Development of a Site-Directed Integration Plasmid for Heterologous Gene Expression in Mycoplasma gallisepticum

Deciphering the molecular basis of the interactions between the parasite Mycoplasma gallisepticum and its avian hosts suffers from the lack of genetic tools available for the pathogen. In the absence of well established methods for targeted disruption of relevant M. gallisepticum genes, we started to develop suicide vectors and equipped them with a short fragment of M. gallisepticum origin or replication (oriC MG). We failed to create a disruption vector, although by adding a further short fragment of the M. gallisepticum tufB upstream region we created a “Trojan horse” plasmid. This is fully integrated into the genomic DNA of M. gallisepticum, always at the same site, oriC MG, and is able to carry and express any gene of interest in the genetic background of M. gallisepticum. Successful expression of a heterologous gene was shown with the lacZ gene of E. coli. When used for gene complementation or expression of hybrid genes in M. gallisepticum, a site-specific combined integration/expression vector constitutes an improvement on randomly integrating transposons, which might have unexpected effects on the expression of chromosomal genes.


Introduction
Since the first successful transformation of Mycoplasma gallisepticum with a transposon in 1994 by Cao et al. [1], there has been little progress in expanding the molecular tool box for this avian pathogen. Mycoplasma gallisepticum is the causative agent for Chronic Respiratory Disease in chickens and Infectious Sinusitis in turkeys (for review see 2) and is being continuously investigated, with the focus on vaccine development, deciphering virulence factors and the interplay of host-pathogen-interactions. Unfortunately, genetic studies in M. gallisepticum are hampered by the lack of suitable methods to genetically modify members of the genus Mycoplasma. In addition, there are obstacles to the practical handling of mycoplasmas: as members of the class Mollicutes, mycoplasmas are bacteria without cell walls that most often depend on an animal or plant host, thereby exhibiting a parasitic or commensal life-style. Streamlined by regressive evolution, mycoplasma genomes have undergone a drastic reduction in size and members of Mycoplasma are among the smallest free-living, self-replicating microorganisms. As a results of the loss of many biosynthetic pathways, mycoplasmas are absolutely dependent on many biosynthetic precursors that have to be provided by the host cells (for review see 3). In the laboratory, mycoplasmas need complex growth media and these, together with their fastidious growth characteristics, make handling the bacteria extremely difficult. The presence in Mycoplasmas of potent endonucleases [4,5] provides a further obstacle to the development of transformation methods or plasmid-based expression systems. For a long time only the two natural genetic mobile elements, the transposons Tn4001 and Tn916 [1], were available for the genetic manipulation of M. gallisepticum. Although the M. gallisepticum genome contains sequences thought to be derived from phages [6], no bacteriophage specific for M. gallisepticum has yet been described.
Transformation experiments with a 2-kbp DNA fragment of the Spiroplasma citri chromosome, comprising the dnaA gene with surrounding DnaA boxes, gave the first evidence that this region is the origin of replication of S. citri [7]. Placed on an E. coli plasmid, it enabled the plasmid to autonomously replicate, even after deletion of the dnaA gene [7] In addition, the plasmids have been stably maintained as free extrachromosomal elements [8]. In Mccp an oriC plasmid was used to successfully express the lacZ gene of E. coli and, despite noticeable codon usage differences between E. coli and Mycoplasma, functional E. coli β-galactosidase (β-Gal) was detected in transformed Mccp cells [9].
oriC plasmids have also been created for the purpose of gene inactivation by homologous recombination, for example in M. agalactiae [10] or in Mccp [9]. Janis et al. concluded that the rare events of homologous recombination in Mycoplasmas were potentiated by the use of heterologous oriC plasmids in which the oriC fragment derives from a related Mycoplasma species. This finding was supported by the work of Lee et al. [11], who used the oriC region upstream of the soj gene of M. gallisepticum and its closest relative, M. imitans, to create plasmids that were able to replicate in both species. A targeted disruption of a M. gallisepticum vlhA gene was achieved with a heterologous oriC plasmid, while the homologous oriC plasmid resulted in the integration of the plasmid into the genomic oriC region. While transformants contained the plasmid almost exclusively as an extrachromosomal element during the first few passages, integration started to occur after 10 to 15 passages.
