Variation in Expression of Inflammation-Related Signaling Molecules with Profibrotic and Antifibrotic Effects in Cutaneous and Oral Mucosa Scars

Wound healing is a complex biologic process evolving in three phases: inflammation, proliferation, and tissue remodeling controlled by numerous growth factors and cytokines. Oral mucosa wounds heal with significantly less important scars with less numerous macrophages and mast cells and more numerous myofibroblasts than cutaneous counterparts. We analyzed 32 cutaneous and 32 oral mucosa scars for TGFbeta1, TGFbeta2, TGFbeta3, TNFalpha, PDGF BB and FGF1 expression in mesenchymal cells, endothelial cells, macrophages, and multinucleated giant cells. We identified differences in the expression of profibrotic and antifibrotic factors in oral mucosa and skin scars; TGFbeta2 was positive in cutaneous multinucleated giant cells, TNFalpha was positive in cutaneous macrophages, and both were negative in oral mucosa while TGFbeta3 was positive in oral macrophages and mostly negative in cutaneous ones. PDGF BB and FGF1 were positive in oral endothelial cells and oral macrophages and negative in macrophages with opposite positivity pattern in cutaneous scars. Based on these findings, macrophage seems to be the key player in modulating pro- and antifibrotic processes in wound regeneration.


Introduction
Wound healing is a complex biologic process evolving in three phases: inflammation, proliferation, and tissue remodeling [1]. These phases are controlled by myriad of growth factors and cytokines and, to some extent, overlap in time.
Inflammation occurs almost immediately after wound formation-after the blood coagulation and lasts 2-4 days; several cells are chemotactically attracted to the wound site-neutrophils (they engulf bacteria and foreign bodies) and monocytes (they differentiate into macrophages that phagocyte necrotic debris, neutrophils, and foreign bodies); later on, macrophages secrete growth factors that stimulate the formation of granulation tissue [2][3][4]. Proliferation includes several processes such as angiogenesis and fibroblasts proliferation and differentiation towards myofibroblasts (formation of granulation tissue) and epithelial cells proliferation (reepithelization) [5][6][7][8][9]. Tissue remodeling (contraction phase, "maturation") is the last and the longest phase of tissue healing; it consists of collagen synthesis and degradation in order to align the newly formed collagen bundles along tension lines [10][11][12].
Immune system intervenes in the regeneration of different processes such as inflammation and debris clearance in the first stages and proliferation and differentiation of the stem cells in further steps [13,14]. Macrophages present two functional variants: M1 ("proinflammatory" macrophages with high interleukin-(IL-) 12, low IL-10, no IL-13α1, and MS4A4A production) and M2 ("anti-inflammatory" macrophages with IL-13α1, MS4A4A, high IL-10, and low IL-12 production) [15,16]. Both types of macrophages intervene in regeneration in different moments and have to be "switched on/off" accordingly-M1 macrophages have to be active in the beginning of the process (mainly local debris clearance and subsequent cytokines secretion-IL6, TNFalpha, IL1beta, and G-CSF stimulate skeletal muscle proliferation), while M2 macrophages are involved in tissue remodeling (in skeletal muscle lesion they promote myogenic differentiation) [17]; prolonged activity of M1 macrophages or too early activation of M2 macrophages will have negative effects on wound healing [13].
Lymphoid cells are involved in wound healing mainly by cytokine secretion. NK T cells and gamma-delta T cells intervene in liver regeneration, and T-reg cells are involved in muscle regeneration and in oligodendrocyte differentiation and myelin regeneration [18][19][20]. Mast cells intervene in wound healing by releasing several factors with diverse functions including cytokines (IL-1 and TNFalpha) and growth factors (TGFbeta1 or PDGF) and favor collagen deposition [21,22]. They are denser in more mature mice embryos than in younger ones, in dermis than oral lamina propria in adult pigs, and in hypertrophic than normotrophic human scars [23][24][25]; also, drugs with effects of inhibiting mast cell degranulation determine reduced contraction in pigs' wounds [26]. All these arguments favor the hypothesis that mast cells are directly related to fibrotic processes.
From all these data, it is obvious that inflammation can have both positive and adverse effects on wound healing, possibly some data obtained in animals being irreproducible in humans. The study of wound healing in human is extremely difficult due to ethical issues that hamper systematic approach; animal models are also difficult to use due to significant differences in healing process between species; data gathered in small mammals (mice, rats, and/or rabbits) are not reproducible in humans; pigs, however, have similar cutaneous structure and relatively similar healing mechanisms [27]. Systematic studies in pigs and studies in humans reveal differences in wound healing between cutaneous lesions and lesions in the oral mucosa. Oral mucosa wounds heal with significantly less important scars, thus deciphering specific healing mechanisms will offer valuable lessons to be applied to prevent severe scarring [28].
Healing in the oral mucosa involves less numerous macrophages and mast cells and more numerous myofibroblasts; also, TGFbeta expression is reduced in oral mucosa lesions; diminished inflammation and reduced tissue remodeling (wound contraction) favor a scarless healing. Local microenvironment is especially important in regulating myofibroblastic function since oral mucosa lesions heal with less collagen deposition than cutaneous ones despite the increased number of myofibroblasts [29]. Local inflammatory response vary according to the moment of loading of dental implants [30]. Cytokines influence on oral mucosa healing may represent the rationales for using autolog platelet-rich fibrin concentrates to favor local repair [31].
Considering all these pro's and con's, we decide to study the expression of several growth factors and cytokines in cutaneous and oral mucosa scars in humans.

