Development of the First Tractable Genetic System for Parvimonas micra, a Ubiquitous Pathobiont in Human Dysbiotic Disease

ABSTRACT Parvimonas micra is a Gram-positive obligate anaerobe and a typical member of the human microbiome. P. micra is among the most highly enriched species at numerous sites of mucosal dysbiotic disease and is closely associated with the development of multiple types of malignant tumors. Despite its strong association with disease, surprisingly little is known about P. micra pathobiology, which is directly attributable to its longstanding genetic intractability. To address this problem, we directly isolated a collection of P. micra strains from odontogenic abscess clinical specimens and then screened these isolates for natural competence. Amazingly, all of the P. micra clinical isolates exhibited various levels of natural competence, including the reference strain ATCC 33270. By exploiting this ability, we were able to employ cloning-independent methodologies to engineer and complement a variety of targeted chromosomal genetic mutations directly within low-passage-number clinical isolates. To develop a tractable genetic system for P. micra, we first adapted renilla-based bioluminescence for highly sensitive reporter studies. This reporter system was then applied for the development of the novel Theo+ theophylline-inducible riboswitch for tunable gene expression studies over a broad dynamic range. Finally, we demonstrate the feasibility of generating mariner-based transposon sequencing (Tn-seq) libraries for forward genetic screening in P. micra. With the availability of a highly efficient transformation protocol and the current suite of genetic tools, P. micra should now be considered a fully genetically tractable organism suitable for molecular genetic research. The methods presented here provide a clear path to investigate the understudied role of P. micra in polymicrobial infections and tumorigenesis. IMPORTANCE Parvimonas micra is among the most highly enriched species at numerous sites of mucosal dysbiotic disease and is closely associated with numerous cancers. Despite this, little is known about P. micra pathobiology, which is directly attributable to its longstanding genetic intractability. In this study, we provide the first report of P. micra natural competence and describe the only tractable genetic system for this species. The methods presented here will allow for the detailed study of P. micra and its role in infection and tumorigenesis.

precipitate, then subsequently examined morphologically by scanning electron microscopy, and finally confirmed with 16S rRNA sequencing (Fig. 1A).
We next generated a selectable marker for use in DNA transformation assays by isolating a spontaneous rifampin-resistant mutant of the wild-type reference strain ATCC 33270. A resulting rifampin-resistant (rif R ) colony of ATCC 33270 was verified by sequence analysis of the rpoB gene encoding the beta subunit of RNA polymerase. A point mutation in this gene confirmed a D501V substitution in RpoB (Fig. 1B), which coincides with rifampin resistance mutations detected in other bacterial species (22,23). Next, genomic DNA (gDNA) was extracted from both the wild-type strain ATCC 33270 (rif S ) and the rifampin-resistant mutant of ATCC 33270 (rif R ) and then transformed into our collection of wild-type P. micra clinical isolates using a transformation protocol (Fig. 1C) similar to that previously employed for Veillonella parvula (24). With this approach, bacteria are exposed to transforming DNA through multiple growth stages and a wide range of cell densities, and the reaction occurs in an environment analogous to that of a biofilm. Surprisingly, we observed natural competence from the majority of strains tested, with particularly high transformation efficiencies in a number of isolates (Table 1). These results confirmed that P. micra is capable of developing natural competence.
Cloning-independent targeted mutagenesis. Following our discovery of P. micra natural competence, it was of interest to determine whether this ability could be exploited for engineering targeted mutations in P. micra. For this, we chose the ermB erythromycin resistance cassette as a selectable marker (25). An allelic insertion mutagenesis construct was created by Gibson assembly of PCR amplicons containing ermB flanked by 1 kb homologous fragments of the EF-Tu-encoding gene tuf and its downstream locus (Fig. 2A). The insertion of the ermB cassette immediately upstream of the FIG 1 Parvimonas micra clinical isolation scheme and transformation assay. (A) (i) Odontogenic abscess samples were collected into anaerobic transport medium and (ii) grown on selective/ differential agar PMM. Putative colonies were (iii) confirmed morphologically and (iv) submitted for 16S rRNA sequencing to verify species identity. (B) Wild-type strain ATCC 33270 (wt rif S ) was grown on rifampin containing sBHI medium to isolate a spontaneous rifampin resistance strain (rif R ). The deduced amino acid sequences of the wild-type rif S and rif R RpoB proteins are shown. The rif R mutation is shown in red font. (C) Plate transformation assays were performed by the addition of (i) DNA and (ii) P. micra to sBHI agar. (iii) After 24 h of growth, bacteria were collected into sBHI medium and plated on sBHI medium supplemented with the appropriate antibiotic.
