Evolutionary Link between the Mycobacterial Plasmid pAL5000 Replication Protein RepB and the Extracytoplasmic Function Family of σ Factors

Abstract

Mycobacterial plasmid pAL5000 represents a family of plasmids found mostly in the Actinobacteria. It replicates using two plasmid-encoded proteins, RepA and RepB. While BLAST searches indicate that RepA is a replicase family protein, the evolutionary connection of RepB cannot be established, as no significant homologous partner (E < 10⁻³) outside the RepB family can be identified. To obtain insight into the structure-function and evolutionary connections of RepB, an investigation was undertaken using homology modeling, phylogenetic, and mutational analysis methods. The results indicate that although they are synthesized from the same operon, the phylogenetic affinities of RepA and RepB differ. Thus, the operon may have evolved through random breaking and joining events. Homology modeling predicted the presence of a three-helical helix-turn-helix domain characteristic of region 4 of extracytoplasmic function (ECF) σ factors in the C-terminal region of RepB. At the N-terminal region, there is a helical stretch, which may be distantly related to region 3 of σ factors. Mutational analysis identified two arginines indispensable for RepB activity, one each located within the C- and N-terminal conserved regions. Apart from analyzing the domain organization of the protein, the significance of the presence of a highly conserved A/T-rich element within the RepB binding site was investigated. Mutational analysis revealed that although this motif does not bind RepB, its integrity is important for efficient DNA-protein interactions and replication to occur. The present investigation unravels the possibility that RepB-like proteins and their binding sites represent ancient DNA-protein interaction modules.

INTRODUCTION

The pAL5000 family (23) consists of a group of plasmids that are found mostly in bacteria belonging to the phylum Actinobacteria. Several of these plasmids have been found to exist in various actinomycetal bacteria, such as Rhodococcus (16, 23, 31, 34, 35, 49), corynebacteria, propionibacteria (26), brevibacteria (1, 37), and mycobacteria (44), and in several bifidobacterial organisms (36, 46). The preponderance of these plasmids in the Actinobacteria suggests that they may have originated within this ancient phylum (63). However, similar plasmids have been reported for Neisseria (29) and Salmonella spp., which belong to the phylum Proteobacteria. It may be that although these plasmids originated in the Actinobacteria, they were transferred subsequently to unrelated bacteria.

The pAL5000 family is distantly related to the ColE2 plasmids (11), which are found in various enteric bacteria such as Escherichia coli. The common feature shared by these two groups is that they both encode a replicase family protein known as RepA, which is conserved in evolution. The major difference, however, lies in the fact that whereas the ColE2 plasmids replicate using only RepA (22), pAL5000-like plasmids require RepB in addition to RepA. The two proteins are produced from a bicistronic operon in a translationally coupled manner (5).

An interesting feature of the pAL5000 family of plasmids is that while RepA happens to be conserved, RepBs are highly divergent. No significant database match for RepB with homologs outside the pAL5000 family can be recovered using BLAST searches. However, conserved domain database (CDD) searches reveal that some members of the rhodococcal and bifidobacterial lineages possess a C-terminal domain corresponding to σ factor region 4 (40).

Considering the lack of information regarding the mechanism of replication of pAL5000, studies were initiated to understand the domain structure of RepB using homology modeling, mutational, and phylogenetic analysis tools. The results presented in this study reveal that RepB-like proteins constitute a novel branch within the σ factor family of transcription factors.

MATERIALS AND METHODS

Bacterial strains and plasmids.

E. coli strain XL1 Blue was used for the routine manipulation of plasmid DNA. For the overexpression of repB and its mutants using the T7 RNA polymerase (56)-based expression system, E. coli BL21(DE3) was used. Mycobacterial transformations were performed by using Mycobacterium smegmatis strain mc²155.

Chemicals.

Ni²⁺-nitrilotriacetic acid (NTA) agarose, used for the affinity chromatographic preparation of 6×His-tagged protein, was purchased from Qiagen (Valencia, CA). Other chemicals for protein expression, purification, and analysis, of the highest grade of purity, were obtained from SRL Laboratories, India. Radiochemicals were purchased from BRIT (Mumbai, India). Restriction enzymes and DNA-modifying enzymes such as polynucleotide kinase were purchased from New England BioLabs (NEB).

Phylogenetic analysis.

CLUSTAL W (60)-based alignments of multiple sequences were performed by using MEGA 4.0 software (58). Pairwise and multiple-alignment penalties of 10 and 0.1 were used for gap opening and extension, respectively, or as stated. The weight matrices chosen were either PAM or BLOSUM (15). The alignments created with MEGA 4.0 were subsequently processed by using either Geneious Pro 5.3 (17) or the Bioedit sequence alignment editor for convenient schematic representations. Phylogenetic trees were constructed by using these alignments with MEGA 4.0. Bootstrap analysis was performed by using 500 replicates. Sequence logos were generated by using the Weblogo program (14).

In silico modeling.

