Interactions between TGF-β1, canonical WNT/β-catenin pathway and PPAR γ in radiation-induced fibrosis

Radiation therapy induces DNA damage and inflammation leading to fibrosis. Fibrosis can occur 4 to 12 months after radiation therapy. This process worsens with time and years. Radiation-induced fibrosis is characterized by fibroblasts proliferation, myofibroblast differentiation, and synthesis of collagen, proteoglycans and extracellular matrix. Myofibroblasts are non-muscle cells that can contract and relax. Myofibroblasts evolve towards irreversible retraction during fibrosis process. In this review, we discussed the interplays between transforming growth factor-β1 (TGF-β1), canonical WNT/β-catenin pathway and peroxisome proliferator-activated receptor gamma (PPAR γ) in regulating the molecular mechanisms underlying the radiation-induced fibrosis, and the potential role of PPAR γ agonists. Overexpression of TGF-β and canonical WNT/β-catenin pathway stimulate fibroblasts accumulation and myofibroblast differentiation whereas PPAR γ expression decreases due to the opposite interplay of canonical WNT/β-catenin pathway. Both TGF-β1 and canonical WNT/β-catenin pathway stimulate each other through the Smad pathway and non-Smad pathways such as phosphatidylinositol 3-kinase/serine/threonine kinase (PI3K/Akt) signaling. WNT/β-catenin pathway and PPAR γ interact in an opposite manner. PPAR γ agonists decrease β-catenin levels through activation of inhibitors of the WNT pathway such as Smad7, glycogen synthase kinase-3 (GSK-3 β) and dickkopf-related protein 1 (DKK1). PPAR γ agonists also stimulate phosphatase and tensin homolog (PTEN) expression, which decreases both TGF-β1 and PI3K/Akt pathways. PPAR γ agonists by activating Smad7 decrease Smads pathway and then TGF-β signaling leading to decrease radiation-induced fibrosis. TGF-β1 and canonical WNT/β-catenin pathway promote radiation-induced fibrosis whereas PPAR γ agonists can prevent radiation-induced fibrosis.


INTRODUCTION
Patients suffering of cancer often receive an external ionizing radiation therapy in association with surgery/ chemotherapy or alone. This ionizing radiation can involve damages in tumor cells but also in healthy tissue in the radiation field. Radiation therapy induces several skin changes such as inflammation, edema, dermatitis, www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 52), pp: 90579-90604 Review ulceration, late radiation-induced fibrosis (RIF), and necrosis [1,2]. Radiation dose, fraction size and volume treated will vary the late side effects of radiation.
Radiation-induced fibrosis is marked by fibroblasts proliferation, myofibroblast differentiation, and synthesis of collagen, proteoglycans and extracellular matrix. [3,4]. RIF can occur 4 to 12 months after radiation therapy, and this process is worsening with time and years. Every radiation-exposed part of the body can be affects by RIF, and the type of tissue exposed will influence the clinical presentation. RIF manifestations are multiple, such as skin induration and thickening, muscle shortening and atrophy, limited joint mobility, lymphedema, mucosal fibrosis, ulceration, fistula, hollow organ stenosis, and pain [5] and generally impact the quality of life [6,7]. Fibrosis present an active retraction of the granulation tissue which is induced by contractile non-muscle cells, named myofibroblasts [8][9][10]. Fibroblasts and myofibroblasts are key effectors involved in the development of fibrosis due to excessive deposition of collagen and inappropriate extracellular matrix (ECM).
TGF-β1 interacts with the canonical WNT/βcatenin pathway and peroxisome proliferator activated receptor γ (PPAR γ) which act in an opposite manner in several pathologies [24]. TGF-β1 stimulates myofibroblast differentiation by the activation of canonical WNT pathway and the downregulation of PPAR γ expression [25]. In response to TGF-β1, resident fibroblasts transdifferentiate into contractile myofibroblasts which express α -smooth muscle actin (α -SMA) and synthesize extracellular matrix proteins, particularly collagen.
We focus this review on the positive link between TGF-β1 and canonical WNT/β-catenin pathway and the opposite role of PPAR γ.

