S. aureus biofilm and planktonic-conditioned medium induced distinct responses in HKs in vitro. The adverse effects of planktonic bacterial cultures on mammalian cells have been well documented in vitro. Bacterial cells grown in broth cultures have long been assumed to retain the same pathogenic properties as bacteria in natural settings. While important discoveries have been realized based on planktonic studies, data presented here provide evidence that bacterial biofilms differentially induce pathogenesis in cultured HKs.
Host-pathogen interactions were investigated between a clinical isolate of S. aureus and HKs. A preliminary analysis of the extracellular proteome of S. aureus biofilm and planktonic cultures was performed by 1D gel electrophoresis and mass spectrometry. Several differences were observed in the 1D gel band patterns of BCM and PCM (Figure 1). The total protein concentrations of BCM and PCM were found to be similar, but BCM clearly contained more features. Smearing of BCM in 1D gels was observed indicating possible bacterial protease activity, although such a protease was not identified by mass spectrometry (Table 1). S. aureus secretes a variety of proteases which are important in pathogenesis . The presence of such a protease could explain some of the observed effects in HKs after treatment with BCM or PCM.
Several 1D gel bands visible in PCM and not BCM contained glycolytic enzymes (Figure 1, Table 1). The presence of intracellular glycolytic enzymes in the extracellular proteome of S. aureus may be due to cell lysis, but cell wall associated glycolytic enzymes have been described for numerous pathogens, including S. aureus [25, 26]. Links between central metabolism and virulence in S. aureus have been described. In S. aureus, when carbon sources are plentiful, glycolysis is active while the tricarboxcylic acid (TCA) cycle is largely repressed . The TCA cycle has been described as a signal transduction pathway capable of regulating toxin production , adhesion synthesis and biofilm formation [29, 30], and antibiotic susceptibility . Additionally, S. aureus deletion mutants for the glycolytic enzymes gapA and gapB have been shown to have attenuated pathogenic capabilities . The presence of several glycolytic enzymes in PCM and not in BCM supports the notion that central metabolic processes are in different states in planktonic and biofilm cultures and that those different metabolic states likely have a large impact on the observed pathogenic effects on HKs described here.
Functional annotation clustering of upregulated transcripts revealed over-represented annotation clusters associated with response to bacteria, regulation of transcription, inflammation, and signal transduction (Figure 2). The gene ontology term "response to glucocorticoid stimulus" was interesting as glucocorticoids are anti-inflammatory hormones. Genes involved in cyclic adenosine monophosphate (cAMP) signaling were also interesting since cAMP is involved in several fundamental cellular processes and may be partially responsible for the observed effects induced by BCM. Functional annotation clustering of downregulated transcripts revealed over-represented annotation clusters associated with transcription and metabolism. The downregulation of genes associated with these processes may indicate a general cessation in BCM treated cells.
Transcriptional responses of HKs to BCM revealed the upregulation of pro-inflammatory genes, including transcripts for pro-inflammatory transcription factors, cytokines, and apoptosis related genes. Among these were members of the AP-1 family of transcription factors and regulators of the NFkB pro-inflammatory transcription factor, TNFAIP3 (A20) and NFkBIA. Expression of these genes indicated active regulation of the NFkB pathway. NFkB regulates the expression of many genes involved in immune and inflammatory responses (i.e. cytokine and chemokine genes) and often acts in synergy with AP-1 to mediate inflammatory responses [33, 34]. NFkB and AP-1 are activated by pro-inflammatory cytokines such as TNF-α and IL-1β which act through MAPK-dependent signal cascades resulting in the production of additional cytokines [35–38]. The transcription factor egr1, which was highly upregulated in BCM treated HKs, is also involved in the regulation of pathophysiologically important genes relating to inflammation, apoptosis, and differentiation [39–41]. The upregulation of these early response transcription factors indicates that four hours of treatment with BCM induces a swift inflammatory response in HKs relative to PCM.
We previously investigated BCM induced apoptosis and HK migration in a scratch wound model . In agreement with that study, S. aureus BCM induced apoptosis in HKs while PCM did not induce a significant amount of apoptosis. BCM mediated induction of apoptosis is discussed in detail in . This striking dissimilarity between PCM and BCM would undoubtedly have substantial impacts on several aspects of wound healing. Cytokine production induced by PCM and BCM were normalized to adherent non-apoptotic HKs.
