Nerve Growth Factor (NGF) modulates in vitro induced myofibroblasts by highlighting a differential protein signature

We previously described the profibrogenic effect of NGF on conjunctival Fibroblasts (FBs) and its ability to trigger apoptosis in TGFβ1-induced myofibroblasts (myoFBs). Herein, cell apoptosis/signalling, cytokines’ signature in conditioned media and inflammatory as well as angiogenic pathway were investigated. Experimental myoFBs were exposed to NGF (0.1–100 ng/mL), at defined time-point for confocal and biomolecular analysis. Cells were analysed for apoptotic and cell signalling activation in cell extracts and for some inflammatory and proinflammatory/angiogenic factors’ activations. NGF triggered cJun overexpression and phospho-p65-NFkB nuclear translocation. A decreased Bcl2:Bax ratio and a significant expression of smad7 were confirmed in early AnnexinV-positive myoFBs. A specific protein signature characterised the conditioned media: a dose dependent decrease occurred for IL8, IL6 while a selective increase was observed for VEGF and cyr61 (protein/mRNA). TIMP1 levels were unaffected. Herein, NGF modulation of smad7, the specific IL8 and IL6 as well as VEGF and cyr61 modulation deserve more attention as opening to alternative approaches to counteract fibrosis.

The prolonged survival of myofibroblasts (myoFBs) characterize the pathological process of fibrosis 1,2 . MyoFBs are transient cells arising from quiescent fibroblasts (FBs); at injured zone FBs are exposed to proinflammatory cytokines, growth factors and extracellular matrix (ECM) products/mediators 2 . A defined deletion of myoFBs is vital to ensure a proper tissue healing (physiological remodeling) and the restoration of local function, while the aberrant apoptosis can account for excessive scarring and overt fibrosis 2,3 . Thus the mechanisms of myoFB persistence still require attention.
Nerve Growth Factor (NGF) promotes wound-healing and regulates ECM remodeling, allowing a wellbalanced tissue repair and recovery of functional activity 4,5 . This pleiotropic factor orchestrates crucial cell activities (proliferation, stimulation, differentiation and survival) on structural and functional cells belonging to the endocrine, immune and visual systems 4 . NGF binds to both specific (trkA NGFR ) and/or pan-neurotrophin (p75 NTR ) receptors to promote (autocrine/paracrine) downstream effects on the surrounding epithelial, endothelial, stromal and immune (mast cells, eosinophils, B/T cells, macrophages) cells [6][7][8][9][10] . Previous studies from our laboratory described the profibrogenic effect played by NGF on healthy-control primary cultures of conjunctival Fibroblasts (FBs), mainly expressing trkA NGFR , highlighting NGF ability to trigger apoptosis in the in vitro TGFβ1-induced myofibroblast model expressing both trkA NGFR /p75 NTR11 .
In visual system, the topical (eye-drop) NGF application improved corneal sensitivity and promoted corneal epithelial healing in both moderate and severe neurotrophic keratitis as well as in neurotrophic or autoimmune corneal ulcers 10,12 . A crosstalk between epithelial cell and FB/myoFB cell balance accounts for the correct tissue remodeling, sustained by the immune related cell pattern 6 . Direct paracrine signaling from early and late apoptotic cells in concert with immune cells (macrophages, neutrophils, infiltrating eosinophils and tissue resident mast cells) can also drive fibrotic outcomes, by a specific release of factors inside fibrotic tissue 6 . Therefore, alternative approaches, other than steroid and proper antifibrotic drugs, have been suggested for counteracting overt fibrotic response and promoting a "remodeling again" inside fibrotic tissues 13 . Few reports highlighted   4 °C) to discard debris and cytosolic/nuclear protein extracts were simultaneously produced by using the NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (78,833; Pierce), following the protocol recommended by the manufacturer. Cytosolic and nuclear sub-fractions (3 μL from each sample) were subjected to A280 quantification against lysozyme referring protein (NanoDrop 1000 Spectrophotometer, Thermo Fisher Scientific, MA, USA). Protein extracts were stored at − 20 °C until SDS-PAGE separation and immunoblotting analysis. Appropriate normalized aliquots were prepared for each sample to assure only one freeze-thaw cycle.
