idea, and can be traced back to Joseph Needham’s “Chemical Embryology”, published in 1930s. Even though Needham’s ideas failed to gain traction and fell by wayside with the advent of genetic studies of embryogenesis, they are now experiencing a comeback, enabled by the powerful merger of metabolomic and genomic techniques. Here we review recent studies that quantify the energy budget of embryogenesis in Drosophila and begin to untangle the interwoven links between core anabolic processes. Dynamic coordination of metabolic, genetic, and signaling networks appears to be essential for seamless progression of development.

Revisiting Chemical Embryology

Chemical embryology, in which developmental processes are analyzed and interpreted at the level of the metabolic reactions, offers insights to both fields of metabolism and development. On the one hand, embryos are powerful platforms to analyze the dynamic regulation of metabolism at the organism level; metabolic fluxes during embryogenesis is tightly controlled to support rapid and precisely timed biochemical transformations. On the other hand, the metabolic perspective adds depth to the gene-based models of development by grounding them with physical principles such as mass and energy conservation. Studies of embryogenesis from a metabolic perspective can be traced back to the collaborations between embryologists and biochemists in early 20^th century [1]. Spearheaded by Joseph Needham, these studies were primarily concerned with the physicochemical characterization of embryogenesis with the rigor of in vitro biochemistry [2]. For instance, Needham and colleagues, which included Conrad Waddington, attempted to identify the chemical that was responsible for the then recently discovered Spemann Organizer’s ability to induce the amphibian embryo’s neural tube. However, with limited biochemical tools, they fell short of discovering chemical mechanisms governing embryogenesis. Although Needham failed to convince contemporary embryologists and biochemists of the value of chemical embryology, his approach to biology foreshadowed the rise of molecular biology a generation later [3,4]. Meanwhile, the powerful collaboration between genetics and embryology became the main driver of developmental biology, and shaped our present understanding of development [5,6].

Recent studies, combining quantitative tools from molecular biology, live imaging, genetics, and analytical chemistry, have shed new light on the complex role of metabolism in multiple developmental contexts [7]. In Drosophila melanogaster, insulin signaling was shown to induce metabolic quiescence in oogenesis [8], while Atg1-Tor signaling and a global transition in gene expression have been separately linked to increasing catabolic activity during embryogenesis [9,10]. Core metabolic pathways have been also identified as the new regulators of epigenetic and post translational modifications. In early mammalian embryos undergoing zygotic genome activation, the acetyl groups required for histone modifications are produced by the TCA cycle enzymes that are transiently localized to the nucleus in the presence of pyruvate [11]. In the 8-cell stage of mouse embryogenesis, glucose does not contribute to ATP production, but is instead used to generate post-translational modifications that govern cell fate specification [12]. Overall, intricate metabolic regulation appears to be essential for supporting the rapidly changing biochemical demands of development. Here, we discuss how Drosophila, one of leading models of developmental biology, can be used to dissect metabolic control of embryogenesis at multiple levels of organization, from a single enzyme to an entire organism.

To begin, we review the basic energy and mass balance of Drosophila embryogenesis, and single out DNA precursor synthesis as the most conspicuous anabolic task during embryogenesis. Next, we summarize recent findings on the regulation of dNTP synthesis during the nuclear cleavage cycles, and on the connection between dNTP metabolism and the midblastula transition. In closing, we present a general overview of dNTP metabolism in cell cycle control and embryogenesis of other animal species.

Physiology and energy balance of embryogenesis

Drosophila embryo is about 10 nL in volume and develops outside the mother, for roughly 24 hours at room temperature [13,14]. During this time, the number of nuclei increases from 1 to roughly 10⁵ [15]. Compared to the dramatic increase in DNA content, the majority of the macromolecular components remain relatively constant throughout embryogenesis. The membrane lipid composition and total RNA contents remain largely unchanged, as measured by lipidomics and biochemical assays [15,16]. Additionally, only modest changes in total protein content and the proteome composition were reported from biochemical and proteomics assays [10,17–19]. Overall, these results imply that embryogenesis is executed largely through the maternally deposited cellular machineries, and that new macromolecules are synthesized by reorganizing the maternally deposited ones. For instance, maternally deposited yolk proteins are degraded into amino acids and polymerized into new functional proteins, and maternal stores of phospholipids are reassembled to form new membranes.