To determine the presence of conserved domains and signature sequences, the InterPro (http://www.ebi.ac.uk/interpro/) and Pfam (52) databases were scanned using the query sequence. Homology modeling of RepB was performed by using the Swiss model server in the alignment mode (3). Quality assessment of the predicted model was performed by using the qualitative model energy analysis (QMEAN) Z score (6). Secondary-structure predictions were done by using the online program PSIPRED (http://bioinf.cs.ucl.ac.uk/psipred/) (25).

DNA manipulations.

DNA manipulations were performed by using standard methods. Site-directed mutations were performed by use of the QuikChange mutagenesis kit (Stratagene) using mutagenic primers (see Table S1 in the supplemental material). Deletion mutants were constructed by directly amplifying the desired region using primers mentioned in Table S2 in the supplemental material and cloning into the expression vector. The transformation of mycobacteria was performed by using the pAL5000-based E. coli-Mycobacterium shuttle vector pMC2 (13), using electroporation. For comparisons of transformation efficiencies, equal amounts of supercoiled wild-type and mutant pMC2 plasmid DNAs were used. Following recovery, after electroporation in Middlebrook 7H9 broth, 100-μl aliquots were plated in triplicate. The average colony counts were normalized with respect to those obtained for wild-type pMC2. The relative transformation efficiencies (RTEs) thus obtained from three independent experiments were averaged and expressed as RTEs ± standard errors (SE).

Protein purification.

C-terminally six-histidine-tagged RepB was synthesized translationally coupled to RepA. Translational coupling was found previously to aid in the correct folding of RepB (5). The recombinant protein was isolated by Ni²⁺-NTA chromatography from isopropyl-d-thiogalactopyranoside-induced E. coli BL21(DE3) cells harboring pTAB3, a pET vector-based recombinant which expresses the repA-repB operon in an inducible manner. The RepB protein obtained by this method was 98% pure, as judged by 13.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by staining with Coomassie blue.

DNA-protein interaction studies.

Electrophoretic mobility shift assays were performed as described previously (12). A 200-bp DNA fragment derived from the pAL5000 origin of replication (nucleotides 4459 to 4663) encompassing the high- and low-affinity binding sites for RepB (55) was PCR amplified by using primers S6 (5′-GGATCCTGGTTGGTACAGGTGGTTGGG-3′) and S7 (5′-GCTGCTCAAATTCGTCGGCG-3′). Ten picomoles of S7 was 5′ end labeled with [γ-³²P]ATP (BRIT, Mumbai, India) and T4 polynucleotide kinase (NEB) and directly used for PCR. The PCR product was purified by using a Qiagen column. The binding reaction mixture (in a 30-μl final volume) contained 3 μl 10× binding buffer (100 mM Tris [pH 8], 600 mM NaCl, 30 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 20% glycerol), 1 μg of salmon sperm DNA, nearly 10,000 cpm of labeled DNA, and the required amount of purified proteins. The reaction mixtures were preincubated for 10 min and then incubated for an additional 20 min on ice after the probes were added. The DNA-protein complexes were separated on a 5% native PAGE gel by electrophoresis in 0.5 Tris-borate buffer (50 mM Tris-borate, 1 mM EDTA) at 200 V for 3 to 5 h at 4°C after a prerun at 100 V for 1 h. Following electrophoresis, the gel was dried, and the bands were visualized by autoradiography. Competition binding experiments were performed by including a molar excess of the desired competitor during the preincubation period. Band intensities were scanned by using a Versadoc imaging system (Bio-Rad). The percent residual binding observed in the presence of the competitor was plotted against the molar excess. The molar excess at which 50% inhibition occurred was determined after curve fitting using KyPlot software.

RESULTS

Phylogenetic analysis of RepA and RepB.

A BLAST search was performed for the RepA protein of pAL5000. The results showed that RepA of pAL5000 has significant homology with replication proteins (RepAs) of several other plasmids found in members of the Actinobacteria. Genomes of these plasmids identified by a BLAST search were then examined for the presence of a downstream gene encoding RepB. Only those plasmids (16 plasmids) that had a repA-repB combination similar to that found in pAL5000 were considered to belong to the pAL5000 family (see Table S3 in the supplemental material). As RepA is highly conserved, it was therefore used to construct a phylogenetic tree for the pAL5000 family. The rooted phylogenetic tree thus obtained (Fig. 1A) indicates that all the plasmids of the Actinomycetes have a common origin, as evident from the 100% bootstrap analysis value at the node from which they emerged.

Fig 1.

(A and B) Evolutionary relationships of RepA (A) and RepB (B) from 16 taxa. RepA or RepB sequences were aligned by using CLUSTAL W with the help of MEGA 4.0. The evolutionary histories were then inferred by using the neighbor-joining method (48). The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) are shown next to the branches. The phylogenetic tree in panel A was rooted by using the RepA protein encoded by plasmid pJD1 from Neisseria gonorrhoeae, a proteobacterium that is phylogenetically distinct from the Actinobacteria. Thick lines are drawn to indicate radiations from a possible common ancestor(s), indicated by arrows. In the case of the RepB tree, the two phylogenetic groups (GrI and GrII) are indicated. Three taxa were chosen (marked by filled circles) as examples for demonstrating the fact that RepAs and RepBs have independent phylogenetic histories. The scale (horizontal line below each tree) signifies evolutionary distance (λ), which was estimated by the Poisson correction method with the equation λ = −ln (1 − p), where p is the fractional dissimilarity between two sequences. (C) Sequence alignment of 15 of the 16 pAL5000 family RepBs. The pCASE01 sequence, which has a longer N-terminal region, was omitted for the sake of convenience of presentation. Residues that are identical in more than 40% of the sequences are shaded black. Stars denote two conserved arginine residues, which were subjected to mutational analysis, and the underlined region indicates the part of the alignment reproduced as a Weblogo in Fig. 2D. A detailed alignment is shown in Fig. S1 in the supplemental material.