To develop an expression plasmid system for M. gallisepticum, we used a fragment of the M. gallisepticum oriC comprising the smallest oriC fragment, a 180-bp sequence, that was shown by Lee et al. to direct plasmid integration. However, this fragment extends into an additional neighbouring DnaA box. Using the 420-bp fragment in combination with a standard E. coli plasmid, we generated a plasmid that did not autonomously replicate in M. gallisepticum but instead integrated fully and almost exclusively into the M. gallisepticum oriC region from the first passage on. Taking advantage of this feature, we equipped the integration vector with a 203-bp fragment of the M. gallisepticum tufB upstream region, which we suspected to contain a promoter-like structure. We tested the integration/expression vector by subcloning the E.coli lacZ gene and measuring the βgalactosidase activities of the transformants. The results show the general applicability of this plasmid for site-directed gene delivery in M. gallisepticum.

Construction of Plasmids
The tentative disruption vector pDGA1-1 was created by subcloning a 2,754-bp gapA fragment (primers gapAFor10/ gapARev11, see Table 1), amplified by PCR from genomic DNA of M. gallisepticum R low and digested with XhoI, into plasmid pGEM5Zf+ (Promega, Mannheim, Germany) which was linearized by SalI, taking into account that both restriction enzymes generate compatible cohesive ends. Subsequently, the gapA fragment was disrupted by replacing an internal 390bp HincII fragment with a tetPO/tetM cassette generated by PCR (primers McTetFor/McTetRev) and digested with EcoRV, thereby creating a 5´gapA-tetPO/tetM-3´gapA gene fusion ( Figure S1). Plasmid pAM120 [14] served as the template for the tetPO/tetM cassette which confers resistance to tetracycline. For creation of pDGA-oriC, a 420-bp fragment of the genomic oriC region of M. gallisepticum RCL1 was amplified by PCR (primers ori1/ori2), and subcloned via T/A cloning into pGEM-Teasy (Promega). From the resulting plasmid pGEM-oriC, the oriC fragment as well as a ColE1 plasmid replicon were obtained as a 2,005-bp fragment by restriction with SacII and FspI, and ligated with the 4.9-kb SacII/BsrBI-fragment of pDGA1-1 which contained the 5´gapA-tetPO/tetM-3´gapA fragment ( Figure S1). The final integration vector pINT emerged from pDGA-oriC by consecutively deleting a 2,241-bp AgeI/BspHI (3´gapA) and a 779bp EcoRV/ BglII restriction fragment (5´gapA). This deletion strategy removed any gapA sequences from the final construct.
LacZ expression plasmids were created by subcloning first a tufPO PCR product (primers 5TufPO/3TufPO), representing 203 bp of the upstream sequence of M. gallisepticum tufB, to pINT linearized by SacII. Then, a 3.6-kb BamHI fragment of plasmid pAW-lac [15], comprising the full-length lacZ gene of E. coli with its native SD lac , was inserted into the BamHI restriction site behind tufPO. The plasmid with lacZ orientated in the same direction as the tufPO was designated p5TlacZ+, while p5TlacZ-carried lacZ in the opposite direction. For a direct fusion of lacZ with the first codon of tufB´, the SD lac -lacZ fragment of p5TlacZ-was replaced by digestion with BamHI and NarI and insertion of a lacZ fragment produced by PCR amplification (primers lacZ-Bam/lacZ-Nar, template pAW-lac). Plasmid p5TSDlacZ was generated by inserting two complementary oligonucleotides (SDlacF/ SDlacR) into the BamHI restriction site of p5TufPOlacZ. The disruption of the tufB specific reading frame was achieved by restriction of p5TlacZ+ with endonuclease BamHI, and filling in the staggered ends by Klenow polymerase reaction.

Stability assay
After transformation of M. gallisepticum RCL1 with plasmid pINT, ten colonies were picked randomly from HFLX agar plates containing 10µg/ml tetracycline, and inoculated in HFLX medium without tetracycline (Tet). After over night growth, the cultures were passaged by transferring an aliquot into fresh Tet-free medium. The passages of days 0, 6, 11, 15 and 20 were checked for the number of colony-forming units (CFU) by spreading dilutions in triplicates on HFLX agar plates with or without tetracycline. After a ten days incubation period, the CFU were counted using a SMZ-U stereomicroscope (Nikon Corp., Tokyo, Japan).