ELISA analysis of cytokine production in HKs revealed that after four hours, BCM induced the production of more cytokines relative to PCM treated HKs. However, after 24 hours, BCM induced cytokine levels were weaker relative to cytokine production induced by PCM. Even though cytokine levels were normalized to non-apoptotic cells, it is important to note that early stage apoptosis may contribute to a general reduction in protein expression contributing to reduced cytokine levels. However, a reduction in MAPK phosphorylation indicates an alternative mechanism to early stage apoptosis for cytokine reduction.
Phosphorylation of the MAPKs JNK and p38 were found to be reduced by BCM while ERK was not. Inhibition of MAPK pathways revealed that MAPK signaling was responsible for a larger percentage of cytokine production in PCM treated HKs compared to BCM treated HKs. Even though there were strong differences in cytokine production between BCM and PCM treated cells after four hours, the representation of the inhibitor data as a percent of the vehicle control helps to reveal to what extent MAPKs are involved in cytokine production. SB203580, U0126, and SP600125 are widely used inhibitors of MAPKs. SB203580 and U0126 show a high degree of specificity towards p38 and ERK while the specificity of SP600125 towards JNK has recently been re-examined . SP600125 was found to inhibit a wider range of kinases than initially thought. Given our goal to determine a generalized relationship between MAPK signaling and cytokine production, the reduced specificity of the JNK inhibitor SP600125 was tolerable. A specific role for p38, ERK, and JNK in S. aureus biofilm mediated host responses remains to be elucidated.
Several studies have investigated the inflammatory effects of planktonic bacterial supernatants on mammalian cells [43–52]. Genes upregulated by PCM were in agreement with the upregulation of pro-inflammatory genes in epithelial cells exposed to planktonic S. aureus supernatant . Similar cytokine gene expression patterns were observed in human vaginal epithelial cells when exposed to late exponential phase S. aureus cultures . Mid-logarithmic-phase cultures of S. aureus planktonic-conditioned medium induced IL-6, CXCL-8, and TNF-α in human-corneal-epithelial cells . Different species of dental bacteria were found to induce various levels of the cytokines IL-1β, IL-6, and CXCL-8 after 4 or 24 hours of challenge in human gingival epithelial cells ; the ability of bacteria to induce cytokine production was correlated to the virulence of the strains tested.
Much less is known about the impacts of biofilm on mammalian cell cultures. S. aureus BCM initially induced higher levels of cytokines in HKs after four hours of exposure followed by reduced levels of cytokine production after 24 hours of exposure relative to PCM. The exception was TNF-α, which was found to be produced at higher levels in BCM treated HKs relative to PCM treated HKs. TNF-α is a cytokine capable of inducing apoptosis in many cell types including keratinocytes  and may be partially responsible for the observed increase in apoptotic HKs after exposure to BCM. In one in vitro host-pathogen model incorporating dental biofilms and human gingival epithelial cells, the cytokines IL-1β, IL-6 and CXCL-8 were degraded by the biofilm after four hours . In that study, direct contact with the biofilm was required for biofilm mediated degradation of cytokines as filtered biofilm supernatant similar to BCM did not induce the degradation of cytokines. Our results showed that direct contact with the biofilm was not necessary for the observed decreases in cytokine production after 24 hours of exposure. A recent study investigating the effects of S. aureus biofilm infection in a mouse model found adaptive immune responses were regulated through cytokine production as the biofilm matured . In that study, the production of key cytokines at certain times during the infection was hypothesized to manipulate the host's adaptive immune response resulting in localized tissue damage allowing S. aureus to establish a mature biofilm and mount a successful infection.
The patterns of cytokine and chemokine production from HKs exposed to either PCM or BCM are analogous to the patterns of cytokines produced during sepsis and chronic inflammatory diseases, respectively. Sepsis is characterized by release of massive amounts of cytokines and is analogous to the effects of PCM on cytokine production in HKs. Chronic inflammation, on the other hand, is similar to the effects of BCM where local inflammation is induced, but a runaway, self-inducing inflammatory response is not produced.
