The role of dNTP metabolites in control of the embryonic cell cycle

ABSTRACT

Deoxyribonucleotide metabolites (dNTPs) are the substrates for DNA synthesis. It has been proposed that their availability influences the progression of the cell cycle during development and pathological situations such as tumor growth. The mechanism has remained unclear for the link between cell cycle and dNTP levels beyond their role as substrates. Here, we review recent studies concerned with the dynamics of dNTP levels in early embryos and the role of DNA replication checkpoint as a sensor of dNTP levels.

1. Introduction

Metazoan eggs are usually bigger than standard somatic cells and contain maternal material for the first steps in embryonic development. The egg material includes substrates for energy production and metabolites for synthesis of polymers, such as proteins, RNAs and DNA, besides the maternal determinants for pattern formation and cellular machinery. The dynamics of metabolites in embryogenesis has been little investigated since appropriate and simple assays were lacking. However, it is clear that some of these metabolites, at least, may have regulatory functions in development. For example, the availability of the four deoxyribonucleotides (dNTPs) is essential for DNA synthesis and thus progression of the cell cycle. Especially in early embryonic development, when cell cycles are unusually fast, the demand on dNTPs is high and may exceed the capacity for de novo dNTP synthesis.

An excellent experimental system to investigate the dynamics of metabolites and their potential link to cell cycle progression is the early Drosophila embryo. These embryos are characterized by extremely fast mode of the cell cycle and DNA replication even when compared to other insects [1]. With S phase lengths in the range of 4 min and 13 rounds of synchronous nuclear divisions, the consumption of substrates for DNA synthesis exponentially increases. The dynamics and functions of many maternally provided materials have been studied in the past, such as decay of the maternal RNAs, incorporation of the maternal proteins into nuclear pores and chromatin [2–4]. Metabolite dynamics had received relatively less attention, so far.

Here, we review the current advances in the link of dNTP metabolites and embryonic cell cycle timing from a Drosophila perspective [5–7], and discuss these findings in the context of reports from other organisms. We provide an overview of metabolic pathways in Drosophila involved in dNTP biosynthesis, the dynamics of dNTPs and consequences of their misregulation in early embryonic development. Lastly, we present insights into the mechanism of how the cell cycle responses to changed levels of dNTP.

2. Early embryonic cell cycle and dynamics of dNTP metabolites

Drosophila embryogenesis starts with 13 rapid and synchronous nuclear cleavage cycles, which are characterized by extremely short S-phases [8,9]. Lacking cytokinesis, these division cycles are referred to as nuclear cycles. During the nuclear cycles, the complete genome of 2 × 180 Mb [10] is duplicated within a few minutes, giving a rate of about 1.5 million dNTP per second and nucleus. The first two nuclear cycles (NC1 and 2) last about 25 min, and NC3 to NC8 last about 40 min, corresponding to about 8.6 min per nuclear cycle [8]. S-phases proceed over about only 4 min in these early cycles. This extraordinary speed is achieved by a high number of origins of replication [11,12]. NC1 to 8 are assigned to the pre-blastoderm stage, when the nuclear divisions occur in the interior of the embryo. During interphase 8 and 9, the nuclei migrate from the interior toward the periphery, finally reaching the cortex of the embryo and forming the syncytial blastoderm. At the cortex the nuclei arrange into an ordered array by internuclear and cortex-nuclear interactions mediated by the microtubule and actin cytoskeleton [13]. Nuclear cycles begin to gradually slowdown, from approximately 10 min in NC11 to 21 min in NC13 [8]. The number of cortical nuclei reaches about 6000 after the end of nuclear cycles. The deviation from 2¹³ = 8192 is due to the loss of a few nuclei during cortical migration. The lost nuclei remain in the yolk, undergo endocycles and differentiate to the polyploid and extraembryonic yolk nuclei. As a feature of the transition from nuclear division cycles to a proper cell cycle, the length of S-phase prominently extends to 50 min in cycle 14 [14]. Furthermore, this transition is associated with two striking changes in the developmental program: (1) activation of the zygotic/embryonic genome and (2) cellularization. Little transcription is detected in early embryos. Only in NC10 and later, clear signs of active RNA polymerase and zygotic transcription are observed. Full levels are reached only in cycle 14. Zygotic genome activation triggers the change from syncytial to cellular development, in that the plasma membrane ingresses between the nuclei, encloses them into individual cells and thus yielding a polarized, single-layered columnar epithelium.

