Molecular analysis of the replication functions of the bifidobacterial conjugative megaplasmid pMP7017

Abstract

pMP7017 is a conjugative megaplasmid isolated from the gut commensal Bifidobacterium breve JCM7017 and was shown to encode two putative replicases, designated here as RepA and RepB. In the current work, RepB was identified as the pMP7017 replicative initiator, as the repB gene, and its surrounding region was shown to be sufficient to allow autonomous replication in two bifidobacterial species, B. breve and Bifidobacterium longum subsp. longum. RepB was shown to bind to repeat sequence downstream of its coding sequence and this region was determined to be essential for efficient replication. Based on our results, we hypothesize that pMP7017 is an iteron‐regulated plasmid (IRP) under strict auto‐regulatory control. Recombinantly produced and purified RepB was determined to exist as a dimer in solution, differing from replicases of other IRPs, which exist as a mix of dimers and monomers. Furthermore, a stable low‐copy Bifidobacterium‐E. coli shuttle vector, pRD1.3, was created which can be employed for cloning and expression of large genes, as was demonstrated by the cloning and heterologous expression of the 5.1 kb apuB gene encoding the extracellular amylopullulanase from B. breve UCC2003 into B. longum subsp. longum NCIMB8809.

Introduction

Bifidobacteria represent anaerobic, non-motile, high G + C% Gram-positive bacteria that belong to the phylum Actinobacteria. Bifidobacteria are considered key commensals in microbe-host interactions, purported to play a pivotal role in maintaining gastrointestinal health through various beneficial or probiotic activities (Leahy et al., 2005; O'Callaghan and van Sinderen, 2016). They have been reported to reduce the risk of microbial infections by enforcing epithelial barrier function through acetate production and secretion of lipophilic antimicrobials (Lievin, 2000; Fukuda et al., 2011). In addition, certain bifidobacterial strains have been shown to interact with their host in vivo via extracellular surface structures (Fanning et al., 2012; Turroni et al., 2013; Schiavi et al., 2016; O'Connell Motherway et al., 2019) and through the production of tryptophan catabolites, such as indole-3-lactic acid via aryl hydrocarbon receptor stimulation (Roager and Licht, 2018; Wong et al., 2020).

These reported health benefits have been the driving force behind a wave of research activities to study the biology and genetics of bifidobacteria and the way in which they interact with their host. Currently, knowledge on the health benefits of bifidobacteria is for a considerable part deduced from in silico analyses due to the recent surge of available bifidobacterial genome sequences, a 16.5-fold increase in four years from 62 (O'Callaghan and van Sinderen, 2016) to 1021 (https://www.ncbi.nlm.nih.gov/taxonomy, accessed May 2020) and to a lesser degree based on findings from functional analyses (Cronin et al., 2007; O'Connell Motherway et al., 2009; Ruiz et al., 2013; Sakanaka et al., 2018; Koguchi et al., 2020). The paucity of functional data is partly due to the inherent resistance of Bifidobacterium to the artificial uptake of DNA due to natural barriers such as restriction-modification systems (Bottacini et al., 2018), but most significantly due to the scarcity of molecular techniques and tools suited to these bacteria (Ventura et al., 2004; Cronin et al., 2011; O'Callaghan and van Sinderen, 2016). Plasmids represent one of these crucial molecular tools, playing an integral role in the expansion of our understanding of gene function. This is in part due to their small size and relatively simple genetic organization, in combination with the ease with which they can be isolated and manipulated in vitro without adversely affecting the host (del Solar et al., 1998). Plasmids are generally defined as autonomously replicating, extrachromosomal genetic elements, which are not essential to host survival, and which may carry a plethora of advantageous functions for rapid environmental adaptation (Sobecky, 1999; Heuer and Smalla, 2012; Dziewit et al., 2015). Plasmids require one or more genes and/or cis-acting sequences, together termed the basic replicon, for replication and faithful transmission to each daughter cell during cell division. The genetic composition of such a basic replicon can vary widely and depends on the mode of replication.

Plasmids were originally thought to be relatively rare in bifidobacteria (Lee and O'Sullivan, 2010), yet a total of 49 plasmids of bifidobacterial origin have been identified to date, isolated from 12 species and ranging in size from 1.8 kb in the case of plasmid pMB1 isolated from Bifidobacterium longum subsp. longum (Rossi et al., 1996) to 190 kb in the case of pMP7017 from Bifidobacterium breve (Bottacini et al., 2015). Furthermore, recent evidence indicates a higher prevalence of megaplasmids in this genus due to the identification of pMP7017 homologues in fifteen B. longum strains (Odamaki et al., 2018). The word ‘megaplasmid’ was initially coined to describe a 450 kb plasmid identified in the Gram-negative rhizosphere bacterium Sinorhizobium meliloti in 1981 (Rosenberg et al., 1981), and in publications to date has been used to describe plasmids > 100 kb (Schwartz, 2009), which appear in both phylogenetically and physiologically diverse groups of archaea and bacteria. While in-depth investigations have been conducted to characterize the replication functions of megaplasmids from Rhizobia (López-Guerrero et al., 2012; Lagares et al., 2014) and Vibrio cholera chromosome II (Stokke et al., 2011; Baek and Chattoraj, 2014; de Lemos Martins et al., 2018; Fournes et al., 2018), relatively little is known about the replication mode of actinobacterial megaplasmids.

The 190 178 bp conjugative megaplasmid pMP7017 of B. breve JCM7017 represents the first and largest, conjugative megaplasmid identified among the actinobacterial gut commensals (Bottacini et al., 2015). The report suggests that pMP7017 is the result of a co-integration of two plasmids, a notion based on the finding that there is a 4.8% deviation in G + C mol% from one arbitrary half to the other, while pMP7017 is also predicted to encode two replicative initiator proteins, denoted here as RepA and RepB. These putative replicases do not correspond or display homology to proteins from the RepABC family of plasmids; thus, this nomenclature is purely for the purpose of differentiation. In the current report, we investigate the replication functions of the conjugative bifidobacterial megaplasmid pMP7017 in an effort to further our understanding of the replicative strategies of this and, by inference, other large plasmids. While research efforts to date have primarily focused on plasmid replication in Gram-negative and pathogenic Gram-positive bacteria, the current work serves to address the knowledge gap of these systems in Gram-positive commensals.

