Downregulation of Autolysin-Encoding Genes by Phage-Derived Lytic Proteins Inhibits Biofilm Formation in Staphylococcus aureus

ABSTRACT

Phage-derived lytic proteins are a promising alternative to conventional antimicrobials. One of their most interesting properties is that they do not readily select for resistant strains, which is likely due to the fact that their targets are essential for the viability of the bacterial cell. Moreover, genetic engineering allows the design of new “tailor-made” proteins that may exhibit improved antibacterial properties. One example of this is the chimeric protein CHAPSH3b, which consists of a catalytic domain from the virion-associated peptidoglycan hydrolase of phage vB_SauS-phiIPLA88 (HydH5) and the cell wall binding domain of lysostaphin. CHAPSH3b had previously shown the ability to kill Staphylococcus aureus cells. Here, we demonstrate that this lytic protein also has potential for the control of biofilm-embedded S. aureus cells. Additionally, subinhibitory doses of CHAPSH3b can decrease biofilm formation by some S. aureus strains. Transcriptional analysis revealed that exposure of S. aureus cells to this enzyme leads to the downregulation of several genes coding for bacterial autolysins. One of these proteins, namely, the major autolysin AtlA, is known to participate in staphylococcal biofilm development. Interestingly, an atl mutant strain did not display inhibition of biofilm development when grown at subinhibitory concentrations of CHAPSH3b, contrary to the observations made for the parental and complemented strains. Also, deletion of atl led to low-level resistance to CHAPSH3b and the endolysin LysH5. Overall, our results reveal new aspects that should be considered when designing new phage-derived lytic proteins aimed for antimicrobial applications.

INTRODUCTION

Over the last few decades, there has been a dramatic rise in the resistance of bacterial pathogens to conventional antimicrobials. In response to this phenomenon, scientists have stepped up their efforts to develop new agents for the control of pathogenic microbes. This research has involved careful reconsideration of some previously dismissed strategies, sometimes giving a new twist to old therapeutics. One such example is phage therapy, which has recently attracted considerable attention in the field of antimicrobial development (1, 2). Moreover, advances in molecular biology allow a more sophisticated design of phage-derived therapeutics. Thus, we can now dissect the specific mechanisms employed by phages to attack bacteria and use them to our advantage. For instance, phage lytic proteins, such as endolysins and virion-associated peptidoglycan (PG) hydrolases, have been proposed as promising antibacterial agents, the so-called enzybiotics (3). These enzymes play a very important role in the lytic cycle, as their muralytic activity enables the phage to enter the bacterial cell (virion-associated PG hydrolases) and ultimately lyse the host cell to release the viral progeny (endolysins). However, both types of proteins can potentially cause exolysis or “lysis from without” when used as antimicrobials (4, 5). Indeed, several studies have proven their ability to lyse planktonic and biofilm-embedded cells from different species (3).

There is evidence that phage lytic proteins can be effective for the control of the notorious human pathogen Staphylococcus aureus (6). This bacterium is particularly problematic because of its ability to form biofilms on a wide range of surfaces, which facilitates the survival of bacterial cells even in hostile environments (7). Additionally, there is a high prevalence of S. aureus strains resistant to most of the antibiotics available on the market. This is why the term “superbug” has often been used in connection with this microorganism (8). Some antistaphylococcal phage endolysins have shown antibiofilm activity, including phage 11 endolysin (9), the lysin CF-301 (10), and SAL-2, an endolysin derived from phage SAP-2 (11), among others (12, 13). Two other promising phage lytic proteins used against S. aureus are the endolysin LysH5 and the PG hydrolase HydH5, both derived from phage vB_SauS-phiIPLA88 (14, 15). In the case of LysH5, Gutiérrez et al. (13) showed not only that this protein is effective for biofilm removal but also that subinhibitory concentrations resulted in decreased biofilm formation. However, the molecular mechanisms mediating this effect remained unknown. Indeed, there is no available information regarding the responses of bacterial cells to phage-derived lytic proteins. The virion-associated PG hydrolase HydH5 consists of two lytic domains, i.e., an N-terminal cysteine/histidine-dependent amidohydrolase/peptidase (CHAP) domain and a C-terminal LYZ2 (lysozyme subfamily 2) domain. However, no cell wall binding (CWB) domain could be identified in the amino acid sequence of HydH5. In a subsequent study, Rodríguez-Rubio et al. (16) demonstrated that the activity of the protein HydH5 could be improved by fusing its CHAP domain to the CWB domain SH3b of lysostaphin. The resulting chimeric protein was designated CHAPSH3b and exhibited specific activity against staphylococci. Moreover, CHAPSH3b displayed the ability to eliminate S. aureus cells in milk and could withstand pasteurization treatment, as well as storage at 4°C for 3 days (17). Nonetheless, the potential of CHAPSH3b for biofilm removal remained to be explored.