Corresponding transitions from fast to slow cell cycles are also characteristic for vertebrates with large eggs, such as Xenopus and zebrafish. Xenopus embryos undergo 12 fast cleavage cycles with lengths of approximately 30 min each. During the next three cycles (13 to 15), the cell cycle decelerates and eventually loses synchronicity [15,16]. Zebrafish embryos pass through 9 fast and synchronous cycles, of about 15 min each. Afterward, the cell cycle decelerates (cycles 10 and 11) and becomes asynchronous [17–19].

Despite the brevity of the S-phases in Drosophila and vertebrate cleavage cycles and the exponentially increasing number of nuclei, the genome is completely duplicated. The exponentially increasing demand on dNTPs is principally satisfied from two sources. As in all other cells, dNTPs are synthesized de novo from precursors of the nucleotide synthesis pathways by the NDP ribonucleotide reductase (RNR) and other enzymes. In addition and specific for the zygote and early embryo, a maternally supplied dNTP pool provides the substrates for DNA synthesis [5–7,20].

During oogenesis, a wide range of nutrients and materials is loaded into the egg, which will control and contribute to various processes of embryonic development. Concomitant with the rapid embryonic nuclear divisions and prior to the zygotic expression, this maternal pool is consumed. Embryos remain their volumes during early nuclear cleavage divisions, thus the continuous DNA replication leads to a dilution of the components of the pool [21,22]. The concentrations, composition, and relevance of the maternal pool have now been an issue for decades. The clarification of their dynamics and the test of their physiological relevance have been hampered by the lack of quantitative and sensitive assays with high temporal and spatial resolution. Concerning RNAs and partially for proteins, good and reliable assays are available now with sensitivity in the scale of single embryos or even less. The quantitative analysis of metabolites and specifically of dNTPs has been lacking behind the technological development concerning RNAs and proteins. Going back to 1959, the dNTP content of Japanese toad Bufo vulgaris formosus blastula has been determined to 38 pmol of dCTP, 76 pmol of dATP, 32 pmol of dGTP and 66 pmol of dTTP per embryo by using ion-exchange chromatography [23]. These values are far higher than later measurement with Xenopus eggs, yielding 19 pmol of dCTP, 13 pmol of dATP, 12 pmol of dGTP and 12 pmol of dTTP [24]. These concentrations would suffice for the DNA content at the 2500-cell stage, corresponding to 11 rounds of cleavage divisions. Later on, a dATP concentration of 50 µM was measured in extracts of 2 h Xenopus embryos by in vitro DNA synthesis [25]. It is not clear, whether eggs in all species contain large pools of maternal dNTP, as only a very small amount sufficient for one genome equivalent was measured in sea urchin eggs [26].

Recent technological developments with strongly improved sensitivity allow for less starting material and more reliable quantification. By liquid chromatography-tandem mass spectrometry (LC-MS/MS), the absolute concentration of dATP, dCTP and dTTP in staged Drosophila embryos was determined [5,6]. With the latest protocols, the dNTP content can be reliably quantified in samples derived from as few as five embryos [5]. The low amount of starting materials allows for high resolution staging by morphological markers such as the nuclear cycle. In freshly laid wild type embryos (stage 1–2, 0–1 h after fertilization), a maternal pool with 1.2 pmol of dCTP, 0.8 pmol of dATP and 1.2 pmol of dTTP was detected per embryo, which is an amount sufficient for the synthesis of about 2700 diploid genomes, a number which is reached after 12 division cycles [5]. Assuming an egg volume to as 9 × 10⁻³ mm³ [27] and a uniform distribution, the concentration of the dATP pool is estimated to 89 µM, which is within the same range as the 50 µM estimated for Xenopus eggs [25]. The dNTP concentration decreases to about 50–80% during NC11 and NC12 when syncytial blastoderm forms, and furthermore to 0.3–0.5 pmol at the transition from syncytial to cellular blastoderm when the cell cycle and DNA replication enter the slow mode [5]. In addition, haploid mutant embryos, which contain half of the DNA content in each nucleus, have a maternal dNTP pool comparable to diploids but show a delayed decrease of dNTP levels during the nuclear cycles [5]. Given these numbers for the total content of dNTP, the embryonic “net production” can be calculated. This yields an estimated rate of approximately 100 µM per hour [6].