Material and Methods
We analyzed 32 cutaneous scars and 32 mucosal scars. The cases were selected from the archives of the Department of Pathology of Colentina University Hospital; cutaneous scars specimens were reexcisions for previously resected cutaneous tumors; some specimens presented residual tumors but fragments without tumors were selected for this study (at least 4 mm distance between the residual tumor and the area of the scar selected for analysis).
IHC tests used several primary antibodies: TGFbeta1, TGFbeta2, TGFbeta3, TNFalpha, PDGF BB, and FGF1; source, clones, specific pretreatments, and dilutions are specified in Table 1. The IHC stains were performed on an automated immunostainer Leica Bond III using Bond TM polymer refine detection (with DAB chromogen) and Bond TM polymer refine red detection. All the stains with one antibody were performed in the same day (to minimize technique-induced variations) with one negative and positive external control for each antibody; negative control consisted of IHC stains without primary antibody; positive external controls consist of normal human spleen for TGFbeta1, TGFbeta2, and TGFbeta3, human breast cancer for TNFalpha, normal human pancreas for PDGF BB, and normal human kidney for FGF1.
Two independent pathologists examined the slides; a three-grade semiquantitative scale for positivity was used: 0, negative; 1, faint positive (positivity evident when slides were examined with 40x); 2, intense positive (positivity evident when slides were examined with 10x) no matter the number of positive cells; the positivity was recorded for mesenchymal cells (fibroblasts and fibrocytes), endothelial cells, macrophages, and, when present, multinucleated giant cells. The level of positivity was interpreted in correlation with gender, localization, presence of residual tumor, and age of the scar.
The statistical analysis of data was performed using EXCEL and EPIINFO programs; the results were considered statistically significant for P (χ2) was lower than 0.05. This study was approved by Colentina University Hospital Ethic Committee, and all patients included agreed to participate in research studies.

Results
The cases were included in two groups: group A for cutaneous scars and group B for oral mucosa scars.
The scars were reexcised after previous resection of benign lesions (14 cases, 43.74%) or malignant tumor (18 cases, 56.25%; 13 melanomas, one basal cell carcinoma, 2 squamous cell carcinomas, one dermatofibrosarcoma protuberans, and one myxofibrosarcoma). Six cases (18.75%) had residual tumors (melanoma, squamous cell carcinoma, dermatofibrosarcoma protuberans, myxofibrosarcoma, melanocytic nevus, and capillary hemangioma), but, as we stated before, the area selected for analysis in our study was located at some distance from the residual tumor.
All the cases presented granulation tissue or cicatricial collagen within the dermis/corion. In cutaneous scars, the inflammatory infiltrate consisted of lymphocytes and macrophages; 17 cases (53.12%) presented multinucleated giant cells, in some cases, in relation to translucid not structured material (suture material). In oral mucosa scars, the inflammatory infiltrate consisted of lymphocytes, macrophages, and, in frequent cases (18 cases, 56.25%), numerous plasma cells were present, occasionally with intracytoplasmic hyaline inclusions (Russell's bodies) as evidence of immunoglobulin production; no multinucleated giant cells were present in oral scars. Very scanty plasma cells were present in cutaneous scars. Table 2) 3.2.1. TGFbeta1, TGFbeta2, and TGFbeta3. TGFbeta1 was expressed mainly in endothelial cells (27 positive cases, 84.37%; 17 cases 1+ and 10 cases 2+) and mesenchymal cells (25 cases, 78.12%; 24 cases 1+ and one case 2+) and less frequently and fainter in macrophages (12 cases, 37.5%; all of them 1+ positive) and multinucleated giant cells (9 cases, 1+ of 17 cases with multinucleated giant cells) (Figure 2(a)).
In the case of TGFbeta1, the only statistical association was recorded for TGFbeta1 expression in mesenchymal cells in correlation with the age of the scar-we noticed a tendency towards lack and/or diminishing of TGFbeta1 expression in mesenchymal cells in the scars of 2-5 weeks compared with younger (less than 2 weeks) or older (more than 5 weeks) cutaneous scars-P = 0 05 (Figure 2(d)). There was a statistically significant correlation between TGFbeta3 expression in  Figure 1: Age of the scars (weeks). mesenchymal cells and localization of the scar-no scars on the trunk had TGFbeta3 expression in mesenchymal cells while all the scars from special areas had intense diffuse positivity-P < 0 001 (Figure 2(e)).
In oral mucosa biopsies, TGFbeta1 was expressed in all the cases in mesenchymal cells, endothelial cells, macrophages, and plasma cells in all cases with obvious positivity in low power (2+) in almost all the cells (Figure 2(h)). TGFbeta2 was negative in all the cases (Figure 2 (Figure 2(f)). We identified a statistically significant association between TNFalpha expression in endothelial cells and the age of the scar-scars 2 to 4 weeks old tend to express TNFalpha compared with very recent scars or older ones-P = 0 036 (Figure 2(g)). Interestingly, in week 4, all the cases showed TNFalpha positivity within endothelial cells (all male patients, scar located on the head, trunk, and axilla, resections after melanoma or basal cell carcinoma, no residual tumor present in either case).
In oral mucosa biopsies, TNFalpha was negative in all the cases in mesenchymal cells, endothelial cells, and macrophages; cases with plasma cells within the inflammatory infiltrate showed 1+ or 2+ positivity in plasma cells (Figure 2(l)).    areas were compared with PDGF BB expression in same cells in scars originating from the head or trunk (P trunk vs. limbs 0.012; P trunk vs. special areas <0.001; P head vs. limbs 0.002; P head vs. special areas 0.0001) (Figure 3(f)). Also, PDGF BB was overexpressed in all types of cells (mesenchymal cells P = 0 009, endothelial cells P = 0 001, and macrophages P = 0 0001) in scars with no residual tumor (Figure 3(g)-3(i)) and in scars in female patients (mesenchymal cells P = 0014, endothelial cells P < 0 001, and macrophages P < 0 0001) (Figures 3(j)-3(l)). Interestingly, PDGF BB overexpression was noted in multinucleated giant cells in a scar with residual tumors and also in scars in male patients (in both circumstances P < 0 001); we looked in the group of scars with multinucleated giant cells that showed a female predominance compared with the general data of group A (70.58% females in scars with multinucleated giant cells compared with 59.38% in group A) and more numerous cases with residual tumor (23.53% in scars with multinucleated giant cells compared with 18.75% in group A), thus explaining the differences with PDGF BB expression in macrophages. In oral biopsies, PDGF BB was positive 2+ in all cases in endothelial cells, macrophages, and also in plasma cells when present; oral mesenchymal cells were negative for PDGF BB (Figures 3(m)-3(n)).

