The de novo Purine Biosynthesis Pathway Is the Only Commonly Regulated Cellular Pathway during Biofilm Formation in TSB-Based Medium in Staphylococcus aureus and Enterococcus faecalis

ABSTRACT

Bacterial biofilms are involved in chronic infections and confer 10 to 1,000 times more resistance to antibiotics compared with planktonic growth, leading to complications and treatment failure. When transitioning from a planktonic lifestyle to biofilms, some Gram-positive bacteria are likely to modulate several cellular pathways, including central carbon metabolism, biosynthesis pathways, and production of secondary metabolites. These metabolic adaptations might play a crucial role in biofilm formation by Gram-positive pathogens such as Staphylococcus aureus and Enterococcus faecalis. Here, we performed a transcriptomic approach to identify cellular pathways that might be similarly regulated during biofilm formation in these bacteria. Different strains and biofilm-inducing media were used to identify a set of regulated genes that are common and independent of the environment or accessory genomes analyzed. Our approach highlighted that the de novo purine biosynthesis pathway was upregulated in biofilms of both species when using a tryptone soy broth-based medium but not so when a brain heart infusion-based medium was used. We did not identify other pathways commonly regulated between both pathogens. Gene deletions and usage of a drug targeting a key enzyme showed the importance of this pathway in biofilm formation of S. aureus. The importance of the de novo purine biosynthesis pathway might reflect an important need for purine during biofilm establishment, and thus could constitute a promising drug target.

IMPORTANCE Biofilms are often involved in nosocomial infections and can cause serious chronic infections if not treated properly. Current anti-biofilm strategies rely on antibiotic usage, but they have a limited impact because of the biofilm intrinsic tolerance to drugs. Metabolism remodeling likely plays a central role during biofilm formation. Using comparative transcriptomics of different strains of Staphylococcus aureus and Enterococcus faecalis, we determined that almost all cellular adaptations are not shared between strains and species. Interestingly, we observed that the de novo purine biosynthesis pathway was upregulated during biofilm formation by both species in a specific medium. The requirement for purine could constitute an interesting new anti-biofilm target with a wide spectrum that could also prevent resistance evolution. These results are also relevant to a better understanding of the physiology of biofilm formation.

INTRODUCTION

Biofilms are ubiquitous in the biosphere since approximately 80% of all bacteria adopt this lifestyle (1). These structures are formed by multicellular communities that adhere to a surface or an interface and are embedded in a matrix of extracellular polymeric substances (2, 3). The intrinsic properties of the matrixes confer them the ability to adapt to their environment and resist hostile conditions (2). Biofilms are known to cause significant problems in different aspects of human activity such as the food industry and the industrial and clinical sectors (4–7).

Biofilms are involved in a high number of nosocomial infections, approximately two-third of which are associated with some type of implanted medical devices (8). In hospitals, patients with biomedical implants such as prostheses are particularly at risk of developing severe complications related to biofilm-mediated infection during or after their surgery. These complications increase the morbidity and mortality rates of patients, as well as the duration and costs of hospitalization (9, 10). The most common pathogens isolated from those types of infection are Staphylococcus aureus, followed by Enterococcus spp., in approximately 36% and 19% of cases respectively (11, 12). Intrinsic properties of their biofilms provide 10 to 1,000 times more tolerance to antibiotics than the MIC required to kill their planktonic counterparts (10, 13–16). Therefore, biomedical catheter and implant infections caused by biofilm can be particularly difficult to treat because they are often unresponsive to antibiotic lock therapy, the currently recommended clinical practice to prevent and treat catheter-related infections (17).

S. aureus is a Gram-positive bacterium that naturally colonizes up to 60% of humans, but it can also become an opportunistic pathogen, causing a wide range of infections acquired in hospital or community environment (18). S. aureus biofilms often contain a matrix mainly composed of the polysaccharide intercellular adhesin (PIA), an important component of cell-cell adhesion (19). PIA production is regulated by the icaADBC operon (20) and is induced by high concentration of glucose and salt in the medium (21, 22). However, although most S. aureus strains possess the ica operon, some of them produce a PIA-independent biofilm (22). In these biofilms, the matrix is mostly composed of eDNA, MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) and other large proteins that form fibrils for intercellular adhesion (23). Methicillin-resistant S. aureus and methicillin-sensitive S. aureus are more likely to produce PIA-independent and PIA-dependent biofilms, respectively (24, 25). Although S. aureus biofilms are well studied, some properties of PIA-dependent and PIA-independent biofilms are still poorly understood (26).