native tuf terminator ensured its robust expression. Transformation of this construct into the clinical strain A28 resulted in the insertion of ermB immediately downstream of the tuf open reading frame (ORF) due to a double-crossover allelic exchange event ( Fig. 2A and B). Next, replicate constructs were generated for each isolate and then transformed into their respective strains, including the reference strain ATCC 33270. In each of these reactions, background erythromycin resistance was undetectable, whereas 4 out of 5 isolates, including ATCC 33270, exhibited detectable levels of natural competence (Fig. 2C).
Comparison of construct homology versus transformation efficiency. In an attempt to further bolster the transformation efficiencies of our isolates, we next tested a variety of mutagenesis constructs (Fig. 3A) to examine the correlation between the sizes of the homologous DNA fragments and the resulting transformation efficiencies (26,27). For this, we focused on isolate A28, which displayed the highest level of natural competence using 1-kb homologous fragments (Fig. 2C). No transformants were detected in the negative-control samples or when using mutagenesis constructs containing homologous flanking regions of 250 bp (Fig. 3B). An increase in the size of the homologous fragments from 1,000 bp to 1,750 bp resulted in a .100-fold increase in transformation efficiency. A further increase in homologous fragment length from 1,750 bp to 2,500 bp resulted in only an additional ;15% increase in transformation efficiency, which suggested that transformation rates were approaching their maximum. Given the significant impact of larger homologous flanking regions upon the transformation rates of strain A28, we subsequently reexamined the transformation-negative strain A1 (Fig. 2C) using 2.5-kb homologous flanking arms. With this larger construct, it was then possible to detect reliable, albeit low, levels of transformation in this strain (Fig. 3B). These results demonstrate that the size of construct homology plays a significant role for natural transformation efficiency in P. micra, with homologous arms of $1,750 bp being optimal for maximal rates of transformation. Targeted mutations of recA in P. micra. We were next interested to determine whether our P. micra transformation protocol could be used for targeted allelic replacement mutagenesis of recA. RecA is a nonessential DNA-dependent ATPase that plays a central role in homologous DNA recombination and repair (28,29). Given the requirement for homologous recombination during allelic replacement mutagenesis, it was anticipated that a recA mutant would exhibit a natural transformation defective phenotype, preventing subsequent mutagenesis using a marked linear DNA construct. Similar to other bacteria, we also anticipated that a recA mutant would exhibit a highly increased sensitivity to DNA damage. We tested this in strain A28 by first transforming a recA deletion construct comprised of an erythromycin resistance cassette (ermB) flanked by 1 kb DNA fragments homologous to the upstream and downstream regions surrounding recA (Fig. 4A). This resulted in a double-crossover allelic exchange deletion mutant in which recA was replaced with ermB ( Fig. 4A). Since the transformation defects resulting from a recA mutation would likely preclude subsequent genetic complementation, it was necessary to first create an ectopically expressed recA knock-in mutant to serve as a recipient strain for a deletion of the native recA gene. A mutagenesis construct comprised of the recA open reading frame (ORF) followed by a kanamycin resistance cassette (aphAIII) (25) was ligated to 2.5 kb segments of the tuf gene and its downstream sequence (Fig. 4B). After transforming this construct into strain A28, a double-crossover event resulted in the insertion of an ectopic copy of the recA ORF and aphAIII directly downstream of tuf (Fig. 4B). In this double recA-expressing background, we subsequently deleted the native recA gene using the same recA deletion construct as that described above (Fig. 4A). Consequently, the resulting erythromycin/kanamycin-resistant transformant harbored a deletion of the endogenous recA while also expressing an ectopic copy of recA together with the tuf gene as part of an artificial operon (Fig. 4B). We subsequently compared the transformation efficiencies of these recA mutant strains using a PCR-generated amplicon of the rif R rpoB gene previously generated in the strain ATCC 33270 (Fig. 1B). The double recA-expressing strain (1A) and the complemented recA deletion strain (1A/DA) both exhibited transformation efficiencies similar to that of the wild type (WT), whereas the DrecA deletion strain (DA) yielded no detectable levels of natural competence (Fig. 4C). As an independent verification of the recA deletion phenotype, we further tested these strains for sensitivity to the genotoxic agent mitomycin C (MMC). As expected, the DrecA strain (DA) displayed a greater sensitivity to MMC at all of the tested dosages (0.5 to 4 mg mL 21 ), with a .3-log reduced survival rate observed at the highest MMC dosage (Fig. 4D). In agreement with the transformation results, both the single and double recAexpressing strains behaved similarly to the wild type. Taken together, these results demonstrate the successful implementation of a highly efficient P. micra mutagenesis protocol that can be employed for both targeted gene deletion and genetic complementation using naturally transformed PCR products.
