Determination of core and CBR-specific accessory biofilm determinants.

In previously reported genetic screens for biofilm determinants, OG1RF Tn mutants were grown as monocultures in microtiter plates (23, 27). This closed, static environment with no competing strains is substantially different from that of CBRs. To extend our understanding of environmental effects on E. faecalis biofilm formation, we sought to determine the overlap between mutants identified from microtiter plate screens and CBR TnSeq, which could constitute core OG1RF biofilm determinants. Because previous screens used TSB-D and not MM9-YEG, we examined the TSB-D TnSeq data from both library 1 and library 2 in this analysis but not the MM9-YEG data sets. Previously, a total of 204 insertions in 179 genes were associated with statistically reduced biofilm formation (23, 27). Only 35 Tn mutants were identified in both TnSeq and microtiter plate screens (Table 2), including the biofilm-associated phosphatase gene bph, autolysin gene atlA, stress response genes hrcA and dnaK, and the ebp pilus operon (23, 37,–39).

Next, we asked which Tn mutants were underrepresented in biofilm TnSeq but did not have reduced biofilm formation in previous studies. These mutants could have biofilm defects in a community of Tn mutants but not monoculture, or they could be accessory biofilm determinants that are important under flow conditions. Using a log₂FC cutoff of −1 for the TnSeq results, we identified 55 Tn mutations in 45 genes that were not found in previous studies (Table 3). These include multiple genes in the epa operon (OG1RF_11710 [epaOY], OG1RF_11714, OG1RF_11715 [epaOX], OG1RF_11716, and OG1RF_11722 [epaQ]), genes encoding predicted LCP family cell wall-modifying enzymes (OG1RF_10350, OG1RF_11288) and putative transcriptional regulators (OG1RF_12423 and OG1RF_12531), and genes annotated as hypothetical (OG1RF_10968 and OG1RF_11630).

We then sought to validate the importance of these genes for in vitro biofilm formation. However, large-scale testing of individual Tn mutants in CBRs is not feasible due to the volume of medium used for each reactor run (~10 liters) as well as the physical size and processing time required for each sample set. Therefore, we chose three previously described in vitro experiments to validate biofilm phenotypes: (i) a 96-well plate assay in which biofilm biomass is stained and quantified relative to cell growth (23, 30), (ii) a submerged substrate assay in which biofilms are grown on an Aclar disc covered by growth medium (16, 26), and (iii) a miniature 96-well flow reactor system (MultiRep reactor) in which 96 samples can be cultured in a total of 12 channels on 5-mm disks (34). Because both M9 and TSB-D were used in the CBR TnSeq screen, we carried out the following experiments with both media.

Phenotypes of “accessory” biofilm determinants in microtiter plate assays.

From the 55 Tn mutants presented in Table 3, we obtained 43 Tn mutants from the arrayed SmarT library stock plates. When multiple Tn insertions in a gene were identified, we chose only the insertion closest to the start codon. Additional mutations were excluded based on their location upstream of known biofilm determinants and the possibility that these insertions had polar effects on previously studied genes. To maintain consistency with previous experiments, we measured the biofilm production of the Tn mutants at 6h and 24h. A strain lacking bph, previously implicated in biofilm development (23), was used as a negative control, and biofilm production was normalized to that of OG1RF (Fig. 2A). In TSB-D, 12 mutants had significantly altered biofilm production relative to OG1RF at 6h (12 decreased, 0 increased) (Fig. 2B, black bars), and 5 mutants had altered biofilm levels at 24h (3 decreased, 2 increased) (Fig. 2B, pink bars). In MM9-YEG, 7 Tn mutants had altered biofilm production at 6h (2 decreased, 5 increased) (Fig. 2C, black bars), and 6 mutants had altered biofilm levels at 24h (5 decreased, 1 increased) (Fig. 2C, pink bars). Overall, ~30% of mutants (13/43) had reduced biofilm formation relative to OG1RF. Interestingly, some mutant strains had higher biofilm production in MM9-YEG than TSB-D, including the Δbph and OG1RF_10576 mutants, demonstrating that growth medium influences which genes are required for biofilm formation. We did not observe a correlation between the change in abundance (log₂FC) of Tn mutants in TnSeq and the biofilm index in microtiter plate biofilm assays (Fig. S1E to H).

Open in a separate window

FIG 2

Tn mutants identified from biofilm TnSeq have variable biofilm production in microtiter plates. (A) Heat map summarizing biofilm index values (A₄₅₀/A₆₀₀ relative to OG1RF) for all mutants. Biofilm index shading legends are shown on the right. (B and C) TSB-D biofilm index values (B) and MM9-YEG biofilm index values (C) for all Tn mutants with significantly altered biofilm production in either medium. For clarity, a dotted line is shown at the OG1RF biofilm index value. Plotted values are the same ones represented in the heat maps in panel A. (D) Biofilm phenotypes were complemented for tig-Tn. Strains carried either an empty pCIEtm plasmid or pCIEtm with the wild-type allele cloned under a pheromone-inducible promoter. Biofilm assays were carried out in the growth medium and for the length of time indicated in x-axis labels. All cultures were grown with 25ng/ml cCF10 to induce expression of the cloned tig gene. For panels B and C, three biological replicates were performed, each with two technical replicates. Statistical significance was evaluated by two-way analysis of variance (ANOVA) with Dunnett’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). For panel D, three biological replicates were performed, each with three technical replicates. Statistical significance was evaluated by two-way ANOVA with Sidak’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001).

Although all 43 Tn mutants were underrepresented in biofilm TnSeq, ~14% (6/43) had increased biofilm levels relative to that of OG1RF in 96-well plates (Fig. 2B and ​andC).C). We chose to complement the high biofilm phenotype of the tig-Tn (OG1RF_10452-Tn) mutant by expression of the wild-type gene from a pheromone-inducible plasmid (23). tig encodes trigger factor, a chaperone involved in folding newly synthesized proteins (40). Expression of tig from a plasmid significantly decreased biofilm relative to the Tn mutant carrying an empty-vector plasmid (Fig. 2D). The opposing biofilm phenotypes observed for some Tn mutants in CBR TnSeq compared to 96-well plates underscores how determinants of biofilm formation may vary across experimental platforms and suggests that molecular changes during biofilm development are highly sensitive to specific assay conditions.