Ella microfluidics-based platform. Specific expression of Cyr61, IL6, IL8, TIMP1 and MMP9 was assessed by using the multiplex Ella platform. Briefly, 1:2 diluted samples were loaded on to the cartridge, according to a standard procedure provided by the manufacturers (Protein Simple, CA, USA). All steps in the procedure were run automatically by the instrument with no user activity. Cartridges include a built-in lotspecific standard curve for the defined protein. The obtained data were displayed as pg/mL and automatically calculated by the internal instrument software.
Chip-based protein microarrays. Conditioned media from myoFBs were hybridized in chip arrays and labeled according to the manufacturer's instructions (G-series arrays; Ray Biotech, Norcross, CA). Each cy3labeled treated/untreated sample was combined with an equal amount of pooled cy5-labeled common reference and 70 µL mixture (well/chip) was hybridized for 18 h at 4 °C. After washes under stringency, the glass slides www.nature.com/scientificreports/ were washed once in MilliQ water to remove salts and quickly spun to dry the chips. Slides were analyzed in parallel: double fluorescence signals were acquired with the GenePix 4100 microarray scanner (Molecular Devices LLC, Sunnyvale, CA) equipped with the GenePix Pro 3.0 software (Axon Instruments, Foster City, CA). Data were expressed as ratio (treated/untreated signal).

RNA extraction, cDNA synthesis and relative real-time PCR.
Total RNA was extracted from 6-well plates cells according to the TRIfast technique (Euroclone), treated with DNAseI (AB1709; Ambion Inc., Austin, TX), spectrophotometrically analyzed for quantification and purity (260/280 ratio; Nanodrop A1000 Spectrophotometer; Celbio, Milan, Italy). A further RNA quality assessment was carried out on randomly selected RNA extracts in a 1% agarose (Promega, Milan, Italy) gel supplemented with Ethidium Bromide (ICN). Only samples with 260/280 > 1.8 were used for amplification studies. cDNA was generated from 3 µg total RNAs by using the GoScript standardized procedure (Promega) in the presence of random hexamers (Promega), in a OneCycler programmable thermocycler (Peqlab; VWR Radnor, Pennsylvania, USA). For amplifications, 3 µL cDNAs for target gene and 1µL cDNAs for referring ones were amplified in a 20 µL final volume of SYBR Green PCR mixture (Applied Biosystems, Foster City, CA), using the  Table 1B (synthesized by Eurofin MWG, Ebersberg, Germany).
Only normalized samples were amplified and cycle threshold (Ct) values from good melting curves were used for analysis in the REST software 19 . Relative gene expression was calculated as the expression level of target gene with respect to that of referring genes (H3 and/or GAPDH), considering both treated vs. untreated cells. As fold-change ratios were expressed in 2-log, only increase/decrease over 2 were considered of interest in the statistical evaluation.
Statistical analysis. Experiments were performed in triplicate, starting from n = 3 primary sets expanded and used in the 5th-7th generation range, and analysed for three times to validate results. Controls (without NGF stimulation) were carried out in every plate and each generation. Descriptive statistic and graphics were performed using the GraphPad Prism 8.01 (GraphPad, San Diego, CA). The standard deviation (SD) and Standard Error from the Mean (SEM) were calculated to assess the variations between different treatments under the same conditions (untreated cells). Data were subject to statistical significance by one-way analysis of variance (ANOVA) using the StatView II Software (Barckley, CA). ANOVA was coupled to post-hoc analysis performed using Tukey's test between subgroups and with respect to untreated cells). All p values ≤ 0.05 were considered as significant and depending on post-hoc analysis, appropriate asterisks were used in panels as follows: *p < .05; **p < .01 and ***p < .001 (highly significant). For chip array analysis, a p < .0083 was used together with a limit of 2-folds. Data are mean ± SD (text) and mean ± SEM (figures), and error bars were calculated from at least three independent experimental sets.

