Fibroblast growth factor-2, but not the adipose tissue-derived stromal cells secretome, inhibits TGF-β1-induced differentiation of human cardiac fibroblasts into myofibroblasts

Transforming growth factor-β1 (TGF-β1) is a potent inducer of fibroblast to myofibroblast differentiation and contributes to the pro-fibrotic microenvironment during cardiac remodeling. Fibroblast growth factor-2 (FGF-2) is a growth factor secreted by adipose tissue-derived stromal cells (ASC) which can antagonize TGF-β1 signaling. We hypothesized that TGF-β1-induced cardiac fibroblast to myofibroblast differentiation is abrogated by FGF-2 and ASC conditioned medium (ASC-CMed). Our experiments demonstrated that TGF-β1 treatment-induced cardiac fibroblast differentiation into myofibroblasts, as evidenced by the formation of contractile stress fibers rich in αSMA. FGF-2 blocked the differentiation, as evidenced by the reduction in gene (TAGLN, p < 0.0001; ACTA2, p = 0.0056) and protein (αSMA, p = 0.0338) expression of mesenchymal markers and extracellular matrix components gene expression (COL1A1, p < 0.0001; COL3A1, p = 0.0029). ASC-CMed did not block myofibroblast differentiation. The treatment with FGF-2 increased matrix metalloproteinases gene expression (MMP1, p < 0.0001; MMP14, p = 0.0027) and decreased the expression of tissue inhibitor of metalloproteinase gene TIMP2 (p = 0.0023). ASC-CMed did not influence these genes. The proliferation of TGF-β1-induced human cardiac fibroblasts was restored by both FGF-2 (p = 0.0002) and ASC-CMed (p = 0.0121). The present study supports the anti-fibrotic effects of FGF-2 through the blockage of cardiac fibroblast differentiation into myofibroblasts. ASC-CMed, however, did not replicate the anti-fibrotic effects of FGF-2 in vitro.

To date, no therapies exist that prevent or reverse fibrosis in vivo, yet it is possible to antagonize TGF-β signaling with specific growth factors such as FGF-2. Fibroblast growth factor 2 (FGF-2) is relevant in wound healing processes in vivo [11][12][13] and in vitro [14][15][16][17] to stimulate proliferation of tissue cells, connective tissue fibroblasts and promote angiogenesis, while it suppresses apoptosis. Moreover, FGF-2 antagonizes TGF-β signaling and thus affects fibrosis, albeit in early stages 18 . In fibroblasts of various origins, FGF-2 suppressed the expression of TGFB1 19 and its protein 16,20 , that also suppressed the deposition of collagen 21,22 . All these studies suggest that FGF-2 is an attractive molecule to target in TGF-β regulated fibrotic disease.
Mesenchymal stromal cells (MSC), such as adipose tissue-derived stromal cells (ASC), as well as their conditioned medium, have been shown to improve cardiac remodeling and thus to modulate cardiac fibrosis [23][24][25][26][27][28][29][30] . Because ASC release a series of anti-fibrotic factors, among which are FGF, IGF and HGF [31][32][33][34] , a possible mechanism underlying their anti-fibrotic effect could be to antagonize TGF-β signaling and thus the inhibition of the transformation of fibroblasts into myofibroblasts as well as the reduction of extracellular matrix production.
In a previous study, we demonstrated that TGF-β1-induced differentiation of dermal fibroblasts to myofibroblasts could be modulated by adipose tissue-derived stromal cells' conditioned medium (ASC-CMed) 35 . Therefore, we hypothesized that cardiac fibroblast differentiation into myofibroblast could also be abrogated by ASC-CMed.

