Leptin and Pro-Inflammatory Stimuli Synergistically Upregulate MMP-1 and MMP-3 Secretion in Human Gingival Fibroblasts

Introduction Gingival fibroblast-mediated extracellular matrix remodelling is implicated in the pathogenesis of periodontitis, yet the stimuli that regulate this response are not fully understood. The immunoregulatory adipokine leptin is detectable in the gingiva, human gingival fibroblasts express functional leptin receptor mRNA and leptin is known to regulate extracellular matrix remodelling responses in cardiac fibroblasts. We therefore hypothesised that leptin would enhance matrix metalloproteinase secretion in human gingival fibroblasts. Methods and Results We used in vitro cell culture to investigate leptin signalling and the effect of leptin on mRNA and protein expression in human gingival fibroblasts. We confirmed human gingival fibroblasts expressed cell surface leptin receptor, found leptin increased matrix metalloproteinase-1, -3, -8 and -14 expression in human gingival fibroblasts compared to unstimulated cells, and observed that leptin stimulation activated MAPK, STAT1/3 and Akt signalling in human gingival fibroblasts. Furthermore, leptin synergised with IL-1 or the TLR2 agonist pam2CSK4 to markedly enhance matrix metalloproteinase-1 and -3 production by human gingival fibroblasts. Signalling pathway inhibition demonstrated ERK was required for leptin-stimulated matrix metalloproteinase-1 expression in human gingival fibroblasts; whilst ERK, JNK, p38 and STAT3 were required for leptin+IL-1- and leptin+pam2CSK4-induced matrix metalloproteinase-1 expression. A genome-wide expression array and gene ontology analysis confirmed genes differentially expressed in leptin+IL-1-stimulated human gingival fibroblasts (compared to unstimulated cells) were enriched for extracellular matrix organisation and disassembly, and revealed that matrix metalloproteinase-8 and -12 were also synergistically upregulated by leptin+IL-1 in human gingival fibroblasts. Conclusions We conclude that leptin selectively enhances the expression and secretion of certain matrix metalloproteinases in human gingival fibroblasts, and suggest that gingival fibroblasts may have an ECM-degrading phenotype during conditions of hyperleptinaemia (e.g., obesity, type 2 diabetes mellitus, exogenous leptin therapy).


Introduction
Gingival fibroblast-mediated extracellular matrix remodelling is implicated in the pathogenesis of periodontitis, yet the stimuli that regulate this response are not fully understood. The immunoregulatory adipokine leptin is detectable in the gingiva, human gingival fibroblasts express functional leptin receptor mRNA and leptin is known to regulate extracellular matrix remodelling responses in cardiac fibroblasts. We therefore hypothesised that leptin would enhance matrix metalloproteinase secretion in human gingival fibroblasts.

Methods and Results
We used in vitro cell culture to investigate leptin signalling and the effect of leptin on mRNA and protein expression in human gingival fibroblasts. We confirmed human gingival fibroblasts expressed cell surface leptin receptor, found leptin increased matrix metalloproteinase-1, -3, -8 and -14 expression in human gingival fibroblasts compared to unstimulated cells, and observed that leptin stimulation activated MAPK, STAT1/3 and Akt signalling in human gingival fibroblasts. Furthermore, leptin synergised with IL-1 or the TLR2 agonist pam2CSK4 to markedly enhance matrix metalloproteinase-1 and -3 production by human gingival fibroblasts. Signalling pathway inhibition demonstrated ERK was required for leptin-stimulated matrix metalloproteinase-1 expression in human gingival fibroblasts; whilst ERK, JNK, p38 and STAT3 were required for leptin+IL-1-and leptin+pam2CSK4-induced matrix metalloproteinase-1 expression. A genome-wide expression array and gene ontology analysis confirmed genes differentially expressed in leptin+IL-1-stimulated human gingival fibroblasts (compared to unstimulated cells) were enriched for extracellular matrix organisation and disassembly, and revealed that matrix metalloproteinase-8 and -12 were also synergistically upregulated by leptin+IL-1 in human gingival fibroblasts.