Biofilm formation of Tn mutants in submerged substrate assays.

We chose 6 of the 43 Tn mutants described above for biofilm assays using submerged Aclar disc assays, in which strains are cultured in multiwell plates containing Aclar discs. These permit sampling of both planktonic and biofilm cells for visualization via microscopy and CFU quantification (16, 30). All 6 mutants were underrepresented in at least one library in biofilm TnSeq (Table 3; Tables S1 and S2) but had a range of phenotypes in the microtiter plate assays described above. Relative to parental OG1RF biofilm levels in 96-well plates, the prsA-Tn (encoding an extracellular peptidyl-prolyl isomerase [PPIase]) and OG1RF_10576-Tn (encoding a predicted DEAD box helicase) mutants had decreased biofilm. The tig-Tn (encoding trigger factor) mutant had increased biofilm, and OG1RF_10350-Tn, OG1RF_11456-Tn, and OG1RF_11288-Tn mutants did not have significantly different levels of biofilm compared to OG1RF (Fig. 2B and ​andCC).

We inoculated strains at 10⁷ CFU/ml and quantified planktonic and biofilm CFU/ml after 6h. In TSB-D, prsA-Tn, OG1RF_10576-Tn, and OG1RF_11456-Tn mutants had significantly lower numbers of planktonic CFU/ml than OG1RF (Fig. 3A, pink bars). The OG1RF_10576-Tn mutant had an ~1 log decrease in biofilm CFU/ml relative to OG1RF (Fig. 3A, green bars), although this difference was not statistically significant. To determine whether mutants had a biofilm-specific decrease in viable cells (as opposed to lower biofilm growth due to growth defects in planktonic culture), we calculated the ratio of biofilm growth to planktonic growth relative to that of OG1RF. By this metric, only the Δbph strain had a significant reduction relative to OG1RF (Fig. 3B).

Open in a separate window

FIG 3

Biofilm formation of selected Tn mutants by use of a submerged Aclar disc assay. (A) Numbers of CFU/ml of strains at 0h and 6h in TSB-D. The dotted line indicates the number of OG1RF biofilm CFU/ml. (B) Ratio of biofilm (BF) growth to planktonic (PL) growth relative to that of OG1RF. (C) Representative microscopy images of Hoechst 33342-stained biofilms from TSB-D cultures. (D) Numbers of CFU/ml of strains at 0h and 6h in MM9-YEG. The dotted line indicates the number of OG1RF biofilm CFU/ml. (E) Ratio of biofilm growth to planktonic growth relative to that of OG1RF. (F) Representative microscopy images of Hoechst 33342-stained biofilms from MM9-YEG cultures. For panels A and D, each data point represents the average of two technical replicates, and a total of four biological replicates were performed. Statistical significance was evaluated by two-way ANOVA with Dunnett’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). For panels B and E, values were obtained using the data points presented in panels A and D, respectively. Statistical significance was evaluated by one-way ANOVA with Dunnett’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). For panels C and F, samples were grown in parallel to cultures used to generate panels A and D. Scale bars,20μm. Two technical replicates were processed for each biological replicate, and representative images are shown.

Biofilms were visualized with fluorescence microscopy after staining with Hoechst 33342, a nucleic acid label. OG1RF biofilms consistently grew as a monolayer of short chains of bacteria with few multicellular aggregates or clumps (Fig. 3C). As previously observed, biofilms formed by the Δbph negative-control strain contained fewer cells than OG1RF (23). The appearance of OG1RF_10350-Tn, tig-Tn, and OG1RF_11288-Tn biofilms was similar to that of OG1RF. Although there was not a significant reduction in OG1RF_10576-Tn biofilm CFU relative to OG1RF (Fig. 3B), these mutant biofilms had visibly less surface coverage than OG1RF biofilms. prsA-Tn biofilms contained some multicellular aggregates, and OG1RF_11456-Tn biofilms had large clumps of cells (Fig. 3C).

We next examined the growth of these mutants in MM9-YEG. Unlike the corresponding experiments in TSB-D (Fig. 3A, pink bars), no mutants had reduced numbers of CFU/ml in planktonic culture (Fig. 3D, pink bars). Additionally, none of the mutants had reduced numbers of CFU/ml in biofilms (Fig. 3D, green bars) or ratio of biofilm growth to planktonic growth relative to OG1RF (Fig. 3E). However, visualization of Aclar substrates revealed substantial differences in biofilm architecture. In MM9-YEG, OG1RF formed a monolayer biofilm composed mainly of single cells and some small aggregates (Fig. 3F). The Δbph biofilm had less surface coverage but was still composed of mostly single cells. All Tn mutants formed biofilms with multicellular aggregates. prsA-Tn, tig-Tn, and OG1RF_10576-Tn biofilms had mixtures of single cells and small multicellular chains, while nearly all cells in OG1RF_11456-Tn biofilms grew as chains and aggregates. Interestingly, fewer multicellular chains and more individual cells were observed in biofilms grown in MM9-YEG than in TSB-D (compare Fig. 3C and ​andF).F). Conversely, more large multicellular aggregates were observed in MM9-YEG than in TSB-D, suggesting that nutritional components could regulate cell chaining and aggregate formation as separate processes during biofilm growth.

Biofilm formation in miniature flow reactors.

