Biofilm formation by Paracoccus denitrificans requires a type I secretion system-dependent adhesin BapA

Abstract

Paracoccus denitrificans is a non-swimming Gram-negative bacterium, with versatile respiration capability which has remarkable potentials for bioremediation, especially in water treatment. Although biofilms are important in water treatment systems, the genetic mechanisms underlying the cellular adherence and biofilm formation of this bacterium remain unknown. We show that P. denitrificans forms a thin biofilm on surfaces at the air–liquid interface under static conditions. The initial step of biofilm formation requires a biofilm-associated protein BapA, which we identified by transposon mutant screening. BapA contains a unique sequence of dipeptide repeats of aspartate and alanine. Our data indicate that BapA is translocated to the extracellular milieu by a type 1 secretion system, where it enables the cells to attach to the substratum. Furthermore, superresolution microscopy shows that BapA is localized on the cell surface, which alters the cell surface hydrophobicity. Our results show a crucial role of BapA that promotes the adhesion and biofilm formation of P. denitrificans.

biofilm, adhesin, bap, type I secretion system, motility, Paracoccus denitrificans

INTRODUCTION

Biofilms are microbial communities comprised of cells surrounded by extracellular polymeric substances such as polysaccharides, proteins and nucleic acids (O’Toole, Kaplan and Kolter 2000). Biofilm formation is a common survival strategy of bacteria and it provides advantages of a physical protection of cells and physiological resistance to various stresses (Hall-Stoodley, Costerton and Stoodley 2004). The initial stage of surface attachment is a decisive event whether to switch bacterial lifestyles from planktonic to sessile modes, and it involves the physical forces that transport cells to a surface and the subsequent surface recognition and attachment: without these, the lifestyle switch from planktonic to sessile may not occur (Tuson and Weibel 2013). Swimming motility, which greatly speeds bacterial surface attachment and affects developmental processes of biofilms (O’Toole and Kolter 1998a; Watnick and Kolter 1999), is driven by a motor rotated flagella. Although the importance of motility in biofilm formation has extensively been studied, direct observations of environmental samples indicate that certain fractions of bacteria, ranging from 30% to 75%, are non-motile (Fenchel 2001; Grossart, Riemann and Azam 2001), and how non-motile bacteria develop biofilms is not fully understood, especially in Gram-negative bacteria.

Paracoccus denitrificans is a non-swimming, metabolically versatile Gram-negative bacterium in the class Alphaproteobacteria (Baker et al.1998). Its metabolic capability shows potentials for water treatment systems, where they are often associated with surfaces or forming biofilms (Neef et al.1996; Uemoto and Saiki 1996). Environmental conditions that influence the biofilm formation in Paracoccus species have been studied (Srinandan et al.2010), but yet the fundamental mechanism of how this bacterium attaches to surfaces and forms biofilms has not been examined in detail.

In this study, we show that P. denitrificans forms biofilms on surfaces at the air–liquid interface under static conditions. By transposon mutagenesis, we identify a gene encoding an adhesin required for biofilm formation. Due to its sequence information and its involvement in biofilm formation, we named this large protein BapA (biofilm-associated protein A). Genes encoding the type 1 secretion system (T1SS) were also found in the same gene cluster, which suggests that they are involved in the transport of BapA to the extracellular milieu. To our knowledge, this is the first report of a T1SS-dependent cell adhesin in Alphaproteobacteria. Our data show that BapA enables the cells to attach to the surface by altering the hydrophobicity of the cells. Our results suggest a pivotal role of this adhesin in initiating biofilm formation in P. denitrificans.

