Cardiac (myo)fibroblasts modulate the migration of monocyte subsets

Cardiac fibroblasts play an important role in the regulation of the extracellular matrix and are newly recognized as inflammatory supporter cells. Interferon (IFN)-γ is known to counteract transforming growth factor (TGF)-ß1-induced myofibroblast differentiation. This study aims at investigating in vitro how IFN-γ affects TGF-ß1-induced monocyte attraction. Therefore, C4 fibroblasts and fibroblasts obtained by outgrowth culture from the left ventricle (LV) of male C57BL6/j mice were stimulated with TGF-β1, IFN-γ and TGF-β1 + IFN-γ. Confirming previous studies, IFN-γ decreased the TGF-ß1-induced myofibroblast differentiation, as obviated by lower collagen I, III, α-smooth muscle actin (α-SMA), lysyl oxidase (Lox)-1 and lysyl oxidase-like (LoxL)-2 levels in TGF-β1 + IFN-γ- versus TGF-ß1-stimulated cardiac fibroblasts. TGF-β1 + IFN-γ-stimulated C4 and cardiac fibroblasts displayed a higher CC-chemokine ligand (CCL) 2, CCL7 and chemokine C-X3-C motif ligand (Cx3CL1) release versus sole TGF-ß1-stimulated fibroblasts. Analysis of migrated monocyte subsets towards the different conditioned media further revealed that sole TGF-β1- and IFN-γ-conditioned media particularly attracted Ly6Clow and Ly6Chigh monocytes, respectively, as compared to control media. In line with theses findings, TGF-β1 + IFN-γ-conditioned media led to a lower Ly6Clow/Ly6Chigh monocyte migration ratio compared to sole TGF-ß1 treatment. These differences in monocyte migration reflect the complex interplay of pro-inflammatory cytokines and pro-fibrotic factors in cardiac remodelling and inflammation.

Cardiac fibroblasts are centrally involved in myocardial remodelling 1 . While their main role is traditionally believed to be the regulation of the extracellular matrix proteins, their function as inflammatory supporter cells has attracted recognition 2,3 . During the development of heart failure, the immune system is activated locally in the myocardium as well as systemically. The following interaction of the different cellular and humoral components of the immune system with the cardiac innate cells orchestrates the myocardial remodelling and compensatory mechanisms 4 . There is currently a growing body of evidence that the immune system exhibits significant effects on cardiac fibroblasts by controlling their activation and transformation into myofibroblasts 3,5 .
TGF-β1, secreted by immune cells, cardiac fibroblasts and cardiomyocytes 6 , induces myofibroblast transdifferentiation and matrix deposition, characterized by the enhancement of collagen I, collagen III and α-SMA, and it is involved in cell proliferation, endothelial-to-mesenchymal transition 7 , immune suppression and inflammation 8 . Maintenance of the extracellular matrix and myofibroblast persistence is essential to preserve functional cardiac integrity after ischemic injury 9 . IFN-γ is the key cytokine in the Th1 reaction and is produced in large amounts by different cell types, mainly CD4 + helper T cells, CD8 + cytotoxic T cells and natural killer (NK) cells 10 . In a previous study, we demonstrated that increased levels of IFN-γ and an excessive Th1 reaction after myocardial infarction leads to decreased myofibroblast differentiation with detrimental effects in scar formation and myocardial remodelling 11 . This supports the general consensus that an extended or prolonged inflammation results in inadequate myocardial wound healing and scar formation associated with a suppression of systolic function and chamber dilation 12 .
Besides T and NK cells, monocytes also play a pivotal role in cardiac remodelling during viral-induced inflammation 13 and myocardial infarction 14 . In general, they can be divided into two main subsets identified by the expression of Ly6C. Pro-inflammatory monocytes are Ly6C high expressing cells, whereas their anti-inflammatory and reparative counterparts are characterized by Ly6C low expression. The chemokines CCL2 and CCL7 mediate Ly6C high monocyte recruitment, whereas Cx3CL1 governs Ly6C low infiltration into the myocardium 15 . In Coxsackievirus B (CVB) 3-induced myocarditis, Müller and co-authors could demonstrate that ablation of CX3CR1 resulted in an increased level of CCL2 accompanied by higher numbers of monocytes/macrophages 16 , further indicating the relevance of the chemokines CCL2 and Cx3CL1 in the attraction of monocytes.