MultiRep reactors are miniaturized 12-channel biofilm flow reactors that permit simultaneous sampling of planktonic cultures and biofilms formed on removable Aclar discs that rest in wells in each channel (Fig. S2A). OG1RF biofilms from MultiRep reactors resemble the monolayer biofilms formed in CBRs (14, 16, 34). The same 6 Tn mutant cultures used for submerged Aclar disc assays in the previous section were inoculated into the MultiRep reactors at 10⁷ CFU/ml and grown with static incubation for 4h, after which medium flowed through each channel at a rate of 0.1ml/min for 20h (t_total = 24h). The flow rate for growth medium was chosen for consistency in turnover rate compared to CBR experiments. Planktonic and biofilm cultures were quantified and visualized at 4h and 24h. After 4h of growth in TSB-D, planktonic cultures of tig-Tn, OG1RF_10576-Tn, and OG1RF_11456-Tn strains had significantly reduced numbers of CFU/ml relative to OG1RF (Fig. 4A, pink bars). The Δbph negative-control strain had significantly reduced numbers of biofilm CFU/ml relative to OG1RF, as did the prsA-Tn, OG1RF_10576-Tn, and OG1RF_11456-Tn mutants (Fig. 4A, green bars). However, only the Δbph strain had a biofilm-specific reduction in growth relative to OG1RF at 4h (Fig. 4B). The tig-Tn mutant, which had increased biofilm formation in microtiter plate assays (Fig. 2D), had a biofilm-specific 1.89-fold ml relative to OG1RF/ml relative to OG1RF (Fig. 4B).

Open in a separate window

FIG 4

Biofilm formation of selected Tn mutants grown in MultiRep reactors in TSB-D. (A) Numbers of CFU/ml of strains at 0h, 4h, and 24h. The dotted lines indicate the numbers of OG1RF biofilm CFU/ml at 24h (top line) and 4h (bottom line). (B) Ratio of biofilm growth to planktonic growth at 4h relative to that of OG1RF. (C) Representative microscopy images of Hoechst 33342-stained biofilms at 4h. (D) Ratio of biofilm growth to planktonic growth at 24h relative to that of OG1RF. (E) Representative microscopy images of Hoechst 33342-stained biofilms at 24h. For panel A, each data point represents the average of two technical replicates, and a total of four biological replicates were performed. For panels B and D, data points were derived using the data points shown in panel A. Statistical significance was evaluated by one-way ANOVA with Dunnett’s multiple-comparison test (*, P <0.05; **, P <0.01; ***, P <0.001; ****, P <0.0001). For panels C and E, samples were grown in parallel to cultures used to generate panel A. Scale bars,20μm. Two technical replicates were processed for each biological replicate, and representative images are shown.

FIG S2

MultiRep biofilm reactors and analysis of OG1RF biofilms grown under multiple experimental conditions. (A) Photograph showing an assembled MultiRep biofilm reactor. Bottles with sterile growth medium are shown on the left, and outflow tubes with waste containers are shown on the right. (B) Additional fluorescence microscopy images of OG1RF biofilms obtained during biological replicates of experiments shown in Fig. 3, ​,4,4, and ​and5.5. Scale bars,20 μm. (C to E) Images of OG1RF biofilms were used for Comstat2 analysis of overall biomass (C), average biofilm thickness (D), and maximum biofilm thickness (E). Statistical significance was evaluated by two-way ANOVA with Tukey’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001) Download FIG S2, TIF file, 1.5 MB.

Open in a separate window

FIG 5

Biofilm formation of selected Tn mutants grown in MultiRep reactors in MM9-YEG. (A) Numbers of CFU/ml of strains at 0h, 4h, and 24h. The dotted lines indicate numbers of OG1RF biofilm CFU/ml at 24h (top line) and 4h (bottom line). (B) Ratio of biofilm growth to planktonic growth at 4h relative to that of OG1RF. (C) Representative microscopy images of Hoechst 33342-stained biofilms at 4h. (D) Ratio of biofilm growth to planktonic growth at 24h relative to that of OG1RF. (E) Representative microscopy images of Hoechst 33342-stained biofilms at 24h. For panel A, each data point represents the average of two technical replicates, and a total of four biological replicates were performed. For panels B and D, data points were derived using the data points shown in panel A. Statistical significance was evaluated by one-way ANOVA with Dunnett’s multiple-comparison test. For panels C and E, samples were grown in parallel to cultures used to generate panel A. Scale bars,20μm. Two technical replicates were processed for each biological replicate, and representative images are shown.

Copyright © 2021 Willett et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

Biofilm appearance was evaluated using fluorescence microscopy of Hoescht 33342-stained cells. After 4h, OG1RF formed biofilms with single cells and multicell chains but few large aggregates (Fig. 4C). Biofilms formed by Δbph and OG1RF_10576-Tn strains had very few cells, in agreement with the average reduction in biofilm CFU/ml at 4h. OG1RF_10350-Tn and tig-Tn mutants formed biofilms with chained cells and small clumps, and prsA-Tn and OG1RF_11456-Tn mutants formed biofilms with larger clumps of cells. The OG1RF_11288-Tn mutant formed biofilms that resembled those of OG1RF.

After 24h, no mutants had reduced numbers of planktonic CFU/ml relative to OG1RF (Fig. 4A, dark purple bars). Although the number of biofilm CFU/ml of the prsA-Tn mutant was ~1 log lower than that of OG1RF (Fig. 4A, lilac bars), this difference was not statistically significant. However, the prsA-Tn mutant had a significant reduction in the ratio of biofilm cells to planktonic cells relative to OG1RF (Fig. 4D). In contrast to biofilm morphology at 4h, OG1RF biofilms at 24h appeared as smooth layers of single cells, and chaining and clumping were not evident (Fig. 4E; Fig. S2B). Unlike 4-h biofilms formed by Δbph and OG1RF_10576-Tn strains, biofilms after 24h of growth covered most of the Aclar surface. OG1RF_10350-Tn and OG1RF_11288-Tn biofilms resembled those of OG1RF, and small clumps of cells were visible in prsA-Tn, tig-Tn, and OG1RF_11456-Tn biofilms.