Enterococcus faecalis is a Gram-positive bacterium primarily found as a commensal organism that colonizes in low abundance in our gastrointestinal tract (27). In hospitals, E. faecalis can become an opportunistic pathogen causing several potentially lethal infections in susceptible patients (28). E. faecalis possesses a high level of tolerance toward environmental stress and antibiotic treatments (29–32), probably due to its ability to form a biofilm (33, 34). E. faecalis biofilm matrix is mostly composed of eDNA, polysaccharides, lipoteichoic acid, and extracellular proteases (35), but the underlying mechanisms leading to its formation remain to be clearly established.

Recent studies revealed a vast remodeling of metabolic pathways during the early phase of biofilm formation in Bacillus subtilis and Bacillus cereus, including modulation in fermentation processes, energy production, primary and secondary metabolism pathways (e.g., fatty acid, carbon, amino acid, and nucleotides metabolism) (36, 37). It is only recently that these pathways have been found to be involved in biofilm formation. In S. aureus, it has been reported that metabolic pathways such as the TCA cycle (38) and respiration (39) are positively or negatively regulated during biofilm formation. Diverse genes involved in metabolism were also transcriptionally modulated during biofilm formation in E. faecalis (40, 41). Identifying and understanding the cellular adaptation that bacteria need to perform when forming a biofilm could lead to the identification of new therapeutic targets to inhibit this process. Such targets would be of great help in designing therapeutic strategies to treat various chronic infections, to improve the effectiveness of current drugs and limit the development of tolerance.

Here, we used a transcriptomic approach to determine if specific cellular pathways were similarly regulated during biofilm formation in S. aureus and E. faecalis. Analysis of differently regulated genes in two strains of each species, and in two different media, allowed us to determine which metabolic pathways were consistently up- or downregulated during biofilm formation and whether the transcriptional regulation was conserved. Only the de novo purine biosynthesis pathway was commonly upregulated in all strains in one medium, and we validated the importance of this finding in S. aureus using gene disruption and drug treatment approaches.

RESULTS

Determination of the growth conditions for biofilm harvest.

Biofilms formed by various strains of S. aureus and E. faecalis can differ in their compositions in a natural environment or in vitro (35, 42). Consequently, to ensure that our assay includes commonly regulated cellular pathways in a strain-independent manner, we analyzed two strains of each pathogen. For S. aureus, we used the Methicillin-resistant S. aureus strain USA300 and the methicillin-sensitive S. aureus strain SH1000, which are phylogenetically apart (43). For E. faecalis, we selected two strains with robust biofilm-forming capacity; the vancomycin-resistant V583, which possess a plethora of mobile elements including three plasmids (44), and the widely referenced vancomycin-sensitive strain ATCC 29212 (45). A core genome analysis of these two E. faecalis strains using PATRIC Proteome Comparison Service (46) revealed that they share an average of 95.4% of protein sequence identity between their coding DNA sequence and 1082 conserved genes with 100% of sequence identity. Since only 36.3% of ATCC 29212 genes were perfectly conserved, it appears that both E. faecalis strains differ significantly, making them good candidates to identify genes whose regulation is well conserved. S. aureus strains share an average of 98.8% sequence identity and 2250 conserved genes with 100% of sequence identity (see Fig. S1A and Table S5 in the supplemental material). Interspecies comparisons showed that the S. aureus and E. faecalis strains have an average of 43.3% of sequence identity between their coding DNA sequence (see Fig. S1A and Table S5). The 455 genes with at least 50% homology between the four strains used in this study are involved in different metabolic pathways, such as amino acid, carbon, fatty acid or nucleotide metabolism (Fig. S1B). Conservation of these pathways between the two species is not surprising, considering that many metabolic subsystems are often conserved in prokaryotes (47).

We analyzed mRNA expression levels during the early stage of biofilm development, which coincides with the most pronounced biofilm cellular adaptations in other Gram-positive species (36, 37). Targeting genes involved in these early stages of cellular adaptation might lead to a more efficient disruption of biofilm formation than targeting genes involved in later stages because matrix production would have already begun in the latter case. However, because early biofilm might be reached at different times by the different strains, we monitored biofilm formation using crystal violet to determine the best time point to harvest the mRNAs. As biofilms of S. aureus and E. faecalis reached their maximum development after 9 h and 7 h, respectively (Fig. 1), cells from both strains were harvested at the beginning of this plateau. In parallel, we monitored planktonic cell growth (OD₆₀₀) under the same conditions but with agitation to harvest the cells just as they reached their growth plateau. These procedures demonstrated that in both conditions, growth dynamics were similar, and it ensured that the comparison of planktonic and biofilm cells were done at similar growth stages. Of note, we used two different media, tryptone soy broth (TSB) and brain heart infusion (BHI) to potentially influence biofilm matrix composition (48–50). Both media were supplemented with 0.5% glucose (TSBg and BHIg) to grow S. aureus USA300 and E. faecalis biofilms, and with 0.5% glucose and 3% sodium chloride to grow S. aureus SH1000 (TSBgs and BHIgs). Growth dynamic in BHIg(s) and TSBg(s) were comparable, except for S. aureus USA300, which formed a more abundant biofilm in BHIg.

