Bacterial strains and plasmids used in this study are listed in Table 1. All strains were grown at 37°C under vigorous agitation in LB medium (VWR, UK) unless specified otherwise.

The cloning and transformation procedures were performed according to established techniques [14] and suppliers' manuals. Restriction enzymes, DNA polymerases, DNase I, RNase I, T4 DNA ligase were obtained from Fermentas Life Sciences (Vilnius, Lithuania) and used as specified by the suppliers. Deoxynucleotide primers for PCR were obtained from Invitrogen (UK), and Table 2 lists the sequences of primers used.

B. licheniformis EI-34-6 was grown in 10 ml Air Membrane Surface (AMS) bioreactors as described previously [15] in NGF medium (Nutrient broth (Oxoid) 13 g/l, 1% glycerol, 1 mM FeCl₂). After 7 days of growth, the medium underneath the filter membranes was collected, pooled, centrifuged at 7800 rpm in 50 ml falcon tubes for 10 min and filtered using a 0.2 µm syringe filter to ensure sterility.

Biofilm dispersal was screened using clear 96 well flat bottom polystyrene tissue culture plates (BD-Falcon, USA). Bacillus licheniformis DSM13 and other bacterial strains tested were grown for 48–96 h and diluted 1∶100 in fresh LB. 200 µl of this culture was added to every well of a 96 well plate. To test for inhibition of biofilm formation, the biofilm dispersal compound was added directly, and the plate was incubated at 37°C, without shaking, for 20–28 h to allow for biofilm development. To test for dispersal activity, the biofilm dispersal compounds were added in varying concentrations after 20–28 h of growth and biofilm development at 37°C, and the plate was further incubated for 1 h at 37°C. Then all non-attached cells were removed by discarding the culture medium and rinsing the plate in a container by immersing and agitating gently four times in tap water. Attached biofilm material was stained by addition of 250 µl of 0.5% crystal violet solution (CV) to each well of the plate for 10 min. Unbound CV stain was removed by aspiration and the plate was rinsed again in tap water until no more CV was observed to dissolve in the water. The plates were air dried and photographed. Subsequently, 250 µl of 96% ethanol containing 2% acetic acid (v/v) was added to each well. Adsorption at 595 nm was measured using a Fluostar Optima plate reader (BMG Labtech, UK), and the data was analysed using the MARS software package (BMG Labtech, UK) and Microsoft Excel.

The proteins in the AMS supernatant were concentrated 50 fold by precipitation with trichloroacetic acid (TCA) (Sigma, UK) as follows: The supernatant of several AMS cultures was pooled, and 6.1 M TCA solution was added to give a final concentration of 0.9 M TCA. This solution was kept on ice for 30 minutes to allow for protein precipitation and the precipitated protein was collected through centrifugation (10 min at 7800 rpm in 50 ml falcon tubes). The protein containing pellets were washed twice using ice cold 96% ethanol, and air dried for 30 min at 45°C. Each pellet was dissolved in 1∶50^th of the original volume with 0.05 M Tris-HCl buffer (pH 7.0). This concentrate was fractionated using a Superose™ 12 (GE Healthcare, UK) gel filtration column (height 40 cm, diameter 3 cm) using ultra pure water as the mobile phase and fractions of 12 ml each were collected. The fractions were tested for biofilm dispersal activity using the 96 well microtitre plate setup, with crude supernatant as the positive control and H₂O as the negative control. Proteins in the active fraction were concentrated again via TCA precipitation (as before) and analysed by SDS page.