In MM9-YEG, no mutants had statistically different numbers of planktonic or biofilm CFU/ml compared to those of OG1RF after 4h of static growth (Fig. 5A, pink and green bars) or an additional 20h of growth under flow conditions (Fig. 5A, purple and lilac bars). We observed more variability in planktonic growth of each Tn mutant after 24h in MM9-YEG than in TSB-D. Accordingly, no strains had biofilm-specific decreases in CFU calculated as the ratio of biofilm growth to planktonic growth relative to OG1RF (Fig. 5B and ​andD).D). Despite the variability in CFU, morphological differences in biofilms were visible. After 4h, OG1RF biofilms grew as single cells with small clumps (Fig. 5C). OG1RF_10350-Tn biofilms had fewer individual cells and more small chains than OG1RF. Reduced surface coverage was observed in Δbph and OG1RF_10576-Tn biofilms, and OG1RF_10576-Tn biofilms had long chains of cells relative to OG1RF. prsA-Tn, tig-Tn, and OG1RF_11456-Tn biofilms all had large aggregates of cells. After 24h, OG1RF formed dense, thick biofilms with visible cellular aggregates (Fig. 5E). Biofilms formed by Δbph, OG1RF_10350-Tn, OG1RF_11456-Tn, OG1RF_11288-Tn strains had some small aggregates. prsA-Tn and tig-Tn biofilms had sparse surface coverage with large clusters of cells, and OG1RF_10576-Tn formed biofilms with large aggregates.

Comparative measurements of biofilm growth of OG1RF in different growth assays.

Because we observed differences in biofilm morphology depending on growth medium, we used Comstat2 (41) to quantify the biomass and thickness of the parental strain by using submerged Aclar disc (6-h) and MultiRep reactor (4-h and 24-h) assays. In general, biofilms grown in MM9-YEG contained more individual cells, whereas biofilms grown in TSB-D had more multicellular chains (Fig. 3C and ​andF,F, ​,4C4C and ​andE,E, and ​and5C5C and ​andE;E; Fig. S2B). In TSB-D, biomass was not significantly different between submerged Aclar and MultiRep biofilms, and biomass of submerged Aclar or 4-h MultiRep biofilms grown in TSB-D was not significantly different from those grown in MM9-YEG (Fig. S2C). However, the biomass of 24-h MultiRep biofilms grown in MM9-YEG was 5.3-fold greater than that of biofilms grown in TSB-D (Fig. S2C). In MM9-YEG, 24-h MultiRep biofilms also had more biomass than 6-h submerged Aclar biofilms (3.45-fold higher) and 4-h MultiRep biofilms (13.0-fold higher) (Fig. S2C).

We next measured biofilm thickness. Biofilms grown on submerged Aclar discs for 6h or the MultiRep reactor for 4h had similar average thicknesses regardless of growth medium (Fig. S2D). However, biofilms grown in the MultiRep reactor for 24h in MM9-YEG had an average thickness of 23.3μm, which is 4.06-fold higher than the average thickness of biofilms grown in TSB-D (5.74μm) and also significantly higher than the other biofilms grown in MM9-YEG (Fig. S2D). All biofilms grown in TSB-D had approximately the same maximum thickness (Fig. S2E). However, 24-h MultiRep biofilms grown in MM9-YEG had a maximum thickness of 27.7μm, which is ~2-fold more than the other MM9-YEG biofilms and ~2.5-fold greater than biofilms grown in TSB-D. Taken together, these measurements show that extended cultivation of OG1RF biofilms in MM9-YEG under flow conditions results in thicker biofilms with more biomass than that in TSB-D, which correlates with the qualitative assessment of biofilm morphology observed using fluorescence microscopy. However, it is currently unknown whether this increase is due solely to the presence of more biofilm cells or to changes in matrix production or composition.

Tn mutant competition against OG1RF in biofilm cocultures.

The 6 Tn mutants described above were originally identified using TnSeq to evaluate mutant abundance in a community. Therefore, we wanted to measure how the mutants competed in a coculture with parental OG1RF. In the data reported below, we used both enumeration on selective agar medium (Tn mutants are resistant to chloramphenicol) and fluorescence microscopy to analyze the results of cocultures. For enumeration, we replaced the Δbph negative control with the bph-Tn mutant, which has the same biofilm phenotype as the deletion strain (23). To differentially label strains for visualization, we transformed OG1RF with a plasmid expressing tdTomato from a strong constitutive promoter (pP₂₃::tdTomato) and each Tn mutant with a plasmid expressing P₂₃::GFP. Prior to coculture, we evaluated whether carriage of the tdTomato or green fluorescent protein (GFP) plasmids resulted in growth defects. Two mutants (OG1RF_11456-Tn and OG1RF_11288-Tn) were excluded from coculture experiments due to poor planktonic growth or unstable fluorescence. With the remaining 4 Tn mutants, we repeated the submerged Aclar disc experiments described above with cultures in which OG1RF was mixed with single Tn mutants. For all experiments, OG1RF pP₂₃::tdTomato was also cultured independently in addition to coculture with Tn mutants to ensure that expression of tdTomato did not negatively affect biofilm formation (Fig. 6A, ​,C,C, ​,E,E, and ​andG;G; Fig. 7A, ​,C,C, ​,E,E, and ​andG;G; Fig. 8A, ​,C,C, ​,E,E, and ​andGG).

Open in a separate window

FIG 6

Cocultures of OG1RF and Tn mutants using the submerged Aclar disc assay. (A) Numbers of CFU/ml of OG1RF grown in TSB-D at 0h and 6h. (B) Numbers of CFU/ml of OG1RF/Tn cocultures grown in TSB-D at 0h and 6h. The dotted line indicates the number of biofilm CFU/ml of OG1RF grown in monoculture (value taken from panel A). (C) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato biofilms grown in TSB-D at 6h. (D) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato/Tn mutant pP₂₃::GFP biofilms grown in TSB-D at 6h. (E) Numbers of CFU/ml of OG1RF grown in MM9-YEG at 0h and 6h. (F) Numbers of CFU/ml of OG1RF/Tn cocultures grown in MM9-YEG at 0h and 6h. The dotted line indicates the number of biofilm CFU/ml of OG1RF grown in monoculture (value taken from panel E). (G) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato biofilms grown in MM9-YEG at 6h. (H) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato/Tn mutant pP₂₃::GFP biofilms grown in MM9-YEG at 6h. For panels A, B, E, and F, each data point represents the average of two technical replicates, and a total of four biological replicates were performed. Statistical significance was evaluated by two-way ANOVA with Sidak’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). For panels C, D, G, and H, samples were grown in parallel to cultures used to generate panels A, B, E, and F. Scale bars,20μm. Two technical replicates were processed for each biological replicate, and representative images are shown.

