E. corrodens 1073 and 1080 were provided by S. S. Socransky (Forsyth Dental Center, Boston, MA, USA) and E. corrodens 23834 was obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). E. corrodens 612-L, 257-4, and 261-2 were isolated clinically from human supragingival plaque [4] . E. corrodens L8Ao3, L9B6, and RV2 were kindly provided by Dr. Giuseppe Valenza (University of Würzburg, Würzburg, Germany). Escherichia coli XL-1 Blue was used for cloning and sequencing. The E. coli/E. corrodens shuttle vector pLES2 [14] was obtained from the ATCC. E. corrodens cells were grown in tryptic soy broth (TSB) containing 2 mg/mL KNO₃ and 5 µg/mL hemin, or on sheep blood agar plates at 37˚C. Bacteria harboring plasmids were cultured on a medium supplemented with 50 µg/mL carbenicillin. Bacterial growth was monitored spectrophotometrically by measuring optical density at 600 nm (OD₆₀₀).

E. corrodens was electrotransformed using a Gene Pulser electroporator (Bio- Rad, Hercules, CA, USA) with 5 to 10 µg plasmid DNA. Briefly, a 100-mL bacterial culture was grown for 12 h, washed three times in solution A (272 mM sucrose, 1 mM MgCl₂ pH 7.4), and resuspended in 100 µL solution A. A 39-µL aliquot of the bacterial suspension was mixed with 1 µL DNA and electroporated at 2.1 kV, 25 µF, 200 Ω. For transformations involving the broad-host-range shuttle vector pLS88 and its derivatives, cells were recovered by being spun for 12 h at 37˚C. After recovery, recombinant cells were cultured on sheep blood agar plates containing 50 µg/mL carbenicillin at 37˚C.

We designed primers A (GGGAAGAAAAGGGAAGTGCT) and B (TCTTCAGGTACC GTCAGCAAAA) based on the 16S rDNA sequence of E. corrodens (GenBank accession no. AF320620), and primers C (TTTTATCCGCAATGGGTATC) and D (TACAAATCT TTGCCCTTCAC) based on the type IV pilin gene sequence of E. corrodens 23834 (GenBank accession no. Z12609). Primers A and B allow the detection of all E. corrodens strains; whereas primers C and D are specific for those strains, in which genomic recombination occurred at the type IV pilin locus. To determine the occurrence of genomic recombination, we used these primers in combination with real-time PCR.

Real-time PCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems, Waltham, MA, USA) as follows: 23 µL master mix was added to 96-well PCR plates containing genomic DNA and primers. The plates were sealed with a clear plastic sheet and placed in an ABI 7300 Sequence Detector (Applied Biosystems). During the course of 40 cycles (94˚C, 15 s and 60˚C, 1 min), data were collected through optical cables connected to each well. Ct values were calculated by the ABI 7300 software and genomic recombination was estimated from the following formula: (Ct value by primers C and D)/(Ct value by primers A and B) = Detection rate of genomic recombination.

The HA activity assay was performed as previously described [4] . Erythrocytes that were obtained by the centrifugation of preserved rabbit blood were washed three times with saline before being suspended in phosphate-buffered saline (PBS, pH 7.2) at a concentration of 2%. The HA assay was performed in microtiter plates (Vdispo; Nalge Nunc International, Roskilde, Denmark). Test preparations (50 µL) were serially diluted two fold in PBS and mixed for 2 min with equal volumes of the 2% erythrocyte suspension. HA activity was examined after 1 h, and HA titers were expressed as the maximum dilution of the test preparation that exhibited HA activity.

E. corrodens strains formed a macroscopically visible biofilm that was firmly attached to the wells of 96-well tissue culture plates (non-treated polystyrene, flat- bottom with lid; BD Bioscience, San Jose, CA, USA), and biofilm production was determined as described previously [15] . The assay measured the primary attachment and accumulation of multilayered cell clusters, and subsequent biofilm production on the polystyrene surface. Briefly, after growth in TSB for 36 h at 37˚C, plates were gently washed four times with PBS, and adherent bacterial cells were fixed with methanol followed by staining with crystal violet. The optical density of the stained adherent bacterial biofilms was measured at 595 nm (OD₅₉₅) with a spectrophotometer.

E. corrodens strains were grown aerobically in TSB medium containing hemin and KNO₃ at 37˚C. Bacterial growth was monitored spectrophotometrically by measuring optical density at 600 nm (OD₆₀₀). Assays were performed at least three times. Mean values and standard deviations are reported.

In the detection of genomic recombination, hemagglutination assay, and growth rate measurement, the results are presented as mean values and standard deviations from triplicate measurements. In biofilm assay, the results are presented as mean values and standard deviations from at least 8 wells. The significance of intergroup differences was analyzed using Student’s t-test (unpaired t-test).

Previously, we reported that genomic recombination increased E. corrodens 23,834 pathogenicity [10] [11] [12] [13] . To examine the effect of genomic recombination in other strains, we introduced the recombinase-encoding ORF4 gene into seven clinical isolates, 261-2, 1080, 257-4, 612-L, L9B6, L8Ao3, and RV2. We checked the colony morphology of each transformant on solid agar plates. Whereas in the absence of ORF4, the strains formed corroding colonies (Figure 1(a)), colony morphology of strains 1080, L9B6, L8Ao3, and RV2 changed to non-corroding after introduction of ORF4 (Figure 1(b)). This change was not observed for strains 261-2, 612-L, and 257-4 (Figure 1(b)). Accordingly, strains were classified into two groups: A (strains 1080, L9B6, L8Ao3, and RV2), and B (strains 261-2, 612-L, and 257-4).