The energy budget of embryogenesis is limited by the maternally deposited fuel sources. Since embryos do not perform work on the environment, the change in internal energy can be measured through the heat dissipation rate. Single embryo calorimetry measurements revealed that heat dissipation of Drosophila embryos increases linearly from 110 W to 170 W (~1 to ~1.6 pmole of ATP/embryo/s) from fertilization to hatching [20] (Figure 1B). This gradual increase in energetic output is in agreement with previous reports of the up-regulation of energy production pathways throughout embryogenesis [10,17]. The net energetic cost of embryogenesis was also estimated from respirometry and biochemical measurements of fuel depletion. Overall, calorimetry, respirometry, and biochemical experiments all suggest that embryogenesis is fueled by burning ~10 mJ worth of maternally deposited glycogen and triacylglycerol stores [20] (Figure 1A).

Figure 1. Energy balance of embryogenesis.

(A) Schematic of energy balance in Drosophila embryogenesis. Embryogenesis is fueled by burning ~10 mJ worth of maternally deposited glycogen and triacylglycerol. (B) Heat dissipation rate per embryo throughout embryogenesis until hatching. The colors represent separate experiments. The spike observed at the beginning of the time course is from introducing the embryo-containing ampoules into the thermal bath at 22°C. The inset shows the representative measurements of oxygen consumption rate during Drosophila embryogenesis at 25°C, adapted from Lints et al., [69]. This figure is reproduced with permission from Figure 1, Song et al., [20].

To understand how Drosophila embryogenesis is programmed with respect to the energy budget of 10 mJ, we must be able to quantitatively account for the energy budget at the level of biomolecular processes. Interestingly, the major biosynthetic processes of protein, RNA, and DNA polymerization consume less than 10% of the energy budget of the early Drosophila embryo. Although many studies from the past century provided valuable measurements of the macromolecular composition and energy usage in embryogenesis [2,21–24], even the major energy sinks of embryogenesis are difficult to pinpoint at the molecular level. Recently, the energy production rates of frog and zebrafish embryos have also been measured [25–27]. In particular, Rodenfels et al., provided quantitative estimates of the energy consumed by cellular processes, and suggested that the plasma membrane synthesis and maintenance costs could account for up to 50% of the early zebrafish embryo’s energy budget. Overall, a quantitative map of the energetic requirements could provide new insights to the design principles of embryogenesis, as has been recently demonstrated for T7 and influenza viruses [28].

Regulation of dNTP synthesis during the nuclear cleavage cycles

Fertilization of the Drosophila egg is followed by 13 nuclear cleavage cycles that occur at a breakneck speed of 10–20 minutes per division (Figure 2A). Around the 13^th cycle (NC13), zygotic genome activation (ZGA) takes place, after which the cell cycles slow down considerably and cellularization occurs (reviewed in [29]). The explosive increase of genome copy number during the cleavage cycles imposes an exponentially growing demand for DNA precursors, deoxynucleoside-triphosphates (dNTPs) (Box 1). The total amount of dNTPs needed for 13 cleavages can be calculated in a straightforward way:

$$\text{dNTPs required}:=\frac{\text{ploidy × [\# of nuclei] × [genome size (bp)]}}{\text{[Avogadro’s \#] × [cell volume (L)]}}\cong \text{1 mM}.$$

Figure 2. Supply and demand of dNTPs during the nuclear cleavage cycles.

(A) Schematic of the nuclear cleavage cycles in early embryogenesis. The duration of the nuclear cycles are taken from Djabrayan et al., [38]. (B) Schematics of the dNTP synthesis pathway in embryos of different genotypes. During the early nuclear cleavage cycles (pre-NC10), the endogenous dATP level inhibits RNR activity. The inhibited RNRs are represented as hexamers of the $\alpha$ subunit. In MTD > RNRL^D68N embryos, dATP cannot inhibit RNR activity, causing dNTP levels to rise far above wild type levels. In SHMT null embryos, the lack of SHMT activity leads to dTTP depletion in NC13. The success/failure of NC13 are depicted in the right most column.

BOX1: dNTP metabolism.

The rate limiting step of de novo dNTP synthesis is catalyzed by the enzyme ribonucleotide reductase (RNR) in all life forms (Figure IA). Eukaryotic RNR is composed of two subunits, $\text{α}$ and $\text{β}$, encoded by RnrL and RnrS in Drosophila [31]. The active form of RNR is the ${\text{α}}_2{\text{β}}_2$ tetramer which catalyzes the substrates ribonucleoside-diphosphates (NDPs) into deoxynucleoside-diphosphates (dNDPs) [57–59] (Figure IB). The $\text{α}$ subunit contains the catalytic site as well as the two allosteric regulatory sites that govern RNR activity and specificity. The $\text{β}$ subunit houses the tyrosyl-radical group essential for catalysis, which can be transferred to the catalytic site of the $\text{α}$ subunit through a proton-coupled electron transfer pathway [60,61]. The dNDPs produced by RNR are subsequently phosphorylated to its corresponding triphosphate forms by NDP kinase (Figure IB). dNTPs can be also synthesized through the salvage pathway, which has been shown to be essential in specific contexts, such as mouse lymphocyte development [62,63] (RNR and dNTP metabolism are extensively reviewed in [42,47,64]).

