Innate Immunity of Adipose Tissue in Rodent Models of Local and Systemic Staphylococcus aureus Infection

Background. The role of adipose tissue in systemic inflammation during bacterial infection is unclear. Effects of Staphylococcus aureus infection on adipocytes in rodent models of experimental endocarditis and peritonitis, the impact of S. aureus infection on gene expression in epididymal and subcutaneous adipose tissue, and effects of S. aureus infection on the toll-like receptor-2- (TLR2-) cathelicidin pathway in vivo and in vitro were investigated. Material and methods. The rat model of catheter-induced S. aureus endocarditis and the mouse model of S. aureus-induced peritonitis were used for infection experiments, gene expression profiling in adipose tissue, and measurement of cytokines. 3T3-L1 adipocytes were analyzed for expression of the TLR2-cathelicidin pathway. Results. Upon systemic bacterial infection by S. aureus, there is a shift from anti- to proinflammatory cytokines in serum and in adipose tissue gene expression. The TLR2-cathelicidin pathway is increasingly expressed during adipocyte differentiation in vitro and is induced upon stimulation by synthetic lipopeptides. Conclusions. Systemic infection by Gram-positive bacteria induces proinflammatory transformation of adipose tissue sites distinct from infection sites, documented on the levels of gene expression and secreted mediators. The TLR2-cathelicidine pathway is expressed and highly inducible in adipocytes in vitro. Lipopeptides are important immune-modulators of adipocytes in both gene expression and protein secretion.

Staphylococcus aureus (S. aureus) is not only a commensal bacterium but also a highly effective pathogen. It represents the most frequently isolated human bacterial pathogen from a range of diseases including blood stream infections, endocarditis, sepsis, pneumonia, arthritis, osteomyelitis, and wound-, skin-, soft tissue-, and foreign body-infections [22]. Challenges in treatment include the severity of life threatening invasive S. aureus infections and resistance issues especially among methicillin-resistant S. aureus (MRSA) strains. Mortality in patients with S. aureus bacteremia 2 Mediators of Inflammation remains high [23]. Additionally, invasive S. aureus infections are also correlated with metabolic disorders and type 2 diabetes mellitus is associated with a higher risk for and poorer prognosis of S. aureus infections [24]. Additionally, obesity is (independent of diabetes) a risk factor for S. aureus colonization and infections following surgery [25,26].
S. aureus is able to infect adipocytes in vitro and to survive inside these cells for up to five days in a glucosedependent manner [27]. Adipocyte infection with S. aureus thereby decreased adiponectin and resistin release whereas visfatin, MCP-1, and IL-6 secretion were increased [27]. S. aureus is also able to attach to and internalize into adipose tissue-derived mesenchymal stem cells [28].
Most recently, Zhang et al. [17] published their data in Science journal showing that subcutaneous adipocytes in the skin release the antimicrobial peptide cathelicidin in response to a subdermal infection with methicillin-resistant S. aureus (MRSA) in mice. This mechanism was accompanied by adipocyte proliferation and hypertrophy as a result of increased adipogenesis driven by two transcription factors, peroxisome proliferator-activated receptor-(PPAR ) and zinc-finger protein 423 (ZFP423). Cathelicidin is produced in human and murine adipocytes also in the absence of infection and is induced by a high-fat diet. Cathelicidin inhibits the growth of S. aureus, acts proinflammatorily, and stimulates neutrophils [29]. However, obese and insulin resistant mice carrying missense mutations within the leptin receptor gene are more susceptible to S. aureus infections [17]. Similarly, humans suffering from obesity and type 2 diabetes mellitus are under higher risk of skin-and soft tissue-infections by S. aureus. This discrepancy [30] might be explained by an interference of the infection-adipogenesiscathelicidin pathway identified by Zhang et al. [17] with leptin and/or insulin receptor signaling pathway. These data open a new window for molecular research on the crossing point of two superhighways, innate immune system and metabolism [30][31][32][33].
Though not yet definitively proven, the recognition of S. aureus by adipocytes [30][31][32][33] might be mediated by the toll-like receptor-2 (TLR2) signaling pathway [30]. TLR2 has been shown to be expressed in adipocytes [9,18,19,[34][35][36] and to recognize lipopeptides (peptidoglycans) produced by bacteria such as S. aureus resulting in enhanced IL-6 expression [18]. Stimulation of murine and human adipocytes with the TLR1/2 ligand Pam3Cys and the TLR2/6 ligand MALP-2 differentially modulated the release of IL-6, IL-8, and MCP-1 [9,13,18,36,37], whereas resistin was either not affected or even downregulated nonsignificantly by both ligands [9]. It seems important to emphasize that the pattern of cytokine and chemokine release caused by lipopeptides is similar to that observed in experiments using viable staphylococci [27]. In accordance with these observations, TLR2 knockout mice have a reduced adipose tissue MCP-1 expression and macrophage infiltration [38]. It is currently unclear whether systemic infection by S. aureus also alters adipose tissue biology at sites distinct from the respective infection, such as peripheral adipose depots in the intraabdominal and inguinal compartments. Therefore, it was the aim of the present study to investigate (i) the effect of severe infection by S. aureus on adipocyte biology in two rodent models of experimental endocarditis and peritonitis; (ii) the impact of S. aureus infection on cytokine and adipokine expression in epididymal (intra-abdominal) and (inguinal) subcutaneous adipose tissue; (iii) the potential effect of S. aureus infection on the postulated adipose tissue-TLR2-cathelicidin pathway in vivo and in vitro in differentiated adipocytes under basal conditions and upon stimulation with the TLR2-agonistic lipopeptides Pam3Cys and MALP-2.  [39,40]. Catheterized rats ( = 7) were challenged intravenously with ∼10 ∧ 5 CFU S. aureus strain Newman 48 h after surgery. Heparinized blood was collected daily from each animal by tail vein puncture and plated quantitatively. Surviving rats were euthanized on day 3 after challenge, the hearts were removed and the correct positioning of the catheter and presence of endocardial vegetation were noted. Epididymal adipose tissue was harvested and was immediately flash frozen in liquid nitrogen.