The phylogenetic tree created by using the corresponding RepBs was then examined (Fig. 1B). The results showed that although genes encoding RepA and RepB are always located in the same operon and the synteny is highly conserved, their evolutionary histories differ. Thus, for example, whereas rhodococcal plasmids pB264 and pRC4 appear to be closely related judged on the basis of the RepA tree, they are distantly related if the RepB tree is taken into consideration. Conversely, pRC4 and pNC500 appear to be closely related in the RepB phylogram but appear to be distantly related in that of RepA. This analysis suggests that the original ancestor of the pAL5000 family may have possessed only the gene for RepA but that it subsequently acquired RepB from other sources to complete the repA-repB operons. The other feature that stands out is that while the RepAs are monophyletic, RepBs appear to be polyphyletic. They appear to have emerged from two most recent common ancestors (MRCAs), giving rise to two distinct clades (group I [GrI] and GrII) (Fig. 1B). Another interesting feature is that while RepAs derived from the pAL5000-related plasmids of Gram-negative organisms such as Stenotrophomonas and Salmonella form a distinct branch emerging from a last common ancestor, the corresponding RepBs cluster with the rhodococcal proteins. This further indicates that the repA and repB genes may have different phylogenetic histories. In spite of the sequence diversity among RepBs, several motifs in both the N- and C-terminal regions appear to be conserved (shaded black in Fig. 1C). The conservation in the N-terminal region indicates that this domain may be associated with some specific function related to plasmid replication.

Homology modeling of RepB.

A search for the presence of signature sequences within pAL5000 RepB was done by using the InterproScan tool. No conserved domain was detected for pAL5000 RepB. However, a similar search conducted using the Rhodococcus pFAJ2600 RepB sequence revealed the presence of a conserved domain at the C-terminal end corresponding to σ factor region 4 type 2 (protein family [Pfam] PF08281). This Pfam represents primarily region 4 of the ECF family of σ factors (53), as is evident from the observation (see Table S4 in the supplemental material) that out of the 43 groups of ECF sigma factors identified (53), 38 possess a σ region 4 corresponding to this Pfam.

To examine closely how RepB-like replication proteins may be related to region 4 of σ factors, homology modeling of pFAJ2600 RepB was attempted by using crystal structures of various σ factor region 4 domains. Three crystal structures of σ region 4 available in the databases were chosen as templates. These structures include (i) Thermus aquaticus SigA (Protein Data Bank [PDB] accession number 1KU) (9), E. coli SigE (accession number 2H27) (30), and (iii) Mycobacterium tuberculosis SigC (accession number 2O8X) (59). In all cases, RepB was first aligned with each of these sequences, and modeling was then performed in the alignment mode using the Swiss model server. Of the three alignments used for modeling, the one with SigC (Fig. 2A and B) yielded the best possible modeled structure (QMEAN Z score of −1.35). This indicates that out of the three σ factors used for comparison, RepB is structurally most related to SigC, an important ECF σ factor of M. tuberculosis that controls its virulence (57). The alignment with SigC (Fig. 2C) was thus considered a seed for further alignments.

Fig 2.

Molecular modeling of pFAJ2600 RepB using SigC region 4 as a template. (A) pFAJ2600 RepB model derived by using the SigC region 4 crystal structure (PDB accession number 2O8X) (chain A). (B) Trimeric form of SigC region 4. Chain A is colored, while the other two chains are shaded gray. The same color scheme is used to highlight the structural similarity between RepB and SigC. The models are drawn so that the recognition helices (red-orange) are in a horizontal position. In this orientation, the three-helix bundle appears as a triangle (2). In the SigC model, arginines 169 and 171, which form a part of a conserved RAR motif (red boxes), are shown as sticks extending from the backbone. In the case of pFAJ2600 RepB, the corresponding residues are glutamic acid 79 and lysine 81. (C) Alignment of pFAJ2600 RepB with SigC region 4. The colored bars above the aligned sequences correspond to helical domains shown in panel A. There are altogether five predicted helical domains (helices 1 to 5). Helix 1 was predicted by PSIPRED analysis. Helices 2 to 5 were predicted by the model shown in panel A. Helices 2 and 3 are most likely to constitute a single helical domain, as predicted by PSIPRED analysis. Identical residues are shown in red, and similar amino acids are shaded in gray. The region of pFAJ2600 RepB that corresponds to σ region 3 (Fig. 3) is indicated (black box). (D) Weblogo presentation of aligned σ region 4 sequences corresponding to Pfams PF04545 and PF08281 and RepB (see Fig. S4 to S6, respectively, in the supplemental material). Seed sequences (130 sequences for PF04545 and 141 for PF08281) were downloaded from the Pfam database and realigned by using a gap-opening penalty of 20. The resulting gapless alignments are presented in the form of Weblogos. The Pfam PF04545 alignment spans 53 amino acid residues, of which 46 residues (residues 3 to 49) are shown. In the case of Pfam PF08281, there are 51 positions, of which 46 (positions 7 to 53) are shown. In the case of RepBs, the region that aligns with Pfam PF08281 (46 positions, corresponding to the underlined region in Fig. 1) is included. A conserved glycine residue (black arrow) known to be conserved within the turn region of a number of HTH motifs was taken as a reference point to align the three Weblogos. The other amino acid residues that are conserved at various positions are indicated by broken arrows. The HTH motif is indicated below, using the same color scheme as that used in panels A and B. Residues that are essential for maintaining the HTH framework, as proposed in the case of the λcI repressor (39), are indicated by asterisks in the Weblogo corresponding to the RepB alignment.