Creation of a renilla luciferase reporter. The high quantum yield and low background of luciferase-based reporter assays can provide a rapid and precise measure of gene expression over an exceptionally wide dynamic range. The superior signal-tonoise ratio of bioluminescent reporters has also supported noninvasive biophotonic imaging studies of mice orally infected with human oral streptococci (30). To determine whether similar luciferase reporters would also function in P. micra, we developed a constitutive green renilla luciferase expression strain by creating a transcriptional fusion of the tuf gene, green renilla (renG) luciferase, and a kanamycin resistance cassette (aphAIII). This construct was inserted immediately downstream of the tuf ORF but upstream of the endogenous tuf terminator (Fig. 5A). Green renilla luciferase (RenG) has been successfully employed in several species of oral Streptococcus and displays a signal intensity much higher than that of firefly luciferase (30). Luciferase assays of the P. micra reporter strain demonstrated extremely high levels of bioluminescence that were .5 logs above the background readings of the unmodified parent strain A28 (Fig. 5B). To gauge the detection limits and precision of the luciferase reporter readings in this strain, we next examined the correlation between population size and luciferase bioluminescence values. A dilution series of the reporter exhibited an extremely strong linear correlation to bioluminescence readings, with precise reporter values detectable in as few as 10 2 to 10 3 total CFU (Fig. 5C). This suggests that luciferase reporter assays can be accurately performed within an exceptionally wide range of P. micra cell densities. Tunable gene expression system in P. micra. The efficacy of the green renilla luciferase reporter system subsequently provided an avenue to further develop a regulated gene expression system in P. micra. Ideally, such systems would display a low basal The tuf gene Rho-independent terminator is indicated by a ball and stick icon. (B) Luciferase activity was measured in wild-type strain A28 and the luciferase reporter strain (c-renG). Normalized luciferase activity (RLU) was calculated by dividing raw luciferase values by the culture OD 600 values. (C) Correlation between viable cell counts (CFU) and luciferase reporter values in the constitutive luciferase reporter strain c-renG (R 2 = 0.99). All data points represent the average values 6 SD from triplicate independent determinations. Statistical analysis was performed using a two-tailed Student's t test. *, P , 0.05. expression along with strong and tunable levels of induction. We first tested the xylose-inducible expression cassette Xyl-S1 developed for oral Streptococcus species (31). However, this yielded only nominal xylose-dependent regulation in P. micra. Therefore, we were curious whether a posttranscriptional regulatory system, such as those of ligand-inducible riboswitches (32), might perform better. To test this, we inserted a theophylline riboswitch (32) between the renG ORF and an ectopic copy of the constitutively expressed rpoB promoter. Like the previous renG reporter construct (Fig. 5A), we similarly inserted the riboswitch-controlled reporter construct immediately after the tuf ORF (Fig. 6A). However, for this construct, we also included a second ectopic copy of the endogenous tuf transcription terminator upstream of the construct to minimize transcriptional read-through from the tuf gene (Fig. 6A). The secondary structure of the theophylline riboswitch is shown in Fig. 6B. This strain was assayed for luciferase reporter activity over a range of theophylline concentrations, and we observed a titratable dose response curve with ;9-fold dynamic range (Fig. 6C). This result is highly consistent with previous studies employing this riboswitch in other species (32). Encouraged by the inducibility of this riboswitch in P. micra, we sought to improve upon its dynamic range by introducing a series of point mutations into the riboswitch. Since the theophylline riboswitch is predicted to fold into a series of stable stem-loop structures (Fig. 6B), we were curious whether the inclusion of a poly-uracil tract immediately after the terminal stem-loop would create a Rho-independent terminator-like element to synergize with the ribosome binding site sequestration already present within the riboswitch, further increasing the magnitude of riboswitch regulation. A similar dual regulatory mechanism has been reported for the flavin mononucleotide riboswitch found on the ribB gene of Escherichia coli (33). Furthermore, mFOLD predictions suggested that these mutations could be introduced without triggering obvious changes to the overall riboswitch secondary structure (Fig. 6B). After examining the dose response curve of the modified riboswitch reporter strain, we indeed observed a greatly enhanced response to theophylline, resulting in a .70-fold dynamic range (Fig. 6C). Together, these results demonstrate the utility of renG-encoded luciferase as a sensitive reporter protein in P. micra as well as a riboswitch-based tunable gene expression system that exhibits low basal expression and robust inducibility.
