Robust helicase loading mechanism underlies diverse architectures of bacterial replication origins

Abstract

The Escherichia coli chromosomal origin, oriC, contains a duplex-unwinding element (DUE) flanked by two clusters of the initiator DnaA-binding sites (DnaA boxes). ATP-DnaA oligomerizes to construct Left- and Right–DnaA subcomplexes on these clusters. These complexes coordinately promote DUE unwinding and DnaB helicase loading onto the unwound region for replication initiation. However, despite the strong conservation of DnaA and DnaB across bacterial species, oriC architectures vary largely. Remarkably, bipartite and bilateral configurations contain the two DnaA box-clusters separated by an insertion of the dnaA gene or positioned on both sides of the DUE, respectively. Here, we demonstrate that E. coli oriC variants mimicking these configurations remain fundamentally functional. Notably, DnaB loading activity of these variants showed increased dependence on single-stranded DUE binding by the relocated DnaA subcomplex and on the AT cluster upstream of the DUE, reflecting reduced stability of DUE unwinding. Consistent with these in vitro results, these variants supported cellular replication initiation with only slow cell growth rates. Together, these findings reveal remarkable architectural flexibility, mechanistic robustness, and a conserved mechanism in oriC. Also, those highlight unwinding stability as a key determinant of oriC adaptation to rapid cell growth, providing insight into the evolution of oriC architecture.

Graphical Abstract

Graphical Abstract.

Introduction

Bacterial chromosomal replication initiates by multiple factors including the conserved key proteins of the initiator DnaA and replicative helicase DnaB. The origin region (oriC) directs construction of highly organized DnaA complexes through a number of DnaA-specific binding sequences, DnaA boxes [1, 2]. This oriC–DnaA complex promotes local DNA unwinding and loading of DnaB helicases onto each strand of single-stranded (ss)DNA, leading to replisome assembly for bidirectional replication [3–11].

DnaA consists of four functional domains (Fig. 1A) [12, 13]. In Escherichia coli, domain I mediates protein–protein interactions, including weak DnaA–DnaA interaction in DnaA complexes and high-affinity interaction with DnaB using Phe46 [5, 14, 15]. Domain II acts as a flexible linker enabling spatially elastic positioning of bound DnaB [16, 17]. Domain III contains the AAA+ motifs responsible for ATP/ADP binding and ATP hydrolysis, and mediates head-to-tail oligomerization of ATP-DnaA via Arg finger motif Arg285 (Fig. 1D) [2, 12, 18–22]. In this domain, the H/B motifs Val211 and Arg245 capture ssDNA via TT[A/G]T(T) motifs and His136 contributes to a low-affinity DnaB interaction (Fig. 1D) [4, 10, 23–25]. Domain IV recognizes DnaA boxes [26].

Figure 1.

Architecture of bacterial oriCs and replication initiation mechanism. (A) Domain structure of DnaA. Domains I–IV are shown with residue numbers. The H/B motifs (square) and the Arg finger (triangle) are indicated. Major functions are summarized at right. (B) Domain structure of DnaB. N-terminal (N-ter; light blue), linker (gray), and C-terminal (C-ter; dim blue) domains are shown with residue numbers. Major functions are summarized at right. (C) Architecture of E. coli oriC. The AT cluster (dim purple), DUE (light purple), and DOR (open bar) are indicated. TT[A/G]T(T) motifs (red bars), DnaA boxes (black and gray arrowheads indicating high- and low-affinity sites), and an IBR (green bar) are also shown. Each DOR subregion is highlighted by a bracket. (D) Replication initiation mechanism in E. coli. Color usage follows above, except for DOR (dark gray). DnaA subcomplexes unwind the DUE capturing the T-rich strand with IHF or HU (gray), and recruit the DnaB–DnaC complexes (blue and brown) via high-affinity DnaA–DnaB binding (left). The Left–DnaA subcomplex loads DnaB onto the A-rich strand, stabilizing the unwinding (middle). This supports DnaB loading to the T-rich strand by the Right–DnaA subcomplex, stimulating translocation of the first-loaded DnaB, which coordinately stimulates translocation of the second DnaB (right). (E) Architectural diversity of bacterial oriCs. DUEs (dim purple) and DORs (open bars) are shown. DnaA boxes are indicated by arrowheads (black, full-consensus sequences; open, degenerated sequences; yellow, Thermotoga maritima-specific sequences). Pink arrows, the dnaA gene. Red underline, a region required for DUE unwinding [24, 43, 45, 46].

DnaB consists of three functional domains and forms a homohexameric ring or spiral (Fig. 1B). The N-terminal and short linker–helix domains contain the low- and high-affinity binding sites for DnaA His136 and Phe46, respectively [5, 10]. The C-terminal domain with the RecA-type ATPase fold interacts with ssDNA [27, 28]. The helicase loader DnaC interacts with the DnaB C-terminal domain and its adjacent upstream region, inducing a spiral conformation that generates a gap for ssDNA entry (Fig. 1D) [3, 29, 30].

The E. coli oriC comprises a duplex-unwinding element (DUE) and a DnaA-oligomerization region (DOR) (Fig. 1C). The DUE contains three TT[A/G]T(T) motifs and connects to an upstream AT-rich cluster consisting of 12-mer sequence TATTAAAAAGAA (hereafter AT cluster) [23, 31, 32]. The DOR is subdivided into Left-, Middle-, and Right-DORs based on the clustering and orientation of DnaA boxes with the consensus TTA[T/A]NCACA. The Left-DOR contains a left-oriented array of DnaA boxes (R1, R5M, τ2, I1, and I2) and an IHF-binding region (IBR) including its binding consensus, whereas the Right-DOR contains a right-oriented array of the DnaA boxes (R4, C1, I3, C2, and C3) [18, 19, 33–35]. ATP–DnaA molecules cooperatively assemble on these elements to construct Left- and Right–DnaA subcomplexes (Fig. 1D). The R2 box-bound DnaA in the Middle-DOR slightly stimulate stability of these subcomplexes but is dispensable for DUE unwinding and DnaB loading [35, 36].

The Left–DnaA subcomplex promotes DUE unwinding by capturing the T-rich (upper) strand via TT[A/G]T(T) motifs, assisted by IHF-mediated DNA bending or by HU-mediated stabilization of DNA bending (ssDUE recruitment mechanism: Fig. 1D, left) [8, 24, 37]. Unlike IHF, HU binds DNA sequence non-specifically, but preferentially binds a bent DNA site, which is created by interaction between R1- and R5M-bound DnaA protomers in oriC [8]. The Right–DnaA subcomplex additionally engages the T-rich strand to stabilize the unwound state, thereby stimulating DnaB helicase loading onto the ssDUE, a process further supported by the adjacent AT cluster [32].

Each DnaA subcomplex recruits a DnaB–DnaC complex via the high-affinity DnaA–DnaB interactions (Fig. 1D, left). The Left–DnaA subcomplex loads the first DnaB onto the A-rich (lower) ssDUE strand, which is free from DnaA (Fig. 1D, middle). The Right–DnaA subcomplex loads the second DnaB onto the T-rich ssDUE strand in a manner depending on the low-affinity DnaA–DnaB interaction (Fig. 1D, middle-right). DnaB helicases then coordinately begin to translocate in the presence of Single-strand binding protein (SSB) for leading bidirectional replication [5, 9, 10, 32, 37, 38].

Despite the widespread conservation of DnaA and DnaB homologs across bacteria [39], oriC architectures vary substantially in the arrangement of DnaA box clusters (Fig. 1E) [1, 40]. Canonical oriCs contain DnaA box clusters on one proximal side of the DUE [41, 42]. Bipartite oriCs contain the dnaA gene inserted between two DnaA box clusters [43, 44]. Moreover, some species have bilateral oriCs where DnaA box clusters are arranged on both sides of the DUE [45]. These variations raise a fundamental question: how does a conserved initiation machinery operate on diverse oriC architectures? Representative oriCs from each architecture require a DnaA box cluster directly adjacent to the DUE for unwinding (Fig. 1E), and the ssDUE recruitment mechanism operates in Thermotoga maritima (T. maritima) and Helicobacter pylori [42, 45–47]. These observations suggest the functional conservation of E. coli Left–DnaA subcomplex across diverse oriC architectures. In contrast, whether the functions of E. coli Right–DnaA subcomplex are conserved remains unclear. We hypothesize that if this is the case, the E. coli initiation system should intrinsically or cryptically adapt to variation in Right-DOR positioning. Consistently, E. coli oriC tolerates a 10-bp insertion or deletion between the Left- and Right-DORs while retaining the in vitro initiation activity. Notably, a 5-bp deletion reduces it, highlighting the functional importance of DNA helical phasing for the Right-DOR [35]. However, the limits of positional flexibility of the Right-DOR while retaining its specific functions have not been determined.

In this study, to test our hypothesis, we investigated engineered E. coli oriC variants in which the Right-DOR was relocated to mimic bilateral and bipartite architectures. In vitro reconstituted systems demonstrated functional DnaB loading mediated by the relocated Right-DORs in comparison with the Left-oriC lacking the Middle- and Right-DORs. Subcomplex-specific mutational analyses revealed that the relocated Right–DnaA subcomplexes required both ssDNA binding and DnaB interaction in these variants for the function, similar to the wild-type (WT) oriC. Notably, these oriC variants displayed increased dependence on ssDUE binding by the relocated Right–DnaA subcomplex and on the AT cluster, reflecting reduced stability of DUE unwinding. Consistently, E. coli mutants harboring these oriC architectures on the chromosome exhibited growth patterns that correlated with in vitro activity in minimal medium. In rich medium, these mutants exhibited only slow growth, suggesting mechanical limitation for efficient initiation in addition to occurrence of inhibitory events by impaired adaptation to physiological changes for rapid cell growth. Collectively, these findings reveal a high degree of architectural flexibility in Right-DOR positioning for DnaB loading and suggest common basis underlying the evolutionary diversification of oriC architecture.