MATERIALS AND METHODS

Strains and growth conditions

Strains used in this study are listed in Table 1. Paracoccus denitrificans PD1222 was routinely cultured in tryptic soy broth (TSB) or on TSB agar (1.5%) at 30°C (de Vries et al.1989). For marker-less deletion mutant construction, we carried out in-frame deletion using pK18mobsacB as previously described (Schäfer et al.1994; Sullivan et al.2013). Mutants were selected with 100 and 50 μg mL⁻¹ kanamycin for P. denitrificans and Escherichia coli, respectively. Rifampicin was used at 100 μg mL⁻¹ for P. denitrificans. Oligonucleotide primers are listed in Table S1 (Supporting Information) . Transposon (Tn) mutants were generated with the Tn5-carrying plasmid pSUP2021 as previously described (Simon, Priefer and Pühler 1983; Shearer et al.1999). Tn insertion sites were identified by sequencing the DNA fragments amplified by arbitrary PCR (O’Toole and Kolter 1998b). Proteinase K and DNase I were used at 1000 and 100 μg mL⁻¹, respectively. Heat inactivation was performed at 95°C for 30 min.

Table 1.

Strains and plasmids used in this study.

Strain, plasmid	Relevant characteristics	Source or reference
Strains
Escherichia coli
DH5α	E. coli strain for transformation (F−, lacZΔM1, recA)	TaKaRa
S17-1	Mobilizer strain	Simon, Priefer and Pühler (1983)
Paracoccus denitrificans
PD1222	Wild type, RifR	de Vries et al. (1989)
PD1222ΔbapA	bapA deletion mutant of PD1222	This study
PD1222ΔbapB	bapB deletion mutant of PD1222	This study
PD1222ΔbapC	bapC deletion mutant of PD1222	This study
PD1222ΔbapD	bapD deletion mutant of PD1222	This study
PD1222ΔRpts	BapA amino acids 114−1895 deletion mutant of PD1222	This study
PD1222ΔflgE	flgE deletion mutant of PD1222	This study
PD1222ΔfliG	fliG deletion mutant of PD1222	This study
Plasmids
pSUP2021	Transposition vector; Tn5	Simon, Priefer and Pühler (1983)
pK18mobsacB	Suicide vector; sacB KmR	Schäfer et al. (1994)

Biofilm formation

Biofilm formation was assessed as previously described (O’Toole and Kolter 1998b). Strains were incubated in polystyrene 24-well microtiter plates (Iwaki, Shizuoka, Japan) at an initial optical density of 0.05 at 660 nm for 48 h at 30°C. Optical densities of the cell cultures were measured, and biofilms were stained with 0.1% (wt/vol) crystal violet solution for 30 min. For quantification, we added 20% (vol/vol) dimethyl sulfoxide in ethanol to each well and measured the absorbance at 595 nm.

For microscopic investigation, biofilms were formed on a piece of polystyrene (15 × 10 mm) at the air–liquid interface, and were visualized by using a LSM880 confocal laser scanning microscope (CLSM; Carl Zeiss, Oberkochen, Germany), as previously described (Obana, Nakamura and Nomura 2014). Cells were stained with SYTO9 nucleic acid stain (Thermo Fisher Scientific, Waltham, MA, USA).

Immunolabeling

Localization of BapA was examined by western blotting with an anti-peptide antibody recognizing a BapA sequence. We generated a rabbit polyclonal antibody against BapA using a 19-aa peptide (VATDFLASTEEDLHGTHTV) synthesized according to the BapA C-terminus sequence (Sigma-Aldrich, St. Louis, MO, USA). For sample fractionation, overnight cultures were centrifuged. Proteins in the supernatants were precipitated in 20% (wt/vol) trichloroacetic acid, washed in acetone and the pellets were eluted in the resuspension buffer (pH 8.0), consisting of 20 mM Tris-HCl and 10 mM MgCl₂ (Hinsa et al.2003). The cell pellets were resuspended in the resuspension buffer with 2% (wt/vol) sodium lauroylsarcosinate, sonicated, centrifuged and the supernatants were collected as the cell fractions. BapA proteins were detected in the fractionated samples by western blotting as previously described (Turnbull et al.2016). In SDS-PAGE, equivalent amounts of 0.036 and 0.180 OD₆₆₀ units for the cell and supernatant fractions, respectively, were loaded to a 4%–20% gradient gel.