Results
A flow chart of overall experimental procedure is shown in Fig. 1A. High density replated and serum-starved confluent TGFβ1-induced myoFBs (herein shorten as myoFBs) were exposed to single or chronic increasing NGF concentration (0-100 ng/mL), over 15 min-30 min-3 h-6 h-24 h (protein/mRNA) time-point sets. Untreated myoFBs were used as control and carried out at each set of experiments. The typical spindle-appearance and the p75 NTR (green) and αSMA (blue) expression on red nuclear counterstaining are visible in Fig. 1B. NGF activates p65NFkB / cJun molecules in AnnexinV positive myoFBs. NGF induced a dosedependent increase of phospho-p65NFkB protein in AnnexinV positive myoFBs. The quantification of immunoreactivity (Integrated Density; IntDen) is shown in the histogram (Fig. 1C). IntDen values are % over control (sister untreated myoFBs). Representative confocal images are shown in Fig. 1C reporting both single (specific) and double (merge) immunostainings. The merge visualization in the upper row (Fig. 1C) highlight that NGF exposed myoFBs positively stained for phospho-p65NFkB (blue/cy5) and AnnexinV (green/FITC). Single immunoreactions for phospho-p65NFkB (middle row, blue/cy5) and AnnexinV (lower row; green/FITC) are shown below. This p65NFkB expression was associated with a significant increase of enlarged, nuclear condensed cells, especially after repeated NGF treatment (see white arrows). TGFβ1 by itself did not trigger significant p75 NTR expression nor AnnexinV, retaining a high trkA NGFR /p75 NTR ratio (data not shown). At the same time, NGF exposure induced the activation of cJun and NFkB pathways in a time-dependent fashion. Data from western blotting analysis corroborated the confocal ones, at least at 100 ng/mL NGF exposure. As shown in Fig. 1D, NGF triggered the cJun protein expression in a time-dependent fashion, as compared to untreated ones (**p < .001 vs. untreated cell extracts; ANOVA-Tukey Kramer post-hoc). In line, p65NFkB protein signal also increased in a time-dependent fashion (*p < .05 vs untreated cell extracts; ANOVA-Tukey Kramer post-hoc; Fig. 1E). Original immunoblots for cJun and p65NFkB are visible in the supplementary Fig. S1. NGF promotes nuclear translocation of phosphorylated p65NFkB in receptive myoFBs. Immunoreactivity for p65NFkB and p75 NTR (merge, left panel) was observed in myoFBs exposed to repeated 100 ng/ mL NGF stimulation ( Fig. 2A). Single staining for p75 NTR (middle panel/green/cy2) and p65NFkB (right panel/ blue/cy5) are also shown. MyoFBs showed a significant increase in phospho-p65NFkB expression at both cyto-  www.nature.com/scientificreports/ plasm and nuclear levels in AnnexinV positive (green/FITC) cells (Fig. 2B). The different AnnexinV (upper frame) and phospho-p65NFkB (lower frame) expression is shown (grayscale of the two cells framed in overlay image is shown in the right side of Fig. 2B). Single cell densitometric analysis was carried out specifically for phospho-p65NFkB in AnnexinV negative and positive cells. A representative expression of different cytoplasm to intracellular localization of phospho-p65NFkB is reported, as shown by greyscale acquired and pseudocolors cell images (Fig. 2C,D). As shown by 3D IntDen graphical representation (right) panels, the intranuclear phospho-p65NFkB expression was more evident in AnnexinV positive cells (Fig. 2D) with respect to AnnexinV negative one (Fig. 2C). Western Blotting analysis confirmed the increased expression and nuclear translocation of phospho-p65NFkB protein, as shown by graphics and a representative immunoblot on both nuclear (Fig. 2E) and cytoplasmic (Fig. 2F) extracts. Quantifications were carried out against untreated nuclear and cytoplasm extracts showing a significant increase in phospho-p65NFkB in nuclear extracts in a dose dependent fashion, as visualized over a baseline dashed line in the bar-graphs (*p < .05 in 100 ng/mL exposed cells with respect to untreated ones; Fig. 2E). A decrease in cytoplasm expression was monitored, with the lowest expression at 1 ng/mL NGF, without explanation (Fig. 2F).