Peritonitis
Model. C57BL/6 mice ( = 11) were challenged intraperitoneally (ip) with 5 × 10 ∧ 8 CFU S. aureus strain PS80 as described previously [41], and the animals were euthanized thereafter at the indicated time points. Heart blood was collected and the peritoneal cavity was lavaged with 5 mL PBS. The S. aureus CFU/mL peritoneal fluid and blood were determined by quantitative plate counts. Epididymal and subcutaneous adipose tissues were harvested and immediately flash frozen in liquid nitrogen.

Measurement of Serum Cytokine and Adipokine Concentrations by ELISA.
Blood was drawn and serum was prepared following standard procedures. The concentrations of resistin, adiponectin, and visfatin in rat and mouse sera were measured by ELISA. Resistin (mouse), leptin (mouse), IL-6 (mouse, rat), and adiponectin (mouse, rat) sandwich ELISA detection systems were purchased from R&D Systems, Wiesbaden, Germany (DuoSet5 ELISA development systems). Visfatin (Nampt)-ELISA Kit (mouse/rat) was a product from AdipoGen (distributed from Biomol, Hamburg, Germany).
Each sample was measured in duplicate by ELISA. Values are depicted as means ± double standard error of the mean (2x SEM).

Measurement of Cytokine and Adipokine
The standard curve method was used for quantification of the results obtained by real-time PCR. Expression of mRNA of investigated genes in each sample was normalized to the mRNA expression level of a housekeeping gene (GAPDH, 18S rRNA or -actin) by calculation of ΔCt. Expression differences between groups were calculated and depicted as ΔΔCt or as fold-expression in comparison to relative control group.

2.7.
Statistics. Data are expressed as mean values ± 2x SEM. Differences between means were tested for significance using the Mann-Whitney U test. One-way ANOVA was used to test for statistically significant differences over time, and two-way ANOVA was used to compare two different groups over time. A value below 0.05 was considered to be statistically significant. For graphical illustration of cell culture experiments, box plot diagrams are shown.

Systemic Infection with S. aureus and Natural Disease Course of the Fatal Endocarditis Model in Rats.
We used the rat endocarditis model as a model for severe and systemic S. aureus infection. Infection in this model is ultimately fatal (Figure 1(a)) and rats were sacrificed after 72 h or whenever they appeared gravely ill. The average weight loss 48 h after infection was 8.2% ± 1.1% of the original body weight. Throughout the course of the experiments, the systemic spread of infection was reflected by continuously increasing levels of bacteremia (Figure 1(b), = 0.001 for trend).

