A PorX/PorY and σP Feedforward Regulatory Loop Controls Gene Expression Essential for Porphyromonas gingivalis Virulence

ABSTRACT The PorX/PorY two-component system in the periodontal pathogen Porphyromonas gingivalis controls the expression of the por genes, encoding a type IX secretion system, and the sigP gene, encoding sigma factor σP. Previous results implied that PorX/PorY and σP formed a regulatory cascade because the PorX/PorY-activated σP enhanced the por genes, including porT, via binding to their promoters. We recently showed that PorX also binds to the por promoters, thus suggesting that an alternative mechanism is required for the PorX/PorY- and σP-governed expression. Here, our in vitro assays show the PorX response regulator binds to the sigP promoter at a sequence shared with the porT promoter and enhances its transcription, mediated by a reconstituted P. gingivalis RNA polymerase holoenzyme. Merely producing σP in trans fails to reverse the porT transcription in a porX mutant, which further argues against the action of the proposed regulatory cascade. An in vitro transcription assay using a reconstituted RNA polymerase-σP holoenzyme verifies the direct role of PorX in porT transcription, since transcription is enhanced by a pure PorX protein. Accordingly, we propose that the PorX/PorY system coordinates with σP to construct a coherent regulatory mechanism, known as the feedforward loop. Specifically, PorX will not only bind to the sigP promoter to stimulate the expression of σP, but also bind to the porT promoter to facilitate the RNA polymerase-σP-dependent transcription. Importantly, mutations at the porX and sigP genes attenuate bacterial virulence in a mouse model, demonstrating that this regulatory mechanism is essential for P. gingivalis pathogenesis. IMPORTANCE The anaerobic bacterium Porphyromonas gingivalis is not only the major etiologic agent for chronic periodontitis, but also prevalent in some common noncommunicable diseases such as cardiovascular disease, Alzheimer's disease, and rheumatoid arthritis. We present genetic, biochemical, and biological results to demonstrate that the PorX/PorY two-component system and sigma factor σP build a specific regulatory network to coordinately control transcription of the genes encoding the type IX secretion system, and perhaps also other virulence factors. Results in this study verify that the response regulator PorX stimulates the expression of the genes encoding both σP and the type IX secretion system by binding to their promoters. This study also provides evidence that σP, like the PorX/PorY system, contributes to P. gingivalis virulence in a mouse model.

feedforward regulatory loop to illustrate gene regulation in a manner dependent on the PorX/PorY system and s P , which also provides an example to elucidate the coordinate interaction between two-component systems and their regulated sigma factors in gene regulation of P. gingivalis. Additionally, our results demonstrate that both the PorX/PorY system and s P are virulence factors that govern transcription of the genetic loci required for P. gingivalis virulence.

RESULTS AND DISCUSSION
PorX/PorY system and sigma factor r P coordinately regulate transcription in P. gingivalis. The PorX/PorY two-component system and extracytoplasmic function sigma factor s P (encoded by the sigP gene) appeared to form a regulatory cascade for upregulation of the T9SS-encoding genes (i.e., the por genes) because PorX/PorY was shown to upregulate s P and, in turn, s P enhanced the transcription of the por genes (5). Particularly, it was observed that the PorX/PorY-stimulated s P bound the promoters of the por genes, including the porT gene which encodes a T9SS component (5). Besides s P , a recent result from our laboratory showed the PorX protein also bound to the porT promoter and actually interacted with two DNA regions (7). This result not only verified the DNA-binding ability of the PorX response regulator, but also provided the possibility that both PorX and s P should directly act on the por promoters. If PorX and s P must coordinately control but not build a regulatory cascade to regulate the por genes, we postulate that s P expressed in trans in the absence of PorX, or vice versa, should not stimulate por transcription. We examined this hypothesis by determining the transcription of the porT gene in a porX deletion mutant (DporX). As predicted, a s P protein that was expressed in trans from a plasmid (pT-COW-P sigP -sigP, referred to as p-sigP) did not exert any effect on the porT expression in the DporX mutant because the porT mRNA level in this mutant harboring p-sigP remained similar to that in the mutant harboring the parental plasmid pT-COW (8), both of which were ;6.4-fold lower than that in the wild-type strain (Fig. 1A). Likewise, PorX had no effect on porT expression in the absence of s P , because a PorX protein expressed in trans from a plasmid (pT-COW-P PGN_1016 -porX, referred to as p-porX) did not stimulate porT transcription in a sigP null (DsigP) mutant (Fig. 1A). In contrast, the alleviated porT transcription was fully reversed to a wild-type level in the DsigP mutant harboring p-sigP and also in the DporX mutant harboring p-porX (Fig. 1A), indicating the trans-expressed s P and PorX proteins were functionally active. We also determined whether this coordinate regulation was effective in controlling two other PorX/PorY-and s P -dependent genes, PGN_0341, which encodes a predicted T9SS component (4), and PGN_1639, which has been known as a s P -dependent gene (5) and recently identified as a PorX/PorY-dependent locus according to our transcriptomic and proteomic analyses (unpublished data). We confirmed that the transcription levels of PGN_0341 and PGN_1639 were upregulated by PorX and s P because their mRNA levels were significantly reduced in the DporX and DsigP mutants compared to those in the wild-type strain ( Fig. 1B and 1C). Comparable to the porT regulation (Fig. 1A), the alleviated transcription of PGN_0341 and PGN_1639 was not stimulated in DporX mutant harboring p-sigP or in DsigP mutant harboring p-porX ( Fig. 1B and 1C).