The model obtained shows that the C-terminal region of pFAJ2600 RepB can potentially attain a three-helical bundle structure characteristic of the helix-turn-helix (HTH) superfamily of DNA binding proteins (2). The major part of the three-helical bundle is taken up by three helical segments (helices 3, 4, and 5) (Fig. 2C), of which the C-terminal helix (helix 5) (red-orange shaded in Fig. 2A and C) corresponds to the DNA-recognizing helix (21) of σ factors that is known to contact the −35 region of promoter elements (9, 30). The quality of the predicted model was found to be optimal over the C-terminal region spanning residues 50 to 93. Further toward the N-terminal end, the quality decreased sharply due to the lack of homology between the query (pFAJ2600 RepB) and the target (Protein Data Bank accession number 2O8X) over this region. To obtain an idea about the secondary structure of the RepB N-terminal region, PSIPRED analysis was performed. The analysis predicted the presence of a helix (helix 1) (Fig. 2, deep blue) at the N-terminal end, which corresponds to σ region 3 (Fig. 2C, sequence within a black box, and 3B). The side chains of the two conserved arginine residues within the SigC structure are highlighted in Fig. 2C. These arginines together with the alanine located between them constitute an arginine-alanine-arginine (RAR) motif (Fig. 2C, red boxes), which is also conserved in the pAL5000 family. In the specific case of pFAJ2600 RepB, the corresponding amino acid residues are DAK, as indicated in Fig. 2C.

Fig 3.

Global alignment of Mycobacterium tuberculosis σ factors with RepB-like proteins of the pAL5000 family. For reference, SigA from T. aquaticus is included. (A) Schematic diagram of the alignment created by Geneious 4.0. Evolutionarily conserved σ regions (9) are indicated above the alignment. (B and C) Parts of the alignment corresponding to σ regions 3 and 4 are shown (details are available in Fig. S2 in the supplemental material). For the sake of simplicity, only a subset of RepBs is included in this analysis.

In the databases, apart from Pfam PF08281, another σ factor region 4 signature, PF04545, exists, representing primary and general stress response sigma factors such as SigA, SigB, and SigF. However, the C-terminal regions of RepBs appear to be more related to PF08281 than to PF04545, as evident from a comparison of the Weblogos of the aligned sequences (Fig. 2D). Several amino acid residues (indicated by broken arrows) were found to occur at comparable frequencies at identical positions within the HTH domains of PF08281 and RepBs. Such a correspondence was not evident when the logos of either PF08281 or the RepB family were compared to that of PF04545. Three conserved residues common to PF08281/RepB, alanine 33/27, glycine 37/31, and valine 43/37 (marked by asterisks in Fig. 2D), occupy the same conserved positions (positions 5, 9, and 15) as those in the λcI repressor HTH motif (7, 38). In this motif, glycine 9 is conserved, as it is located in the turn region. Alanines 5 and 15, located in helices 2 and 3, respectively, are known to interact with each other, resulting in the stabilization of the angle between the two helices (38).

Another important feature that emerges from the above-described comparison is the conservation of an RAR motif at the C-terminal end (red boxes in Fig. 2). The degree of conservation of the arginine residues within the RAR motif varies, with the first arginine being more conserved than the second in case of Pfam PF08281 and vice versa for RepB. The alanine residue within the RAR motif, however, is highly conserved between the two. In PF04545, a similar motif, RVR, is present, although it is located 3 residues closer to the conserved G than in either PF08281 or RepB. The valine residue, which appears to be equivalent to alanine within the RAR motif, is also highly conserved. The RVR and RAR motifs are involved in DNA binding (30), and this explains at least partly why they are so conserved.

Global alignment of RepB and σ factors.

