Metabolic cofactors act as initiating substrates for primase and affect replication primer processing

Recently a new, non-canonical type of 5’-RNA capping with cellular metabolic cofactors was discovered in bacteria and eukaryotes. This type of capping is performed by RNA polymerases, the main enzymes of transcription, which initiate RNA synthesis with cofactors. Here we show that primase, the enzyme of replication which primes synthesis of DNA by making short RNA primers, initiates synthesis of replication primers using the number of metabolic cofactors. Primase DnaG of E. coli starts synthesis of RNA with cofactors NAD+/NADH, FAD and DP-CoA in vitro. This activity does not affect primase specificity of initiation. ppGpp, the global starvation response regulator, strongly inhibits the non-canonical initiation by DnaG. Amino acid residues of a “basic ridge” define the binding determinant of cofactors to DnaG. Likewise, the human primase catalytic subunit P49 can use modified substrate m7GTP for synthesis initiation. For correct genome duplication, the RNA primer needs to be removed and Okazaki fragments ligated. We show that the efficiency of primer processing by DNA polymerase I is strongly affected by cofactors on the 5’-end of RNA. Overall our results suggest that cofactors at the 5’ position of the primer influence regulation of initiation and Okazaki fragments processing. Visual abstract A. Non-canonical capping of RNA by RNA polymerase. RNA polymerase uses cellular cofactor as initiating substrate for RNA synthesis, instead of NTP. Then RNA chain grows, while cofactor remains attached and serves as cap. B. Proposed mechanism of non-canonical initiation of RNA primer synthesis by DnaG primase during replication. DnaG primase initiates synthesis of the primer for DNA replication using cellular cofactor. Primer stays annealed with the DNA template. DNApolI encounters cofactor, which affects the removal of primer.


Introduction
This work was inspired by the recent discovery of the non-canonical RNA capping phenomenon. Many RNA species in bacteria and eukaryotes bear metabolic adenine-containing cofactors at their 5'-end; NAD + (nicotinamide adenine dinucleotide) and DP-CoA (dephospho-coenzyme A) and FAD (flavin adenine dinucleotide), as well as cell wall precursors UDP-GlcNAc (uridine 5'-diphospho-N-acetylglucosamine) and UDP-Glc (uridine 5'-diphosphate glucose) (1, 2). Unlike the classic cap m 7 G, noncanonical caps are installed by the main enzyme of transcription, RNA polymerase (RNAP) (3). It happens during initiation of transcription in a template-dependent manner -ADP-containing cofactors are incorporated at promoters with +1A start sites, i.e. promoters dictating ATP as the initiating substrate (3,4), UDP-containing cell wall precursors -on +1U promoters (3).
By analogy with the classic cap, there are decapping enzymes for non-canonical caps, in E. coli it is NudC (NADH pyrophosphohydrolase of the NUDIX family). NudC processes NADylated RNAs into a monophosphorylated species that are quickly degraded in the cell (5).
Overall it appears that unrelated multi-subunit eukaryotic and bacterial RNAPs as well as the single-subunit RNAPs of mitochondria and viruses can utilize non-canonical initiating substrates (NCISs) and perform RNA capping (6). Another DNA-dependent enzyme initiating de novo synthesis of RNA is primase (DnaG in bacteria) which makes primers for replication and present in all organisms. It is not structurally related to either single subunit e.g. mitochondrial RNAP or multi subunit bacteria or eukaryotic RNAPs.
In E. coli, DnaG recognises a consensus GTC motif and makes a 10-12 nucleotides long RNA primer. Primase acts in concert with other replication proteins. Primase requires DnaB helicase to start synthesis on double-stranded DNA (7). Primase is then displaced and the primer is elongated by the DNA polymerase III to produce the RNA/DNA polynucleotide of a leading strand or an Okazaki fragment of a lagging strand (8). RNA primers need to be removed via the combined actions of DNA polymerase I (PolI) and/or RNaseH before the Okazaki fragments and leading strand are ligated to complete genome replication. Thus, synthesis of an RNA primer by primase is believed to be a rate limiting step of replication (9), tightly coupled to other steps of the replication process. Primase plays a key role during assembly of the replisome (10), regulation of replication elongation and Okazaki fragments length (11,12) in both bacterial and eukaryotic systems. In bacteria, DnaG primase plays a part in a global concerted response to starvation via starvation alarmone ppGpp which inhibits primer synthesis in nutrient limited conditions (13).
Here we show that E. coli primase DnaG starts synthesis of a replication primer using a number of ADP-containing cofactors in vitro, including NAD+/ NADH, FAD and DP-coA. This reaction requires amino acid residues of the DnaG "basic ridge" region, and is inhibited by global starvation alarmone ppGpp. We also show that cofactors on the 5'-end of RNA specifically and differentially affect processing of this RNA by DNA polI. Our data suggest that 5'-cofactors influence initiation efficiency and the rate of processing of replication intermediates.