Inhibition of the de novo dNTP synthesis in the early embryo allows revealing the size of the maternal pool. The exhaustion of dNTP leads to an arrest in S-phase. Hydroxyurea (HU) is a potent inhibitor of RNR, which is required for the synthesis of all four types of dNTP [28]. HU application to yeast [29], Escherichia coli (E.coli) [30], and embryos from mouse [31], Xenopus [20,32], zebrafish [18,33] and Drosophila [5,6,34] result in severe cell cycle defects. Specifically, Xenopus embryos incubated with HU cause a small fraction of prolonged S-phases, but frequently prematurely arrest the cell cycle after cleavage division 12 [20,32]. Likewise, in Drosophila embryos, injection of HU leads to a lengthened S-phase [35], and consequent arrest in mitosis 12, which is followed by a mitotic catastrophe [5,6]. In addition, genetic aberrations of enzymes, which are involved in de novo dNTP synthesis, elicit comparable phenotypes as HU treatment. For instance, serine hydroxymethyl transferase (SHMT), a central enzyme feeding precursors into the one-carbon metabolism, is required for the synthesis of dTMP from dUMP (Figure 1). Embryos lacking maternal SHMT undergo one fewer round of nuclear cleavage cycle [36]. These cell cycle arrest phenotypes in HU injected embryos or SHMT mutants are due to the depletion of dNTPs, since they can be “rescued” by co-injection of a mixture of dNTPs or dTTP, respectively [5,6]. In particular, SHMT mutant embryos contain a full maternal pool of dNTPs, but are specifically depleted of dTTP by NC13 [5].

Figure 1.

Metabolic pathways concerning dTTP biosynthesis. Indirect catalyzes that would involve multiple steps are shown with dashed arrows. Names of the enzymes are indicated in blue boxes. The pathway is modified from [50].

Embryos injected with dNTPs usually fail to achieve extra rounds of cleavage cycles. A report in Xenopus claimed that injecting dNTPs could increase the number of synchronous embryonic cell cycles [37]. Later studies showed that injecting dNTPs to 2.5–7.5 folds of their endogenous concentration, did neither change the number of synchronous cleavage cycles [20,32], nor the duration of S-phase [38]. Correspondingly, in Drosophila, injection of all four dNTPs did neither affect the number nor the lengths of the embryonic cell cycle. In fact, the dNTP concentration converged to normal levels 2 h after the injection [6], which may be due to the feedback control of RNR activity by its products. An effective upregulation of dNTP levels was achieved with a RNR mutant, RNRL^D68N, with an impaired feedback inhibition [6,39]. In such embryos, the dNTP levels are approximately 6–25 times higher in blastoderm embryos [5]. Indicating a role in cell cycle control, about 10% of RNRL^D68N embryos exhibited an extra round of nuclear division (NC14) and a faster cycle in all embryos [7].

