Staphylococcus epidermidis surfactant peptides promote biofilm maturation and dissemination of biofilm-associated infection in mice (original) (raw)
PSMβ peptides are the main PSM type produced in S. epidermidis biofilm culture. To investigate the role of PSMs in biofilms, we first determined production of PSM peptides in planktonic versus biofilm culture. We found that PSM production was overall lower in biofilm culture, while relative production of β-type PSMs was significantly increased (Figure 1). Of note, β-type PSMs were virtually the only PSM type produced in the biofilm mode of growth. We detected similar production patterns in all S. epidermidis strains in our collection, with the exception of 24 strains that completely lacked production of PSMs, including the _agr_-encoded δ-toxin. Thus, they are most likely naturally occurring mutants in the agr quorum–sensing system, whose deletion or lack of functionality leads to absence of PSM production (16–18). Of note, all those 24 strains were shown to contain _psm_β genes by analytical PCR (data not shown).
PSM production under biofilm and planktonic modes of growth. PSM production under planktonic (shaken 125 ml flasks containing 50 ml TSB/0.5% glucose, 18 hour growth, 37°C) or biofilm (24 hour static culture in microtiter plates, 37°C) modes of growth was assayed by RP-HPLC/ESI-MS. PSMs were determined in supernatants of planktonic or biofilm cultures after centrifugation. Absolute amounts are shown at the top, relative composition at the bottom. S. epid. 1457, S. epidermidis 1457.
The psmβ locus forms a transcriptional unit. β-type PSMs are distinguished from α-type PSMs and δ-toxin (~20–25 amino acids) by their larger size (~45 amino acids) and absence of cytolytic activity at physiological concentrations (18). As PSM production patterns suggested that especially the β-type PSMs may play a role in biofilm development, we first analyzed organization and transcription of the _psm_β genes (Figure 2A). The _psm_β operon is composed of 4 genes in strains S. epidermidis ATCC12228 and RP62A (19, 20), whose genome sequences are available, while the clinical isolate 1457 (21) used in our study has only 3 _psm_β genes. This is due to an exact duplication of the psm_β_1 gene in strains ATCC12228 and RP62A, while we determined by DNA sequencing that this duplication is missing from strain 1457. The remaining sequence of the _psm_β locus of strain 1457 was exactly the same as in ATCC12228. In the shotgun-sequenced strain M23864, the _psm_β operon appears to consist of 5 genes with slightly different peptide products. In keeping with the psm_β_1 gene duplication, strains RP62A and ATCC12228 secreted more PSMβ1 than PSMβ2, but combined production of the PSMβ peptides was not higher than in strain 1457 (data not shown). Furthermore, we never detected a peptide in culture filtrates of any S. epidermidis strain corresponding to the gene product of the psm_β_3 gene, indicating that the hypothetical PSMβ3 peptide is not secreted. To analyze whether the _psm_β genes form an operon, and thus are transcribed together, we performed Northern blot analysis (Figure 2B). We detected 1 predominant band, whose size corresponded exactly to the postulated size of a transcript comprising all 3 _psm_β genes of strain 1457, which therefore form an operon.
The _psm_β operon. (A) Organization of the _psm_β locus in genome-sequenced S. epidermidis strains and the 1457 strain used in this study. Construction of the _psm_β mutant allelic replacement strain is shown at the bottom. (B) Northern blot analysis of S. epidermidis 1457 RNA from different growth phases using a _psm_β probe. The _psm_β signal is marked by an arrow.
PSM peptides influence S. epidermidis biofilm formation in vitro. To investigate whether PSMβ peptides impact biofilm development, we first assayed in vitro biofilm formation with synthetic PSMβ peptides. The peptides were added to growing biofilm cultures of an agr mutant of strain 1457 (22), which owing to strict regulation of PSMs by agr (16–18), is devoid of PSM production. The PSMβ peptides promoted biofilm formation in a medium concentration range, while they inhibited biofilm formation at higher concentrations (Figure 3A). Of note, this phenotype is in accordance with a putative function of PSMβ peptides in cell-cell disruption on the molecular level (3, 23): at lower concentrations, this process is thought to facilitate the formation of channels, which increases biofilm formation on the macroscopic level, whereas higher concentrations promote detachment and thus a reduction in biofilm mass (Figure 3B).
S. epidermidis in vitro biofilm formation under influence of PSMβ peptides. (A) Biofilm formation in microtiter plates (24 hours, 37°C). PSMβ peptides at different concentrations were added at the time of inoculation with S. epidermidis agr (devoid of PSMs). Biofilms were made visible using safranin staining (see example for PSMβ1 at the bottom), and biofilm formation was measured using an ELISA reader. Error bars depict mean ± SEM. (B) Schematic presentation of biofilm cell-cell disruptive processes leading to channel formation and detachment.