Confocal microscopy showed the localization of perinuclear p75 NTR and p65NFkB immunoreactivity inside nuclei after NGF treatment (100 ng/mL; Fig. 3A). By contrary, p75 NTR siRNA (oligonucleotides) transfected myofibroblasts showed a drastic reduction of p75 NTR immunoreactivity, which resulted in a retained cytoplasm immunoreactivity of p65NFkB (Fig. 3B). The original immunoblots specific for both cytoplasmatic and nuclear expression phospho p65NFkB are shown in Fig. S1.   Figure 3. NGF driven p65NFkB expression/localization. Confocal images of p75 NTR siRNA cultures were acquired at 24 h from NGF stimulation. Representative images of p65NFkB (green) and p75 NTR (blue) immunoreactivity over red nuclei (propidium iodide) of 100 ng/mL NGF-exposed myofibroblasts, either untreated (A) and p75 NTR oligonucleotides transfected ones (B). Upper squares display single staining. Images are representative of three independent experiments acquired under the same channel-series parameters by confocal microscope Nikon ci and analyzed by NIS software.  20 . As shown in Fig. 4A, changes in expression of Bcl2 and Bax transcripts were monitored upon 100 ng/mL NGF exposure. Particularly, Bcl2mRNA deregulation was higher at 3 h (− 3.691 ± 0.037 log2 ) while Bax transcript upregulation (2.547 ± 0.749 log2 , *p < .05) was maximum at 24 h, as observed by REST-ANOVA coupled analysis on treated versus untreated RNA extracts (Fig. 4A). As shown in Fig. 4B, smad7 transcript (the inhibitory factor of TGFβ1 signaling) was significantly increased at 24 h upon 100 ng/mL NGF exposure (5.946 ± 3.320 log2 , *p < .05; fold changes vs. untreated expression; REST-ANOVA coupled analysis).

NGF influences myoFBs' protein signature (conditioned media).
To identify potential mediator between inflammatory (cytokines, chemokines, adhesion molecules), growth (neurotrophins, fibrogenic and angiogenic factors) and tissue remodeling (TIMPs) ones, potentially produced/released into conditioned media by NGF-exposed myoFBs, a chip array with sixty (60) potential candidates was personalized. According to multiple comparison and Bonferroni's corrections, ANOVA analysis pointed to 3 out of 60 pre-selected candidates. A trend to an increase and significant expression at 10 ng/mL NGF was observed for IL8 (p < .001) proteins (Fig. 5A,B). A decrease of IL6 was assessed at increasing NGF doses (p < .01; Fig. 5C,D). Besides an increase at lower concentration (Fig. 5E), no significant changes were observed for TIMP1 expression (Fig. 5E,F). Protein chip array data (Fig. 5A,C,E) were confirmed by the ELLA microfluidic platform (Fig. 5B,D,F).

NGF increased VEGF and cyr61 proteins/gene transcripts.
To verify the NGF contribution in angiogenesis, both VEGF and cyr61 molecules were analyzed in conditioned media (protein) and cell extracts (mRNA). Of interest, specific increase of VEGF protein was quantified at 1 ng/mL NGF exposure (Fig. 6A), as confirmed by ELISA (data not shown), and sustained by molecular analysis (Fig. 6B). Of interest, a significant increase of cyr61 protein was detected at 10 ng/mL NGF, as compared to untreated ones (p < .05; Fig. 6C) and confirmed by molecular data showing a specific cyr61mRNA upregulation after 1 ng/mL NGF exposure (− 2.918 log; p > .05, Fig. 6D).