Immunofluorescent staining of BapA was performed with fluorescent labeling of the anti-BapA antibody with an anti-rabbit CAGE552-conjugated antibody (Abberior, Göttingen, Germany). Cells grown in TSB media were (i) fixed with 4% (vol/vol) formaldehyde in PBS; (ii) blocked with 1% (wt/vol) BSA and 50 mM NH₄Cl in PBS; (iii) BapA was probed with the antibody diluted 1:100 in PBS containing 1% BSA; and (iv) this was followed by a secondary probing with the anti-rabbit CAGE552-conjugated antibody diluted 1:100 in PBS containing 1% BSA. Immunolabeled BapA and nucleic acids of cells stained with SYTO9 were imaged with an ELYRA PS.1 photoactivated localization microscope (PALM; Carl Zeiss).

BATH tests

Relative hydrophobicity of the cell surface was evaluated by the bacterial adherence to hydrocarbons (BATH) test as previously described (Yoshida et al.2015). Briefly, 3 mL of washed cell suspensions with an optical density of 0.7 in PBS and 1 mL of p-xylene were mixed and incubated at 30°C for 1 h. Then optical densities of the aqueous phases were measured. The ratios of reduction in the cell densities were calculated as relative hydrophobicity of the cells.

Nucleotide sequence accession number

The nucleic acid sequence of bapA is deposited in DDBJ with an accession number of BR001397.

RESULTS AND DISCUSSION

Paracoccus denitrificans forms a biofilm at the air–liquid interface under static conditions

Paracoccus denitrificans formed a biofilm at the air–liquid interface under static conditions that was firmly attached to a microtiter dish (Fig. 1A). Paracoccus denitrificans possesses most of the flagellar genes except fliD, which encodes a filament cap protein shown to be critical to swimming motility in other bacteria (Arora et al.1998; Sampaio et al.2016) (GenBank accession NC_008686-008688), and we confirmed that they do not swim in the liquid medium (data not shown). To examine if the remnants of the flagellar components are involved in biofilm formation, flgE and fliG mutants were tested. Deletions of flgE and fliG genes, encoding flagellar hook and motor proteins, respectively, did not affect biofilm formation (Fig. 1A), which indicates that these flagellar genes are not necessary for biofilm formation in this strain. Using CLSM with nucleic acid staining, we observed a very thin and dense biofilm with a thickness ∼4 μm (Fig. 1B). We frequently observed cells in chain on the surface, which are presumably formed by division of the cells (Fig. 1C). We assume that P. denitrificans PD1222 forms biofilms by translocating on surfaces by growth-driven spreading (Recht et al.2000), which explains why the cells are so densely packed in the biofilm. To estimate the extracellular components involved in P. denitrificans biofilm formation, we added DNase I or proteinase K to the media prior to incubation. Proteinase K significantly inhibited biofilm formation (Fig. 1A), while DNase I did not affect biofilm formation; this suggests that proteinous components as opposed to extracellular DNA are essential to biofilm formation.

Figure 1.

Biofilm formation in P. denitrificans. (A) Biofilm formation by the WT, ΔflgE and ΔfliG mutants in microtiter plates under static conditions. Proteinase K (designated ProK, 1000 μg mL⁻¹) or DNase I (100 μg mL⁻¹) was added to the media of the WT prior to incubation. Biofilm was stained by crystal violet (above) and quantified by measuring absorbance of the eluted stain at 595 nm (below). Growth of the static cultures was measured by optical density at 660 nm. Means of three independent experiments are shown with s.d. ^#P < 0.05, two-tailed Student's t-test with Welch's correction for A₅₉₅/OD₆₆₀ values in comparison with the WT. (B and C) CLSM images of biofilms formed by the WT and the ΔbapA mutant. (C) Magnified view of the squared field in the WT biofilm is shown. Arrows indicate representative images of cells in chain. Biofilms were formed on polystyrene at the air–liquid interface. Cells were incubated in microtiter plates for 48 h and nucleic acids were stained by SYTO9. Scale bars, 10 μm.