3. Drosophila metabolic pathways linked to dNTP biosynthesis

The metabolic pathways for nucleotide biosynthesis are conserved in all eukaryotes and many bacteria. RNR catalyzes the reduction of for all four types of dNTPs from ribonucleotide precursors [28,40]. From bacteria to mammals, RNR is evolutionarily conserved and its E.coli homolog is the best characterized [41,42]. Briefly, active RNR contains two nonidentical dimeric subunits α2 and β2, which are encoded by separate genes [41]. α2 contains an allosteric activity site for regulating the general activity of RNRs, an allosteric specificity site for regulating substrate specificity as well as a catalytic site [28], whereas the activity of β2 subunit is sensitive to HU in vitro [43]. The allosteric activity site modulates the overall size of the dNTP pool [44]. Meanwhile, specific NTP or dNTP bind to the allosteric specificity site, to change the preference of RNR for the four NDP substrates. In general, RNR plays a central role in maintenance of the balanced dNTP pools that are required for accurate DNA replication, whose function is regulated on multiple levels, including transcription, protein modification, protein degradation and allosteric regulation [39,42].

In Drosophila, RnrL and RnrS genes encode the large and small subunits of RNR, corresponding to α2 and β2, respectively [45]. RnrS mRNA is maternally supplied. Following maternal RNA degradation by blastoderm in cycle 14, RnrS is present again in late cycle 16 in rapidly proliferating tissues [45]. In embryos from stage 13 onwards, RnrS is involved in G1–S progression and found associated with E2F and PCNA [46]. Its counterpart RnrL is also maternally expressed, and which was identified in an RNAi screen as a melanotic suppressor associated with RPA2 [47]. AhcyL1 and AhcyL2 are Drosophila homologs of S-adenosyl-homocysteine (SAH) hydrolase-like proteins, which contain IRBIT domains and therefore inhibit RNR [48]. Down-regulation of AhcyL1 and AhcyL2 leads to an increased life span but unchanged dNTP levels in adult flies, whereas down-regulation of RnrL alone does not change the life span [49]. As mentioned above, blastoderm embryos expressing the hyperactive RNR mutant RNRL^D68N, contain 6–25 times higher dNTP levels than wild type [5,6]. This change is due to the impaired feedback by dATP, so that the excess dATP levels cannot turn down RNR activity [39].

Apart from de novo synthesis, dNTPs can also be produced via ribonucleotide salvage pathway, by reusing substrates from degradation of DNA. Mammalian salvage pathway requires deoxycytidine kinase (dCK), thymidine kinase 1/2 (TK1/2) and deoxyguanosine kinase (dGK) to phosphorylate deoxyribonucleoside monophosphates (dNMPs) from deoxyribonucleosides (dNs) [50,51]. In Drosophila and other insects, deoxyribonucleoside kinase (dNK) can phosphorylate all of the four dNs with a high turnover, instead of the four specific kinases in mammals [50,52,53]. As a TK2-like enzyme, Drosophila dNK functionally complements TK2-deficient mice and leads to increased dNTP pools [54,55]. However, freshly laid Drosophila embryos contain very low dN levels, suggesting that salvage pathway plays no significant role in early embryonic development [6]. Figure 1 and Table 1 illustrate the biosynthesis of dTTP in Drosophila.

Table 1.

Enzymes involved dTTP biosynthesis in Drosophila.

Enzyme	Abbreviation	Drosophila gene[s)	Genetic position	Key reference
NDP ribonucleotide reductase	RNR	RnrL, RnrS	2L: 31D8, 2R: 48D5	[45]
Nucleoside-diphosphate kinase	NDP kinase	awd, nmdyn	3R: 100D2, X: 12F1/3R: 85E4	[56]
dUTP diphosphatase	dUTPase	dUTPase	2L: 32C1	[57, 58]
dCMP deaminase	-	CG6951	3L: 77A3- 77A4	-
UMP-CMP kinase	-	Dak1	3R: 96F10	[59]
Deoxyribonucleoside kinase	dNK	dNK	3R: 91E2	[53]
Thymidylate synthase	dTMP synthase/TYMS	Ts	2L: 23C3	[60]
Serine hydroxymethyl transferase	SHMT	SHMT	X: 5C7	[36]
Dihydrofolate reductase	DHFR	Dhfr	3R: 89B18	[61]
Glycine decarboxylase	GLDC	CG3999	3R: 86A2	-
dTMP kinase	-	CG5757	2R: 55B4	[62]
dTDP kinase	-	-	-	-
Methylenetetrahydrofolate dehydrogenase 1	MTHFD1	pugilist	3R: 86C4	[63]