Introduction
Gingival connective tissue is predominantly composed of stromal cells, such as fibroblasts, and a collagen-rich extracellular matrix (ECM). Fibroblasts control the formation and remodelling of the gingival ECM, through regulated expression of ECM components and ECM-remodelling enzymes such as matrix metalloproteinases (MMPs) [1]. Fibrillar collagens are degraded by the collagenolytic MMPs (MMP-1, MMP-8 and MMP-13) [2]. Excessive destruction of ECM proteins, such as collagen, can be irreversible and is a feature of the common, chronic inflammatory disease periodontitis. Therefore, ECM remodelling is tightly regulated [3]. Exogenous proinflammatory stimuli such as bacterial lipopolysaccharide (LPS) are implicated in the pathogenesis of periodontitis [4] and up-regulate the expression of several MMPs in human gingival fibroblasts (HGFs) [5]. Additionally, the IL-6 family cytokine oncostatin-M (OSM), and IL-1 synergistically increase the secretion of MMP-1 by human gingival fibroblasts (HGFs) [6].
Obesity and type 2 diabetes mellitus are both positively associated with periodontitis [7], however the molecular mechanisms underpinning this relationship are poorly characterised [8]. The profile of circulating cytokines, growth factors and adipokines is altered in obese individuals and those with type 2 diabetes [9], and HGFs respond to many of these molecules. For example, the adipokine adiponectin partially suppresses IL-1-stimulated IL-6 and IL-8 production by HGFs [10]. Changes in the regulation, interactions and functionality of these mediators may link obesity, diabetes and periodontitis [8]. Circulating levels of the adipokine leptin are proportional to total adipose tissue mass, and the primary functions of leptin are to inform the brain of energy reserves and to regulate energy expenditure [11]. However, leptin also plays roles in angiogenesis, fertility, bone metabolism, immunity, wound repair and haematopoiesis, and is classified as a member of the IL-6 cytokine family [12,13]. Leptin has been detected in the gingiva by immunohistochemistry [14] and HGFs express functional leptin receptor (LEPR) mRNA [15]. Functionally, leptin increases the secretion of IL-6 and IL-8 by HGFs [15]. In this study we examined whether leptin, either alone or in combination with inflammatory mediators, could alter the expression of genes involved in ECM remodelling in HGFs, with a focus on the collagenase MMP-1, and the stromelysin MMP-3. We found that leptin enhanced the secretion of MMP-1 and MMP-3, and in combination with IL-1 synergistically up-regulated MMP-1, MMP-3, MMP-8 and MMP-12 expression by HGFs. Mechanistically, we found that leptin+IL-1-induced MMP-1 expression in HGFs was regulated by MAPK and STAT3 signalling.