Discussion
The NGF healing properties have been prospected since 50s' , from NGF-driven wound healing under physiological states to tissue repair under acute and chronic inflammatory conditions, either Th1 or Th2 driven 6,7 . Herein, our findings extend previous data on NGF-induced apoptosis in trkA NGFR /p75 NTR -bearing and αSMA-expressing myoFBs 11 , by providing evidence on cJun activation, Bcl2:Bad ratio reduction (Bad overexpression), p65NFkB nuclear translocation and smad7 overexpression. The protein signature of conditioned media indicates that long-lasting myoFBs synthesize and release IL8 and VEGF as well as cyr61. Of interest, IL6 and TIMP1 were significantly deregulated. A graphical explanation of these findings is shown in Fig. 7. Prolonged stimulation might trigger survival of myoFBs and inappropriate ECM remodeling (overt fibrosis), causing irreversible alterations of organ anatomy and function [21][22][23] . Apoptotic process represents a physiological strategy removing long-lasting myoFBs, the source of ECM deposition and prolonged matrix contraction 2,3,20,24 . We previously showed the profibrogenic NGF effects on primary cultures of FBs outgrew from skin, lung and conjunctival/corneal tissue (NGF/TGFβ1 expression, αSMA protein metabolism and contractive activity) 11 . Subsequently, we observed that NGF-treated myoFBs showed apoptosis restricted to a p75 NTR expressing myoFB phenotype, an effect that was counteracted by specific trkA NGFR and/or p75 NTR inhibitors 11,25 . Herein, our findings point at the NGF-mediated cJun increase and Bcl2;Bax ratio decrease in apoptotic myoFBs, supporting previous studies on p75 NTR -transduced apoptosis in association with Rac-GTPase and c-Jun activation 26 . The observation of a decreased Bcl2:Bax ratio in Annexin www.nature.com/scientificreports/ positive myoFBs would suggest that Bcl2:Bax ratio might serve as a rheostat to determine the susceptibility to death process [27][28][29] . Our NGF-mediated apoptotic effect was also associated with p65NFkB nuclear translocation. NFkB transcription factor is composed of p50/p65 subunits close to cytoplasmic IkB inhibitors to prevent nuclear translocation, a route observed in several cell types in the presence of the pan-neurotrophin p75 NTR receptor activation 30,31 . Our finding on NGF-mediated p65-NFkB nuclear translocation in these myoFBs strongly suggest the ability of NGF to increase myoFB apoptosis while in unresponsive myoFBs it might modulate the release of some proinflammatory as well as pro-angiogenic factors, through a p75 NTR mediated signal 11 . The p75 NTR neutralization showed a retaining in p65NFkB translocation, corroborating the NGF/p75 NTR specific involvement. We hypothesize that the unresponsive cells were those with a lower trkA NGFR /p75 NTR ratio (likewise with high Bcl2:Bax ratio). It is noteworthy highlight that tissues remodeling is strongly driven by TGFβ1 working on a microenvironment enriched of others profibrogenic mediators released by accessory and immune cells, epithelial and endothelial cells and even activated myoFBs 13,17 . The profibrogenic TGFβ1-RI/RII routes the down streaming of smad2/3 and smad4 activation in the presence of a blocked Smad7 activity 32,33 . As observed in this in vitro  IL8 (A,B), IL6 (C,D) and TIMP1 (E,F) in conditioned media from myoFBs exposed to increasing NGF doses (0-10 ng/mL). Note the increase of IL8, the decrease of IL6 and the quite unchanged expression of TIMP1. Data are mean ± SEM (fold changes or pg/mL) of three independent experiments. ANOVA analysis followed by Tukey-Kramer post-hoc; sign: *p < .05 and **/ < .01, as for multiparametric array analysis. www.nature.com/scientificreports/ model, NGF might dampen TGFβ1 signaling thought activation of smad7 expression, explaining at least in part the physiological remodeling observed upon exogenous addition of NGF (corneal ulcers' closure) 10 . As known, smad7 works as an inhibitory factor in TGFβ1 pathway, the Smad7 expression in these myoFBs might justify at least in part the absence of insistent myoFB signaling in human tissues when repair is NGF assisted 32 . As illustrated in Fig. 7, exogenous NGF administration (shorten as + NGF in red) might revert Bcl2:Bax ratio of fibrotic myoFBs and promote p65NFkB nuclear translocation restoring physiological condition and function of tissue, with resolution of fibrosis.
To identify potential mediator between inflammatory (cytokines, chemokines, adhesion molecules), growth (neurotrophins, fibrogenic and angiogenic factors) and tissue remodeling ones, we used a customized array chip approach with the screening of 60 potential candidates, as previously tested in other studies 34 . Herein, the increased expression of IL8 and the decreased expression of IL6, as detected in conditioned media of NGFexposed myoFBs, would imply that NGF does not contribute to the profibrogenic microenvironment.
