BipA Is Associated with Preventing Autoagglutination and Promoting Biofilm Formation in Bordetella holmesii

Bordetella holmesii causes both invasive and respiratory diseases in humans. Although the number of cases of pertussis-like respiratory illnesses due to B. holmesii infection has increased in the last decade worldwide, little is known about the virulence factors of the organism. Here, we analyzed a B. holmesii isolate that forms large aggregates and precipitates in suspension, and subsequently demonstrated that the autoagglutinating isolate is deficient in Bordetella intermediate protein A (BipA) and that this deletion is caused by a frame-shift mutation in the bipA gene. A BipA-deficient mutant generated by homologous recombination also exhibited the autoagglutination phenotype. Moreover, the BipA mutant adhered poorly to an abiotic surface and failed to form biofilms, as did two other B. holmesii autoagglutinating strains, ATCC 51541 and ATCC 700053, which exhibit transcriptional down-regulation of bipA gene expression, indicating that autoagglutination indirectly inhibits biofilm formation. In a mouse intranasal infection model, the BipA mutant showed significantly lower levels of initial lung colonization than did the parental strain (P < 0.01), suggesting that BipA might be a critical virulence factor in B. holmesii respiratory infection. Together, our findings suggest that BipA production plays an essential role in preventing autoagglutination and indirectly promoting biofilm formation by B. holmesii.


Introduction
Bordetella holmesii is a Gram-negative coccobacillus that was first reported in 1995 [1]. The organism is associated with a variety of invasive infections, including bacteremia, septicemia, endocarditis, pneumonia, and septic arthritis, in patients with underlying medical conditions, such as asplenia and sickle cell anemia, and has been isolated from such patients' blood, pericardial effusion, and synovial fluids [1][2][3][4][5]. B. holmesii has been detected in respiratory specimens, such as nasopharynx and sputum samples from patients with pertussis-like symptoms [6,7], and therefore may also be responsible for causing a disease similar to pertussis (whooping cough) in otherwise healthy individuals. Indeed, B. holmesii has been isolated in patients Bacterial suspensions were incubated for 4 h at 36°C in disposable cuvettes under static conditions, and the OD 650 of each suspension was measured every 30 min. In addition, the cell morphology of B. holmesii was visualized after 4 h incubation using an Olympus IX83 microscope (Olympus, Tokyo, Japan).

Biofilm assay
Biofilm formation was assessed by scanning-electron microscopy (SEM) as described previously [25], with minor modifications. Briefly, B. holmesii strains were cultured on BG agar plates and then suspended in modified Stainer-Scholte (mSS) broth [26] to an OD 650 of 0.2. Bacterial suspensions (2 ml) were then cultured with shaking at 36°C for 24 h and diluted to an OD 650 of 0.2 in mSS broth. The resulting suspensions (2 ml) were incubated at 36°C under static conditions on vertically submerged glass slides (6 × 8 mm) in 6-well plates. After culturing for 5 or 72 h, the bacteria were washed 3 times with phosphate-buffered saline (PBS), fixed with 2.5% glutaraldehyde and 2.5% paraformaldehyde, and then post-fixed with 1% osmium tetroxide. The fixed cells were dehydrated through a graded series of ethanol and freeze-dried. Subsequently, the adherence of cells to glass slides and biofilm growth at the air-liquid interface were visualized using a SEM SU6600 (Hitachi, Tokyo, Japan). In addition, the levels of biofilm formation were assessed by crystal violet assay analysis [27]. A 1 ml suspension of each B. holmesii strain (OD 650 = 0.2) was incubated at 36°C under static conditions in a 5 ml polystyrene tube. After culturing for 24, 48, and 72 h, the biofilms were stained by treatment with 1.5 ml of a 0.1% crystal violet solution at room temperature for 30 min, and washed with distilled water. The remaining crystal violet was solubilized with 1.5 ml of 95% ethanol and quantified by measuring the OD 540 using a Multiskan FC Microplate Reader (Thermo Fisher Scientific, Inc., Waltham, MA, USA).

DNA sequencing
The bipA gene of B. holmesii F627 (complement 1618935..1623335, GenBank: AOEW01000001.1) was amplified from B. holmesii strains and sequenced using appropriate primers (S1 Table). The gene was named as the "outer membrane ligand binding protein" in the GenBank database; however, it was selected as the bipA gene based upon sequence similarity to other Bordetella bipA genes (see the Results section for more details). Sequencing reactions were carried out with a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Waltham, MA, USA), and the products were sequenced using a 3130xl Genetic Analyzer (Applied Biosystems) or a 3730 DNA Analyzer (Applied Biosystems).

