Short chain fatty acids produced by Cutibacterium acnes inhibit biofilm formation by Staphylococcus epidermidis

Biofilm formation by bacterial pathogens is associated with numerous human diseases and can confer resistance to both antibiotics and host defenses. Many strains of Staphylococcus epidermidis are capable of forming biofilms and are important human pathogens. Since S. epidermidis coexists with abundant Cutibacteria acnes on healthy human skin and does not typically form a biofilm in this environment, we hypothesized that C. acnes may influence biofilm formation of S. epidermidis. Culture supernatants from C. acnes and other species of Cutibacteria inhibited S. epidermidis but did not inhibit biofilms by Pseudomonas aeruginosa or Bacillus subtilis, and inhibited biofilms by S. aureus to a lesser extent. Biofilm inhibitory activity exhibited chemical properties of short chain fatty acids known to be produced from C. acnes. The addition of the pure short chain fatty acids propionic, isobutyric or isovaleric acid to S. epidermidis inhibited biofilm formation and, similarly to C. acnes supernatant, reduced polysaccharide synthesis by S. epidermidis. Both short chain fatty acids and C. acnes culture supernatant also increased sensitivity of S. epidermidis to antibiotic killing under biofilm-forming conditions. These observations suggest the presence of C. acnes in a diverse microbial community with S. epidermidis can be beneficial to the host and demonstrates that short chain fatty acids may be useful to limit formation of a biofilm by S. epidermidis.

Chemical properties of the C. acnes metabolites that inhibit S. epidermidis biofilm formation. To identify metabolic products of C. acnes that can inhibit S. epidermidis biofilm formation we examined the chemical properties of CS from C. acnes ATCC29399. To exclude the possibility that the low pH of C. acnes CS was responsible for inhibition of the biofilm, we measured the media pH after the addition of C. acnes CS (Table 1A). 25% C. acnes CS acidified tryptic soy broth (TSB) medium from a pH of 7.2 to a pH of 6.0. However, acidification of TSB medium to a pH of 6.0 by the addition of hydrochloric acid did not inhibit S. epidermidis 1457 biofilm formation or cell growth ( Fig. 2A). Thus, media pH reduction by C. acnes was not responsible for inhibition of S. epidermidis biofilm production.
Stability analysis of the biofilm inhibitory activity produced by C. acnes further defined the chemical nature of the molecule(s) in the C. acnes CS with activity to inhibit biofilm (Table 1B). The inhibitory activity could not be precipitated from CS by the addition of ammonium sulfate and was resistant to inactivation by digestion with proteinase K or lysozyme. This suggested the bioactive compound(s) were not proteinaceous. Biofilm activity was resistant to heating in a sealed tube at 100 °C for 10 min but was lost when CS was lyophilized. Additionally, the biofilm inhibitory activity was retained after passage through a 500 Da MW filter. These results indicated that the bioactive molecule(s) produced by C. acnes were heat stable and volatile.