Results
Formation of F-actin stress fibers in differentiating NHCF-V is refractory to suppression by ASC-CMed. After five days of pro-fibrotic stimulus, NHCF-V undergo myofibroblast differentiation, phalloidin detected F-actin and cells had an increase in αSMA expression (Fig. 1). Human cardiac fibroblast without TGF-β1 stimuli had a mild phalloidin staining distributed all over the cytoplasm, on the other hand, the TGF-β1 stimulated cells had the detection of phalloidin in the contractile fibers which had colocalization with the αSMA staining. FGF-2, but not ASC-CMed, could inhibit the phalloidin and αSMA in the contractile fibers. The expression of αSMA seems to be more evident in the TGF-β1 stimulated cells. As expected, TGF-β1 induced myofibroblast differentiation of cardiac fibroblasts which leads to the formation of smooth actin fibers at cytoskeletal level. Only FGF-2 inhibited the formation of these fibers. FGF-2, but not ASC-CMed, inhibits the expression of mesenchymal markers in NHCF-V stimulated with TGF-β1. The expression of mesenchymal genes TAGLN, encoding SM22α, and ACTA2, encoding αSMA, was increased TGF-β1 stimulated fibroblasts, irrespective of ASC-CMed (Fig. 2). FGF-2 suppressed the expression of TAGLN (One-way ANOVA, p < 0.0001; Sidak's multiple comparison test, p < 0.0001) and ACTA2 in cardiac fibroblasts stimulated with TGF-β1 (One-way ANOVA, p = 0.0007; Sidak's multiple comparison test, p = 0.0056). The influence of ASC-CMed no more than tended to decrease TAGLN expression (One-way ANOVA, p < 0.0001; Sidak's multiple comparison test, p = 0.0820) but not ACTA2 expression.
TGF-β1 upregulated the expression of αSMA in human cardiac fibroblasts which was blocked by FGF-2 ( Fig. 3) (One-way ANOVA, p = 0.0413; Sidak's multiple comparison test, p = 0.0338). ASC-CMed did not alter the upregulation of αSMA by TGF-β1 (Fig. 3). Under these culture conditions, cardiac fibroblasts had a basal expression of SM22α which increased 1.4-fold after TGF-β1 stimulation. Thus, although FGF-2 inhibits the production of αSMA, it could not reverse the expression of the already formed SM22α.
FGF2, but not ASC-CMed, modulates extracellular matrix production in NHCF-V stimulated with TGF-β1. The gene expression of collagens, as well as of matrix metalloproteinases (MMPs) -enzymes responsible for ECM degradation -and the tissue inhibitors of metalloproteinases (TIMPs), was analyzed. The stimulation with TGF-β1 upregulated the transcription of both COL1A1 and COL3A1, genes responsible for the protein synthesis of the alpha-1 chain of collagens I and III respectively (Fig. 4A,B). This increase was more pronounced for COL1A1. Treatment with FGF-2 in samples stimulated by TGF-β1 was responsible for a statistically significant downregulation in COL1A1 (One-way ANOVA, p < 0.0001; Sidak's multiple comparison test, p < 0.0001) and COL3A1 (One-way ANOVA, p = 0.0572; Sidak's multiple comparison test, p = 0.0290), demonstrating the strong inhibition of NHCF-V towards the myofibroblast phenotype. The gene expression of MMPs and TIMPs did not change irrespective of treatment, except for FGF-2 ( Fig. 4C-G). Expression of MMP1, encoding matrix metalloproteinase1 i.e. collagenase, was downregulated by TGF-β1. In contrast, FGF-2 upregulated MMP1 expression compared to control groups and TGF-β1 stimulation (One-way ANOVA, p < 0.0001; Sidak's multiple comparison test, p < 0.0001). Treatment with ASC-CMed did not affect the TGF-ß-induced downregulation of MMP1 in NHCF-V. The expression of MMP2 gene was not regulated by TGF-ß, FGF or ASC-CMed (co)stimulation of NHCF-V. Although TGF-β1 did not affect the expression of MMP14, FGF-2 upregulated its expression (One-way ANOVA, p < 0.0001; Sidak's multiple comparison test, p = 0.0027). Treatment with ASC-CMed tended to increase the expression of MMP14.  Expression of TIMPs genes TIMP1 and TIMP2 which are regulators of MMP activity had differential expression. Expression of TIMP1, remained unchanged irrespective of TGF-ß1 or ASC-CMed treatment. Expression of TIMP2 was decreased by FGF-2 (One-way ANOVA, p = 0.0005; Sidak's multiple comparison test, p = 0.0023), while neither TGF-ß1 nor ASC-CMed affected its expression. Interestingly, TIMP2 is a co-factor of membrane-bound MMP14, which is e.g. responsible for activation of ECM-bound MMPs.