Initially, our main goal was to evaluate the antibiofilm ability of the protein CHAPSH3b. However, our preliminary data led us to investigate the transcriptional responses of S. aureus cells to subinhibitory concentrations of this enzybiotic. The results of this analysis suggest that the biofilm inhibitory effect of these proteins may be due to downregulation of endogenous cell wall hydrolases, which are known to participate in biofilm development in staphylococci. Furthermore, we show here that deletion of the gene encoding the major autolysin of S. aureus, AtlA, confers low-level resistance to phage lytic proteins on S. aureus cells. Thus, this work reports for the first time gene expression changes associated with exposure to phage lytic proteins and reveals that these changes may be linked to adaptive resistance and antibiofilm properties.

RESULTS

Treatment of preformed biofilms with CHAPSH3b protein.

Previous studies had demonstrated the antistaphylococcal activity of the fusion protein CHAPSH3b at room temperature and at 37°C (16, 17). In this study, we sought to investigate whether this protein can also be used for the control of biofilms formed by S. aureus, which generally exhibit resistance to many conventional antimicrobial agents. Two strains were chosen for these experiments on the basis of their different origins and abilities to form sessile communities. One of them, S. aureus IPLA 1, was isolated from a dairy industry sample and is a poor biofilm former, while the other, S. aureus ISP479r, was derived from clinical isolate NCTC 8325 and can form strong biofilms. Interestingly, both S. aureus strains showed identical susceptibility to CHAPSH3b, with MICs of 3.91 and 62.48 μg/ml at 25°C and 37°C, respectively.

To evaluate the antibiofilm activity of this protein, biofilms of the two strains were grown for 24 h at 25°C or 37°C and then treated with 4× MIC of CHAPSH3b (15.64 and 250.24 μg/ml at 25°C and 37°C, respectively) for 6 h. The results showed a significant decrease in the adhered biomass of both strains at both temperatures (Fig. 1A and C). Thus, treatment of biofilms formed by S. aureus IPLA 1 led to reductions of 70% and 53% at 25°C and 37°C, respectively (Fig. 1A). In the case of S. aureus ISP479r, biofilm formation decreased by 30% and 61% at 25°C and 37°C, respectively, compared to that of the untreated controls (Fig. 1C). A significant reduction in bacterial cell counts was also observed in all biofilms treated with CHAPSH3b (Fig. 1B and D). Nevertheless, the results showed that treatment with this protein was more effective at 37°C than at 25°C. Thus, the viable cell counts of both strains only decreased by <1 logarithmic unit in biofilms treated at 25°C (Fig. 1B). In contrast, treatment at 37°C led to 1- and 2-log reductions in strains ISP479r and IPLA 1, respectively (Fig. 1D).

FIG 1.

Treatment of preformed biofilms of S. aureus IPLA 1 and ISP479r with the fusion protein CHAPSH3b. Biofilms were formed for 24 h and then treated with 4× MIC of CHAPSH3b for 6 h at 25°C (A and B) or 37°C (C and D). Control wells were incubated with NaPi buffer alone. After incubation, the adhered biomass was quantified by reading the A₅₉₅ after crystal violet staining (A and C). Alternatively, viable cell counts of bacteria adhered to the well were determined (B and D). Black and gray bars correspond to control and treated samples, respectively. The values represented correspond to the means and standard deviations of three independent biological repeats. *, P < 0.05.

After confirming the ability of CHAPSH3b to kill S. aureus cells forming biofilms, microscopy analyses were performed to observe the effects of the lytic protein in greater detail. To do that, biofilms of strain ISP479r were grown for 24 h at 37°C and subsequently treated for 6 h with 250.24 μg/ml CHAPSH3b (4× MIC) or with buffer alone at the same temperature. These samples were then stained with SYTO 9, which dyes live cells, and propidium iodide, which dyes dead cells and extracellular DNA (eDNA), and observed by confocal laser scanning microscopy (CLSM). The control samples showed well-organized biofilms that completely covered the glass surface (Fig. 2A), whereas the treated biofilm contained fewer bacterial cells and displayed a general loss of structure together with accumulation of eDNA, probably the result of widespread cell lysis (Fig. 2B). Additionally, time-lapse microscopy allowed us to monitor the changes in the S. aureus ISP479r biofilm throughout the 6-h treatment (see Video S1 in the supplemental material). This experiment confirmed the data described above, as areas of lysis appeared in the biofilm and gradually increased in size during the incubation period.

FIG 2.