C. acnes is a facultative anaerobe that produces short chain fatty acids (SCFAs) when provided a carbon source such as glycerol 13 . These SCFAs are volatile, heat stable and resistant to proteases and thus matched well with the chemical properties of the biofilm-inhibiting activity in C. acnes CS. To determine if SCFA production by C. acnes correlated with inhibitory activity, we investigated if the addition of glycerol to C. acnes culture media increased the production of SCFAs. Indeed, CS of C. acnes grown in the presence of glycerol has greater potency for biofilm inhibitory activity compared to CS without glycerol supplementation (Fig. 2B). This observation further implied that SCFAs may inhibit S. epidermidis biofilm activity. SCFAs inhibit biofilm formation by S. epidermidis. SCFAs known to be produced by C. acnes include acetic acid, propionic acid, isobutyric acid, and isovaleric acid 13 . Therefore, to directly test the hypothesis that SCFAs can inhibit S. epidermidis biofilm, we added these pure SCFAs to S. epidermidis 1457 cultures. Similar to C. acnes CS, SCFAs inhibited biofilm formation at concentrations that did not inhibit cell growth (Fig. 3A). Of note, this inhibition occurred at physiologic concentrations of SCFAs produced by C. acnes on skin 13 , and was weakest for acetic acid, a SCFA produced by S. epidermidis as a metabolic byproduct. Furthermore, a mixture of SCFAs that mimicked the composition of SCFAs in C. acnes CS (acetic acid, 3.17 mM; propionic acid, 4.59 mM; isobutyric acid, 0.11 mM; isovaleric acid, 2.06 mM) strongly inhibited biofilm formation (Fig. 3B,C). These observations suggested the production of SCFAs by C. acnes inhibits the capacity of S. epidermidis to produce a biofilm. Figure 1. Cutibacteria inhibited the capacity of S. epidermidis to form a biofilm. (a) Culture supernatant (CS) of C. acnes ATCC29399 inhibited biofilm formation by S. epidermidis s 1457 as seen by crystal violet staining. C. acnes was cultured in reinforced clostridial media (RCM) and fresh RCM was used as control. C. acnes CS or RCM was added to a final concentration of 25% (v/v) during growth of S. epidermidis for 6 or 24 h. (b,c) Dose-dependent inhibition of biofilm formation but not cell growth by C. acnes CS when applied to S. epidermidis 1457 (b) or S. epidermidis clinical isolate (c). (d) CS of several species of Cutibacteria inhibited S. epidermidis 1457 biofilm formation. (e) CS of several strains of C. acnes inhibited S. epidermidis 1457 biofilm formation. Data were expressed as mean ± SEM of a single experiment (n = 6) that was representative of 3 independent experiments. Differences were analyzed using the unpaired Student's t test (b,c) or one-way ANOVA with Dunnett's test (d,e). Significance was shown as *P < 0.05, **P < 0.01, *** P < 0.001.  Figure 2. S. epidermidis biofilm formation is not observed at low pH but is increased during fermentation of C. acnes. (a) S. epidermidis 1457 was grown for 6 h in TSB culture medium at the indicated pH range following the addition of hydrogen chloride or sodium hydroxide. Biofilm formation compared to cell growth after the addition of C. acnes ATCC29399 CS or RCM as control were shown. Data are expressed as mean ± SEM of a single experiment (n = 6) that is representative of 3 independent experiments. Differences were analyzed using one-way ANOVA with Dunnett's test. (b) C. acnes ATCC29399 was cultured under anaerobic conditions with the addition of glycerol as a carbon source for fermentation. Sterile media from these cultures was then added at the indicated final concentrations to S. epidermidis 1457 culture. Biofilm formation assayed at 6 h was compared to the results with C. acnes ATCC29399 CS. Data are expressed as mean ± SEM of a single experiment (n = 6) that is representative of 3 independent experiments. Differences were analyzed using the unpaired Student's t test. Significance was shown as *P < 0.05, **P < 0.01, ***P < 0.01. www.nature.com/scientificreports/ C. acnes and SCFAs increase capacity of ampicillin and doxycycline to kill S. epidermidis. Since biofilm formation is associated with resistance to killing by antibiotics, we tested whether C. acnes CS would enable antibiotics to kill bacteria grown under conditions that would otherwise lead to formation of a biofilm. S. epidermidis 1457 was cultured with C. acnes CS and with increasing concentrations of ampicillin or doxycycline. After incubation for 6 h, S. epidermidis was killed at lower concentrations of ampicillin or doxycycline when grown with C. acnes CS compared to culture medium that was not conditioned (RCM) as a control (Fig. 4A). The same result was obtained with SCFAs ( Fig. 4B). This observation suggested that by inhibiting the biofilm formation, C. acnes or pure SCFAs can increase S. epidermidis susceptibility to antibiotics.