It has been shown that T9SS mediates secretion of gingipains, which are required for pigmentation of P. gingivalis on a blood plate (for review see reference 9), and consistently both DporX and DsigP mutants display a nonpigmented phenotype (6,10). We conducted a phenotypic analysis to evaluate the coordinate interaction between the PorX/PorY system and s P . While the DporX and DsigP mutants carrying pT-COW exhibited nonpigmented colonies on a brain heart infusion (BHI) blood plate, both the DporX mutant harboring p-porX and the DsigP mutant harboring p-sigP formed vigorous black-pigmented colonies (Fig. 1D). However, the DporX mutant harboring p-sigP and the DsigP mutant harboring p-porX exhibited a nonpigmented phenotype when they were grown on a BHI blood plate (Fig. 1D). Taken together, these genetic approaches suggest the PorX/PorY system and s P should govern transcription of the por genes via a coherent regulatory network rather than a direct regulatory cascade.
PorX response regulator directly binds to the promoter of the sigma factor gene sigP. Evidence suggests that transcription of the sigP gene is activated by the PorX/PorY system (5). This is confirmed by our result derived from a reverse transcription-PCR, since the sigP mRNA level in the DporX mutant (lane 2, Fig. 2A) was 4.3-fold lower than that in the wild-type strain (lane 1, Fig. 2A). Our result also confirmed the plasmid p-sigP should be able to express the sigP gene in trans because it fully restored , and the PGN_1639 gene (C) in the 33277 wild-type strain, the DporX mutant (YS19181), and the DsigP mutant (YS17717) carrying pT-COW, p-porX (pYS18679, pT-COW-P PGN_1016 -porX), or p-sigP (pYS19107, pT-COW-P sigP -sigP). The mRNA level in the wild-type strain was set to 1 for calculation. Results are representative of three independent experiments. *, P , 0.05; **, P , 0.01; versus wild type by t test. (D) The growth of wild-type 33277 strain with pT-COW (vector), DporX mutant (YS19181), and DsigP mutant (YS17717) carrying pT-COW, p-porX, or p-sigP, respectively, on a blood BHI plate containing tetracycline (0.5mg/ml). Results are representative of four independent experiments.  Fig. 2A). To determine whether the PorX/PorY system can directly upregulate the sigP gene, we first characterized the sigP promoter region and investigated the PorX binding to this promoter by conducting an electrophoretic mobility shift assay (EMSA) using a 275-bp DNA fragment (marked as T 1 ), including the 149-bp intergenic region of the sigP-PGN_0275 genes. We found that a PorX protein with a C-terminal His 6 tag (referred to as PorX-C -His 6 ) gel-shifted this DNA fragment in a concentration-dependent manner (Fig. 2B), thus suggesting this T 1 fragment should contain the sigP promoter and also the sequence(s) that binds the PorX protein (i.e., the PorX-binding site). Therefore, we conducted a DNase footprinting assay to map the PorX-binding site in T 1 and found that the PorX-C -His 6 protein bound to an AT-rich DNA sequence (59-tcgaaaaaaatgtttttctttgc-39) in a concentration-dependent manner (Fig. 2C). This PorX-binding site, which is located 297 to 275 nucleotides (nt) upstream of the start codon (underlined nucleotides, Fig. 2D), shared a partial sequence with the PorX-binding site II (59-gattcgcgcaaaaatacaatatcttt-39) in the porT promoter, recently characterized by our laboratory (7). We postulate that PorX can recognize a sequence (59-CG(A/C)AAAAA-N 5 -T(T/A)TCTTTGC-39) that is conserved in these two promoters. Interestingly, the 5 nucleotides located between the conserved segments in the PorX-binding sites of the sigP and porT promoters were complementary (nucleotides labeled with arrows in Fig. 2E). Therefore, these results and our recent data (7) not only verify that PorX directly regulates transcription of the sigP gene and the por genes such as porT, but also elucidate that PorX is a DNA-binding protein and capable of recognizing specific DNA sequences in a manner similar to many other TCS response regulators.