Open in a separate window

FIG 7

Cocultures of OG1RF and Tn mutants in TSB-D in MultiRep reactors. (A) Numbers of CFU/ml of OG1RF at 0h and 4h. (B) Numbers of CFU/ml of OG1RF/Tn cocultures at 0h and 4h. The dotted line indicates the number of biofilm CFU/ml of OG1RF grown in monoculture (value taken from panel A). (C) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato biofilms at 4h. (D) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato/Tn mutant pP₂₃::GFP biofilms at 4h. (E) Numbers of CFU/ml of OG1RF at 0h and 24h. (F) Numbers of CFU/ml of OG1RF/Tn cocultures at 0h and 24h. The dotted line indicates the number of biofilm CFU/ml of OG1RF grown in monoculture (value taken from panel E). (G) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato biofilms at 24h. (H) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato/Tn mutant pP₂₃::GFP biofilms at 24h. For panels A, B, E, and F, each data point represents the average of two technical replicates, and a total of three biological replicates were performed. Statistical significance was evaluated by two-way ANOVA with Sidak’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). For panels C, D, G, and H, samples were grown in parallel to cultures used to generate panels A, B, E, and F. Scale bars,20μm. Two technical replicates were processed for each biological replicate, and representative images are shown.

Open in a separate window

FIG 8

Cocultures of OG1RF and Tn mutants in MM9-YEG in MultiRep reactors. (A) Numbers of CFU/ml of OG1RF at 0h and 4h. (B) Numbers of CFU/ml of OG1RF/Tn cocultures at 0h and 4h. The dotted line indicates the number of biofilm CFU/ml of OG1RF grown in monoculture (value taken from panel A). (C) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato biofilms at 4h. (D) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato/Tn mutant pP₂₃::GFP biofilms at 4h. (E) Numbers of CFU/ml of OG1RF at 0h and 24h. (F) Numbers of CFU/ml of OG1RF/Tn cocultures at 0h and 24h. The dotted line indicates the number of biofilm CFU/ml of OG1RF grown in monoculture (value taken from panel E). (G) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato biofilms at 24h. (H) Representative microscopy images of Hoechst 33342-stained OG1RF pP₂₃::tdTomato/Tn mutant pP₂₃::GFP biofilms at 24h. For panels A, B, E, and F, each data point represents the average of two technical replicates, and a total of three biological replicates were performed. Statistical significance was evaluated by two-way ANOVA with Sidak’s multiple-comparison test (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). For panels C, D, G, and H, samples were grown in parallel to cultures used to generate panels A, B, E, and F. Scale bars,20μm. Two technical replicates were processed for each biological replicate, and representative images are shown.

For submerged Aclar disc assays, we inoculated both strains at 10⁷ CFU/ml and quantified OG1RF and Tn mutants after 6h. In planktonic cultures grown in TSB-D, only the OG1RF-10576-Tn mutant had a significant difference in number of CFU/ml (~1 log decrease) relative to OG1RF in the same coculture (Fig. 6B). The number of biofilm CFU/ml of OG1RF_10576-Tn was also decreased to the same extent relative to OG1RF in coculture. Interestingly, the prsA-Tn mutant outgrew OG1RF in these coculture biofilms by ~1 log (Fig. 6B) and had a 4.23-fold increase in the ratio of biofilm CFU to planktonic CFU relative to OG1RF (Fig. S3C), suggesting that this mutant outcompeted OG1RF under these conditions. Cocultures were visualized by fluorescence microscopy (Fig. 6C and ​andD;D; Fig. S3A). bph-Tn/OG1RF biofilms had sparse surface coverage compared to that of OG1RF alone. The OG1RF_10350-Tn/OG1RF biofilm resembled biofilms formed by the individual strains grown in monoculture. In accordance with CFU quantification, the prsA-Tn/OG1RF biofilm had more prsA-Tn cells and small clumps than that of OG1RF. In contrast to tig-Tn monoculture biofilms, the tig-Tn mutant formed large clumps in coculture with OG1RF. OG1RF_10576-Tn biofilms had low surface coverage when cultured alone, yet this mutant formed large clumps when cocultured with OG1RF. Interestingly, these large clusters appeared to colocalize with patches of OG1RF cells (Fig. S3A).

In MM9-YEG, none of the mutants had significantly different numbers of planktonic or biofilm CFU/ml than OG1RF (Fig. 6F). Overall, the MM9-YEG biofilms had more surface coverage than the TSB-D biofilms (compare Fig. 6C and ​andDD and Fig. 6G and ​andH),H), and all strains had higher numbers of biofilm CFU/ml in MM9-YEG than in TSB-D. The bph-Tn/OG1RF and OG1RF_10350-Tn/OG1RF biofilms resembled those of the mutants and OG1RF grown individually (Fig. 6G and ​andH;H; Fig. S3B). However, prsA-Tn, tig-Tn, and OG1RF_10576-Tn biofilms contained fewer aggregates when cocultured with OG1RF than when grown individually. Additionally, biofilms from the tig-Tn mutant cocultured with OG1RF in MM9-YEG contained more individual tig-Tn cells (as opposed to multicellular chains) than when cocultured in TSB-D. The OG1RF_10576-Tn mutant formed clumps and chains with visibly less surface coverage than OG1RF (Fig. S3B) when cocultured with OG1RF in MM9-YEG, although there was no statistical difference between numbers of OG1RF_10576-Tn and OG1RF biofilm CFU/ml.