FIG 1.

Growth dynamic of S. aureus and E. faecalis strains in a planktonic state or during biofilm formation. Planktonic growth (in blue) was performed at 37°C in shaking condition and measured by OD₆₀₀, while biofilm growth (in red) was performed at 37°C without shaking and measured by crystal violet assay (OD₅₉₀). Growth was observed in either BHIg(s) (circle) or TSBg(s) (square). Each point represents the mean of at least three biological replicates ± standard deviation (SD). RNA harvesting time is indicated by a gray arrow.

Transcriptomic analysis reveals global changes in gene expression during biofilm formation.

To determine changes in gene expression levels, we used an RNA-seq approach on cells harvested from planktonic and biofilm conditions from biological triplicates of each experimental condition. A principal-component analysis of differential expression showed that biological replicates mostly clustered together (see Fig. S2 in the supplemental material). Genes were considered differentially expressed in biofilm versus planktonic when showing a statistical significance (adjusted P value < 0.05) and a fold change (log₂ ratio) ≥ |2.0|. All media combined, 466 unique differentially expressed genes (DEGs; ∼17.3% of the genome) were found for S. aureus USA300, 280 unique DEGs (∼9.6% of the genome) for S. aureus SH1000, 242 unique DEGs (∼7.7% of the genome) for E. faecalis V583 and 461 unique DEGs (∼16.4% of the genome) for E. faecalis ATCC 29212 (see Table S4 and Gene Expression Omnibus database under the accession number GSE162709). Upregulated (log₂ fold change ≥ 2.0) and downregulated (log₂ fold change ≤ −2.0) DEGs distribution among S. aureus and E. faecalis strains and media are displayed in the Volcano plots shown in Fig. 2.

FIG 2.

Transcriptional profiles of biofilm-grown cells and planktonic cells. The DEGs (blue dots) adjusted P value cutoff was 0.05. Genes were considered upregulated in biofilm if the log₂ ratio (fold change) was ≥ 2.0 and downregulated in biofilm if the fold change was ≤ −2.0. DEGs from the media effect transcriptome analysis (BHIg[s] vs TSBg[s]) were then compared with the DEGs from the biofilm transcriptome analysis. Genes found in both analyzes are represented by orange dots and represent DEGs possibly emerging due to an effect of medium instead of biofilm formation. (A) S. aureus USA300 and S. aureus SH1000 DEGs in TSBg(s) or BHIg(s). (B) E. faecalis V583 and E. faecalis 29212 DEGs in TSBg or BHIg.

Media effect on gene expression.

Metabolic pathways have been previously shown to be involved in biofilm formation in several Gram-positive bacteria (36, 37, 51), but these pathways can also be modulated by nutrients in the growth medium. Thus, we examined the influence of two different biofilm growth media on gene expression patterns to determine if the regulation we observed was specific to the medium or to biofilm formation. To do that, we determined genes that were differentially expressed in TSBg(s) compared to BHIg(s), all growth condition combined. For S. aureus, 75 genes of USA300 and 149 genes of SH1000 had their expression significantly modulated by the two media. Of these, a total of 9 and 18 genes were also found differentially expressed in biofilms for USA300 and SH1000, respectively (Fig. 2). For E. faecalis strains, we observed that only 11 genes of V583 and 31 genes of ATCC 29212 were differentially expressed in one or the other media, all growth conditions combined. Consequently, the growth medium used had a small effect on the gene expression pattern of E. faecalis strains during biofilm formation, with less than 1% of all DEGs being also differentially expressed when comparing the transcriptomes obtained from different growth media. Altogether, our results suggest that the medium has a minor influence on gene expression during biofilm formation, but this difference needs to be considered.

Gene set enrichment analysis reveals global cellular adaptations during biofilm formation.