Results In silico analyses of the regions flanking repA and repB

As data pertaining to replication features of Gram-positive megaplasmids and bifidobacterial replicons are scarce, we focused our attention on two previously identified putative replicase genes (Bottacini et al., 2015) to assess the intrinsic features of pMP7017 replication. Inspection of the repA gene (nucleotide coordinates 97 215…96 166; GenBank Accession No. KM406416) and surrounding sequences identified features characteristic of plasmid replicons within the 5’ region of the repA coding sequence, such as an A + T rich region, a 35 bp palindrome and eight 15 bp direct repeats, designated here as A1 through to A8, with the consensus 5′-RGYGTCARWATTCTA-3′ (Fig. 1A). An exploration of the non-coding region upstream of the repB gene (nucleotide coordinates 120 394…119 397) did not unveil any features typical of a plasmid origin. In silico scrutiny of the sequence located downstream of repB revealed a small 267 bp open reading frame (ORF) encoding a hypothetical, basic (pI 11.2) 10 kDa protein without any discerning motifs and referred to here as hptX. Furthermore, seventeen irregularly spaced 20 mer repeats, designated B1 through to B17 (Fig. 1B), were identified conforming to the consensus 5′-hvrTGGmCTATCrsTGhmdh-3′ with an additional sequence, B18, identified outside of the repeat region (nucleotide coordinates 115 314…115 334). Analysis of these repeats revealed that 20 mers B5 and B9 differ in their boundary sequences from the other repeats which appear to contain overlapping inversely orientated sequences, 5′- TTTTACACC-3′, resembling those of putative DnaA boxes observed in the chromosomal origin of Bifidobacterium species (Lee and O'Sullivan, 2010).

Fig. 1. pRD derivative plasmids; stability and copy number.A. Representation of the genomic architecture of repA and flanking regions. Grey arrowheads represent eight direct repeats preceding and overlapping the repA promoter, A1-A8. The region tested for replication is indicated by a red line.B. Genetic organization of repB and surrounding regions. tRNAs are depicted as black arrows, while the seventeen 20 mer repeats, B1-B17, are represented as black arrowheads. The region tested for replication is indicated by a red line. A schematic representation of the pMP7017 replicon within each pRD1 derivative plasmid is depicted below.C. Graph illustrating the segregational stability of the pRD1 derivatives over 132 generations, each data point is averaged from three independent replicates, and standard deviations are indicated as error bars.D. Table of pRD1 plasmid copy numbers and MIC of chloramphenicol resistance.

When replication occurs bi-directionally from a single origin such as with theta-type replication, there are two replication forks of opposite polarity, which have been shown to exhibit a GC-skew (Arakawa and Tomita, 2012; Zhao et al., 2015; Zhang and Gao, 2017). Thus, in an effort to define the location of the replication origin and terminus the GC-skew properties across pMP7017 were plotted. A clear asymmetry was observed with two strong polarity switches at nucleotide positions 117 611 bp and 37 431 bp, presumably representing the approximate locations of the replication origin and terminus respectively. This trend in disparity is reflected in a Z-curve analysis which predicts the oriV to be located at position 120 395 bp (Luo et al., 2019). These findings, therefore, suggest that the replication origin is located in the vicinity of repB.

Analysis of the protein sequence of RepA revealed a RepA_C domain (PF04796) at the N-terminus (interval: 57–215 aa, E-value: 1.6⁻¹⁶). RepB also presents with this domain at its C-terminus (PF04796, interval: 193–361aa, E-value: 1.1⁻¹²), which is part of the Rep_3 superfamily (CDD: cl19398, interval: 193–361aa, E-value: 5.09⁻⁰⁸). Members of this family of replicases are found in a number of IRPs from Gram-negative bacteria such as RK2 (Fang and Helinski, 1991), R6K (Filutowicz et al., 1985), pSC101 (Kazuo and Mitsuyo, 1984), F (Murotsu et al., 1981) and pPS10 (Kazuo and Mitsuyo, 1984), and low GC, non-conjugative plasmids of Gram-positive origin such as pSK639 from Streptococcus epidermidis (Apisiridej et al., 1997). Homology modelling further supports this finding, as 57% of the total amino acid sequence of RepB (residues 102–291) was modelled heuristically at > 90% confidence, using PDB templates of the C-terminus of IRP initiators RepE from F plasmid and the R6K π monomer, represented by c2z9oB (Nakamura et al., 2007) and c2nraC (Swan et al., 2006) respectively. RepB is therefore structurally most similar to replicases of IRPs, which are known to replicate by means of the theta-type mechanism.

The repB gene encodes the pMP7017 replication initiator protein

To establish the potential involvement of either (or both) of the putative replication initiator-encoding genes in pMP7017 replication, the repA and repB genes, in addition to the regions flanking their coding sequences, were individually cloned into the bifidobacterial replication probe vector pBif (Alvarez-Martín et al., 2007), generating pBif_repA and pBif_repB respectively. Recombinant plasmids for each of the replication genes were successfully obtained in E. coli and were then introduced into B. breve UCC2003 by electroporation to assess their ability to independently facilitate replication. Failure to obtain B. breve UCC2003 pBif_repA transformants suggests that the 1829 bp fragment encompassing repA and upstream repeat region was insufficient to allow autonomous replication (Fig. 1A). In contrast, recovery of B. breve UCC2003 transformants containing pBif_repB demonstrated that this 3258 bp region harbours all relevant features to facilitate autonomous replication in Bifidobacterium and thus is assumed to represent the basic pMP7017 replicon (Fig. 1B). To corroborate this finding, we adopted a strategy of site-directed insertional mutagenesis to investigate the dispensability of either repA or repB for replication using a non-replicating vector which integrates into the targeted gene via homologous recombination. Successful disruption of the repA coding sequence, as confirmed by Illumina sequencing, demonstrates that this gene is not essential for pMP7017 replication, while the inability to obtain a repB mutant despite multiple efforts indicates the indispensability of repB.

Iterons B12–B15 and flanking AT-rich region are essential cis-acting elements for pMP7017 replication

To determine the minimal boundaries of the functional repB-associated replicon, successively shorter fragments of the 3258 bp repB-containing replication region were cloned into the Bifidobacterium replication probe vector pBif and assayed for their capacity to support autonomous replication in B. breve UCC2003 (Fig. 1B). In this manner, the minimal replicon was determined to be the 2366 bp fragment of pRD1.4, which contains the repB promoter and coding sequence, hptX and seventeen 20 mer repeats. In order to determine the essentiality of individual iterons and the co-transcribed hptX, five additional constructs were made, pRD1.7 – pRD1.12, represented graphically in Fig. 1B. Plasmid derivatives pRD1.7, pRD1.8, pRD1.9 and pRD1.10 were replication-proficient in B. breve UCC2003, whereas derivatives pRD1.11 and pRD1.12 failed to replicate. It can be deduced from these data that the hptX coding sequence and iterons B2, B3, B4 and B9 are dispensable for replication and that iterons B12-B15 and flanking AT-rich region represent essential cis-acting elements.