PorX protein activates sigP transcription in vitro. To further validate the direct role of the PorX/PorY system in sigP transcription, we conducted an in vitro transcription assay using a P. gingivalis RNA polymerase holoenzyme (referred to as pg-RNAPs D ) that was reconstructed from N-terminal His 6 -tagged subunit proteins, including a (PGN_1841), b (PGN_1571), b' (PGN_1570), and the major sigma factor s D (PGN_0638) (for details see the Materials and Methods section). When the T 1 fragment, which was tested for PorX binding ( Fig. 2B and C), was used as the template for the in vitro transcription reactions supplemented with 50 nM pg-RNAP-s D , two transcripts labeled as P 1 and P 2 , respectively, were produced (Fig. 3A). Both transcriptions were stimulated by the PorX-c-His 6 protein because the amount of P 1 and P 2 increased in a PorX concentration-dependent manner (lanes 1 to 4, Fig. 3A). These results suggest that sigP transcription is initiated from two regions that are located at 65 to 60 nt (labeled as p 1 ) and 99 to 94 nt (p 2 ) upstream of the start codon, respectively (illustrated in the T 1 sequence, Fig. 3B). To verify whether these transcripts were produced specifically, we compared the in vitro transcripts from the wild-type T 1 template and a mutated T 1 template (T 1-Sub ) which carried 17-nt substitutions at a 103-to 87-nt sequence located upstream of the start codon. Our results showed that levels of both P 1 and P 2 transcripts from a reaction using the T 1-Sub template were much lower than those using the T 1 template (lane 2 versus lane 1, Fig. 3C). Since this substituted sequence overlaps a partial region of the PorX-binding site for the P 1 transcription and the p 2 region (Fig. 3B), we reasoned that these substitutions must simultaneously interfere with transcription initiated from p 1 and p 2 in T 1-Sub . To further verify that the transcription initiation from p 1 and p 2 was specific, we used another template, i.e., T 2 , which was a longer template (291 bp) and contained an additional 16-bp sequence extending from downstream of the T 1 template (275-bp). The in vitro transcription using this T 2 template could still produce two transcripts, labeled as P 1 ' and P 2 ', in a PorX concentrationdependent manner (lanes 3 and 4, Fig. 3D), and both products were exactly 16-nt longer than  PorX/PorY and s P Coordinately Regulate Transcription P 1 and P 2 , respectively (lanes 3 and 4 versus lane 2, Fig. 3D). Therefore, the in vitro transcription of the sigP gene must be specifically initiated from two DNA regions, p 1 and p 2 , thus allowing the T 1 template to produce P 1 and P 2 transcripts and also the 16-bp longer T 2 template to produce 16-nt longer P1' and P 2 ' transcripts.