Immunofluorescence microscopy of intracellular collagen I demonstrated an increase in the protein content for all the groups stimulated with TGF-β1 (Fig. 5). FGF-2 could not decrease the intracellular collagen I at the microscopical level, so that virtually all the TGF-β1 stimulated cells and ASC-CMed/TGF-β1 expressed the protein in their cytoplasm. Both groups without TGF-β1 stimulation showed a very limited expression of collagen I.
FGF2 and ASC-CMed restore the TGF-ß1-inhibited proliferation of NHCF-V. NHCF-V proliferation was measured after five days of stimulation with TGF-β1. The pro-fibrotic stimulus led to a decrease in cell proliferation, as detected by Ki-67 staining. TGF-β1 suppressed the proliferation of human cardiac fibroblasts; control cells had 40.8% of proliferating cells compared to 22.3% in TGF-β1 stimulated cells (Fig. 6). Treatment with FGF-2, recovered the proliferation to 37.8% (One-way ANOVA, p < 0.0001; Sidak's multiple comparison test, p = 0.0002) while the proliferation of fibroblasts cells stimulated with TGF-β1 was virtually rescued by Wound healing is not affected by NHCF-V stimulation with TGF-β1. Wound healing was examined with a scratch assay and assessed after 24 hours. The percentage of gap closure did not differ among all groups (One-way ANOVA, p = 0.3716). Still, the treatment with FGF-2 or ASC-CMed tended to increase wound healing potential (Fig. 7).

Discussion
We demonstrated that FGF-2 -but not the secreted factors of adipose tissue-derived stromal cells -downmodulate TGF-ß1-induced fibroblast into myofibroblast differentiation. Our results show that FGF-2 reduced differentiation via the reduction of mesenchymal gene expression (TAGLN and ACTA2), and the reduction of αSMA expression subsequently. Concomitantly, we demonstrated that FGF-2 downregulated expression of collagens (COL1A1 and COL3A1), upregulate expression of matrix metalloproteinases (MMP1 and MMP14), and downregulated expression of inhibitor of metalloproteinase TIMP2. Finally, we showed that not only FGF-2 but also ASC-CMed restored the proliferation of TGF-β1-impaired fibroblasts. TGF-β1 signaling is a pivotal mechanism of the activation of fibroblasts into myofibroblasts and, thus, fibrosis [6][7][8][9][10] . The ability of FGF-2 to antagonize TGF-β signaling, in turn, has been demonstrated in a series of studies 17,36-39 , each of these suggesting different mechanisms of interference on the TGF-β pathway. Among these suggested mechanisms are: the activation of ERK and JNK pathways 17 ; the expression of let-7 miRNA, TGFβ  We hypothesized that, because ASC are known to secrete pro-regenerative growth factors such as FGF-2, treatment of human cardiac fibroblasts with ASC-CMed would abrogate TGF-β1-induced differentiation into myofibroblasts. Both gene and protein expression of mesenchymal and extracellular matrix-related markers showed that, differently than expected, ASC-CMed did not exert a noticeable effect on TGF-β1-induced myofibroblast activation. We did find, however, that ASC-CMed restored fibroblast proliferation impaired by TGF-β1 stimulation. This finding might point to potential differences between the FGF-2 thresholds required for restoring proliferation and blocking differentiation. However, another possibility is that ASC produce other growth factors that also impact proliferation but are not involved in the differentiation process. In this regard, it has been previously shown that epidermal growth factor (EGF) from ASC was capable to enhance the proliferation of skin fibroblasts in vitro 41 .
To our knowledge, this is the first report to investigate the effects of conditioned medium from mesenchymal stromal cells -here ASC -in the TGF-β1-induced differentiation of cardiac fibroblasts into myofibroblasts. Several other studies, however, have investigated the effects of MSC secretome in the spontaneous differentiation of cardiac fibroblasts [42][43][44][45][46] . Thus, although our findings oppose to many of the results described in these studies, there are important differences -which deserve attention -between published data and our current work (Table 1).
