Rational Design of a Plasmid Origin That Replicates Efficiently in Both Gram-Positive and Gram-Negative Bacteria

Background Most plasmids replicate only within a particular genus or family. Methodology/Principal Findings Here we describe an engineered high copy number expression vector, pBAV1K-T5, that produces varying quantities of active reporter proteins in Escherichia coli, Acinetobacter baylyi ADP1, Agrobacterium tumefaciens, (all Gram-negative), Streptococcus pneumoniae, Leifsonia shinshuensis, Peanibacillus sp. S18-36 and Bacillus subtilis (Gram-positive). Conclusions/Significance Our results demonstrate the efficiency of pBAV1K-T5 replication in different bacterial species, thereby facilitating the study of proteins that don't fold well in E. coli and pathogens not amenable to existing genetic tools.


Introduction
The laboratory work horse, E. coli, cannot efficiently translate or fold many foreign proteins. Most prokaryotic plasmids replicate in a particular eubacterial genus or family. The employment of different bacterial species as expression systems therefore necessitates the acquisition or development of new expression vectors [1,2]. Broad host range plasmids based on RK2 [3], IncaP [4] or rolling circle replication (RCR) [5] origins have been developed for the production of proteins, or the study of poorly characterized bacterial pathogens, but most are limited in their host range, genetic stability, size or capacity to accept large inserts [2,6]. pWV01 is a cryptic plasmid originally purified from Streptococcus cremoris [7]. Its RCR origin has been used to create over 20 cloning vectors [5]. Among them, pGK12 is most widely used by other researchers. It replicates in Bacillus subtilis, Lactococcus lactis, E. coli, Borrelia burgdorferi, and numerous Lactobacilli (namely reuteri, fementum, casei, acidophilus, pentosus and helveticus) [6]. Unfortunately, pGK12 is unstable and does not replicate to high copy number in these species. Its performance in E. coli is particularly poor, so it was never widely adopted by researchers [6].
Little was known about RCR when pGK12 was first constructed [5]. Three decades of subsequent study [6,8,9] have laid the foundation for the rational design of better plasmid origins. RCR plasmids exist in eubacteria (Gram-positive and Gram-negative) and archaeabacteria [6]. Replication begins when the Rep protein, which is encoded on the plasmid (ORF A), recognizes a specific site on the plasmid (double-strand origin, or DSO) and catalyzes the nicking of one DNA strand. The Rep protein remains bound to the 59 phosphate after the nicking action. The newly released 39 hydroxyl on the opposite end serves as a primer for DNA synthesis. The host DNA polymerase uses the unnicked circular strand as a template, so that a single replication fork moves around a plasmid until it regenerates the DSO. A second copy of Rep protein catalyzes the cleavage of the newly formed DSO, effectively releasing a single stranded copy of the plasmid. In the absence of Rep, the replication fork continues to move around the template, forming a single stranded concatemer. The single strand origin (SSO), a non-coding element that forms extensive secondary structure, is required for synthesis of the lagging strand. SSO sequences vary considerably among different RCR plasmids, but are extremely important for robust replication of the plasmid in the cell [10]. Here we describe the engineering of the pWV01 RCR origin to create pBAV1K-T5, a very broad-host range expression vector.

Results and Discussion
We hypothesized that the RCR of pWV01-based plasmids in non-native hosts was inefficient. Cryptic plasmids by definition have no detectable effects on their hosts, so the copy number of pWV01 in its native host must be stringently controlled. If the copy number control mechanisms of a plasmid are more efficient than its RCR mechanisms in non-native hosts, the plamid would not be stable under non-selective conditions. The elimination of regulatory elements, particularly those not widely conserved among RCR plasmids, should allow the altered plasmid to replicate more freely. We sought to create a minimal plasmid origin that included only pWV01 ori elements that were shown to be necessary by other researchers [11]. Runaway replication is toxic to host cells, of course, but we speculated that the inefficiency of RCR in non-native hosts would moderate this risk.