Generation of BipA mutant
BipA-deficient mutant was constructed by double-crossover homologous recombination, as described previously but with minor modifications [28]. First, the ΔbipA sequence containing a 636-bp deletion was constructed by overlap extension PCR using B. holmesii BH2 genomic DNA as the template (S1A Fig). Briefly, two DNA fragments, which were 1.4 kbp and 1.2 kbp, were amplified by PCR using the attB1-bipA and MP1-bipA, and MP2-bipA and attB2-bipA primers, respectively (S1 Table). The DNA fragments were joined by overlap extension PCR using the attB1-bipA and attB2-bipA primers, and a 3rd round of PCR was then performed using the attB1-adaptor and attB2-adaptor primers, with the 2nd PCR product being used as the template. The resulting PCR product was cloned into pDONR221 to obtain pDONR221-ΔbipA using the adaptor PCR method and the Gateway cloning system (Invitrogen, Waltham, MA, USA). Vectors pDONR221-ΔbipA and pABB-CRS2 [29] were then combined using the Gateway cloning system to obtain pABB-ΔbipA, which was introduced into E. coli SM10λpir and trans-conjugated into Sm-resistant B. holmesii BH2 (BH2Sm r ). The resulting mutant was designated BH2Sm r -ΔBipA. The lack of BipA protein expression in the strain was confirmed by immunoblot analysis (S1B Fig). The BipA mutant had an in vitro growth rate that was similar to that of the parental strain BH2Sm r .

Intranasal infection of mice with BipA mutant
To examine the role of BipA during respiratory infection, we employed a mouse intranasal infection model using 4-week-old female BALB/c mice (Japan SLC) [30]. Briefly, strain BH2Sm r -ΔBipA was cultured on BG agar containing SM and suspended in casamino acid solution at an OD 650 of 1.0. The bacterial suspension was diluted 15-fold with casamino acid solution, and 50 μl of the resulting solution (10 7.9 colony forming unit (CFU)/ml) was instilled intranasally into mice (n = 3 mice per time point) anesthetized by intraperitoneal injection of pentobarbital. After 3, 6, or 24 h, individual lungs of the mice were removed and homogenized in 10 ml of PBS, using an ULTRA-TURRAX tube dispenser and DT-20M sterile tubes (IKA, Staufen, Germany). After sample dilution (10 0 -to 10 4 -fold) in PBS, each homogenate (100 μl) was spread on BG agar containing SM and incubated at 36°C. The number of viable bacteria was calculated using the number of CFU logarithmically transformed. The parental strain BH2Sm r was used as a control, exhibiting a viable cell number of 10 7.6 CFU/ml. This study was approved by the Animal Research Committee of Kitasato University, and experiments were conducted according to the guidelines of the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Quantitative reverse transcriptase-PCR (qRT-PCR)
B. holmesii strains were cultured on BG agar. Total RNA was isolated using the RNeasy Mini Kit (Qiagen) and treated with RNase-Free DNase (Qiagen) to degrade contaminating DNA. Total RNA (0.1 μg) was reverse-transcribed into cDNA using the PrimeScript RT Master Mix (TaKaRa Bio, Inc.) with random hexamers. The relative levels of bipA and recA transcripts were determined using SYBR Premix Ex Taq II (TaKaRa Bio) with the ABI PRISM 7500 Fast Real-Time PCR System (Applied Biosystems). The qRT-PCR conditions used were 10 s at 95°C, followed by 40 cycles of 95°C for 3 s and 60°C for 30 s. Primer sets (qbipA-F/qbipA-R and qrecA-F/qrecA-R) were used for bipA and recA amplification, respectively (S1 Table). Relative bipA transcript levels were calculated using the ΔΔCt method and were normalized to those of recA, which was used as an internal control for each sample. Data are presented as fold-changes in expression compared to those observed in BH2 bacteria cultured on normal BG agar.

Statistical analysis
Data are presented as means ± standard deviations. Student's t-test was used for the statistical evaluation of data. p < 0.05 was considered statistically significant.