Iterons B2, B3 and B4 confer stability of the pMP7017 replicon

While the hptX gene, iterons B2, B3 and B4, and iteron B9 individually were shown to be non-essential for pMP7017 replication, these features may nonetheless play a role in other functions, such as plasmid segregation, stability and/or copy number. To ascertain the role of these elements, if any, in the stability of the replicon, B. breve UCC2003 harbouring each of the replication-proficient pRD derivatives was grown in the absence of selection pressure for 132 generations and then assayed for the presence of the plasmid as a measure of stability, presented in Fig. 1C. These data show that plasmids containing all putative iterons, pBif_repB, pRD1.2, pRD1.3, pRD1.4 and pRD1.7, are quite stable as they are retained in the population for at least 132 generations without selective pressure. Instability was observed for plasmids pRD1.9, lacking two iterons (B2 and B3), and pRD 1.10 lacking three iterons (B2, B3 and B4) with a total loss of the plasmid occurring at ≤ 52 and ≤ 35 generations respectively. Furthermore, the absence of iteron B9 from plasmid pRD1.8 was shown to cause only a minimal loss of stability. The removal of the hptX gene appears to have no adverse effect on stability nor does the absence of the region located 5′ of the repB promoter. Plasmids pRD1.2 and pRD1.3 were shown to be the most stable among the assessed plasmids as they were maintained in > 85% of cells at 100 generations. Together these findings indicate that the iterons contribute to plasmid stability and highlight the 2542 bp fragment of pRD1.3 as the basic replicon due to the greater stability of this plasmid derivative, apparently containing all features necessary to support replication and ensure stable maintenance.

Iterons B2, B3, B4 and B9 do not participate in copy control

The plasmid copy number (PCN) of the eight replicating pRD derivatives and megaplasmid pMP7017 was determined by the relative quantification method using single-copy housekeeping genes on the B. breve UCC2003 chromosome as a reference. As data displayed in Fig. 1D illustrate, the PCN of all pRD1 plasmids approximate 1 copy per chromosome, with the exception of pRD1.3 with a somewhat higher value of 1.68 ± 0.16. These data illustrate the copy number of all pRD1 derivates is comparable to that of pMP7017 from which the replicon derives, existing at approximately 1 copy/chromosomal equivalent in B. breve UCC2003, irrespective of the absence of the 5′ region, hptX or iterons B2, B3, B4 and B9, thus suggesting these features do not play a major role in copy control. These non-essential iterons may nonetheless encode functions not explored in this work, considering their lack of participation in copy control they will subsequently be referred to as direct repeats.

repA and repB are transcriptionally regulated in pMP7017

To gain insight into repA and repB transcription and possible transcriptional regulation, the presumed promoter regions and 5′ UTRs of each rep gene were cloned upstream of the promoterless β-glucuronidase (β-GUS)-encoding gene in the reporter plasmid pNZ272 (Platteeuw et al., 1994; Cronin et al., 2007), generating plasmids pNZPP_repA and pNZPP_repB respectively. These plasmids were individually introduced into B. breve UCC2003 and B. breve UCC2003 pMP7017_199, facilitating transcriptional quantification of these promoter regions in both the absence and presence of the megaplasmid. B. breve UCC2003 pMP7017_199 contains a tetracycline-tagged pMP7017 derivative with an insertion in ORF199 (Bottacini et al., 2015) encoding a surface-associated protein which displays no phenotypic deviation from the wild type and is used in the current work for the purpose of antibiotic-based plasmid selection. A bar chart of the data in Fig. S1A illustrates a moderate level of transcriptional activity from both promoters in pNZ272 when present in B. breve UCC2003 pMP7017_199, while the intracellular β-GUS activity of the pNZ272 transcriptional fusions in a B. breve UCC2003 background was higher. B. breve UCC2003 pNZPP_repA revealed a marked increase in activity ˜ 3.7-fold from its B. breve UCC2003 pMP7017_199 counterpart, while activity from the repB promoter increased ˜1.6-fold in the absence of the megaplasmid. These findings indicate that the promoters of both repA and repB are repressed in strain B. breve UCC2003 pMP7017_199, suggesting that pMP7017 encodes trans-acting factors negatively regulating repA and repB transcription. To precisely define the location of the repA and repB promoters, the repA and repB transcription start sites (TSSs) were experimentally determined by primer extension (PE) analysis. A single extension product was obtained for repA corresponding to a position 71 nucleotides upstream of the predicted translational start site (Fig. S1B). Investigation of the preceding sequence revealed hexamers resembling the consensus −10 and −35 recognition sequences of bifidobacterial vegetative promoters (Bottacini et al., 2017). PE analysis for repB resulted in an extension product 131 nucleotides 5′ of the predicted translational start site (Fig. S1CC) that is associated with a putative vegetative promoter sequence.

RepA binds to its corresponding promoter region

While it was unclear what role, if any, RepA plays in pMP7017 replication, mobility shift electrophoresis was performed to determine whether the RepA protein interacts with probe 124P, containing the repA promoter and eight direct repeats A1–A8. The electrophoresis mobility shift assay (EMSA) depicted in Fig. 2A clearly shows the reduced mobility of DNA probe 124P upon addition of RepA, highlighting the affinity of RepA for this region. Assuming that the eight direct repeats contained within the probe are the recognition sequences of RepA, the observation of one strong shift despite eight potential binding sites suggests that RepA binding involves cooperativity, hence site saturation appears instantaneous rather than incremental. To further elucidate this interpretation, the fraction of bound probe (Φ) was calculated and plotted as a function of protein concentration to generate a specific binding curve. From the curve obtained, we can see that RepA-124P complex formation is a non-linear function of RepA concentration consistent with cooperative interaction, and further analysis revealed a Hill coefficient of 6.27 ± 1.36 reflecting marked positive cooperativity. Together this substantiates the notion that binding of RepA to each of the eight repeats within probe 124P does not occur independently. Furthermore, an apparent K_d of ˜ 121.8 nM ± 8.0 nM was determined, from this the K_a was calculated to be 5.26 pM⁻¹ suggesting a high affinity of RepA for its DNA substrate.

Fig. 2. The binding behaviour of RepA to its corresponding promoter region and of RepB to multi-site probes in EMSA.A. EMSA image displaying the interaction of RepA with promoter probe 124P. The binding isotherm and Scatchard plot (inset) of this data are presented to the right.B. A representative EMSA gel displaying two clear shifts representing the binding of RepB to the 2-site probe, F, containing two tandemly arranged iterons. The binding isotherm and Scatchard plot (inset) of this data are presented to the right.C. EMSA image of RepB binding to 2-site probe, I, containing two invertedly orientated iterons. The binding isotherm and Scatchard plot (inset) of this data are presented to the right.

RepA exists predominantly as a homodimer in solution

The pronounced cooperativity observed in RepA binding to its cognate recognition sequence signalled a propensity for RepA-RepA interaction. To establish its oligomeric state in solution, size exclusion chromatography was performed. From the chromatogram (Fig. S1D), one main peak was observed at 17.87 RV (ml) representing a molecular weight of 82.49 kDa and 67.31 % of the total protein, suggesting RepA exists predominantly as a homodimer. Two additional peaks can be seen at 164.8 kDa (17.23 RV; representing 16.98% of total protein) and 328 kDa (16.57 RV and representing 15.71% of the total protein) which we presume denote tetrameric and octameric forms of RepA, respectively, thus illustrating the multimerization ability of this protein.

RepB does not display any affinity for its promoter or coding region

In an effort to ascertain where within the pMP7017 replicon RepB binds and to assess whether the presumed RepB replication protein was able to physically interact with any of the identified DNA features flanking its coding sequence, nine IRD700-labelled probes were generated, designated here as A through to I, encompassing various segments of the replicon (Table S2). EMSAs with probes A and B, encompassing the repB promoter-containing region, and probes C and D, spanning the repB coding sequence, revealed an absence of binding activity (Fig. S2).