Materials and methods

Protein purification

Cells and plasmids in this section are listed in Supplementary Tables S1 and S2. DnaA and its derivatives were overproduced in KA450 cells using pKA234, pDnaAF46A, pKW44-1, pChidnaA, pChidnaA F46A, and pChidnaA V211A R245A, and purified as previously described [23, 24, 34]. HU, IHF, and N-terminally His₆-tagged DnaB in addition to DnaC, SSB, GyrA, and GryB were overproduced and purified as previously described [5, 8, 10, 37]. Total protein concentrations of the final fractions were determined using the Bradford assay. Protein purity was assessed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) by loading 1 µg of total protein, confirming that contaminating proteins were barely detectable. The concentrations of the proteins of interest were subsequently determined by SDS–PAGE using a BSA (bovine serum albumin) standard curve.

Plasmid construction

Plasmids and primers described in this section are listed in Supplementary Tables S2 and S3.

Plasmids were constructed by outward-directed PCR unless otherwise indicated. pBSinvroriC was constructed from pBSoriC using primers Rightinv Fw and Rightinv Rev. pBSinvrd5oriC was constructed from pBSinvroriC using primers Rightinvdelins Fw and Rightinvdel5 Rev. pBSinvri3oriC was obtained as an unintended single-deletion variant from pBSinvroriC generated using primers Right invdelins Fw and Rightins4 Rev.

pBSoriC-BL, pBSoriC-BL+5, and pBSoriC-BL+10 were constructed from pBSoriC-Left using primer pairs RightATL Fw/RightATL Rev, Upperins5 Fw/Upperins5 Rev, and Upperins10 Fw/Upperins10 Rev, respectively. Similarly, pRSoriC-mioC-BL was constructed from pRSoriC-mioC-Left using the RightATL Fw/RightATL Rev primer pair.

pRSoriC-mioC-Left and pRSoriC-mioC-ΔDUE were constructed from pRSoriC-mioC as a template using primer pairs oriD-1/Dr2 and ori1/dDUE rev, respectively. pRSoriC-mioC-BP was constructed from pRSoriC-mioC-Left using primers mioCRight1 Fw and mioCRight1 Rev. pRSoriC-mioC-BP-5 was constructed from pRSoriC-mioC-Left using primers mioCRight2 Fw and mioCRight2 Rev.

pCLoriC was constructed from pR1Tma12 using primers R5-I2allmTma Fw and R5-I2allmTma Rev. For pCLoriC-BL and pCLoriC-mioC-BP, the R1Tma mutation was first introduced into pBSoriC-BL and pRSoriC-mioC-BP using primers ori1tmafb and R1tma12r, followed by introduction of four miniTmaDnaA boxes at R5M, τ2, I1, and I2 box positions using primers L-R5-I2allmTma Fw and R5-I2allmTma Rev.

The ΔAT mutation was introduced using primers upperLMR and dDUE rev to generate pBSoriC-BL-ΔAT from pBSoriC-BL+10, pRSoriC-mioC-ΔAT from pRSoriC-mioC, pRSoriC-mioC-Left-ΔAT from pRSoriC-mioC-Left, and pRSoriC-mioC-BP-ΔAT from pRSoriC-mioC-BP. pBSoriC-BL-Δint was constructed from pBSoriC-BL+10 using primers ATLM and dUpDUE.

psgRNA-R-DOR and psgRNA-PmioC were constructed from psgRNA using primer pairs psgRev/psgleftoriC and psgRev/psgPmioC.

For constructing pBDoriC-BsdnaA, pBDoriC-BsdnaA-Left, pBDoriC-BsdnaA-BP, pBDoriC-BsdnaA-BP-5, and pBDoriC-BsdnaA-ΔDUE, the mioC gene coding region was substituted by the BsudnaA gene in pRSoriC-mioC, pRSoriC-mioC-Left, pRSoriC-mioC-BP, pRSoriC-mioC-BP-5, and pRSoriC-mioC-ΔDUE using Gibson assembly protocol. The BsudnaA fragment with 40-bp homologous arms was amplified from pUC-BsuOriC* using primers gibsonbsudnaA fw and gibsonbsudnaA rev. The BsudnaA region in pUC-BsuOriC* contains a deletion, resulting in a 1340 bp insert (BsudnaA*). Vector fragments were amplified from each pRSoriC derivatives using primers gibsondmioC fw and gibsondmioC rev.

Construction of MIT162 derivative strains

MIT162 (MG1655 rnhA::cat) was used as the parental strain for constructing all oriC mutant derivatives by the HoSeI method with minor modifications [48, 49]. Constructed strains are listed in Supplementary Table S1.

Electrocompetent MIT162 cells carrying the pCas plasmid were prepared in the presence of 10 mM arabinose to induce Cas9 protein. To construct MIT162-Left and MIT162-BL, Left-oriC and Bilateral-oriC fragments were amplified from pRSoriC-mioC-Left and pRSoriC-mioC-BL using primers CasgidAFw and CasmioCrev. Resulting fragments were co-transformed with psgRNA-R-DOR into the MIT162-pCas cells. For construction of MIT162-BP, Right-DOR fragments were amplified from pRSoriC-mioC-BP using primers CrPmioC fw and CrPmioC rev and co-transformed with plasmid psgRNA-PmioC. Transformants were selected on M9 glucose agar plates containing 100 µg/ml ampicillin and 50 µg/ml kanamycin, and correct recombination was confirmed by colony PCR. To eliminate psgRNA derivatives, positive clones were incubated in 1 ml LB medium containing 50 µg/ml kanamycin and 0.5 mM IPTG at 28°C overnight, followed by incubation on LB agar plates containing 50 µg/ml kanamycin. Loss of psgRNA derivatives was confirmed by ampicillin sensitivity. To cure pCas, cells are grown on LB agar plate at 42°C overnight, and loss of pCas was verified by kanamycin sensitivity. All genomic mutations were confirmed by DNA sequencing.

Form I* assay

The Form I* assay was performed as previously described with minor modifications [10, 32, 50, 51]. Reactions were carried out in a buffer containing 20 mM Tris–HCl (pH7.5), 125 mM potassium glutamate, 10 mM magnesium acetate, 8 mM dithiothreitol (DTT), 0.5 mg/ml BSA, and 3 mM ATP. DnaA derivatives were preincubated with ATP in a buffer containing 50 mM HEPES–KOH (pH 7.6), 2.5 mM magnesium acetate, 0.3 mM ethylenediaminetetraacetic acid (EDTA), 7 mM DTT, 0.007% (v/v) Triton X-100, and 20% (v/v) glycerol on ice for 3 min to generate the ATP-bound state.

For analyses using pBSoriC derivatives, 1.6 nM plasmids were incubated with the indicated amount of ATP-EcoDnaA, ATP-ChiDnaA, or these mutant derivatives for 15 min at 30°C in the presence of 40 nM IHF or 20 nM HU, together with 100 nM His-DnaB, 100 nM DnaC, 76 nM GyrA, 100 nM His-GyrB, and 760 nM SSB. GyrA and GyrB comprise gyrase as a complex. Reactions were terminated by the addition of SDS to a final concentration of 0.5%, and DNA was purified by phenol–chloroform extraction. Samples were analyzed by 0.65% agarose gel electrophoresis in 0.5× Tris–borate–EDTA (TBE) buffer at 23 V for 15 h, followed by ethidium bromide staining.

For analyses using derivatives of pRSoriC-mioC and pBDoriC-BsdnaA, reactions were performed as described above except that 1.6 μM SSB was used. In addition, purified plasmids were resolved by 0.65% agarose gel electrophoresis in 0.5× TBE buffer at 48 V for 6 h, followed by ethidium bromide staining.

Gel images were taken using the GelDoc GO imaging system (Bio-Rad Laboratories, Hercules, CA). The band intensities of each lane in the gel image were analyzed by densitometric scanning using Image J software. The proportion of Form I* DNA was calculated as the percentage of total input DNA [Form I* (%)], and values are presented as the mean ± standard deviation (SD).

P1 nuclease assay

The P1 nuclease assay was performed as previously described with minor modifications [8, 52]. Reactions were carried out in a buffer containing 60 mM HEPES–KOH (pH 7.6), 125 mM potassium glutamate, 0.1 mM zinc acetate, 8 mM magnesium acetate, 30% (v/v) glycerol, 0.32 mg/ml BSA, and 5 mM ATP. DnaA derivatives were preincubated with ATP as described above.

For analyses using pBSoriC derivatives, 1.6 nM plasmids were incubated with the indicated amount of ATP-EcoDnaA, ATP-ChiDnaA, or these mutant derivatives for 9 min at 30°C in the presence of 40 nM IHF or 20 nM HU, followed by further incubation with 1.5 units of P1 nuclease (Wako) for 5 min. The reaction was stopped by the addition of SDS and EDTA to final concentrations of 1% and 25 mM, respectively. Resultant DNA fragments were purified by phenol–chloroform extraction and ethanol precipitation. One-third of each purified DNA was digested with AlwNI (NEB), which yielded 2.6 and 1.0 kb fragments after DUE unwinding. The resultant DNA fragments were analyzed by 1% agarose gel electrophoresis with 1× Tris–acetate–EDTA (TAE) buffer at 100 V for 30 min, followed by ethidium bromide staining.

For analyses using derivatives of pRSoriC-mioC and pBDoriC-BsdnaA, purified DNA products were digested with NcoI-HF (NEB) instead of AlwNI (NEB), yielding 4.5 and 2.9 kb fragments or 4.8 and 3.5 kb fragments, respectively.

Gel images were taken using the GelDoc GO imaging system (Bio-Rad Laboratories, Hercules, CA). The band intensities of each lane in the gel image were analyzed by densitometric scanning using Image J software. The proportion of unwound DNA products was calculated as the percentage of total input DNA [DUE unwinding (%)], and values are presented as the mean ± SD.