Three out of the five studies reported anti-fibrotic effects of MSC conditioned-medium on cardiac fibroblasts, including a reduction in expression of αSMA, collagens and TIMPs and increased MMPs expression and/or activity 43,45,46 . These studies were all performed with rat -not human -cardiac fibroblasts and none of these included a control group treated with Fibroblast Growth Medium (FGM), a medium capable to inhibit the spontaneous differentiation of fibroblasts, as we did. Thus, in this respect our results are not comparable to these studies for several reasons, the main being that the induction of cardiac fibroblast differentiation with TGF-β1 is a much more effective and powerful mechanism than the spontaneous differentiation, being consequently much more difficult to be blocked or reversed. Furthermore, both other studies which investigated the effects of MSC conditioned-medium on the spontaneous differentiation of cardiac fibroblasts reported either no differences 42 or even a trend towards the pro-fibrotic phenotype, with reduced MMP2 and MMP9 protein expression and activity and increased TIMP1 protein expression 44 . An important observation is that the concentration of FGF-2 in the ASC-CMed could be too low to counteract TGF-β1 action, while it is enough to block spontaneous differentiation. Previous studies demonstrated that the concentration of FGF-2 in ASC-CMed was less than 150 pg/ mL 33,34 , much lower than the FGF-2 concentration necessary to block the TGF-β1-induced differentiation in this study. Another point which deserves attention is the variety of factors secreted by ASC which could interfere with FGF-2 activity. Among the factors that might be involved in the FGF-2 blockage of TGF-β1 signaling are inflammatory factors, as INFα, TNFα, and IL-1β, which block the FGF receptor and, thus, do not allow FGF-2 to inhibit TGF-β1 action 39 . In fact, studies demonstrated that ASC may release TNFα and IL-1β, besides many other interleukins 33,47,48 . These factors, together, could abrogate the effect of FGF-2 in the low doses it is found in ASC-CMed. Still, it is essential to highlight the fact that ASC also produce TGF-β, as demonstrated by several studies 33,34,49,50 . In this regard, Rehman et al. showed that ASC produce around ten times more TGF-β than FGF-2 34 . The conjoint action of all these factors, which are contained in the ASC secretome, could explain, in part, the insufficiency of ASC-CMed to block the fibrotic stimule.
Still, in our view, the partial recapitulation in vitro of the fibroblast-related component of cardiac fibrosis, with the use of primary human cardiac fibroblast cultured in FGM and induced with TGF-β1 is a suitable in vitro model of the natural conditions in the fibrotic heart. Thus, considering the controversial findings between the studies using the spontaneous differentiation in addition to the results of the present study, we consider that in TGF-β-driven cardiac fibrosis ASC-CMed may not suffice as a sole therapeutic modality to block or reverse ongoing fibrosis. FGF-2, however, corroborating with the findings of two previous studies 21,22 , did block TGF-β1-induced pro-fibrotic conversion of cardiac human fibroblasts.
The present study supports the anti-fibrotic effects of FGF-2 through the blockage of NHCF-V differentiation into myofibroblasts, as demonstrated by the modulation of gene and/or protein expression in a series of

Methods
Experimental groups. Ventricular normal human cardiac fibroblasts (NHCF-V) were allocated into 5 groups with different induction/FGF-2/ASC-CMed combinations, as described in Table 2.
Human ASC were isolated as described previously 51 . Briefly, human abdominal fat was obtained by liposuction, washed with phosphate-buffered saline (PBS) and digested enzymatically with 0.1% collagenase A (#11088793001, Roche Diagnostic, Mannheim, Germany) in PBS with 1% bovine serum albumin (BSA; #A9647, Sigma-Aldrich, Boston, USA). The tissue was shaken constantly at 37 °C for 2 h. After this, the digested tissue    was mixed with 1% PBS/BSA, filtered, centrifuged and the cell pellet was resuspended in Dulbecco's Modified Eagle's Medium (DMEM; #12-604F, Lonza, Basel, Switzerland) with 10% fetal bovine serum (FBS; #F0804, Sigma-Aldrich, Missouri, United States), 1% penicillin/streptomycin (#15140122, Gibco Invitrogen, Carlsbad, USA) and 1% L-glutamine (#17-605E, Lonza Biowhittaker, Verviers, Belgium). Cells were cultured at 37 °C in a humidified incubator with 5% CO 2 . The medium was refreshed every 2 days. Cells were passed at a ratio of 1:3 after reached confluency. ASC conditioned medium (ASC-CMed) was obtained from confluent cultures of ASC between passages 3 and 6 from at least 3 different donors. For ASC-CMed, cells were cultured in DMEM without serum. Cells were kept at 37 °C with a minimum relative humidity of 95% and an atmosphere of 5% CO 2 in the air. The conditioned medium was harvested after 24 hours, passed through a 0.22 µm filter and stored t −20 °C until use. Before the experiment, the conditioned medium was concentrated 30 times using Amicon ® Ultra 15 mL filters (UFC900324, Merck, Darmstadt, Germany) and resuspended to the initial volume using Fibroblast Growth Medium-2 (FGM2; #CC-3132, Lonza, Basel, Switzerland).