RepB binds specifically to iterons

In order to establish whether RepB binds directly and specifically to the putative iterons identified 3′ of its coding sequence, EMSAs were performed with DNA probes E through to I, each containing ≥ 1 20 mer repeat. EMSAs with each probe within this region resulted in clear mobility shifts, indicating that RepB is able to bind to the repeat-containing region downstream of the repB gene. EMSAs with short DNA fragments E and H (Fig. S3A and B), each containing an individual 20 mer site, resulted in a single mobility shift interpreted as a single species of protein–DNA complex, RepB-E and RepB-H respectively. To determine the quantitative parameters of these interactions, a titration curve was generated from the experimental data and fit to the Hill equation revealing a coefficient of ˜ 1 in both cases with an apparent K_d of 254 ± 34 nM and 236.6 ± 66.5 nM for probe E and probe H respectively. This result suggests an approximate K_a of 4.1 pM⁻¹ ± 0.3 pM, reflecting high-affinity binding of RepB to the 20 mer sites B1 and B9. Additionally, the near equivalent estimations of dissociation suggest no predilection in terms of site orientation, while highlighting the interchangeability of residues 13 (A/G), 14 (C/G) and 18 (C/A) within the apparent recognition sequence, without any significant effect on affinity estimates.

RepB exists as a dimer in solution with no monomeric fraction

While quantitative evaluation of the binding isotherms allows for affinity approximation and estimation of the effective cooperativity behaviour, it was necessary to determine the oligomeric state of the protein to support our interpretation of the graphical presentations of the ligand-binding data. To achieve this, size exclusion chromatography was performed, revealing that when RepB was assayed at a high protein concentration (0.8 mg ml⁻¹), it exists as a homodimer in solution illustrated by one main peak at 18.75 retention volume in the chromatogram (Fig. S1E) representing a molecular weight of 79.85 kDa and 97% of the total sample.

The binding behaviour of RepB to multi-site probes illustrates cooperative interactions

The EMSA (Fig. 2B) employing DNA probe F containing presumed binding sites B2 and B3 clearly depicts two distinct electrophoretic species, which can be explained by assuming that the initial shift is the result of a single binding event, while the double shift occurs due to subsequent RepB binding to the second site, representing nucleoprotein complexes RepB-F and 2RepB-F respectively. Analyses of the titration curve data generated from the binding reactions suggest positive cooperativity (nH = ˜ 2), with an apparent dissociation constant of 13.7 nM ± 1.8 nM for RepB-F and 21.8 nM ± 4.9 nM for 2RepB-F. Correspondingly, the EMSA displayed in Fig. S3C demonstrates incremental binding of RepB to the 20 mer sites B2, B3 and B4 within probe G. Global sensitivity measures of the specific binding curves reveal positive cooperativity in the formation of complexes RepB-G (nH = 1.5 ± 0.34) and 2RepB-G (nH = 2.8 ± 0.62); conversely, a Hill coefficient of ˜1 associated with 3RepB-G suggests independent binding at a third site. In consolidation with the results obtained from the EMSA employing DNA fragment F, it can be inferred that positive cooperativity occurs between sites B2 and B3 due to their relative proximity (3 bp intervening), while site B4 is bound independently. Although cooperativity has been indicated for contiguous tandem sites, it was interesting to establish what binding behaviour would be observed with convergent proximal sites on opposing strands, a spatial arrangement observed for sites B16 and B17. EMSAs with probe I containing these sites display the formation of two protein–DNA complexes, RepB-I and 2RepB-I (Fig. 2C). The persistence of nucleoprotein complex RepB-I, despite increasing RepB concentration, may be indicative of a faster K_off rate, potentially as a result of steric hindrance. Together these data illustrate an increased affinity of RepB for one recognition sequence in the presence of a second adjacent site, being reflected by an average ˜ 20.7-fold reduction in the apparent K_d of the first binding event in addition to nH ≥ 2 suggesting two interacting sites.

GGnCnAnnR are critical bases within iterons for binding of RepB

To verify that RepB specifically binds to the putative iterons downstream of its coding sequence, thirteen core nucleotides (4–16) within the 20 mer motif were individually substituted by PCR site-directed mutagenesis, and each probe (P1–P18) was then assessed for its ability to bind RepB by EMSA. The results of this assay, displayed in Fig. 3, revealed that substitutions of residues 5G>T (P2), 6G>T (P3), 8C>A (P5), 10A>C (P7) and 13G>T (P10) completely abolished the ability of RepB to bind, indicating that these residues are essential for RepB–iteron complex formation. Furthermore, the substitution of the remaining residues resulted in RepB–iteron binding events exhibiting a range of much-lowered affinities, which suggests a role in the stability of the complex. These data clearly show that alterations within the motif impact on RepB-mediated DNA binding affinity and indicate that RepB specifically binds to GGnCnAnnR.

Fig. 3. Identification of the critical nucleotide residues for the binding of RepB to iterons. A bar chart representing the percentage of probe bound by RepB, the presence or absence of RepB in the reaction is denoted by a + or – symbol respectively. The sequence of each probe is given, and the incorporated mutation/s are indicated in red text. An EMSA gel image representing a replicate from which the bar chart is based is displayed at the bottom.

The RepB C-terminal domain is required for dimerization and DNA binding

In silico structural homology modelling revealed a strong positive correlation of the C-terminus of the predicted 3D structure of RepB to the conserved iteron-binding domains of plasmid initiators RepE from F plasmid (Nakamura et al., 2007) and π from R6K (Swan et al., 2006) (Fig. S4A). Furthermore, structural analyses indicate RepB to be pseudo-symmetric with two putative DNA binding domains, a central winged helix-turn-helix 1 (wHTH1) and a C-terminal HTH2 flanked by a low complexity region rich in proline, serine and threonine residues (PST) (Fig. S4B). Oligopeptides enriched in these amino acids are known to form flexible structures which participate in ligand binding under specific conditions (Matsushima et al., 2008); additionally, PST residues are known to serve as phosphorylation sites modulating protein activity (Yaffe and Smerdon, 2001; Johnson, 2009). To define the role of the C-terminal domain in both DNA binding and protein–protein interactions, a deletion mutant, RepB_∆C, was created. RepB_∆C was overexpressed and purified, followed by size exclusion chromatography to establish its oligomeric tendency in solution. From the chromatogram (Fig. S4C), one peaks at 18.7 RV (ml) representing 28.81 kDa and 100% of the sample, indicating RepB_∆C exists as a monomer in solution. Purified RepB_∆C was then tested for its ability to bind iterons by EMSA. The deletion variant showed no detectable binding to probe P17 as compared to the RepB control, revealing a functional requirement of the C-terminal 98 residues of the RepB protein in dimerization and DNA binding (Fig. S4D).