Confocal laser scanning micrographs of S. aureus biofilms treated or not with CHAPSH3b. Biofilms of strain ISP479r were grown for 24 h at 37°C and then treated with NaPi buffer alone (A) or containing 4× MIC of CHAPSH3b (250.24 μg/ml). After incubation, the planktonic phase was removed and samples were stained with the LIVE/DEAD BacLight staining kit. Intact cells appear green, whereas dead cells and eDNA are stained red.

Subinhibitory concentrations of CHAPSH3b decrease biofilm formation by S. aureus.

Gutiérrez et al. (13) had previously shown that exposure to the endolysin LysH5 at the beginning of biofilm formation inhibited this process in some S. aureus strains. Here, we wanted to determine if this phenomenon also occurs as a result of exposure to the fusion protein CHAPSH3b. To do that, biofilms of strains IPLA 1 and ISP479r were grown for 24 h at 25°C or 37°C in the presence of increasing concentrations of CHAPSH3b and the adhered biomass was then compared to a control grown without the protein. Interestingly, the protein showed the ability to significantly reduce the adhered biomass at both growth temperatures, even at concentrations well below the MIC (Fig. 3). Indeed, biofilm formation by strain IPLA 1 was reduced by about 50% at concentrations corresponding to 0.125× MIC at both 25°C (Fig. 3A) and 37°C (Fig. 3C). In the case of strain ISP479r, 0.125× MIC led to respective reductions of 34% and 52% at 25°C (Fig. 3B) and 37°C (Fig. 3D). In contrast, growth of IPLA 1 and ISP479r at subinhibitory concentrations of the antibiotic vancomycin did not significantly inhibit biofilm formation (see Fig. S2).

FIG 3.

Biofilm formation by S. aureus strains in the presence of subinhibitory concentrations of CHAPSH3b. Biofilms of strains IPLA 1 (A and C) and ISP479r (B and D) were grown for 24 h at 25°C (A and B) or 37°C (C and D). The planktonic phase was then removed, and the adhered biomass was quantified by crystal violet staining and subsequent A₅₉₅ determination. The percent biomass production compared to the untreated controls was then calculated. The values shown correspond to the mean and standard deviation of three independent biological repeats. *, P < 0.05. The MICs for both strains were 3.91 and 62.48 μg/ml at 25 and 37°C, respectively.

The data described above showed that very low doses of CHAPSH3b are necessary for significant inhibition of the biofilm formation process at room temperature (25°C). This would be very interesting to test the use of this protein to reduce biofilm formation on surfaces in the food industry or the hospital environment, as well as in human therapy. For this reason, the effects of subinhibitory CHAPSH3b on biofilm structure at 25°C were studied in greater detail by confocal laser microscopy. After 24 h of development without protein, strain ISP479r could form thick biofilms that contained a well-defined network of eDNA surrounding the bacterial cells, with visible areas of accumulation (Fig. 4A and B). In contrast, when biofilm formation took place in the presence of 0.25× MIC (0.98 μg/ml) of CHAPSH3b, the resulting structure was different. On the one hand, the thickness of the biofilm was reduced overall and cells tended to accumulate in specific areas (Fig. 4C and D). Also, eDNA accumulation appeared to be reduced, whereas the number of dead cells had increased (Fig. 4C and D). Thus, propidium iodide was predominantly bound to small, cell size dots instead of forming a network around the cells (Fig. 4C and D). Furthermore, differences in the structure of the biofilms formed by strain ISP479r with and without subinhibitory lytic protein could even be observed macroscopically in biofilms stained with crystal violet (Fig. 4E). Of particular note is the observation of intensely stained areas located on a seemingly thin biofilm layer. In the case of strain IPLA 1, the changes displayed by biofilms developed in the presence of the fusion protein were not as obvious as in strain ISP479r. Nevertheless, greater coverage of the glass surface could be observed in the untreated control (Fig. 5A and C) than in the sample grown in the presence of CHAPSH3b (Fig. 5B and D). This would be consistent with the lower biomass values determined by crystal violet staining.

FIG 4.

Effect of subinhibitory concentrations of CHAPSH3b on S. aureus ISP479r biofilm formation. (A to D) Micrographs obtained by CLSM of biofilms grown in the absence (A and B) or presence (C and D) of 0.98 μg/ml CHAPSH3b at 25°C. The samples were stained with the LIVE/DEAD BacLight kit. Panels B and D show the three-dimensional structures of the biofilms. (E) Photograph of biofilms formed by S. aureus ISP479r in the presence of increasing concentrations of CHAPSH3b ranging from 0 to 3.91 μg/ml at 25°C and subsequently stained with crystal violet.

FIG 5.