Two enzymes, SHMT and glycine decarboxylase (GLDC) feed the most of the one-carbon units into the tetrahydrofolate (THF) cycle. Usually, the cytoplasmic/nuclear and mitochondrial THF/one-carbon pools are separately controlled. The cytosolic-mitochondrial counterparts of the enzymes are encoded as distinct isoforms from the same gene or by distinct genes. Mammals contain two SHMT genes, the cytosolic isozyme gene SHMT1 and the mitochondrial isozyme gene SHMT2 [64]. SHMT2 has a second transcript SHMT2α, which encodes a protein that localizes to the nucleus during S-phase. Except for the nuclear SHMT2α, SHMT1 together with dihydrofolate reductase (DHFR) and dTMP synthase (TYMS) are post-translationally modified by the small ubiquitin-like modifier (SUMO) to translocate to the nuclear lamina [65–67]. Alternatively, methylenetetrahydrofolate dehydrogenase 1 (MTHFD1, encoded by pugilist in Drosophila) catalyzes THF, ATP, and formate to yield 10-formylTHF and further N5,N10-methylentetrahydrofolate (5,10-meTHF) in the cytosol and nucleus [68]. The Drosophila genome contains a single SHMT gene that encodes three transcripts, which are proteolytically derived from a longer precursor with a putative mitochondrial presequence. When the mitochondrial presequence is preserved, E.coli SHMT can functionally substitute its Drosophila homolog, suggesting the protein isoforms properly reach their subcellular localizations [36].

SHMT mutant Drosophila embryos are specifically depleted of dTTP by cycle 13. In addition and surprisingly, the dCTP levels are also reduced [5]. The cell cycle arrests one round later in SHMT embryos than HU treated embryos. This difference may partially be due to the misincorporation of dCTP instead of dTTP, or conversion of dCDP to dCMP and furthermore to dUMP [40,69,70] (Figure 1). SHMT is involved in folate-mediated one-carbon metabolism, together with TYMS (Ts in Drosophila) and DHFR [64] (Figure 1). These enzymes mediate synthesis of purines, thymidylate as well as methionine cycle through the one-carbon donor THF. In particular, SHMT delivers one-carbon unit from serine to glycine and 5,10-meTHF, and TYMS catalyzes the conversion of dUMP and 5,10-meTHF to dTMP and dihydrofolate (DHF), then DHFR catalyzes DHF to generate THF for subsequent thymidylate synthesis cycles [64] (Figure 1, Table 1). The second enzyme beside SHMT generating one-carbon units is GLDC, which cleaves glycine into CO₂ and a methylene unit accepted by THF [71]. The homolog of GLDC in Drosophila (CG3999) has not been studied in functional terms.

Metabolites, nucleotides and their precursors are not able to cross the cell border by themselves from neighboring cells or extracellular matrix. To fulfill the intracellular requirements, metabolites need transporters or channels in the plasma membrane. Likewise, many enzymes that are needed for conversion of nucleotides and other metabolic substrates may be autonomous. Drosophila imaginal discs and ovarian follicle epithelium are good experimental systems to investigate autonomous functions of genes, since clones of mutant cells within a wild type environment can be easily generated (Figure 2). SHMT has been shown to be non-autonomous [36]. This conclusion is based on the fact that SHMT mutant clones within the imaginal discs and somatic follicle cells can grow and pass through the cell cycle. The rescue may be due to uptake of 5,10-meTHF or dTMP across cell membrane from neighboring cells or extracellular fluid. Similarly, oogenesis proceeds as normal, even if the germline is deficient of SHMT [36] (Figure 2). A comparable non-autonomy of metabolites has been reported for “activated sulfate”, PAPS. Somatic follicle cells lacking PAPS can be rescued by wild type germline cells, and a mutant germline can be rescued by wild type follicle cells [72]. Altogether, these non-autonomous functions of enzymes indicate that some metabolites, at least, can be exchanged between tissues or extracellular hemolymph (Figure 2).

