Establishment of a Tractable Genetic Transformation System in Veillonella spp.

Abstract

We have constructed the first Escherichia coli-Veillonella shuttle vector based on an endogenous plasmid (pVJL1) isolated from a clinical Veillonella strain. A highly transformable Veillonella strain was also identified. Both the shuttle vector and the transformable strain should be valuable tools for future Veillonella genetic studies.

TEXT

Veillonellae are among the most prevalent and numerically predominant bacteria in the oral microbiota (18). One of the unique characteristics of veillonellae is their inability to ferment sugars and their utilization of lactic acid excreted by other fermentative bacteria as a carbon source. This feature makes veillonellae a central player in the development of multispecies oral biofilms (7, 11, 16). In addition to the human oral cavity, veillonellae are also regular colonizers of the human gastrointestinal tract. Although veillonellae have generally been regarded as commensal bacteria, numerous epidemiological studies have linked them to dental caries (tooth decay) (2, 10) and to the initiation of periodontitis (gum disease) (9, 17). Veillonella species were also found to cause localized and systemic infections in patients with underlying medical conditions (3, 15).

So far, the genus Veillonella consists of 11 species (12). Despite their importance in human health and disease, little is known about their biology and pathogenic potential, largely due to our inability to genetically manipulate this group of bacteria. We report here development of the first genetic transformation system in Veillonella. The system consists of a shuttle vector and a genetically transformable strain of Veillonella atypica. The successful establishment of this tractable genetic system provides an unprecedented opportunity for the research community to understand the biology and pathogenesis of this group of important members of the human microbiota.

Isolation of veillonellae from saliva samples.

We began this study by isolating Veillonella strains from saliva samples upon institutional review board (IRB) approval (no. 14107). An easy isolation method was developed in this study, based on our fortuitous findings that veillonellae could form colonies on top of a lawn of Streptococcus mutans on BHI plates (J. Liu, Z. Xie, J. Merritt, and F. Qi, unpublished data). Briefly, 20 μl of diluted saliva was mixed with 100 μl of overnight cultures of S. mutans and plated on brain heart infusion (BHI) plates. Colonies forming on top of the S. mutans lawns after 48 h of incubation were randomly selected and streaked onto BHIL (BHI supplemented with 0.6% sodium lactate) plates to isolate single colonies. The identities of these colonies were then determined by 16S rRNA gene amplification and sequencing using the primers 27f and 1492r (13). This simple approach proved to be highly selective, since nearly 100% of the colonies were identified as Veillonella. Overall, we obtained 12 different Veillonella isolates from 11 saliva samples, which belong to 4 different Veillonella species: V. atypica (isolates OK2, OK5, OK6, and OK7-1), V. parvula (OK1), V. rogosae (OK3 and OK11), and V. dispar (OK4, OK7-2, OK8, OK9, and OK10). Among the 12 isolates, 3 are resistant (OK7-2, OK10, and OK11) while the rest are sensitive to tetracycline at a concentration of 2.5 μg ml⁻¹.

Identification of an endogenous plasmid.

We performed plasmid isolation from the 12 Veillonella strains using the alkaline lysis method (5). One endogenous plasmid (designated pVJL1) was detected in strain OK1. Sequence analysis of pVJL1 revealed a circular molecule of 4,813 bp with an overall 29.9% G+C content. pVJL1 contains 4 open reading frames (ORFs) encoding proteins larger than 50 amino acids (aa) in length (Fig. 1A). Among the 4 ORFs, only ORF2 encodes a protein which shows similarity to plasmid replication proteins. The deduced amino acid sequence shared 42% and 39% identities with the replication proteins of plasmid pTRACA22 from uncultured human gut bacteria and plasmid pTS1 from oral spirochetes, respectively (4, 8). It also exhibits low homology to two functional regions of the Rep protein encoded by ColE2, a theta-replicating plasmid from Escherichia coli (6), suggesting that pVJL1 might replicate via the theta replication mode. Based on these sequence homologies, ORF2 was designated rep. Like many theta replicons, a cluster of 3 direct repeats (DRI-DRIII) was also found in the upstream region of rep in pVJL1 (Fig. 1B), suggesting that this direct repeat region could function as the origin of replication. We also found two putative promoters for rep, each of which partially overlaps with DRII and DRIII, respectively (Fig. 1B). Three inverted repeat regions (IRI to IRIII) were also identified, which could function as rho-independent terminators (Fig. 1A) (1).