Highly conserved RepA and RepB homologues are present in all bifidobacterial megaplasmids

A comparative analysis of the RepA and RepB protein sequences in twelve recently identified bifidobacterial megaplasmids confirmed the presence of both RepA and RepB homologues in all plasmids. RepB is conserved at a high level (protein per cent identity range: 88.75–89.43%), while, interestingly, RepA shows a higher level of conservation (protein per cent identity range: 89.97–96.56%). This indicates a potential functional importance of RepA despite its dispensability for pMP7017 replication.

The putative pMP7017 origin of replication is conserved between bifidobacterial replicons

A comparative analysis of the nucleotide sequence of the pMP7017 basic replicon between bifidobacterial megaplasmids, displayed in Fig. S5, illustrates the preservation of the repB coding sequence and iteron-containing region. Each replicon retains a minimum of eleven iterons with between one and three additional iterons located outside the replicon (data not shown). Megaplasmids from B. longum subsp. longum strains, MCC10043, MCC10070 and MCC10096 contain two genes, orfA and orfB, constituting an IS200 transposable element that is not present in the other plasmids. The hptX gene identified in the pMP7017 replicon is absent in all B. longum megaplasmids, supporting the experimental data that this hypothetical gene is not required for replication. In consideration of these data, the hptX, orfA and orfB coding sequences were removed and the nucleotide sequences aligned to better assess the iteron-containing region in terms of its conservation (Fig. S6). The alignment displays the absence of sequences corresponding to pMP7017 20mers B2, B6, B7 and B9 in one or more B. longum megaplasmids, while a moderate level of variability is observed in the intervening sequences between the first ten repeats (B1–B10) which is not observed between the remaining iterons. Furthermore, no sequence resembling the DnaA consensus motif was identified in the B. longum megaplasmids. Analysis of the curvature, bendability and GC content of this highly conserved region reveals a conspicuous peak of curvature (> 7.5) from positions 118 659 bp to 118 686 bp, overlapping iterons B12, B13 and B14, followed by a region of high bendability in the AT-rich region with a GC content 14% less than the average from positions 118 772 bp to 118 264 bp.

pRD1.3: a novel bifidobacterial shuttle vector for cloning of large fragments

To verify the suitability of the pRD1 plasmids for cloning and expression purposes in bifidobacteria, the apuB gene from B. breve UCC2003, encoding an extracellular type II amylopullulanase (Motherway et al., 2008), was directionally cloned into the multiple cloning site of plasmid pRD1.3, resulting in plasmid pRD1.3_apuB. Both pRD1.3 and pRD1.3_apuB were introduced into B. longum subsp. longum NCIMB8809, a strain unable to metabolize pullulan (Ryan et al., 2006; Arboleya et al., 2018). B. longum NCIMB8809 pRD1.3 and B. longum NCIMB8809 pRD1.3_apuB were then assessed for their ability to utilize pullulan by measuring optical density after 24 h of incubation in MRS supplemented with the sugar. From Figure 4A, it is obvious that B. longum NCIMB8809 pRD1.3 shows little to no growth (final OD₆₀₀ 0.13 ± 0.013), whereas B. longum NCIMB8809 pRD1.3_apuB displays substantial growth (final OD₆₀₀ 1.80 ± 0.16). The results show that pRD1.3_apuB confers pullulan-dependent growth in B. longum NCIMB8809, which validates the capacity of pRD1.3 as a novel bifidobacterial cloning and expression vector (Fig. 4B). Plasmid pRD1.3 is presumed to carry all replication and stability functions of the pMP7017 megaplasmid, and it may therefore be possible to employ pRD1.3 for cloning large (> 25 kb and up to 100 kb) DNA fragments.

Fig. 4. Development of a novel bifidobacterial shuttle vector, pRD1.3.A. Bar chart displaying the final optical density values of B. longum NCIMB8809 pRD1.3 and B. longum NCIMB8809 pRD1.3_apuB growth after 24 h of incubation with 10 % w/v pullulan as a sole carbon source.B. Genetic map of plasmid pRD1.3 indicating the relevant features, the pMP7017 replicon is highlighted by a grey bar.

Discussion

In the current work, repA was found to be non-essential for pMP7017 replication through successful site-directed mutagenesis, in addition to the inability of repA and surrounding regions to support autonomous replication in a replicase-deficient vector. As this gene is not critical for pMP7017 replication, repA may be a gene remnant encoding a dysfunctional replicase despite its high level of conservation among bifidobacterial megaplasmids. RepA was determined to exist predominantly as a dimer in solution with a tendency towards multimerization and was shown to bind its cognate promoter with high affinity.

We show that replication of the bifidobacterial megaplasmid pMP7017 depends on the plasmid-encoded initiator RepB, which exists as a dimer in solution at high concentrations with no observable monomeric fraction. RepB was demonstrated to be competent in iteron binding, displaying both iteron-dependent protein–protein interaction and iteron-independent interactions. The critical residues within the iteron sequence which facilitates the binding of RepB were established to be GGnCnAnnR, situated asymmetrically across the core of the 20 mer iteron. In all characterized IRPs, with the exception of the R6K π initiator, Rep proteins bind iterons as monomers, while Rep dimers are incompetent in iteron binding (Kunnimalaiyaan et al., 2004). The study of Rep–iteron interactions is predominantly achieved with the use of dimerization-defective Rep mutants, through the addition of molecular chaperones in the binding reactions (DasGupta et al., 1993; Ishiai et al., 1994; Chattoraj et al., 1996) or the use of chaotropic agents to induce dimer dissociation and facilitate the accumulation of iteron-binding active monomers. Interestingly, HIS-tagged WT RepB was shown to associate with a high affinity to iterons in EMSA in the absence of stimulation by host cell chaperones or the use of chaotropic agents. In addition, we did not observe a reduction in binding upon increasing RepB concentration nor did we observe a second shift representing a higher-order nucleoprotein complex in EMSA with single-site probes. Thus, it is tempting to speculate that the binding observed in this work is due to dimeric RepB. If this assumption is true, binding of iterons by dimeric RepB may inhibit replication as observed in the replication of plasmid R6K (Kunnimalaiyaan et al., 2004). Dimeric RepB may then rely on the participation of temporally expressed cell-cycle chaperones, such as DnaK/DnaJ for monomerization as described for many initiators of IRPs (DasGupta et al., 1993; Chattoraj, 2000; Zzaman et al., 2004; Nakamura et al., 2007).

The C-terminal defective mutant, RepB_∆C, was determined to be both monomeric and incompetent in iteron binding. While this suggests that the 98 aa C-terminal end of RepB is required for oligomerization and DNA binding, the observed phenotype may also be due to protein instability or indeed signify a lack of affinity of RepB monomers for the 20 mer sites in favour of alternative sites. It has previously been reported that even a single point mutation within the C-terminal domain of structurally similar replication initiators, i.e. π from R6K plasmid and RepE from F plasmid, is enough to severely affect or abolish the ability of the protein to bind iterons (Komori et al., 1999; Swan et al., 2006). Thus, a more conservative approach to mutagenesis is necessary to substantiate this finding.