The production of dTTP requires more steps than that of dATP, dCTP, and dGTP. Briefly, dUTP pyrophosphatase catalyzes dUTP into dUMP, and thymidylate synthase catalyzes dUMP into dTMP, which is phosphorylated into dTTP. In this process, thymidylate synthase activity requires a single carbon unit from N5,N10-methylene tetrahydrofolate. N5,N10-methylene tetrahydrofolate is in turn produced by the enzyme, serine hydroxymethyl transferase (SHMT), which transfers a single carbon unit from serine onto the carrier molecule, tetrahydrofolate, to produce glycine and N5,N10-methylene tetrahydrofolate. Ultimately, SHMT activity contributes to the synthesis of glycine, purines, thymine nucleotides, and SAM, the methyl donor for many methylation reactions in the cell (one-carbon metabolism reviewed in [65]).

As the gate-keeper of dNTP metabolism, RNR activity is regulated at multiple levels, including transcription, protein modification, and allosteric regulation [42]. In particular, RNR is allosterically regulated by feedback from its downstream products, dNTPs (Figure IB,C). dTTP, dGTP, dATP, and ATP bind to the specificity-site of RNR and regulate the specificity of the enzyme by tuning the affinity towards the four NDP substrates. dATP inhibits the overall activity of RNR by binding to the activity-site, which promotes the formation of the enzymatically inactive hexamers (${\text{α}}_6$) in eukaryotic RNR [66]. ATP competes with dATP for binding at the activity-site, and also induces the formation of ${\text{α}}_6$. However, the ATP bound $\text{α}$ subunits can form active quaternary conformations with the $\text{β}$ subunit, thus promoting RNR activity. Overall, these allosteric regulations are crucial in keeping balanced levels of dNTPs, which is in turn important for high fidelity DNA replication [67,68].

Box 1, Figure I. Regulation of Ribonucleotide Reductase.

(A) Ribonucleotide reductase (RNR) catalyzes the rate limiting step of dNTP synthesis from ribonucleotides. In eukaryotes, the active form of RNR is the ${\alpha }_2{\beta }_2$ tetramer, which catalyzes the substrates ribonucleoside-diphosphates (NDPs) into deoxynucleoside-diphosphates (dNDPs). The radical group essential for catalysis is generated at the $\beta$ subunit and transferred to the $\alpha$ subunit where the catalysis takes place. (B) The pathway of dNTP synthesis includes RNR, NDP kinase, and enzymes involved in the synthesis of dTTP from the uridine species. The extra steps required for dTTP synthesis are abbreviated by the single arrow from dUTP to dTTP for simplicity. The single carbon unit required for the synthesis of dTTP is provided by the activity of enzyme SHMT. The black single (resp. double) arrows indicate irreversible (resp. reversible) chemical reactions. The blue (resp. red) dashed arrows represent the positive (resp. negative) allosteric effects that govern RNR activity. (C) Schematic of the inhibition of RNR activity by dATP. dATP can bind to the ATP-cone motif of the $\alpha$ subunit and induce the formation of enzymatically inactive ${\alpha }_6$.

Interestingly, metabolomic profiling of early embryos revealed that the concentration of dNTPs at the beginning of embryogenesis is enough to synthesize at most 1/3 of the nuclei required by the time of zygotic genome activation (ZGA) [30]. The rate limiting step of dNTP synthesis in all life forms is catalyzed by the enzyme ribonucleotide reductase (RNR), which is maternally deposited in the Drosophila embryo in both the mRNA and protein forms [31,32]. Accordingly, the injection of the RNR inhibitor, hydroxy-urea (HU) [33], into the pre-NC10 embryo leads to checkpoint activation and cell cycle arrest after the 11^th synchronous division, which is rescued by the co-injection of dNTPs [30,34]. Similarly, the one carbon units required for dTTP synthesis during the nuclear cleavage cycles are synthesized on the go through enzyme serine hydroxymethyl transferase [34,35]. Overall, dNTPs are synthesized on the go during early Drosophila embryogenesis through well characterized, canonical metabolic pathways (Figure 2 and Box 1).