Biofilm formation in staphylococci has been shown to be exopolysaccharide dependent or independent (24, 25). To analyze whether PSMβ peptides mediate biofilm detachment in both exopolysaccharide-dependent and –independent S. epidermidis biofilms, we first determined presence of the ica genes responsible for production of the main biofilm exopolysaccharide polysaccharide intercellular adhesin (PIA, or poly-N-acetylglucosamine [PNAG]) in the 24 _agr_-negative S. epidermidis strains. Then we determined in vitro biofilm formation by these strains. Interestingly, biofilm-forming capacity was significantly higher among the strains that contained ica genes than among the _ica_-negative strains (Figure 4A), confirming the importance of PIA/PNAG for efficient biofilm formation in S. epidermidis (24). We selected the 2 _ica_-negative strains with the highest biofilm-forming capacity and 1 _ica_-positive strain, in addition to strain 1457, to confirm the universal importance of PSMβ peptides for biofilm detachment in vitro. In all strains, PSMβ1 had a biofilm detachment effect very similar to that observed with strain 1457 (Figure 4D), demonstrating that the impact of PSMβ peptides on biofilm detachment in vitro is not dependent on whether biofilm formation is exopolysaccharide dependent or independent.
Specific impact of PSMβ1 on in vitro biofilm detachment. (A) Biofilm formation by functionally _agr_-negative (without PSM production) _ica_-positive versus _ica_-negative strains. Biofilms were grown in microtiter plates for 24 hours and stained with safranin. ***P < 0.0001, t test. Colored symbols represent the strains used for the analyses shown in D. Error bars depict mean ± SEM. (B) α-Helical wheel presentation and (C) sequences of PSMβ1 and the mutated PSMβ1* peptide. The α-helical part (positions 17–44) is shaded in yellow. In the sequence, amino acids changed in PSMβ1* in comparison with PSMβ1 are shown in red; in the wheel presentation, exchanges are shown using red arrows. (D) Impact of PSMβ1 versus PSMβ1* on in-vitro biofilm development by 2 _ica_-positive and 2 _ica_-negative S. epidermidis strains. Conditions are the same as those used for the experiment shown in Figure 3A. *P < 0.05; **P < 0.01, t tests comparing values for the PSMβ1- versus PSMβ1*-treated samples at corresponding concentrations of added peptide.
To confirm the specificity of PSMβ detachment activity and analyze the importance of the PSMβ amphipathic α-helix in the observed detachment process, we synthesized a PSMβ1 derivative, PSMβ1*, in which 5 amino acids were altered to prevent the formation of an amphipathic α-helix (Figure 4, B and C). As the formation of an amphipathic α-helix is the basis of surfactant properties, these experiments were also conducted to gain insight into whether the detachment mechanism is dependent on the surfactant properties of PSMβ1. According to analysis by HeliQuest (http://heliquest.ipmc.cnrs.fr), the amphipathic α-helical part of PSMβ1 extends from amino acids 17 to 44 (C terminus). This was predicted by the hydrophobic face that is characteristic of an amphipathic α-helix and was found for 18–amino acid stretches comprising those, but not amino acids N-terminal of position 17 (stretches 1–18 to 16–43). In the C-terminal part of PSMβ1, 2 lysine residues (positive charge) were changed to proline residues (K22P, K40P), and 3 uncharged amino acids were altered to negatively charged amino acids (W20E, I27E, V35E) to abolish α-helicity and amphipathy in PSMβ1* (Figure 4, B and C). With PSMβ1*, the detachment effect was absent (in strains 1457, BM26, and BM33) or strongly decreased (strain BM32) compared with PSMβ1 (Figure 4D), indicating that the impact of PSMβ1 on biofilm detachment is specific and dependent on the PSMβ amphipathic α-helix. Furthermore, these results strongly suggest that the mechanism by which PSMβ peptides contribute to biofilm maturation and detachment is based on their surfactant properties.
Biofilm structuring and detachment by PSMβ peptides is likely accomplished during biofilm development, while mature S. epidermidis biofilms that are stabilized by an exopolysaccharide network may be largely resistant to macroscopic decomposition by externally added PSMβ peptides. To test this hypothesis, we added PSMβ1 peptide at 1 mg/ml to a 24-hour biofilm of the PIA/PNAG-producing strain 1457. We observed only a very minor impact on the biofilm (data not shown). In agreement with a surfactant-based mechanism of PSMβ detachment, these results suggest that PSMβ peptides must be present during biofilm development to promote biofilm structuring and detachment, while the presence of an established matrix network prevents the large-scale detachment of biofilm clusters by PSMβ peptides.