While the pMP7017 minimal replicon was defined as a 2366 bp region (corresponding to pMP7017 nucleotide coordinates: 118 220…120 584) encompassing the repB coding sequence and seventeen repeats spanning 1159 bps, it was established that not all features contained within the replicon are essential. Deletion of the hptX coding sequence did not have any negative effect on replication. In addition, it was demonstrated that 20 mers B2, B3, B4 and B9 are not essential. While repeats B2-B4 were shown to participate in the stability of the replicon, the removal of these sequences displayed no perceived effect on plasmid copy number suggesting the pMP7017 copy control functions are encoded by alternative iterons. Contrastingly, the removal of iterons B12–B17 or the AT-rich region located between iterons B15 and B16 abolished the plasmid’s capacity for replication, indicating this region contains essential cis-acting elements and thus likely constitutes the pMP7017 origin of replication. These data are supported by a comparative analysis of the pMP7017 basic replicon between bifidobacterial megaplasmids which revealed a high degree of variably between and within repeats constituting the putative stability region, which is not observed in the highly conserved region containing the essential iterons B12–B17.

Although experimental validation is required to fully unravel the intricacies of pMP7017 replication, we here propose a mechanistic model for this bifidobacterial replicon (Fig. 5). We hypothesize that the pMP7017 iterons constitute a framework for the binding of RepB monomers that interact to form a structure that disrupts the DNA helix, facilitating DNA unwinding at the flanking AT-rich region. Reminiscent of the layout of DnaA boxes within the oriC of bacterial genomes, RepB iterons are arranged with varying affinities, distances and orientations suggesting a dynamic system regulating the formation of specific higher-order nucleoprotein complexes in the occlusion and stimulation of origin opening (Ekundayo and Bleichert, 2019). In addition to their importance for RepB assembly at the origin, the difference in sequence may constitute additional regulatory measures through weaker vs stronger binding sites. Furthermore, iteron-bound Rep protomers may inhibit origin melting as a result of coupling two origins in a process of replication inhibition, referred to as ‘handcuffing’. This dual functionality of iterons is a copy control strategy commonly adopted by large, low copy number plasmids preventing plasmid over-replication through origin pairing (Chattoraj, 2000).

Fig. 5. Working model of pMP7017 replication initiation. Iteron saturation by RepB dimers prevents origin firing. Cell-cycle chaperones facilitate the conversion of RepB dimers to active monomers. Cooperative interaction (represented by green arrows) of RepB monomers bound to tandemly arranged iterons facilitates the unwinding of the flanking AT-rich region. While iteron-bound monomers on opposite template strands may participate in oligomerization to prevent over-replication in a process termed ‘handcuffing’, iterons are depicted as solid black lines and non-essential direct repeats are shown as broken arrows.

As the availability of genetic tools for the study of bifidobacteria is limited, the characterization of bifidobacterial replicons is necessary to develop new cloning and shuttle vectors. Considering the pMP7017 replicon is sufficient to drive replication of a plasmid of > 190 kb, it can be inferred that pRD1.3 represents a novel low-copy shuttle vector with the capacity to allow cloning of large fragments in bifidobacteria without instability or toxicity issues.

Experimental procedures Bacterial strains, plasmids and cultivation conditions

Bacterial strains and plasmids are listed in Table S1. Bifidobacterial strains were routinely cultivated in reinforced clostridial medium (RCM [Oxoid, Basingstoke, Hampshire, United Kingdom]) or modified de Man, Rogosa and Sharpe (mMRS) medium prepared from first principles (Egan et al., 2018). The mMRS medium was supplemented with cysteine-HCl (0.05 % [wt/vol]) and lactose (1.0 % [wt/vol]) (Sigma) prior to inoculation. Bifidobacterial strains were cultivated in an anaerobic hood (Davidson and Hardy, Belfast, Ireland) for incubation at 37°C, maintaining an anaerobic environment. All Escherichia coli strains were cultivated exclusively in Luria–Bertani (LB) broth (Sambrook and Russell, 2001) at 37°C with agitation. Where applicable, growth media contained tetracycline (10 μg ml⁻¹), ampicillin (100 μg ml⁻¹), erythromycin (100 μg ml⁻¹), chloramphenicol (5 μg ml⁻¹) or kanamycin (50 μg ml⁻¹).

DNA manipulations and molecular cloning

Chromosomal DNA was isolated from B. breve JCM7017-199, a B. breve JCM7017-derivative strain containing a tetracycline resistance marker disrupting the open reading with locus tag pMP7017_199 of plasmid pMP7017, as previously described (Bottacini et al., 2015). Plasmid DNA from E. coli was extracted using the High Pure Plasmid Isolation Kit (Sigma-Aldrich, Gillingham, UK). Standard oligonucleotides (Table S2) used in this study were synthesized by Eurofins (Ebersberg, Germany). Both standard and colony PCRs were performed using DreamTaq PCR 2× Master Mix (Thermo Fisher Scientific, Gloucester, United Kingdom), and colony PCRs were carried out as described previously (Mazé et al., 2007). High-fidelity PCR was achieved using Q5 High-Fidelity 2X Master Mix (New England Biolabs, Hitchin, United Kingdom). PCR fragments were purified using the High Pure PCR Product Purification Kit (Sigma). Restriction endonucleases and T4 DNA ligase were used according to the manufacturer’s instructions (Roche, Mannheim, Germany). Electroporation of plasmid DNA into Bifidobacterium spp. was conducted as previously described (Rossi et al., 1997), while E. coli transformations were performed as described by Sambrook et al. (1989). All electroporation experiments were carried out using a GenePulser apparatus (Bio-Rad Laboratories, Richmond, CA, USA).

Insertional mutagenesis

Insertional mutagenesis was performed based on a previously described protocol (O'Connell Motherway et al., 2009). For this purpose, a 525 bp internal fragment of ORF pMP7017_124 (repA) from codon positions 92 to 267 of the repA coding sequence and an internal 501 bp fragment of ORF pMP7017_146 (repB) from codon positions 101 to 268 of the repB coding sequence were amplified by high-fidelity PCR using B. breve JCM7017_199 total DNA as a template with primer combinations repAmut_F and repAmut_R, and repBmut_F and repBmut_R respectively (Table S2). The amplicons were ligated into the multiple cloning site of the integration vector pORI19 following HindIII and XbaI digestion. 5 µl of each ligation mixture was introduced into E. coli EC101 competent cells by electroporation. Recombinant E. coli derivatives were selected by blue-white screening on LB agar supplemented with 40 g ml⁻¹ X-gal (5-bromo-4-chloro-3-indolyl-d-galactopyranoside) and 1 mM IPTG. Confirmation of recombination was achieved by restriction analysis prior to subcloning of the tetracycline resistance gene, tetW, from plasmid pAM5 into the unique SacI restriction site of each plasmid. The resulting constructs, designated pORI19_repA and pORI19_repB, were introduced into B. breve JCM7017 by electroporation, and transformants were selected on tetracycline-containing media. Site-specific recombination in such tetracycline-resistant transformants was confirmed by colony PCR with primers tetW_F and tetW_R to verify the presence of the tetW gene. Total DNA was isolated from such tetW-positive transformants and used as a template for PCR with primer pair repAconf and TetR_final to confirm integration within the expected genetic position of the megaplasmid. The B. breve JCM7017_repA mutant was then verified by genome sequencing performed employing an Illumina MiSeq Sequencing Platform.