Screening of genes required for biofilm formation

To identify genes required for biofilm formation, we generated transposon mutants with Tn5 and screened for strains unable to form a biofilm. Approximately 4500 mutants were screened and we isolated 26 mutants that are defective in biofilm formation. Sequencing analyses of the transposon insertion sites revealed that 11 of the transposons were inserted in the genomic region between PDEN_RS11975 (old locus tag, Pden_2411) and PDEN_RS12005 (old locus tag, Pden_2416) genes. This indicates that this region is essential for biofilm formation (Fig. S2, Supporting Information), and therefore was further investigated in detail.

In this region, genes homologous to the HlyBD-TolC T1SS are present. The T1SS secretes its specific substrate including hemophore, cytotoxin, protease, lipase and BAPs from the cytosol to the extracellular space (Thomas, Holland and Schmitt 2014). In search of the specific substrate of the putative T1SS, we identified a 6636-bp open reading frame (ORF) with a potential ribosome-binding site (GGAG) located 7 nt upstream of the start codon, upstream of the gene encoding a HlyB-type inner membrane protein (PDEN_RS11985) (Fig. 2A). This ORF was not identified in the genome database but two ORFs, namely PDEN_RS11990 and PDEN_RS11995, are misannotated in this locus (GenBank accession NC_008686-008688). The putative gene that we found encodes a 2211-aa protein with a large repeat region (Fig. S3, Supporting Information), sharing common features with the BAPs found in other Gram-negative bacteria, typically found with large sizes consisting of >2000 aa and contains at least one extensive repeat region (Satchell 2011). BAPs in Gram-negative bacteria are large proteins involved in biofilm formation, which are secreted with T1SS encoded by genes homologous to the Escherichia coli hlyB, hlyD and tolC genes (Satchell 2011). Given the protein features and its role in biofilm formation, we named this gene bapA. Accordingly, the PDEN_RS11985, PDEN_RS11980 and PDEN_RS11975 genes, which were homologous to the T1SS genes of hlyB, hlyD and tolC (Thomas, Holland and Schmitt 2014), were designated bapB, bapC and bapD, and were further characterized for their roles in biofilm formation.

Figure 2.

Biofilm formation by the different bap mutants. (A) Genetic organization of the bap region. (B) Schematic of the BapA protein. BapA contains 891 tandem repeats of two amino acids, aspartate and alanine. (C) Biofilm formation by the WT, ΔbapA, ΔbapB, ΔbapC, ΔbapD and ΔRpts mutant strains. Biofilms stained with crystal violet were quantified by absorbance at 595 nm, and growth of the static cultures was measured by optical density at 660 nm. Means of three independent experiments are shown with s.d. ^#P < 0.05, two-tailed Student's t-test with Welch's correction for A₅₉₅/OD₆₆₀ values in comparison with the WT.

bapA is required for biofilm formation

To confirm the results of our transposon mutagenesis experiment and examine the roles of each bap gene in biofilm formation, we constructed in-frame mutants of bapA, bapB, bapC and bapD and performed biofilm formation assays. Although the growth of these mutants slightly decreased compared with the WT under static conditions, the mutants did not form biofilms (Fig. 2C), which indicates that these genes are required for biofilm formation. Compared to the WT, very few ΔbapA mutant cells were attached to the surface (Fig. 1B), indicating that BapA plays a crucial role in cell-to-surface attachment. We note that the growth of the mutants did not decrease under shaking conditions (Fig. S4, Supporting Information).