Next, we assessed hemolysis of E. corrodens strains on sheep blood agar media. Hemolysis can be observed as a transparent zone surrounding the colony. Although no hemolysis was observed in the absence of ORF4 (Figure 1(a)), strains 1080, L9B6, L8Ao3, and RV2 presented hemolytic activity after introduction of ORF4 (Figure 1(b)). No hemolytic activity was observed in strains 261-2, 612-L, and 257-4 after introduction of ORF4 (Figure 1(b)). These results were consistent with those relating to colony morphology, as hemolysis was detected

in all strains whose colony morphology was altered by introduction of ORF4.

In a healthy human body, the concentration of free iron should be maintained at 10⁻¹⁸ M [16] . However, bacteria require an iron concentration of 0.05 - 0.5 µM for growth [17] . To this end, bacteria can satisfy their metabolic needs and acquire iron by applying hemolytic factors that lyse host erythrocytes and cause the release of intracellular iron [18] . Therefore, hemolysis is thought to be important for the pathogenicity of many in vading bacteria and increasing hemolytic activity through introduction of ORF4 might be vital for in vivo survival of oral bacteria.

We showed previously that introduction of ORF4 into strain 23834 resulted in genomic recombination at the type IV pilin locus. Here, we investigated whether the same occurred in other E. corrodens strains. In strains 261-2, 612-L, and 257-4, we could not detect any increased genomic recombination compared to the non-transformed strains (Figure 2). Instead, introduction of ORF4 resulted in enhanced genomic recombination in strains 1080, L9B6, L8Ao3, and RV2 (Figure 2). These results were consistent with those pertaining to colony morphology and hemolysis. Accordingly, group A strains showed both colony morphology changes and genomic recombination at the type IV pilin locus, whereas group B strains displayed none of the above.

Given that the pili on the cell surface are involved in the cells’ twitching motility as well as colony morphology [19] [20] , it is likely that the change from corroding to non-corroding colonies reflected a loss of pili following genomic recombination at the type IV pilin locus.

It is believed that the GalNAc-specific lectin contributes to the pathogenicity and virulence of E. corrodens. To assess the effect of introducing ORF4 on pathogenicity, we measured HA activity in seven E. corrodens strains. As shown in Figure 3, HA activity was high in group A strains following introduction of ORF4, but did not change in group B strains even after introduction of ORF4.

We reported previously that genomic recombination by plasmid-mediated recombinase stimulated simultaneous GalNAc-dependent lectin activity and hemolytic activity in E. corrodens 23834 [11] [12] . Recently, we demonstrated that hemolytic activity decreased upon addition of GalNAc [21] . Here, we show that the introduction of ORF4 enhanced both hemolytic activity and GalNAc- dependent lectin activity (Figure 1 and Figure 3). These findings suggest that hemolytic activity correlates with lectin activity. Moreover, we recently isolated the hemolytic factor from E. corrodens 1073 and demonstrated that in its absence lectin activity was the same in this as in the wild-type strain [21] . It has been suggested that hemolysin and lectin are not the same protein, because absence of hemolysin does not have any effect on lectin. It is thought that once lectin mediates adhesion to the blood cell, the latter becomes susceptible to attack by hemolysins present on the bacterial surface. Therefore, enhancement of hemolytic activity following introduction of ORF4 may depend on increased lectin activity.

Previously, we reported that introduction of ORF4 into strain 23834 resulted in increased biofilm formation [11] . Here, we investigated whether the same occurred in other E. corrodens strains. As shown in Figure 4, biofilm formation increased in group A strains following introduction of ORF4 on pMU4, but not in group B strains.

We suggested earlier that the GalNAc-specific lectin and other factors contributed additively to biofilm formation in some strains of E. corrodens [11] . In this study, we demonstrate that the introduction of ORF4 enhanced both GalNAc-specific lectin activity and biofilm formation. Accordingly, the GalNAc-specific lectin might be involved in biofilm formation by E. corrodens.

Given that in our previous study, the introduction of ORF4 stimulated growth rate [11] , we investigated whether the same occurred in other E. corrodens strains. As shown in Figure 5, growth rate in group A strains increased after in-

troduction of ORF4 on pMU4, whereas that in group B strains did not change. Biofilm formation is thought to be a means by which pathogenic bacteria fix to a solid surface and enhance their pathogenicity [22] . In this light, increased growth and biofilm formation following transformation with pMU4 might represent one of the strategies by which E. corrodens can survive in the oral cavity and, at the same time, promote pathogenicity and virulence.

In the present study, colony morphological change, genomic recombination at the pilin gene locus, and enhanced GalNAc-specific lectin activity, hemolytic activity, biofilm formation, and growth rate were observed simultaneously following introduction of ORF4 into E. corrodens in four of the seven clinical isolates tested. These results suggest that the changes recorded in the four strains might be attributed to the same mechanism. Given that ORF4 is homologous to the gene encoding, a recombinase specific for the type IV pilin gene, genomic recombination at the pilin locus is likely caused by introduction of ORF4. Moreover, changes to colony morphology may be associated to the loss of pili derived from genomic recombination. However, it is presently unknown why enhancement of biofilm formation, lectin activity, hemolytic activity, and growth rate was observed following introduction of ORF4. One possibility is that the introduction of ORF4 may cause genomic recombination at multiple loci. Furthermore, as we did not observe any changes in three of the strains, it may be that the target sequences for genomic recombination is not located in their genomes. It remains to be seen how and where ORF4-mediated genomic recombination occurs, and how it promotes pathogenicity and virulence of E. corrodens.