S. aureus Endocarditis Alters the Systemic Cytokine and
Adipokine Profile in Rats. Although we observed ∼3-fold higher levels of the proinflammatory cytokine IL-6 in animals with endocarditis when compared to control animals, these differences did not reach statistical significance due to considerable variability of serum concentrations between animals within the subgroup of S. aureus infected rats (Figure 1(c)). In contrast, systemic levels of the anti-inflammatory adipokine adiponectin significantly decreased in serum from rats with endocarditis when compared to control animals (4670.5 ± 669.3 ng/mL versus 6844.3 ± 680.3 ng/mL at day 2 after infection and 4146.0 ± 669.5 ng/mL versus 7604.5 ± 709.7 ng/mL at day 3 after infection; = 0.004 in two-way ANOVA) (Figure 1(d)). In contrast to the anti-inflammatory adiponectin, the proinflammatory adipokines visfatin and leptin increased significantly. Visfatin concentrations (Figure 1(e)) were 5.03± 0.65 ng/mL versus 3.71 ± 0.31 ng/mL at day 2 after infection and 8.09 ± 2.59 ng/mL versus 4.33 ± 0.42 ng/mL at day 3 after infection ( = 0.007 in two-way ANOVA). Leptin concentrations (Figure 1

Nonfatal S. aureus Peritonitis Modulates Serum and Peritoneal Fluid Adipokine Concentrations Differentially.
In contrast to the endocarditis model we found no significant changes in adiponectin serum levels in S. aureus infected mice (Figure 2(d)). The initial decline at 48 h did not reach statistical significance. Peritoneal adiponectin concentrations also remained unchanged (Figure 2(d)). Visfatin levels, however, were increased over time in both infection models.
In the mouse peritonitis model, visfatin serum levels were 71.9 ± 25.6 ng/mL after 168 h compared to 24.2 ± 6.2 ng/mL after 24 h of infection and 25.8 ± 6.6 ng/mL in uninfected controls (Figure 2(e), = 0.004 in one-way ANOVA). We also found increased visfatin levels in peritoneal fluid of S. aureus infected mice ( < 0.001 for trend), albeit at a lower level. Resistin was the only adipokine found with similar concentrations in serum and peritoneal fluid of S. aureus infected mice (Figure 2(f)). There was no visible and significant trend of resistin levels during the course of infection in both compartments.

Systemic and Local S. aureus Infection Alters Gene Expression Levels in Adipose Tissue Depots Distinct from the Sites of
Infection. Infection with S. aureus altered expression levels of several genes investigated. Genes that showed at least a 2fold change in the initial screen or whose expression levels were significantly altered are summarized in Table 1 regarding  the endocarditis rat model and in Table 2 regarding the peritonitis mouse model. Genes expressed significantly stronger in epididymal adipose tissue from rats with S. aureus endocarditis (Table 1) compared to uninfected animals were the genes for PYCARD (PYD and CARD domain-containing), resistin, and protein tyrosine phosphatase-1 (Ptpn1). Genes expressed significantly lower in epididymal adipose tissue obtained from rats with endocarditis (Table 1) were the genes for the adiponectin receptor type 1 (AdipoR1) and phosphoenolpyruvate carboxykinase-1 (PEPCK-1). Although there was a trend towards lower PPAR expression, this trend did not reach a statistically significant level. The expression of the adipokines visfatin, leptin, and adiponectin was not different between the groups of experimental animals ( Table 1). The gene expression of markers specifically indicating induction of brown adipocyte differentiation ("browning") was investigated and was completely negative for the marker genes UCP-1, TMEM26, CD136, and TBX1 (Table 1).
In the peritonitis mouse model (Table 2), PYCARD, IL-6, and lectin-like oxidized-LDL receptor-1 (LOX-1) were significantly upregulated in epididymal and in subcutaneous adipose tissue at all time points, whereas MCP-1 was induced only in epididymal but not in subcutaneous adipose tissue. In contrast, resistin, leptin, and PPAR were downregulated in epididymal but not in subcutaneous adipose tissue (Table 2). Adiponectin, AdipoR1, and visfatin gene expression were not significantly altered (Table 2). Gene expression of markers specifically indicating induction of brown adipocyte differentiation ("browning") was investigated and there was no clear and consistent induction of the marker genes UCP-1 and CD137 (Table 2).

Expression of TLR2 in Adipose Tissues
Samples of Animals Infected with S. aureus. Since staphylococci and their lipopeptides are being recognized by TLR2, we investigated the expression of TLR2 on mRNA level (Tables 1 and 2) in the two murine infection models. TLR2 gene expression remained unchanged in rat epididymal adipose tissue and in mouse adipose tissues when compared to control animals.