The single active fraction from gel filtration on Superose™ 12 was concentrated 10× via TCA precipitation in 2 ml micro-tubes and separated on a 4–12% Tris-Tricine gel using MES buffer (Invitrogen, UK). A Novex Sharp Pre-stained protein standard (Invitrogen, UK) was also loaded to determine protein size. After electrophoresis the gel was stained using Biosafe Coomassie (Biorad, UK) according to the manufacturer's protocol. Three bands were visible on the gel, one abundant band at 12 kDa and two higher bands around 30 and 34 kDa. These three bands were analysed by LCMS, following in-gel tryptic digest (North East Proteome Analysis Facility (www.nepaf.com), Newcastle, UK). The peptide fragments were analysed against NCBI NC_006270.faa 2008.04.22 (Bacillus_licheniformis_ATCC_14580), NCBI NC_006322.faa 2008.04.22 (Bacillus_licheniformis_DSM_13), NCBI C_000964.faa 2008.04.22 (Bacillus_subtilis) and the common Repository for Adventitious Proteins.

Primers were designed to amplify both identified nuclease genes based on the published genome sequence of B. licheniformis DSM13 [16].

For barnase, primer sets were designed to amplify the gene only and also the barnase-barstar operon. PCR was performed using Phusion DNA polymerase (Finnzymes, Finland). Table 2 lists the nucleotide sequences of the primers used. The barnase gene and barnase-barstar operon were successfully amplified in one.

The PCR reaction to amplify nucB did result in several amplified fragments, and a faint band of the correct size was present. This band was isolated from the agarose gel (gel isolation kit, Invitrogen, UK) and used as a template for a new PCR. The amplified genes were digested with Eco91I, XbaI (nucB) and Eco91I, HindIII (barnase, barnase-barstar) and ligated into vectors pNZ8901 (Cm^R) and pNZ8902 (Ery^R). The ligation mixture was transformed to E. coli DH5 alpha. Colonies were screened using colony PCR with the unique primers mentioned above and plasmids were isolated from positive clones. Plasmids were analysed by restriction and correct plasmids were sequenced. Constructed plasmids are listed in Table 1. The constructed plasmids were transformed to Bacillus subtilis NZ8900 [17] using natural competence [18].

B. subtilis NZ8900+pNZ8901/2-nucB clones were screened on DNase test agar containing methyl Green (Oxoid, UK) as follows. A colony was streaked onto the DNase test agar and grown overnight at 30°C. Subsequently a drop of B. subtilis ATCC6633 culture supernatant containing subtilin and 1% agar was spotted next to the colony. The plates were further incubated for 2 h at 37°C and colonies developing a halo due to the degradation of DNA were judged positive. Correct B. subtilis clones containing the Barnase gene were characterized by colony-PCR followed by plasmid isolation and restriction analysis of the obtained plasmid.

Correct clones were picked from single colonies and transferred to a shake flask containing LB and the appropriate antibiotics (Kanamycin and Chloramphenicol/Erythromicin). At an OD₆₀₀ of ∼1.0, 5% cell free supernatant of an overnight B. subtilis ATCC6633 culture was added to provide subtilin to induce expression and the total culture supernatant was harvested 2 h after induction.

Overproduction of the NucB or Barnase was visualised on SDS-page after 10× concentration via TCA precipitation. The concentration of overproduced NucB was estimated on SDS-page by comparing band intensity after staining the gel with Bio-Safe Coomassie (Biorad) against a BSA standard. When production of Barnase was induced the culture stopped growing and no overproduction of Barnase could be detected on a Coomassie stained PAA gel. To circumvent this problem the gene downstream of Barnase, barstar, was also included in the overexpression construct. This strategy yielded an improvement in Barnase overexpression.

DNase activity was tested by incubating purified plasmid DNA with the DNase containing fractions for 30 min at 37°C. The samples were run on a 1% agarose gel containing ethidium bromide to visualize DNA degradation.