PSMβ peptides promote biofilm structuring and detachment in vitro. To validate the hypothesis that PSMβ peptides promote biofilm detachment, we analyzed biofilm development under flow using confocal laser scanning microscopy (CLSM). To measure expression of the _psm_β operon, we constructed a _psm_β promoter-egfp transcriptional fusion. The egfp gene was synthetically assembled using codon usage optimized for staphylococci (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI42520DS1) to increase fluorescence intensity. Expression of _psm_β was limited to the outer layers of the biofilm (Figure 5A) in accordance with a function in detachment and control by agr, for which we have previously shown a similar spatial expression pattern in S. epidermidis biofilms (26). Importantly, we observed that strong _psm_β expression was followed by the appearance of void spaces (Figure 5, B and C). Furthermore, there were significantly more _psm_β-expressing cells in the effluent compared with the biofilm cells (Figure 5D). These results confirmed our hypothesis that _psm_β expression leads to biofilm cluster detachment during biofilm development.
Role of PSMβ peptides in S. epidermidis biofilm development: cell detachment. A _psm_β-promoter egfp fusion construct (see Supplemental Figure 1 for the synthetic egfp gene with optimized Staphylococcus codon usage) was produced to monitor _psm_β expression in dynamic S. epidermidis biofilms using flow cells with CLSM (green). Entire biofilms were stained with propidium iodide (red). (A) Mature (24 hours) biofilm showing expression of _psm_β at the outer biofilm layer. (B and C) Biofilm cluster detachment observed during biofilm development. Lower (A) and higher (B) temporal and spatial resolution is shown. (D) Comparative analysis of the relative amount of cells expressing _psm_β (green fluorescence) in effluent and biofilm. The ratio of fluorescent versus nonfluorescent cells was determined using IMARIS software in effluent and biofilm samples. **P < 0.01, t test. Error bars depict mean ± SEM.
To further investigate the role of PSMβ peptides in S. epidermidis biofilm development, we constructed a deletion mutant of the entire _psm_β operon in S. epidermidis 1457 (Figure 2A). CLSM images revealed that static biofilms of the _psm_β operon deletion mutant in comparison with the isogenic WT strain lacked channel formation and were thicker (Figure 6, A and C). Genetic complementation restored the phenotype observed in the WT (Figure 6A). Image analysis of the biofilms using IMARIS software showed a significant increase in the total and average biovolumes, indicative of an increase in total biofilm formation and decrease in the degree of channel formation, respectively, comparing the mutant to the WT, while complementation restored WT levels (Figure 6B). These findings showed that PSMβ peptides promote channel formation in S. epidermidis biofilms. In addition, they further substantiated that PSMβ peptides facilitate detachment, as the lack of detachment leads to biofilm expansion.
Role of PSMβ peptides and agr in S. epidermidis biofilm development: channel formation and biofilm expansion. (A) CLSM pictures of S. epidermidis 1457 WT, isogenic _psm_β mutant, and _psm_β-complemented strains. The WT and _psm_β mutant strains were transformed with the control plasmid pT181mcs to ensure comparability. Static biofilms were grown for 24 hours, stained with propidium iodide, and imaged using CLSM. View is from the top. (B) Analyses of total and average biovolumes, which are measures of total biofilm and degree of channel formation, respectively, using IMARIS software of the biofilm samples shown in part A. Note that increased average biofilm volume corresponds to decreased channel formation. **P < 0.01; ***P < 0.001, 1-way ANOVA with Bonferroni’s post tests. Error bars depict mean ± SEM. (C) CLSM pictures of S. epidermidis 1457 WT, isogenic agr, and _psm_β operon deletion mutant biofilms. Growth conditions are as in A.
Furthermore, we compared the phenotype of the _psm_β deletion mutant with that of an isogenic agr mutant. The biofilm phenotypes of these strains were very similar (Figure 6C), indicating that PSMβ peptides represent the main effector molecules of quorum-sensing regulated biofilm maturation in S. epidermidis.
PSMβ peptides promote dissemination from in vivo biofilms. To analyze the role of β-type PSMs in biofilm detachment in vivo, we performed a murine model of device-related infection (Figure 7 and Table 1). The development of human sepsis is a rare event compared with the frequency of catheter infection and possibly due to yet unknown roles of host factors; it is thus difficult to mimic in an animal model. We therefore focused on analyzing bacterial spread from the catheters to the lymph and organs. Furthermore, we used immune-compromised Nu/Nu mice, as we showed previously that those mice develop more pronounced and longer lasting catheter-related infection (27).