C. acnes and SCFAs inhibit polysaccharide-dependent biofilm formation by S. epidermidis. The
process of biofilm formation has at least two distinct phases: initial attachment by surface proteins and biofilm accumulation. The second phase requires cell-to-cell interaction that is mediated by intercellular polysaccharide adhesin (PIA aka PNAG). In the majority of S. epidermidis strains 14 , including S. epidermidis strain 1457, the production of polysaccharide is important for accumulation of the biofilm.
To determine if C. acnes acts at the phase of attachment or polysaccharide assembly, we stained culture plates during formation of the biofilm with probes to detect total protein, DNA or carbohydrate. At 2 h, protein deposition by S. epidermidis 1457 was unchanged by C. acnes or SCFAs, thus suggesting no effect on initial attachment (Fig. 5A). However, after 6 h, the amount of polysaccharide and DNA was markedly reduced (Fig. 5A). Considering that SYTO 9 stains both intracellular DNA and extracellular DNA, the reduction of SYTO 9 staining can be also interpreted as a reduction of adherent bacteria.
In addition, to directly test the effects of C. acnes CS and SCFAs on cell adhesion compared to assembly of the biofilm, we tested the Δica mutant strain of S. epidermidis 1457 which lacks the production of PIA but can still assemble a protein-based biofilm 15 . As expected, biofilm formation capacity by the S. epidermidis 1457 Δica mutant was less than in the wild-type strain. This biofilm formed by the Δica mutant was not further inhibited by C. acnes CS (Fig. 5B,C), thus demonstrating that the effect of C. acnes does not occur in absence of polysaccharide deposition. To further confirm the inhibition of polysaccharide production, we stained biofilm with periodic acid-Schiff (PAS). PAS is a staining method used to detect polysaccharides such as glycogen, and mucosubstances such as glycoproteins, glycolipids and mucins. As expected, C. acnes CS inhibited the production of Crystal violet **** **** **** **** **** **** **** **** **** **** **** *** Figure 3. SCFAs inhibit biofilm formation by S. epidermidis. (a) SCFAs as indicated were added to culture media of S. epidermidis 1457 at concentrations that did not inhibit cell growth. Biofilm formation at 6 h was assayed by crystal violet staining. (b) A mixture of SCFAs simulating the composition measured in C. acnes CS inhibited biofilm formation of S. epidermidis 1457. (c) Representative images of biofilm inhibition by SCFAs were shown. SCFAs or distilled water as a control were added to a final concentration of 25% (v/v) during growth of S. epidermidis for 6 or 24 h. Data were expressed as mean ± SEM of a single experiment (n = 6) that was representative of 3 independent experiments. Differences were analyzed using one-way ANOVA with Dunnett's test. Significance was shown as *P < 0.05, **P < 0.01, *** P < 0.001, **** P < 0.0001. www.nature.com/scientificreports/ PAS-positive substance that was also undetectable in biofilms by S. epidermidis 1457 Δica mutant (Fig. 5D,E). Consistent with this observation, the expression of icaA and icaD, two major genes involved in the synthesis of intercellular adhesin by S. epidermidis 16 , were significantly decreased in the presence of C. acnes CS. In contrast, the expression of accumulation-associated protein (Aap) was unchanged (Fig. 5F), which is consistent with our observation that the S. epidermidis Δica mutant biofilm is unaffected by C. acnes CS (Fig. 5A,B). Taken together, these results suggest that C. acnes CS may act directly or indirectly to inhibit the synthesis or assembly of polysaccharide in the biofilm, potentially through suppression of icaA and icaD expression.