PorX stimulates in vitro transcription of the porT gene carried out by a reconstructed RNA polymerase-r P holoenzyme. Since PorX directly binds to the s Pdependent porT promoter (7), we postulated that it should be able to stimulate porT transcription in vitro. To examine this hypothesis, we conducted an in vitro transcription assay using a P. gingivalis RNA polymerase s P holoenzyme (referred to as pg-RNAP-s P ) which was reconstructed from purified N-terminal His 6 -tagged a, b, b' and C-terminal His 6 -tagged s P proteins (for details see the Materials and Methods section). When a 301-bp DNA fragment, including the porT promoter sequence, was used as the template, two transcripts labeled as S 1 and S 2 were produced by the reconstructed pg-RNAP-s P (at 50 nM) and both transcriptions were enhanced by PorX in a concentration-dependent manner (lanes 2 to 5, Fig. 4A). S 1 transcription was initiated from the adenosine (labeled as s 1 , Fig. 4B) located 29 nucleotides downstream of PorX binding site II, identified in our previous study (7). S 2 transcription was initiated from the guanosine (labeled as s 2 , Fig. 4B) located 49 nucleotides downstream of PorX binding site I in the porT promoter. Thus, we postulated that PorX should bind to site I and site II and enhance the transcription initiated at s 2 and s 1 , respectively. Synthesis of both S 1 and S 2 was significantly stimulated when the pg-RNAP-s P concentration was elevated from 25 nM to 50 nM (lanes 2 and 3, left panel, Fig. 4C). In contrast, the pg-RNAP-s D holoenzyme was not as efficient as pg-RNAP-s P because only S 2 could be produced to a detectable level by pg-RNAP-s D at a high concentration of 200 nM (lane 4, right panel, Fig. 4C). These observations suggest that s P should be the preferred sigma factor to mediate the porT transcription and that both PorX and s P act directly on its promoter. Interestingly, the s 1 and s 2 sites did not overlap the transcription initiation site (11) detected from a primer extension assay using a total wild-type mRNA sample (7). This is probably because other factors in the bacterial cell might interact with PorX and RNA polymerase-s P holoenzyme to initiate the porT transcription from the 11 position.
PorX/PorY system is essential for the virulence of P. gingivalis in a mouse model. According to our previous results (7), the PorX/PorY system is a virulence regulator of P. gingivalis because a virulent W83 wild-type strain, but not the DporX mutant, could cause infection in a mouse model described previously (11). To determine whether the PorX/PorY-activated s P contributes to bacterial virulence, we compared the pathogenesis of this wild-type strain and its isogenic DsigP mutant in this mouse model. Sixweek-old BALB/c mice were subcutaneously injected on the dorsal surface with the strains that were grown in BHI medium for 12 h, and all five mice that were challenged by W83 wild-type cells at a dose of 4.72 Â 10 10 CFU died in 48 h ( Fig. 5A and 5B). On the other hand, four out of the five mice challenged with the isogenic DsigP mutant cells at a dose of 4.58 Â 10 10 CFU survived the 30-day observation period (Fig. 5A and  5B), thus demonstrating that the sigma factor s P is a virulence determinant. The DsigP mutant was highly attenuated but not as avirulent as the DporX mutant, which, at a The DNA sequence of the sigP promoter region. Underlining corresponds to the PorX-protected region. Blue dashed frames correspond to the regions labeled as p 1 and p 2 , respectively, where transcription was initiated. The highlighted sequence corresponds to the wild-type sequence which was substituted by the sequence (Sub) in red capital letters. Numbering begins from the adenine nucleotide of the start codon (underlined capital letters). (C) In vitro transcription of the sigP templates (T 1 and T 1-sub ) with the wild-type sequence and a substituted sequence, respectively. Blue left braces indicate the transcripts, P 1 and P 2 , produced from the reaction with template T 1 . (D) In vitro transcription of the sigP templates (T 1 and T 2 ) containing the first 29 and 45 coding nucleotides, respectively. Blue right braces indicate the transcripts, P 1 and P 2 , produced from the reaction with template T 1 , and red right braces indicate the transcripts P 1 ' and P 2 ', produced from the reaction with template T 2 . Double arrows indicate that P 1 and P 2 are 16 nucleotides shorter than P 1 ' and P 2 ', respectively. Results in A, C, and D were repeated two times. dose of 4.32 Â 10 10 CFU, did not kill even one mouse in the 30-day observation period (Fig. 5A and 5B). The result of the DporX mutant also reconfirmed that the PorX/PorY system is essential for P. gingivalis virulence (7). Accordingly, we postulated that the PorX/PorY system should also be able to activate other P. gingivalis virulence factors whose regulation is independent of s P . Based on these observations, it is reasonable to assume that the PorX/PorY system renders P. gingivalis virulent in part by activating the sigP gene in this mouse model. This assumption should be further confirmed by our ongoing RNA sequencing analysis, which compares the expression of overall PorX/ PorY-and s P -regulated genes in P. gingivalis cells recovered from the animal against those grown in vitro.