Based on the finding that IFN-γ counteracts TGF-ß1-induced cardiac fibrosis 11 on the one hand, and the relevance of monocyte subpopulations during inflammation and wound healing on the other hand 15 , we sought to examine in vitro how IFN-γ can influence the chemokine profile of cardiac (myo)fibroblasts and subsequent attraction of pro-inflammatory versus anti-inflammatory monocytes.

IFN-γ reduces TGF-ß1-induced expression and activity of matrix metalloproteinase (MMP)-2.
Given the relevance of MMP-2 on CCL7 cleavage 18 , we next analyzed the effects of TGF-ß1, IFN-γ, and TGF-ß1 + IFN-γ on the expression and activity of MMP-2 (Fig. 4). First, we evaluated their impact on CCL7 mRNA expression and confirmed the observed regulation of CCL7 on protein level. In detail, TGF-ß1 treatment (G) α-SMA marked by a Cy3-conjugated streptavidin antibody (red) and DAPI for nuclear staining (blue). Recording of the images was done in a 100× magnification (scale bar = 100 μm). To characterize collagen crosslinking, the gene expression of the crosslinking enzymes (H) Lox-1 and (I) LoxL-2 were measured. Bar graphs represent the mean ± SEM of 3 independent experiments after normalization to the corresponding control fibroblasts. Statistical analysis was performed by One way-ANOVA or Kruskal Wallis test ( # p < 0.05 vs co, *p < 0.05 vs TGF-β1, § p < 0.05 vs IFN-γ).

IFN-γ in combination with TGF-β1 induces migration of pro-inflammatory monocytes.
To verify how the different CCL2, CCL7, and Cx3CL1 chemokine expression affects the attraction of different monocyte subsets, a migration assay was performed with splenocytes from control C57BL6/j mice towards the supernatants of TGF-ß1-and/or IFN-γ-stimulated C4 fibroblasts and cardiac fibroblasts (Fig. 5). The lower CCL2/Cx3CL1 and CCL7/Cx3CL1 ratio in the supernatant of TGF-ß1-stimulated versus control C4 fibroblasts was paralleled by a 1.1-fold (p < 0.05) higher ratio of Ly6C low towards Ly6C high migrated cells, indicating a lower number of attracted pro-inflammatory Ly6C high monocytes compared to anti-inflammatory Ly6C low monocytes (Fig. 5A). In contrast, the ratio of Ly6C low /Ly6C high migrated monocytes was 1.3-fold (p < 0.0001) lower versus the supernatant of TGF-ß1 + IFN-γ-compared to TGF-ß1-stimulated fibroblasts. In agreement to the observed results for C4 fibroblasts, supernatant of TGF-β1-stimulated cardiac fibroblasts also attracted more Ly6C low monocytes than Ly6C high monocytes (Fig. 5B), whereas TGF-β1 + IFN-γ conditioned media mainly induced the migration of Ly6C high monocytes.

Discussion
The prominent findings of the present study are that TGF-ß1 induces the production of CCL2, CCL7, and Cx3CL1 in cardiac fibroblasts, which results in an enhanced migration of (anti-inflammatory) Ly6C low monocytes. IFN-γ combined with TGF-ß1 induces a chemokine profile in cardiac fibroblasts in favor of attracting pro-inflammatory monocytes, as indicated by a reduction of the Ly6C low /Ly6C high monocyte ratio.
Myocardial remodelling is a complex process, which involves multiple different steps 19 . After an initial damage to the heart, the immune system is activated and coordinates the function of several cell types, both of immune and non-immune origin, thereby affecting the pathological remodelling of the heart and the progression to heart To determine the inflammatory phenotype after stimulation with TGF-ß1, IFN-γ, or both, the mRNA expression of (A) TNF-α, (B) IL-6, (C) ICAM-1, and (D) VCAM-1 was investigated. Bar graphs represent the mean ± SEM of 3 independent experiments after normalization to the corresponding control fibroblasts. Statistical analysis was performed by One way-ANOVA or Kruskal Wallis test ( # p < 0.05 vs co, *p < 0.05 vs TGF-β1, § p < 0.05 vs IFN-γ).