In order to understand the relationship between RepBs and σ factors, a global alignment of RepB-like sequences and a repertoire of mycobacterial σ factors was attempted. The repertoire includes mycobacterial σ factors classified in three major groups (40), ECF, general stress response, and primary (33). The repertoire also includes Thermus aquaticus SigA as a reference, as its structure and function are well known (9). The results of such an alignment revealed that, as expected, the RepBs can be aligned with region 4 of σ factors (Fig. 3A and C). In the N-terminal region, the extent of the sequence homology with σ factors was less. This region, which can potentially take up a helical structure, as predicted by a PSIPRED analysis, nevertheless was found to align with region 3 (Fig. 3A and B), which was previously considered to be region 2.5, a part of region 2 (4). This is consistent with the known fact that region 3 of σ factors is a helix (9, 10, 59). The alignment (Fig. 3 and see Fig. S2 in the supplemental material) shows that all the ECF family factors possess this region but lack region 3.1/3.2 (9). This feature (the lack of region 3.1/3.2) is also shared by RepBs, and hence, these proteins appear to be structurally related to the ECF family of σ factors (47, 53, 59).

In order to assess the extent of the conservation of region 4 of the ECF family of σ factors, a sequence alignment was performed by using representative members from each of the 43 ECF groups identified in a previous study (53). The alignment (see Fig. S3 in the supplemental material) reveals a significant conservation of several residues. In order to obtain a phylogenetic interpretation of the alignment, a cladogram was constructed by using the neighbor-joining (NJ) method. The tree thus constructed (Fig. 4) shows that the C-terminal HTH domain of RepB region 4 constitutes a novel branch in the σ factor family. The tree also indicates that, as described above, GrI and GrII RepBs belong to different clades and that GrI RepBs appear to be closely related to a subset of ECF σ factors, particularly ECF16 (53) and the mycobacterial σ factor SigM (43, 47).

Fig 4.

Cladogram depicting the evolutionary relationships between C-terminal HTH domains of RepBs and ECF family σ factors. RepB C-terminal HTH domains, corresponding to residues 32 to 91 of pFAJ2600 RepB (Fig. 1C), were aligned with the region 4 sequences derived from representative members of each of the 43 groups of ECF sigma factors (53) (see Fig. S4 in the supplemental material). To avoid bias, the first member in the list of entries for each group (see the supplemental material in reference 52) was arbitrarily chosen as the representative for that group (see Table S4 in the supplemental material). Several other σ factor region 4 domains were included in the alignment and are indicated by their respective conventional names followed by three-letter codes indicating the bacteria in which they are found. The abbreviations are as follows: Eco, Escherichia coli; Sen, Salmonella enterica; Bsu, Bacillus subtilis; Ccr, Caulobacter crescentus; Mtu, Mycobacterium tuberculosis. The alignment thus obtained was used to construct the cladogram. The GrI and GrII RepB clades are highlighted.

Mutational analysis of RepB.

In order to understand the domain structure of RepB, two deletion mutants were designed, in which either the N- or C-terminal domains were deleted (Fig. 5A). These deletion mutants were then tested for DNA binding activity (Fig. 5B). The results of such an analysis revealed that in the case of the RepB(1–82) C-terminal deletion mutant, a complete loss of DNA binding activity was observed. However, for the RepB(34–119) N-terminal deletion mutant, a low level of DNA binding activity was observed, which was nonspecific, as it could not be competed out by specific competitor (data not shown). The deletion analysis therefore indicates that while the C-terminal region of RepB corresponding to the three-helix bundle alone can interact with DNA, its specificity depends on the presence of N-terminal sequences.

Fig 5.

DNA binding activities of the individual domains. (A) Domain structure of RepB. The location of the two conserved σ domains 3 and 4 are indicated. Shaded boxes correspond to predicted helices. The five helices are labeled. PSIPRED analysis predicts that helices 2 and 3 constitute possibly a single helical domain. The sequence motifs that are conserved in the pAL5000 family of proteins are shown at the top. The mutations are indicated with vertical up arrows. The regions present in the two deletion derivatives RepB(34–119) and RepB(1–82) are shown. (B to D) Origin binding assays were performed by using a probe derived from the origin of pAL5000 spanning the H and L sites (nucleotides 4459 to 4663), with either RepB or its deletion mutants (B) or point mutants (C and D). The protein concentrations were 100, 250, and 500 nM (B); 200, 400, and 600 nM (C); and 200, 400, 600, and 800 nM (D).

Point mutational analysis of RepB was performed to specifically identify the residues involved in the interaction with RepB (Fig. 5C and D). Mutations were introduced within both conserved domains (Fig. 5A). In the C-terminal region, mutations were introduced within the RAR motif. Either of these two R's were mutated to alanines. In the N-terminal region, mutations were introduced within the RKKR motif, which is particularly conserved in the RepB family (Fig. 1C). The results of this analysis revealed that the R94A mutation resulted in the partial loss of origin binding activity, but for the R96A mutation, a complete loss of origin binding activity was observed (Fig. 5C). These mutants were then tested for their replication efficiencies. The results (Table 1) showed that in the case of the R96A mutation, a complete loss of transformation activity was observed. However, in the case of the R94A mutation, no significant loss of transformation activity was encountered. In the N-terminal region, the R23A mutation resulted in a partial loss of origin binding activity (Fig. 5D) and a complete loss of transformation activity (Table 1), but no effect on either origin binding or transformation activity was found in the case of the R20A mutation. These results indicate that RepB interacts with DNA primarily through R96, located within the DNA recognition helix of the σ region 4-homologous region within RepB. The results also indicate that R23 within the σ region 3-like region of RepB has an important role to play in the replication of pAL5000 and related plasmids, although it does not appear to be involved in RepB-DNA interactions, at least not in a major way.