Development of in vitro transposon mutagenesis for Tn-seq studies. Transposon sequencing (Tn-seq) is a powerful forward genetic screening tool that has been successfully employed for genetic fitness studies of a number of pathogens (34) and pathobiont microbiome species (25,35). Given the high level of natural competence observed in P. micra, we were curious to determine the feasibility of implementing in vitro transposon mutagenesis using recombinant MarC9 transposase and the mini-mariner transposon commonly used in Tn-seq studies ( Fig. 7A and B) (25,34,36). Following in vitro transposon mutagenesis of P. micra gDNA, the reactions were transformed directly into the wild-type strain A28. With this approach, we readily obtained .6,000 transposon mutants per microgram gDNA ( Fig. 7C and D). Considering that the P. micra genome is ,2 Mb, a single in vitro mutagenesis reaction with several micrograms of P. micra gDNA would be sufficient to create densely saturated libraries of transposon insertions that are directly compatible with Tn-seq protocols. Overall, P. micra should be highly amenable to Tn-seq studies of its pathobiology.

DISCUSSION
P. micra is a notable pathobiont in the oral cavity due to its strong association with a variety of inflammatory oral diseases. P. micra is also commonly identified as a major constituent of acute and chronic infections at numerous other sites in the human body, including the abdomen, chest, spine, brain, blood, skin, and urogenital tract (37,38). Many P. micra infections have been described as polymicrobial, though a number of reports have implicated it as the sole source of infection as well (39,40). Interestingly, the presence of P. micra has also been strongly associated with multiple types of cancer. Numerous reports have identified a major enrichment of P. micra in colorectal tumors (5, 41-43), while others have reported an association of P. micra with oral and gastric cancers (4,43,44). Given the substantial body of evidence for its role in human disease, the development of new genetic tools for P. micra is an essential step to reveal its unknown pathogenic mechanisms. The P. micra genome contains a number of genes with putative functions essential for natural competence in organisms such as Haemophilus influenzae and Bacillus subtilis, including predicted traffic NTPases (PilB, ComGA) and pilin/pseudopilins (PilA, ComGC). This led us to consider the possibility that P. micra may also be naturally transformable. To the best of our knowledge, our demonstration of P. micra natural competence is not only a first for the species but is also likely the first report of natural competence within the entire Tissierellia class, which is a large subgroup in the Firmicutes phylum composed largely of microbial dark matter. The use of natural competence for DNA transformation has numerous advantages over artificial methods, such as chemical treatments (e.g., CaCl 2 ) and electroporation (45). These approaches are also difficult to optimize for many fastidious anaerobes like P. micra, due to their sensitivity to excessive handling. Natural transformation can also be more efficient than artificial methods, as naturally transformed DNA enters the bacterial cytosol single-stranded (46). Unlike duplex DNA introduced by artificial means, single-stranded DNA (ssDNA) is largely resistant to degradation by most bacterial restriction enzymes (47,48). Furthermore, ssDNA can serve as a more efficient substrate for homologous recombination than duplex DNA. The recombinase RecA is specifically activated by the presence of ssDNA to form presynaptic filaments, which in turn promote RecA to scan for homologous sequences, induce synapse formation, and ultimately mediate recombination (49). In our study, all of the P. micra isolates tested displayed some level of natural competence (Fig. 2C and Fig. 3B), though it is currently unclear why some strains exhibited higher transformation rates than others. A number of factors could contribute to these disparities, such as strain-specific differences in the expression of competence related genes, variable responses to competence-specific environmental cues, and/or an altered expression of transformation-limiting bacterial factors (e.g., nucleases) (50). Our results demonstrate that some of these limiting factors can be mitigated by increasing the lengths of the homologous fragments contained on the mutagenesis constructs (Fig. 3B) (51). Based on this, we speculate that endogenous exonuclease activity is likely to be a major determinant of the transformation efficiency observed with naturally transformed linear DNA constructs. However, it is also possible that longer segments of homology simply lead to higher rates of recombination.