Effect of subinhibitory concentrations of CHAPSH3b on S. aureus IPLA 1 biofilm formation. (A to D) Micrographs obtained by CLSM of biofilms grown in the absence (A and B) or presence (C and D) of 0.98 μg/ml CHAPSH3b at 25°C. The samples were stained with the LIVE/DEAD BacLight kit. Panels B and D show the three-dimensional structures of the biofilms.

Transcriptional responses to phage-derived lytic proteins.

Previous data had shown a biofilm-inhibiting effect of the endolysin LysH5 on some S. aureus strains. However, we still had no information concerning the underlying molecular mechanisms behind this phenomenon. In an attempt to understand how phage-derived lytic proteins exert their antibiofilm activity, we explored the transcriptional responses of S. aureus cells exposed to these antimicrobials. A preliminary study performed with 24-h biofilms of S. aureus IPLA 1 exposed to 0.25× MIC of LysH5 (10.94 μg/ml) had provided some hints about transcriptional responses to lytic proteins. In that experiment, preformed biofilms were incubated for 30 min with the buffer containing the protein or with the buffer alone. After treatment, RNA was purified from all of the samples and analyzed by transcriptome sequencing (RNA-seq) to assess the effect of LysH5 exposure on the transcriptome of the biofilm cells. The results revealed that 61 genes were significantly dysregulated (adjusted P values of <0.05) between the two conditions (see Table S3). However, in most cases, these expression changes were <2-fold. Interestingly, several genes encoding proteins with PG hydrolase activity were identified among the differentially expressed transcripts. These genes included lytM, sceD, sle1, and atl, which were downregulated by 1.43-, 1.76-, 1.35-, and 1.31-fold, respectively. Transcriptional analysis of biofilms presents a significant problem, as it is difficult to ensure that all cells are exposed to the protein. In order to obtain a more homogeneous response of the bacterial culture, transcriptional analysis of liquid cultures was also performed. For this purpose, S. aureus IPLA 1 cultures grown to mid-logarithmic phase at 25°C were exposed for 10 min to the same concentration of LysH5 used for the biofilm samples (10.94 μg/ml) and RNA was isolated and subsequently analyzed. Reverse transcription (RT)-quantitative PCR (qPCR) analysis of these samples confirmed that exposure to the endolysin resulted in decreased expression of endogenous PG hydrolases (Table 1). Dysregulation of other genes identified by RNA-seq in the biofilm samples could not be confirmed in the liquid cultures exposed to LysH5 (data not shown). As a result, we did not further study the role of these genes in the response to phage lytic proteins.

TABLE 1.

Transcriptional analysis of S. aureus IPLA 1 cells exposed to 0.25× MIC of endolysin LysH5^a

Gene	Fold change in biofilms	Avg fold change in liquid culture ± SD
atl	−1.31	−3.08 ± 0.86
sle1	−1.35	−5.14 ± 0.82
lytM	−1.42	−4.30 ± 1.90
sceD	−1.76	−14.40 ± 6.17

^aTotal RNA was purified from biofilms or a liquid culture grown at 25°C and then exposed to the lytic protein. Gene expression in these samples was then compared to that of an untreated control by RNA-seq or RT-qPCR. Results shown are the averages of three independent biological replicates.

With these results in mind, we examined whether exposure to subinhibitory CHAPSH3b also resulted in the downregulation of genes encoding bacterial PG hydrolases in strains IPLA 1 and ISP479r. To do that, we challenged mid-logarithmic-phase cultures of the two strains grown at 25°C with 0.25× MIC of CHAPSH3b (0.98 μg/ml) and isolated RNA for transcriptional analysis. RT-qPCR data confirmed that the autolysin-encoding genes atl, sle1, lytM, and sceD were indeed repressed following exposure to the lytic protein in both strains (Table 2).

TABLE 2.

Effect of CHAPSH3b on the transcription of autolysin-encoding genes in S. aureus strains IPLA 1 and ISP479r^a

Gene	Mean fold change ± SD
	IPLA 1	ISP479r
atl	−3.15 ± 0.28	−13.35 ± 3.25
sle1	−9.19 ± 2.06	−29.70 ± 6.27
lytM	−2.31 ± 0.38	−14.07 ± 1.38
sceD	−5.24 ± 1.84	−12.96 ± 3.45

^aRNA was isolated from liquid cultures grown at 25°C to mid-logarithmic phase and then exposed to 0.98 μg/ml CHAPSH3b (0.25× MIC) for 10 min. Values represent the means and standard deviations of three independent biological replicates.

Mutation of the major autolysin AtlA in S. aureus SA113 prevents biofilm inhibition by CHAPSH3b but leads to low-level resistance to phage lytic proteins.