KMnO₄ footprint assay

This assay was performed as previously described with minor modifications [32, 38]. Reactions were carried out in a buffer containing 60 mM HEPES–KOH (pH 7.6), 125 mM potassium glutamate, 0.1 mM zinc acetate, 8 mM magnesium acetate, 30% (v/v) glycerol, 0.32 mg/ml BSA, and 5 mM ATP. Forty nanomolars ATP-EcoDnaA and 100 nM His-DnaB K236A were incubated for 9 min at 30°C with 1.6 nM pRSoriC-mioC or its mutant derivatives in the presence of 20 nM HU, followed by further incubation with 10 mM KMnO₄ for 5 min. The reaction was stopped by the addition of β-mercaptoethanol and SDS to final concentrations of 140 mM and 0.33%, respectively. The resulting DNA was purified by phenol–chloroform extraction and ethanol precipitation. One-fifth of the resultant DNA were analyzed by primer extension using 0.4 units of Vent (exo-) DNA polymerase and 0.1 pmol ³²P-labeled Dr2 for the upper strand. Extension products were analyzed by electrophoresis on a 5% sequencing gel, followed by exposure of the dried gels to imaging plates. The imaging plates were scanned with the Typhoon FLA9500 (GE Healthcare). The band intensities of each lane in the gel image were analyzed by densitometric scanning using Image J software. The Relative intensity of specific DNA bands was calculated using both upstream and downstream reference bands (C1 and C2), and values are presented as the mean with two independent data points.

Flow cytometry analysis

Flow cytometry analysis was performed as previously described with minor modifications [51, 53]. MIT162 (rnhA::cat) and MIT162-BL (rnhA::cat oriC::Bilateral-oriC) cells carrying pRRNH were grown at 37°C in M9 glucose medium including 100 µg/ml ampicillin until the absorbance of the culture (A₆₀₀) reached 0.1. Portions of the cultures were diluted 1 000-fold into 5 ml M9 glucose medium including 100 µg/ml ampicillin, which were incubated at 37°C until an A₆₀₀ reached 0.1. The remaining portions were further incubated to determine the doubling time (Td) by measuring A₆₀₀ every 20 min. At an A₆₀₀ of 0.1, the aliquots of the cultures were fixed in 70% ethanol to analyze cell mass using the Multisizer 3 Coulter counter (Beckman Coulter). The remaining cultures were further incubated for 4 h with 0.3 mg/ml rifampicin and 0.01 mg/ml cephalexin for run-out replication of the chromosomal DNA. The resultant cells were fixed in 70% ethanol. After DNA staining with SYTOX Green (Life Technologies), cellular DNA contents were analyzed on the FACSCalibur flow cytometer (BD Biosciences). The number of origins per cell (ori/cell) was deduced from histograms of the flow cytometry analysis. The ori/cell values were divided by the mean cell mass deduced from the cell mass analysis, resulting in the values of ori/mass. The values of ori/cell, ori/mass, and Td are presented as the mean ± SD.

qPCR of oriC copy number

Real-time quantitative PCR (qPCR) was performed as previously described with minor modifications [51]. Cells of MIT162 (rnhA::cat), MIT162-Left (rnhA::cat oriC::Left-oriC), MIT162-BL (rnhA::cat oriC::Bilateral-oriC), and MIT162-BP (rnhA::cat oriC::Bipartite-oriC) were grown at 37°C in LB medium until the absorbance of the culture (A₆₀₀) reached 0.1. Portions of the cultures were diluted 1 000-fold into 5 ml LB medium. The remaining portions were further incubated to determine the doubling time (Td) by measuring A₆₀₀ every 20 min. When an A₆₀₀ reached to 0.1, aliquots of the cultures were fixed in 70% ethanol to analyze cell mass using the Multisizer 3 Coulter counter (Beckman Coulter). Other aliquots of the cultures were harvested, and boiled in saline for 5 min at 95°C to extract genome DNA. The levels of oriC (84.6 min) relative to ypaB (50.5 min) were quantified by real-time qPCR with THUNDERBIRD^® Next SYBR^TM qPCR Mix (Toyobo, Osaka, Japan) and primer pairs ori1/Dr2 for oriC, and STM419/STM420 for ypaB.

Results

Right-DOR orientation has a limited impact on DnaB loading

Because DnaB helicases should be loaded onto ssDNA with defined polarity at oriC, the orientation of the DnaA box clusters could influence DnaB loading. To assess the flexibility of Right-DOR orientation, we constructed three derivatives of oriC plasmid (pBSoriC) in which the Right-DOR was inverted at different distances from the Left-DOR; Inv R-DOR-oriC with the native spacing, Inv R-DORdel5-oriC with a 5-bp shorter spacing, and Inv R-DORins3-oriC with a 3-bp longer spacing (Fig. 2A). These plasmids were analyzed using an in vitro system for DnaB loading and translocation with helicase activity (Form I* assay). In this assay, DnaA unwinds the DUE with IHF or HU and loads a pair of DnaB helicases onto ssDUE in the presence of DnaC. Translocating DnaB, together with SSB, expands the ssDNA region bidirectionally, generating positive supercoils that are relaxed by DNA gyrase to produce Form I* topoisomers [10, 32, 50].

Figure 2.

In vitro replication initiation of oriC with the inverted Right-DOR. (A) Schematic of inverted Right-DOR mutant oriCs. DNA sequence between R2 and R4 boxes are shown. Each oriC name is indicated at right. (B) Form I* assay with HU. The supercoiled form (Form I) of oriC plasmids (pBSoriC [WT], pBSinvroriC [Inv R-DOR], and pBSinvrd5oriC [Inv R-DORdel5]; 1.6 nM) was incubated with the indicated amount of ATP-EcoDnaA and HU at 30°C for 15 min in the presence of DnaB, DnaC, gyrase, and SSB. The resulting plasmids were purified and analyzed using 0.65% agarose gel electrophoresis. DNA bands were visualized using ethidium bromide. DNA migration positions of Form I and Form I* are indicated. (C) Quantification of Form I* (%) in panel (B). Data are presented as mean ± SD (n = 3). (D) Form I* assay with IHF. The supercoiled form (Form I) of oriC plasmids (pBSoriC [WT], pBSinvroriC [Inv R-DOR], pBSinvrd5oriC [Inv R-DORdel5], and pBSinvri3oriC [Inv R-DORins3]; 1.6 nM) was incubated with the indicated amount of ATP-EcoDnaA with IHF and treated as in panel (B). (E) Quantification of Form I* (%) in panel (D). Data are presented as mean ± SD (n = 3). (F) P1 nuclease assay with IHF. The supercoiled form of oriC plasmids as in panel (D) in addition to that of pBSoriC-ΔDUE [ΔDUE] was incubated with the indicated amount of ATP-EcoDnaA and IHF for 9 min at 30°C, followed by P1 nuclease digestion. The purified DNA was further digested by AlwNI and analyzed using 1% agarose gel electrophoresis. The DUE-unwound form shows 1.0- and 2.6-kb fragments. (G) Quantification of unwound DNA (%) in panel (F). Data are presented as mean ± SD (n = 3). Abbreviations: Inv R, Inv R-DOR; Inv Rd5, Inv R-DORdel5; Inv Ri3, Inv R-DORins3.

Inv R-DORdel5-oriC and Inv R-DORins3-oriC displayed higher activities than Inv R-DOR-oriC (Fig. 2B–E). Especially, Inv R-DORdel5-oriC was fully active in the presence of HU, indicating that Right-DOR orientation is basically flexible in DnaB loading function when the spacing to Left-DOR is adjusted. The reduced activity of Inv R-DOR-oriC is consistent with the requirement for proper DNA helical phasing of the Right–DnaA subcomplex [35].

Inv R-DORdel5-oriC and Inv R-DORins3-oriC showed moderate Form I* formation activity in IHF condition (Fig. 2D and E). To investigate the underlying mechanism of this deduced activity, DUE unwinding activities of these oriC variants were analyzed using P1 nuclease assay in the presence of IHF and DnaA. In this assay, unwound DUE regions are selectively digested by single-strand-specific nuclease P1, and subsequent single-site digestion by AlwNI yields characteristic 2.6- and 1.0-kb fragments. All inverted Right-DOR variants exhibited DUE unwinding activities intermediate between WT and ∆DUE oriCs (Fig. 2F and G). These results suggest that Right-DOR orientation modestly affects DUE unwinding or stability of the unwound state, likely due to suboptimal engagement of the Right–DnaA subcomplex with the T-rich strand, as previously indicated [32]. Unlike IHF, HU may compensate for suboptimal positioning of the Right–DnaA subcomplex by facilitating DNA bending at multiple sites within oriC.

Bilateral oriC architecture facilitates efficient DnaB loading

Mechanistic understanding of diverse oriC architectures has been limited. To address this limitation, we examined using Form I* assay whether E. coli proteins could support DnaB loading on oriC variants that mimic non-canonical architectures (Fig. 3A and D). Bilateral oriC (BL-oriC) and Bipartite oriC (BP-oriC) architectures were constructed by relocating Right-DOR. The R2 box was omitted in these architectures for simplicity as our previous study demonstrates that the R2 box has no substantial effect on the activities of the Left–DnaA subcomplex in DUE unwinding and Form I* formation [37] (also see below).

Figure 3.

