Bacterial strains used in this study are listed in Table 1. Overnight cultures were incubated at 37°C and shaken at 200 rpm in LB medium with the appropriate antibiotics. To express heterologous ComX, a 2% (v/v) of overnight culture of Escherichia coli ED367 grown in LB medium supplemented with 100 μg/ml ampicillin was inoculated into the fresh M9 minimal medium (Ansaldi et al., 2002). Cells were grown to OD₆₅₀ of ∼0.7 a.u. at 37°C and 200 rpm. Then IPTG was added in final concentration 0.4 mM. Cells were incubated further for 4 h at 200 rpm and 37°C and then centrifuged at 8,000 g for 10 min. M9 spent medium was sterilized through filters with 0.2 μm pores and stored at 4°C until use.

To prepare B. subtilis spores, the overnight culture [1% (v/v)] was inoculated into sporulation medium (SM) (Warriner and Waites, 1999). After 5 days of incubation, the culture was exposed to 80°C for 30 min and then centrifuged at 10,000 g for 10 min. The pellet was re-suspended and washed three times with physiological saline. Before freezing at -20°C 10% (v/v) glycerol was added to the spore suspension. Spores were enumerated using the MPN method (data not shown). A 1% (v/v) solution of triphenyltetrazolium chloride (TTC; BioLife, Italy) was sterilized through filters with 0.2 μm pores. 180 μl of the liquid LB medium with TTC [0.01% (v/v) final concentration] was dispensed in standard 96-well sterile microtiter plates. A 10-fold serial dilution of the spore suspension was prepared in microtiter plates with eight technical replicates for each 10-fold dilution down to 10^-11. Microtiter plates were then incubated at 37°C overnight. Positive wells were identified by red color development due to the bacterial growth. Spore suspensions with MPN count in the 10⁸ MPN/mL range were used in the experiments.

To measure gene expression in floating biofilms (pellicles) a spore suspension 1% (v/v) was inoculated into the liquid MSgg medium. In some experiments the MSgg medium was supplemented with 20% (v/v) of the spent medium containing ComX, which was heterologously produced by E. coli ED367 (Ansaldi et al., 2002). The spent M9 medium of E. coli ED367 that was not induced by IPTG was used as a negative control.

To harvest pellicle spent media 1% (v/v) of B. subtilis spores were inoculated in 4 ml of MSgg medium (Branda et al., 2001), in some cases complemented with 20% (v/v) spent M9 minimal medium and incubated in sterile 12-well microtiter plates in static conditions at 37°C.

Construction of the mutant strains was performed by transformation of specific markers into competent B. subtilis strains grown in competence medium (CM) at 37°C (Albano et al., 1987). Antibiotic selections were carried out on LB agar plates at 37°C containing chloramphenicol (Cm) 5 μg/ml, kanamycin (Kan) 50 μg/ml, spectinomycin (Spec) 100 μg/ml, tetracycline (Tet) 10 μg/ml, erythromycin 0.5 μg/ml, and lincomycin 12.5 μg/ml (mls). The comQ::spec mutants were constructed by transforming the B. subtilis PS-31, PS-53 with the DNA isolated from the strain BD2962 (Tortosa et al., 2001) and by transforming PS-196 and PS-218 with the DNA isolated from the strain BD3019 (Ansaldi et al., 2002). The comQ::kan mutant (BM1400) was constructed by transforming the B. subtilis PS-216 with the DNA isolated from the strain BD2876 (Tortosa et al., 2001). The degQ::tet mutant was constructed by transforming the genomic DNA from the strain BD7123 (Miras and Dubnau, 2016) to appropriate PS-216 strains. Mutants with PaprE-gfp were constructed by transforming O8G57 genomic DNA (Veening et al., 2008) into appropriate PS-216 strains. When constructing the amyE::PsrfAA-yfp mutants, the genomic DNA from the strain DL722 (López et al., 2009) was transformed into the strain BM1400. When transforming DNA into comQ mutants, the competence was achieved by the addition of exogenous ComX in the form of spent medium [5% (v/v) of the E. coli ED367 M9, which was grown in the presence of IPTG].

