Arresting chromosome replication upon energy starvation in Escherichia coli

Most organisms possess several cell cycle checkpoints to preserve genome stability in periods of stress. Upon starvation, the absence of chromosomal duplication in the bacterium Escherichia coli is ensured by holding off commencement of replication. During normal growth, accumulation of the initiator protein DnaA along with cell cycle changes in its activity, ensure that DNA replication starts only once per cell cycle. Upon nutrient starvation, the prevailing model is that an arrest in DnaA protein synthesis is responsible for the absence of initiation. Recent indications now suggest that DnaA degradation may also play a role. Here we comment on the implications of this potential new layer of regulation.


Introduction
Nutrient availability affects the cell cycle of all organisms. In Eukaryotes, checkpoints known as 'Restriction' or 'Start' prolong the G1 phase following nutritional stress until a favorable cellular energy state is reached (Johnson and Skotheim 2013;Solaki and Ewald 2018). In the well-studied prokaryotic organism Escherichia coli, no such checkpoints have been identified. Instead, accumulation of the initiator protein DnaA, whose expression is coupled to mass increase, is thought to play a determining role in deciding when DNA synthesis starts (Lobner-Olesen et al. 1989;Si et al. 2017).

Initiation of DNA replication depends on accumulation and activation of DnaA
DnaA is a multi-domain protein that binds the origin of replication and loads the replicative helicase (Hansen and Atlung 2018;Katayama et al. 2017;Leonard and Grimwade 2015;Riber et al. 2016). Like other initiator proteins, it possess an AAA + domain with ATPase activity that determines its ability to form a nucleoprotein structure on the origin and to load the helicase; the initiator being active when bound to ATP while inactive in its ADP bound form. Because DnaA has an equally high affinity for ATP and ADP and because ATP is about seven times more abundant in the cell than ADP during normal growth, it is assumed that newly synthesized DnaA mainly binds ATP. In other words, the energy state of the cell could influence the ratio of active to inactive initiator. Thus it is expected that overproduction of DnaA will increase the amount of DnaA ATP relative to DnaA ADP . Accordingly, overproduction of a mutant DnaA that binds ATP and ADP equally well but has lost its ATPase activity (DnaAR334A), results in an increase of DnaA ATP from ~ 10% to ~ 60% and an enormous increase in initiation frequency (Nishida et al. 2002). Contrary to the expectation, overproduction of wild-type DnaA does not increase or even reduce the ratio DnaA ATP /DnaA ADP (Kurokawa et al. 1999;Nishida et al. 2002) and has a relatively minor effect on the initiation frequency (Nishida et al. 2002;Riber et al. 2006). The fact that an accumulation of DnaA ATP alone is much more potent in starting DNA replication than an accumulation of DnaA ATP together with DnaA ADP indicates that DnaA ADP has an inhibitory function on initiation. Thus, DnaA needs to be both in sufficient amount and predominantly in active form (high DnaA ATP /DnaA ADP ratio) to start DNA replication (Fig. 1). DnaA is predominantly found in the ADP bound form at the population level (~ 70%) (Kato and Katayama 2001). This is because the nucleotide bound state of DnaA is determined by a balance between control mechanisms.