SCIeNTIFIC RepoRts | (2018) 8:5575 | DOI:10.1038/s41598-018-23881-7 failure 2 . In patients with chronic heart failure, serum levels of pro-inflammatory cytokines and chemokines are upregulated and correlate with the severity of the disease and the prognosis 20,21 . We previously demonstrated that IFN-γ decreases the differentiation of fibroblasts from the infarcted area in a mouse model of myocardial infarction, leading to decreased production of collagen I and III and impaired wound healing 11 . Based on this finding, we wanted to further explore the impact of IFN-γ on TGF-ß1-induced chemokine expression of cardiac fibroblasts and their subsequent potential of attracting monocyte subsets.
Cardiac fibroblasts get activated in the myocardium mainly by pro-fibrotic factors such as TGF-β1 22 and differentiate into myofibroblasts, which are characterized by high expression of α-SMA, increased production of extracellular matrix proteins, mainly collagens 2 , and upregulated expression of the crosslinking enzymes Lox-1 and LoxL-2, which on their turn promote fibroblast-to-myofibroblast activation [23][24][25] . In accordance with previous reports showing anti-fibrotic effects of IFN-γ 26-28 , we demonstrate that simultaneous stimulation of cardiac fibroblasts with TGF-β1 and IFN-γ resulted in lower gene and protein expression of collagens and α-SMA, and led to lower expression levels of Lox-1 as well as LoxL-2. Interestingly, Lox-1 is besides its role in crosslinking also known to resolve inflammation by reducing CCL2 expression 29 . In accordance to this negative correlation between Lox-1 and CCL2, higher CCL2 levels were observed in the conditioned media of TGF-ß1 + IFN-γ-versus sole TGF-ß1-stimulated cardiac fibroblasts. However, Lox-1 also has own chemoattractant potential 30 . This hints to the complex role of Lox in inflammation and monocyte attraction on the one hand, and further confirms cardiac fibroblasts as inflammatory supporter cells on the other hand 2,3 . Further investigations are needed to clarify the specific role of cardiac fibroblast-derived Lox-1 and/or LoxL-2 in the attraction of monocyte (subsets).
Besides CCL2, also CCL7 and Cx3CL1 levels were enriched in cell culture supernatant of TGF-ß1-stimulated cardiac fibroblasts compared to control fibroblasts. Since MMP-2 is known to degrade CCL7 31 , we evaluated whether the found CCL7 protein expression could be explained by changes in MMP-2 activity. Conforming Westermann et al. 22 , who demonstrated an increase in MMP-2 activity upon TGF-ß1 stimulation of cardiac fibroblasts derived from endomyocardial biopsies of HFpEF patients, TGF-ß1 increased MMP-2 expression and activity. In contrast to that, co-stimulation of TGF-ß1 with IFN-γ reduced MMP-2 expression and activity, which might explain the higher CCL7 expression under these stimuli versus TGF-ß1 stimulation. This finding is in line with the observation that reduced MMP-2 activity in viral-induced myocarditis was associated with increased levels of CCL7/MCP-3 and translated in exacerbation of cardiac immune cell infiltration 32 . In addition, we demonstrated that TGF-ß1 + IFN-γ further promotes an inflammatory phenotype of cardiac fibroblasts, as indicated by higher expression of the cell adhesion molecules ICAM-1 and VCAM-1, and of the inflammatory cytokine TNF-α compared to sole TGF-ß1 supplementation. To investigate the MMP-2 activity, zymography (C) of the cell culture supernatant was performed followed by quantification of proMMP-2 (D) and MMP-2 (E). Data are depicted as mean ± SEM from 3 independent experiments. Statistical analysis was performed by One way-ANOVA or Kruskal Wallis test ( # p < 0.05 vs co, *p < 0.05 vs TGF-β1, § p < 0.05 vs IFN-γ).