Transcriptional analysis revealed a direct link between exposure to phage lytic proteins and downregulation of autolysins in S. aureus strains. For instance, the gene coding for the S. aureus major autolysin AtlA showed 3- and 13-fold downregulation in strains IPLA 1 and ISP479r, respectively. Sometimes, bacterial responses to an antimicrobial challenge provide us with hints about resistance mechanisms (18). For this reason, we studied the participation of the major autolysin atl in resistance to phage lytic proteins. To do that, the MICs of lytic proteins for strain SA113 and its isogenic atl mutant were determined. These assays indicated that the atl mutant was slightly more resistant to CHAPSH3b. Thus, the MIC for the mutant was 7.81 μg/ml, while the MIC for the parental strain was 3.91 μg/ml. Notably, the mutant was also more resistant to the endolysin LysH5, with an MIC of >43.75 μg/ml compared to a MIC of 21.88 μg/ml for the parental strain. This resistance phenotype was reverted in the complemented strain. It must be noted that no statistically significant difference was found between the growth rate of the atl mutant and that of the wild-type strain, which had respective generation times of 41.27 ± 1.04 and 38.30 ± 0.15 min (P = 0.14).

It is well established that the major autolysin AtlA plays a key role in S. aureus biofilm formation (19). Therefore, we sought to investigate the possible participation of the gene that encodes it in biofilm inhibition by phage-derived lytic proteins. This was achieved by comparing biofilm formation by strain SA113 and its isogenic atl mutant strain in the presence of increasing concentrations of CHAPSH3b. The attached biomass was then quantified by crystal violet staining, and the resulting values were compared to those of a control biofilm of each strain grown without the protein (Fig. 6). Strain SA113 showed a trend similar to that observed with strain ISP479r, with decreases in biofilm formation of 26 and 51% at 0.125× MIC (0.49 μg/ml) and 0.25× MIC (0.98 μg/ml), respectively. Conversely, no significant change in biofilm formation by the atl mutant was observed at either 0.125× MIC (0.98 μg/ml) or 0.25× MIC (1.96 μg/ml) (P > 0.05). The complemented strain showed the same behavior as parental strain SA113, with respective reductions in biomass of 37 and 63% at 0.125× MIC (0.49 μg/ml) and 0.25× MIC (0.98 μg/ml). As a negative control for the effect of CHAPSH3b on biofilm formation, strain SA113 and the atl mutant were also grown in the presence of subinhibitory concentrations of vancomycin, but no significant biofilm inhibition was observed in either strain (see Fig. S2).

FIG 6.

Effect of atl mutation on the antibiofilm effect of CHAPSH3b. Biofilms of S. aureus SA113 (black), its isogenic atl mutant (white), and the complemented strain (gray) were formed at 25°C in the presence of increasing concentrations of the lysin (0 to 7.81 μg/ml). The attached biomass was quantified by crystal violet staining and subsequent A₅₉₅ determination. The data corresponding to the treated wells were compared to those of the respective controls. The values represent the average and standard deviation of three independent biological repeats. *, P < 0.05. The MIC for the parental and complemented strains was 3.91 μg/ml, whereas the MIC for the mutant was 7.81 μg/ml.

DISCUSSION

The “honeymoon” period of antibiotic therapy is over. By now, we have encountered microbes resistant to practically all of the antibiotics available on the market, including superbug strains of S. aureus. This situation is only worsened by the ability of this bacterium to form attached multicellular communities with inherent resistance to external challenges. Indeed, biofilms are known to withstand very high concentrations of antibacterial agents that would be lethal to cells in the planktonic state. As a response to this threat, researchers have been steadily working on the development of novel antimicrobials. An interesting possibility is the use of bacteriophages, one of bacteria's natural predators, or phage-derived proteins as therapeutics. For instance, phage PG hydrolases have shown promising results for the control of S. aureus, even when they are part of a biofilm (9–11, 13). Moreover, some phage lytic proteins, like LysH5, display biofilm-inhibiting activity (13). Also importantly, lysin CF-301 has been shown to be effective for the treatment of systemic S. aureus infections in a mouse model (10).

Here, we set out to study the ability of the fusion protein CHAPSH3b to kill S. aureus biofilm cells and/or to prevent biofilm development. Regarding treatment of preformed biofilms, we have demonstrated that this protein can reduce viable cell counts, as well as the total adhered biomass, at both 25°C (room temperature) and 37°C (infection temperature). Microscopy analysis also confirmed that treatment of S. aureus biofilms with CHAPSH3b could lyse part of the population and reduce the number of bacterial cells in the biofilm. However, it must be noted that the activity of the protein was higher at 37°C than at 25°C, even though MIC determination assays indicated that bacteria were actually more sensitive at room temperature. To explain this phenomenon, it must be considered that biofilm treatment was performed under conditions that did not allow bacterial growth. Consequently, the results may reflect the activity of the enzyme, which is higher at 37°C than at 25°C (15). In contrast, MIC determination assays are affected by differences in the bacterial growth rate, which is higher at 37°C than at 25°C. Similarly, CHAPSH3b was more effective in killing S. aureus cells in raw and pasteurized milk at room temperature than at 37°C. It is very likely that the bacterial cell population can increase too rapidly at the latter temperature, making it necessary to use a higher enzyme concentration to control the whole population.