Radiation-induced fibrosis (RIF)
Radiation-induced fibrosis (RIF) is a major late effect which contributes to patient morbidity and occur in the skin, and subcutaneous tissue, lungs, gastrointestinal as well as any other organs in the treatment field. The mechanisms linking radiation to tissue sclerosis, fibrosis and atrophy are complex. In both tumors and normal tissues, radiation causes the induction of apoptosis or clonogenic cell death through free radical-mediated DNA damage [26]. In normal tissues, radiation toxicity induces changes in cell functions and causes activation of coagulation system, inflammation, epithelial regeneration, and tissue remodeling, which are precipitated by several signaling such as cytokines, chemokines and growth factors [27].

Etiology of RIF
Several factors can increase the risk of RIF. Total dose of radiotherapy and dose per fraction, volume of tissue treated, and time of treatment delivery are considered as primary factors. Increased levels of radiation dose, hypofractionation and increased field size are directly correlated with degree of RIF [28][29][30][31][32]. Patients with certain diseases are more susceptible to develop severe RIF, such as systemic scleroderma, systemic lupus erythematosus (SLE), or Marfan syndrome [33,34]. Genetics factors have also a role in the predisposition to RIF [35,36]. Single-nucleotide polymorphisms have been observed in genes encoding proteins such as TGF-β1, and superoxide dismutase 2 (SOD2) [37,38]. Additional genes like IL18 (interleukin 18), MMP12 (matrix metalloproteinase 12), PER3 (period circadian protein homolog 3 protein), LTF (lactoferrin) stimulate the degradation of post-radiation ECM [39]. Several DNA modifications have been associated with RIF, like epigenetic modifications to DNA and histones [40]. Mitochondrial DNA damage enhance the removal of reactive oxygen species (ROS) [41].

Clinical presentation of RIF
RIF usually occurs 4 to12 months after radiation therapy and can progress over many years. The type of tissue exposed to irradiation is responsible for the clinical presentation. In general, RIF can manifest as skin induration and thickening, muscle shortening and atrophy, limited joint mobility, lymphedema, mucosal fibrosis, ulceration, fistula, hollow organ stenosis, and pain [5]. Other manifestations more regionally and specific include trismus, xerostomia, decreased vocal quality, osteoradionecrosis, dysphagia, and aspiration in patients with head and neck malignancy [42][43][44][45][46][47]; cervical plexopathy, brachial plexopathy, interstitial fibrosis, dyspnea, and oxygen requirement in patients with breast or lung malignancy [48,49]; and urinary urgency, increased urinary frequency, diarrhea, loss of reproductive function, and dyspareunia in patients with abdominopelvic malignancy [50][51][52]. Currently, there is no uniform consensus to objectively quantify the degree of fibrosis in RIF [53].

Pathogenesis of RIF
Three histopathological phases of RIF are described. The prefibrotic phase shows chronic inflammation in which endothelial cells have a major role. The organized fibrosis phase contains a high density of myofibroblasts in an unorganized matrix adjacent to poorly cellularized fibrotic areas of senescent fibrocytes in a dense sclerotic matrix. The third phase named late fibroatrophic phase shows retractile fibrosis and gradual loss of parenchymal cells [54]. RIF is initially characterized by an injury which incites an acute response leading to inflammation, followed by the accumulation of fibroblasts, differentiation into myofibroblasts, and activation of extracellular matrix proteins like collagen [22]. Radiation induces direct DNA damages and the apparition of reactive oxygen species (ROS) [55] resulting in oxidative stress [56]. ROS involves interactions of ionizing radiation with water molecules and then the formation of free radicals such as superoxide, hydrogen peroxide and hydroxyl radical [57]. Hydroxyl radical production is responsible for the major part of damages [58,59] The first inflammatory cells which arrived at injured sites are neutrophils [69]. Neutrophils encounter fibronectin and collagen fragments and then lead to the release of inflammatory cytokines such as tumor necrosis factor alpha (TNF-α), interleukin 1 (IL-1), and interleukin 6 (IL-6) for the initiation of ROS and local inflammation [3, 70- TGF-β and radiation therapy (cf. Figure 1) Transforming growth factor β (TGF-β) is made of three similar proteins: TGF-β1, TGF-β2 and TGF-β3. TGF-β receptor 1 and TGF-β receptor 2 are transmembrane proteins. TGF-β1 binds TGF-βR2 which recruits TGF-βR1. TGF-β1 is secreted and then deposited into ECM and considered as a large latent complex. ECM is a main reservoir of cytokines [104]. TGF-β1 stimulates the Smad pathway and non-Smad pathways such as phosphatidylinositol 3-kinase/serine/ threonine kinase (protein kinase B) (PI3K/Akt), Rho, and MAPK. Radiation-induced fibrosis is characterized by an upregulation of TGF-β1 [21, 23, 84, 105, 106].
TGF-β is responsible for the control of many functions, like breakdown of connective tissue, inhibition of epithelial cell proliferation and synthesis of extracellular matrix proteins (collagen and MMPs) [106].
A close relationship is observed between irradiation dose and liver cell viability [117]. Indeed, ionizing irradiation induces a progressive change in mRNA levels which gradually increase with time [118,119]. Evolution of α-SMA in liver fibrosis induction suggests a time-dependent effect of irradiation [120,121]. TGF-β1 expression gradually increases in a dose-dependent manner and peaking 4 weeks after irradiation [122].