HGF isolation and culture
Primary HGFs were derived from healthy gingival tissue obtained from patients with written informed consent undergoing canine tooth exposure surgery, with ethical approval from the National Research Ethics Service Committee North East (Reference: 07/Q1003/41). The tissue was collected on ice into F-12 Ham's medium, washed in PBS (both supplemented with 200 U/ml penicillin, 200 μg/ml streptomycin and 80 U/ml nystatin), dissected free of gingival epithelium and cut into 2 mm 3 sections. Sections were incubated (5% CO 2 at 37°C) in DMEM (containing 4.5 g/l glucose and supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM L-glutamine and 10% FBS) to allow HGF outgrowth. HGFs were maintained in DMEM as previously described [17] and examined for morphology and expression of the mesenchymal marker vimentin (Fig 1). Briefly, HGFs were fixed in methanol, then washed in TBS (pH 7.6), prior to incubation with mouse monoclonal anti human vimentin Ab (IgG2b, Clone Vim 3B4) (Dako) or diluent alone (3% FBS/TBS). After washing in TBS, HGFs were incubated in Envision polymer (Dako) for 30 min. HGFs were washed again in TBS before incubation in diaminobenzidine solution (Dako) for 5 min, then counterstained with haematoxylin. Excess diaminobenzidine and haematoxylin were removed by washing in water. HGFs between passages 5 and 9 were serum-starved for 18 h before a 30 min pre-treatment with chemical pathway inhibitors or DMSO (<0.1%), followed by stimulation with leptin (0.1-25 μg/ml), IL-1 (0.05 ng/ml), OSM (5 ng/ml), pam2CSK4 (50 ng/ml) and/or LPS (10 ng/ ml) for 24 h as described previously [18][19][20]. Inhibitors and DMSO were used at concentrations that were empirically demonstrated to have no adverse effect on HGF viability at 24 h (not shown).  Table 1. TaqMan assays for RNA polymerase II (RNAP) (POLR2A Hs00172187_m1) and Collagen Type 6A3 (COL6A3 Hs00915125_m1) were from Life Technologies. For realtime RT-PCR, relative gene expression levels were determined in cDNA samples using Taq-Man Universal Master Mix (Life Technologies) and the ABI Prism 7900HT sequence detection system (Life Technologies). Expression levels were normalised to RNAP. Primers used for conventional real-time RT-PCR are shown in Table 2. For RT-PCR, cDNA was amplified using Bio-Mix Red PCR reaction mix (Bioline, London, UK) and analysed on 3.5% agarose gels stained with Gel-Red (Biotium, Cambridge Bioscience, Cambridge, UK) using β2-microglobulin (β2m) or 18S rRNA as a reference gene.

ELISA, flow cytometry and immunoblotting
Culture supernatants from HGFs were analysed for total human MMP-1 and MMP-3 by ELISA (R&D systems) according to the manufacturer's protocol. These assays measure total MMPs (i.e. both proenzymes and active forms) Expression of LEPR on the cell surface of HGFs was analyzed by flow cytometry. 10,000 gated events were acquired on a FACSCalibur flow cytometer (Becton Dickinson, Oxford, UK). Data were analyzed with WinMDI 2.8 software (Joseph Trotter, The Scripps Research Institute, Purdue University, West Lafayette, IN, USA). Western blotting was performed as described previously [18].

Microarray transcriptional profiling
HGFs from three donors were stimulated with leptin (10 μg/ml) ± IL-1β (0.05 ng/ml) for 24 h in independent experiments and RNA was isolated using a RNeasy Mini Kit. Quality control, RNA amplification, hybridisation onto the HumanHT-12 v4 expression beadchip (Illumina, Little Chesterford, UK) and data generation were performed by Cambridge Genomic Services (Cambridge, UK). Table 1. Primer and probe sequences used for real-time RT-PCR.

Gene
Forward Reverse Probe doi:10.1371/journal.pone.0148024.t001 Table 2. Primer sequences used for conventional RT-PCR. Expression datasets were normalised using the robust spline method and principal component analysis was performed in R (Bioconductor, Seattle, WA, USA). Principle component analysis identified that the datasets from one donor were different to the others in one dimension (not shown). This variation was eliminated using ComBat/surrogate variable analysis [21,22]. The Human HT-12 v4 expression beadchip consists of > 47,000 probes and covers 31,000 genes. Differentially expressed genes for selected comparisons were determined as those having a corrected p value of <0.01 (as assessed using Limma with Benjamini-Hochberg correction for multiple comparisons) and a fold change 2. Gene ontology (GO) overrepresentation analysis (biological processes only) was performed using the GOstats Bioconductor library in R. The datasets used in this study are deposited in Gene Expression Omnibus (Ref: GSE68685) in accordance with Minimum Information About a Microarray Experiment (MIAME).

Statistical analysis
Statistical analysis was performed using SPSS 15.0. Shapiro-Wilk testing for normality and Levene testing for homogeneity of variance were performed prior to post hoc analyses of parametric and non-parametric data by Student's t test and Mann Whitney U test respectively. Real-time RT-PCR data were analysed using ΔCt values [23]. P values were corrected for multiple comparisons using the Bonferroni-Holm correction. A P value of <0.05 was considered significant.