Regulation of DnaA activity
On one hand, there is Regulatory Inactivation of DnaA (RIDA) (Kato and Katayama 2001) and datA-dependent DnaAATP hydrolysis (DDAH) (Kasho and Katayama 2013) that stimulate DnaA ATPase activity to reduce the DnaA ATP /DnaA ADP ratio. On the other hand, the rejuvenation of DnaA ADP to DnaA ATP at DnaA Rejuvenating Sequences DARS1 and DARS2 serves to increase the DnaA ATP /DnaA ADP ratio by promoting the dissociation of ADP from DnaA (Fujimitsu et al. 2009); the Apo-DnaA produced binds preferentially ATP (Fig. 1). These regulatory mechanisms act in concert during the cell cycle to ensure that DnaA ATP accumulates pre-initiation and is subsequently converted to DnaA ADP post-initiation. Mechanistically, DDAH at the datA locus and rejuvenation at DARS sites distinguish themselves by the type of DnaA-DNA complex formed and the auxiliary proteins required to function. DARS1 is constitutively active and does not require helper proteins while IHF is required for DDAH and DARS2. Rejuvenation at DARS2 rejuvenation is also depends on Fis function and is expected to be growth phase regulated (Frimodt-Moller et al. 2016;Inoue et al. 2016;Kasho et al. 2014) because Fis is abundant in exponential phase but absent during stationary phase. RIDA, is orchestrated by the Hda protein interacting with the DNA loaded β-clamps. During replication, β-clamps tether the DNA polymerase to the template DNA. However, numerous β-clamps are left loaded on double stranded DNA behind the replication forks for several minutes (Moolman et al. 2014). These DNAbound β-clamps act as a platform for several activities such as DNA repair and in the case of RIDA serve to complex DnaA with the protein HdA resulting in activation of DnaA ATPase. RIDA is expected to be less active in absence of DNA replication where few β-clamps are left loaded on DNA (Katayama et al. 1998;Kurokawa et al. 1999). Because RIDA is the predominant process lowering the DnaA ATP /DnaA ADP ratio, we speculate that DnaA becomes more ATP-bound during stationary phase and energy starvation (Fig. 1). In cells where only DDAH and DARS1 are functional, the DnaA ATP /DnaA ADP ratio increases (Fujimitsu et al. 2009 is slowly converted into DnaA ATP in a RIDA-deficient cell when protein synthesis is blocked by chloramphenicol treatment (Fujimitsu et al. 2009;Kurokawa et al. 1999).

DnaA synthesis arrest upon starvation
Upon energy starvation, DNA replication stops and it was assumed that a general reduction in protein synthesis (Holm et al. 2010), including DnaA, is responsible for this arrest (Fig. 2)  During short-term energy starvation (lower panel), DnaA does not accumu-late to sufficient level and DnaA ATP /DnaA ADP is low. During longterm energy starvation, we propose that DnaA ATP /DnaA ADP becomes sufficiently high to allow for replication initiation despite of fewer DnaA molecules, unless DnaA level is lowered by degradation sufficient to promote a new round of replication. The DNA replication arrest is specific to the initiation step, with ongoing replication allowed to proceed to completion. This is also the case when DnaA expression is blocked by treatment with antibiotics that arrest protein synthesis such as chloramphenicol and rifampicin or during the stringent response when the alarmone (p)ppGpp is induced (Chiaramello and Zyskind 1990;Lark 1972;Schreiber et al. 1995;Skarstad et al. 1986). This is corroborated by the fact that the hyperactive DnaA mutant DnaAcos continues to initiate DNA replication in the presence of chloramphenicol (Kellenberger-Gujer et al. 1978) and by the observation that overexpression of dnaA bypasses the DNA replication arrest seen in (p)ppGppinduced cells (Riber and Lobner-Olesen 2020).

DnaA degradation during ATP starvation
Recently, a manipulation of the cellular ATP level indicates that additional regulation mechanism exist. Normally, DnaA is extremely stable during steady state growth, with protein half-life exceeding 24 h (Atlung and Hansen 1999;Torheim et al. 2000). It is shown now that depletion of ATP results in a replication initiation arrest as expected, but surprisingly, this is accompanied by degradation of about 30% of the DnaA proteins after 1-hour ATP starvation (Charbon et al. 2021). The nucleotide binding of DnaA does not affect DnaA stability during energy starvation (Charbon et al. 2021). However, the mechanism responsible for the degradation remains elusive and neither of the ATP-dependent proteases Lon or ClpP, implicated in the initiator proteolysis in the bacterium Caulobacter crescentus, are involved. Following carbon starvation, a similar DNA synthesis arrest is observed despite the fact that DnaA is stable during the initial stages of starvation where ATP/ADP level is remains unchanged. Nonetheless, following long-term carbon starvation, the ATP/ADP level falls (Chapman et al. 1971), and DnaA is degraded (Charbon et al. 2021).