Discussion
C. acnes is one of the most abundant commensals on human skin 7,17 . Other commensal skin commensal organisms such as specific strains of CoNS can kill S. aureus 18,19 or selectively inhibit the growth of C. acnes 20 but limited information has been found to suggest that C. acnes can benefit its host. In contrast, although typically present without deleterious effect, C. acnes can cause infection of implanted medical devices 21 , and is most frequently thought of due to its involvement in the pathogenesis of acne vulgaris 22,23 . This study sought to determine if C. acnes could benefit it's host by influencing the function of S. epidermidis to form a biofilm. We conclude that short chain fatty acids produced by C. acnes will limit biofilm formation by S. epidermidis. This observation

Control SCFAs
Control SCFAs Ampicillin (µg/mL) Doxycycline (µg/mL) epidermidis in TSB at 1 × 10 7 CFU/mL was cultured for 6 h with several concentrations of antibiotics. After the incubation total CFU was counted. Data were expressed as mean ± SEM of a single experiment (n = 3) that was representative of 3 independent experiments. Differences were analyzed using the unpaired Student's t test.
Significance was shown as *P < 0.05, **P < 0.01. www.nature.com/scientificreports/ may explain in part why highly abundant and dense growth of S. epidermidis in the human hair follicle does not typically result in formation of a biofilm.

Scientific Reports
To test the capacity of S. epidermidis to form a biofilm, we examined defined laboratory strains of S. epidermidis isolated from infection and clinical isolates obtained from healthy skin. Similarly, we examined multiple strains of C. acnes as well as other related bacterial species relevant to this issue. Our initial reference strain of C. acnes inhibited S. epidermidis isolated from health and disease equally well. Importantly, this occurred at concentrations of C. acnes supernatant that did not inhibit the growth of the opposing species and acted only before the biofilm was formed. Other strains of C. acnes as well as other major members of the Cutibacterium genus also prevented S. epidermidis from forming a biofilm. This suggests the activity produced from C. acnes is likely conserved across the genus. Furthermore, the action against biofilm formation was selective. C. acnes CS strongly inhibited biofilms by S. epidermidis, inhibited biofilms by S. epidermidis to a lesser extent, and did not inhibit biofilms by P. aeruginosa or B. subtilis. Although these findings cannot exclude the potential that some S. epidermidis strains could be resistant, or that some C. acnes may be inactive, our observations support a general conclusion that C. acnes can inhibit S. epidermidis biofilm formation. www.nature.com/scientificreports/ To determine the mechanism of biofilm inhibition, we considered the possibility that simple acidification of the environment by the C. acnes CS could be the source of activity. The pH at the surface of the skin is normally acidic, ranging in pH values of 4-6 24,25 . Many bacterial species, including S. epidermidis, also can produce substances that change the pH of the environment 26,27 , and lower pH has been associated with increased biofilm formation, not a decrease 28 . Thus, we considered it unlikely that low pH would be the mechanism of inhibition. Analysis of C. acnes culture medium showed a drop in pH from 5.5 to 4.85 after 14 days of anaerobic culture, and 25% mixture of C. acnes CS with TSB had a pH of 6.0. Since acidification of S. epidermidis media from 7.2 to 5.3 did not affect the formation of biofilm in our system, we conclude acidic pH is not a responsible for our observations.
We considered the possibility that C. acnes may produce a specific protein or peptide with the capacity to inhibit biofilm formation. Stability analysis of the biofilm inhibitory activity produced by C. acnes suggested this was likely not the case since the activity was volatile, protease resistant and heat resistant. Since prior reports have shown Cutibacteria can produce SCFAs 29 and some SCFAs have similar chemical properties to the observed bioactivity from C. acnes 30 , we tested if pure SCFAs could have an effect similar to C. acnes CS. These experiments showed direct addition of SCFAs had a similar action to C. acnes CS. Although these results do not rule out the potential that other bioactive C. acnes metabolic products, the totality of our observations strongly support the hypothesis that SCFAs may be at least part of the explanation for how C. acnes acts against S. epidermidis biofilms. Further work to define these as the cause, and understand the mechanism of action against S. epidermidis, is still needed.