RNR activity is self-regulated through the feedback inhibition by dATP, a downstream product of RNR activity (See also Box 1). In pre-NC10 Drosophila embryos, the endogenous dATP concentration allosterically inhibits the enzymatic activity of RNR. As DNA polymerization depletes dATP, RNR is reactivated, providing newly synthesized dNTPs for the later nuclear cleavage cycles (Figure 2). Accordingly, the injection of dATP into freshly fertilized embryos leads to prolonged inhibition of RNR and the failure of the 11th division cycle, similar to the injection of HU. The expression of a feedback insensitive mutant (${\text{RNRL}}^{\text{D68N}}$) through the MTDGal4 construct (${\text{MTD > RNRL}}^{\text{D68N}}$) results in a 10-fold increase in dNTP concentration compared to wild type, while ${\text{RNRL}}^{\text{WT}}$ overexpression leads to a 2-fold increase. Therefore, the dNTP levels in the early embryo are primarily controlled through the negative feedback of RNR by dATP.

Surprisingly, the overexpression of feedback insensitive RNR leads to embryonic lethality with severe anterior defects [30]. Studies addressing the origins of this effect revealed new connections between metabolism and one of the key events of early embryogenesis. To explain this connection, we must first briefly introduce the midblastula transition (MBT). The first 10 nuclear cleavage cycles are dictated entirely by the maternally deposited cell cycle machinery, and progress rapidly at ~10 minutes/cycle. Between NC10 and NC13, the cell cycle checkpoint activation gradually increases the duration to ~20 minutes, while ZGA takes place. From NC14 and beyond, the cell cycles are desynchronized in a spatially stereotyped way, and interphase duration is increased to at least 70 minutes [36]. This crucial transition point of cell cycle control, also called MBT, is common to well-studied developmental systems including fly, frog, and fish embryos (Figure 3). In addition to the numerous factors linked to the gradual lengthening of the cell cycle duration in MBT, such as replication stress, transcription, and changes to the cell cycle regulating protein concentration (MBT in Drosophila embryos reviewed in [37]), recent studies revealed the effect of the metabolic regulation of dNTP levels [34,38].

Figure 3. The effect of dNTP level perturbation on the nuclear cleavage cycle duration.

MBT in Drosophila embryos takes place mainly during NC13. In SHMT mutants, the depletion of dTTP leads to strong check point activation and cell cycle arrest at NC13. In MTD > RNRL^D68N embryos, NC12 and NC13 are shorter in duration, which leads to the insufficient transcriptional output and the failure of the germband extension.

The first phenotype of dNTP overexpression in embryos with genetically deregulated RNR is a modest but statistically significant acceleration of the NC12 and NC13 interphases: ~1 minute for NC12 and ~2.5 minutes for NC13 [38]. This effect is more pronounced for the $\mathbf{\text{α}}$-amanitin induced extra synchronous cell cycle (NC14), whose duration is ~12 minutes faster in embryos with deregulated RNR than in wild type embryos [34]. In wild type embryos, the gradual lengthening of the nuclear cleavage cycles is required for the sufficient accumulation of crucial zygotic gene products [39–41]. In embryos with deregulated RNR, live imaging of zygotic transcription revealed that the shortened duration of the late cleavage cycles leads to strongly reduced gene expression. Furthermore, the protein levels of cellularization-regulating genes, nullo and sry-a, are also reduced. Interestingly, embryos with deregulated RNR undergo normal nuclear cleavages, axial patterning, cellularization, and the first stages of gastrulation. The first morphological defect is noticed in germband extension, during which the ventral midline veers off-course. While the mechanistic links among reduced gene expression and the failure of germband extension remain unresolved, these findings clearly link dNTP synthesis to the temporal control of cell cycles at MBT.

Concluding remarks

The relationship between dNTP levels and the cell cycle has been investigated primarily in yeast and mammalian cell cultures, in which dNTP homeostasis is accomplished through the combination of allosteric, transcriptional, and post translational control of RNR and other proteins of dNTP metabolism [42]. Similar to embryos, administering HU to dividing cells results in the depletion of dNTPs and the cell cycle checkpoint activation [43,44]. Additionally, increasing dNTP concentrations through the manipulation of RNR activity has been shown to accelerate the replication fork progression in yeast [45]. Though it may seem surprising that the cellular concentration of dNTPs has not been evolutionarily optimized for DNA replication speed, increasing dNTP levels can damage DNA replication fidelity. At physiological dNTP levels, the DNA-polymerase stalls after incorporating an erroneous deoxynucleotide, which allows the exonuclease to cleave the incorrect insertion. At excessively high dNTP levels, the DNA-polymerase moves on with replication before the exonuclease can correct mistakes, leading to higher mutation rates [46,47]. In yeast, the feedback mechanisms acting on RNR have been shown to safeguard against the depletion or the excess accumulation of dNTPs, thus ensuring the robust progression of the replication forks and high DNA replication fidelity [44,48]. In animal models, however, the mechanistic links between RNR, dNTP levels, and mutation rate are yet to be elucidated (see Outstanding Questions).