Figure 2.

Non-autonomous function of SHMT. (a) Germline cells homozygous for SHMT or deficient for PAPS, can be rescued by surrounding wild type cells, such as the follicle cells. Missing metabolites or precursors are imported into the deficient germline cells and rescue the mutant phenotype, as previously shown for PAPS. A similar mechanism may rescue SHMT mutant germline cells. (b) A non-autonomy of SHMT function was detected in somatic clones, such as in larval imaginal discs. The SHMT deficiency is rescued presumably by uptake of metabolites from neighboring cells or hemolymph.

4. Mechanisms of cell cycle control by dNTP

During Drosophila embryonic nuclear cleavage cycles, activation of Cdk1-cyclin complex promotes entry into mitosis. A key regulator, Twine/Cdc25 phosphatase, activates the Cdk1-cyclin complex during the fast and synchronized nuclear cycles [4,9]. Following mitosis 13 and onset of zygotic transcription, Twine/Cdc25 is destabilized. The loss of Twine/Cdc25 leads to the first gap phase and change of the slow mode of cell cycle. This switch of the cell cycle mode is referred to as cell cycle remodeling.

Zygotic transcription acts as a timer for cell cycle remodeling. Specific zygotic genes, such as frühstart, inhibit mitotic entry through binding to the hydrophobic patch of CyclinA [73–75]. Another zygotic gene tribbles, which also potently inhibits mitotic entry and thus promotes cell cycle remodeling [75], destabilizes Twine/Cdc25 [76,77]. Besides these specific genes, cell cycle remodeling is also triggered by global zygotic expression, and the subsequent activation of replication checkpoint, due to the interference of zygotic transcription and DNA replication [34]. Precocious zygotic transcription is sufficient for precocious cell cycle remodeling [78]. Moreover, when transcription is fully blocked by the RNA polymerase II inhibitor α-amanitin, embryos undergo an extra nuclear cleavage division and arrest in interphase 15 [5,34,79]. Reduction of zygotic transcription by a mutation of the pioneer transcription factor Vielfältig/Zelda leads to one extra round of nuclear cycle in about half of the embryos [5,78,80,81]. Nonetheless, this indicates that additional mechanisms are also involved in timing of the cell cycle remodeling, since the embryonic cell cycle progression is eventually arrested even with the absence of zygotic transcription.

It has recently been shown that some critical metabolites, such as dNTPs, contribute to the timing of cell cycle remodeling to a certain extent [5,6]. In addition to the activation of global zygotic transcription, insufficient supply of dNTPs causes replication stress and triggers the checkpoint activation. Injection of HU to inhibit de novo synthesis of all four dNTPs induces replication stress and lengthens interphase 12 [5,34]. Depletion of dTTP by the absence of SHMT leads to replication stress in interphase 13 [5]. DNA replication stress results in the activation of replication checkpoint. When removing the Checkpoint kinase 1 (Chk1) by mutation of grapes gene, more than half of the SHMT mutant embryos passed through mitosis 13, indicating that cell cycle arrest in SHMT embryos is largely mediated by the DNA replication checkpoint [5]. In SHMT and vielfältig/zelda double mutants with depletion of dTTP and reduced global zygotic transcription, half of the embryos pass through mitosis 13 and show replication stress in lengthened interphase 14. This is consistent with the phenotype of α-amanitin injected SHMT mutant embryos, in which almost all the embryos enter cycle 14 [5]. The prolonged additional cycle in embryos with impaired zygotic transcription, and with both deficient dNTP and impaired zygotic transcription, indicates these two sources additively contribute to the cell cycle remodeling through activation of the replication checkpoint (Figure 3). This mechanism may be independent of induced Twine/Cdc25 degradation, as SHMT mutant embryos arrest in interphase 13 even that Twine is present [36].

Figure 3.