We compared biofilm and planktonic gene expression patterns from the four strains in TSBg(s) or BHIg(s) using DAVID Bioinformatic resources (52). Since S. aureus USA300 and E. faecalis ATCC 29212 locus tags were not recognized by DAVID, we attributed the corresponding recognized locus tag of their species counterpart based on sequence homology and orthologs with the help of BioCyc Pathway tools version 23.0 (53) and AureoWiki (54). Genes up- or downregulated in the heatmap of S. aureus and E. faecalis represented 497 and 316 Gene Ontology:biological process (GO:BP) terms, respectively (Fig. 3A and B). The 20 most up- and downregulated GO:BP terms for each species, all strains and media combined, are shown in Panel C of Fig. 3 as a portrait of species-specific global regulation. The same analysis was performed with the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways as the annotation source showed that 103 KEGG functions of S. aureus and 97 KEGG functions of E. faecalis were represented in up- or downregulated genes (Fig. S3 in the supplemental material). Many metabolism-related pathways were upregulated during biofilm formation, all annotation sources combined, such as carbon, fatty acid, nucleotides and some amino-acid metabolism genes. However, many of these pathways appeared differently regulated according to the growth medium, as shown in Fig. S4, which depicts the most up- and downregulated GO:BP terms by species and by medium.

FIG 3.

Average relative expression of GO:BP terms during biofilm formation. Each row of both heatmaps represents a single functional annotation term from DAVID Bioinformatic software using GO:BP annotation source. A term’s relative expression is the average of the log₂ fold change of every associated gene with an adjusted P value cutoff was 0.05. Genes were considered upregulated in biofilm if the log₂ ratio (fold change) was ≥ 2.0 and downregulated in biofilm if the fold change was ≤ −2.0. DEGs from the media effect transcriptome analysis (BHIg[s] vs TSBg[s]) were then compared to the DEGs from the biofilm transcriptome analysis. Genes found in both analyzes are represented by orange dots and represent DEGs possibly emerging due to an effect of medium instead of biofilm formation. (A) S. aureus USA300 and SH1000, in BHIg(s) or TSBg(s) medium. (B) Functional annotation terms’ (total = 316) relative expression in E. faecalis V583 and 29212, in BHIg or TSBg medium. (C) Table of the top 20 upregulated and downregulated GO:BP terms for each species, by calculating the average of the relative expression of each term for the 2 strains in each medium.

Absence of common cellular adaptation between S. aureus and E. faecalis strains during biofilm formation.

The differences in GO:BP terms up- and downregulated in S. aureus and E. faecalis suggested that genes differently regulated during biofilm formation are very different between both species. To confirm this observation, we compiled all the genes showing a statistically significative (adjusted P value < 0.05) fold change of (log₂ ratio) ≥ |1.0| in both media and for both strains of each species (Table S3 in the supplemental material). For S. aureus, 26 genes were consistently up- or downregulated in all conditions tested. Interestingly, genes encoding for urease accessory proteins were significantly upregulated in USA300 and downregulated in SH1000. Previous studies have reported that urease genes were upregulated in mature biofilms formed by UAMS-1 and SA113 (20, 55). We also observed strong upregulation of the formate dehydrogenase gene fdh and of the DNA-binding protein HU gene hup in both media, similarly to what was observed previously (55). Our observations correlate with a recent study showing that S. aureus gene expression within the biofilm is largely strain-dependent (56). For E. faecalis a total of 50 genes, including many transporters, were either up- or downregulated. Several of these genes were also previously reported as being modulated in biofilm, such as the downregulation of nusA and of the cation-transporting ATPase EF0871 (40), and the upregulation of the S-adenosylmethionine synthetase gene metK and of the quorum sensing gene luxS (57). Importantly, exhaustive examination of the significantly regulated genes in biofilms in all conditions revealed no pathways for which gene regulation was common to both species.

Purine biosynthesis remodeling during biofilm formation.

Bacteria can metabolize purines de novo from 5′-phosphoribosylamine-1-pyrophosphate through biosynthesis of IMP (IMP), or they can recycle purines from the nucleic acids present in their environment via the purine salvage pathway (58). Interestingly, GO:BP terms related to purine biosynthesis were among the 20 most upregulated term between both species when the two media were examined together (Fig. 3B). This metabolic pathway, with each gene involved in these reactions, is depicted in Fig. 4. In adjacent boxes we show their expression level in biofilm (log₂ fold change) in TSBg(s) (dark outline) and in BHIg(s) (thin outline). Overall, genes in the de novo purine biosynthesis pathway were homogeneously strongly upregulated (log₂ fold change ≥ 2.0) in all S. aureus and E. faecalis strains in TSBg(s), and also in BHIg for E. faecalis V583. Only purB, found also in the purine salvage pathway, did not show a strong upregulation. Interestingly, we did not find significant and conserved regulation of the purine biosynthesis regulator PurR, suggesting that another mechanism is involved here. Gene expression levels of the purine salvage pathway were heterogeneously and weakly regulated.