In conclusion, pathogenic bacteria have developed many sophisticated mechanisms to control the expression of the genes that contribute to virulence. Growing evidence suggests that the PorX/PorY system in P. gingivalis plays an essential role in the regulation of numerous virulence determinants, exemplified by the set of por genes encoding the T9SS components. This study has revealed that the PorX/PorY system and sigma factor s P construct a regulatory pathway to coordinate the regulation of the por genes. We provide evidence that the PorX response regulator binds to the sigP promoter ( Fig. 2B and 2C) and activates the sigP transcription in an in vitro transcription reaction system using a reconstructed RNA polymerase holoenzyme (Fig. 3A, C, and D), thus demonstrating that the PorX/PorY system directly regulates transcription of the sigP gene.
When two related regulators build a regulatory cascade, the first regulator regulates the second regulator, and then the second regulator regulates their target genes. Therefore, in the absence of the first regulator, the target genes will still be regulated by the second regulator when this regulator can be produced in trans. Although PorX/ PorY activates s P , and then s P activates the por genes, this regulatory cascade model is inapplicable to the regulation dependent on the PorX/PorY system and s P because both the first regulator (PorX) and the second regulator (s P ) are shown to bind to the por promoters (5, 7), and s P produced in trans from p-sigP is unable to activate these genes in the DporX mutant ( Fig. 1A to C). We also show that the PorX protein can directly enhance in vitro porT transcription catalyzed by an RNA polymerase-s P holoenzyme ( Fig. 4A and 4C), which further confirms the direct action of PorX on the porT promoter. Therefore, regulation of the por genes should be controlled coordinately by the PorX/PorY system and s P , in which PorX stimulates the production of s P , and both PorX and s P regulate the porT transcription. We postulate that this mechanism of action of the PorX/PorY system and s P should fall under the criteria of a regulatory motif, which is known as the feedforward loop (12) (Fig. 6). Our previous results have shown that the PorX/PorY system responds to hemin and enhances transcription of the porT gene (7). In many cases, the feedforward loop has the capability to integrate multiple signaling molecules into a gene regulation (12). It remains to be investigated whether the feedforward loop contributing to the PorX/PorY-and s P -governed signal transduction pathway is able to respond to signal molecules besides hemin.
It is worth noting that the transcription of the sigP gene in the DporX mutant is not completely repressed (5) (Fig. 2A). According to a previous study (13), s P exerts an inhibitory effect on P. gingivalis biofilm formation, as biofilm formation is induced in a sigP null mutant in an enriched BHI medium. However, the DporX mutant grown in this BHI medium did not induce biofilm formation (unpublished result). We reason that the expression of the sigP gene remaining in the DporX mutant is sufficient to inhibit biofilm formation.
The PorX/PorY system has been shown as an essential regulator for P. gingivalis virulence since the DporX mutant is avirulent in mouse infection (7) (Fig. 5A and B). In this study, the murine virulence assay has demonstrated that s P contributes to P. gingivalis virulence and the DsigP mutant becomes attenuated. Further in vivo analysis will be needed to confirm the role of s P in the PorX/PorY-controlled mechanism required for P. gingivalis pathogenesis.

MATERIALS AND METHODS
Bacterial strains, plasmids, media, and growth conditions. Strains and plasmids used in this study are listed in Table 1. The P. gingivalis ATCC 33277 and W83 wild-type strains used in this study were obtained from Koji Nakayama (4). P. gingivalis cells were grown at 37°C in an anaerobic chamber (Model 2000, Coy Lab Products) that maintained 90% N 2 /5% CO 2 /5% H 2 in the atmosphere. Blood agar plates (5% sheep defibrinated blood, 1.5% agar) or brain heart infusion (BHI, purchased from BD) medium supplemented with hemin (5 mg/ml) were used to culture P. gingivalis strains. When necessary, erythromycin (0.5 mg/ml) or tetracycline (0.5 mg/ml) was supplemented. P. gingivalis cells were harvested by centrifuging liquid cultures at 10,000 Â g (;8,500 rpm) in a Sorvall ST 8R centrifuge with a HIGHConic III fixed angle rotor (maximum 9,500 rpm) at 4°C for 10 min. E. coli DH5a and BL21(DE3) strains were used for cloning and protein production, respectively. E. coli cells were routinely grown in Luria broth (LB) supplemented with antibiotics when necessary (kanamycin, 50 mg/ml; ampicillin, 50 mg/ml) at 37°C. To prepare cell lysates, bacterial cells were opened with a sonicator (Misonix Sonicator 3000).