In vitro replication initiation of non-canonical oriC plasmids with HU. (A) Architectures of Bilateral-oriC (BL-oriC) and Left-oriC in pBSoriC. Each oriC name is indicated at right. (B) Form I* assay of Bilateral-oriC with HU. The supercoiled form (Form I) of pBSoriC derivatives (pBSoriC [WT], pBSoriC-BL [BL], and pBSoriC-Left [Left]; 1.6 nM) was treated as described in Fig. 2B. (C) Quantification of Form I* (%) in panel (B). Data are presented as mean ± SD (n = 3). (D) Architectures of WT oriC, Left-oriC, and Bipartite-oriC (BP-oriC) in pRSoriC-mioC derivatives. Each oriC name and spacing between Left- and Right-DORs are indicated at right. (E) Form I* assay of Bipartite-oriC with HU. The supercoiled form (Form I) of pRSoriC-mioC derivatives (pRSoriC-mioC [WT], pRSoriC-mioC-BP [BP], and pRSoriC-mioC-Left [Left]; 1.6 nM) was treated as described in Fig. 2B. (F) Quantification of Form I* (%) in panel (E). Data are presented as mean ± SD (n = 3). (G) P1 nuclease assay of Bipartite-oriC plasmid with HU. The supercoiled form of oriC plasmids as in panel (E) in addition to pRSoriC-mioC-ΔDUE [ΔDUE] was incubated with the indicated amount of ATP-EcoDnaA and HU for 9 min at 30°C, followed by P1 nuclease digestion. The purified DNA was further digested by NcoI and analyzed using 1% agarose gel electrophoresis. (H) Quantification of unwound DNA (%) in panel (G). Data are presented as mean ± SD (n = 3). See also Supplementary Figs S1 and S2.

We first analyzed a plasmid bearing the oriC variant with bilateral architecture [pBSoriC-BL carrying Bilateral-oriC (BL-oriC)]. This plasmid was constructed from the Left-oriC derivative of pBSoriC (pBSoriC-Left), by inserting an inward Right-DOR upstream of the AT cluster at different positions (Figs 3A and 4A). Notably, in both conditions with HU and IHF, BL-oriC showed Form I* formation activity comparable to WT oriC (Figs 3B and C, and 4B and C), indicating positional flexibility in the Right-DOR for DnaB loading function. In this architecture, the relocated Right-DOR is closer to the DUE than in the canonical architecture, which might underly its DnaB loading function. In addition, as BL-oriC lacks the R2 box, these results indicate that an effect of the R2 box is negligible under our experimental setup, consistent with its minimal impact both in vitro and in vivo [37, 54].

Figure 4.

In vitro replication initiation of bilateral oriC plasmids with IHF. (A) Schematic of Left-oriC and Bilateral-oriC (BL-oriC) derivatives are shown. DNA sequences between the AT cluster and R4 box in the relocated Right-DOR are shown. Each oriC name is indicated at left. (B) Form I* assay with IHF. The supercoiled form (Form I) of oriC plasmids (pBSoriC [WT], pBSoriC-BL [BL], pBSoriC-BL+5 [BL+5], pBSoriC-BL+10 [BL+10], and pBSoriC-Left [Left]; 1.6 nM) was incubated with indicated amount of ATP-EcoDnaA and IHF at 30°C for 15 min in the presence of DnaB, DnaC, gyrase, and SSB. Plasmid DNA was then analyzed as described in Fig. 2B. (C) Quantification of Form I* (%) in panel (B). Data are presented as mean ± SD (n = 3). (D) P1 nuclease assay of Bilateral-oriC plasmids with IHF. The supercoiled form of oriC plasmids (pBSoriC [WT], pBSoriC-BL [BL], pBSoriC-Left [Left], and pBSoriC-ΔDUE [ΔDUE]; 1.6 nM) was incubated with the indicated amount of ATP-EcoDnaA and IHF at 30°C for 9 min, followed by P1 nuclease digestion. The purified DNA was further digested by AlwNI and analyzed using 1% agarose gel electrophoresis. The DUE-unwound form shows 1.0 and 2.6 kb fragments. (E) Quantification of unwound DNA (%) in panel (D). Data are presented as mean ± SD (n = 3). (F) P1 nuclease assay of Bilateral+5-oriC and Bilateral+10-oriC plasmids with IHF. The supercoiled form of oriC plasmids (pBSoriC [WT], pBSoriC-BL+5 [BL+5], pBSoriC-BL+10 [BL+10], and pBSoriC-ΔDUE [ΔDUE]; 1.6 nM) was treated as described in panel (D). (G) Quantification of unwound DNA (%) in panel (F). Data are presented as mean ± SD (n = 3).

The Form I* formation activity was higher with IHF than with HU for WT and BL-oriC, likely because IHF stabilizes the Left–DnaA subcomplex, as previously reported [24], facilitating the DnaB loading more efficiently. This activity of Left-oriC was less stimulated by IHF due to the impaired interaction with the second DnaB.

Next, to assess the effect of helical phasing in BL-oriC, we similarly analyzed helical phasing variants of BL-oriC, Bilateral+5-oriC (BL+5-oriC) and Bilateral+10-oriC (BL+10-oriC) (Fig. 4A). In the presence of IHF, BL+5-oriC and BL+10-oriC exhibited similarly moderate Form I* formation activities compared to WT, but more active than Left-oriC (Fig. 4B and C). Given the importance of helical phasing of the Right–DnaA subcomplex for Form I* formation in the canonical oriC architecture [35], the present results indicate a considerable flexibility in the helical phasing and spacing of the relocated Right-DOR in the bilateral architecture. This flexibility is consistent with the above idea that the proximity of the DOR positions to the DUE has predominant importance for the DnaB loading function.

P1 nuclease assays showed that BL-oriC and BL+10-oriC retained DUE unwinding comparable to WT oriC, and BL+5-oriC showed only a slight reduction (Fig. 4D–G). This is consistent with the supportive roles for the Right-DOR in DUE unwinding. Although DUE unwinding activity of Left-oriC was slightly higher than that of WT oriC, this is likely because the relative concentrations of DnaA to Left-oriC are elevated, basically consistent with our previous data [32]. Together, these results demonstrate that the E. coli DnaA complexes can accommodate the bilateral oriC architecture with efficient DUE unwinding and DnaB loading-translocating activities.

Bipartite oriC architecture supports DnaB loading

Bipartite oriC plasmid (pRSoriC-mioC-BP) was generated using pRSoriC-mioC plasmid containing the genomic oriC–mioC region [34, 55, 56], by relocating the Right-DOR to a site downstream of the mioC gene (Fig. 3D). Notably, in the presence of HU, this Bipartite-oriC (BP-oriC) exhibited Form I* activity comparable to WT oriC at low DnaA concentrations (Fig. 3E, lanes 1–3 and 7–9, and F) and moderate activities at higher DnaA concentrations (Fig. 3E, lanes 4–6 and 10–12). Also, these activities were considerably higher than those of the Left-oriC derivative (pRSoriC-mioC-Left) (Fig. 3E, lanes 13–18), indicating a high degree of flexibility of the Right-DOR position for supporting DnaB loading-translocating processes even in the bipartite oriC architecture.

In addition, Form I* formation efficiencies of WT oriC and Left-oriC in the pRSoriC-mioC derivatives (7.4 kb) were lower than those in the pBSoriC derivatives (3.6 kb) (Fig. 3B, C, E, and F). One possible explanation is that the longer non-oriC DNA region acts as a competitor for DnaA in specific and non-specific binding manners, thereby lowering the effective DnaA concentration at oriC and reducing Form I* formation. Indeed, the mioC promoter region bears multiple DnaA boxes [55, 56]. Alternatively, the larger DNA size of the supercoiled plasmid may necessitate a longer distance of DnaB translocation to generate the detectable Form I* formation, reducing the detection efficiency.

In the presence of IHF, BP-oriC showed reduced Form I* activity but still supported higher levels than Left-oriC (Supplementary Fig. S1), basically consistent with the above results. The preference for HU is consistent with the evolutional conservation of HU, but not IHF, in species with bipartite oriCs. HU may facilitate DNA bending at multiple sites within this oriC architecture, promoting interaction of the Right–DnaA subcomplex with ssDUE.

P1 nuclease assay of pRSoriC-mioC derivatives demonstrated similar DUE unwinding activity between BP-oriC and WT oriC in the presence of HU (Fig. 3G and H), supporting the above idea in the high degree of flexibility of the Right-DOR position for DnaB loading-translocating processes in addition to a negligible effect of DnaA box R2 for BP-oriC unwinding.

To more closely mimic native bipartite oriCs, mioC regions of pRSoriC-mioC derivatives were replaced with the B. subtilis dnaA gene, resulting in pBDoriC-BsdnaA derivatives (Supplementary Fig. S2A). In these derivatives, the bipartite oriC, BP(BsdnaA)-oriC exhibited Form I* formation activity at higher levels than Left-oriC (Supplementary Fig. S2B and C). Consistently, the BP(BsdnaA)-oriC showed the DUE unwinding activity comparable to WT oriC (Supplementary Fig. S2D and E). These results demonstrate that the Right–DnaA subcomplex basically supports DnaB loading even with a dnaA-length interspace in the bipartite oriC architecture, a further indication for a high degree of flexibility in the Right-DOR position. In addition, a helical phase-shifted variant [BP-5(BsdnaA)-oriC] showed reduced Form I* formation, but not DUE unwinding (Supplementary Fig. S2B and C), underlining the importance of precise control for helical phasing of the Right-DOR even in the bipartite oriC architecture.

Chimeric Left-Tma oriC supports in vitro DnaB loading

Next, using a chimeric DnaA–oriC system, the specific mechanisms of the Right–DnaA subcomplex in non-canonical architectures were dissected. The chimeric DnaA protein (ChiDnaA) consists of domains I–III of E. coli DnaA (hereafter EcoDnaA) conjugated to domain IV of T. maritima DnaA [34]. T. maritima DnaA box sequence differs substantially from the E. coli DnaA box sequence, and these two are incompatible in binding between EcoDnaA and ChiDnaA [8, 24, 34]. Consistently, Form I* formation by ChiDnaA alone was substantially inactive in canonical and non-canonical oriC architectures with E. coli DnaA box (Supplementary Fig. S3). Leveraging this system, we replaced five DnaA boxes (R1, R5M, τ2, I1, and I2) in the Left-DOR with T. maritima DnaA boxes, either the full 12-mer (TmaDnaA box; AAACCTACCACC) or the minimal 9-mer (mini-TmaDnaA box; ACCTACCAC) (Fig. 5A and B) [57]. In this chimeric oriC (CLoriC) bearing the TmaDnaA boxes within the Left-DOR, ChiDnaA and EcoDnaA predominantly construct the Left– and Right–DnaA subcomplexes, respectively (Fig. 5C).