To construct the ΔcomQ marker less deletion strain, the region upstream of the comQ gene was PCR amplified using the primer pair Up-F/Up-R (Table 2) and digested with EcoRI and BamHI. Also the region downstream of comQ gene was PCR amplified using the primer pair Down-F/Down-R (Table 2) and digested with BamHI and SalI. The two fragments were then simultaneously ligated into the EcoRI and SalI sites of pMiniMAD2 (Parashar et al., 2013), which carries a temperature-sensitive origin of replication and an erythromycin resistance cassette to generate pMiniMAD2-updowncomQ. The constructed plasmid was transformed into B. subtilis PS-216 at the restrictive temperature for plasmid replication (37°C) using 0.5 μg/ml erythromycin and 12.5 μg/ml lincomycin (mls) as a selection. To evict the plasmid, the strain was harvested according to an established protocol (Patrick and Kearns, 2008). Chromosomal DNA from colonies that had excised the plasmid was isolated and screened by PCR using primers Up-F/Down-R to determine which isolates carried a deletion in comQ gene.

Skim milk powder 20% (w/v) was reconstituted with distilled water and autoclaved at 110°C. An agar solution 3% (w/v) was also prepared and autoclaved at 121°C. After autoclaving, the suspensions were carefully mixed in a 1:1 ratio. Finally 20 ml of the mixture was poured into 9 cm diameter Petri dishes, yielding 10% skim milk (w/v) and 1.5% (w/v) agar. For overnight cultures B. subtilis strains were incubated in LB medium supplemented with appropriate antibiotics at 37°C and 200 rpm. Cultures were then diluted 100-fold in physiological saline and a 5 μl droplet of diluted culture was surface spotted in the center of the skim milk agar plate. Photos of the proteolytic clearing zones were taken after 16 h of incubation at 37°C.

To prepare casein gelatin agar plates 1% (w/v) casein sodium salt from bovine milk (Sigma–Aldrich, United States) and 1% (w/v) gelatin from porcine skin (Sigma–Aldrich, United States) were thoroughly dissolved in 0.02 M NaOH and the pH was equilibrated to 7 ± 0.2 as described by Montville (1983). Wherever, the proteolytic inhibition by ethylenediaminetetraacetic acid (EDTA) was tested, EDTA was added to the casein gelatin medium in 1 mM final concentration. Finally, 1.5% (w/v) of agar was added to the casein gelatin medium. The casein gelatin medium was autoclaved at 110°C. 40 ml of the medium was poured into Petri dishes with 9 cm diameter. Wells (6 mm diameter) were cut into the casein gelatin agar using an agar punch cutter.

Proteolytic activity was determined in B. subtilis spent medium at different time points. The spent medium below the biofilm was centrifuged for 5 min at 8000 g to remove remaining planktonic cells and sterilized through filters with 0.2 μm pores. Spent medium was diluted in 10 mM sodium acetate buffer with 5 mM calcium acetate (pH 7.5). If EDTA was added to experimental samples, spent medium was diluted in PBS buffer (10 mM sodium phosphate dibasic, 1 mM potassium phosphate monobasic, 137 mM NaCl, and 2.7 mM KCl pH 7). The diluted spent medium (100 μl) was dispensed into the wells punched into the casein gelatin agar. The plates were incubated for 16 h at 37°C when the proteolytic zones were measured. To estimate the proteolytic activity of spent media in U/ml of subtilisin, the proteolytic zone diameters of the spent media were compared to the proteolytic zone diameters obtained by different concentrations of commercial subtilisin (Sigma–Aldrich). One such subtilisin dose-response curve is shown in Supplementary Figure 1. The subtilisin concentration equivalents were displayed as units where 1 unit of subtilisin hydrolyzes casein to produce color equivalent to 1.0 μmol (181 μg) of tyrosine per minute at pH 7.5 at 37°C using Folin-Ciocalteu reagent. According to Sigma–Aldrich and to our in-house control of subtilisin proteolytic activity following the Sigma’s protease activity assay (Cupp-Enyard, 2008), 1 mg of subtilisin corresponds to roughly 13 units. Protease activity estimation was done routinely after storage of subtilisin for over 6 months at -20°C, to determine that the storage conditions did not lower the subtilisin standard proteolytic activity.