Addition of CHAPSH3b at the beginning of the biofilm-forming process also had an inhibitory effect on biofilm development at both 25°C and 37°C. This is a very interesting property, as some antimicrobials like vancomycin have precisely the opposite effect; that is, they promote biofilm formation (20, 21). Other compounds that exert a negative effect on biofilm development by S. aureus include the lantibiotic gallidermin (22) and the synthetic cationic peptide IDR-1018 (23). Gallidermin was shown to decrease the expression of the gene encoding autolysin AtlA and the ica operon, involved in polysaccharide intercellular adhesin synthesis (22). In contrast, IDR-1018 prevents biofilm formation in different bacterial species by interfering with the stringent response (23). As mentioned above, Gutiérrez et al. (13) had already demonstrated that endolysin LysH5 inhibited biofilm formation in different S. aureus strains. In the case of CHAPSH3b, we also investigated the effect on the biofilm structure by confocal microscopy. The structure of the biofilm formed by the food industry strain IPLA 1 did not exhibit major structural alterations, although surface coverage was diminished in the samples grown in the presence of the protein. In contrast, the strong biofilm former ISP479r displayed more noticeable changes in the lysin-exposed biofilm than in the control. Indeed, the control biofilms of this strain were thicker and contained more eDNA surrounding the bacterial cells, whereas treated biofilms were overall thinner with areas of cell accumulation. This phenotype could even be observed macroscopically in crystal violet-stained biofilms, where the surface of the well appeared to be covered by a thin layer with scattered intensely stained spots.

In many cases, the phenotypic effects associated with exposure to subinhibitory antimicrobials can be understood by studying the transcriptional responses of bacterial cells under these conditions (18, 24). Until now, there was no information on the gene expression changes triggered by exposure to phage lytic proteins. In this work, we show that S. aureus cells challenged with two different lytic proteins, namely, LysH5 and CHAPSH3b, display downregulation of several genes coding for endogenous PG hydrolases, including the atl gene, which encodes the major autolysin. The atl gene had been previously shown to be downregulated upon exposure to other cell wall-active antimicrobials, namely, bacitracin, d-cycloserine, and oxacillin (25). The S. aureus autolysin AtlA and its homolog in S. epidermidis, AtlE, participate in the biofilm formation process of these microbes. In fact, autolysins are thought to be especially important during the initial attachment phase (26). Taking this into account, we thought that there might be a link between lysin-induced downregulation of atl expression and biofilm inhibition. Study of an atl mutant strain derived from SA113 indicated that this was indeed a possibility. Thus, strain SA113 displayed a phenotypic response similar to that of strains IPLA 1 and ISP479r, with decreased biofilm formation in the presence of subinhibitory concentrations of CHAPSH3b. In contrast, its isogenic atl mutant strain did not show a significant reduction in biofilm-forming ability at low doses of the protein, a phenotype that could be complemented by expressing the gene in trans. Interestingly, while performing these experiments, we also realized that atl mutation leads to low-level resistance to both CHAPSH3b and LysH5. To our knowledge, this would make atlA the first known phage lysin resistance determinant. Moreover, the downregulation of atl triggered by lytic proteins suggests that this may be a defense mechanism against these antimicrobials, thereby constituting an example of adaptive resistance. The phenomenon of adaptive resistance had been originally dismissed, but it is now gaining attention in the field of antibiotic development (27). It is particularly interesting because there seems to be a connection between mechanisms of adaptive resistance and mechanisms of mutational resistance (18). For example, a given gene may be downregulated upon exposure to subinhibitory antimicrobial concentrations or be mutated in a particular strain, leading in both cases to reduced susceptibility to an antimicrobial compound. Therefore, attaining a better understanding of bacterial responses to nonlethal antimicrobial concentrations may provide hints about mutations that lead to resistance. In the case of phage-derived lytic proteins, our results show that mutational resistance is possible but that the increase may be so small that these mutants would be difficult to identify in routine screenings. This would explain why no resistant strains have been generally isolated after rounds of exposure to phage lytic proteins (3, 28). Additionally, autolysins are very important for cell wall turnover and cell division, among other functions, which probably makes these mutants less competitive in a real-life situation. Of note, Becker et al. (12) recently reported the selection of increased resistance in S. aureus Newman following 10 rounds of exposure to sublethal concentrations of LysK and chimeric endolysins, although they did not identify the genetic mechanisms responsible for this phenotype.