We therefore sought to delete copy number control mechanisms from the pWV01 origin. Two inverted repeats (IRI and IRII), out of the six within the pWV01 origin, are sufficient for the conversion of single-stranded DNA to the double-stranded form Figure 1. Construction of pBAV1K-T5-lux, a very broad host range expression vector. The cryptic plasmid, pWV01, exhibits broad host range but is unstable in many species. The ORF D, and inverted repeats IV, V and VI were deleted from its plasmid origin; terminators t0 and T1 were inserted on opposite ends of the shortened origin (upper right). The selectable marker, the Enterococcus 39,50-aminoglycoside phosphotransferase type III, and a T5 promoter within a BioBrick multiple cloning site were cloned into the plasmid (top circle). The lux genes of Photorhabdus luminescens were individually PCR amplified, cloned, assembled with ribosome binding sites (middle) and cloned into the plasmid to create pBAV1k-T5-luxABCDE (bottom circle). doi:10.1371/journal.pone.0013244.g001 [12]. The ORF D protein may play a role in copy number regulation, but is not essential for replication [7]. Three different inverted repeats (IRIV, IRV and IRVI), which may serve as an alternative SSO, and ORF D were deleted (Fig. 1). The ORF C protein is a negative regulator of P1 promoter that regulates the expression of repA [7]; we chose to retain it to moderate the risk of runaway plasmid replication. In addition, the T1 and t0 transcription terminators [13] were inserted on opposite sides of the origin of replication to prevent RNA read-through to the plasmid origin part. The terminators should prevent possible antisense interference by the wild-type RNA transcript with repA expression and SSO function.
We completed our expression vector by cloning a selectable marker, a multiple cloning site, a promoter and reporter genes into the plasmid (Fig. 1). Each element was selected for its ability to function in the widest variety of hosts. The 39,50-aminoglycoside phosphotransferase type III, under its own promoter, confers kanamycin resistance upon many bacterial species [14]. We introduced a BioBrick multiple cloning site (EcoRI-NotI-XbaI-insert-SpeI-NotI-PstI) so that our vector, pBAV1K-T5, which does not otherwise contain those restriction sites, would be compatible with this standard [15]. We subsequently cloned a BioBrick containing a T5 promoter and two lacO operators [16] into the multiple cloning site (pBAV1K-T5). The gfp (0.8 kb), gusA (1.8 kb) and lacZ (3 kb) reporter genes, and the luxABCDE reporter operon (5 kb), were separately cloned downstream of the promoter (pBAV1K-T5-gfp, pBAV1K-T5-gusA, pBAV1K-T5-lacZ and pBAV1K-T5-lux).
Like most other molecular biologists, we use E. coli as a cloning vehicle, even if the final construct intended to be used in other species. We transformed E. coli with pBAV1K-T5-gfp, or control RCR plasmids pGK12 (derived from pWV01) and pLZ12-T5-gfp (derived from pSH71 [17] and has high level of homology to pWV01) or ColE1 plasmids (pQBAV, or pIMBB). The transformants were propagated in liquid LB cultures and lysed. The yield of plasmid from the pBAV1K-T5-gfp transformant was comparable to those of the ColE1-derived plasmids (Fig. 2a). The visibly higher amount of purified vector for pBAV1K-T5-gfp, relative to parental plasmids pGK12 and pLZ12-T5-gfp, was confirmed by Real-Time qPCR analysis (Fig. 2b). pBAV1K-T5-gfp replicates to 357 copies per cell in MDS42recA-Blue and 251 copies per cell in INV alpha F9. In contrast, pGK12 replicated only to 60 copies/ cell in E. coli [5].
We were concerned about the effect of high plasmid copy number (runaway replication) on cell fitness, so we used flow cytometry to assess the genetic stability of the plasmid of the Figure 2. Plasmid pBAV1K-T5-gfp replicates to high copy number in Escherichia coli. (A) E. coli was transformed with plasmids pBAV1K-T5gfp, pLZ12-T5-gfp, pGK12 (two other RCR plasmids), pQBAV3Cm-T5-gfp or pIMBB-T5-gfp (two ColE1 derived plasmids). The transformants were propagated in liquid LB cultures supplemented with the appropriate antibiotics. The plasmids were purified, and 2 microliters of each were analyzed on a 0.8% agarose gel. The higher yield and faster mobility of the pBAV1k relative to the larger pWV01 derivatives indicates supercoiling. (B) Five different species of bacteria (namely Agrobacterium tumefaciens, Streptococcus pneumoniae, Bacillus subtilis, Acinetobacter baylyi ADP1 and E. coli) were transformed with pBAV1K-T5-luxABCDE ( Table 3). The APH(39)-IIIa gene present on the plasmid was used as a target to estimate the copy number in reference to the chromosomal relA/spoT gene (or its homolog) by quantitative real-time PCR. Each bar represents the average of three replicates. Error bars represent standard error. doi:10.1371/journal.pone.0013244.g002 transformed cells. We created T5-lacO-lacO-gfp versions of our vectors (pBAV1K-T5-gfp, pLZ12-T5-gfp, pIMBB-T5-gfp, and pQE30-T5-gfp) in parallel, using overlap extension PCR cloning (Materials and Methods). It proved far easier to clone genes into pBAV1K-T5 than into the other RCR-replicating plasmids (pLZ12 and pGK12). When E. coli were transformed with the in vitro recombination reactions, 172 pBAV1K-T5-gfp transformant colonies formed. In contrast, a single pLZ12-T5-gfp transformant colony grew on the second try, and no pGK12-T5-gfp transformants grew in four attempts. We hypothesized that the poor performance of pLZ12 and pGK12 as cloning vectors reflected their inability to replicate stably in E. coli. Indeed, flow cytometry analysis of mid-log (OD600 = 0.8) showed that populations of E. coli cells transformed with pBAV1K-T5-gfp had much lower proportions of non-fluorescent (and potentially dead) cells than populations of isogenic pLZ12-T5-gfp transformants (Fig. 3).