DP-coA.
DnaG primase functions as a low-processive RNA polymerase able to start de novo RNA synthesis on a DNA. We wanted to test if primase can initiate synthesis using ADPcontaining metabolic cofactors, (structures on Fig. 1A), by analogy with other polymerases.
We used a general priming system, i.e. minimal system in the absence of singlestrand DNA-binding protein. This set-up requires only DnaG for RNA primer synthesis on the short single-stranded DNA template containing GTC recognition motif (scheme on We found that DnaG makes a 13nt long RNA product in a subset of NTPs using either ATP, NAD+, NADH, FAD and DP-coA (Fig. 1B). To avoid confusion and for simplicity, we will refer to the length of RNA with conventional substrates even though cofactors are dinucleotides. Notably, DnaG incorporates NAD+ much less efficiently than NADH, in contrast to other RNAPs, which do not discriminate between NAD+ and NADH (4). We found that affinity to the non-canonical initiation substrates is comparable with their physiological concentrations. Michaelis constants we measured for ATP, NADH and FAD as initiating substrates, were 46.6, 109 and 390 µM, correspondingly. In actively growing in rich media E. coli cells, concentrations of ATP, NADH and FAD are 9.6 mM, 100-1000 µM and 200 µM, correspondingly (14,15). NADH capped primers were not susceptible to decapping by NudC, unless primase was washed away with high salt buffer, like in the experiment shown on Figure   1C. This result is in agreement with the view that full length primer stays bound to the DNA template and in complex with primase (12), in our case even if the primer contains extra moiety on its 5'-end. Replication primer synthesis on single-stranded DNA template, scheme above, with ATP, NAD + , NADH, FAD, and DP-coA as initiation substrates. C. RNA primer with 5'-NADH is susceptible for cleavage by NudC nuclease after DnaG was removed from the complex with high salt wash. Note that absence of ATP from the reaction results in an increased amount of non-specific product, which is not susceptible to NudC in lanes 5, 6. We assume that this band results from initiation with GTP present in the reaction.
Interestingly, after primase was removed, 5'-NADH of "naked" RNA primer annealed to DNA was susceptible to NudC, in contrast to the published result with 5'-NADH of RNA annealed to RNA (5). Therefore, NudC may have a limited window of opportunity to remove the cap from the primer during active replication and participate in processing of replication intermediates.  A template at -1 position relative to the start site (scheme on Fig. 2). These contacts may affect initiation specificity by DnaG which recognises GTC motif on a template strand.

The central T codes for A and G is a template -1 base. To test if cofactors incorporation
depends on the nature of -1 base, we tested synthesis of RNA13 on templates with -1 changed to three other bases. This experiment we performed with initiation substrates concentrations roughly in the region of their corresponding K m s (50 µM for ATP and 500 µM for cofactors). We found that primase preferred purines in this position, and the least preferred base is C (Fig. 2). Initiation with cofactors did not change these preferences, suggesting that cofactors do not make specific contacts with -1 base of the template. This result also implies that cofactors as substrates do not change specificity of DnaG initiation and do not lead to spurious initiation. Synthesis efficiency of RNA13 started with ATP or NADH, FAD or DP-CoA on DNA templates with C, A or T at position -1 was compared to consensus -1G template. Relative efficiency of synthesis is shown in percentage from -1G template, error bars reflect standard deviations from three independent experiments.