Expression of TLR2 and Cathelicidin mRNA during 3T3-L1 Adipocyte Differentiation.
We studied expression of TLR2 and cathelicidin mRNA in the 3T3-L1 fibroblast cell line before and after induction of hormonal differentiation (day 0 to day 9) into mature adipocytes. As shown in Figure 3(a), cathelicidin expression is very weak in 3T3-L1 fibroblasts but strongly ( = 0.002) induced during adipocyte differentiation in a stepwise manner with highest levels in mature adipocytes at day 9 of differentiation.
Adipocyte di erentiation (days)    (Figure 3(f)) are also highly inducible in fibroblasts. In these cells, effects of MALP-2 were not documented as superior to those of Pam3Cys.

Effect of Lipopeptide Stimulation on Adipocytic Secretion of Adipokines, Chemokines, and Cytokines into the Supernatant.
Since TLR2 is expressed and upregulated during adipocyte differentiation and upon stimulation with lipopeptides, we studied the effect of lipopeptides on the secretion of adipokines (adiponectin, resistin, and lipocalin-2), chemokines (MCP-1), and cytokines (IL-6) into the supernatant of mature adipocytes and undifferentiated fibroblasts via ELISA (Table 3). In mature adipocytes, the secretion of the proinflammatory mediators MCP-1 and lipocalin-2 was significantly upregulated by each of the lipopeptides Pam3Cys and MALP-2, whereas the secretion of the antiinflammatory mediator adiponectin was significantly downregulated upon stimulation with Pam3Cys. Both lipopeptides were not significantly affecting resistin secretion. In undifferentiated fibroblasts which normally exhibit a more proinflammatory potential lacking the expression of classical adipokines such as adiponectin, secretion of IL-6 and MCP-1 was analyzed. Basal IL-6 concentrations were generally very low. Pam3Cys and MALP-2 were able to stimulate MCP-1 and IL-6 secretion significantly.