Table 1.

Effects of point mutations on RepB activity

Protein or mutation	Mean RTEa ± SE
RepB (wild type)	100
R20A	95 ± 13
R23A	0
R94A	73 ± 06
R96A	0

^aRTE, relative transformation efficiency.

A conserved A/T-rich sequence within the origin is essential for replication.

The origin region of plasmid pAL5000 comprises two RepB binding sites, H (high affinity) and L (low affinity). However, in between the H and L sites, there is a region that is highly conserved within the pAL5000 family (Fig. 6A) (16). This region includes an A/T-rich sequence (hatched box in Fig. 6A). A theoretical analysis of the conformation of this DNA sequence revealed that this region may be prone to curvature. This situation is similar to that of several DNA-protein interaction modules, where the presence of a bendable A tract plays an important role (18, 27, 41). In order to test whether the A/T stretch between the H and L sites has any role to play, mutations which should abolish the intrinsic bend were introduced. Thus, the A-C mutation (Mut-1) was expected to abolish the intrinsic bend, whereas the T-A mutation (Mut-2) should allow bending, as in the case of the wild type (Fig. 6B, right). The results of the transformation assays revealed that the A-C mutation (Mut-1) indeed resulted in the loss of replication activity but that the T-A mutation (Mut-2), in which replication activity was retained to the extent of ∼30%, did not. The results indicate that, in a manner reminiscent of many DNA-protein interaction modules (18, 45, 62), the noncontacting residues in between the two RepB binding sites affect replication to different degrees, with the maximum effect being observed in the case of the mutation that alters the AAA stretch. It is likely that this sequence affects DNA bending, although this could not be demonstrated, at least not by using gel electrophoresis-based methods (data not shown). By using more sensitive structural analysis tools, it may be possible to determine more accurately structural aberrations, including bending, induced by the A/T-rich conserved sequence.

Fig 6.

Effect of mutations within the 15-bp conserved sequence. (A) Organization of the pAL5000 origin of replication showing the high- and low-affinity RepB binding sites, H (black) and L (white), respectively. The hatched box indicates the 15-bp sequence that is conserved in pAL5000 family origins. The mutations introduced to study the role of this conserved sequence are shown at the top. The middle panel shows the relative transformation efficiency (RTE) (±SE) of pMC2 mutants Mut-1 and -2 with respect to the wild type (Wt). (A) The hypothetical trajectories of the H- and L-containing origin sequence, derived by using the Bendit server (64), are incorporated below each sequence. (B and C) Comparison of RepB binding to the wild type and mutant templates Mut-1 and Mut-2, as judged by competition binding assays. RepB–wild-type interactions were performed in either the absence or presence of a competitor, either the wild type or the mutant, at the indicated molar excesses. The percent binding (as estimated by densitometric scanning) was plotted against the molar excess of competitive DNA. The IC₅₀ (50% inhibitory concentration) (which in this case is the molar excess of competitor used) was estimated by curve fitting using the equation Y = A2 × A1/(X + A1), where X is the inhibitor dose, Y equals A2 at an X of 0, and A1 is the IC₅₀. Curve fittings were done by using KyPlot software. The IC₅₀ values (±standard errors) are shown.

It was expected that if the mutations alter the structure of the DNA, an effect on the binding efficiency would be observed. Direct binding assays did not lead to an all-or-none phenomenon, and hence, to investigate whether subtle changes in binding affinities occurred, a competition binding assay was performed by using the wild-type and mutant competitors. The results of the competition experiments (Fig. 6B and C) revealed that in the case of Mut-1, 50% inhibition occurred at about a 75-fold molar excess, whereas in the case of self-competition, the same inhibition occurred at about a 35-fold molar excess. This indicated an approximately 2-fold drop in affinity due to Mut-1. On the other hand, Mut-2 competed as well as the wild type, if not better (50% inhibition was obtained at about a 35-fold molar excess in the case of the wild type and at a 24-fold molar excess in the case of Mut-2). This finding indicates that the A-C mutation resulted in the loss of the affinity of the origin for RepB. Hence, the stretch of A residues within the conserved box has a crucial role to play in modulating the origin binding efficiency of RepB, even though it does not actually interact with RepB. However, the possibility that the conserved box makes secondary contacts with either RepB or any other ancillary factor present in the final origin complex is not ruled out.

DISCUSSION

The RepB class of proteins is encoded by genes that form a part of the replication region of a large number of plasmids that may be described as the “pAL5000 family.” Mycobacterial plasmid pAL5000 is the most studied member of this family and also the most used in the context of recombinant DNA research. These plasmids are considered to be related to ColE2 plasmids, which have been reported for a large number of enteric bacteria, as both possess a homolog of the gene encoding the replication protein RepA (65). RepA, in turn, is a part of a superfamily comprised of not only bacterial plasmid replication proteins but also similar proteins involved in the replication of eukaryotic and archaeal genomes (19, 24).