FIG 4.

The de novo purine biosynthesis pathway is strongly upregulated during biofilm formation in TSBg(s). The purine biosynthesis pathway includes the de novo purine biosynthesis pathway (in blue) and the purine salvage pathway (in orange). Plain arrows indicate a unique reaction while dotted arrows indicate multiple reactions. The expression level (log₂ fold change) during biofilm formation of the genes involved in the pathway, as obtained by the transcriptomic analysis, are indicated by a 4-square stripe (from left to right: S. aureus USA300, S. aureus SH1000, E. faecalis V583 and E. faecalis 29212). The stripe with thick black lines represents expression in TSBg(s), and the thin line is in BHIg(s). The red and blue color scale indicates high and low expression levels, respectively. For abbreviations, see Table S6.

Genes involved in the purine biosynthesis pathway are important for biofilm production.

Since de novo purine biosynthesis was found highly upregulated in biofilm formation in TSBg(s) for all the strains used here, we wanted to confirm that this pathway is important for S. aureus biofilm formation. Thus, we performed biofilm growth assays using S. aureus USA300 strains with transposon insertion in purD, purH, purM, purN, and purQ genes from the purine biosynthesis pathway (Nebraska Transposon Mutant Library; NTML) (59). We also used a strain with a transposon insertion in purA from the purine biosynthesis and salvage pathway and in folD from the 1 C pool by folate pathway. The bifunctional protein FolD synthetizes 10-formyl tetrahydrofolate (10-formyl-THF), a cofactor used by purN and purH (60). As shown in Fig. 5A, disruption of purH and purQ significantly impaired biofilm production (Fig. 5A), and this effect was not due to a delayed or decreased cell division rate (Fig. 5B). This defect in biofilm formation was also visible under confocal microscopy (Fig. S6); ΔpurH had a generally weaker biofilm, and ΔpurQ biofilm showed good exopolysaccharide signal but appeared to contain less cells. Interruption of purA, a gene also part of the salvage pathway, had a strong impact on biofilm formation (Fig. 5A). Its biofilm appeared very sparse and contained few cells (Fig. S6). However, this mutation also had a mild effect on planktonic growth (Fig. 5B), which could partly explain the effect on biofilm production. Importantly, complementation of ΔpurA, ΔpurH, and ΔpurQ by addition of the final product of the pathway IMP restored biofilm formation to wild-type levels (Fig. S5 in the supplemental material). USA300 biofilm formation relies heavily on eDNA, thus possibly exacerbating the necessity of a functional purine biosynthesis pathway. To assess the importance of this pathway in PIA-dependent biofilms, we treated SH1000 with ATIC dimerization inhibitor, a drug targeting the bifunctional enzyme PurH (61). As shown in Fig. 5C, this drug successfully reduced biofilm formation by SH1000 without affecting cell growth (Fig. 5D). While the high concentration of drugs required is likely because the ATIC inhibitor was developed specifically for cancer treatment in mammalian cells, this result supports the notion that the purine biosynthesis pathway constitutes an interesting target to reduce biofilm formation in S. aureus.

FIG 5.

Inhibition of de novo purine biosynthesis pathway affects biofilm formation in S. aureus. (A) Biofilms of S. aureus USA300 containing transposon insertion (NTML) in gene encoding for the purine biosynthesis pathway were formed in TSBg and quantified by crystal violet. Each bar represents biofilm production (in OD₅₉₀) of 9 biological replicates ± SEM (one-way ANOVA and Dunnett’s multiple comparison; *, P value < 0.05; **, P value < 0.01; ****, P value < 0.0001). (B) Planktonic growth of S. aureus USA300 mutants in TSBg. Each point of the curve represents the cell density (OD₆₀₀) of 9 biological replicates ± SEM. (C) Biofilms of S. aureus SH1000 were formed in TSBgs containing different concentration of ATIC dimerization inhibitor and quantified by crystal violet. Each bar represents biofilm production (in OD₅₉₀) of at least 3 biological replicates ± SD (one-way ANOVA and Dunnett’s multiple comparison; ***, P value < 0.001; ****, P value < 0.0001). (D) Planktonic growth of S. aureus SH1000 in the presence of various concentration of ATIC inhibitor in TSBgs. Each point of the curve represents the cell density (OD₆₀₀) of at least 3 biological replicates ± SD.