Southern blot analyses
Mycoplasma genomic DNA of RCL1, and of transformants harbouring pINT was digested with PvuII (Promega), restriction fragments were separated on a 0.8% agarose gel, and blotted on Hybond N+ Nitrocellulose Membranes (GE Healthcare, Freiburg, Germany) following the Neutral transfer protocol provided by the manufacturer. Hybridizations were performed using a DIG-labeled tetM probe and were detected using anti-DIG alkaline phosphatase antibody conjugates (Roche Diagnostics, Mannheim, Germany) and the chemoluminescent substrate CPD-Star (Roche) according to the manufacturer´s instructions. The tetM probe was produced by PCR amplification of pAM120 DNA using primers TetF and TetR ( Table 1).

Quantitation of β-galactosidase activity
For determination of β-Gal activities, colonies of E. coli DH10B transformed with a lacZ expression plasmid were grown in LB medium containing the appropriate antibiotic at 37°C to an optical density at 600 nm wave length (OD 600 ) of 0.4 to 0.9, while M. gallisepticum RCL1 transformants were grown in HFLX to mid-log phase as indicated by a colour change of the growth medium from red to orange. After centrifugation of bacterial samples at 10,000 x g (5 min for E. coli cultures, 10 min for M. gallisepticum cultures), pelleted bacteria were resuspended in 1ml Z-buffer, and the LacZ activities were determined with o-nitrophenyl-β-galactopyranoside according to the method of Miller [16]. Standardization of RCL1 samples was achieved by measuring the total protein content at 562 nm after employing a bicinchoninic acid (BCA) protein assay (Pierce Kit; Thermo Fisher Scientific, Rockford, IL), while for E. coli samples, the cell turbidity expressed as OD 600 values served as a correction factor in the formula for calculating the Miller units: (OD 420 × 1000) / (OD 600 or 562 × ml bacterial sample added × min reaction time).

Creation of a M. gallisepticum integration vector
To create a vector for the targeted disruption of selected genes of the M. gallisepticum genome, the MG gapA gene was subcloned into pGEM5Zf and disrupted by inserting a tetM expression cassette, resulting in plasmid pDGA1-1. The gapA gene was chosen as a candidate target gene to prove the concept of integration, as natural gapAmutants have already been described [12,17], demonstrating that GapA is not essential for the survival of M. gallisepticum. The tetM cassette allows screening of tetracycline-resistant (Tet R ) mycoplasmas, which should arise from successful integration of the gapA::tetM construct by homologous recombination with the genomic gapA sequence. Unfortunately, our attempts to transform M. gallisepticum RCL1 with plasmid pDGA1-1 were not successful. To increase the probability of a homologous recombination event, pDGA1-1 was equipped with a 420-bp fragment of the oriC region of M. gallisepticum, which harbors several DnaA boxes ( Figure 1A). The hope was that this would prolong the residency time of the plasmid within the bacterial cell.
Transformation with the modified plasmid, pDGA-oriC, resulted in tetracycline-resistant(Tet R ) M. gallisepticum transformants, although mainly at low frequencies of about 1.8 x10 -7 per μg plasmid DNA. To our surprise, Southern blot analyses of the transformants indicated that pDGA-oriC did not integrate into its target, the genomic gapA. Instead, the plasmid was found exclusively in the genomic region responsible for the origin of replication (data not shown).
Because the tool box for M. gallisepticum is still limited, we decided to shape this plasmid into a general integration vector by removing the gapA sequences around the tetM gene, designating the result pINT ( Figure 1B). Transformation of M. gallisepticum with pINT resulted in the same transformation efficiencies as seen with pDGA-oriC, and Southern blot analyses showed integration into the oriC MG fragment in all transformants ( Figure 1C). Based on size calculations it was speculated that the entire 6.9-kb plasmid could have integrated into the genomic DNA. To investigate this possibility we undertook PCR analyses with mixed primer pairs specific for plasmid and genomic sequences. The results (data not shown) confirmed that the entire pINT integrated into the genomic DNA and the sequences of the PCR products supported the notion that integration occurred via a single cross-over recombination event, initiated by the short oriC fragment of pINT. Plasmid pINT seemed to become stably integrated: Tet R clones of RCL1oriC::pINT could be grown for 20 passages on tetracycline-free liquid medium and Tet R colonies were still found when the bacteria were replated on tetracycline agar plates. Determination of the CFU confirmed that even in the absence of a positive selection pressure the antibiotic resistance acquired by the integration of the pINT plasmid was not lost over a period of three weeks of continuous growth ( Table 2).