After observing robust natural competence in P. micra, we examined the feasibility of creating recA deletion mutants. We chose recA as a test case due to its anticipated and easily measured phenotypes. As expected, the recA mutant exhibited a total loss of detectable natural transformability as well as an MMC-sensitive phenotype (Fig. 4C and D). We were also able to complement these phenotypes by employing allelic exchange to insert the recA ORF immediately downstream of the constitutively expressed tuf gene, creating an artificial tuf-recA operon (Fig. 4B). This further illustrates a straightforward approach for genetic complementation without requiring a shuttle vector.
The subsequent expression of luciferase in P. micra led us to consider its use as a reporter for the development of a regulated gene expression system. Previously, we developed a highly efficient xylose induction system (XylR/O) that functioned in several oral Streptococcus species (31). Repressor/operator-based systems such as XylR/O (xylose) and LacR/O (IPTG) are contingent upon the uptake of inducer molecules as well as the heterologous expression of the corresponding repressor proteins. These can be of limited utility for bacterial species in which repressor protein expression or inducer permeability may be limiting factors. Indeed, our attempts to develop a XylR/O system in P. micra were largely unsuccessful. This led us to consider an alternative approach using a riboswitch-based regulatory system for posttranscriptional control of gene expression. Using a synthetic theophylline riboswitch (32), we demonstrated the potential utility of theophylline induction in P. micra, achieving up to an ;9-fold maximum induction of luciferase activity (Fig. 6C). In an attempt to further improve upon its performance, we introduced a series of riboswitch point mutations that were predicted to create a Rho-independent terminator-like structure in the absence of theophylline ( Fig. 6A and B). Further studies would be required to confirm whether the mutant riboswitch truly functions in transcription termination. Regardless, these mutations were able to further increase the dynamic range of the theophylline riboswitch by nearly an order of magnitude (Fig. 6C). This new version that we now refer to as the Theo1 riboswitch yielded low basal expression while providing robust and tunable inducibility. In addition to genetic complementation, this induction system is likely applicable for studies of both essential and toxic genes. It may also be useful for controlling target transcript levels and the timing of gene expression during infection studies. Tn-seq is a powerful forward genetic screening tool used for genome-level assessments of genetic fitness. Based upon the efficacy of P. micra natural transformation, we predicted that it would be well suited for in vitro mariner-based transposon mutagenesis. Considering that the Himar transposase specifically targets TA dinucleotides (36), the high A1T content of the P. micra genome (.70% A1T) makes it especially well suited for Tn-seq analysis with Himar/mariner. The availability of a high-efficiency natural competence protocol also circumvents the requirement for complex in vivo transposition methods employing conjugation, temperature-sensitive plasmids, and ectopically expressed transposases. Based upon the results shown in Fig. 7C and D, the current protocol should easily achieve the mutagenesis thresholds required for highdensity Tn-seq analysis. Thus, when combined with the other genetic tools described in this study, one can now reliably perform nearly all of the molecular genetic approaches required to characterize P. micra pathobiology. As such, P. micra should also be added to the list of genetically tractable oral microbes.

MATERIALS AND METHODS
Isolation of Parvimonas from clinical odontogenic abscess samples. Clinical sample collection and strain isolation protocols were reviewed by the Oregon Health and Science University Institutional Review Board (IRB) prior to the initiation of the study and deemed to be not human subjects research. All clinical specimens were collected by clinicians in the OHSU Pediatric Dental clinic using materials generated during routine treatment procedures. Specimens were deidentified of all protected health information (PHI) as part of the collection protocol. Clinical specimens were derived from a pediatric cohort undergoing tooth extraction due to odontogenic abscesses.
Abscess samples were collected using Dacron swabs and immediately placed in prereduced transport medium (21). Abscess samples were vortexed and plated onto agar PMM (12). Plates were incubated in an anaerobic chamber (85% N 2 , 10% CO 2 , 5% H 2 ) at 37°C for 4 to 7 days. Bacterial colonies producing a black precipitate were passaged on fresh PMM plates. Candidate isolates were imaged by microscopy for the expected cellular morphologies and then screened by PCR using the primers 16SFOR/REV. Sequence identification of 16S rRNA was confirmed using the expanded Human Oral Microbiome Database (www.homd.org) (52).