DISCUSSION

Recent developments in biofilm research have highlighted metabolic remodeling during biofilm formation in some Gram-positive bacteria (36, 37). However, regulatory mechanisms leading to biofilm formation are often species or even strain-specific, making it difficult to develop anti-biofilm strategies that target a wide spectrum of bacteria (62, 63). Our transcriptomic approach with Gram-positive strains of clinical importance confirm that no cellular pathways or genes are commonly regulated in the two strains of S. aureus and E. faecalis during biofilm formation in both media tested (Table S3 in the supplemental material).

TSB and BHI medium supplemented with glucose and/or sodium chloride are two frequently used media to grow S. aureus and E. faecalis biofilms (49, 64). However, it was suggested that BHI promotes a protein-based matrix, while TSB-based media promotes either a PIA-dependent or PIA-independent biofilm (48). According to our transcriptomic analysis, several genes were found differentially expressed in one or the other medium, all strains and growth conditions combined. Many of those DEGs were attributed to primary metabolism pathways and some to secondary metabolism pathways (Table S4 in the supplemental material). We observed that the growth media used for biofilm formation had some effect on the transcriptome profiles in S. aureus, possibly due to environmental factors, but that these profiles were mostly unchanged in E. faecalis strains (Fig. 2). However, a recent study in E. faecalis highlights that the genetic determinants for robust biofilm formation in the OG1RF strain differ according to the medium and the experimental conditions (65). Different growth media were also reported to influence biofilm phenotypic properties and matrix composition in other Gram-positive strains such as B. subtilis (66).

Some of the genes and biological functions we found upregulated in S. aureus during biofilm growth (Fig. 3), such as fermentation, nucleotide metabolism and the TCA cycle, were already observed in other transcriptomic, metabolomic or proteomic studies using other S. aureus strains (55, 67–70). In E. faecalis, only a few metabolic adaptations during biofilm formation were reported (40, 57). Therefore, our data provide new information about biofilm formation of E. faecalis that could be exploited in further studies and consolidate previous observations on S. aureus biofilm formation.

Several studies previously reported purine biosynthesis as being upregulated in both Gram-positive and Gram-negative bacteria during biofilm formation (36, 51, 71–75). In keeping with those studies, we found a strong upregulation of the de novo purine biosynthetic process during biofilm formation in E. faecalis and S. aureus strains in TSBg(s) (Fig. 4). Inhibition of this pathway either by genetic disruption or by using a drug targeting a key enzyme led to a decrease in S. aureus USA300 and SH1000 biofilm production (Fig. 5). These results points toward an important need for purine during biofilm formation, which would be reflected by an upregulation of de novo purine biosynthesis in an environment where this nutrient is not abundant. An interesting perspective from this study will be to use drugs targeting this pathway to investigate this phenomenon in other Gram-positive strains that lack efficient transformation tools or an available mutant library.

Purines are required for eDNA production, an important constituent of the extracellular matrix of S. aureus (76), E. faecalis (77) and other Gram-positive bacteria such as B. subtilis or B. cereus (78). However, it was reported that in B. cereus, purD and purH deletions did not significantly affect the level of eDNA in B. cereus matrix compared to the wild-type strain (51). Importantly, genes of the de novo purine biosynthesis pathway were shown to be important for biofilm formation in Streptococcus sanguinis (75), Pseudomonas fluorescens (74), and a Burkholderia symbiont (72). Recent studies also showed that upregulation of the de novo purine biosynthesis pathway led to increased virulence in mice (79–81). Our results show that this pathway is important in S. aureus and is upregulated in E. faecalis and indicate that the requirement for high amounts of purine during biofilm formation might be widespread among bacteria.

MATERIALS AND METHODS

Strains and media.

Strains used for the transcriptomic analysis are S. aureus USA300, S. aureus SH1000, E. faecalis V583, and E. faecalis ATCC 29212 (a kind gift from F. Malouin, Université de Sherbrooke). Biofilm phenotypic assays were performed with S. aureus USA300 mutants from the Nebraska Transposon Mutant Library. The USA300 JE2 strain was used as a wild-type control (NTML; a kind gift from J.-P. Côté, Université de Sherbrooke) (82) (see Table S1 in the supplemental material). Strains were maintained on Tryptic Soy agar or Brain Heart Infusion agar plates, and for the NTML mutants, 5 µg/mL of erythromycin was added to the media. To induce biofilm formation, Tryptic Soy broth or BHI broth supplemented with 0.5% (wt/vol) glucose (TSBg or BHIg) was used for E. faecalis strains and S. aureus USA300, and TSBg or BHIg supplemented with 3.0% (wt/vol) sodium chloride (TSBgs or BHIgs) was used for S. aureus SH1000.