Fig 1.

Properties of plasmid pVJL1. (A) Physical and genetic map of pVJL1. The four ORFs are indicated by solid arrows. The three direct repeats (DRI to DRIII) are indicated by gray arrows. The three inverted repeats (IRs) are depicted as stem-loop structures. (B) Nucleotide sequence analysis of the putative ori of pVJL1. The direct repeat sequences (DRI to DRIII) are underlined. Two predicted promoters, P1 and P2 (indicated by −35 and −10), are in bold. The predicted ribosome binding site (RBS) is underlined, and the translation start codon of rep is indicated by a box.

Construction of an E. coli-Veillonella shuttle plasmid.

To construct an E. coli-Veillonella shuttle plasmid, a tetracycline resistance cassette (PgyrA::tetM) was first constructed by overlapping PCR. The 0.21-kb promoter region of gyrA and the 2.0-kb coding region of tetM were PCR generated with the primer pairs PgyrAF/PgyrAR (5′-AAAACTGCAGTAAAATGCAGGCGAGTGAAG-3′/5′-AATCTCACCTCATTTTACTAAC-3′) and tetMF/tetMR (5′-GTATCAGTGTTAGTAAAATGAGGTGAGATTATGAAAATTATTAATATTGG-3′/5′-CGCGGATCCCAGGACACAATATCCACTTG-3′), respectively, using V. parvula PK1910 genomic DNA as the template, since this strain was sequenced by our group and contains a Tn916-like transposon with a functional tetM gene (F. Qi, unpublished). These two fragments were mixed and used as a template for overlapping PCR with the primer pair PgyrAF/tetMR to generate PgyrA::tetM. PgyrA::tetM was digested with PstI and BamHI and cloned into pBluescript II KS(+) digested with the same enzymes to create pBS-gtetM. pBS-gtetM was linearized with EcoRI and inserted into the EcoRI site of pVJL1 to generate pBSJL1, which was used as the prototype shuttle plasmid for subsequent studies.

Transformation of Veillonella strains.

Since the well-studied laboratory strain PK1910 is tetracycline resistant, we used the nine tetracycline-sensitive Veillonella strains that we isolated as recipients for transformation with the shuttle plasmid pBSJL1. The electroporation procedure was performed as previously described with minor modifications (14). THL medium (Todd-Hewitt broth supplemented with 0.6% sodium lactate) was used to grow Veillonella, and tetracycline (2.5 μg ml⁻¹) was added for transformant selection. Among the 9 strains tested, one (V. atypica PK5) was found to be highly transformable with plasmid pBSJL1; the transformation efficiency reached 5.8 × 10⁴ CFU/μg DNA (data not shown). None of the other 8 strains yielded any transformants. The successful transformation of strain OK5 by pBSJL1 was further confirmed by plasmid preparation from the transformants. Thus, we have successfully constructed an E. coli-Veillonella shuttle plasmid and identified an easily transformable Veillonella strain.

The minimal replicon of pVJL1.