Proteins in the active fraction were identified by SDS-page and peptide mass fingerprinting. The lowest molecular weight band on the SDS-page gel, approx 12 kDa, contained two small proteins, both of them nucleases. The most abundant protein was Barnase (locus_tag: BL03601), a secreted ribonuclease, and the other protein was NucB (locus_tag: BL00126), a secreted deoxyribonuclease. The second band, cut out at approximately 30 kDa, contained three different proteins. The most abundant protein was protein YckK (locus_tag: BL01829) from the solute-binding family. Also present was the glycine betaine ABC transporter (opuAC; locus_tag: “BL01556”) and the ribonuclease present in the 12 kDa band. The third band, cut out at approximately 36 kDa, contained three different proteins. The most abundant protein was the same glycine betaine ABC transporter found in the 30 kDa band. Also present was an ABC transport system substrate-binding protein and probably also the putative extracellular solute-binding protein YckB (locus_tag: BL01818). We tested the correlation between biofilm dispersal activity and DNase activity and found that all culture media and fractions capable of dispersing biofilms also contained DNase activity (data not shown). Based on these results the two most likely candidates to have biofilm dispersal activity, the predicted ribonuclease Barnase and the predicted deoxyribonuclease NucB, were cloned.

Cloning and overexpression of NucB and Barnase was performed in Bacillus subtilis NZ8900. Primers were designed to amplify both identified nuclease genes based on the published genome sequence of B. licheniformis DSM13 [16]. Both genes were successfully amplified from B. licheniformis EI-34-6 chromosomal DNA and cloned into the SURE expression vectors pNZ8901 and pNZ8902 using E. coli as an intermediate host. Both vectors were transformed to the SURE expression strain B. subtilis NZ8900 [17]. Direct over-expression of NucB was successful with expression being under the control of pSpaS induced by subtilin (Fig. 4), whilst over-expression of Barnase required the inclusion of the gene downstream of Barnase, Barstar. Barstar is known to inhibit the intracellular RNase activity of the pre-barnase [19], thus allowing the over-expression and secretion of the Barnase itself, although on a lower level than for NucB.

The supernatant of B. subtilis NZ8900 containing the induced overexpression construct of NucB had strong DNase activity (data not shown) and was capable of dispersing established biofilms, whereas a control with an induced empty vector was not. The non-purified supernatant could disperse biofilms at NucB concentrations as low as 3 ng/ml (Fig. 5).

Heterologously expressed Barnase was also tested, and compared to commercially available RNaseI in its ability to disperse biofilms. No dispersal was observed using either heterologously produced Barnase or RNaseI in concentrations up to 10% supernatant/well or 200 units/ml respectively. We also tested the activity of Barnase in combination with bovine DNaseI or NucB, but no significant increase in dispersal was found compared to NucB or DNaseI alone (data not shown).

The B. licheniformis EI-34-6 genes for NucB and Barnase were sequenced (MWG operon, UK) to identify potential differences with the known sequenced strain. Both nucB (22bp of 428bp = 5.1%) and barnase (21bp of 455bp = 4.6%) contained base pair substitutions, leading to 4 amino acid changes in the NucB protein and 6 amino acids changes in the Barnase protein, compared to the DSM13 sequence. The sequences are available through Genbank, accession numbers HQ112343 (nucB) and HQ112344 (barnase).

Bovine DNaseI gave full dispersal of the bacterial biofilm above 15 ng/ml. At concentrations below 15 ng/ml there was still partial dispersal, and at the lowest level tested (0.7 ng/ml) ∼35% of the biofilm was dispersed (Fig. 5). For NucB, above 3 ng/ml there was full biofilm dispersal whilst below this concentration the dispersal activity dropped rapidly, and at 0.7 ng/ml only 15% of the biofilm was dispersed. From this experiment it is clear that eukaryotic DNaseI and bacterial NucB have a very different dose response curve in relation to biofilm dispersal. Importantly, NucB is fully dispersing the biofilm at a w/v concentration 5 times lower than that of DNaseI. (Fig. 5) The defined cut off of activity with NucB, compared to the more gradual loss of activity of DNaseI, suggests that the bacterial nuclease is better adapted to disrupt eDNA present in bacterial biofilms.