The present investigation focused on proteins of the RepB class, which are highly divergent and do not possess any significant homology with proteins outside their own family. Even within the family, the proteins diverge considerably. Thus, RepB may be classified as an orphan open reading frame (ORFAN) (51). The origin binding activity of pAL5000 RepB has been established by several studies (12, 55). It binds to two sites within the origin of replication located upstream of RepB with either high (H) or low (L) affinity. The two sites within the origin are separated by two to three turns of the DNA helix. The DNA sequence required for RepB binding has been characterized. One of the major consequences of RepB binding is the asymmetric bending of the DNA and the subsequent opening up of the origin of replication (12).

Although several studies have been performed to obtain insights into the DNA sequence requirement for RepB binding, relatively little information is available regarding the structure-function relationship of the protein. In a previous investigation, it was shown that the active form of RepB possesses significant alpha-helical content (5). In order to attain the alpha-helical active structure, the protein has to be synthesized in a translationally coupled manner in combination with RepA (5). It may be that since coupled expression is a prerequisite for optimum activity, the synteny within the repA-repB operon is then highly conserved.

The phylogenetic analysis of the pAL5000 family of plasmids performed by using the conserved RepA protein revealed that the actinobacterial members of the pAL5000 family originated from a common ancestor. The phylogenetic tree obtained by using RepB, however, yielded a significantly different pattern. RepBs were found to form two clusters, one of which is comprised mostly of RepBs of the rhodococcal plasmids and the other of which is comprised of RepBs closely related to pAL5000. The two clusters are highly divergent and do not appear to have emerged from the same immediate ancestor. There are also considerable differences in the phylogenetic affinities. Plasmids with closely related RepAs possess RepBs that are distantly related. This indicates different evolutionary origins of RepAs and RepBs. It may have so happened that a plasmid similar to ColE2 which replicated using only RepA was horizontally acquired by the Actinomycetes. The replication regions of these plasmids subsequently acquired RepBs independently and evolved into pAL5000-type plasmids. During the process of this transformation, a novel origin, located in the upstream region of RepA, instead of downstream as in the case of ColE2, evolved, which started functioning in a RepB-dependent manner.

RepBs share structural homology with C-terminally located region 4 of the ECF family σ factor SigC (59). Structurally, σ region 4 is comprised of a three-helical helix-turn-helix (HTH) domain. The basic HTH domain possesses a three-helix bundle structure (2). In this structure, there are three helices, designated helices 1, 2, and 3 as per convention. Of helices 2 and 3, which constitute a typical HTH motif, helix 3 is directly involved in the recognition of the target DNA (recognition helix) (21, 50). Secondary-structure prediction analysis and homology modeling indicated that there may be five helical regions in the RepB family proteins, the last three of which constitute the three-helical HTH domain. The fifth helix is the counterpart of helix 3 (recognition helix). The mutation of a conserved R residue in helix 5 of RepB leads to the loss of DNA binding as well as replication activity. This confirms that helix 5 indeed functions as a recognition helix.

The simplest three-bundle helix is considered to be the closest to the ancestral form of the HTH domain (2). The structure of RepB reveals that it is a stand-alone protein possessing the minimal three-helix bundle. Although HTH domains constitute a part of many globular proteins having diverse functions, there are very few examples of such stand-alone HTH domains. Since RepB appears to be evolutionarily linked to σ region 4, it is tempting to speculate that RepB-like proteins represent an extinct version of the σ region 4 ancestral HTH domain.

The fact that RepB, a stand-alone HTH domain protein, binds to its cognate sequence in a specific manner appears intriguing. In other HTH family regulatory proteins, such as the λcI repressor, the DNA binding domains themselves have a low affinity for their cognate sequences and have to depend on separate multimerization domains for high-affinity interactions (39). RepB, although it does not possess any additional domain apart from the DNA binding one, forms multimers in vitro and binds to the H site within the origin in a cooperative manner (12, 54), a feature that is typical of multimeric DNA binding proteins. Cooperativity leads to high affinity, as it increases the efficiency of the recruitment of factors to the binding site (12, 42). However, for the manifestation of cooperativity, interaction between subunits is necessary. RepB is unique in that even though it is essentially a single-domain protein, it forms multimers and displays cooperativity. A possible multimeric structure for the RepB-origin complex was proposed previously (12). It is comprised of a RepB dimer bound cooperatively to the H site and a monomer at the L site. The bound subunits are likely to interact further, resulting in the stabilization of a phasing-sensitive looped structure involving a trimeric RepB unit. How RepB, a single-domain HTH protein, forms dimers, trimers, or higher multimers, is not clear at present. However, considering the structural homology with SigC region 4, it is possible that interchain contacts similar to that observed for the trimeric crystal structure of SigC region 4 (59) may be involved.