Biofilm formation by B. holmesii
Cell adherence to an abiotic surface and biofilm formation by B. holmesii isolates were assessed by SEM. The non-autoagglutinating isolate BH2 was observed as scattered cells across the glass slides after culturing for 5 h. At a 72 h culture, the BH2 isolate formed an unusual network of thread-like structures, which radiated from the bacterial surface, indicating that BH2 formed a biofilm in vitro (Fig 2). In contrast, the autoagglutinating isolate BH7 adhered to glass slides after culturing for 5 h, but its level of cell adherence was apparently lower than that of BH2. Moreover, the adherent BH7 isolate did not form biofilms on glass slides during the 72 h cultivation. These results suggested that BH7 failed to form biofilms, due to its low adherence to the abiotic surface.

Identification of BipA associated with bacterial autoagglutination and a defect in biofilm formation
To identify the protein(s) associated with bacterial autoagglutination, the total protein expression profile of B. holmesii BH7 was analyzed by SDS-PAGE and compared with that of BH2, BH6, and BH8 (non-autoagglutinating isolates). There were 2 protein bands (approximately 150-kDa) that were present in the BH2, BH6, and BH8 samples but absent from the BH7 lysate ( Fig 3A). Subsequent nano-LC-MS/MS and Mascot search analyses of the bands present in the BH2 sample identified the two proteins as an outer membrane ligand binding protein (Gen-Bank: EMD74886.1) of B. holmesii F627. Furthermore, the amino acid sequence of the protein was 57−59% identical to that of BipA of B. bronchiseptica and B. pertussis, and the protein contained a conserved intimin domain and seven 90-amino acid repeats similar to those of other Bordetella BipA proteins [21]. We therefore designated the outer membrane ligand binding protein as BipA. As shown in Fig 3B, immunoblot analysis using anti-BipA antisera detected two protein bands (corresponding to BipA) in isolates BH2, BH6, and BH8, but not BH7. Instead, a protein band that was approximately 70-kDa was detected in BH7. Sequencing analysis revealed that the bipA gene of isolate BH7 contained a guanine (G) deletion at nucleotide position 2039 (c.2039delG), resulting in the generation of a premature stop codon at nucleotide position 2066 (amino acid 689, p.Ala681ProfsX9) (Fig 3C). As such, our immunoblot analysis might have detected the truncated form BipA in isolate BH7.

Autoagglutination of constructed BipA mutant
As shown in Fig 4A, the BipA-mutant strain BH2Sm r -ΔBipA, constructed by homologous recombination, formed aggregates and precipitates in suspension under static conditions. Specifically, the turbidity of the bacterial suspension significantly decreased from 1.08 to 0.58 (OD 650 ) after a 4 h of incubation. Meanwhile, the parental strain, BH2Sm r , did not form aggregates or precipitates during the 4 h incubation period (OD 650 = 1.00 ± 0.07; Fig 4B). Microscopic observations subsequently detected high levels of autoagglutination by the BH2Sm r -ΔBipA cells (Fig 4C). Notably, the bacterial aggregates were easily broken into single cells by repeat pipetting.

Biofilm formation by constructed BipA mutant
We cultured BH2Sm r -ΔBipA in polystyrene tubes under static conditions and examined the levels of cell adherence to the polystyrene surface. As shown in Fig 5A, BH2Sm r -ΔBipA was defective in its ability to form biofilms. Indeed, the biofilm biomass of this strain remained near the background level of staining after culturing for 72 h (Fig 5B). In contrast, the biofilm biomass of BH2Sm r significantly increased over time, reaching 0.86 (OD 540 ) after 72 h of cultivation. Subsequent SEM observations confirmed that BH2Sm r formed biofilms after culturing for 72 h, while BH2Sm r -ΔBipA adhered poorly to glass slides and failed to form biofilms during the 72 h cultivation period ( Fig 5C). These data suggested that the autoagglutination phenotype of BH2Sm r -ΔBipA inhibited the strain's ability to adhere to the abiotic surface and, consequently failed to form biofilms in vitro.