Growth measurement (planktonic cells and biofilm).

Colonies from overnight cultures on Tryptic Soy agar or BHI agar plates at 37°C were suspended in 1 mL of the appropriate biofilm inducing medium. Then, 200 µL of biofilm inducing medium was added in wells of a 48-wells culture plate and inoculated to an OD₆₀₀ of 0.005 with the bacterial suspension. For planktonic growth measurement (at least three biological replicates), plates were incubated in a Synergy HT plate reader (BioTek, Winooski, VT, USA) at 37°C with continuous agitation at 1,140 rpm to prevent biofilm formation and monitored at 600 nm every 30 min for 12 h (Fig. 1), 24 h (Fig. 5B), or 10 h (Fig. 5D). Absence of biofilm in these conditions was confirmed by crystal violet. For Fig. 5C and D, ATIC dimerization inhibitor (Millipore-Sigma) was added from a 1 mg/mL stock in distilled water.

For biofilm formation, polystyrene plates were incubated at 37°C without agitation for 12 h (Fig. 1), 24 h (Fig. 5A) or 9 h (Fig. 5C); quantification (at least 3 biological replicates) was performed using the crystal violet method (83). Briefly, biofilms were gently washed with phosphate-buffered saline (PBS), covered with 200 µL of 0.01% (wt/vol) crystal violet and incubated for 20 min at room temperature, protected from light. Each well was then gently washed with sterile distilled water to remove the excess of dye. Stained biofilms were solubilized in 200 µL of 33% acetic acid and quantified by spectrophotometry at 590 nm.

mRNA extraction and isolation.

Planktonic cells (three biological replicates) were washed with cold PBS and kept in RNA stabilization solution (25 mM sodium citrate, 10 mM EDTA, 47% ammonium sulfate, adjusted to pH 5.2) until all samples were harvested. Biofilms (three biological replicates) were washed with cold PBS and then suspended in RNA stabilization solution by scraping the bottom of the well. Planktonic and biofilms cells were then centrifuged, the supernatant discarded, and the pellet kept at −80°C until RNA extraction.

Cells were suspended in 250 µL of fresh lysis buffer (45 mg/mL of lysozyme, 16 µL/mL of lysostaphin 5 mg/mL, and 10 µL/mL of mutanolysin 5 U/µL in Tris-EDTA) and incubated for 1 h at 37°C, while mixed by inversion every 15 min. They were then transferred in O-Ring tubes with 25–50 mg of acid-washed beads and mixed with three volumes of TRI Reagent, to be lysed mechanically with a FastPrep-24 Classic (MP Biomedicals, Santa Anna, CA, USA) (speed of 4.5 m/s, 2 × 20 s). Total RNA was extracted with the Direct-zol RNA Miniprep Plus kit (Zymo Research, Irvine, CA, USA) following the manufacturer's instruction, including the DNase I (New England Biolabs, Ipswich, MA, USA) treatment in the extraction column. RNA quantity and the integrity number were evaluated with an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), and all samples had integrity number between 8 and 10. To remove rRNA (rRNA), samples were treated using a Pan-Prokaryote riboPOOL kit (siTOOLs Biotech Gmbh, Planegg, Germany) coupled with streptavidin magnetic beads (New England Biolabs, Ipswich, MA, USA). mRNA was eluted in 50 µL of nuclease-free water and then cleaned and concentrated using a Clean & Concentrator kit (Zymo Research, Irvine, CA, USA). mRNA samples were kept at −20°C if used on the same day or frozen at −80°C for long-term storage. Ribo-depletion success was assessed by verifying the diminution or disappearance of the 16s rRNA and 23s rRNA peaks with an Agilent 2100 Bioanalyzer, between a treated sample and a nontreated sample.

Construction of prokaryotic Illumina libraries.