Our results indicated that C. acnes CS did not inhibit the growth of S. aureus but prior studies have shown that SCFAs did inhibit S. aureus at concentrations over 250 mM 31 . The discrepancy with our results may therefore be due to the lower concentrations of pure SCFAs present in C. acnes CS and which we used in these experiments. However, other molecules in the complex C. acnes CS mixture may also influence our observations. For example, some C. acnes phylogroups encode biosynthesis genes for a thiopeptide with possible antimicrobial activity against S. epidermidis, which conversely secretes bacteriocins such as epidermin that kills C. acnes 32 . The antagonism between S. epidermidis and C. acnes is also noted in acne vulgaris, in which not only SCFAs but many factors like antimicrobial peptides secreted from keratinocytes have an impact 33 . Considering that S. epidermidis also produces SCFAs, further study of additional, unidentified factors other than SCFAs should be addressed in the future. One example is N-acetylcysteine, which inhibits the growth, adhesion, and biofilm formation of Gram-positive skin bacteria 34 .
One of the clues to understanding a mechanism of action for C. acnes to inhibit biofilm formation was the observation that pure SCFAs that are produced by C. acnes had a similar effect to C. acnes conditioned medium. SCFAs may have multiple beneficial effects and have been studied in the setting of the intestinal microbiome 35 and contribute to the reduction of luminal pH which could inhibit pathogenic microorganisms in gut 36 . SCFAs also can have direct antimicrobial activity 37,38 , can increase mucin production 39 , influence immune responses 40 and suppress calcium phosphate-induced itching through activation of IL-6/p-ERK signaling 41 . In the context of the present study we also observed that higher concentrations of SCFAs can inhibit S. epidermidis survival. Our observations add to this list and suggest that the production of SCFAs may activate host defense, inhibit bacterial survival or act to limit biofilm production. As these effects are dose dependent they will be influenced by the environment since hypoxic conditions within the follicle will favor greater production of SCFAs. Further study is needed to determine if activity observed from C. acnes is solely due to SCFAs, as well as the most relevant functions of SCFA in different specific contexts seen in epithelial biology.
The bacterial biofilm matrix is mainly composed of polysaccharides, proteins, nucleic acids and lipids 42 . Since S. epidermidis 1457 produces a significant amount of PIA-dependent biofilm, it is considered as an excellent model strain to understand icaADBC transcriptional regulation 43 . Regulation of biofilm formation may vary depending on the type of biofilm produced as well as the species of organism that produces the biofilm. A previous report suggested C. acnes could induce S. aureus biofilm formation by producing coproporphyrin III 44 . S. epidermidis was also reported to inhibit S. aureus biofilm formation and nasal colonization 45 . Our observations did not find lower biofilm formation with S. aureus as we did with S. epidermidis. These vastly different responses from two somewhat similar species of Staphylococci suggest that the mechanisms by which the products of C. acnes act on S. epidermidis are specific. We hypothesize that polysaccharide synthesis or assembly is a primary target for SCFAs and C. acnes CS and we are working to define this mechanism of action. A series of experiments supported this idea. Staining with SYTO 9 showed less staining (intracellular DNA and extracellular DNA) when SCFAs are added, but bacterial growth itself was not inhibited by these concentrations of SCFAs. This suggests that less bacteria were able to adhere and form a biofilm in the presence of SCFA rather than a decrease in DNA synthesis.