Biofilm formation of OG1RF and Tn mutant cocultures in miniature flow reactors.

Biofilm formation of cocultures was evaluated using the MultiRep biofilm flow chambers described above, and each strain was inoculated at 10⁷ CFU/ml. After 4h in TSB-D, there were no statistically significant differences in numbers of planktonic CFU/ml between OG1RF and any mutants, but OG1RF_10350-Tn/OG1RF biofilms contained 3.6-fold more OG1RF_10350-Tn CFU than OG1RF CFU (Fig. 7B). Visualization of biofilms revealed that OG1RF_10350 and tig-Tn mutants formed biofilms with aggregates containing both mutant and OG1RF cells (Fig. 7D; Fig. S4A). The prsA-Tn mutant formed large aggregates in coculture with OG1RF, but these aggregates contained relatively few OG1RF cells (Fig. 7D; Fig. S4A). bph-Tn/OG1RF and OG1RF_10576-Tn/OG1RF biofilms had less surface coverage than OG1RF grown alone (Fig. 7C and ​andD).D). After 24h of growth in TSB-D, there were no significant differences in numbers of coculture planktonic or biofilm CFU/ml (Fig. 7F). OG1RF pP₂₃::tdTomato and coculture biofilms grew as monolayers of mostly individual cells, with fewer multicellular aggregates and less chaining than observed after 4h (Fig. 7G and ​andH).H). Fewer bph-Tn and OG1RF_10576-Tn cells were present than OG1RF cells (Fig. 7H; Fig. S4B), although only the bph-Tn mutant had significantly reduced numbers of biofilm CFU/ml relative to OG1RF (Fig. 7F).

After 4h in MM9-YEG, there were no significant differences in numbers of planktonic or biofilm CFU/ml between OG1RF and Tn mutants in coculture (Fig. 8B). Very few mutant cells were visible in the bph-Tn/OG1RF and OG1RF_10576-Tn/OG1RF biofilms (Fig. 8D; Fig. S5A). OG1RF_10350/OG1RF biofilms had larger aggregates of cells than those grown in TSB-D for 4h. prsA-Tn/OG1RF and tig-Tn/OG1RF biofilms resembled those grown in TSB-D for 4h and contained large aggregates of cells. After 24h of growth in MM9-YEG, there were no significant differences between numbers of OG1RF or Tn mutant CFU/ml in planktonic or biofilm cultures (Fig. 8F). Coculture biofilms contained thick multicellular aggregates of both OG1RF and Tn mutants, with the exception of prsA-Tn coculture biofilms, which had fewer large aggregates (Fig. 8H; Fig. S5B). None of the Tn mutants had significant differences in the ratio of biofilm cells to planktonic cells relative to OG1RF at either 4h or 24h (Fig. S5C and D).

Cloning and Tn mutant verification.

Nucleotide sequences of primers are listed in Table S3. All restriction enzymes were purchased from New England Biolabs. For construction of the cCF10-inducible tig complementation vector, tig was amplified from purified OG1RF genomic DNA using Pfu Ultra II polymerase (Agilent), digested with BamHI-HF/NheI-HF, and ligated to pCIEtm (23) treated with the same restriction enzymes. The plasmid construct was verified by Sanger sequencing (Eurofins). For generation of constitutive fluorescent protein constructs, P₂₃ was excised from pDL278p23 (57) by digestion with EcoRI-HF/BamHI-HF and ligated to pTCV-LacSpec digested with the same restriction enzymes. A fragment encoding promoterless GFP (58) flanked by BamHI and BlpI sites was inserted to create pP₂₃::GFP, and the BamHI-SphI fragment from pJ201::187931 was inserted to create pP₂₃::tdTomato. The Tn insertions in strains used for submerged Aclar disc and MultiRep reactor experiments were verified by colony PCR using the oligonucleotides listed in Table S4. The Tn insertion adds ~2.1kb to the size of the wild-type allele.

CDC biofilm reactors.

Reactors were assembled as previously described (14, 16, 28) and incubated at 37°C overnight to ensure a lack of contamination. Polycarbonate (BioSurfaces Technologies Corp.) and Aclar (Electron Microscopy Sciences) discs were used as biofilm substrates (at least 12 discs per replicate). Immediately prior to inoculation, single-use Tn library aliquots were removed from storage at −80°C and thawed on ice. Growth medium (either MM9-YEG or TSB-D) was inoculated with 6×10⁸ to 2×10⁹ CFU. Batch cultures were grown without flow for 4 to 6h, after which the peristaltic pump (Cole Parmer) was turned on at a flow rate of 8ml/min for 18 to 20h (total experiment time,24h). In a previous study, this resulted in at least 10⁶ biofilm CFU/disc after 4 to 6h of static culture and 10⁸ biofilm CFU/disc after 24h (28), reducing the possibility of a bottleneck restricting growth of the Tn libraries. Two biological replicate reactors were run for each Tn library-growth medium combination.

DNA isolation, library preparation, and transposon sequencing.

Substrates were removed from the CDC biofilm reactor chamber and processed to remove adherent biofilm cells. Polycarbonate discs were aseptically removed and placed in 6-well plates (4 discs/well) containing 5ml distilled water and incubated for 5min at room temperature to remove nonadherent cells. To obtain attached biofilm cells, 12 discs were placed in 50-ml conical tubes containing 30ml KPBS (potassium phosphate-buffered saline, pH 7.0) with 2mM EDTA and vortexed in a BenchMixer multitube vortexer (Benchmark Scientific) at 2,000rpm for 5 min. Biofilms grown on Aclar membranes were rinsed in 50-ml conical tubes with 30ml KPBS, followed by inversion to remove nonadherent cells. Rinsed Aclar membranes were submerged in 4ml KPBS with 2mM EDTA, and biofilms were removed by scraping with a sterile razor blade. Biofilms from multiple substrates from each reactor were pooled in a conical tube, pelleted at 6,371×g for 15min, and frozen at –80°C until further use. Pellets were resuspended in 180μl enzymatic lysis buffer (20mM Tris-HCl [pH 8.0], 2mM EDTA, 1.2% Triton X-100) with 30mg/ml lysozyme and 500 U/ml mutanolysin. After 30min of incubation at 37°C, 25μl proteinase K and 200μl buffer AL (DNeasy blood and tissue kit; Qiagen) were added. Tubes were incubated at 55°C for 30min, after which DNA was extracted using a DNeasy blood and tissue kit by following the manufacturer’s instructions. Samples were submitted to the University of Minnesota Genomics Center for library preparation and sequencing. Sequencing libraries were prepared using the Illumina TruSeq nano library preparation kit as previously described (22). Libraries were sequenced as 125-bp paired-end reads on an Illumina HiSeq 2500 in high-output mode (440M reads total).