pBAV1K-T5-luxABCDE plasmid DNA purified from E. coli was used to transform E. coli, A. baylyi, S. pneumoniae, A. tumefaciens and B. subtilis (underlined in Fig. 4). Every species we tested was successfully transformed, except Deinococcus radiodurans, possibly because this species harbors its own incompatible cryptic plasmid [18]. The transformants were propagated in rich media, and the luminescence of 100 microliters of liquid culture was measured. Each species produced light, confirming the broad host specificity of the promoters, ribosome binding sites and selectable markers in our vector (Fig. 5). Light production is a function of many factors, including the plasmid copy number, codon bias, protein folding and the metabolic network of each host species, so the variations between species are difficult to rationalize. Plasmid DNA copy number in E. coli, A. baylyi, B. subtilis and A. tumefaciens was measured by real-time quantitative PCR using whole DNA purified from the bacterial cells; a relative quantification was used for copy number calculation (Fig. 2b). In B. subtilis, pBAV1K-T5-luxABCDE replicates to 10 times the reported copy number of pGK12 in the same species [5].
All bacterial cells transformed with pBAV1K-T5-gfp retain the plasmid when propagated under selective conditions (i.e. in media containing kanamycin). To determine whether the vector could stably replicate under non-selective conditions, E. coli, and B. subtilis were separately transformed with pBAV1K-T5-gfp, pGK12, or pLZ12-T5-gfp. E. coli cells were also separately transformed with pIMBB-T5-gfp or pQBAV3Cm-T5-gfp in parallel. The transformants were separately grown in nonselective broth for 80 generations by serial dilution and agitation, and the percentage of antibiotic-resistant colonies in the total viable count was determined. In E. coli, plasmids pBAV1K-T5-gfp (approximately 18% kanamycin-resistant colonies after 80 generations) exhibited higher stability than either pLZ12-T5-gfp or pGK12 (0% chloramphenicol-resistant colonies after 80 generations, Fig. 6). The specialized ColE1 based plasmids were, however, more stable than pBAV1K-T5-gfp in E. coli (80% and 90% ampicillin-resistant colonies after 80 generation for pQBAV3Cm-T5-gfp and pIMBB-T5-gfp respectively). When B. subtilis transformants were cultured for 80 generations in nonselective broth, pBAV1K-T5-gfp and pLZ12-T5-gfp exhibited greater stability than pGK12 (Fig. 6b). It is possible that the expression of the erythromycin resistance marker from pGK12 plasmid increased the cost of the plasmid in nonselective conditions, resulting in the earlier elimination of the plasmid from the bacterial population.
The pBAV1K-T5-gfp expression vector was also used to transform natural bacterial species in two arbitrarily collected soil samples. These experiments extended the known host-specificityof the plasmid and revealed naturally competent bacterial species. Gene transfer by natural transformation allows bacteria to adapt rapidly to changing environmental conditions. It probably occurs all the time, but is too infrequent to detect under natural conditions, particularly in soil [19,20,21]. The soil was mixed with the aqueous sample of the plasmid, agitated overnight at room temperature and allowed to grow in rich medium supplemented with kanamycin. Fluorescent colonies formed on the plates were collected and later identified by 16S RNA gene sequencing as Peanibacillus sp. S18-36 and YN14-0; Leifsonia sp.L89, WPCB149 and shinshuensis. The pBAV1K-T5 vector will allow researchers to study natural transformation, and evolution, of the soil bacteria. These species are unrelated to the others that we transformed (E. coli, A. baylyi, S. pneumoniae, A. tumefaciens, B. subtilis, B. burgdorferi, A. tumefacienes, Fig. 4). We have also shared the vector with collaborators, and have heard that it also replicates efficiently in A. baumannii, S. pyogenes and F. novicida (personal communication, Justin Gallivan, Julia Bugrysheva and David S. Weiss). We are therefore optimistic that pBAV1K-T5 will function in many other mesophilic eubacteriaThe genetic features of the natural occurring plasmids, including promoters, ribosomal binding sites, terminators, and codon usage, typically coincide with those of their hosts. The plasmid must serve as a template for host replication factors, namely DNA gyrase, DNA ligase, RNA polymerase, DNA polymerase I, and DNA polymerase III; the host ribosome must recognize the ribosome binding site of the repA transcript. Eubacterial genomes have been diverging from their last common ancestor for over three billion years. We were therefore surprised that pBAV1K-T5 replicated in so many species (Fig. 4), imparting kanamycin resistance and reporter protein production upon each. The factors that mediate transcription and translation in these Gram-negative and Gram-positive species must therefore be conserved to a heretofore underappreciated extent.