Regarding cell migration, TGF-ß1-stimulated cardiac fibroblasts exhibited lower CCL2/Cx3CL1 and CCL7/ Cx3CL1 ratios in their conditioned media compared to control fibroblasts. In accordance to the chemoattractant potential of CCL2 and CCL7 attracting pro-inflammatory Ly6C high monocytes, whereas Cx3CL1 mainly attracts anti-inflammatory Ly6C low monocytes 15,33 , this TGF-ß1-induced chemokine profile was reflected in the attraction of a higher number of anti-inflammatory Ly6C low cells compared to pro-inflammatory Ly6C high monocytes. Simultaneous stimulation with TGF-ß1 and IFN-γ promoted the pro-inflammatory phenotype, as shown by the higher ratio of CCL7/Cx3CL1 compared to sole TGF-ß1 treatment. This was reflected in the enhanced migration of pro-inflammatory Ly6C high monocytes towards this media. Similar results were observed for rat cardiac fibroblasts by which TGF-β1 conditioned media induced M2 macrophage polarization, whereas LPS-treated media favored M1 macrophage polarization 34 . These findings are further conforming own data from previous experimental CVB3-induced myocarditis mice, which exhibited elevated CCL2/Cx3CL1 and CCL7/Cx3CL1 ratios, and were associated with an increased cardiac presence of pro-inflammatory towards anti-inflammatory monocytes 13 . Finally, the TGF-ß1-and IFN-γ-induced chemokine profile of cardiac fibroblasts and subsequent attraction of pro-and anti-inflammatory monocytes corresponds to the sequential attraction of Ly6C high and Ly6C low monocytes post myocardial infarction 15 and their function in cardiac wound healing and scar formation 33,35 .
In conclusion, TGF-ß1 boosts chemokine expression stimulating the attraction of Ly6C low monocytes, whereas IFN-γ in conjunction with TGF-ß1, particularly attracts Ly6C high monocytes (Fig. 6). It is tempting to speculate that primed fibroblasts contribute to the repair process via attracting specific monocytes subsets, and that this homing is regulated via a balance between pro-inflammatory (IFN-γ) versus pro-fibrotic (TGF-ß1) factors. It is clear that during the pathogenesis of heart failure more cytokines and growth factors are involved in the cardiac remodelling process than the pro-inflammatory IFN-γ and the pro-fibrotic TGF-ß1. Nevertheless, this study is the first study to our knowledge, which points out the interplay among IFN-γ and TGF-ß1 on cardiac fibroblasts influencing the functional phenotype of the (myo)fibroblasts with respect to the attraction of monocyte subsets. However, further studies are warranted to get further insights into the complex regulation of fibroblasts in the inflammatory and remodelling process of distinct cardiac disorders.

Materials and Methods
Isolation and expansion of cardiac fibroblasts. Murine cardiac fibroblasts were obtained from the LV of 8-12 weeks old healthy male C57BL6/j mice (n = 6, Charles River, Sulzfeld, Germany), cut in approximately 2 mm large pieces and fixed in 12-well culture plates (n = 1/well). Therefore, mice were euthanized via cervical dislocation, which is approved by the local ethics committee (Landesamt für Gesundheit und Soziales, Berlin, T0098/11) and in accordance to the European legislation for the Care and Use of Laboratory Animals (Directive 2010/63/EU). The outgrowing cardiac fibroblasts were cultured in Dulbecco's modified eagle medium (DMEM; PAA, Cölbe; Germany) containing 20% fetal calf serum (FCS; Biochrom, Berlin, Germany), 100 U/mL penicillin and 100 µg/mL streptomycin (both PAA, Cölbe, Germany) at 37 °C with 95% air and 5% CO 2 . Stimulation experiments were performed with cells in passage 4 to 8. Culture and stimulation of C4 fibroblasts. Murine C4 fibroblasts, a murine fibroblast cell line derived from embryonic BALB/c mice by SV40 infection in vitro 36 , were cultured in Iscove medium (Sigma Aldrich, Munich, Germany) supplemented with 10% FCS (Biochrom) and 1% penicillin/streptomycin (Biochrom). For stimulation experiments, 60,000 cells/well were plated in 12-well-plates. One day after plating, cells were washed once with phosphate buffered saline (PBS; Thermo Fisher Scientific, Waltham, Massachusetts, USA), and starvation medium (Iscove medium + 0.05% FBS + 1% P/S) was added. After 24 hours (h), cells were stimulated with or without 5 ng/mL TGF-β1 (Peprotech, Hamburg, Germany), 10 ng/mL IFN-γ (Peprotech) or both in starvation medium. 72 h after stimulation, the supernatant was harvested, centrifuged (10 minutes (min) at 4 °C, 12000 × g) and aliquoted to further investigate chemokine secretion and perform chemotaxis assays.

Stimulation experiments of cardiac fibroblasts.