The present study gives strong indications that RepBs belong to the σ factor family. This is evident from the cladogram representation of the alignment of region 4 sequences derived from RepBs and a variety of σ factors, including representative members from each of the 43 groups of ECFs identified in previous studies (53). Whether the RepBs evolved through convergent or divergent mechanisms is an issue that is debatable. Given the extensive sequence divergence within the σ factor family, it is very difficult to decide on the phylogenetic histories of σ and related factors. One has to rely on other pointers, such as structural homologies and the presence of signature sequences. Homology modeling indicates that the tertiary structure of the HTH domain of RepB is more related to the region 4 crystal structures of SigC and SigE, which are ECF family factors, than to that of SigA. It was proposed previously that all HTH motifs evolved from a single common ancestor (2). Such a proposition is based on the fact that, despite the sequence divergence, the basic framework is conserved. Using the λcI repressor prototype, it was shown that certain residues (alanine 5, glycine 9, and valine 15) are highly conserved in the basic framework, as they are structurally important determinants of the HTH motif (7, 38). It is interesting to note that in PF08281 and also in RepBs, these residues are more conserved at their respective positions than in PF04545. This leads to the argument that PF08281, which represents mostly ECFs, is more related to the common ancestor of σ factor and λcI families than PF04545. Hence, evolutionarily, PF08281 possibly predates PF04545. Given that RepB region 4 is related to PF08281, it follows that RepBs may be just as ancient as ECFs or, perhaps, considering the simplicity of their domain structures, more ancient.

Both Pfam PF08281 and the RepB family possess a conserved RAR-type motif at the C-terminal end. Structurally, this RAR motif occupies the same position as R(176)A(177)R(178) in the crystal structure of E. coli SigE (30). The central A residue within this motif is highly conserved between ECFs and RepBs. The degrees of conservation of the corresponding R176 and R178 residues in ECFs and RepBs differ. In the case of ECFs, R176 is more conserved than R178, whereas in case of RepBs, it is just the reverse. The degree of conservation tallies well with the extent to which these residues contribute to DNA binding and/or biological activity. Thus, as indicated by the mutational analysis performed in this study, it is the R178-corresponding arginine which is important for RepB activity, whereas in the case of SigE (an ECF prototype), it is R176 (9). It seems, therefore, that following the divergence from the same ancestor, evolutionary pressures forced the conservation of R176 in the case of ECFs and R178 in the case of RepBs.

The alignment between σ factors and pAL5000 family RepBs further revealed the presence of a small stretch at the N-terminal end of RepBs with a low level of homology with σ region 3. To what extent this region functionally resembles σ region 3 is unclear at present. Mutational analysis showed that at least one mutation, R23A, within this region resulted in a complete loss of transformation activity. The mutation, however, did not significantly impair the protein's DNA binding activity. It is most likely that this region is involved in protein-protein interactions with other proteins involved in pAL5000 replication.

Apart from characterizing the domain structure of RepB, the role played by a 15-bp sequence located within the H and L binding sites, which is highly conserved within the origins of the pAL5000 group of plasmids, was investigated (16). Within this conserved region, there is a stretch of three A residues which, according to the trinucleotide model for DNA bending propensity (8), is likely to contribute to the curvature of the DNA. A mutational analysis of this region, performed for the first time, revealed that, indeed, this A tract plays a crucial role, even though it does not contact RepB. Evidence presented here shows that the A tract influences the DNA binding affinity of RepB, which is reduced about 2.5 times in a mutant in which the A tract is altered. The effect on the replication efficiency was, however, much greater. It is likely that a relatively small difference in the RepB binding efficiency observed in vitro may be amplified in vivo due to reasons which may include the involvement of other replication proteins and alterations in the structure of the origin due to supercoiling. A tracts generally influence DNA transaction processes by altering the structure of the DNA (28, 62). In prokaryotic promoters, for example, there is an A/T-rich sequence between the −35 and −10 elements, which is believed to influence promoter function by supporting DNA bending (32, 41). Similarly, the phased occurrence of AA dinucleotide steps is necessary for DNA bending around the nucleosomal core (61). In both eukaryotic and prokaryotic replication origins, A-tract-induced bending is a common feature (18). The presence of a bendable A/T-rich sequence is thus a common theme in diverse DNA-protein interaction modules found in all three domains of life. In the case of pAL5000, the conserved 15-bp core, which includes a bendable A tract, does not interact with RepB, as is evident from footprinting studies performed previously (12). Hence, it may be that the primary role of this conserved region is to alter the structure in such a way that the origin can form an initiation complex efficiently. That this possibility is most likely to be true is indicated by the observation that an A-to-C mutation, which changes the rigid AAA trinucleotide (8) sequence within the 15-bp conserved region to the flexible ACA sequence, abolishes replication activity. Theoretically, this mutation should lead to a loss of DNA curvature, as it destroys a rigid-flexible junction, which, according to the junction model (20), is prone to DNA bending. Hence, the 15-bp conserved region appears to control origin activity primarily by influencing the conformation of the DNA. This does not rule out the possibility that in the ultimate origin complex, the conserved domain may be involved in protein binding in a manner similar to that of the RNA polymerase β′ subunit, which, in the RNA polymerase-promoter complex, makes secondary contacts with the spacer between the −35 and −10 promoter elements (66). Considering the various novel features of the pAL5000 replication protein RepB and its binding site unearthed in this study, it is tempting to speculate that RepB and its interactive site within the origin constitute an evolutionarily ancient module that may have existed in the last universal common ancestor (LUCA).