Further Development into an Integrative Expression Vector
To test the usefulness of pINT for heterologous gene expression, we constructed a marker gene expression cassette for the easy detection of gene expression in M. gallisepticum. A 203-bp fragment from the upstream sequence of the M. gallisepticum house-keeping gene tufB, encoding the EF-Tu protein, was generated by PCR and combined with the lacZ gene of E. coli, which encodes the β-galactosidase and has a long history of use as a marker gene (Figure 2) as it can produce a stable, insoluble blue compound from the chromogenic substrate X-Gal [18]. As shown by Knudtson and Minion in 1993, the lacZ gene is expressed in M. gallisepticum, where it is under the transcriptional control of a M. gallisepticum-specific promoter [19]. In our expression cassette, the lacZ gene is preceded by its own ribosome binding site (Shine-Dalgarno sequence, SD) but lacks the lacPO promoter. The cassette was placed between the promoter of the tetracycline resistance gene tetM (tetPO) and the oriC fragment of pINT, giving rise to plasmid p5TlacZ+ ( Figure 3). Colonies of M. gallisepticum RCL1oriC:: p5TlacZ+ turned light blue when grown on agar plates and overlayed with X-Gal, clearly indicating that β-Gal is synthesized by the transformed mycoplasmas (not shown). Quantification of the β-Gal activities of M. gallisepticum RCL1 transformed with p5TlacZ+ or, as a control, with pISMlacZ+, which contains the transposon Tn4001mod with the same lacZ fragment subcloned behind the P out promoter of the IS256L element [20] showed that LacZ expression levels were much higher in RCL1oriC::p5TlacZ+ (186 U/ml) than in RCL1oriC::pISMlacZ+ (1.2 U/ml) (Figure 3). When the lacZ fragment was oriented in the opposite direction to the presumptive tufB promoter in plasmid p5TlacZ-, no β-Gal activity (0 U/ml) was detected in RCL1 transformants. The 203-bp tufB´ fragment thus seems to contain a transcriptionally active element that is able to promote the transcription of lacZ in M. gallisepticum. A recent publication by Panicker et al. supports this notion [21]: using a larger fragment of the tufB upstream region, these workers succeeded in expressing E.coli phoA when the gene was fused to signal and acylation sequences of the MG vlhA1.1 gene and placed behind the suspected tufPO.
To address the contribution of the ribosome binding site of lacZ (SD lac ) or of tufB (SD tuf ) for expression of lacZ in RCL1oriC::p5TlacZ+, we introduced a frame shift in the beginning of the tufB gene, leading to the occurrence of a stop codon 54 nucleotides downstream of the start codon. As the SD-lacZ fragment was unintentionally cloned in frame behind the tufB 5´ region, it was conceivable that we had created a tufB-SD-lacZ hybrid gene that would encode a LacZ protein with a N-terminal fusion that consists of the first five amino acids of TufB, followed by 32 unrelated amino acids. However, plasmid p5TlacZdis, with both, the disrupted tufB frame and the stop codon (Figure 3), was also able to establish lacZ expression in RCL1 transformants (130 U/ml), indicating the importance of SD lac for translation. However, a readthrough of the small ribosomal subunit and reinitiation of the translation at SD lac cannot be ruled out. Further supporting evidence comes from p5TufPOlacZ, where the lacZ fragment of p5TlacZ+ was shortened for the SD lac sequence by directly fusing the lacZ open reading frame with the first codon of the tufB gene. No β-Gal activity (0.01 U/ml) was detected in RCL1oriC::p5TufPOlacZ transformants ( Figure 3). Interestingly, no β-Gal activity (0.02 U/ml) was observed in RCL1 oriC::p5TSDlacZ transformants ( Figure 3). Here, the SD lac sequence was placed immediately upstream of the tufB´-lacZ fusion of p5TufPOlacZ to turn the non-functional p5TufPOlacZ into a lacZ expression plasmid. The absence of β-Gal in RCL1oriC::p5TSDlacZ indicates that SD lac alone is not sufficient for the expression of a TufB-LacZ hybrid product ( Figure 3).