Discussion
Systemic inflammation by circulating Gram-positive bacteria such as Staphylococcus aureus or their respective bacterial lipopeptides such as MALP-2 and Pam3Cys might affect adipose tissue biology at sites located distinctly from primary infection. Particularly, systemic inflammation together with the exposure to bacteria or bacterial lipopeptides might cause adipose tissue inflammation. During this process, activation of TLR2 and nucleotide-binding oligomerization domain protein-1 (NOD1) could represent a reasonable molecular mechanism capable of altering adipose tissue biology [49]. In the present study, we could demonstrate a shift in secretion profile from anti-inflammatory adipokines/cytokines such as adiponectin towards proinflammatory adipokines/cytokines such as visfatin, leptin, IL-6, PYCARD, and MCP-1. Gene expression profiles were similar between epididymal and subcutaneous adipose tissue depots in the mouse peritonitis model (PYCARD, IL-6, adiponectin, AdipoR1, visfatin, and LOX-1). Only for MCP-1, resistin, and leptin, we observed a differential gene expression profile. Whereas MCP-1 was upregulated in epididymal adipose tissue at all time points, its expression remained unchanged in subcutaneous adipose tissue. Resistin was only downregulated in epididymal adipose tissue but not in subcutaneous adipose tissue. Finally, leptin remained unchanged in subcutaneous adipose tissue but decreased in epididymal adipose tissue. PPAR , a transcriptional and specific regulator of adipocyte gene expression, also remained unchanged in subcutaneous adipose tissue but decreased in epididymal adipose tissue. Since epididymal adipose tissue in mice resembles visceral adipose tissue in humans, this site-specific difference in gene expression might be related to the pathophysiological process of peritonitis.
It is a matter of debate why functional toll-like receptors such as TLR2 are expressed in adipocytes. However, from an evolutionary point of view this pathway might coordinate energy utilization and distribution during severe infection and immune response. Moreover, the direct defense of adipose tissue against bacteria such as staphylococci might represent an additional and evolutionarily conserved mechanism of the innate immune system. The latter mechanism seems highly reasonable and meaningful since adipocytes and adipose tissue often reside at sites and tissue interfaces involved in complicated infections and diseases like bowel perforation (mesenteric adipose tissue), acute pancreatitis (retroperitoneal adipose tissue), chronic inflammatory bowel disease (mesenteric adipose tissue), arthritis (periarticular adipose tissue), ophthalmitis (retro-orbital adipose tissue), soft tissue-infection (subcutaneous adipose tissue), and many others [11]. Thus, the TLR2-cathelicidin pathway seems to be of major interest. Most importantly, we could demonstrate a strong induction of cathelicidin and TLR2 gene expression in vitro in differentiating adipocytes. Moreover, two of the major TLR2 agonistic lipopeptides, Pam3Cys and MALP-2, were shown to induce cathelicidin and TLR2 expression with high significance in both fibroblasts and mature adipocytes. These in vitro data strongly argue for a physiological significance of these evolutionary highly conserved signaling pathways and open new prospects for future research on adipocyte innate immunity. In contrast to our in vitro data, in the rodent models of Staphylococcus aureus infection used, we could not demonstrate a significant induction the TLR2 gene expression. However, TLR2 gene expression might be already at high levels in these mice and does not further increase during infection.
Transformation of white adipose tissue into beige or brown adipose tissue, a process recently named "browning" [50][51][52][53], might be triggered by systemic inflammation. However, we could exclude this possible mechanism by showing that UCP-1, one of the most prominent and highly specific markers of brown adipose tissue, was not induced in adipose tissue in endocarditis animals. Furthermore, other markers indicating "browning" such as TMEM26 and TBX1 were also negative.
Recent studies elucidated the role of epigenetic mechanisms in altered adipocyte phenotypical and functional programing during inflammation, involving regulation of adiponectin expression by modified promoter methylation [54]. These are intriguing novel aspects on adipocyte transdifferentiation processes and should be addressed by future investigation with respect to intra-abdominal and inguinal adipose tissue.
A deteriorated adipogenic differentiation program or redifferentiation into more premature adipocytes (fibroblasts) could possibly be caused by systemic inflammation. However, the white differentiation markers adiponectin, visfatin, and leptin were not changed in our experimental animals. Moreover, we could demonstrate via gene expression profiling that the main transcriptional inducer of white adipocyte differentiation, PPAR , was induced in epididymal adipose tissue in the murine peritonitis model whereas its expression levels were not changed significantly in the endocarditis model as well as in cultured adipocytes treated with TLR2-agonistic lipopeptides. Additionally, expression levels of common markers of lipid and carbohydrate metabolism in mature adipocytes such as PEPCK, Ptpn1, and LOX-1 were not downregulated but significantly upregulated. These data argue against a disturbed differentiation program during systemic inflammation in these animals.
Taken together, it is intriguing to speculate based on our data that adipocytes might be involved in the clearing of bacteremia and in the defense against systemic infection by Staphylococcus aureus [17]. This implication is meaningful with respect to the development of novel antibiotics and to the understanding of the pathophysiology of soft tissueinfections [17]. Importantly, stimulation by lipopeptides in vitro caused a significant and differential regulation of adipokines and cytokines in adipocytes on the basis of protein secretion into the supernatants. Both Pam3Cys and MALP-2 upregulated the secretion of proinflammatory proteins whereas the secretion of anti-inflammatory adiponectin was downregulated by Pam3Cys. These data strongly argue for the TLR2 pathway being both inducible and functional in adipocytes since effects were not only seen on the basis of gene expression but also seen on the basis of protein secretion.
Most in vivo studies have focused so far on either animal models with endotoxemia (lipopolysaccharide) or seriously diseased patients with sepsis that are presumably more heterogeneous. Only few studies are available using defined infection models with a single and well defined pathogen and only few studies investigated the effect of systemic infection on changes in adipose tissue itself directly. To our knowledge, no study has been published so far addressing the molecular effects of systemic infection by S. aureus on adipose tissue sites located distinctly from the infection site.

Summary
Following systemic bacterial infection by Staphylococcus aureus, there is a shift from secreted anti-to proinflammatory adipokines and cytokines in serum and on the level of gene expression in adipose tissue at sites distinct from infection. In contrast, the adipocyte differentiation program or the inducible mechanism of "beiging" or "browning" seems not to play a role in this context. The TLR2-cathelicidin pathway is expressed and inducible in murine 3T3-L1 adipocytes and fibroblasts in vitro. Classical lipopeptides are able to induce TLR2 and cathelicidin gene expression as well as the secretion of proinflammatory mediators from adipocytes.

Conclusions
In addition to the well-known process of metabolic inflammation ("adipoflammation") in the context of insulin resistance, systemic infection by Gram-positive bacteria also induces proinflammatory transformation of adipose tissue. This effect is documented on the levels of gene expression and secreted mediators. The TLR2-cathelicidin pathway is expressed and highly inducible in adipocytes in vitro. Lipopeptides are important immune-modulators of adipocytes on the level of gene expression and protein secretion.

Conflicts of Interest
All authors declare no conflicts of interest.