Initial colonization of constructed BipA mutant in murine lungs
Mice were infected with BH2Sm r -ΔBipA or BH2Sm r , and after 3, 6, and 24 h the numbers of viable bacteria recovered from the lungs of the infected animals (n = 3) were compared. As shown in Fig 6, the average levels of colonization by BH2Sm r -ΔBipA were significantly lower than those of BH2Sm r (log 10 CFU, P < 0.01) at 3 and 6 h post-infection: 10 5.0 CFU versus 10 6.2 CFU at 3 h, and 10 4.6 CFU versus 10 5.8 CFU at 6 h, respectively. In contrast, both lung CFUs of BH2Sm r -ΔBipA and BH2Sm r drastically decreased at 24 h post-infection, and the number of CFUs was near or below the detection limit of 100 CFU (not applicable to the Student's t-test). The average recovery rates of BH2Sm r -ΔBipA were 2.8% and 1.1% at 3 and 6 h post-infection, respectively, while those of BH2Sm r were 77% and 35%, respectively. Significant differences were observed in the initial colonization levels at 3 and 6 h post-infection between the BipA mutant and its parental strain.

Relationship between BipA production and biofilm formation in other B. holmesii strains
We examined the levels of BipA production in other B. holmesii strains by immunoblot analysis. Interestingly, BipA production was detected in each of 5 isolates (BH1, BH3, BH4, BH5, and BH9); however, the protein was not detected in strains ATCC 51541 or ATCC 700053, which were isolated from blood infections. Likewise, the two ATCC strains were the only isolates that failed to exhibit in vitro biofilm formation. While sequence analyses revealed that these strains encoded an intact bipA gene, qRT-PCR detected markedly lower levels of bipA gene expression (0.07-to 0.09-fold) in the ATCC strains than in BH2 (Fig 7). Interestingly, sequence analysis identified some mutations in bipA upstream region and bvgA gene of the ATCC 700053 strain (S2 Fig). Although the ATCC 51541 strain also was found to have an insertion mutation in the bvgA gene [31], the relation between these mutations and BipA downregulation is presently unknown. Taken together, our findings indicate that BipA production is downregulated at the transcriptional level in ATCC 51541 and ATCC 700053. The characteristics of clinical and ATCC strains are summarized in Table 1.