While the impact of SCFAs on epithelia is being gradually elucidated, little is known about how SCFAs interact with other microbes on skin. Our data add here a new level of insight and suggest that production of SCFAs by C. acnes is an important mechanism to maintain homeostasis of the microbiome in the cutaneous environment. This may be particularly important in the approximately 5 × 10 6 follicles present on adult human skin where the density of S. epidermidis is high and hair shafts are present. Such an environment might be expected to foster the frequent development of a biofilm. Despite high density colonization by S. epidermidis, biofilms rarely appear on healthy intact skin. We speculate the observations reported here may be one of the factors that limits biofilm formation and enables homeostasis between S. epidermidis and the host environment. Understanding of mechanisms to maintain the normal balance between humans and commensal microbes may be applicable for development of new strategies to prevent biofilm formation in wounds and medical devices. www.nature.com/scientificreports/

Methods
Experimental design. This study was designed to biochemically characterize the activity of C. acnes inhibition of S. epidermis biofilm formation. Pilot experiments were performed to determine the activity. Experimental replicates of at least three (indicated in figure legends) were performed and analyzed to determine statistical significance as defined by P < 0.05. Sample analysis was performed quantitatively in an unblinded manner and confirmed by at least three independent experiments as indicated in the figure legends.
Bacterial culture. Preparation  Preparation of Cutibacterium culture supernatant. All Cutibacteria species, including all the C.
acnes strains, were cultured in RCM media (Sigma-Aldrich, St. Louis, MO), anaerobically for 14 days 13 . Culture media was then centrifuged for 10 min and this media was then filtered through a 0.22 micron filter (Fisher Scientific, Waltham, MA) to produce culture supernatant (CS). In some experiments ammonium sulfate was added to C. acnes CS, and the solution was centrifuged at 10,000 g for 10 min. At the concentration of 60%, 70%, and 80% (w/v) of ammonium sulfate, precipitate was confirmed. The precipitate was collected and used for further analysis of anti-biofilm activity. C. acnes CS was also tested by lyophilization using SpeedVac Vacuum Concentrators (Thermo Fisher Scientific, Waltham, MA). Volatile portion of a sample was removed by evaporation.
For dialysis, C. acnes CS was centrifuged with cellulose membrane (Amicon Ultra Centrifugal Filters; Millipore Sigma, Burlington, MA) to determine the rough molecular weight of the activity. After confirming that the molecular weight was under 3,000 Da, flow-through from the column was set to the dialysis tubes (Float-A-Lyzer Dialysis Devices; Spectrum Chemical Manufacturing, New Brunswick, NJ), and dialyzed in a clean floating water for 24 h. The concentration of SCFAs produced by laboratory strains of C. acnes strain ATCC29399 was measured as previously determined 13 . Briefly, bacteria were cultured under anaerobic conditions for 14 days. SCFAs concentrations in culture supernatants were measured by gas chromatography-mass spectrometry after ethyl acetate extraction. Concentrations were as follows: acetic acid, 3.17 mM; propionic acid, 4.59 mM; isobutyric acid, 0.11 mM; isovaleric acid, 2.06 mM. All SCFAs were purchased from Sigma-Aldrich (St. Louis, MO).
Colony forming assay. S. epidermidis was inoculated into 3% TSB medium, and cultured at 37 °C overnight. Then, the culture was diluted in fresh TSB with 25% of C. acnes CS or RCM to 1 × 10 7 CFU/mL by 600-nm optical density. Ampicillin sodium salt (Sigma-Aldrich, St. Louis, MO) or doxycycline hyclate (Sigma-Aldrich, St. Louis, MO) with several final concentrations were also added. A total of 100 μL of each diluted culture were transferred to flat-bottom 96-well microtiter polystyrene plates in which a 5 mm plastic cover slip coupon was put inside. The plates were then incubated for 6 h at 37 °C without shaking. A coverslip was collected from the plates, and we extracted bacteria in biofilm using vortex mixer and sonication 51 . Colony forming unit was counted on trypticase soy agar plate. Periodic acid-Schiff colorimetric assay. Periodic Acid Schiff (PAS) Stain Kit (ab150680; Abcam, Cambridge, MA) was used to detect polysaccharide. The methods to quantify in a microtiter plate format is described elsewhere 52 . Briefly, after the formation of bacteria, 100 μL of periodic acid was added to the plate and incubated for 30 min. After the washing, 100 μL of Schiff 's reagent was added and incubated for 15 min. Absorbance was measured at 550 nm in a plate reader.