The dual-faced of angiogenesis at induction of fibrogenesis and resolution of fibrosis has been described in the last years 35,36 . Angiogenesis-the formation of new blood vessels from pre-existing vessels-is a complex and dynamic process occurring both physiologically and pathologically 37 . The tissue healing with proper matrix remodelling would benefit from a controlled induction of vascular activity. A controlled apoptosis and a tidy epithelia-stroma interaction should occur at involved and uninvolved surrounding tissues 36,37 . Some recent evidence suggested that experimental inhibition of angiogenesis ameliorates the development of liver fibrosis, while other recent studies indicate that neutralization or genetic ablation of VEGF can delay tissue repair and fibrosis resolution in damaged tissues 38,39 . The imbalance between pro/anti-angiogenic mediators might contribute to apoptotic or sustain the process of fibrosis, as observed for CCN1/CYR61 involved in attenuating and/or scavenging of TGFβ, mitigating the process of fibrogenesis 13,40,41 . The findings herein reported suggest that the NGF mechanisms might involve a deregulation of TGFβ1 signalling, due to Smad7 gene expression, and alternative a VEGF/cyr61 activation 17,32,41 .
Some open questions still persist (1) as in vitro study on TGFβ1-induced myoFBs, and as known TGFβ1 does not represent the lone differentiating factor in vivo and other soluble mediators can trigger this differentiation; (2) the trkA NGFR /p75 NTR heterodimer distribution on cell membrane, as other trkA NGFR /trkA NGFR and p75 NTR / p75 NTR homodimers can also occur influencing the cellular pathway and finally (3) this NGF-driven p65 NFkB translocation could result in three different pathways (the canonical, the non-canonical, and the atypical one) (C,D) Bar-graph showing the expression of cyr61: note the higher protein expression at 10 ng/mL NGF dose, as confirmed by biomolecular analysis. Data are mean ± SEM (MFI or pg/mL for proteins and fold changes as log2-expression for transcripts) of three independent experiments. ANOVA analysis followed by Tukey-Kramer post-hoc; sign: *p < .05 and **p < .01, as for multiparametric array analysis.  42 . On the contrary, the possibility to use topical NGF will reduce all the disappointed effects played by the circulating addition of NGF. Taken together, several mediators of inflammation and tissue remodelling, either endogenous or exogenously administered (including natural (omega 3/5) and synthetic ones), might participate in the myoFB-driven remodelling differentially over again, representing candidate factors for driving the correct repair 43 . The NGF-driven upregulation of smad7, an inhibitory component of TGFβ1 pathway, and VEGF/cyr61 deserve further investigation, as opening to alternative approaches in counteracting fibrosis, especially for fibrotic eye diseases.

Data availability
All data are available in the manuscript.  In injured tissues, trkA NGFR /p75 NTR -myoFBs arising from resident "quiescent" trkA NGFR -FBs (fibroblasts), mainly upon TGFβ1 stimulation and a specific microenvironment, achieve appropriate tissue remodeling (collagen production, ECM refining and contraction). NGF might function as a modulator, in concert with a plethora of inflammatory and matrix remodeling mediators. Overt fibrosis indicates a not well-balanced tissue remodeling. As known, smad7 works as an inhibitory factor in TGFβ1 pathway. Exogenous NGF administration (shorten as + NGF in red) might revert Bcl2/Bax ratio and promotes p65NFkB nuclear translocation restoring physiological condition and function of the tissue, with the resolution of fibrosis. As pleiotropic factor, NGF can also drive neovessel formation and trough VEGF and cyr61 (red) modulation, it might contribute to functional recovery. For representative purpose (on left upper side), a B/W image (a conjunctival frame from a section belonging to a very old historical collection; Anatomia Patologica, Università Cattolica del Sacro Cuore) as background for layered inflammatory cells. Illustration was developed on Microsoft Office Power Point 365 software (Microsoft corp., New York, NY, USA).