Sequencing reads were processed using a published workflow (22). Briefly, reads were trimmed and aligned to the OG1RF genome (NC_017316.1), and Tn insertions at TA sites were quantified. The statistical significance of the relative abundance of Tn reads at each TA site in biofilm compared to planktonic samples was evaluated using a chi-square test and an additional Monte Carlo-based method as an evaluation of the accuracy of the chi-square test. Scripts for all processing steps are publicly available (https://github.com/dunnylabumn/Ef_OG1RF_TnSeq). Output files were filtered for nucleotide positions of Tn mutants known to be present in the library based on previous sequencing (22). Log₂ fold changes were calculated from relative Tn abundances in biofilm compared to planktonic samples. Statistical significance was defined as a Pof <0.05 and a Monte Carlo simulation value of 1.119552, the lowest value obtained in these calculations. Venn diagrams showing overlap between data sets were generated with the VennDiagram package for R (59).

Tn mutants used for additional experiments were obtained from frozen library stock plates and grown on BHI/FA agar plates. Single colonies were picked and patched onto BHI/Erm to ensure loss of the plasmids used in Tn mutagenesis and BHI/Cm to confirm functionality of the Cm resistance gene located in the Tn. Single colonies were picked from BHI/Cm plates and grown in BHI/Cm/FA to generate freezer stocks. Tn insertions were verified by colony PCR using primers flanking the gene of interest (Table S3). The Tn adds ~2.1kb to the size of the parental allele (27, 60).

SEM.

Biofilms were removed from the CBRs and rinsed with KPBS three times and then processed for scanning electron microscopy (SEM) using the cationic dye stabilization methods described previously (14,–16, 34). Briefly, biofilms were subjected to primary fixation in sodium cacodylate buffer containing methanol-free EM-grade formaldehyde (2%), glutaraldehyde (2%), sucrose (4%), and Alcian blue 8GX (0.15%) overnight. Discs were then rinsed three times with sodium cacodylate buffer and subjected to secondary fixation in sodium cacodylate buffer containing 1% osmium tetroxide and 1.5% potassium ferrocyanide for 1h. Fixed samples were rinsed three times with sodium cacodylate buffer and chemically dried using a graded ethanol series, processed in a CO₂-based critical-point dryer (Tousimis, Rockville, MD), and sputter coated with ~2nm iridium (EM ACE600; Leica, Buffalo Grove, IL). Imaging was done using a Hitachi SU8230 field emission instrument at 0.8kV using the low-angle backscatter and secondary electron detectors.

Biofilm assays.

Ninety-six-well plate biofilm assays were carried out as described previously (23, 26, 30). Overnight cultures for complementation assays were grown with 5μg/ml tetracycline and 25ng/ml cCF10, and experiments were performed in the indicated growth medium supplemented with 25ng/ml cCF10. Briefly, overnight cultures were diluted 1:100 in the appropriate growth medium, and 100μl was added to a 96-well plate (Corning 3935). For the secondary screens using 43 Tn mutants, two technical replicates were performed for each strain. For complementation assays, three technical replicates were performed for each strain. For all experiments, values shown are the results of three independent biological replicates. Plates were incubated in a humidified plastic container at 37°C for the indicated length of time. Cell growth was measured in a Biotek Synergy HT plate reader as the absorbance at 600nm (A₆₀₀). Plates were gently washed three times with ultrapure water using a BioTek plate washer, dried in a biosafety cabinet or on a lab bench overnight, and stained with 100μl 0.1% safranin (Sigma). Stained plates were washed three times and dried. The A₄₅₀ was measured to quantify safranin-stained biofilm biomass. Biofilm production was evaluated as the ratio of stained biofilm biomass to overall growth (A₄₅₀/A₆₀₀), and values were normalized to the biofilm production of OG1RF.

For submerged Aclar biofilm assays, overnight cultures were adjusted to 10⁷ CFU/ml in the appropriate growth medium, and 1ml was added to 1 well of a 24-well plate (Costar 3524) with a 5-mm Aclar disc. Plates were incubated at 37°C in a plastic container on a tabletop shaker (Thermo Scientific MaxQ 2000) at 100rpm. After 6h, planktonic cells were transferred to microcentrifuge tubes. Aclar discs were washed by gentle shaking in KPBS and transferred to microcentrifuge tubes with 1ml KPBS (1 Aclar disc/tube). Tubes with planktonic cultures and Aclar discs were vortexed at 2,500rpm for 5min in a BenchMixer multi-tube vortexer (Benchmark Scientific) and then diluted (10-fold serial dilutions) in KPBS and plated on BHI/FA medium to enumerate colonies. For coculture experiments, diluted cultures were plated on BHI/FA (total CFU counts) and BHI/Cm plates (Tn mutant CFU counts). CFU/ml values for OG1RF in coculture were obtained by subtracting the CFU/ml counts from BHI/Cm plates from the CFU/ml counts from BHI/FA plates. At least three biological replicates (each with two technical replicates) were performed for all strains.