Libraries were prepared for RNA sequencing according to the instructions of the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New Englands Biolabs, Ipswich, MA, USA), with the following exception. RNA samples were fragmented 7 min instead of 15 min to produce fragments longer than 100 bp. Instead of the adaptors provided with the kit, we used TruSeq compatible YIGA adaptors (see Table S2 in the supplemental material). Nucleic acid purification was performed using Mag-Bind TotalPure NGS beads (Omega Bio-Tek Inc, Norcross, GA, USA) following the recommended ratio in the NEBNext kit. Ligation products were subjected to PCR amplification using a reverse primer containing a unique 8 bp index for each sample (see Table S2 in the supplementary material). The concentration of nucleic acid in the libraries was measured using a Quant-iT PicoGreen dsDNA assay kit (Invitrogen, Carlsbad, CA, USA) and 1.13 ng of each sample were pooled in 80 µL of nuclease-free water. Quantity and quality of the final mix were confirmed with an Agilent 2100 Bioanalyzer before submitting to a Nextseq500 High Output 75 bp sequencing.

RNA sequencing data analysis.

Approximately 8,000,000 reads per library were obtained using an Illumina Nextseq system providing single-ended reads of 75 bp each (Rnomics platform, Université de Sherbrooke). The data was analyzed using Geneious Prime 2020.0.3 (https://www.geneious.com). Fastq files were trimmed with the BBDuk plugin using the default parameters and eliminating reads shorter than 10 bp. Trimmed data were then aligned with Bowtie2 (84) using an “end-to-end” alignment type, a “High Sensitivity/Medium” preset and using the “do not trim option,” against the following reference chromosomes: GenBank accession number GCA_000013425.1 (for S. aureus SH1000, a close descendant of the ancestral strain NCTC 8325) (85); GenBank accession number GCA_000017085.1 (for S. aureus USA300); GenBank accession number GCA_000742975.1 (for E. faecalis ATCC 29212) and GenBank accession number GCA_000007785.1 (for E. faecalis V583). Finally, the differentially expressed genes (DEGs) were assessed with DESeq2 (86) by combining the 3 biological replicates of each condition and comparing biofilm cultures vs planktonic cultures in either TSBg(s), BHIg(s) or both. Only the genes with an adjusted P value ≤ 0.05 were retained for further analysis. DEGs were scored as upregulated if they had a log₂ fold change ≥ 2.0 and downregulated if they had a log₂ fold change ≤ −2.0. For the transcriptomic study on growth media effect on genes expression, DEGs with a log₂ fold change ≥ 2.0 were considered upregulated in BHIg(s) while DEGs with a log₂ fold change ≤ −2.0 were considered upregulated in TSBg(s).

Functional annotation.

The various gene set enrichment analysis (GSEA) (or functional annotation) was performed using DAVID Bioinformatics Resources version 6.8 (52). The locus tag of the upregulated and downregulated DEGs were submitted as a list to DAVID and then analyzed with the Functional Annotation tool. The UP_KEYWORDS of the Functional_category, the KEGG_PATHWAY of the Pathways category, the GOTERM_BP_DIRECT and/or GOTERM_MF_DIRECT of the Gene_Ontology category were selected as possible annotation sources. The analysis was performed with the functional annotation chart, including the default thresholds of a count of 2 and an EASE of 0.1 and applying the following statistical test: fold enrichment, Benjamini, and FDR. Following DAVID recommendations, function with a fold enrichment ≥ 1.5 and a Benjamini value < 0.05 were considered interesting. The GSEA comparing biofilm vs planktonic conditions was confirmed with the ShinyGO v0.61 software. The locus tags of the same DEGs were submitted as a list, using either the E. faecalis STRINGdb or the S. aureus NCTC8325 STRINGdb and the default P value cutoff (FDR) of 0.05. Both KEGG pathways and GO:BP were considered.

Core genome circular comparison.

To identify potential conserved cellular pathways among the strain used in this study, each of them was submitted to the Proteome Comparison Service performs, a protein sequence-based genome comparison tool from PATRIC (46), using the default parameters and the following reference genomes: Staphylococcus aureus subsp. aureus NCTC 8325 (Genome ID: 93061.5), Staphylococcus aureus subsp. aureus USA300_TCH1516 (Genome ID: 451516.9), Enterococcus faecalis V583 (Genome ID: 226185.9) and E. faecalis ATCC 29212 (Genome ID: 1201292.8). Genome comparison was illustrated in a scalable vector graphic image file produced by Circos. All the recognized locus tags of the genes sharing over 50% of sequence identity in the genome comparison text file of every strain were subjected to a GSEA as described earlier.

Statistical analysis.

Statistical analysis was performed using GraphPad Prism 8.4.0. Comparisons were done using Student's t test or one-way ANOVA followed by Tukey’s multiple-comparison test, both with 95% confidence intervals.

Data availability.

The data sets generated and analyzed during the current study are available in the Gene Expression Omnibus database under the accession number GSE162709.