Plasmid constructions

For the construction of plasmids pBif_repA, pBif_repB, pRD1.2, pRD1.3, pRD1.4, pRD1.5 and pRD1.6, various DNA fragments representing sections of the repA or repB gene and surrounding sequences were obtained by PCR amplification using total DNA isolated from B. breve JCM7017 as a template in all reactions and primer combinations as listed in Table S2. Recombinant DNA sequences for plasmids pRD1.7, pRD1.8, pRD1.9 and pRD1.10 were generated by overlap extension PCR containing two PCR segments with primer combinations outlined in Table S2. Oligonucleotides were designed with an SphI restriction site incorporated at the 5′ end to allow direct cloning of amplicons. The generated amplicons and the bifidobacterial replication probe-E. coli vector pBif (Alvarez-Martín et al., 2007) were digested with SphI and ligated. Ligation mixtures were dialysed and transferred into E. coli XL1blue competent cells by electroporation. pBif consists of a chloramphenicol resistance cassette derived from plasmid pC194 (Sangrador-Vegas et al., 2007) in combination with the pSC101 replication functions of pWSK29. Colony PCR was performed to confirm the presence of the repA or repB gene using primer pairs pMP_0124F and pMP_0124R, or pMP_0146F and pMP_0146R, designated to generate a 237 bp or 244 bp amplification product respectively. Purified miniprep plasmid DNA of positive clones was used as a template for a second round of PCR with primer pRDconf_F in combination with the forward or reverse repA or repB cloning primers to determine the orientation of the gene, and plasmids containing repA or repB in the forward orientation were selected for further analysis.

Primer pair apuBF_Xba and apuBR_H3 was used to amplify a 5437 bp region containing the apuB coding sequence and promoter region from B. breve UCC2003 genomic DNA. Both the resulting apuB-encompassing amplicon and pRD1.3 plasmid were digested with restriction enzymes Xba1 and HindIII prior to ligation and dialysis. 5 µl of ligation mixture was electroporated into E. coli XL1blue competent cells, and the cells were resuspended in 950 µl of LB broth and incubated for 1 h incubation at 37°C shaking (160 r.p.m.) prior to spread-plating on LB agar containing 5 μg ml⁻¹ of chloramphenicol and 10 μg ml⁻¹ of tetracycline. Colony PCR was performed with primers apuBF_Xba and apuBR_H3 to select colonies containing the insert, prior to plasmid preparation and sequencing. Plasmid pRD1.3_apuB was isolated and transformed into B. longum NCIMB8809 by electroporation.

Overexpression constructs pQE30_repA and pQE30_repB were produced by cloning each coding sequence in frame with the plasmid-encoded N-terminal hexa-histidine using primer pairs pQE30-124N_Sph1 for repA, and pQE30-124N_HindIII or pQE30-146N_Sph1 and pQE30-146N _HindIII for repB.

To construct plasmids pNZPP_repA and pNZPP_repB, 200bp preceding the predicted translational start site of each gene was amplified by PCR with primer pairs 124PP_BgLII and 124PP_Pst1, or 146PP_BgLII and 146PP_Pst1 respectively. The resulting amplicons were cloned upstream of the promoterless β-glucuronidase gene in reporter plasmid pNZ272 and transformed into L. lactis NZ9000. The recombinant plasmids were then isolated and introduced into B. breve UCC2003 and B. breve UCC2003 pMP7017_199 by electroporation.

To create a C-terminal deletion in RepB, 669 bp of the repB gene excluding 297 bp of the 3′-end of this gene (representing the C-terminal amino acid positions 235 through to 332) was amplified by PCR using primers pQE146C-_SphIF and pQE146C-_H3R. The resulting amplicon was cloned into pQE30 in a manner to fuse a histidine tag to the N-terminus of the encoded, C-terminally truncated RepB protein (designated here as RepB∆C), resulting in the expression construct pQE30_RepB∆C.

The sequence integrity of inserts in all recombinant plasmids was checked by Sanger sequence analysis (performed by MWG, Ebersberg, Germany).

Plasmid stability

The stability of the pRD constructs was assayed by growing B. breve UCC2003 strains harbouring each of the individual constructs in non-selective media with successive subcultures at 12 h for approximately 145 generations. At regular time intervals, 100 µl of each culture was serially diluted to 10⁻⁹ and spot plated on both non-selective and selective agar plates. Colonies were counted, and the CFU/ml obtained on both non-selective and selective plates was determined. Total DNA was isolated from five resistant colonies for each construct and plasmid content assessed by PCR with primers pRDconf_F and pRDconf_R.

Carbohydrate utilization assay

To assess the pullulan-dependent growth potential of strains B. longum NCIMB8809 and B. longum NCIMB8809 pRD1.3_apuB, modified de Man Rogosa and Sharpe (mMRS) medium was freshly prepared from first principles (Man et al., 1960) and supplemented with 1 % (v/v) of pullulan solution and 0.05 % (v/v) of l-cysteine. This medium was inoculated with 1 % (v/v) of stationary phase B. longum NCIMB8809 or B. longum NCIMB8809 pRD1.3_apuB cells, from cultures cultivated overnight in 1 % (v/v) glucose. The cultures were grown anaerobically at 37°C, and optical density was measured at OD_600nm after 24 h using a UV-1280 spectrophotometer (Shimadzu Corporation, Kyoto, Japan). Analyses were carried out in triplicate.

Antibiotic resistance of vectors

The minimal inhibitory concentration (MIC) of chloramphenicol supported by the constructs in B. breve UCC2003 as a host was measured by the E-test method according to the manufacturer’s instructions (AB Biodisk, Solna, Sweden). MIC assays were performed using Reinforced Clostridial Agar (RCA [Oxoid, Basingstoke, Hampshire, United Kingdom]).

Determination of relative plasmid copy number

The relative copy number of pRD1-derived plasmids was assessed by quantitative real-time PCR using the culture and PCR conditions reported previously (Lee et al., 2006). Amplification and detection were performed using a Fast real-time PCR system (Applied Biosystems, Foster City, CA) with QuantiNova SYBR Green PCR Kit (Qiagen, Manchester, UK). Primers 146A_F and 146A_R and, 146B_F and 146B_R (Table S2) were designed based on the pRD1 repB sequence, resulting in amplicons of 120 bp and 150 bp respectively. The 1-deoxy-d-xylulose 5-phosphate synthase gene (dxsA) (GenBank Accession No. ABE95261) and the housekeeping gene dnaA of B. breve UCC2003 were used as single-copy chromosomal reference genes, respectively, generating a 120 bp amplicon with primers dxsA_F and dxsA_R, and a 150 bp amplicon with primers UCC776B_F and UCC776B_R (Table S2). The relative copy number of the derivatives was calculated using the formula N_relative = 1 + E^−ΔCT, where E is the amplification efficiency and ∆CT is the difference between the threshold cycle number of the dxsA- and dnaA-targeting reactions and that of the repB target. Experiments were performed in triplicate; mean results are indicated.