The contractile non-muscle cells: myofibroblasts
Myofibroblasts contain actin filaments bundles with α-SMA. α-SMA represents peripheral focal adhesions and gap junctions for the connection between myofibroblats and granulation tissues [123]. Fibroblasts become protomyofibroblats after this interaction.
Protomyofibroblasts synthetize ECM, collagen, fibronectin which are essential for differentiation [ Myofibroblasts are contractile non-muscle cells, characterized by the type IIA non-muscle myosin (NMIIA) [140]. NMII are responsible for cell polarity, cell migration NMIIA is characterized by three pairs of chains, two heavy chains of 230 kDa, two 20 kDa regulatory light chains (RLCs) which stimulate the NMII activity and two 17 kDa essential light chains (ELCs) which stabilize the heavy chain structure. Calcium-calmodulin-myosin light chain kinase (MLCK) and Rho/ROCK/myosin light chain phosphatase increases NMII activity [90,143,144].
Myosin filaments link actin filaments in thick bundles like as stress fibers. NMMIIA molecules assemble into bipolar filaments. A title of the head enables a conformational change that moves actin filaments in an anti-parallel manner. The cross-bridge actin-myosin cycle of NMIIA is overall like smooth and striated muscle myosin. An ATP molecule binds the NMIIA-ATPase site on the myosin head. This allows the dissociation of actin from the NMIIA head. ATP is then hydrolyzed and subsequently, NMIIA binds to actin. Then, the power stoke occurs with a tilt of the NMIIA head, which generates a CB single force and a displacement of few nanometers. ADP is then released from the acto-NMIIA complex. A new ATP molecule dissociates actin from myosin head, and a novel CB cycle begins.
However, the major feature of NMIIA is its extreme slowness, with dramatic slow of its kinetics of contractile [145,146]. The cross-bridge actin-myosin detachment constant, attachment constant, catalytic constant are widely low compared with striated or smooth muscles. Nevertheless, the cross-brigde actin-myosin of NMIIA single force is same order of magnitude compared with muscle myosin II (MII). Thermodynamic force, thermodynamic flow and thermodynamic entropy production rate are rarely low [147]. This explains why this stationary contractile system operates nearequilibrium. The low isometric tension observed in placental stem villi [141,148] can be surely explained by the low placental myosin content [149, 150]. The extremely slow shortening velocity can be accounted for by the very low placental myosin ATPase activity which has an essential role for the association/dissociation of actin from the NMIIA head [145,149]. In placental stem villi, a low isometric tension has been reported [147]. This can be explained by the low placental myosin content and the low placental myosin ATPase activity [148]. The acto-myosin apparatus functions as in smooth muscles in myofibroblasts of human placenta. In experimental bath, addition of KCl or electrical field induce the contraction Figure 2: Interactions between TGF-β1, canonical WNT/β-catenin pathway, and YAP/TAZ signaling. In the absence of WNT ligands, Smad7 associates with β-catenin and Smurf2 to degrade β-catenin. Smad7 also binds phosphorylated YAP and Smurf to form a high affinity complex for TGF-β type 1 receptor to decrease TGF-β signaling expression. In the absence of WNT ligands, phosphorylated YAP and TAZ are associated with the β-catenin destruction complex. Phosphorylated YAP is also associated with β-catenin to inhibit its nuclear translocation and promote its degradation. Upon WNT stimulation, the beta-catenin destruction complex is inhibited and then beta-catenin accumulates in the cytosol and translocates to the nucleus. TAZ inhibits the phosphorylation of DSH and dissociates it from the β-catenin destruction complex. The destruction complex is inhibited because YAP and TAZ dissociate from the complex. β-catenin accumulates in the cytosol and then translocates to the nucleus for activating WNT targets. Upon TGF-β stimulation, Axin promotes the tail-phosphorylation of Smad2/3. Axin also forms a complex with Arkadia and Smad7 for enhancing TGF-β signaling. The activated Smad complex associates with TAZ and YAP and then translocates to the nucleus for activating Smad targets. phase. 2,3-butanedione monoxime, or isosorbide dinitrate, two inhibitors of NMII, can induce the relaxation phase [146]. Changes in the volume of the intervillons chamber can alter the length of the placental stem villi. Contraction of myofibroblasts induces a modulation of the distal resistance of the umbilical artery and then a modulation of the umbilical blood flow, due to the Starling phenomenon. In fibrotic processes, myofibroblasts generate rather a phenomenon of contraction-retraction lasting without relaxation and the pathological tissue undergoes an irreversible retraction, evolving towards fibrosis favored by the synthesis of collagen.