For stimulation experiments, murine cardiac fibroblasts were seeded out in 24-well culture plates (30,000 cells/well) for RNA isolation and immunofluorescence or 12-well culture plates (10,000 cells/well) for protein analysis and zymography. Stimulation with TGF-β1 (Peprotech) was chosen to induce differentiation of cardiac fibroblasts into myofibroblasts, mimicking the environment present during cardiac remodelling 37  Protein extraction and western blot analysis. After 72 h incubation time, supernatants were collected and cells were washed with PBS. Then, fibroblasts were lysed with 150 µL cell lysis buffer containing a protease and phosphatase inhibitor cocktail (both Sigma-Aldrich). Denaturized proteins were electrophoresed on 10% SDS-polyacrylamide gels, followed by transfer onto PVDF membranes and blocking with 5% skim milk for 30 min at room temperature. Blots were incubated over night with primary antibodies against collagen I (1:1000; Merck Millipore, Darmstadt, Germany) or GAPDH (1:1000; Cell Signaling, Frankfurt am Main, Germany) at 4 °C. The bound antibodies were detected by the respective horseradish peroxidase-conjugated secondary antibodies (1:2500, Cell signaling), which were visualized with super signal solution (Thermo Scientific, Darmstadt, Germany). The resulted chemiluminescence was measured by using the Chemo Star Imager camera system (INTAS Science Imaging Instruments, Göttingen, Germany). The relevant blots were analyzed by ImageJ software (NIH, Maryland, USA). For quantification of the protein levels, GAPDH was used as an internal control for each sample.
Zymography. Gelatin zymography from cell culture supernatant of stimulated cardiac fibroblasts was performed to determine the gelatinolytic activities of MMP-2 in response to TGF-β1 and IFN-γ. After 72 h incubation, supernatants were collected and mixed with 5× sample buffer. Samples were loaded on 10% SDS-polyacrylamide gels copolymerized with 1 mg/mL of gelatine (VWR, Darmstadt, Germany). Following electrophoresis, gels were washed in renaturing buffer with gentle shaking for 20 min. Digestion of gelatine was carried out twice in developing buffer, at first for 20 min at room temperature and then for 18 h at 37 °C. Next, the solution was replaced and gels were incubated again with cold developing buffer for 20 min. After development, gels were stained with 0.5% coomassie G250 blue solution (Sigma-Aldrich) and finally slowly destained with destain solution. Recording of images was carried out under UV light using the BioDoc 2.0 Imaging system (UVP, Cambridge, UK). ImageJ (NIH) was used for quantification of the bands and expressed as n-fold to the corresponding control.

ELISA.
To determine the secreted amount of CCL2 and CCL7 in the cell culture media of stimulated cells, the respective ELISA kits (both Peprotech) were used. In brief, ELISA microplates were coated overnight with the capture antibody (1:100) followed by 4 times washing with PBS + 0.05% Tween 20, and blocking with the blocking buffer for at least 1 h. Next, standards or samples were added to the wells for 2 h, followed by washing and incubation with the detection antibody (1:200) for additional 2 h. After washing, the avidin-HRP conjugate (1:2000) was added to each well. For color development, ABTS liquid substrate (Sigma-Aldrich) was used and the absorbance was measured at 405 nm. To determine the secreted amount of Cx3CL1, the murine Cx3CL1 ELISA kit (Abcam, Cambrigde, UK) was used. The absorbance was measured at the announced wavelength according to the manufacturer's protocol.
CytoSelect ™ cell migration assay. To determine the chemotactic potential of the supernatants of the (un)stimulated fibroblasts, a cell migration assay (Cell Biolabs, San Diego, USA) was performed according to the manufacturer's protocol. Therefore, splenocytes of 8-12 weeks old healthy male C57BL6/j mice (n = 10, Charles River) were isolated, pooled and placed at a density of 0.1 × 10 6 cells/ml on the upper membrane chamber. Under sterile conditions, 150 µL of the respective supernatants (=further named as conditioned media) were added to the wells of the feeder tray to explore chemotaxis. After 24 h incubation time, migratory cells were harvested and further analyzed by flow cytometry. For migration experiments versus conditioned media of C4 fibroblasts, migrated splenocytes of 8 wells/condition were pooled to obtain enough cells for subsequent flow cytometry. The migration assay was repeated 6 times/condition (n = 6). With respect to cardiac fibroblasts, migrated splenocytes of 4 wells/condition were pooled and performed three times (n = 3).