Figure 5.

In vitro replication initiation of the chimeric oriC plasmid. (A) Schematic of the chimeric oriC bearing TmaDnaA boxes in Left-DOR (CLoriC). Full (12-mer) and mini (9-mer) TmaDnaA boxes are indicated (dim- and light-yellow, respectively) in the chimeric oriC. (B) Sequences of WT Left-DOR and the chimeric Left-DOR. Escherichia coli DnaA boxes (black) and TmaDnaA boxes (full: yellow and mini: light-yellow) are highlighted. (C) Model of DnaA complexes constructed in the chimeric oriC. Left–ChiDnaA subcomplex is construced with ChiDnaA and the chimeric Left-DOR. Domain IV of ChiDnaA is highlighted (red). (D) P1 nuclease assay of pCLoriC. The supercoiled form (1.6 nM) of oriC plasmids, pBSoriC (WT) and pCLoriC (CL), was incubated for 9 min at 30°C with IHF (40 nM) and the indicated amounts of ChiDnaA WT, F46A, or VR in the presence or absence of EcoDnaA (30 nM), followed by P1 nuclease digestion. The purified DNA was further digested by AlwNI and analyzed using 1% agarose gel electrophoresis. ChiDnaA VR bears the H/B-motif mutations, V211A and R245A. (E) Quantification of unwound DNA (%) in panel (D). Data are presented as mean ± SD (n = 3–4). (F) Form I* assay of pCLoriC. The supercoiled form (Form I) of oriC plasmids as above was incubated with the indicated amounts of ATP-EcoDnaA and ATP-ChiDnaA or these F46A mutant derivatives at 30°C for 15 min in the presence of DnaB, DnaC, gyrase, IHF, and SSB. Plasmid DNA was then analyzed as described in Fig. 2B. FA, DnaA F46A. (G) Quantification of unwound DNA (%) in panel (F). Data are presented as mean ± SD (n = 3). Dark gray bar, WT oriC; Light gray bar, CLoriC; N.T., Not Tested. See also Supplementary Fig. S3.

Before assessing non-canonical oriC architectures, we applied P1 assay to assess the formation of a functional Left–DnaA subcomplex with ChiDnaA in the canonical oriC architecture (CLoriC), using pCLoriC, a derivative of pBSoriC (Fig. 5C–E). In WT oriC, DnaA F46 residue and H/B-motifs V211 R245 residues are essential, respectively, for high-affinity interaction with DnaB and for specific binding to ssDUE (Fig. 1). WT ChiDnaA and its F46A mutant derivative impairing the DnaB high-affinity interaction efficiently promoted DUE unwinding (Fig. 5D, lanes 5–13, and E). In contrast, the H/B-motifs mutant derivative (ChiDnaA-VR), which bears the V211A R245A substitutions and impairs ssDUE-binding activity, did not unwind DUE in either the presence or absence of EcoDnaA (Fig. 5D, lanes 14 and 24). These results are consistent with the essential role of the Left–DnaA subcomplex in DUE unwinding and suggest that ChiDnaA predominantly constitutes the Left–DnaA subcomplex [24]. In addition, EcoDnaA might only slightly stimulate DUE unwinding in the presence of 5 nM ChiDnaA and might slightly inhibit it at higher ChiDnaA concentrations (Fig. 5D, lanes 5–9 and 15–19, and E). However, these subtle changes were not observed in the presence of ChiDnaA F46A (Fig. 5D, lanes 10–13 and 20–23, and E). Given that ChiDnaA and ChiDnaA F46A showed comparable DUE unwinding activities in the absence of EcoDnaA (Fig. 5D, lanes 5–13, and E), consistent with our previous report [15], the slight differences in the DUE unwinding profile observed with the combination of EcoDnaA and ChiDnaA are likely within a technical fluctuation.

DnaB loading function of Left–DnaA subcomplex formed with ChiDnaA was also analyzed by Form I* assay using pCLoriC (Fig. 5F and G). This activity should require both ChiDnaA and EcoDnaA, as both DnaA-subcomplexes are crucial for efficient DnaB loading [10, 32, 37, 51]. Indeed, ChiDnaA alone only slightly stimulated Form I* formation with CLoriC, whereas EcoDnaA alone did not (Fig. 5F, lanes 6–8 and 12, and G), consistent with the DUE unwinding and basal DnaB-loading activities of the Left-oriC [32]. Notably, WT ChiDnaA robustly stimulated Form I* formation in the presence of EcoDnaA, whereas ChiDnaA F46A did not (Fig. 5F, lanes 9 and 11). These results demonstrate that the high-affinity interaction between the Left–ChiDnaA subcomplex and DnaB is required for efficient DnaB loading, consistent with the proposed mechanism of DnaB loading (Fig. 1D).

Non-canonical oriC architectures increase dependence of ssDNA binding by the relocated Right–DnaA subcomplex for DnaB loading

We next compared specific functions of the Right–DnaA subcomplex in DnaB loading between canonical and non-canonical architectures (Fig. 6). Like the canonical architecture (CLoriC of Fig. 5A), non-canonical chimeric-oriC architectures carried identical substitutions of TmaDnaA boxes in the Left-DOR (Fig. 6C and F). To evaluate the major functions of the Right–DnaA subcomplex, namely DnaB interaction and ssDUE binding, we employed specific negative-control EcoDnaA mutants defective in either DnaB binding (F46A) or ssDUE binding (V211A) [15, 23].

Figure 6.

In vitro DnaB loading of chimeric Left-DOR oriC with bilateral and bipartite configurations. (A) Form I* assay of pCLoriC. The supercoiled form (Form I) of pCLoriC (1.6 nM) was incubated and analyzed as described in Fig. 5F. (B) Quantification of Form I* (%) in panel (A). Data are presented as mean ± SD (n = 3). (C) Schematic of BL-CLoriC. Full (12-mer) and mini (9-mer) TmaDnaA boxes are indicated as in Fig. 5A. (D) Form I* assay of pCLoriC-BL. The supercoiled form (Form I) of pCLoriC-BL plasmid (1.6 nM) was incubated and analyzed as described in Fig. 5F. (E) Quantification of Form I* (%) in panel (D). Data are presented as mean ± SD (n = 3). (F) Schematic of BP-CLoriC. TmaDnaA boxes are indicated as above. (G) Form I* assay of pCLoriC-mioC-BP. The supercoiled form (Form I) of pCLoriC-mioC-BP plasmid (1.6 nM) was incubated and analyzed as described in Fig. 5F except for using HU instead of IHF. (H) Quantification of Form I* (%) in panel (G). Data are presented as mean ± SD (n = 3). Abbreviations: EW, E. coliDnaA WT; CW, ChiDnaA WT; CF, ChiDnaA F46A; CV, ChiDnaA V211A. See also Supplementary Fig. S3.

In the canonical oriC architecture (pCLoriC), EcoDnaA F46A only slightly stimulated Form I* formation even with ChiDnaA (Fig. 6A, lanes 7–9, and B), consistent with the requirement of DnaB high-affinity interaction by the Right–DnaA subcomplex [32]. EcoDnaA V211A retained moderate activity (Fig. 6A, lanes 10–12), supporting a contributory role of ssDUE interaction by the Right–DnaA subcomplex. The residual activity might be attributed to Arg245 (B-motif), as reported previously [32].

Next, we assessed the bilateral oriC architecture using a pBSoriC-derivative pCLoriC-BL carrying the chimeric Left-DOR within the bilateral configuration (BL-CLoriC) (Fig. 6C). WT EcoDnaA robustly stimulated Form I* formation of pCLoriC-BL in concert with ChiDnaA (Fig. 6D, lanes 1–6 and 13–14, and E), consistent with the efficient DnaB loading activity of BL-oriC (Fig. 3). EcoDnaA F46A again showed only minimal stimulation (Fig. 6D, lanes 7–9), supporting the importance of DnaB high-affinity binding by the relocated Right–DnaA subcomplex. In contrast to its moderate activity in the canonical architecture, EcoDnaA V211A was largely inactive in BL-CLoriC (Fig. 6D, lanes 10–12), indicating an increased dependence on ssDUE binding by the relocated Right–DnaA subcomplex in the bilateral architecture.

We also analyzed the bipartite architecture using a pRSoriC-mioC-derivative pCLoriC-mioC-BP carrying the chimeric Left-DOR within the bipartite configuration (BP-CLoriC) (Fig. 6F). Because Bipartite-oriC showed more comparable activity to WT oriC under the HU condition (Fig. 3 and Supplementary Fig. S1), further experiments were performed with HU rather than IHF. WT EcoDnaA moderately stimulated Form I* formation of BP-CLoriC in the presence of WT ChiDnaA (Fig. 6G, lanes 1–6 and 13–14, and H), consistent with the moderate DnaB loading activity of BP-oriC (Fig. 3). However, neither EcoDnaA V211A nor F46A derivatives substantially supported Form I* formation (Fig. 6G, lanes 7–12). These results indicate that both ssDUE binding and DnaB high-affinity interaction of the Right–DnaA subcomplex are essential in the bipartite architecture.

In summary, a series of chimeric experiments (Fig. 6) indicate that ssDUE binding and DnaB recruitment by the Right–DnaA subcomplex are commonly important across both canonical and non-canonical oriC architectures, suggesting a conservation of fundamental mechanisms among oriC variants. Notably, the increased importance of ssDUE binding by the Right–DnaA subcomplex in non-canonical oriC architectures likely reflects suboptimal spatial positioning of the complex, which compromises efficient ssDUE engagement, and thereby increases the requirement for stable ssDUE–DnaA interactions for efficient DnaB loading.