We intend to distribute pBAV1k-T5 without regard to intellectual property considerations. Microbiologists could employ it as genetic tool to elucidate the pathogenic mechanisms of poorly characterized bacteria. Protein engineers could use it to express their favorite genes in a variety of eubacteria. It is difficult to predict a priori which species will produce the highest yield of any particular protein, but we showed here that E. coli is not always the best choice. Synthetic biologists could better leverage the complex molecular machines that already exist within the vast domain of the eubacteria.

Materials
All primers used in the study (Table 1) were purchased from IDT (Coralville, IA). All plasmids ( Table 2) and bacterial strains (Table 3) were obtained from commercial vendors or collaborators. Restriction enzymes and DNA modification enzymes were from NEB (Ipswich, MA), and reactions were carried out under the recommended conditions. All other chemicals used in the study were from Sigma-Aldrich (molecular biology grade). Plasmid DNA from E. coli was purified with the QIAprep spin miniprep kit as directed by the manufacturer (QIAGEN, Valencia, CA).

Recombinant DNA
We used Overlap Extension PCR cloning [22] or restriction enzymes and ligase, to create recombinant plasmids; BioBricks were assembled as directed in the BioBrick Assembly Manual (NEB, Ipswich, MA). When necessary, DNA fragments were purified from agarose gels with QIAquick-gel extraction kits from QIAGEN (Valencia, CA). DNA fragments and PCR mixtures were analyzed on 0.8% Seakem LE agarose gels (Lonza Rockland, Rockland, ME) using the 1 kb DNA ladder (New England BioLabs, Ipswich, MA) a molecular size standard. DNA was sequenced by Macrogen (Rockville, MD).We propagated and transformed established laboratory bacteria according to published procedures (Table 3).

Transformation and identification of bacteria from soil samples
The soil samples (1 g each) were each mixed with water (100 microliters) and pBAV1K-T5-gfp (10 micrograms) and agitated overnight at room temperature. The sample was diluted with LBkanamycin (1 mL, 100 microgram/mL), agitated for another two hours and spread on LB agar plates supplemented with kanamycin.. After several days of incubation at room temperature, transformed, fluorescent colonies were picked and propagated in liquid LB-kanamycin. Total DNA was prepared with the DNeasy kit (as directed by the Qiagen for gram-positive bacteria). The 16S RNA genes were PCR amplified (with the primers 27F and 1492R [23] ) and sequenced with the same primers and two others (946F and 518R [23]).

Plasmid stability test
Cultures of E. coli cells, each harboring a plasmid (pBAV1K-T5gfp, pGK12, pLZ12-T5-gfp, pQBAV3Cm-T5-gfp or pIMBB-T5-gfp, Fig. 6A) were propagated in LB medium containing the appropriate antibiotic at 37uC from a single colony. An aliquot (10 microliters) of each overnight culture was inoculated in 10 milliliters of fresh LB medium without antibiotic and grown for 24 hrs until the cells reached stationary phase. At this point the OD 600 of the culture was typically between 2.0-2.4, which corresponds to a titer of ,2610 9 cells/milliliter. An aliquot (100 microliters) of this stationary phase culture was used to inoculate 100 milliliters of fresh medium and this process of sub-culturing was repeated for eight days. Since each inoculum was 0.1% (100 microliters in 100 ml) it represented a 1000 fold increase in the cell number. The OD and cfu/milliliter values were similar at the end of each round of sub-culturing. Thus every 24 hr period of growth represents 10 doublings and 10 generations. The fraction of untransformed cells emerging within each culture was calculated by plating appropriate dilutions of this culture onto LB plates (without antibiotics) to get isolated colonies. From each LB agar plate, 96 colonies were transferred onto 96-well plates containing LB plus antibiotic; the fractions of plasmid containing cells were calculated by counting the number of wells that had visible bacterial growth.