Leptin upregulates MMP expression in HGFs
Leptin increased the expression of the collagenase MMP-1 in HGFs in a dose-dependent manner (Fig 2A; Table 3). Similar effects on MMP-1 protein secretion were also observed in HGFs ( Fig 2B). This increased MMP-1 secretion was observed in a time-dependent manner ( Fig 2C). The collagenase MMP-8 was also upregulated by leptin in HGFs (Table 3). In contrast, the collagenase MMP-13 was not induced by leptin (Table 3). Leptin had no effect on the expression of the gelatinases (MMP-2/MMP-9), the matrilysin MMP-7, the metalloelastase MMP-12 and tissue inhibitor of metalloproteinase (TIMP)-1-3, although most of these genes were expressed basally in HGFs (Table 3). Similarly, leptin had no effect on the expression of the stromelysins MMP-3 and MMP-10 in HGFs; however, the expression levels of MMP-3 were low and not consistently detected in all donors by RT-PCR. MMP-3 protein secretion in HGFs was increased by leptin though the absolute levels of expression were donor-dependent ranging between 51.9±5.2-1760±510 pg/ml (Fig 2D). Leptin increased the expression of the membrane-type MMP-14 in HGFs in a dose-dependent manner (Fig 2E). The related IL-6 family cytokine OSM, but not IL-1, pam2CSK4 or LPS, similarly increased MMP-14 expression ( Fig  2F and 2G).

Leptin activates MAPK, STAT and Akt signalling in HGFs
HGFs express cell surface LEPR and the long isoform of the LEPR (Fig 6A). Stimulation of HGFs with leptin, alone or in combination with IL-1 or pam2CSK4, activated multiple signalling pathways (Fig 6B). Leptin activated the MAPK pathway in HGFs, as evidenced by phosphorylation of p38, JNK and ERK (Fig 6B). Leptin and OSM, but not IL-1 or pam2CSK4, induced STAT1 and STAT3 tyrosine phosphorylation in HGFs. Similarly, phosphorylation of Akt (at S473) in HGFs was induced by leptin and OSM, but not IL-1 or pam2CSK4 (Fig 6B).

Multiple signaling pathways control leptin+IL-1-induced MMP-1 expression
To investigate whether these identified signaling pathways regulated the expression of leptininduced MMP-1, HGFs were cultured in the presence of selective pathway inhibitors. The ERK HGF cultures from 3 HGF donors stimulated in independent experiments (n = 4 from each donor). Remaining data (Fig 2D) are expressed as mean+SD from 3 individual HGF donors (n = 4 from each donor). *P<0.05, **P<0.01, ***P<0.001 compared to the unstimulated control at the same time point.    inhibitor U0126 significantly reduced MMP-1 expression in HGFs stimulated with leptin, IL-1 and pam2CSK4 both alone and in combination (Fig 6C). The JNK inhibitor SP600125 and STAT3 inhibitor VI significantly reduced MMP-1 expression in HGFs stimulated with leptin +IL-1 or leptin+pam2CSK4 (Fig 6D and 6E), whilst the p38 inhibitor SB203580 only significantly reduced MMP-1 expression in HGFs stimulated with leptin+IL-1 (Fig 6F). Akt inhibitor VIII at a concentration of 0.3 μM failed to inhibit MMP-1 gene expression in HGFs stimulated with leptin, IL-1 and/or pam2CSK4 for 24 h (not shown). Thus, multiple signaling pathways are involved in mediating leptin-induced MMP-1 expression in HGFs.