DnaA degradation during stringent response
Although DnaA was originally described as stable or even more abundant in stationary phase (Ali Azam et al. 1999;Sakakibara and Yuasa 1982;Sekimizu et al. 1988), recent work indicates that DnaA is degraded during the stringent response (increased levels of (p)ppGpp) occurring upon entry into stationary phase, or upon amino acid starvation (Gross and Konieczny 2020). Indeed, the amount of DnaA per cell decreases over time following induction of (p)ppGpp, a decrease that is not observed in a lon or ppk mutant (polyphosphate synthesis). In vitro, Lon specifically degrades DnaA ADP in the presence of polyphosphate (PolyP) (Gross and Konieczny 2020). Thus it is proposed that accumulation of PolyP acts as a starvation signal to degrade DnaA and prevent initiation of DnaA replication. Note that the degradation of DnaA apparently depends on how (p)ppGpp is induced as it is not observed elsewhere (Riber and Lobner-Olesen 2020).

A new regulatory mechanism?
Because dnaA expression is reduced or arrested during energy starvation and the stringent response respectively, future work should address the need for DnaA degradation as extra layer of regulation. For example, for cells entering carbon starvation, the DNA replication block does not require DnaA degradation (Charbon et al. 2021). In this case, cessation of de novo DnaA synthesis is sufficient to arrest initiation and degradation of DnaA comes only much later. The question therefore remains why DnaA degradation in ATP depleted or (p)ppGpp accumulating cell is at all necessary. One possibility is that DnaA degradation is incidental; i.e., DnaA instability may be a mere consequence of a protein degradation in general. This explains why mutants in proteolytic pathways have not been isolated in genetic screens for suppression of DnaA hyperactivity (Charbon et al. 2018). On the other hand, although the block in expression of DnaA is a sufficient short-term solution, we suggest that it becomes ineffective if the DnaA ATP /DnaA ADP balance increases over time (Fig. 2). We speculate that upon long-term energy starvation during which RIDA and DARS2 are inactive, DnaA ADP is converted into DnaA ATP overtime, thereby lowering the number of DnaA molecules needed to trigger initiation of replication (Fig. 2). Thus, DnaA is capable to re-initiate DNA replication despite being at a level normally deemed insufficient. This is in accordance with the observation that a hyperactive DnaA protein mimicking DnaA ATP can continue to initiate new rounds of replication despite protein synthesis arrest (Kellenberger-Gujer et al. 1978). Degradation of DnaA in general during ATP depletion (Charbon et al. 2021) could lower the cellular amount of protein and prevent this (Fig. 2). The observation that specifically DnaA ADP is degraded during PolyP accumulation (Gross and Konieczny 2020) is counterintuitive, as this would increase the DnaA ATP /DnaA ADP ratio and theoretically lower the amount of DnaA needed to start replication; fewer DnaA molecules but more active. Since DnaA ADP is the most abundant species, accounting for 70% of total DnaA, DnaA ADP proteolysis could provide a fast way to degrade the majority of DnaA molecules. However this is not supported by the modest ~ 20% reduction in total DnaA observed after 30-min amino acid starvation (Gross and Konieczny 2020). On the other hand, degrading the inactive form of DnaA could be a way to quickly restart DNA replication following starvation. Post starvation, cells would be left with only DnaA ATP and therefore fewer DnaA molecules would be required to re-initiate. Finally, we do not exclude that the role of DnaA degradation is to control transcriptional activity (Messer and Weigel 1997) instead of or in addition to its initiator activity. Indeed, DnaA overproduction during late phase has been reported to induce the transcription of polA gene encoding DNA polymerase I in a manner independent of the late phase sigma factor RpoS (Quinones, et al. 1997). Thus, DnaA transcriptionial activity could clash with other master regulators specialized in promoting survival during starvation, justifying DnaA degradation.
Authors' contributions GC, JFM and ALO drafted the manuscript.
Funding This research was funded by grants from the Danish National Research Foundation (DNRF120), from the Novo Challenge Center for Peptide-Based Antibiotics NNF16OC0021700 (Cepan), and from the Villum Foundation through the Villum Experiment Programme, Grant no. 35894.

Data availability Not relevant.
Code availability Not relevant.

Conflicts of interest There are no conflicts of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.