Construction of plasmids and strains with chromosomal mutations. All plasmids used in this study are listed in Table 1. Polymerase chain reactions (PCR) were performed using a Bio-Rad T100 thermal cycler with Taq DNA polymerase (New England BioLabs [NEB]). Custom oligonucleotides were synthesized by Integrated DNA Technologies (IDT) and are listed in Table 2. PCR products were isolated using a QIAquick PCR purification kit (Qiagen). Restriction enzymes were purchased from New England BioLabs and used according to the manufacturer's instructions. Digested DNA fragments were separated in 0.8 to 1% agarose gels and then isolated using a QIAquick gel extraction kit (Qiagen). Plasmids were purified from overnight cultures of E. coli DH5a in LB at 37°C using plasmid minikit or midi kit (Qiagen). Plasmid pYS19107 for complementation assays was constructed using PCR fragments containing a 500bp sequence of the upstream region followed by the sigP coding region, which was amplified with primers 3809 and 2827, digested with HindIII and BamHI, and ligated between the HindIII and BamHI sites of pT-COW (14). Plasmid pYS17676 for mutagenizing the sigP gene in both 33277 and W83 strains was constructed using a DNA fragment containing the 8-to 305-nt sigP coding region amplified with primers 2768 and 2769, digested with PstI, and then ligated with PstI-digested pGEM-ermF plasmid. Plasmid pYS18051 was constructed using PCR fragments containing the rpoA (PGN_1841) coding region amplified with primers 3158 and 3159, digested with NcoI and BamHI, and then ligated between the NcoI and BamHI sites of plasmid pET28a. Plasmid pYS18943 was constructed using PCR fragments containing the rpoB (PGN_1571) coding region amplified with primers 3160 and 3161, digested with NheI, and then ligated between the NheI sites of plasmid pET11a. Plasmid pYS18165 was constructed using PCR fragments containing the rpoC (PGN_1570) coding region amplified with primers 3162 and 3163, digested with NcoI and XhoI, and then ligated between the NcoI and XhoI sites of plasmid pET28a. Plasmid pYS18052 was constructed using PCR fragments containing the rpoD (PGN_0638) coding region amplified with primers 3164 and 3165, digested with NcoI and HindIII, and then ligated between the NcoI and  agc ata ttc gcc aaa agg concentration of protein samples were determined using a Silver Staining kit (Pierce) and BCA Protein assay kit (Pierce) by following the instructions from the manufacturer.
DNase footprinting analysis. The DNase I footprinting assay was performed as described (15) with the following modifications. 32 P-labeled DNA (25 pmol, as was used for EMSA) was mixed with 0, 70, 140, or 280 pmol of the PorX-C -His 6 protein in a 100 ml reaction. DNase I digestion was carried out using 0.05 units DNase I (Invitrogen) per reaction. Samples were analyzed by 6% denaturing polyacrylamide electrophoresis by comparison with a DNA sequence ladder generated by Maxam and Gilbert A1G reaction, using the same 32 P-labeled PCR product. The positions of radioactive DNA fragments in the gels were detected by autoradiography.
Reconstitution of RNAP holoenzymes from isolated subunits. A procedure for reconstitution of E. coli RNAP holoenzyme developed and described in detail (16) was successfully used for reconstitution of other bacterial RNAPs. We used a modified procedure presented in a previous study (17) to carry out reconstitution of P. gingivalis RNAP holoenzymes with the following modifications. Briefly, prior to the in vitro reconstitution, RNAP subunits isolated from the procedure above were suspended in a denaturation buffer (6 M guanidine-HCl, 50 mM Tris-HCl [pH 7.9], 10 mM MgCl 2 , 10mM ZnCl 2 , 10% glycerol, 1 mM EDTA, and 10 mM DTT). The mixtures were left for 30 min on ice and then spun in a 4°C microcentrifuge at 10,000 Â g for 30 min. The supernatants were transferred into fresh tubes and the protein concentration was determined using the BCA protein assay kit (Pierce) with bovine serum albumin (BSA) as a standard. RNAP subunits were mixed in a molar ratio of 2:8:4 (a:b:b9) and dialyzed against 250-volume reconstitution buffer (50 mM Tris-HCl [pH 7.9], 200 mM KCl, 10 mM MgCl 2 , 10mM ZnCl 2 , 10% glycerol, 1 mM EDTA, and 10 mM 2-mecaptoethonal) at 4°C for 16 h with two changes. One molar equivalent of isolated RNAP s subunit (s D or s P ) in PBS was added to the supernatant and the mixture was incubated at 30°C for 1 h. The resulting RNAP preparations were used directly in transcription assays or stored under (NH 4 ) 2 SO 4 (65% saturation) until further use.