The DUE-upstream AT cluster supports DnaB loading in non-canonical oriC architectures

Given the increased dependence on ssDUE binding by the relocated Right–DnaA subcomplex in non-canonical architectures (Fig. 6), we examined a possible role for the AT cluster in the architectures. We and others previously demonstrate that the AT cluster promotes replication initiation in Left-oriC [32, 41].

First, we constructed a BL-oriC ΔAT plasmid (pBSoriC-BL-ΔAT), in which the 12-bp AT cluster was deleted from BL+10-oriC, for adjusting the spacing between the relocated Right-DOR and DUE. Also, we constructed a BL-oriC Δint plasmid (pBSoriC-BL-Δint) for a spacing control, in which a 12-bp sequence was deleted in the intervening region between the relocated Right-DOR and the AT cluster (Fig. 7A). Also, pBSoriC and pBSoriC-ΔAT, which carries WT oriC and its AT cluster-deleted derivative, were used as controls [32]. WT oriC and the oriC ΔAT derivative exhibited similar activity of Form I* formation, consistent to our previous data [32]. The BL-oriC Δint almost fully retained Form I* formation activity, similar to WT oriC, oriC ΔAT, and BL-oriC (Figs 1C, 4B, and 7B, lanes 1–12, and C). Given that BL+10-oriC showed a moderate activity (Fig. 4), these results further support the importance of spacing between the DUE and the Right-DOR in bilateral configurations. Notably, the BL-oriC ΔAT exhibited only moderate Form I* formation (Fig. 7B, lanes 13–16), indicating that the AT cluster stimulates DnaB loading in the bilateral architecture.

Figure 7.

In vitro DnaB loading of oriC ∆AT with bilateral and bipartite configurations. (A) Schematic of BL-oriC derivatives. DNA sequences across the AT cluster and R4 box in the relocated Right-DOR are shown. Each oriC name is indicated at left. (B) Form I* assay of BL-oriC ∆AT plasmid. The supercoiled form (Form I) of oriC plasmids (pBSoriC [WT], pBSoriC∆AT [∆AT], pBSoriC-BL-∆int [BL ∆int], and pBSoriC-BL-∆AT [BL ∆AT]; 1.6 nM) was incubated and analyzed as described in Fig. 3B except for using IHF instead of HU. (C) Quantification of Form I* (%) in panel (B). Data are presented as mean ± SD (n = 3). (D) Schematic of BP-oriC and BP-oriC ΔAT in pRSoriC-mioC derivatives.(E) Form I* assay of BP-oriC ∆AT plasmid. The supercoiled form (Form I) of oriC plasmids (pRSoriC-mioC [WT], pRSoriC-mioC-∆AT [∆AT], pRSoriC-mioC-Left [Left], pRSoriC -mioC-Left-∆AT [Left ∆AT], pRSoriC-mioC-BP [BP], and pRSoriC -mioC-BP-∆AT [BP ∆AT]; 1.6 nM) was incubated and analyzed as described in Fig. 3E. (F) Quantification of Form I* (%) in panel (E). Data are presented as mean ± SD (n = 3).

Next, we analyzed Form I* formation of a BP-oriC ΔAT plasmid (pRSoriC-mioC-BP-ΔAT), in which the 12-bp AT cluster was similarly deleted from BP-oriC shown in Fig. 3D (Fig. 7D). Even in the HU condition, consistent with our previous observations in the IHF condition [32], deletion of the AT cluster severely impaired Form I* formation of Left-oriC but had little effect on WT oriC (Fig. 7E, lanes 1–16, and F). Based on this, Form I* formation of BP-oriC derivatives was analyzed in the HU condition. The AT cluster deletion markedly abolished Form I* formation also in the bipartite architecture (Fig. 7E, lanes 17–24, and F), demonstrating that the AT cluster is essential for DnaB loading in this architecture.

Together, these results further support the idea that stable and expanded DUE unwinding is stimulatory or even essential for DnaB loading in the bilateral and bipartite oriC architectures. In these architectures, the spatial positioning of the relocated Right–DnaA subcomplexes is likely suboptimal for ssDUE binding, thereby increasing the dependence on the duplex-unstable AT cluster in supporting efficient DnaB loading.

DUE unwinding is destabilized in non-canonical oriC architectures

The importance of DnaA–ssDNA binding and the AT cluster for the DnaB loading in non-canonical oriC architectures inspired us to analyze DUE unwinding at the nucleotide resolution using KMnO₄ footprinting. In this assay, KMnO₄ preferentially oxidizes exposed thymine residues in ssDNA, leading to strand scission detectable by primer extension [32, 38]. KMnO₄ concentration was optimized to modify about one nucleotide per DUE, thereby yielding a broad footprint pattern and preferentially detects more stable DUE unwinding events than the P1 nuclease assay. Due to the thymine residue specificity, we focused on the T-rich strand. All plasmids were derivatives of pRSoriC-mioC (Supplementary Fig. S4A) and were analyzed in the presence of HU.

DnaA induced hypersensitivity of multiple thymine residues within the DUE to KMnO₄, whereas only a moderate increase was observed in Left-oriC (Fig. 8A, lanes 1–4, B, and C, and Supplementary Fig. S1B–D), consistent with previous findings [32, 38]. Thus, the reduced KMnO₄ sensitivity of Left-oriC likely reflects destabilized DUE unwinding in the absence of interaction between the T-rich strand and the Right–DnaA subcomplex. Both BL- and BP-oriCs exhibited footprint patterns similar to Left-oriC (Fig. 8A, lanes 5–8), suggesting that the relocated Right–DnaA subcomplexes in these oriCs do not stably interact with the T-rich strand. These findings are consistent with the shared dependence on the AT cluster for efficient DnaB loading among Left-, BL-, and BP-oriCs (Fig. 7).

Figure 8.

Mapping of the unwinding region within DUE. (A) KMnO₄ footprint analysis of BP- and BL-oriCs with DnaA and HU. The supercoiled form of oriC plasmids (pRSoriC-mioC [WT], pRSoriC-mioC-Left [Left], pRSoriC-mioC-BP [BP], and pRSoriC-mioC-BL [BL]; 1.6 nM) was incubated with the 40 nM ATP-EcoDnaA and 20 nM HU for 9 min at 30°C, followed by further incubation with KMnO₄ for 5 min. Modified bases on the T-rich strand were analyzed by primer extension and sequencing gel. Positions of the DUE and the AT cluster, determined by sanger-sequence maker, are shown at left. Red lines indicate TTA[G]T(T) motifs. (B, C) Quantification of Relative intensity of bands “r” and “m” to control “C1” in panel (A). Indicated data of two independent experiments are shown with average. (D) KMnO₄ footprint analysis of BP- and BL-oriCs with DnaB K236A. The supercoiled form of oriC plasmids as above was incubated with the 40 nM ATP-EcoDnaA, 100 nM DnaB K236A, and 20 nM HU for 9 min at 30°C, followed by further incubation with KMnO₄ for 5 min. Modified bases on the T-rich strand were analyzed as described above. (E–H) Quantification of Relative intensity of bands “r,” “m,” “l,” and “at” to control “C1” in panel (D). Indicated data of two independent experiments are shown with average. See also Supplementary Fig. S4.

Next, we analyzed the DUE unwinding state after DnaB loading. To immobilize DnaB at the loading site, we employed the ATPase-deficient DnaB K236A mutant. In WT oriC, DnaB K236A together with DnaA extended KMnO₄ sensitivity across overall DUE and into the AT cluster (Fig. 8D, lanes 1–3, and E–H, and Supplementary Fig. S4E–H), consistent with our previous data [32]. This extension is likely caused by steric hindrance of DnaB hexamers loaded onto the ssDUE. KMnO₄ reactivity in Left-oriC was moderate even in the presence of both DnaA and DnaB K236A (Fig. 8D, lanes 4–6), indicating reduced efficiency of DnaB loading. This is also consistent with our previous data [32], results of the Form I* formation (Fig. 3) and previous suggestion that the Left–DnaA subcomplex preferentially loaded DnaB onto the lower strand and its loading to the upper strand is impeded [10]. Notably, BL-oriC exhibited a footprint pattern similar to WT oriC in the presence of both DnaA and DnaB K236A (Fig. 8D, lanes 10–12), consistent with its high activity in the Form I* assay (Fig. 3). The increase in KMnO₄ sensitivity of BL-oriC was slightly lower than that of WT oriC, possibly reflecting slight reduction in stability of unwound DNA state due to the relocated Right–DnaA subcomplex under these conditions. In addition, BP-oriC displayed an intermediate footprint pattern between those of Left- and BL-oriC constructs (Fig. 8D, lanes 7–9), indicating intermediate efficiency of DnaB loading, consistent with the results of Form I* assay (Fig. 3).

Non-canonical oriC architectures stimulate replication initiation in vivo

To examine in vivo activity, we generated mutant E. coli strains in which BP- and BL-oriCs were introduced at the native chromosomal oriC locus. To ensure construction of mutant oriC strains, these strains were constructed in an rnhA::cat background, which sustains replication initiation via constitutive stable DNA replication (cSDR) independently of the DnaA–oriC system [58, 59]. Cells bearing the entire oriC deletion (ΔoriC) and rnhA mutation (KW262-5) very slowly grew with the doubling time of ∼70 min even at 37°C in LB medium (Supplementary Table S4). The Left-oriC strain (MIT162-Left) grew with the doubling time of ∼47 min at 37°C in LB medium, which is faster than the ΔoriC strain and is slower than the WT oriC strain (MIT162) (Fig. 9A), consistent with a previous report [41]. Also, growth of MIT162-Left cells was confirmed to be dependent on the presence of Left-oriC (Supplementary Table S4). Although the BP-oriC strain (MIT162-BP) exhibited slow growth comparable with MIT162-Left, the BL-oriC strain (MIT162-BL) showed a moderate growth rate, suggesting that BL-oriC is more active in replication initiation in vivo than Left-oriC and BP-oriC.