In S. aureus, decreased expression of autolysins may help bacterial cells cope with the deleterious effect of phage lytic proteins. However, it also results in a lesser ability to form biofilms, at least in some strains. This would, in turn, make the population more sensitive to treatment with other antimicrobials as cells would not be able to attach so readily to surfaces and would remain in the planktonic phase. Taking all of these factors into consideration, it may be a good compromise to change low-level resistance by biofilm inhibition. Nonetheless, this would still be an important aspect to consider for the design of phage-derived lytic proteins. For instance, our results show that the atl mutant displays a greater increase in resistance to LysH5 than to CHAPSH3b compared to the parental and complemented strains. Thus, depending on their specific targets in the PG or their binding sites, some lytic proteins may be more affected by lower expression or mutation of autolysin genes.

To sum up, the main achievement of this work lies in the identification of transcriptional responses to phage lytic proteins and their potential role in biofilm inhibition and adaptive resistance to enzybiotics. These are factors that definitely need to be considered for the improvement of products based on these novel antimicrobials and especially for the design of new lytic enzymes. It will also be interesting to find out if this phenomenon is specific to S. aureus or if similar transcriptional trends exist in other microorganisms. Despite the negative connotations generally associated with the identification of novel resistance mechanisms, the overall conclusion of this study is, in our opinion, a positive one. On the one hand, our results seem to confirm that mutational resistance to phage lytic proteins would involve loss of important cell functions. Moreover, we demonstrate that this same resistance mechanism works in our favor by decreasing the ability of S. aureus to form dreaded biofilms. Therefore, enzybiotics remain a promising therapeutic alternative to conventional antimicrobials.

MATERIALS AND METHODS

Bacterial strains, culture conditions, and purification of phage-derived lytic proteins.

The S. aureus strains used in this study are listed in Table 3. S. aureus cultures were routinely grown at 37°C on Baird-Parker agar plates (AppliChem, Germany) or in TSB (tryptic soy broth; Scharlau, Barcelona, Spain) with shaking. When necessary, the antibiotics spectinomycin and chloramphenicol, both purchased from Sigma (St. Louis, MO, USA), were added to the growth media at concentrations of 150 and 20 μg/ml, respectively.

TABLE 3.

Bacterial strains used in this study

S. aureus strain	Description	Reference
IPLA1	Isolate from dairy industry sample	34
ISP479r	NCTC 8325-4 derivative with rsbU restored	35
SA113	NCTC 8325 derivative, agr, 11-bp deletion in rsbU	36
SA113 Δatl::spc	Δatl::spc, Spcr	37
SA113 Δatl::spc(pRBatlE)	Δatl::spc complemented with pRC14, Spcr, Cmr	38

The PG hydrolases CHAPSH3b and LysH5 were purified as described previously (29). The purified proteins were then checked by SDS-PAGE analysis and quantified by using the Quick Start Bradford Protein Assay kit (Bio-Rad).

Treatment of preformed biofilms.

Biofilms were grown according to the protocol described by Herrera et al. (30), with some modifications. Briefly, overnight cultures of the different S. aureus strains were diluted in fresh TSBg (TSB supplemented with 0.25% [wt/vol] d-(+)-glucose) medium to a final concentration of 10⁶ CFU/ml. Two hundred microliters of these suspensions was poured into each well of a 96-well microtiter plate (Thermo Scientific, Nunc, Madrid, Spain). These microtiter plates were then incubated for 24 h at 25°C or 37°C to allow biofilm formation. The planktonic phase was then removed, and the adhered cells were washed twice with phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄ [pH 7.4]). These preformed biofilms were subsequently treated with sodium phosphate (NaPi) buffer (50 mM, pH 7.4) alone or containing 4× MIC of CHAPSH3b protein (15.64 or 250.24 μg/ml for samples incubated at 25°C or 37°C, respectively) for 6 h at the same temperature of biofilm formation. After treatment, the adhered phase was washed again with PBS and stained with 200 μl of 0.1% (wt/vol) crystal violet for biomass quantification as described previously (13). After 15 min, the excess crystal violet was washed with water and the dye attached to the well was solubilized with 33% (vol/vol) acetic acid. Finally, the absorbance at 595 nm (A₅₉₅) was measured with a Benchmark Plus Microplate Spectrophotometer (Bio-Rad Laboratories, Hercules, CA, USA). Alternatively, the adhered viable cells were counted by scratching the wells twice with sterile cotton swabs as previously described (13). The cells were then resuspended in 9 ml of PBS by vigorous vortexing, and serial dilutions were plated onto tryptic soy agar plates and incubated at 37°C.