Briefly, 200 μl aliquots of inoculated MSgg medium, sometimes supplemented with M9 spent medium were dispensed in a sterile 96-well black transparent bottom microtiter plate in four technical replicates. The lid was sealed with micropore tape and the space between wells was filled with sterile distilled water to minimize the effect of medium evaporation. The microtiter plate was incubated in the Cytation 3 imaging reader (BioTek, United States) at 37°C without shaking. Optical density at 650 nm and fluorescence intensity were measured in half hour intervals for up to 60 h. Fluorescence intensity of GFP (green fluorescent protein) was used to monitor PaprE-gfp expression with excitation at 480 nm and emission at 510 nm. The gain was set on 50. In a parallel experimental setup the same strains without the fluorescent marker were always cultured. To calculate the final expression the autofluorescence of unmarked strains was deducted from the fluorescence of the marked strains. Fluorescence intensity was normalized per OD₆₅₀ of fluorescently labeled strains at each time point.

The ComX biological activity was quantified in spent media of various B. subtilis strains harvested after 36 and 48 h of static growth, using biosensor strains: B. subtilis BM1456 and BM1400. We chose the 48 h time point because then all strains have comparable floating biofilm (pellicle) thickness, estimated by measuring OD₆₅₀ (Supplementary Figure 2B). The BM1400 strain only served to determine background fluorescence, since it carries no fluorescent marker. Both biosensor strains do not produce their own ComX. The BM1456 biosenor responds to ComX in the spent medium by inducing the PsrfAA-yfp reporter (Oslizlo et al., 2014). Since the induction of PsrfAA is proportional to the quantity of ComX (Ansaldi et al., 2002) we used this assay to estimate the quantity of biologically active ComX in the spent media of various pellicles.

Both biosensor strains were grown in 5 ml of CM medium supplemented with 1% (v/v) of filtered spent media of selected strains for 6 h at 37°C with shaking (200 rpm). After 6 h of incubation 200 μl aliquots were dispensed in a sterile 96-well black transparent bottom microtiter plate. The PsrfAA-yfp expression was quantified by YFP (yellow fluorescent protein) fluorescence intensity. YFP was excited at 510 nm and emissions were measured at 530 nm. The gain was set to 100. Each PsrfAA-yfp reading was normalized to OD₆₅₀. The normalized autofluorescence background of BM1400 grown in the same experimental conditions was subtracted. Additionally, we subtracted the PsrfAA-yfp expression of cells exposed to ΔcomQ spent medium (lacks ComX), to account for ComX independent YFP expression.

ComX biological activity was determined in the spent media immediately after harvest (T₀) and after a 24 h (T₂₄) incubation at 37°C. We reasoned that if exoproteases in the spent medium degrade ComX the ComX biological activity at T₀ will be greater than at T₂₄. To calculate the ComX biological activity decay we used the following equation:

The same approach was used to determine ComX biological activity decay in the spent media supplemented with subtilisin (0.1 mg/ml) or metalloprotease inhibitor EDTA (1 mM) after harvesting or to estimate the effect of commercially available subtilisin on ComX heterologously produced in E. coli ED367.

ComX was purified as described previously (Ansaldi et al., 2002; Oslizlo et al., 2014). Briefly, 5 ml of ED367 spent medium was acidified and separated with a C-18 reverse phase HPLC column. ComX presence was determined by measuring the absorbance at 214 nm, where it appears as two chromatographic peaks, both of which exhibit biological activity (Oslizlo et al., 2014). The medium was purified after adding subtilisin (0.1 mg/ml) and incubating it for 24 h at 37°C. To ensure heterologous ComX stability over time bovine serum albumin (BSA, 50 μg/ml) was added to the spent media prior to tests. The resulting chromatogram was compared to a control experiment, where subtilisin was not added. To control for the presence of other signals potentially produced by E. coli ED367, the spent medium of the non-induced E. coli was subjected to the same procedure.

All results were statistically analyzed using the non-parametric Mann–Whitney U-test. The threshold level for accepting significant differences was p = 0.05. All of the error bars show the standard error of mean (SEM).