MultiRep biofilm reactors (Stratix Labs, Maple Grove, MN) were loaded with 5-mm Aclar discs (6 Aclar discs per channel). Influx (MasterFlex HV-96117-13) and efflux (MasterFlex EW-06424-16) tubing was attached to each channel and capped with foil prior to autoclaving. The 10% growth medium was autoclaved in a separate bottle with sterile connecting tubing and attached to the influx reactor tubing immediately prior to inoculation. Overnight cultures were diluted to 1×10⁷ CFU/ml, and 4ml was added to each channel (1 channel per strain). The reactor was sealed by placing 2 silicon sheets in the lid and clamping the lid on the reactor using Irwin Quick-Grip ratcheting bar clamps. The influx tubing was connected to peristaltic pumps (MasterFlex 77202-60), and the efflux tubing was placed horizontally over waste containers. Reactors were kept at 37°C with static incubation for 4h, after which the pumps were turned on at a flow rate of 0.1ml/min for 20h. For disassembly and sample processing, the reactor lids were removed, and 2ml planktonic culture was transferred to microcentrifuge tubes. Aclar discs were removed and rinsed in KPBS and then placed in microcentrifuge tubes with 1ml KPBS. Tubes with planktonic cultures or Aclar discs were vortexed and diluted, and the numbers of CFU were enumerated as described above.

Fluorescence microscopy.

For all experiments, Aclar discs (2 per strain) were rinsed 3 times in KPBS and stained for 15min in Hanks’ balanced salt solution with CaCl₂ and MgCl₂ (Gibco) and 5μg/ml Hoechst 33342 (Molecular Probes) with gentle agitation. After staining, Aclar discs were washed 3 times in fresh KPBS, transferred to a 48-well plate (Costar 3548) with 1ml 10% buffered formalin (Fisher Scientific) with gentle agitation, and shielded from light for 12 to 16h. After fixing, Aclar discs were washed in KPBS and mounted on a Superfrost Plus microscope slide (Fisher Scientific) in a 0.24-mm double-sided adhesive Secure-Seal spacer (Grace BioLabs) with a 7-mm hole punched to accommodate the Aclar disc. Aclar discs were covered with 7μl ProLong glass antifade mountant (Invitrogen) and a Gold Seal cover slip (no. 1.5; Fisher Scientific). Slides were cured at room temperature shielded from light for 4 to 8h and stored at 4°C until imaging.

Microscopy and image processing.

All images were acquired on a Zeiss Axio Imager M1 widefield microscope with a Plan-APO 20× (0.8 numerical aperture [NA]) using an X-Cite 120 metal halide light source (EXFO, Inc.) illuminating 365-nm, 470-nm, and 550-nm excitation filters for Hoechst 33342, GFP, and tdTomato, respectively. Images were captured using the Zeiss AxioCam 503 mono microscope camera and Zen imaging software (v2.1; Zeiss). For each Aclar disc, two independent images were obtained, yielding four images per sample from which a final representative image was chosen. Representative images were processed using the Fiji ImageJ package (version 1.48; NIH) and subjected to background subtraction with a rolling-ball radius of 50 pixels using the internal ImageJ function as well as uniformly applied brightness and contrast adjustments of the entire image prior to cropping (61). For biofilm coculture images, the Hoechst, GFP, and tdTomato images were false colored cyan, yellow, and magenta, respectively, using Fiji. For coculture images, tdTomato (OG1RF) and GFP (Tn mutant) maximum intensity projections (MIPs) were processed independently and merged. Images were cropped to 500by500 pixels using GIMP (v 2.0) and exported as PNG files. The GFP (mutant) and tdTomato (OG1RF) MIPs were processed independently and merged.

Biofilm thickness and distribution were analyzed using Comstat2. Cells were imaged using an Axio Observer.Z1 confocal microscope equipped with an LSM 800-based Airyscan system in normal confocal mode (Zeiss). Confocal images were acquired with a 20by0.8NA objective and 405-nm lasers for excitation of Hoechst 33342 stain. For image analysis, two representative z stacks were taken per Aclar disc with a 1-μm interval. Each experiment used three independent biological replicates with at least 2 Aclar discs in each. The maximum thickness of the biofilms was determined from the Hoechst channel using the Comstat2.1 plugin for ImageJ (41, 62). All image processing adheres to the standards outlined by Rossner and Yamada (63).

Gelatinase assays.

Overnight cultures were grown in the respective growth medium and spotted onto TSB-D agar plates supplemented with 3% (wt/vol) gelatin. Plates were incubated overnight at 37°C and then moved to 4°C for 1 to 3h prior to imaging. Plate photos were obtained using a ProteinSimple (Cell Biosciences) FluorChem FC3 imager. Strains were considered gelatinase positive if they developed a halo around colony growth and gelatinase negative if no halo was present.

Bioinformatic analysis.

Functional annotations of proteins were obtained from KEGG and NCBI. Protein sequences were obtained from NCBI and used as input for Phyre2 in intensive mode (45). Transmembrane predictions were obtained with TMHMM (64). Additional protein structure files were downloaded from the Protein Data Bank (PDB), and structures were rendered in PyMOL 2.1 (65).

Statistical analysis.

All statistical analysis was carried out using GraphPad Prism (version 9.0.1). Statistical tests and significance are described in the figure legends. Corrections for multiple comparisons were performed using the test recommended by GraphPad.

We thank the University of Minnesota Genomics Center for assistance with TnSeq and the Minnesota Supercomputing Institute (MSI) at the University of Minnesota for providing computational resources. We thank Elizabeth Cameron for providing the pCIEtm::tig plasmid and Dawn Manias for assistance with constructing pP₂₃::GFP and pP₂₃::tdTomato.

Parts of this work were carried out in the Characterization Facility, University of Minnesota, which receives partial support from the NSF through the MRSEC (award no. DMR-2011401) and the NNCI (award no. ECCS-2025124) programs. This work was supported by grant no. 1RO1AI122742 to G.M.D. from the NIH. J.L.E.W. was supported by American Heart Association grant no. 19POST34450124/Julia Willett/2018. M.L.K. was supported by grant no. TL1R002493 and UL1TR002494 from the NIH’s National Center for Advancing Translational Sciences. A.M.T.B. received support via NIH training grant no. AI055433 for portions of this work.