B. subtilis cells harboring the plasmid were similarly grown from a single colony in LB medium containing the appropriate antibiotic at 37uC. An aliquot of 10 microliters of this overnight culture was inoculated in 10 milliliters of fresh LB medium without antibiotic and grown for 24 hours until the cells reached stationary phase. At this point the OD 600 of the culture was typically between 1.0-1.3. An aliquot (100 microliters) of this stationary phase culture was used to inoculate 100 milliliters of fresh medium and this process of subculturing was repeated for eight days. To avoid fails-positive results due to integration of the plasmid into the chromosome of B. subtilis, a method different from that used for E. coli was used to calculate the percentage of B. subtilis cells containing the plasmid. Total DNA was purified from 2 mL aliquots of each overnight saturated culture with Qiagen DNeasy kit. The eluate (1 microliters out of a 250 microliter elution volume) was used to transform chemically competent E. coli INV alpha F9 cells. The transformants were spread on LB plates supplemented with appropriate antibiotic; the colonies (corresponding to different generations of the B. subtilis culture) were counted. The number of the colonies obtained with the generation 0 culture was considered 100%. Test plasmid purifications and agarose gel electrophoresis were performed from several colonies to confirm plasmids identities.

Real-Time PCR assays
The copy numbers of the pBAV1K-T5-luxABCDE within different cell types were assessed by real-time PCR. Amplification and detection were carried out in LightCyclerH 480 (Roche) using sequence specific fluorescent probes from ''Universal ProbeLi-braryH'' (Roche). PCR primers were designed using Primer3 software located at the ''Assay Design Center'' of the Roche website. Total DNA from bacterial species was purified with Qiagen DNeasy kit (as directed by the manufacturer for gram-positive bacteria) and quantified in a ThermoFisher Nanodrop spectrophotometer. The ACIAD3326 (relA/spoT homolog), a single-copy gene on the chromosome of A. baylyi, and orthologues in other bacterial species, were used as a references (genes, primers and probes are listed in Table 4). A 59 bp fragment of the gene was amplified with the primers AVB 126 and AVB 127; probe #47 from Universal ProbLibrary (Roche) was used for detection of the product. The APH(39)-IIIa gene was used as a target to estimate the copy number for the vector. A 59 bp fragment of the APH(39)-IIIa gene was amplified with the primers AVB 128 and AVB 129; probe #48 from Universal ProbLibraryH (Roche) was used to detect the product. Both target and reference DNA standards were diluted in 8 to12 serial steps, each applied in duplicate. LightCyclerH 480 Probes Master was used for the preparation of all samples. The PCR conditions included a single denaturation cycle of 95uC for 7 min, followed by 45 cycles of 95uC for 10 s, and combined annealingelongation for 1min at 55uC. All real-time PCRs were done in triplicate and average results are reported.
Bacterial detection GFP+/GFP-discrimination by flow cytometry E. coli INV alpha F9 cells transformed with pBAV1K-T5-gfp, pLZ12-T5-gfp or pIMBB-T5-gfp were grown in LB at 30uC with aeration to the OD 600 = 1.2. A. baylyi ADP1 cells were grown in LB at 30uC with aeration to the OD 600 = 0.8. Cells were then washed twice with M9 minimal media, resuspended in M9, and analyzed by flow cytometry with a FACSCalibur flow cytometer (BD Biosciences; San Jose, California). Cell samples were diluted to approximately 5610 5 cells per ml with M9 minimal media and delivered at the flow rate of 50 to 150 cells/sec. The FSC (Forward Scatter), SSC (Side Scatter), and fluorescence signal were measured. A band pass filter of 530 nm (515 to 545 nm) was used to collect the green fluorescence. All signals were collected by using logarithmic amplifications. A combination of FSC and SSC were used to discriminate bacteria from background. A total of 20,000 events for each sample were collected and analyzed with the CellQuest Pro software.
Luminescence assay E. coli MDS42 recA, A. tumefaciens, B subtilis and A baylyi were transformed with pBAV1K-T5-luxABCDE vector. The transformed cells, and untransformed negative controls were grown in rich media to mid-exponential phase with aeration. S. pneumoniae cells were grown in Todd-Hewitt medium containing 0.5% yeast extract without aeration but were aerated for 2 hours at room temperature before measurement. After incubation, luminescence from 100 microliters of each bacterial culture was measured with a SpectraMax M5 Multi-Mode Microplate Reader (Molecular Devices; Sunnyvale, CA). All the measurements were performed in triplicate independent experiments, each in octuplicate.