To isolate spontaneous rifampin-resistant mutants, the P. micra reference strain ATCC 33270 was plated on sBHI agar containing rifampin (5 mg L 21 ). A rifampin-resistant mutant (rif R ) was screened by PCR using the primers rpoB-1240FOR and rpoB-2095REV. Sequence analysis of this conserved region of rpoB confirmed an A-to-T transversion resulting in an aspartic acid-to-valine substitution at amino acid 501 of RpoB.
Electron microscopy. Bacteria were grown overnight in sBHI on indium-tin-oxide (ITO) coverslips and fixed in 2.5% glutaraldehyde plus 2.5% formaldehyde in sodium cacodylate buffer (pH 7.4). Samples were rinsed in buffer, postfixed in 2% osmium tetroxide for 30 min, rinsed in water, dehydrated in a graded ethanol series, and critical point dried in a Leica CPD300. Samples were mounted on aluminum stubs with a carbon tab and coated with 8 nm carbon using a Leica ACE600 coater. Imaging was performed on a ThermoFisher Scientific Helios NanoLabG3 DualBeam scanning electron microscope.
Extraction of genomic DNA from P. micra. Genomic DNA (gDNA) was purified using a phenol-chloroform extraction procedure (53). Bacteria were harvested from sBHI agar plates, suspended in STES buffer (0.5 M NaCl, 0.02 M EDTA, 0.2 M Tris-HCl [pH 8], 20 mg mL 21 lysozyme), and incubated at 55°C for 2 h. SDS was added to a final concentration of 1% followed by the addition of proteinase K (0.1 mg mL 21 ) and incubation at 55°C for 1 h. One volume of phenol/chloroform/isoamyl alcohol (25:24:1; Fisher, Fair Lawn, NJ) was added to the suspension and mixed gently. After centrifugation at 16,000 Â g, the aqueous layer was collected and treated with RNase (50 mg mL 21 ) at 37°C for 2 h followed by a second extraction with phenol/chloroform/isoamyl alcohol. A 1:10 volume of 3 M sodium acetate (pH 5.5) and 3 volumes of ice-cold ethanol were added and incubated at 220°C for 24 h. DNA was pelleted by centrifugation at 16,000 Â g at 4°C for 30 min and subsequently washed with ice-cold ethanol (70%). The DNA pellet was dried in a laminar flow hood and resuspended in double-distilled water (ddH 2 0).
DNA transformation experiments. Transformation assays were performed similarly as described previously (24). Briefly, P. micra was grown on sBHI agar for 72 h. Cells were harvested from plates, suspended into liquid sBHI, and adjusted to an optical density at 600 nm (OD 600 ) of 0.4. Unless otherwise indicated, a total of 1 mg of gDNA (dissolved in 20 mL of ddH 2 0) was spotted onto sBHI agar plates. After complete absorption of the gDNA into the agar, 20 mL of each bacteria suspension was pipetted over the DNA spots and absorbed. Plates were incubated in anaerobic conditions at 37°C for 24 h. Following the incubation period, cells were then harvested into 50 mL of sBHI and vortexed. Serial dilutions were spread onto sBHI agar plates containing the appropriate antibiotic as well as on nonselective sBHI plates to determine total CFU counts. Transformation efficiency was calculated as the ratio of antibiotic-resistant transformants to total CFU. For allelic replacement mutagenesis and genetic complementation, transformations were performed similarly as described above.