Initiation of RNA synthesis with cofactors requires basic ridge amino acid residues of DnaG
Since cofactors do not make strong contacts with DNA, they most probably contact DnaG protein itself. A number of amino acid residues, including several in a "basic ridge", were implicated in initiation nucleotide binding based on sequence conservation amongst primases and structural information for the Staphylococcus aureus primase (13).
We tested synthesis of a primer by DnaG with amino acid substitutions, K229A, Y230A K241A (basic ridge) and D309A (participating in metal chelation), which were all previously shown to influence initiating substrate incorporation (13,16), in the presence of DnaB. We found that "basic ridge" substitutions K229A, Y230A, and to some extent K241A specifically inhibited capping with NADH and FAD (Fig. 3), in contrast with repli cati on pri m er D309A. We assumed that NMN and FMN moieties of the corresponding cofactors might make contacts with these amino acid residues either during binding of the initiating substrates or during the very first step of RNA synthesis.

Eukaryotic primase catalytic subunit P49 uses modified initiating substrate
In addition to prokaryotic system, we wanted to test if the human primase catalytic subunit P49 utilises ADP containing cofactors. We analysed the formation of the first dinucleotide product by this enzyme on the single stranded DNA template (sequence on Fig. 6). We were unable to make this enzyme to start synthesis with ATP at a specific start site, and initiation with GTP was very inefficient on this template (lane 2). Yet, P49 efficiently produced a dinucleotide using m 7 GTP as initiating and UTP as the substrate for the second position (lane 4). This ready incorporation of a modified substrate hints at P49 general low fidelity and propensity to non-canonical initiation. Therefore, our result suggests that P49 might utilise a variety of non-canonical substrates with possible consequences for initiation kinetics and elongation to full length primer (18).

5'
polI ATP-RNA12  Figure 6. Human primase catalytic subunit P49 efficiently utilises m 7 GTP as initiating substrate. Scheme of the singlestranded DNA template is above the gel image. On the gel products of first step of synthesis with different initiating substrates and UTP as extending nucleotide are loaded.

Discussion
Here we showed that DnaG primase initiates synthesis of a replication primer with ADP-containing cellular cofactors. This ability is reminiscent of that of RNA  (9). We showed that the rate of processing of replication primer by PolI is affected specifically by 5'-cofactors. NADH greatly stimulates, and FAD and DP-CoA inhibit the processing. We also found that decapping nuclease NudC can remove 5'cofactors from RNA even if RNA is annealed to DNA. Therefore, hypothetically, NudC may assist removal of primers aberrantly left unprocessed in the cell.
The propensity of the human primase catalytic subunit to incorporate efficiently a modified substrate, an analogue of the classic cap m 7 GTP, suggests the probability of non-canonical initiating of replication in eukaryotes.  (3). We show that replication may also be affected by cofactor incorporation. Currently, more potential nucleotide analogues, including dinucleotide polyphosphates incorporated into 5' position of cellular RNAs are being discovered in both kingdoms (2). The role of these emerging substrates potentially extends to replication regulation.
Our results strongly suggest that cofactor initiation of replication happens in vivo, and future research would confirm this. At present we were unable to detect presence of cofactors on RNA primers in vivo, most probably due to the transient nature of a replication primer, relatively low number of primers per cell, and the sensitivity limitation of the methods we used. Nevertheless, we keep trying.
Media and selection E. coli were always grown in Luria bertani (LB) medium (broth or agar). In presence of pET28a Kanamycin (50 mg/L) was given to select for plasmid, pCA24N was selected with Chloramphenicol (34 mg/L).

Cloning of WT and mutant DnaG
Gibson Assembly and QuickChange protocolls were used according to manufacturer's specifications and transformed into E. coli DH5α cells. After sequencing, the plasmid was retransformed into E. coli T7 overexpression strain.

Protein purification
DnaG wt and mutants were grown in 1 L LB with Kanamycin to an OD600= 0.5 at 37° C before induction with IPTG (1 mM) and continued growth at room temperature overnight. Cells were pelleted and stored at -80° C until lysis using sonication in Grinding Buffer (20 mM Tris-HCl pH 7.9, 200 mM NaCl, 5% Glycerol) and Ni-column