Immunofluorescence, gene expression, and immunoblotting. Immunofluorescence. NHCF-V were cultured in 48 well tissue culture plates. After 5 and 21 days of induction, cells were fixed at room temperature with 2% paraformaldehyde (PFA) for 30 minutes. Cells were permeabilized with 1% Triton-X100 in PBS for 15 minutes at room temperature and blocked with 5% donkey serum in PBS and 1% BSA for another 15 minutes at room temperature. Subsequently, cells were incubated with primary antibodies diluted in 5% donkey serum in PBS for 2 hours at room temperature. The following primary antibodies were used: Gene expression analysis. NHCF-V were cultured in 25 cm 2 flasks. After 5 days of induction, RNA isolation was performed using TRIzol reagent (#15596018, Invitrogen Corp, Carlsbad, United States) according to the manufacturer's protocol. RNA concentration and purity were determined using NanoDrop technology (Thermo Scientific, Hemel Hempstead, United Kingdom). Between 300 ng and 500 ng of total RNA was used for cDNA synthesis, which was performed using RevertAid TM First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, United States) according to the manufacturer's protocol. The cDNA-equivalent of 5 ng total RNA was used per single qPCR reaction. PCR was performed using SYBR Green (Bio-Rad, Hercules, United States) with the ViiA7 Real-Time PCR system (Applied Biosystems, Foster City, United States). Each analysis was done in duplicate for each one of the independent experiments. The primers used are listed in the Supplementary  Table S1. Data were analyzed using ViiA7 software (Applied Biosystems, Foster City, United States) and normalized with the ∆Ct method, using the geometrical mean of 18S ribosomal RNA (18S RNA) cycle threshold (C T ) values. The fold-change in gene expression versus the no treatment control group (ECM) was calculated using the ∆∆C T method.
Immunoblotting analysis. NHCF-V were cultured in 25 cm 2 flasks. After 5 days of induction, cells were rinsed with cold PBS and lysed in 100 µL of cold lysis buffer (RIPA; #89900, Thermo Fisher Scientific, Waltham, United States) containing 1% protease inhibitor cocktail (PIC; #P8340, Sigma Aldrich, St. Louis, United States) and 1% Halt ™ phosphatase inhibitor cocktail (#78420, Thermo Fisher Scientific, Waltham, United States). The lysed cells were collected in 2 mL microcentrifuge tubes and the contents homogenized by sonication at 30 W for 30 seconds and centrifuged at 7,500 g at 4 °C for 5 minutes. The supernatant was collected for the protein concentration determination using the Bio-Rad DC protein assay (#5000112; Bio-Rad, Hercules, United States) according to the manufacturer's protocol. Gels (12%) were loaded with 4-10 μg of protein. After electrophoresis, gels were blotted onto nitrocellulose membranes (#170-4270; Bio-Rad, Hercules, United States). Blots were blocked with Odyssey ® Blocking Buffer (#927-40000, LI-COR, Lincoln, USA) in a dilution of 1:1 with PBS overnight at 4 °C.
Wound healing assay. NHCF-V were seeded in 48-well tissue culture plates at a density of 15,000 cells/ cm 2 and cultured in FGM3, 37 °C, and 5% CO 2 until 70% confluence was reached. Them, cells were divided into 5 groups and stimulated for 5 days, as prescribed previously. After this period, the confluent monolayers were scored with a 10 µl sterile pipette tip to leave a scratch of approximately 0.4-0.5 mm in width. Culture medium