P. micra ermB insertion mutagenesis constructs. Primers and mutagenesis constructs were designed using Serial Cloner software (https://serial-cloner.en.softonic.com). All primers used in this study are listed in Supplemental File 1. All PCR amplifications were performed using Phusion high-fidelity polymerase (Fisher, Fair Lawn, NJ). For insertion mutagenesis constructs, an erythromycin resistance cassette (ermB) was inserted directly downstream of the tuf gene to ensure robust expression. The ermB cassette, originally isolated from Staphylococcus aureus, has shown diverse utility in both Gram-positive and Gram-negative bacteria (54). This cassette naturally lacks a transcription terminator. Therefore, the ermB cassette containing its endogenous promoter and ORF was amplified using primers ermR-FOR/REV and the plasmid pJY4164 as a PCR template (25). Upstream homologous regions were generated using the primer pair tuf-FOR/REV for all isolates. For downstream homologous regions, the primer pair down-tuf-FOR/REV was used with isolates A1, A3, and A28, while down-tufFORa/tuf-REV was used for strains A11 and ATCC 33270. gDNA from each respective P. micra strain was used as the PCR template. All amplicons were screened for size, column purified (Qiagen, Germantown, MD), and assembled using Gibson assembly master mix (NEB, Ipswich, MA) per the manufacturer's instructions. The assembled construct was amplified by PCR using the primer pair tuf-FOR/down-tuf-REV, screened for size, and column purified. Transformation assays using these constructs were performed as described above. The primer pair tuf-FOR*/down-tuf-REV* was used to screen transformants for ermB insertion.
To generate ermB insertion constructs with flanking regions of various sizes, gDNA from an ermB insertion mutant (ermR A28) was used as the template. The primer pairs tuf-250-FOR/REV, tuf-FOR/ down-tuf-REV, tuf-1750-FOR/REV, and tuf-2500-FOR/REV were used to generate homologous fragments of 250 bp, 1.0 kb, 1.75 kb, and 2.5 kb, respectively. For transformation assays, a molar equivalent to 1 mg of the 1.0 kb homologous flank construct was used for all constructs. Transformation assays with these constructs were performed as described above.
P. micra recA deletion and genetic complementation. For construction of the recA deletion mutant, a mutagenesis construct consisting of an ermB cassette flanked by homologous upstream and downstream regions to recA was assembled. The ermB cassette was amplified using the primer pair ermR-FOR/REV and plasmid pJY4164 as the PCR template. Upstream and downstream homologous regions to recA were generated using the primer pairs up-recA-FOR/REV and down-recA-FOR/REV, respectively, with strain A28 gDNA serving as the PCR template. All amplicons were screened for size, column purified, and assembled using Gibson assembly master mix. The assembled construct was amplified by PCR using the primer pair up-recA-FOR/down-recA-REV, screened for size, and column purified. Transformants were selected on agar plates supplemented with erythromycin.
To generate the recA knock-in strain, a construct was made containing the recA ORF along with a kanamycin resistance cassette aphAIII inserted immediately downstream of the tuf gene. The aphAIII gene was originally isolated from Enterococcus faecalis and has exhibited broad utility in numerous bacterial genetic systems (55). To accomplish this, the PCR fragments homologous to tuf and its downstream sequence were generated using the primer pairs tuf-FOR/tuf-REVa and down-tuf-FORb/tuf-2500-REV, respectively, with A28 gDNA serving as the PCR template. The recA gene was amplified using the primer pair recA-FOR/REV and A28 gDNA as the template. The kanamycin resistance cassette aphAIII was amplified with its endogenous transcription terminator using the primer pair kan-FOR/REV with plasmid pWVTK as the template (25). All amplicons were screened for size, column purified, and assembled using Gibson assembly master mix. The assembled construct was PCR amplified using the primer pair tuf-FOR/ tuf-2500-REV and then column purified. Transformants were selected on agar plates supplemented with kanamycin. A complemented DrecA mutant strain was generated by transforming the recA knock-in strain with the recA deletion construct described above. Transformants were selected on agar plates supplemented with kanamycin and erythromycin.
To compare transformation efficiencies of the various recA mutants, PCR amplicons were generated with the primer pair rpoB-1240FOR/2095REV using gDNA from the spontaneous P. micra rifampin-resistant strain described above. PCR products were screened for size and column purified. Transformation assays were performed as described above using sBHI 6 rifampin. For mitomycin C (Sigma, St. Louis, MO) sensitivity experiments, each strain was suspended in sBHI at an OD 600 of 1.0 and then serial dilutions were plated on sBHI agar plates containing various dosages of MMC (0 mg mL 21 , 0.5 mg mL 21 , 1 mg mL 21 , 2 mg mL 21 , and 4 mg mL 21 ). Plates were incubated in an anaerobic chamber at 37°C for 72 h.