Canonical WNT/β-catenin pathway
Wingless and integration site (named WNT) pathway is a cascade of numerous signaling which are involved in development, metabolism, growth cellular, and maintain of stem cells [151]. WNT pathway is formed by secreted lipid-modified glycoproteins [152]. Deregulation of the canonical WNT pathway is observed in several pathologies [24].
WNT/β-catenin pathway is increased in fibroblasts in response to radiation during skin fibrogenesis [157]. WNT ligands are activated after irradiation and promote survival in head and neck cancers [158,159]. Radiation stimulates ERK pathway through a reactive oxygen species (ROS) generation and inactivates GSK-3β [114].
Ionizing radiation activates NF-κB which contributes to cell sensitivity to radiation [207]. NF-κB is activated in endothelial cells after irradiation and is necessary for radiation-induced IL-6 release [208]. Inflammation, through the activation of NF-κB, promotes the production of collagen and the release of inflammatory chemokines in the development of liver fibrosis [209]. NF-κB expression is stimulated by the activation of Akt pathway observed in RIF [210]. In fibrosis, CCN4, a WNT-inducible signaling pathway protein-1 (WISP1) is increased and participates to the fibroblast proliferation and ECM expression [211,212]. CCN4 overexpression induces morphological transformation in skin fibrosis [213]. CCl4mAb, a specific inhibitor of CCN4, reduces NF-κB activity and then decreases the expression of profibrotic factors, such as TGF-β1, in hepatic fibrosis [214].
Fibroblasts stimulation with activated WNT3a enhances the expression of TGF-β and then the phosphorylation of the MH2 domain of Smad2 [216]. Conversely, the absence of WNT ligands decreases the expression of TGF-β and then attenuates the fibrotic response [220].
Without tail-phosphorylation by the TGF-β type I receptor, Smad2/3 cannot interact with Smad4 and then cannot engaged the DNA. Smads inactivated are phosphorylated by activated GSK-3β and then degraded [221]. Moreover, activation of TGF-β also enhances WNT pathway stimulation through the inhibition of dickkopfrelated protein 1 (DKK1) [222].
Myofibroblast activation and fibrosis induction have been recently translated by the mechanical properties of YAP and TAZ. YAP and TAZ (transcriptional coactivator with PDZ-binding motif) elevated levels are observed in fibrosis [228]. YAP and TAZ knockdown in lung and liver fibroblasts cultured reduces the levels of protein associated with myofibroblast differentiation, such as pro-collagen and α-SMA [229]. During fibrosis, F-actin polymerization inactivates the Hippo core kinase complex leading to the dephosphorylation of YAP and TAZ [230].
Stimulation by WNT3a inhibits the destruction complex because YAP/TAZ dissociates from the β-catenin destruction complex and then allowed β-catenin nuclear translocation. TAZ binds DSH and dephosphorylates it upon stimulation with WNT3a to dissociate DSH from the destruction complex [231]. Indeed, in the absence of WNT ligands, phosphorylated YAP and TAZ bind β-catenin with activated GSK-3β and Axin to degraded β-catenin in the proteasome [232]. In fibrosis, activated Smad2/3-Smad4 complex is associated with YAP and TAZ for its translocation to the nucleus [233]. A crosstalk of several components of the TGF-β, WNT, and YAP/TAZ signaling is playing in the tuning of nucleocytoplasmic shuttling of fibrosis [230].