Protein sequence analysis predicts that BapA is composed of three main regions (Fig. 2B) (Mitchell et al.2015): (i) the N-terminal region consisting of the first 113 amino acid residues; (ii) the subsequent 1782 amino acid residues that consist of tandem repeats of the two amino acids, aspartate and alanine; and (iii) the C-terminal region consisting of 316 amino acid residues. While the N-terminal and C-terminal regions of BapA show sequence similarity to some of the previously characterized adhesins, the extensive repeat region shows no similarity. The N-terminal region is 35% identical to the N-terminal region of BapA in Salmonella enterica (Latasa et al.2005), while the C-terminal region is 31% identical to the C-terminal region of LapF in P. putida (Martínez-Gil, Yousef-Coronado and Espinosa-Urgel 2010) in amino acid sequence. Interestingly, while the repeats in T1SS-dependent adhesins typically consist of 80 to 300 aa (Satchell 2011), the repeats in BapA consist of only two amino acids of aspartate and alanine (Fig. S3), which BLAST analysis against prokaryotes revealed unique to P. denitrificans. Because it has been reported that the large repeat regions in BAPs are required for their functions (Boyd et al.2014; Nishikawa et al.2016), we hypothesized that this region is important for the BapA function. Indeed, deletion of the tandem repeat region (amino acid residue 114 to 1895, designated ΔRpts) resulted in similar defects in biofilm formation of the bap mutants (Fig. 2C).

BapA secretion is BapBCD dependent and mediates biofilm formation

To examine whether BapA secretion depends on the BapBCD T1SS, the presence of BapA was examined in the cell lysate and supernatant by western blotting in a series of mutants. To this end, we generated an anti-BapA C-terminus antibody using a 19-aa peptide of the C-terminus. A band of ∼300 kDa was detected by western blotting in both cell and supernatant fractions of the WT with the anti-BapA antibody, which was not detected in the ΔbapA mutant (Fig. 3A); this indicates that this band is BapA and confirms that the bapA gene we annotated is translated. Smeared bands that were detected in the WT were absent in the western blot of the ΔbapA mutant, suggesting that they represent degraded BapA proteins. For the ΔbapB, ΔbapC and ΔbapD mutants, levels of BapA in the supernatant fractions were significantly reduced (Fig. 3A), indicating that BapBCD mediate BapA transport. We note that the theoretical molecular mass of BapA is 210 kDa, whereas the BapA band in the western blot appeared at ∼300 kDa. Our Southern blotting result excluded the possibility of the bapA sequence misassembling, which often occurs with repeated sequences (Fig. S5, Supporting Information). Because acidic proteins have lower mobility in SDS-PAGE (Shirai et al.2008), we reasoned that the BapA band shift is due to the extreme acidity of the Rpts repeat region that has a predicted isoelectric point of 1.1 (http://web.expasy.org/compute_pi/). In agreement with this, a band corresponding to the theoretical molecular mass (46 kDa) of BapA whose Rpts repeat region was deleted (designated BapA-Rdel) was detected in the cell fraction of the ΔRpts mutant (Fig. 3A). This BapA-Rdel was also clearly detected in the supernatant fraction (Fig. 3A), indicating that polar effects are negligible as the BapBCD transporter that is encoded downstream of bapA was functional in this mutant.

Figure 3.

BapBCD is involved in BapA secretion and an anti-BapA antibody inhibits biofilm formation. (A) Western blot of BapA. BapA was detected from cell lysate (C) and supernatant (S) fractions of the WT, ΔbapA, ΔbapB, ΔbapC, ΔbapD and ΔRpts mutant strains with an anti-BapA antibody recognizing 19 C-terminal residues of BapA. The arrow indicates BapA bands. ΔRpts, deletion mutant of the BapA Rpts repeat region. (B) WT biofilm formation in the presence of the anti-BapA antibody. The antibody diluted 1:100 or the heat-inactivated (HI) antibody was added to media prior to incubation. Biofilms were stained with crystal violet and quantified by absorbance at 595 nm, and growth of the static cultures was measured by optical density at 660 nm. Means of three independent experiments are shown with s.d. ^#P < 0.05, two-tailed Student's t-test with Welch's correction for A₅₉₅/OD₆₆₀ values.

To determine if BapA functions extracelluarly, we added the anti-BapA antibody to the media prior to incubation. The addition of the antibody inhibited biofilm formation by the WT, presumably by counteracting the effect of BapA, while the heat-inactivated antibody had no effect on biofilm formation (Fig. 3B). Together with the observation that proteinase K treatment inhibits biofilm formation (Fig. 1B), these results indicate that the extracellularly secreted BapA plays a major role in biofilm formation by supporting the cell attachment to surfaces.