Green renilla luciferase expression in P. micra. To express green renilla luciferase in P. micra, a construct was assembled containing the renG ORF along with a kanamycin resistance cassette aphAIII inserted immediately downstream of the tuf gene. PCR fragments homologous to tuf and its downstream sequence were generated using the primer pairs tuf-2500-FOR/tuf-REVa and down-tuf-FORb/tuf-2500-REV, respectively, with A28 gDNA as the template. A PCR fragment of renG was generated using the primer pair renG-RBS-FOR/renG-REV with gDNA from strain brsRM-renG as the template (25). The kanamycin resistance cassette aphAIII was amplified using the primer pair kan-FOR/REV and plasmid pWVTK as the template. All amplicons were screened for size, column purified, and assembled using Gibson assembly master mix. The assembled construct was amplified by PCR using the primer pair tuf-2500-FOR/REV, screened for size, and column purified. The amplicon was transformed into wild-type P. micra strain A28 and the transformants were selected on agar plates supplemented with kanamycin.
Construction of P. micra-inducible luciferase strain. For inducible expression of renilla luciferase in P. micra, a construct containing the tuf gene rho-independent terminator followed by a rpoB promoter and riboswitch preceding renG followed by a kanamycin cassette was inserted directly downstream of the tuf gene. PCR fragments homologous to tuf, including its terminator and its downstream sequence, were generated using the primer pairs tuf-2500-FOR/tuf-53-REV and down-tuf-FORb/tuf-2500-REV, respectively, using strain A28 gDNA as the template. The riboswitch regulatory region was PCR amplified with the primer pair ribo-FOR/REV using a synthetic riboswitch DNA fragment (IDT, Coralville, IA) as the template (see Supplemental File 1). The renG gene was amplified with the primer pair renG-FOR/REV using gDNA from strain brsRM-renG. The kanamycin resistance cassette was amplified using the primer pair kan-FOR/REV and plasmid pWVTK as the template. All amplicons were screened for size, column purified, and assembled using Gibson assembly master mix. The resulting construct was PCR amplified using the primer pair tuf-2500-FOR/REV, screened for size, and column purified. This product was transformed into wild-type P. micra isolate A28, and transformants were selected on agar plates supplemented with kanamycin.
For the Theo1 riboswitch variant, a construct was generated using the primer pairs tuf-2500-FOR/theoplus-ribo-REV and theo-plus-renG-FOR/tuf-2500REV using gDNA from the inducible mutant described above. PCR products were assembled using Gibson assembly master mix and amplified using the primer pair tuf-2500-FOR/REV. Transformants were selected on sBHI agar plates containing kanamycin.
Luciferase assays. RenG expression strains were grown for 72 h on sBHI agar plates supplemented with kanamycin. Bacteria were suspended in liquid sBHI medium with a range of theophylline concentrations (0 mM, 0.0078 mM, 0.0156 mM, 0.03125 mM, 0.0625 mM, 0.125 mM, 0.25 mM, 0.5 mM, 1 mM, 2 mM, 4 mM, and 8 mM) at an OD 600 of 0.1 and then incubated in an anaerobic chamber at 37°C for 20 h. Coelenterazine-h solution (Prolume, Pinetop, AZ) was added to each sample (3.4 mg mL 21 ), and luciferase activity was measured on a Promega Glomax Discover luminometer. Optical densities were measured immediately after measuring luciferase activity for normalization. Normalized activity was expressed as relative light units (RLU), which is the luminescence value/OD 600 .
Generation of a transposon insertion library in P. micra. A transposon library was generated in P. micra strain A28 using a previously described protocol (34). Briefly, in vitro transposon mutagenesis was performed by combining 1 mg of gDNA from isolate A28, 1 mg of the mariner transposon-containing vector pT-MGL-erm (25), and various amounts of the MarC9 transposase (0 mg, 0.125 mg, 0.25 mg, 0.5 mg, and 1 mg). The mixture was incubated for 30°C for 1 h followed by a 75°C incubation for 10 min and then incubation on ice. The DNA was precipitated with ethanol and transposon junctions were subsequently repaired. The resulting transposon reactions were transformed directly into P. micra isolate A28. Serial dilutions were plated on nonselective sBHI agar plates for total counts, while transposon mutants were selected on sBHI agar plates supplemented with erythromycin.
Statistical analysis. All statistical analyses were performed using GraphPad Prism software to calculate significance via two-tailed Student's t test. Statistical significance was assessed using a cutoff value of P , 0.05.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, XLSX file, 0.01 MB.