PPAR γ
Peroxisome proliferator receptor γ (PPAR γ) is a ligand activated transcriptional factor which forms a heterodimer with retinoid X receptor (RXR) to activate specific peroxisome response elements (PPRE) [234]. PPAR γ expression is involved in several mechanisms such as glucose and lipid metabolism, immune response, and inflammation [235,236]. PPAR γ decreases NF-κB activity and then represses inflammation [237]. PPAR γ is expressed in several cells, such as adipocytes, muscle cells, brain cells, immune cells and in fibroblasts [238].

PPAR γ and fibrosis
An inverse relationship is observed between the expression of PPAR γ and the apparition of fibrosis. Aberrant downregulation of PPAR γ is correlated with the development of fibrosis in skin, lung, pancreas, heart, and liver [245]. PPAR γ□ expression is decreased in lung fibrosis [246], liver fibrosis [247], kidney fibrosis [248] and scarring alopecia fibrosis [249]. Furthermore, reduced PPAR γ expression precede fibrosis in several human diseases, suggesting a causal role [25,250]. Fibroblasts, with a decreased level of PPAR γ, present an increase expression of TGF-β1, type 1 collagen, and α-SMA [251,252].

PPAR γ and radiation-induced fibrosis
Radiation therapy causes a downregulation of PPAR γ expression [59]. PPAR γ levels are decreased in 3 to 12 hours after a total body irradiation of C57BL/6 mice [59], after UVB irradiation in skin models [269] and in 3 days following a single abdominal dose of 10 Gy [270].
Oxidative stress and inflammation are involved in radiation-induced brain injury [271,272]. Administration of PPAR γ agonists, such as pioglitazone, can decrease the severity of radiation-induced cognitive impairment in rat models [273]. PPAR γ ligands block the radiation-induced activation of NF-κB [274]. PPAR γ agonists administration can contribute to reduce the inflammation and oxidative stress [248,275].
PPAR γ ligands enhance radiation-induced DNA damage in lung cancer cells in vitro [276,277].

Interactions through the Smad pathway
Anti-fibrotic PPAR γ effects can be largely explained through the Smad pathway. PPAR γ promoter has two Smad binding elements and inhibition of PPAR γ favors canonical Smad2/3 signaling [316].

CONCLUSION
Radiation-induced fibrosis is characterized by DNA damage and inflammation. These two processes lead to the activation of TGF-β1 and canonical WNT/β-catenin pathway. TGF-β1 plays a major role in the differentiation of fibroblasts into myofibroblasts. Myofibroblasts appear able to physiological contraction and relaxation. However, in fibrosis myofibroblasts generate a phenomenon of contraction-retraction lasting without relaxation and with an irreversible retraction favored by the synthesis of collagen. Myofibroblasts show a main role in cellular fibrosis in numerous organs such as kidney, heart, lung, liver and wound. TGF-β1 operates in a canonical WNT/β-catenin pathway dependent manner. These two pathways stimulate each other through the Smad pathway or non-Smad pathways like PI3K/Akt pathway. TGF-β1 stimulates myofibroblast differentiation by the stimulation of canonical WNT pathway and the downregulation of PPAR γ expression. TGF-β1 appears to be an interesting therapeutic target in fibrosis [337,338]. PPAR γ agonists stimulate Smad7 for inhibiting Smads pathway what blocks TGF-β signaling. PPAR γ agonist can decrease canonical WNT/β-catenin pathway by activating both Smad 7, GSK-3β and DKK. PPAR γ agonists can also stimulate PTEN, an inhibitor of PI3K/Akt pathway and inhibit the Smad pathway for downregulation TGF-β1 expression. Thus, by both inhibiting TGF-β1 and the canonical WNT/β-catenin pathway, PPAR γ agonists could be interesting preventive targets for radiation-induced fibrosis treatment (cf. Figure 4). www.impactjournals.com/oncotarget Abbreviations APC: adenomatous polyposis coli; DSH: disheveled; FZD: frizzled; GSK-3β: glycogen synthase kinase-3β; LRP 5/6: low-density lipoprotein receptor-related protein 5/6; PI3K-Akt: phosphatidylinositol 3-kinase-protein kinase B; PPAR γ: peroxisome proliferator-activated receptor γ; RIF: radiation-induced fibrosis; TCF/LEF: T-cell factor/ lymphoid enhancer factor; TGF-β: transforming growth factor β.

Author contributions
All the authors have contributed equally to this review.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.