Surprisingly, p5TlacZ+, where lacZ is under control of the M. gallisepticum-derived tufPO, was also able to turn E. coli DH10B into a lacZ-expressing strain, as revealed by the presence of blue colonies on X-Gal agar plates. At 3793 U per ml, the β-Gal expression levels were in the same range as those for pISMlacZ+ (5402 U/ml). The similarly high β-Gal level seen in DH10B(p5TlacZ-) (1920 U/ml) was unexpected. In this plasmid, the lacZ gene is oriented against the tufPO. β-Gal activity can be explained by the occurrence of a remnant lac promoter (lacPO), a constituent of plasmid pGEM5Zf, which is  ). For control reasons, plasmids were created where the SD lac was inserted into tufB´ (p5TSDlacZ), or the tufB frame was terminated by creating a frame shift (p5TlacZdis), or the SD lac -lacZ unit was subcloned in reverse orientation to tufPO (p5lacZ-). For control reasons, the SD lac -lacZ unit was also subcloned into transposon Tn4001mod behind the P out promoter [20] (pISMlacZ+) and in opposite direction (pISMlacZ-). LacZ activities of transformed E. coli or M. gallisepticum were analyzed in 2 (E. coli) or 4 (M. gallisepticum) independent assays and are given as Miller units [16]; however, they should not be compared to each other as different methods for standardization of E. coli and M. gallisepticum samples had to be used. doi: 10.1371/journal.pone.0081481.g003 oriented against the tufPO-lacZ in all other plasmid constructs. As no β-Gal activity was found in RCL1oriC:: p5TlacZ-, the E. coli lacPO does not appear to be functional in M. gallisepticum. It should be emphasized that LacZ units given for E. coli and M. gallisepticum should not be compared to one another, as different methods were used for the standardization of samples of these diverse bacteria. Lacking cell walls, mycoplasmas do not grow as a turbid broth culture even at high numbers, so the OD 600 of mycoplasma cells does not correlate with the number of cells, as it does in E.coli. For M. gallisepticum transformants, protein contents in samples of fresh cultures were quantified by a BCA protein assay, while E. coli was quantified turbidimetrically.
The data using the modified lacZ expression cassettes makes it likely that M. gallisepticum can utilize the SD sequence of lacZ to express β-galactosidase encoded by the E.coli lacZ gene once it is placed behind a 203-bp upstream fragment of tufB. Attempts to create gene fusions of lacZ with tufB failed, presumably because of improper spacing of the presumed SD tuf sequence to the start of the lacZ gene. Interestingly, the fusion of other M. gallisepticum genes such as crmA or mgc2, in frame with the first codon of tufB resulted in successful expression of CrmA or Mgc2 (data not shown). It is unclear why the expression of foreign genes in M. gallisepticum is more sensitive than that of autologous genes.
In summary, we describe the successful delivery of heterologous recombinant DNA to M. gallisepticum. Heterologous DNA is integrated into the bacterial genome at a predetermined site and allows the synthesis of autologous and heterologous gene products in the genetic background of M. gallisepticum. Figure S1. Construction of plasmid pDGA-oriC. Plasmid pDGA-oriC was created by the ligation of a ColE1 / oriC fragment of pGEM-oriC with the tetM-disrupted gapA fragment of plasmid pDGA1-1. It should be mentioned that for creation of pGEM-oriC, a PCR fragment encoding an oriC fragment of M. gallisepticum was subcloned into pGEM-Teasy by exploiting the fact that Taq DNA polymerase exhibits terminal transferase activity. This leads to PCR fragments carrying single 3´-A overhangs at both ends, which can be fused to the 3´-T overhangs of pGEM-Teasy, often referred as T/A cloning. Another noteworthy fact is that for creation of pGA2, a PCR fragment encoding a gapA fragment of M. gallisepticum was subcloned into pGEM5Zf+ via compatible cohesive ends. Such ends were generated by cutting pGEM5Zf+ with SalI, and the gapA PCR fragment with XhoI. Ligation of the compatible cohesive ends did not produce recleavable ligation products, therefore the SalI and XhoI restriction sites are denoted in brackets. Abbreviations: bla, ampicillin resistance gene; blaPO, promoter of bla; tetM, tetracycline resistance gene; tetPO, promoter of tetM; ColE1, origin of plasmid replication; oriC MG, origin of M. gallisepticum genome replication; T/A, insertion site of pGEM-Teasy for DNA fragments carrying single 3´-A overhangs at both ends. (TIF)