Discussion
In the present study, we analyzed a B. holmesii isolate exhibiting autoagglutination and found that this phenotype was the result of a defect in BipA production. Indeed, a BipA-deficient mutant generated by homologous recombination also exhibited the autoagglutination phenotype. Notably, the BipA mutant, as well as other autoagglutinating B. holmesii isolates, did not show biofilm formation. Our findings therefore suggest that BipA production plays essential roles in preventing bacterial autoagglutination and indirectly promoting biofilm formation in B. holmesii. To the best of our knowledge, this is the first report to demonstrate a function for the B. holmesii BipA protein.
Biofilms are microbial communities in which bacteria adhere to each other and to biotic or abiotic surfaces. Biofilm development occurs through multiple sequential processes, including adherence to a surface, microcolony formation, and establishment of a 3-dimensional structure. Three classical Bordetella species, B. pertussis, B. parapertussis, and B. bronchiseptica, are capable of forming biofilms on abiotic surfaces and within the mouse respiratory tract [25,27,[32][33][34]; however, the biofilm-forming capacity of B. holmesii was previously unreported. The present study clearly demonstrates that BipA-producing B. holmesii isolates formed biofilms in vitro. Conversely, strains that were deficient in BipA production adhered poorly to an abiotic surface consequently failed to form biofilms, likely due to the high levels of bacterial autoagglutination associated with these strains. These findings therefore suggest that BipA may act as an autoagglutination inhibitor and contributes indirectly to biofilm formation in B. holmesii. In contrast, a previous study showed that while B. bronchiseptica exhibited autoagglutination under Bvg i phase, this phenotype was not BipA-dependent [21], indicating that there are functional differences between the BipA proteins of B. bronchiseptica and B. holmesii. BipA localizes to the outer membrane of the bacterial cell, and the C-terminus of the protein is exposed to the environment [21]. The C-terminal domain (278 amino acids) of B. bronchiseptica BipA is nearly identical to those of BipA proteins produced by B. pertussis and B. parapertussis (>88% identity). Conversely, the Cterminal domain (294 amino acids) of B. holmesii BipA exhibits much lower sequence identity (42-43%) to those of the classical Bordetella species. Thus, this distinct C-terminal domain may play a role in the anti-autoagglutination effect of BipA in B. holmesii, perhaps by inhibiting cellto-cell binding. While this is a potential explanation for B. holmesii autoagglutination, we have not excluded other possible explanations. For instance, the absence of BipA may allow the expression of an unknown factor that causes autoagglutination, and the absence of BipA may affect the surface charge or hydrophobicity, leading to the autoagglutination. Further studies are required to elucidate the molecular mechanism governing B. holmesii autoagglutination.
In our autoagglutination assay, the constructed BipA mutant BH2Sm r -ΔBipA showed slower autoagglutination kinetics than that of the naturally occurring BipA-deficient strain BH7 (Figs 1 and 4). Our immunoblot analysis revealed that BH2Sm r -ΔBipA does not produce any detectable truncated BipA, which allows for a partially functional BipA (S1B Fig). These findings suggest that the strength of autoagglutination may depend on the strain used. In addition, we observed that the BipA + back-mutant (BH7Sm r -BipA + ) showed slower autoagglutination kinetics than that of its parental strain BH7Sm r (S3 Fig). However, the restoration of agglutinBipA production failed to completely abrogate the autoagglutination. This observation may have been due to the acquisition of Sm resistance, as the autoagglutination kinetics of BH7Sm r was faster than that of the parental isolate BH7. These results support our hypothesis that BipA is associated with preventing autoagglutination in B. holmesii. were cultured for 4 days on BG agar. Total protein (2 μg) was subjected to immunoblot analysis with anti-BipA antisera. (B) After culturing for 3 days on BG agar, total RNA was isolated, reverse transcribed into cDNA, and then analyzed by qRT-PCR analysis. Relative bipA transcript levels were calculated using the ΔΔCt method and were normalized to those of recA. The recA transcript was used as an internal control for each sample. Data are presented as fold-changes in expression compared to those observed in BH2. The means ± standard deviations of results obtained from 3 separate experiments are shown. In a previous study, a BipA homolog was identified and designated as Bordetella colonization factor A (BcfA), which contains an intimin domain (similar to the BipA protein) in its Nterminal region, in B. bronchiseptica [22]. In this study, we identified an outer membrane ligand binding protein that was absent from an autoagglutinating isolate of B. holmesii ( Fig  3A). The amino acid sequence of the protein was 57% and 44% identical to that of B. bronchiseptica BipA (GenBank: CAE32799.1) and BcfA (GenBank: CAE30611.1), respectively. In addition, the outer membrane ligand binding protein contained not only an intimin domain but also a highly conserved 90-amino acid repeats, similar to BipA of B. bronchiseptica and B. pertussis [21]. We therefore designated the B. holmesii protein as BipA, but not BcfA.