Figure 9.

In vivo analyses of non-canonical oriC strains. (A) Doubling times of MIT162 [WT], MIT162-Left [Left], MIT162-BP [BP], and MIT162-BL [BL] cells at 37°C. Data are presented as mean ± SD (n = 5). (B) Relative oriC copy number to ypaB of cells as in panel (A) at 37°C measured by qPCR. Data are presented as mean ± SD (n = 5). (C) Cell mass of cells as in panel (A) measured by Coulter counter at 37°C. Data are presented as mean ± SD (n = 5). (D, E) Spotting assays. MIT162 [WT], MIT162-Left [Left], MIT162-BP [BP], and MIT162-BL [BL] cells carrying pRRNH or pMZ1 [vector] were incubated on M9 glucose agar plates (D) and LB agar plates (E) with 25 µg/ml ampicillin at 37°C for 36 h. Representative plates of four independent experiments are shown. (F) Flow cytometry analysis following replication run-out. DNA content and cell mass of MIT162 [WT] and MIT162-BL [BL] cells carrying pRRNH were quantified; mass, ori/mass ratio, and doubling time are shown as mean ± SD with representative histogram of DNA content per cell (n = 5).

To further dissect activities of Left-oriC and BP-oriC, we next quantified relative oriC copy numbers by qPCR. Because the ter region (32.4 min) contains origins of cSDR, oriC (84.6 min) copy number was inferred using the mid-chromosomal locus ypaB (50.5 min), as performed previously [51, 60]. Whereas Left-oriC cells exhibited approximately half the oriC copy number of WT oriC cells, both BP- and BL-oriC cells showed 1.4-fold higher oriC copy numbers relative to Left-oriC cells (Fig. 9B), indicating that the relocated Right-DOR stimulates replication initiation in both strains. In addition, the cell mass of BL-oriC cells was smaller than that of BP-oriC cells (Fig. 9C), consistent with a higher initiation frequency in BL-oriC cells. This is also consistent with the doubling times of those cells (Fig. 9A) and with the activities observed in Form I* assay (Fig. 3).

We further assessed growth by spotting assays using strains carrying either an rnhA-expressing plasmid (pRRNH) for complementation or vector control (pMZ1). On M9 glucose agar plates, all strains with vector grew similarly. In contrast, expression of rnhA severely impaired growth of the Left-oriC strain due to inactivation of cSDR (Fig. 9D). Growth of the BP-oriC strain was moderately impaired but was faster than the Left-oriC strain. The BL-oriC strain sustained growth comparable to the WT oriC strain, yet with a delay at early time points (Fig. 9D and Supplementary Fig. S5A). These results indicate that the relocated Right-DOR moderately enhances growth of BP-oriC strains and more effectively that of BL-oriC strains, consistent with the present in vitro data. In addition, on LB agar plates, whereas WT oriC strain grew rapidly, all mutant oriC strains expressing rnhA exhibited only slow growth (Fig. 9E and Supplementary Fig. S5B). Among these mutant oriC strains, the BL-oriC strain grew with a faster rate than the BP-oriC strain, and the BP-oriC strain did with a slightly faster than the Left-oriC strain (Supplementary Fig. S5B), consistent with the data shown Fig. 9A. Under rich-medium conditions, the non-canonical architectures likely support replication initiation with limited mechanistic efficiency, possibly owing to specific defects in the initiation process (also see the “Discussion” section).

For precise evaluation of replication initiation by BL-oriC in vivo, we measured using flow cytometry the ori/mass ratio in Bilateral-oriC cells carrying pRRNH. Exponentially growing cells in M9 glucose medium were treated with rifampicin and cephalexin to allow run-out replication, resulting in cells containing chromosome numbers equivalent to the number of oriC copies prior to drug treatment. Combined with cell mass measurements obtained using Coulter counter, ori/mass was calculated as an indicator of initiation frequency. BL-oriC strains expressing rnhA exhibited a moderate ori/mass ratio and asynchronous initiation, as evidenced by an increased population of cells containing three oriC copies (Fig. 9F), indicating partial inhibition of BL-oriC activity even under slow-growth conditions.

Together, these results indicate that, even in vivo, both bilateral and bipartite oriC architectures are basically competent in replication initiation, although the bipartite oriC architecture is less efficient in E. coli cells. In addition, both architectures are not well tolerated under conditions that support rapid growth. These findings are overall consistent with the in vitro data and support a high degree of architectural flexibility in oriC in vivo, albeit with substantial limitations in functional efficiency.

Discussion

At E. coli oriC, the DnaA complexes promote ssDUE-specific loading of DnaB helicases to initiate bidirectional chromosome replication. This mechanism has been extensively characterized using the robust in vitro system and versatile genetic tools [5, 9, 10, 32, 37, 38]. However, it has remained unclear whether this mechanism can operate on diverse oriC architectures, ranging from canonical one-sided arrangements of DnaA complexes to bilateral and bipartite arrangements of those. Here, using E. coli oriC variants with large-scale rearrangements, we show that the E. coli DnaB loading system is intrinsically adaptable to such non-canonical oriC architectures. The Right-DOR retained its stimulatory activity for DnaB loading even when inverted (Fig. 2), or relocated upstream of DUE to mimic the bilateral architecture, or when a mioC or dnaA gene was inserted between the Right- and Left-DORs to mimic the bipartite architecture (Figs 3 and 4). A chimeric DnaA–oriC system demonstrated that relocated Right–DnaA subcomplexes engage both DnaB and ssDUE, similar to the native configuration (Figs 5 and 6). However, non-canonical architectures showed increased dependence on ssDUE binding by the relocated Right–DnaA subcomplexes and on the AT cluster (Figs 6 and 7). Consistently, DUE unwinding was less stable in these oriC architectures, as well as in Left-oriC, than in WT oriC (Fig. 8). In vivo, E. coli strains harboring non-canonical oriCs at the native chromosomal oriC locus sustained growth in M9-glucose minimal medium in proportion to their in vitro activities (Fig. 9). As summarized in Fig. 10A, these findings define a high degree of architectural flexibility in oriC as well as mechanistic requirements for DnaB loading conserved across diverse oriC architectures and highlight the specific importance of DnaA–ssDUE interactions and the DUE-flanking AT cluster in non-canonical architectures.

Figure 10.

Summary of data and model of DnaB loading processes in non-canonical oriCs. (A) Summary of the experimental data. The first and second rows indicate the name of activity and assay conditions, respectively. DUE unwinding activity is omitted because all oriC derivatives exhibited activity comparable to that of the WT oriC. DnaB loading activity is defined based on the Form I* assay using the DnaA concentration at which the WT oriC reached approximately half of its maximal Form I* formation. Unwinding expansion activity is determined using the background-subtracted intensity of band “l” (Fig. 8). Activity levels are classified as follows: +++, 70%–100%; ++, 40%–70%; +, 10%–40%; and −, <10% of the WT oriC activity. In vivo activity was scored as follows: +++, activity comparable to the WT oriC strain in both M9 and LB media; ++, activity comparable to the WT oriC strain in M9 medium but reduced in LB medium; +, moderate activity relative to the WT oriC strain in M9 medium and little activity in LB medium; and +/−, little activity in both M9 and LB media. Abbreviation: w/ or w/o, with or without; AT, the AT cluster; RC, Right–DnaA subcomplex; N.A., not applicable. (B) Model of DnaB loading in WToriC and non-canonical oriCs. In WT oriC, two DnaA subcomplexes efficiently capture the T-rich DUE strand and unwind the DUE, resulting in maximal expansion of the unwound region and efficient loading of two DnaB helicases (upper panel). In non-canonical oriCs, upon the first DnaB loading, the unwound DUE region is unstably expanded with the support of the AT cluster (middle panel). In bilateral configurations, the Right–DnaA subcomplex efficiently engages ssDUE and load DnaB due to the close proximity to the DUE (lower left panel). In bipartite configurations, the Right–DnaA subcomplex flexibly moves due to a long linker DNA and DnaB loading becomes relatively inefficient (lower right panel). L, the l region of DUE (Fig. 8); AT, the AT cluster.

Integrating these findings with previous studies and the idea that essential mechanisms in DnaB loading are conserved diverse oriC architectures, we propose a mechanistic model for DnaB loading in non-canonical oriC derivatives (Fig. 10B). In WT-oriC, the two DnaA subcomplexes cooperate to stabilize and expand the unwound region, thereby promoting efficient DnaB loading (Figs 8 and 10B, upper panel) [10, 32]. In contrast, the relocated Right–DnaA subcomplex has limited access to ssDUE due to suboptimal spatial positioning, reducing the stability and expanding of unwinding (Figs 8 and 10B, middle-lower panels). Loading of the first DnaB by the Left–DnaA subcomplex moderately stabilizes DUE unwinding through steric hindrance (Fig. 10B, middle), unstably expanding the unwound region with the support of the AT cluster, as previously observed [10, 32]. This unwound region enables the relocated Right–DnaA subcomplex to access to ssDUE, further stabilizing DUE unwound state and promoting the second DnaB loading (Fig. 10B, lower panels). In the bilateral oriC architecture, the proximity of the relocated Right–DnaA subcomplex to the DUE allows efficient access to ssDUE (Fig. 10B, lower left), supporting rapid loading of the second DnaB (Figs 3 and 6). In contrast, in the bipartite oriC architecture, increased spacing from the DUE likely permits free swiveling of the relocated Right–DnaA subcomplex and reduces the efficiency of ssDUE capture (Fig. 10B,lower right). HU binding would promote favorable DNA looping, thereby enhancing ssDUE capture and facilitating the loading of two DnaB helicases onto the unwound region (Fig. 3 and Supplementary Fig. S1).