Heterologous protein overexpression and purification

RepA, RepB and RepB∆C were overexpressed and purified as N-terminal hexa-histidine fusion proteins using pQE30, an E. coli overexpression vector conferring an N-terminal six histidine tail to facilitate downstream purification. E. coli EC101 transformed with pQE30-repA, pQE30-repB or pQE30_RepB∆C was cultured in LB medium supplemented with 100 μg ml⁻¹ ampicillin for growth at 37°C until the logarithmic phase (at OD_600nm of ˜ 0.5–0.6) and induced by IPTG at a final concentration of 0.2 mM for 2 h at 37°C. Cells were harvested at 5000 × g for 15 min at 4°C and the pellet resuspended in 10 ml lysis buffer (50 mM Tris-HCl, 50 mM CaCl₂, 300 mM NaCl, 10 mM imidazole, pH 8.0). The cells were disrupted with 3 × 1 min rounds of bead-beating using the Mini-BeadBeater-16 (BioSpec Products, Bartlesville, Oklahoma, USA) with 1 min intervals on ice. Cellular debris was removed by centrifugation at 12 000 × g for 30 min at 4°C. Proteins were purified from crude cell extracts via metal affinity chromatography.

Electrophoretic mobility shift assays (EMSAs)

EMSAs were performed essentially as described previously (Hamoen et al., 1998). Binding reactions were carried out in a final volume of 20 μl in the presence of poly[d(I-C)] in binding buffer (20 mM Tris-HCl [pH 7.0], 5 mM MgCl₂, 0.5 mM DTT, 1 mM EDTA, 200 mM KCl, 10 % glycerol). Varying amounts of purified protein and the fixed concentration of probe were mixed on ice and subsequently incubated for 30 min at 37°C. Samples were loaded onto a 6 % non-denaturing polyacrylamide (PAA) gel prepared in TAE buffer (40 mM Tris-acetate [pH 8.0], 2 mM EDTA) and run in a 0.5–2.0 × gradient of TAE at 100 V for 90 min in an Atto Mini PAGE system (Atto Bioscience and Biotechnology, Tokyo, Japan). Fluorescent signals were detected using the Odyssey Infrared Imaging System (LI-COR Biosciences UK, Cambridge, UK) and captured using the supplied software (Odyssey V3.0). Band intensity of free probe and DNA–protein complex was calculated using the ImageJ software (Schindelin et al., 2012; Schneider et al., 2012). Graphical representations and statistical analysis were performed using Prism5 (GraphPad Software, San Diego, CA). Details pertaining to each probe used for EMSAs are outlined in Table S3.

Transcriptional start site determination by primer extension analysis

Bifidobacterium breve JCM7017 was grown in mMRS supplemented with 1% lactose to early exponential growth phase and total RNA isolated using a previously described method (Kuipers et al., 1993). Primer extension analyses were performed by annealing 1 pmol of IRD700 labelled oligonucleotides to 15 μg of RNA as described previously (Ventura et al., 2005). Sequence ladders of the presumed promoter regions immediately upstream of repA or repB were amplified from genomic DNA with the aforementioned primers using the Thermo Sequenase Primer Cycle Sequencing Kit (Thermo Scientific, Waltham, MA, USA). Separation was achieved on a 6.5% LI-COR Matrix KB Plus acrylamide gel. Signal detection and image capture were performed utilizing a LI-COR sequencing instrument (LI-COR Biosciences, Cambridge, United Kingdom).

GUS assay

Intracellular β-glucuronidase (β-GUS) activity was determined essentially as described previously (Sangrador-Vegas et al., 2007). GusA activity was expressed in Miller Units and calculated using the equation: 1000*((OD₄₂₀ − (1.75*OD₅₅₀))/(t × v × OD₆₀₀)), where t is the reaction time [min], v is the cell volume [ml] and OD₄₂₀, OD₅₅₀ and OD₆₀₀ are absorbance values at 420, 550 and 600 nm respectively.

SEC RALS/LALS/RI analysis

Size exclusion chromatography was carried out on an AKTA Pure HPLC system (GE Healthcare, Cork, Ireland) using a Superose 6 10/300 G/L column (GE Healthcare, Cork, Ireland) run in a buffer containing 50 mM Tris-HCl, 300 mM NaCl and 50 mM CaCl₂ at pH 7.5 with a flow rate of 0.5 ml min⁻¹. Proteins were injected at a final concentration of > 0.8 mg^-ml. Detection was performed using OmniSec REVEAL, a dual-angle light-scattering apparatus and refractometer (RALS/LALS/RI) (Malvern Instruments, Malvern, UK). Molecular weight calculations were performed with the OmniSec software (v10.4).

Nucleotide sequence analysis and comparative analysis

Sequence data were obtained from the Artemis-mediated (Rutherford et al., 2000) genome annotations of the pMP7017 sequence (Bottacini et al., 2015). Database searches were conducted using non-redundant sequences accessible at the National Centre for Biotechnology Information (http://www.ncbi.nlm.nih.gov) using BlastP or BlastN. Sequence verification and analysis were performed using the SeqMan and Seqbuilder programs of the DNASTAR software package (DNASTAR, Madison, WI, USA v10.1.2). Amino acid sequences obtained from NCBI© were aligned using ClustalOmega© prior to gap removal, and the finished alignment was generated using Jalview© in order to highlight conserved residues. Pfam (El-Gebali et al., 2019) and NCBIs CDD search tool were used to identify functional domains. Structural analyses were performed using DALI (Holm, 2020) and Phyre2 (Kelley et al., 2015), and structural comparisons and overlays were achieved with The PyMOL Molecular Graphics System, Version 2.3.5 Schrödinger, LLC. The Genomic Nucleotide Skew application (Developed by TU Munich; maintained by Department of Computational Systems Biology of the University of Vienna, Austria) was used to plot the GC-skew across the pMP7017 sequence using a window size of 200 bp and a step size of 100 bp. Nucleotide sequences of repB from pMP7017 and B. longum homologues were extracted from the NCBI database and aligned with Clustal Omega (Sievers et al., 2011), and phylogeny was inferred using MEGAX (Stecher et al., 2020) following gap removal. The Blast+ terminal (Camacho et al., 2009) was used to determine percentage identity, and the analysis was conducted in R (‘R: The R Project for Statistical Computing’, https://www.r-project.org) and the comparative figures produced using the genoPlotR package (Guy et al., 2010).

Author contributions

R.D. and D.v.S. designed and directed the project and experiments. R.D. performed all experiments outlined in this work. C.P. aided primer extension analysis. P.K. performed HPLC analysis of Rep proteins. M.J.B.B. assisted in the comparative analysis of the pMP7017 replicon. R.D. and D.v.S. analysed and interpreted the data. R.D drafted the manuscript and figures. R.D., D.v.S. and M.O.C.M. provided commentary and edits to the manuscript and figures in preparation of the final version of the article.

Conflict of interest

None declared.