The size of pBSJL1 is 10.0 kb, which would hinder the routine use of this plasmid for large DNA fragment insertions. To construct a robust shuttle vector, we wanted to minimize the size of the plasmid to its minimal replicon. We constructed a series of mutants with a deletion in the pVJL1 portion of pBSJL1 (Fig. 2A). The primer pairs JLF1 (5′-CGGACTAGTGATTGTTGCCTAGAAGTGTC-3′)/JLR1 (5′-TCCCCGCGGGTAGAGATGCCAATATTCCA-3′), JLF2 (5′-CGGACTAGTATGAGTGATATTTTTTTAGT-3′)/JLR1, JLF3 (5′-CGGACTAGTATGAGTGATATTGTGATG-3′)/JLR1, and JLF1/JLR2 (5′-TCCCCGCGGAAATACCTCTAACTTAACTT-3′) were used, respectively, in PCRs with pBSJL1 as the template. These PCR amplicons were digested with SpeI and SacII and cloned into the same sites of pBS-gtetM to create plasmids pBSJL2, pBSJL3, pBSJL4, and pBSJL5, respectively. The replication of these constructs was tested in strain OK5. As shown in Fig. 2A, typical numbers of transformants were obtained with pBSJL1, pBSJL2, and pBSJL3, while no transformants were obtained with pBSJL4 and pBSJL5. Thus, the minimal replicon of pVJL1 was mapped to a region of about 2.2 kb, which consists of the rep gene and its 109-bp upstream region. Based upon these results, we concluded that the upstream distal region containing the two direct repeats (DRI and DRII) and the putative promoter P1 of rep are dispensable whereas the rep ORF and its proximal upstream region containing DRIII and the putative promoter P2 are essential for plasmid replication.

Fig 2.

(A) Identification of the minimal replicon of pVJL1. Shown on the top line is the pVJL1 portion of pBSJL1. Each derivative plasmid contains part of the pVJL1 portion. The direct repeats upstream of rep are indicated by gray arrows, and the inverted repeats are indicated by a stem-loop symbol. The two putative promoters of rep (P1 and P2) are indicated by bent arrows. (B) Stability of pBSJL1 and derivatives in V. atypica OK5 under nonselective conditions. Shown here are the averages of three independent experiments.

Stability of pBSJL1 and derivatives.

The stability of pBSJL1 and its replicable derivatives (pBSJL2 and pBSJL3) was further characterized. We serially passaged strain OK5 carrying the respective plasmid for up to 40 generations in the absence of antibiotic selection, and the percentage of tetracycline-resistant colonies in the population was calculated at each time point. As expected, no plasmid loss was observed in transformants carrying the full-length plasmid pBSJL1 after 40 generations (Fig. 2B). Similar stability was also found for pBSJL2 (Fig. 2B), which contains only the rep gene plus all three DRs of its upstream region (Fig. 2A), indicating that ORF1, ORF3, and ORF4 are not required for replication or stability. Interestingly, the stability for pBSJL3 was dramatically altered: it showed a slow decline from 0 to 16 generations and a drastic reduction from 16 to 24 generations, and at 40 generations, only ∼5% of the population retained the plasmid (Fig. 2B). Since pBSJL3 contains only DRIII in the upstream region of the rep gene, while pBSJL2 contains all 3 DRs of the upstream region, we concluded that DRI and DRII are required for plasmid stability, although they are not required for plasmid replication. Taken together, our results demonstrate that pBSJL2 is the optimal shuttle vector for future studies.

In summary, starting from isolation of Veillonella strains from saliva and plasmids from these clinical strains, we have isolated a Veillonella plasmid, pVJL1, and constructed a shuttle vector, pBSJL1 from this plasmid. Using pBSJL1 as transforming DNA, we then identified an easily transformable strain (OK5) from the Veillonella clinical isolates. Further studies allowed us to identify a minimal replicon of pVJL1, from which a robust shuttle vector, pBSJL2, was constructed. pBSJL2 is much smaller in size than pBSJL1 (7.8 kb versus 10.0 kb) and has the same stability in Veillonella as pBSJL1. Thus, pBSJL2 and V. atypica strain OK5 can be used as a genetic system for future studies of Veillonella. To the best of our knowledge, this is the first and only genetic transformation system ever developed in veillonellae. We believe that the availability of this system will open new ways to better understand Veillonella species, comprising one of the most prevalent yet least understood bacteria in the human microbiota.

Nucleotide sequence accession numbers.

The sequences reported in this article have been deposited in GenBank under accession numbers JN695642 (Veillonella atypica OK5 16S rRNA gene), JQ004006 (pVJL1), and JQ354980 (pBSJL2).