MIC determination and biofilm formation in the presence of subinhibitory concentrations of PG hydrolases.

Determination of MICs was performed according to the broth microdilution technique in accordance with the CLSI guidelines (31, 32) but with TSBg as the growth medium. The MIC was determined as the lowest concentration of the protein that inhibited visible bacterial growth after 24 h of incubation at 37°C or 25°C. At the end of the incubation, the attached biomass was quantified by crystal violet staining as described above. All MIC assays were repeated five to eight times.

Analysis by CLSM and time-lapse microscopy.

For microscopy analyses, biofilms were grown on two-well μ-slides with a glass bottom (ibidi, USA). Each well was inoculated with 1 ml of a cell suspension containing 10⁶ CFU/ml in TSBg and 1 ml of TSBg alone or 1 ml of TSBg containing a subinhibitory concentration of CHAPSH3b. Biofilms were then allowed to grow for 24 h at 37°C or 25°C. To test the effect of subinhibitory concentrations of the protein, the planktonic phase was removed after biofilm development and wells were washed with PBS prior to staining with the LIVE/DEAD BacLight kit (Invitrogen AG, Basel, Switzerland). To analyze the impact of CHAPSH3B treatment on preformed biofilms, the planktonic phase was replaced with NaPi buffer (negative control) or buffer containing 4× MIC of CHAPSH3b and then incubated for an additional 6 h at the same temperature. After treatment, biofilms were stained with the LIVE/DEAD BacLight kit (Invitrogen AG, Basel, Switzerland). All samples were then observed under a confocal scanning laser microscope (DMi8; Leica Microsystems) with a 63× oil objective.

For time-lapse microscopy, 24-h biofilms grown at 37°C were washed with PBS and then placed in an incubation chamber set at 37°C connected to an inverted microscope (DMi8; Leica Microsystems) equipped with a Leica DFC365FX digital camera. The biofilm was then treated with NaPi buffer containing CHAPSH3b at 250.24 μg/ml (4× MIC) and monitored for 6 h. Images were acquired every 15 min with LasX software (Leica Microsystems).

Transcriptional analysis.

Samples from liquid cultures were grown to an optical density at 600 nm of 0.5 at 25°C and then exposed to a subinhibitory concentration of CHAPSH3b (0.98 μg/ml) or LysH5 (10.94 μg/ml) for 10 min. Cells were then centrifuged, and the pellets were frozen at −80°C until further processing. Biofilms were grown for 24 h at 25°C, supernatant was then removed and replaced with NaPi buffer alone or containing 10.94 μg/ml of protein LysH5. Biofilms were then incubated for 30 min before the adhered cells were scraped as previously described (33). Following treatment with RNAprotect (Qiagen), the samples were stored at −80°C prior to RNA purification. Total RNA from S. aureus samples was isolated as previously described (33). Briefly, cell lysis was performed by mechanical disruption with a FastPrep-24 in a solution of 1:1 phenol-chloroform, glass beads (Sigma), and 80 mM dithiothreitol. RNA isolation was performed with the Illustra RNA spin minikit (GE Healthcare). Afterwards, RNA samples were treated with Turbo DNase (Ambion) to remove traces of genomic DNA and 1 μl of SUPERase inhibitor (Ambion) was added to each 50-μl sample for storage at −80°C. RNA quality was checked by agarose gel electrophoresis, and the RNA concentration was determined with an Epoch microplate spectrophotometer (BioTek).

For RNA-seq analysis, 8 μg of RNA from each sample was sent to Macrogen Inc. (South Korea). Sequencing was performed with the Illumina HiSeq2000 platform (Illumina, San Diego, CA, USA), and quality control of the reads was carried out by using FastQC. Mapping of the RNA-seq reads to the S. aureus NCTC 8325 genome was carried out with BowTie2, and uniquely mapped reads were used for subsequent analyses. The differential gene expression analysis was carried out with EDGE-pro software. Bioinformatic analysis was performed at Dreamgenics (Dreamgenics, Oviedo, Spain).

To perform RT-qPCR, 0.5 μg of each RNA sample was converted into cDNA by using iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad). Then, 2.5 μl of a 1:25 dilution of cDNA was added to each well together with Power SYBR green PCR master mix (Applied Biosystems). Three independent cultures, each repeated in duplicate, were tested for each condition. Calculation of the fold changes was carried out according to the C_T method with the rplD gene as a control.

Statistical analyses.

Data corresponding to three independent biological replicates were analyzed with a two-tailed Student t test. P values of <0.05 were considered significant.

Accession number(s).

The RNA-seq data obtained in this study have been deposited in NCBI's Gene Expression Omnibus (GEO) and can be accessed through GEO series accession number GSE94512.