Leptin+IL-1 alter the HGF transcriptome towards an ECM-degrading phenotype
In order to assess the extent of genes involved in ECM remodelling altered by leptin+IL-1 in HGFs, we conducted genome-wide transcriptional profiling by microarray. Numerous genes relevant to ECM remodelling were significantly regulated in HGFs by leptin, IL-1 and   (Table 4). Among these genes, the most upregulated by leptin+IL-1 were MMP3 and MMP1, whilst several collagen genes (e.g. COL6A3, COL14A1, COL15A1) were down-regulated by leptin+IL-1 compared to the unstimulated control. Indeed, most of the genes related to ECM remodelling were only significantly regulated in HGFs stimulated with leptin+IL-1 together, rather than these mediators alone, suggesting a synergistic regulation. We found that several GO terms related to ECM remodelling were significantly overrepresented in the differentially expressed gene list for leptin+IL-1-stimulated HGFs compared to unstimulated controls (Table 5). HGFs from three donors were stimulated with leptin (10 μg/ml), IL-1β (0.05 ng/ml) or lep-tin+IL-1β for 24 h in independent experiments. RNA was extracted and used for genome-wide expression analysis (Illumina microarray). Genes displayed were significantly differentially expressed after leptin, IL-1 or leptin+IL-1 stimulation compared to unstimulated HGFs. The numbers displayed indicate fold change in stimulated HGFs as compared to unstimulated HGFs.-Indicates no change in gene expression.
These GO terms and example gene were significantly overrepresented in leptin+IL-1-stimulated HGFs compared to unstimulated HGFs.
The effect of leptin and IL-1 on the expression of several genes highlighted by transcriptional profiling (MMPs and COL6A3) was confirmed by real-time RT-PCR (Fig 7). Thus, leptin and IL-1 synergistically increased the expression of MMP-3, MMP-8 and MMP-12 in HGFs (Fig 7A-7C). We also confirmed that the expression of COL6A3 was significantly lower in HGFs stimulated with leptin+IL-1 compared to either leptin or IL-1 alone (Fig 7D). Finally, we confirmed that leptin+IL-1 enhanced MMP-14 and MMP-2 expression in HGFs compared to unstimulated cells (Fig 7E and 7F). However, increases in MMP-14 and MMP-2 expression in HGFs were not due to synergy between leptin and IL-1; MMP-14 expression was stimulated by leptin independently of IL-1 (as observed in Fig 2), whilst IL-1 stimulated MMP-2 expression independently of leptin. Although expression of leptin itself by HGFs was detected in the microarrays this was only detected at very low levels and was not influenced by any of the experimental treatments. Similarly, there was no evidence from the microarray experiments for any effect of leptin in IL-1 expression.