Role of PSMβ peptides in the dissemination of biofilm-associated infection. Catheter pieces with equal cell numbers of preformed biofilms of either the S. epidermidis 1457 WT or isogenic _psm_β mutant strains were inserted under the skin of mice at the dorsum. Each mouse received 2 pieces, left and right, 1 of which was coated with WT and 1 with mutant biofilms. At days 2 and 4 after insertion (this timing found in a pilot experiment to be optimal), catheter pieces were removed and dissemination was analyzed. In a first experiment, body fluids (from the i.p. cavity) and organs were analyzed. Almost all disseminated bacteria detected were of the WT, and significantly more WT than mutant bacteria were detected in the body fluids and liver. Number of mice: n = 9 (day 2); n = 6 (day 4). In a second experiment, lymph nodes at day 2 after infection were analyzed and showed predominantly WT bacteria, except for in the nodes adjacent to the catheter infected with _psm_β mutant bacteria. Numbers represent percentages of WT among total bacteria from all tested mice (n = 7) in the respective lymph nodes. Blue, WT; green, _psm_β mutant strains. See Table 1 for details.
Catheter-related animal infection model
In this model, 2 catheters, precolonized with equal numbers of either the _psm_β mutant or WT strain, were inserted s.c. on the right and left dorsum, respectively. In a first experiment, we measured systemic dissemination into body fluids as assessed by taking body fluid samples from the peritoneal cavity. Bacteria harvested from body fluids were overwhelmingly and significantly more of the WT than _psm_β mutant strain (Table 1), indicating dissemination is favored in the WT over the mutant strain. Bacteria were rarely found in organs, but those that were found were almost exclusively of the WT. Owing to low and strongly varying numbers, differences in the organs only reached statistical significance in the liver sample at day 4 (Table 1). Finally, bacteria on the WT infected catheter were only of the WT strain, while on the _psm_β mutant–infected catheter, rare infiltration of WT bacteria was detected, in accordance with dissemination occurring only for WT bacteria.
In a second experiment, we focused on determining bacterial dissemination into the lymph nodes, as those would be first encountered by bacteria after dissemination. Bacteria in all lymph nodes were almost exclusively of the WT strain (Figure 7 and Table 1). Only bacteria in the right brachial and axillary lymph nodes were mainly of the mutant strain, most likely due to involvement of adjacent tissue during surgical insertion of the catheters. In contrast, no bacteria were found in the blood. This suggests that bacteria found in the body fluids from the peritoneal cavity in the first experiment originated from the lymph, while host defenses had cleared bacteria in the blood. Together, these data demonstrated that systemic dissemination from the indwelling device was strongly and significantly favored in the WT compared with the _psm_β deletion strain and thus, PSMβ peptides play a key role in the dissemination of biofilm-associated infection. Finally, results from a bacteremia control experiment indicated that PSMβ peptides do not have an impact on the development of disease that is unrelated to biofilm formation, as no significant difference was detected when monitoring death of animals injected with WT or _psm_β mutant bacteria, respectively (Supplemental Figure 2).
Antibodies to PSMβ peptides block dissemination of biofilm-associated infection. To further substantiate the involvement of PSMβ peptides in the dissemination of biofilm-associated infection, we investigated whether antibodies against PSMβ peptides prevent dissemination. To that end, we first produced antibodies to PSMβ1 and PSMβ2 in mice. PSMβ peptides proved immunogenic and elicited a strong IgG response (Supplemental Figure 3). Antisera were treated using extracts from the _psm_β deletion mutant, and the resulting blocked antisera were highly specific for PSMβ1 and PSMβ2, respectively (Supplemental Figure 3). We pooled the PSMβ1 and PSMβ2 antisera and analyzed the efficacy of the obtained serum in reducing the dissemination of biofilm-associated infection using the same device-related infection model. In contrast to animals treated with PBS or control serum, no bacteria were found in the livers, spleens, or kidneys of animals treated with the anti-PSMβ antibodies, and bacterial load in the lymph nodes was significantly reduced (Figure 8). Reduced bacterial counts were most likely not due to an opsonophagocytosis-enhancing effect of the anti-PSMβ antibodies, as in an in vitro control experiment, we did not observe enhanced bacterial killing with anti-PSMβ compared with control serum (Supplemental Figure 4). Thus, these results confirmed that PSMβ peptides are essential for the dissemination of biofilm-associated infection.
Blocking dissemination of biofilm-associated infection using anti-PSMβ serum. Specific antisera against PSMβ1 and PSMβ2 were mixed 1:1 and used to protect mice in the biofilm infection/dissemination model. *P < 0.05, versus PBS and control serum for the lymph nodes, versus PBS for the kidney results; 1-way ANOVA with Bonferroni’s post-tests. For spleen and liver, differences were not statistically significant owing to the fact that dissemination into organs only occurred in a limited number of mice and therefore SEM values were high. However, dissemination into the organs of mice that received anti-PSMβ serum was never observed (average CFU/100 μl ≤ 1). Number of mice: controls, n = 9; anti-PSMβ, n = 7. Control serum was from animals that received injections containing CFA and IFA. Error bars depict mean ± SEM.