Transcription of sigP and porT in vitro. The in vitro transcription was conducted in a 50-ml reaction mixture containing 1Â in vitro transcription buffer (80 mM HEPES-KOH [pH 7.5], 24 mM MgCl 2, 2 mM spermidine, 40 mM DTT with 500 mM ATP, CTP, GTP, and UTP, respectively) and 1 mg of linear doublestranded DNA (dsDNA) template with the desired amounts of PorX-c-His 6 protein and an RNA polymerase holoenzyme. Reaction mixtures were incubated for 2 h at 37°C and transcripts were precipitated using three volumes of cold 100% ethanol and 1/10 volume of 3 M sodium acetate (pH 5.8) and resuspended with RNase-free water. For sigP transcription in vitro, the template T 1 was amplified from 33277 chromosomal DNA using primers 3043 and 3044, while T 1-Sub with substituted PorX binding sequence (from gttttgtcgaaaaaaat to caggcgctgggatccgc) was prepared with primers 3043 and 3044 and 4105 and 4106 by using an overlap extension PCR (18). The longer template T 2 was amplified from 33277 chromosomal DNA using primers 3044 and 4111. For porT transcription in vitro, the template was amplified from 33277 chromosomal DNA using primers 4025 and 4026. Transcripts in vitro were monitored after being converted into cDNAs through a primer extension performed as described (19) with the following modifications. RNA pellets derived from templates T 1 and T 1-Sub were reverse transcribed using 2 ml of 32 P-labeled primer 3043 in a 20-ml mixture containing 25 units of M-MuLV reverse transcriptase (NEB) at 42°C for 2 h. 32 P-labeled primer 4111 was used for primer extension of the transcripts derived from template T 2 . Transcripts derived from the porT template were reverse transcribed by using 32 P-labeled primer 4025. The cDNA samples were precipitated with 2.5 volumes of ethanol and 0.3 M sodium acetate (pH 5.8) and resuspended in 5 ml of Gel Loading Buffer II (Thermo Fisher), and then analyzed by a 6% denaturing polyacrylamide gel. DNA ladders were amplified from 33277 chromosomal DNA using three primer pairs ( 32 P-labeled 3043 and 3044; 32 P-labeled-4111 and 3044; and 32 P-labeled-4025 and 4026) for the products from T 1 , T 2 , and the porT template, respectively, and generated by Maxam-Gilbert reaction.
Virulence assay in a mouse model. All animal experiments conform to our animal protocols (18-1655R) approved by the Institutional Animal Care and Use Committee (IACUC), Office of Research Integrity and Assurance, Arizona State University (ASU protocol number 18-1655R). Groups of 6-weekold female BALB/c mice (purchased from Charles River Laboratories) were randomly allocated into different groups. Determination of virulence of the P. gingivalis W83 and mutant strains was performed using mouse subcutaneous infection experiments, as described previously (20), with slight modifications. Briefly, bacterial cells were grown in enriched BHI broth at 37°C for 12 h. The culture was diluted 20-fold in 100 ml of fresh BHI medium and grown for the time periods indicated. The cells were harvested by centrifugation at 10,000 Â g for 20 min and washed once with PBS, then adjusted to a concentration of approximately 5 Â 10 11 CFU/ml in PBS. Resulting bacterial cultures were serially diluted and plated for bacterial CFU to determine the exact titer of all strains used for infections. Mice were challenged with subcutaneous injections of 0.1 ml at each of the two sites on the depilated dorsal surface (0.2 ml per mouse). Infected mice were examined daily for survival.