Discussion
HGF-mediated ECM remodelling is implicated in the pathogenesis of periodontitis [1], but the role of adipokines in regulating this effect is not fully understood. Leptin upregulates the secretion of IL-6 and IL-8 by HGFs [15] and in the current study we demonstrated for the first time that leptin upregulates MMP-1 and MMP-3 secretion in HGFs, alone and synergistically in combination with IL-1 or pam2CSK4.
doi:10.1371/journal.pone.0148024.g006 Table 4. Genes functionally related to ECM homeostasis and proteolysis that were differentially expressed in HGFs treated with leptin and/or IL-1. we found no differences in TIMP gene expression in HGFs stimulated by leptin, IL-1, OSM, LPS or pam2CSK4 suggesting that leptin could promote ECM remodelling by HGFs by increasing the active MMP:TIMP ratio.
In agreement with previous work investigating cytokine synthesis in HGFs [28,29] we observed that MMP responses to both LPS and pam2CSK4, and leptin-stimulated MMP-3 responses were qualitatively but not quantitatively similar between HGFs from different donors. These differences might be explained by genetic and/or epigenetic differences between individuals although this requires systematic investigation.
We demonstrated that leptin and IL-1 synergistically increase the expression of multiple MMPs (MMP-1, MMP-3, MMP-8, MMP-12) in HGFs. The fold change in MMP-1 expression in leptin+IL-1-stimulated HGFs in this study was approximately 10 times higher than that observed in a previous study in chondrocytes using the same concentrations of leptin and IL-1 [18]. Also, to our knowledge, this is the first report of synergy between leptin and a TLR2 agonist (pam2CSK4) in any primary human cells. These results indicate that the potential for HGFmediated ECM remodelling may be greatly enhanced under simultaneous hyperleptinaemic  and proinflammatory conditions, which could disturb ECM homeostasis and promote deleterious ECM degradation as is observed in periodontitis. Furthermore, leptin increases MMP-14 protein production and surface expression in primary rat cardiac fibroblasts [30]; similarly, leptin (and OSM) increased MMP-14 expression in HGFs in this study suggesting a role for IL-6 family cytokines in regulating pericellular ECM remodelling by HGFs.
All the short isoforms of the LEPR are expressed at the mRNA level by HGFs [15], including the long LEPR isoform in accordance with the data presented herein. Leptin activates the MAPK, JAK/STAT, and PI3K/Akt signalling pathways in HGFs; these are some of the well characterised signalling responses activated downstream of the long LEPR isoform [31]. Together with the finding that HGFs express cell surface LEPR, these data suggest that HGFs are sensitive and responsive to leptin.
ERK is known to regulate MMP-1 expression in IL-1-stimulated HGFs [32,33], and in leptin-stimulated chondrocytes/cartilage [18,26]. We showed that ERK regulates MMP-1 expression in HGFs stimulated with leptin, IL-1 and pam2CSK4 alone and in combination. Similarly, MAPK and STAT signalling pathways clearly regulate MMP-1 expression in HGFs stimulated by leptin+IL-1 as has been shown in chondrocytes [18]. We also showed that MMP-1 expression stimulated by leptin and pam2CSK4 is regulated by MAPK and STAT signalling. Multiple mechanisms, such as direct binding to the MMP-1 promoter [34,35], crosstalk between signalling pathways [36], and chromatin remodelling [37] could underpin the requirement for STAT3 in regulating MMP-1 expression in HGFs. Additionally, functional redundancy between STAT family members could add another dimension to this mechanism [38]. Interestingly, Akt inhibitor VIII did not inhibit leptin (+IL-1/pam2CSK4)-induced MMP-1 expression by HGFs, which could be attributed to a requirement to use lower concentrations of this inhibitor than other studies [18,39] due to a cytotoxic effect in HGFs (not shown). Together, our results suggest that leptin and pro-inflammatory mediators stimulate specific and coordinated gingival fibroblast signalling events that determine the production of the collagenase MMP-1.
Transcriptional profiling revealed that MMP-8 and MMP-12 were most upregulated in HGFs after stimulation with both leptin and IL-1 together. While MMP-8 in the periodontium is predominantly thought to be neutrophil-derived [40,41], there is evidence that other cells in the periodontium such as plasma cells, 'macrophage-like cells', sulcular epthielial cells as well as HGFs synthesise MMP-8 [25,41]. The elastase MMP-12 can activate itself and other MMPs through autolytic processing [42], providing a putative mechanism by which the initial activation of HGF-derived MMPs upregulated by leptin+IL-1 could occur. Interestingly, leptin+IL-1, as compared to leptin or IL-1 alone, synergistically regulated a wide range of genes relevant to ECM remodelling not limited to MMPs including collagens and other enzymes as well as certain cytokines and chemokines (unpublished observations). Studies that investigate the full extent of the HGF phenotype induced by leptin and the functional implications of this, in relation to ECM remodelling and to broader inflammatory responses in the gingiva are clearly required, and may help to address the possible redundancy between leptin and other IL-6 family members (e.g. OSM).
Circulating leptin concentrations are proportional to adipose tissue mass, and as such leptin has been shown to indicate organismal energy status to several tissues/organs [11] and also influence the immunocompetence of leukocytes [43]. We hypothesise that leptin could serve to inform HGFs of the energy status of an organism, thereby ensuring that the responses of HGFs (e.g. ECM remodelling) are in line with energy availability but are influenced by inflammatory context. For example, the presence of high concentrations of leptin in the inflamed gingiva (and hence a pro-inflammatory cytokine milieu) may synergistically drive HGF-mediated ECM degradation.
We conclude that leptin, in synergy with pro-inflammatory stimuli (IL-1 and the TLR-2 agonist pam2CSK4) selectively enhances MMP-1 and MMP-3 secretion in HGFs and suggest that gingival fibroblast-mediated ECM degradation, may be deleteriously enhanced during conditions of hyperleptinaemia (e.g. obesity, type 2 diabetes mellitus, exogenous leptin therapy).