Similar to the WT oriC, coordinated translocation of DnaB helicases should be preserved even in non-canonical oriC architectures, as indicated by robust Form I* formation (Figs 3, 6, and 8). In this mechanism, translocation of the first-loaded DnaB is initially restrained, likely by binding to DnaA subcomplex [10]. Subsequent loading of the second DnaB onto the opposite strand relieves this constraint, enabling translocation of the first DnaB, likely through disruption the first DnaB–DnaA binding. Given that DnaB helicases are loaded onto the limited ssDUE region, the steric clash or specific interaction between the two DnaB helicases would promote this transition [10]. Consistently, loading of the second DnaB onto the limited ssDUE region is observed even in non-canonical oriC architectures (Figs 3, 6, and 8). Translocation of the first DnaB promotes SSB binding to the expanded ssDNA region, which in turn stimulates translocation activity of the second DnaB, allowing passing through DnaA complexes for enabling bidirectional replication [9]: these processes are conserved across canonical and non-canonical oriC architectures.

Non-canonical oriC architectures supported cell growth in M9 glucose minimal medium at levels comparable to their in vitro activities, but in comparison with WT oriC cells, cell growth of the non-canonical oriC cells in LB rich medium was more severely impaired than in M9 medium (Fig. 9 and Supplementary Fig. S5), similar to the previously reported phenotype for cells lacking the right-half regions of oriC [41]. These observations suggest that the native E. coli oriC has evolved specialized mechanisms to support rapid growth. Notably, the canonical oriC exhibited greater stability of DUE unwinding than the non-canonical variants (Fig. 8), likely maximizing the efficiency of DnaB loading by both DnaA subcomplexes. In addition, an increase of AT content within the E. coli oriC DUE enhances DUE unwinding and in vitro initiation activity [61]. Similar to E. coli, Vibrio species have canonical oriC. Among them, Vibrio natriegens exhibits exceptionally short doubling time of 9.4 min [62] and has a relatively AT-rich DUE compared to other Vibrio species (Supplementary Fig. S6). These observations suggest that mechanical optimization of DUE unwinding represents a general strategy for coordinating initiation efficiency with fast cell growth.

Also, for the non-canonical oriC cells growing in LB medium, in addition to the mechanical limitation on efficient initiation, inhibitory events should occur, causing further reduction of the growth rate of cells. As suggested for the cells lacking the right-half oriC [41], impaired timely initiation at the mutant oriC might exacerbate defects in coordination between replication and cell cycle in rich medium, thereby compromising multiple cellular processes such as chromosome integrity and cell division. In addition, physiological state of cells also differs between rich and poor medium. DNA supercoiling levels have been reported to be higher in LB medium than in a minimal medium [63], raising a possibility that such supercoiling levels may be inhibitory to the non-canonical oriC but not the canonical oriC. A nucleoid-associating protein Fis is abundant in cells rapidly growing in rich medium [64]. Cells bearing R4 box mutation have been reported to require specific binding of Fis to a site flanking the R2 box for viability [54]. In contrast, cells with oriC variants lacking the right-half as well as the full-length oriC cells remain viable without the Fis binding site [41, 43], consistent with our present observation for the oriC-Left cells lacking the Fis binding site. In addition, the non-canonical oriC bilateral and bipartite variants used in this study contain the relocated Right-DOR, including the R4 box, but not the Fis binding site. The role for Fis in various oriC-mutant cells remains to be further examined.

The proposed mechanistic model (Fig. 10B) is supported by previous studies of other bacterial species. In the native bipartite oriC of Helicobacter pylori, the DnaA subcomplex proximal to the DUE is sufficient for DUE unwinding, whereas the distal DnaA subcomplex promotes unwinding expansion [43]. These two DnaA subcomplexes interact by DnaA domain I dimerization, resulting in DNA loop formation across the interspace [65]. These observations suggest an adaptive mechanism in bipartite oriCs in which inter-DnaA subcomplex interactions facilitate recruitment of the distal DnaA subcomplex to the unwound DUE, consistent with our model (Fig. 10B). Given that oriC loop formations have been observed in diverse bacteria, including ε-proteobacteria to the Gram-positive Bacillus subtilis [66, 67], this strategy may be broadly conserved among species that have bipartite oriCs. Proteins such as DiaA or HobA, which stimulate DnaA–DnaA interaction by direct linking, could enhance such inter-DnaA subcomplex interactions in addition to DnaA–DnaA interaction [15, 65, 68, 69].

Thermoanaerobacter tengcongensis (T. tengcongensis) oriC is the only experimentally validated example of a bilateral oriC [45]. According to the transition probability to the unwound state, calculated by Stress-Induced Duplex Destabilization method available from the DoriC 12.0 database [70], the DUE of T. tengcongensis oriC exhibits relatively high transition probability (~3.5e⁻³) compared to those of canonical oriCs (e.g. E. coli: ~1.8e⁻⁴ and T. maritima: ~3.0e⁻⁴) and bipartite oriCs (e.g. H. pylori: ~1.2e⁻⁵ and B. subtilis: ~3.0e⁻⁵) (Supplementary Fig. S7). Moreover, the presence of bilateral oriCs is predicted in several species within the genus Thermoanaerobacter as well as in diverse proteobacteria and cyanobacteria. Some of these oriCs contain regions of exceptionally high transition probability between DnaA box clusters (Supplementary Fig. S8). Such intrinsic duplex-instability of oriC may facilitate recruitment of DnaA subcomplexes to the unwound DUE and support efficient DnaB loading. Further detailed experimental studies using various bacterial species are evidently important.

In E. coli oriC, DnaA molecules orderly assemble using R5M box and R4 box as an assembling core within Left-DOR and Right-DOR, respectively [8, 24, 34]. Although this study insists the importance of forming at least two DnaA subcomplexes, some species do not appear to possess two DnaA box clusters in their oriC based on the prevalent E. coli-like DnaA box consensus sequence. However, DnaA box can be highly degenerated, as exemplified by the low-affinity DnaA boxes in E. coli (e.g. the τ2 box shares only three of the nine-mer consensus sequence). Also, some species establish species-specific DnaA box sequences, such as T. maritima DnaA box (AAACCTACCACC) and the W-box (NNNTCCCA) of Caulobacter crescentus (C. crescentus) [42, 71]. Furthermore, the orientation of individual DnaA boxes within a DnaA box cluster is not always strictly required when DnaA contains a sufficiently long linker between domains III and IV that allows swiveling of these domains, as shown by T. maritima DnaA [57]. Therefore, even though some oriC regions do not appear to contain ideal DnaA box arrays, they may be capable of assembling two DnaA complexes through previously unrecognized or highly degenerate DnaA-binding sites with greater flexibility in DnaA box orientations.

The high flexibility of oriC architecture and sequence may have facilitated the evolutionary acquisition of binding sites for species-specific origin biding proteins (OriBP). For instance, C. crescentus recruits the cell cycle master regulator CtrA to oriC through TTAANNNNNNNTTAA motifs, regulating the timely assembly of DnaA [71]. In B. subtilis, entry into sporulation is governed by the response regulator Spo0A, which binds to TGTCGAA motifs within oriC and inhibits DnaA–oriC complex formation [72]. Similarly, the essential response regulator MtrA in Mycobacterium tuberculosis binds to oriC via GTCACAGCG moitfs, thereby modulating chromosome replication initiation [73]. Architectural and sequence plasticity of oriC may have enabled the evolutionary incorporation of these species-specific OriBP-binding sites, allowing chromosome replication initiation to be integrated with diverse developmental programs and environmental adaptation.

Typical archaeal oriCs contain two binding regions for the initiator protein Orc1/Cdc6, separated by an AT-rich region [74]. The oriC-bound two Orc1/Cdc6 molecules are suggested to load two MCM helicases onto the double-stranded DNA within the AT-rich region [75]. However, in the oriCs of the crenarchaeal species Aeropyrum pernix (A. pernix), unwinding of the AT-rich region is observed in the presence of Orc1/Cdc6 [76]. Moreover, A. pernix Orc1/Cdc6 binds to the single-stranded DNA of this AT-rich region with sequence specificity [77]. Notably, the experimentally determined unwound region of this species is predicted to have a high transition probability (A. pernix: ~1.5e⁻³), similar to bacterial bilateral oriCs (Supplementary Fig. S9). These observations suggest that some archaeal oriCs as representated by A. pernix oriC may share a helicase-loading principle with the Bilateral-oriC model (Fig. 10B).

Although we demonstrated the adaptability of E. coli mechanisms to non-canonical oriC architectures, differentiation in DnaB loading mechanisms should be noted, reflecting diversity of helicase loaders and primosomal proteins. For example, in Bacillus subtilis, the helicase loader BsuDnaI assembles individual BsuDnaC helicase monomers into a hexameric helicase complex on ssDNA in cooperation with several co-loader proteins and DnaA [78, 79]. Additionally, DciA, which is structurally unrelated to DnaC, represents a widely conserved class of DnaB helicase loaders [80, 81, 82]. Furthermore, E. coli oriC can exploit an alternative “rescue” mode that bypasses the DnaA–DnaB interaction through the primosomal protein PriC, which bears binding activities to ssDNA and DnaB [51]. These observations underscore the diversity of helicase-loading strategies. Therefore, the relationships between hallmarks of oriC architectures and the distinct modes of helicase loading mediated by diverse loaders and primosomal proteins warrant further investigation.

Notes

Present address: Molecular Biology of Archaea, Microbiology, Faculty of Biology, University of Freiburg, Schänzlestr. 1, Freiburg im Breisgau 79104, Germany

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was supported by the Japan Society for the Promotion of Science (JSPS KAKENHI ;JP24K23190, JP23K27131) and a JSPS pre-doctoral fellowship (JP22J11077 to R.Y.). Funding to pay the Open Access publication charges for this article was provided by the JSPS JP23K27131.

Data availability

All data are included in the article and in the Supplementary data.