BapA is localized on the cell surface and alters cell hydrophobicity

It has been reported that a number of adhesins secreted by T1SS are localized on the cell surface to mediate cell-to-surface interactions (Nishikawa et al.2016). Given that BapA is secreted and involved in the cell-to-surface interactions (Figs 1B and 3), we hypothesized that BapA resides on the cell surface to initiate biofilm formation. The localization of BapA was analyzed by immunofluorescence microscopy with a CAGE552-conjugated antibody using a superresolution microscope. In the WT, BapA was present in small discrete spots on the cell surface, which were not observed in the ΔbapA mutant (Fig. 4A), suggesting that BapA resides on the cell surface.

Figure 4.

BapA localizes on the cell surface and increases cell hydrophobicity. (A) Immunostaining images of BapA. BapA was stained with the antibody and imaged by PALM. Nucleic acids were stained with SYTO9. (B) Cell surface hydrophobicity of the WT, ΔbapA, ΔbapB, ΔbapC, ΔbapD and ΔRpts mutant strains. Relative hydrophobicity of cells was measured by BATH test. Means of three independent experiments are shown with s.d. ^#P < 0.05, two-tailed Student's t-test with Welch's correction in comparison with the WT.

It is known that BAPs support cell adhesion by increasing the cell surface hydrophobicity (Huber et al.2002). To examine if BapA functions in a similar manner in surface attachment, we measured cell surface hydrophobicity. Deletions of the bapA gene as well as the ΔbapB, ΔbapC and ΔbapD genes significantly reduced hydrophobicity of the cells compared with the WT (Fig. 4B), supporting the idea that BapA is a hydrophobic protein that resides on the cell surface. Furthermore, deletion of the Rpts repeat region in BapA resulted in a similar decrease in cell hydrophobicity as the bapA mutant (Fig. 4B), indicating that the repetitive region consisting of aspartate and alanine is critical for cell surface hydrophobicity. Cell surface hydrophobicity is one of the critical properties involved in bacterial adhesion (van Loosdrecht et al.1987), which is altered by BAPs involved in cell adhesion to certain surfaces (Toledo-Arana et al.2001; Huber et al.2002). Although not conclusive, the length scale of attraction of hydrophobic interactions is <10 nm (Meyer, Rosenberg and Israelachvili 2006), which is in line with our result that the BapA-specific antibody, whose size is >10 nm, inhibits surface adhesion. Most of the works that characterize BAPs have been carried out in bacteria that possess swimming motility, showing that motility significantly accelerates initial attachment of the bacteria whereas BAPs play an important role in the transition from reversible to irreversible attachment once the bacteria are in contact with the surface (Huber et al.2002; Hinsa et al.2003). Cell surface adhesins are also reported to be essential to the attachment of non-motile Gram-positive bacteria to surfaces (Lasa and Penades 2006; Latasa et al.2006), such as Bap in Staphylococcus aureus (Cucarella et al.2001), shedding light on the universal role of BAPs in biofilm formation. Given the inability to actively move in liquids, our results imply a key strategy used by P. denitrificans, where they alter their surface properties by expressing BapA to initiate biofilm formation.

SUPPLEMENTARY DATA

Supplementary data are available at FEMSLE online.

Acknowledgments

We thank Prof. Alfred Pühler (Bielefield University) for providing us with pSUP2021 vector and Prof. David J. Richardson (University of East Anglia) for kindly providing us with Paracoccus denitrificans PD1222. We thank Prof. Andrew S. Utada (University of Tsukuba) for critical comments on the manuscript.

FUNDING

This study was funded by Japan Science and Technology Agency ERATO, and a Grant-in-Aid for Scientific Reseach to NN (60292520) and MT (25701012) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. KY was supported by a scientific research fellowship from the Japan Society for the Promotion of Science (JSPS). MT was supported by the JSPS Postdoctoral Fellowship for Research Abroad.

Conflict of interest. None declared.