In the present study, the BipA mutant BH2Sm r -ΔBipA was associated with lower levels of initial colonization in murine lungs than the parental strain BH2Sm r , suggesting that BipA may function as an adhesion in B. holmesii. However, we also observed that the BipA mutant adhered to cultured human lung cells as large aggregates via autoagglutination (S4 Fig). Thus, while there is no evidence that the B. holmesii BipA acts as an adhesin, expression of the protein appears to be crucial during early stages of lung colonization. Consistent with this conclusion, previous studies demonstrated that the B. bronchiseptica BipA was not directly involved in respiratory tract colonization [21]. Furthermore, Vergara-Irigaray et al. suggested that BipA is not strictly required for respiratory tract colonization by B. pertussis, but is instead necessary immediately before and/or after a transmission event [35]. In regard to our observations, one possible explanation for the lower lung colonization level of the BipA mutant is that bacterial aggregates may be less efficient at adhering to the respiratory epithelium than single cells. Consequently, such aggregates would be easily cleared from respiratory tract sites. Alternatively, bacterial aggregates may be more rapidly eliminated from respiratory tract sites or efficiently killed by phagocytes than individual cells [36]. BipA may have the ability to promote suppression or evasion of the host innate immune system.
In B. pertussis, B. parapertussis, and B. bronchiseptica, BipA is characterized as a Bvg-intermediate phase protein. These BipA proteins are maximally expressed under intermediate Bvg i phase, but are expressed at very low or undetectable levels during the virulent Bvg + and avirulent Bvgphases, respectively [21,37]. Bvg phase transitions are dependent on environmental signals such as culture temperature and chemical modulators (MgSO 4 and nicotinic acid). In our preliminary study, however, we found that the transcript and protein levels of B. holmesii BipA were largely unaffected by the presence or absence of chemical modulators (S5 Fig). While B. holmesii encodes a BvgAS sensory transduction system [18,30], it is possible that, unlike in other Bordetella species, BipA may not be regulated by the BvgAS system in B. holmesii. In addition, we confirmed that putative Bvg binding sites are not present in the upstream region of bipA.
B. holmesii causes both invasive and respiratory diseases in humans, and the number of cases of respiratory infection has increased in the last decade worldwide [6,[8][9][10][11][12][13]. Genome analyses revealed that circulating B. holmesii strains are closely related, and that important virulence genes that are present in B. pertussis are not found in B. holmesii [16,17]. Moreover, no traits specific to B. holmesii respiratory or blood isolates were identified previously [38]. Notably, while each of the respiratory isolates tested in the present study produced BipA, 3 of the 5 invasive isolates tested, including 2 ATCC strains, did not. These findings suggest that BipA may play an important role during B. holmesii respiratory infections, but may be dispensable during invasive infections. This hypothesis is supported by our observation that the BipA-deficient mutant exhibited a lower initial colonization level than did its parental strain (Fig 6). However, because five of the 6 B. holmesii respiratory isolates tested were collected from the same outbreak in Japan [9], further study, using a larger number of B. holmesii isolates, is required to fully test this hypothesis.
Recently, BipA was identified as one of the most abundant proteins in B. pertussis biofilms [39]. Furthermore, immunization with recombinant B. pertussis BipA protected against B. pertussis infection in an animal model, suggesting that BipA may be potentially considered for inclusion as an antigen in pertussis vaccines to improve their effectiveness. Meanwhile, current pertussis vaccines confer little protection against B. holmesii infection [14]. As such, the development of a novel vaccine that is effective against B. holmesii will be necessary. Given that BipA was produced by all B. holmesii respiratory isolates tested in this study, this protein may also comprise an ideal antigen for use in the production of a vaccine to prevent B. holmesii respiratory infections. Indeed, attempts to develop a B. holmesii vaccine using BipA are now in progress.
In conclusion, BipA plays an essential role in preventing autoagglutination and indirectly promoting biofilm formation in B. holmesii. Bacterial biofilms are important contributors to chronic and persistent diseases. Because B. holmesii BipA production was detected in all respiratory isolates tested, we consider BipA a crucial virulence factor during respiratory infections by this organism. However, further studies using additional isolates are needed to elucidate the role of BipA in the pathogenesis of B. holmesii infections. The BipA + back-mutant BH7Sm r -BipA + was constructed from BipA-deficient isolate BH7 by double-crossover homologous recombination. A 2.5-kbp DNA fragment encoding the intact bipA gene was amplified by PCR with the attB1-BH2-bipA and attB2-BH2-bipA primers (S1 Table) using BH2 genomic DNA as the template. The resulting PCR product was cloned into pDONR221 and then combined with pABB-CRS2 using the Gateway cloning system to obtain pABB-bipA, which was trans-conjugated into BH7Sm r via E. coli SM10λpir. The resulting mutant was designated BH7Sm r -BipA + . (A) BipA production in the strain was confirmed by immunoblot analysis using anti-BipA antisera. (B) BH7Sm r -BipA + , its parental strain BH7Sm r , and BH7 were suspended in casamino acid solution and incubated for 4 h at 36°C under static conditions. The turbidity (OD 650 ) of the bacterial suspensions was measured every 30 min. (EPS) S4 Fig. Adherence of BipA mutant to A549 cells. A549 cells were grown in DMEM containing 1% FBS on a glass coverslips in 6-well tissue culture plates. After a 24 h culture, the medium was removed and replaced with fresh DMEM. BH2, BH2Sm r , or BH2Sm r -ΔBipA (6 × 10 7 CFU in 1% casamino acid solution) was added to the A549 cells at a multiplicity of infection (MOI) of 200. The plates were centrifuged at 200 × g for 5 min, then incubated at 36°C. After 2 h, the cells were washed, fixed with 4% paraformaldehyde, permeabilized with 0.2% TritonX-100, and quenched with 50 mM ammonium chloride. The fixed cells were incubated with a B. holmesii BH7-specific primary antibody followed by a fluorescein isothiocyanate (FITC)-conjugated secondary antibody (green). Actin and nuclear DNA were visualized by rhodamine phalloidin (red) and TO-PRO3 (blue) staining, respectively. Subsequently, the bacterial cells were examined by fluorescence microscopy using an Olympus IX83 (objective magnification, 100×).