Multiple pathways impinge on cell cycle control in early Drosophila embryos. Zygotic transcription is required for the remodeling of the nuclear cycle to a cell cycle with slow DNA replication and a G2 phase. The onset of zygotic transcription sets the time for cell cycle remodeling. Zygotic transcription inhibits the Cdk1-Cyclin complex in two ways at least: (1) Expression of zygotic mitotic inhibitors (Tribbles, Frühstart), including induced degradation of Twine/Cdc25, and (2) induction of replication stress and consequent DNA checkpoint activation. In parallel, the levels of dNTP contribute to cell cycle regulation via activation of the DNA checkpoint.

The dependence on dNTP levels may vary to a great extent between the states of the cells and among cell types. The fast embryonic cell cycles may have very different requirement for dNTP levels than cultured cells in vitro, for example [82]. It has been proposed that defective dNTP pool halts DNA replication during S-phase in cultured mammalian cells [83,84]. The defective pool may include shortages of dNTP precursors, enzymes or dNTPs themselves, and always leads to replication stress and checkpoint activation [85,86]. Addition of exogenous nucleotide precursors can rescue the dNTP shortage induced cell cycle progression [87–88]. In yeast, transient increase of the dNTP pool to around 30 folds improves the survival following DNA damage, but at the same time leads to higher mutation rates [89]. However, continuous presence of 30 folds higher dNTP concentration arrests cell cycle progression, similar to HU treated low dNTP situation [90,91]. In contrast, a moderately increased dNTP pool (2–10 folds) protects cells from replication stress and accelerates the cell cycle [92]. This contradiction may be explained by the recent finding that the high speed of a progressing fork can induce DNA replication stress and slowdown the cell cycle [93]. On the other hand, the dNTP concentrations in yeast are controlled by multiple genes, and some of them are associated with mitotic entry or replication stress response [94,95], raising the complexity of interpretation compared with cultured cells. Moreover, it can be speculated that the embryonic cell cycle of metazoans controlled by dNTP may be underlying by further yet unknown mechanisms.

5. Concluding remarks

Besides their roles in development, dNTP levels appear to be important in proliferation control of tumor cells [96]. Genetic associations have been identified between enzymes in dNTP metabolism and melanoma [97], lung cancer [98], cervical cancer [99] and breast cancer [100]. Potential genetic lesions and target genes have been found to be crucial in dNTP induced tumorigenesis. Genetic alterations in specific enzymes involved in dNTP biosynthesis pathways are associated with specific cancers: Aberrant GLDC expression due to an amplified GLDC locus correlates with cell cycle control in multiple cancer types [71]; expression levels of SHMT are associated with vascularized tumor region and B-cell lymphoma [101,102]; the binding of TK and RNR is required for preventing breast cancer cells [103]; and SAMHD has tumor suppressive properties [104,105]. Apart from these biosynthesis enzymes, altering dNTP levels often induces changes in other regulatory correspondences, and subsequent genome instability [99,106], DNA damage [107], DNA replication fidelity [108] and dNTP imbalance caused misincorporation [109,110]. Therefore, a profound understanding of dNTP functions under physiological conditions is needed.

Embryonic cell cycles constitute an excellent, highly reproducible and tractable model to investigate physiological functions and dynamics of dNTP levels beyond their basic biochemical function as substrates for DNA synthesis. In Drosophila embryos, the cell cycle is controlled by multiple mechanisms in a developmental scenario. In brief, a proper maternal dNTP pool supports early rounds of cell cycle progression, which go beyond the capacity of de novo synthesis. Embryos with an excessive maternal dNTP and deregulated de novo synthesis have developmental abnormalities such as abnormal zygotic expression [7]. Whereas insufficient capacity for de novo dNTP synthesis triggers checkpoint activation and thus precocious cell cycle remodeling [5]. With the help of new assays, a higher sensitivity of dNTP measurement is reached [111,112]. These advances may help us in studying dynamic dNTP functions in development with higher spatiotemporal resolution.

Funding Statement

BL was in part supported by China Scholarship Council. The work was in part supported by the Deutsche Forschungsgemeinschaft [DFG GR1945/3-1, FOR1756 GR1945/6-2 and equipment grant INST1525/16-1 FUGG].

Disclosure statement

The authors declare no competing interests.