Three sub-types of MAPKs have been identified in mammals, ERK, JNK, and p38. JNK and p38 activation in HKs by PCM agree with other reports of JNK and p38 activation in mammalian cell cultures in response to bacterial cultures similar to the planktonic cultures described in this research [44, 56–60]. Suppression of JNK and p38 phosphorylation in BCM-treated HKs below that of control and PCM-treated HKs occurred after 4 hours. Transcriptional analysis of BCM-treated HKs revealed the upregulation of dual specificity MAPK negative regulators, which may be responsible for the de-phosphorylation of JNK and p38 (Additional file 1). ERK is involved in the regulation of differentiation, apoptosis, and motility . The activation of ERK may be associated with the regulation of these processes in HKs treated with BCM.
Chemical inhibition of MAPKs confirmed that PCM treatment induced more MAPK-dependent cytokine production than BCM in HKs after 4 hours of stimulation. The relative ineffectiveness of the MAPK inhibitors on BCM mediated cytokine production in addition to the reduced phosphorylation status of JNK and p38 suggests that BCM induces cytokine production through MAPK independent signaling mechanisms and the production of different factors by S. aureus biofilm compared to planktonic cultures.
The suppression of MAPK signaling by BCM could impact other wound-related activities involving MAPK cascades in HKs including HK differentiation , secretion of antimicrobial peptides , response to mechanical stress , and response to osmotic stress . Suppression of MAPK signal transduction in HKs would be detrimental to all phases of wound healing, possibly contributing to the formation and/or persistence of chronic wounds. The observed upregulation of pro-inflammatory transcription factors at four hours may be an attempt by the cell to compensate for reduced MAPK signaling. The consequence of the overproduction of pro-inflammatory transcription factors could be the cause for the greater production of cytokines in BCM-treated HKs at four hours. Several transcription factors are differentially regulated in BCM treated HKs. Certain transcription factors induce or inhibit AP-1. One such transcription factor is A20 which is known to activate AP-1 and inhibit activation of JNK . A20 was upregulated 3.09 fold in BCM treated HKs relative to PCM treated cells (Additional file 1). It is possible that other MAPK independent pathways are activated or inhibited by BCM mediated MAPK inactivation resulting in A20 expression, leading to the initial increase of AP-1 family transcription factors.
Guggenheim et al. found that cytokines were degraded by direct contact with an in vitro dental biofilm . The smearing of BCM proteins on 1D gels indicates the possible presence of a S. aureus protease that may be responsible for the degradation of excreted cytokines. However, the suppression of MAPK phosphorylation and MAPK independent production of cytokines in BCM treated HKs suggests that cytokine production is at least partially limited through this important signaling pathway. MAPK suppression in various mammalian cell types by bacterial toxins has been observed. Bacillus anthracis secretes lethal toxin, which cleaves most isoforms of MAPKs, reducing pro-inflammatory cytokine secretion from immune cells . Shigella flexneri, Yersinia spp., and Salmonella spp. deliver toxins which inhibit MAPK signal transduction through a type III secretion mechanism resulting in the repression of genes such as TNF-α, IL-6, and CXCL-8 [68, 69]. To our knowledge, a toxin has not been identified in S. aureus that inhibits MAPK signaling, but it is tempting to speculate that such a toxin exists and is responsible for the observed suppression of p38 and JNK phosphorylation. The results presented here provide the basis to characterize the response of HKs to BCM and allow the formulation and testing of hypotheses as to specific components in BCM that cause the observed HK response. Metabolomic and proteomic characterization of BCM are beyond the scope of the present work, but it is relevant to mention that preliminary MS and NMR-based metabolomics analysis revealed numerous metabolites specific to S. aureus BCM (Our unpublished observations).
A hypothetical mechanism of pathogenesis induced by S. aureus infection as related to this work is presented here. The initial infection of wounded tissue is assumed to be primarily by planktonic S. aureus. That infection could result in a normal inflammatory response where the invading bacteria are destroyed and the tissue progresses through a normal healing response. If the host were immune-compromised, had an underlying disease (i.e. diabetes, pulmonary disease, or other inflammatory diseases), or conditions were favorable for the pathogen, S. aureus could successfully evade the immune system. If S. aureus were successful in evading the host's immune response, the resulting infection could continue to spread, reach the bloodstream and induce sepsis, resulting in death (i.e. a planktonic S. aureus infection). Alternatively, S. aureus could revert to a biofilm growth phase where HK apoptosis and cytoskeletal rearrangements would inhibit the re-epithelialization of the wound  and a deranged inflammatory response could establish a localized, persistent infection.