Outstanding Questions.

• What are the major energy consuming cellular processes during embryogenesis? How does the energy budget influence the composition of the embryo’s proteome?

• How does the overproduction of dNTPs lead to embryonic lethality? What are the molecular mechanisms that link high dNTP levels, acceleration of DNA replication, and the failure of germ band extension?

• Why is early embryogenesis the only developmental stage at which the overexpression of RNRL^D68N is lethal?

• Is dNTP homeostasis crucial in embryos of other animal species? What are the relationships among RNR, dNTP concentration, mutation rate, and the animal’s fitness?

Embryogenesis presents an interesting platform to study dNTP metabolism in the context of cell cycle regulation and DNA replication fidelity. Not only are the embryonic cell cycles highly regulated, but extra caution in DNA replication is expected at this stage of development to prevent mutations from spreading to the entire organism. Similar to fruit flies, frogs and sea urchins must also synthesize dNTPs concurrently with DNA polymerization from early embryogenesis. Xenopus oocytes contain dNTPs sufficient for half of that required by the ZGA, which occurs at the 12th division. Sea urchin oocytes store dNTP sufficient for one cell cycle, and ZGA occurs as early as the first cell cycle [29,49–51]. Interestingly, sea urchin embryos injected with HU are able to undergo more nuclear division cycles than allowed using only the maternally deposited dNTPs, possibly by utilizing the dNTP salvage pathway [51] (Box 1). In the surf clam, the expression of the two subunits of RNR are differentially regulated. The $\alpha$ subunit is maternally deposited in the protein form, while the $\beta$ subunit is deposited as mRNA and translated within 20–30 minutes of fertilization [52] (Box 1). Meanwhile, in mouse embryogenesis, the elimination of RNR expression in the heart perturbs the balance among the 4 dNTP species, leads to heart abnormalities, and ultimately to embryonic lethality [53]. Overall, even within the single metabolic pathway related to dNTP synthesis, studies from different animal species hint at the diverse trove of regulatory mechanisms employed during development.

In the spirit of Needham’s vision of chemical embryology [1–3], we examined Drosophila embryogenesis from the metabolic perspective of mass and energy balance. In particular, we reviewed how accounting for the mass balance of dNTPs led to a series of experiments which demonstrated that dNTP homeostasis is required for MBT. Similarly, studies in mouse embryogenesis have found new links between metabolism and epigenetics by tracing the fates of the central carbon metabolism metabolites [11,12]. However, the lessons from physical chemistry extend beyond mass and energy balance laws. For instance, just as biologists continue to refine models of development, physicists recently discovered a new thermodynamic principle that imposes an upper bound on the speed and precision of a biomolecular process given its thermodynamic cost [54–56]. In this light, chemical embryology, which combines physical chemistry with models of development, will continue to provide insights on the design principles of embryogenesis.

Highlights.

• Dynamic coordination of metabolic, genetic, and signaling networks appears to be essential for seamless progression of development.

• Throughout fly embryogenesis, the energy production rate increases linearly in time until hatching, summing up to a total of ~10 mJ.

• Fly embryos synthesize dNTPs “on the go” while undergoing the nuclear cleavage cycles, during which the dNTP concentration is maintained through a negative feedback mechanism.

• Without negative feedback, uncontrolled dNTP production leads to the failure of the deceleration of the nuclear cleavage cycles during the midblastula transition, insufficient gene expression, and embryonic lethality.

Glossary

Spemann Organizer

a group of cells in the Xenopus laevis embryo that establishes the dorsoventral and antero-posterior axes

Nuclear cleavage cycles

the synchronous cell cycles that occur without cytokinesis in the early Drosophila embryo

Heat dissipation rate

the amount of thermal energy transferred per second, in units of watts (Joules per second). In the experimental setup discussed in the text, the calorimeter measures the embryo’s heat dissipation as it is transferred to the surrounding bath

Zygotic genome activation

the initiation of the transcription of zygotic genes after fertilization

Cellularization

after the 13^th nuclear cleavage cycle, cell membranes are created around each nucleus

$\text{α}$-amanitin

an RNA polymerase II inhibiting peptide that induces an extra nuclear cleavage cycle when injected into early Drosophila embryos

Germband extension

the morphological process during which the ventral germband doubles in length along the antero-posterior axis