FGF1
. FGF1 expression was noted in half of the cases (50.00%) in mesenchymal cells (11 cases 1+ and 5 cases 2 +), two-thirds (68.75%) in endothelial cells (16 cases 1+ and 5 cases 2+), and almost three quarters of cases (78.12%) in macrophages (23 cases 1+ and 2 cases 2+). All but one case (94.11%) showed FGF1 positivity in multinucleated giant cells (14 cases 1+ and 2 cases 2+) (Figures 4(a)-4(b)). We looked for FGF1 expression related to location. There was a statistically significant overexpression in the mesenchymal cells of scars from the limbs compared with the mesenchymal cells of scars from special areas (P < 0 001) and in the endothelial cells of scars from the head and trunk compared with the limbs and special areas (P = 0 05). Macrophages expressed more intense FGF1 in scars from the trunk or limbs compared with the head (P = 0 007 and P = 0 0005,

Discussions
We studied several growth factors expression in cutaneous and oral mucosa humans scars: TGFbeta1, TGFbeta2, TGFbeta 3, TNFalpha, PDGF BB, and FGF1. 4.1. TGFbeta1, TGFbeta2, and TGFbeta3. Our data showed the expression of TGFbeta1 (profibrotic growth factor) in all cases in oral specimens and numerous cases in skin fragments; however, there is an obvious relation to the age of the cutaneous scars, the young and old ones having less if any TGFbeta1 expression. TGFbeta2 (the other profibrotic factor) was positive in more than ¾ of the cases of skin specimens in multinucleated giant cells and absent in oral specimens. TGFbeta3 (with probable antifibrotic effects) was positive in all oral specimens in macrophages either intense or mild positivity, while less than half of the cutaneous fragments showed macrophagic positivity. TGFbeta1, TGFbeta2, and TGFbeta3 belong to the TGFbeta/activin/Nodal subfamily of TGFbeta superfamily [32]. They are involved in numerous biological processes such as cellular differentiation, cell migration, apoptosis, cell-cycle arrest, production of extracellular matrix, epithelial to mesenchymal transition, or wound healing [33][34][35]. The overexpression of TGFbeta determines pathologic tissue fibrosis. TGFbeta1 is upregulated in early granulation tissue thus determining an increase in number and activity of SMA positive myofibroblasts thus stimulating neovascularization, collagen deposition, and wound contraction; all these processes are responsible for scar formation in adulthood. Wound in embryos predominantly express TGFbeta3 and to a significantly lesser extent, TGFbeta1; moreover, adding TGFbeta3 to a wound in adult tissues (or neutralizing TGFbeta1 and TGFbeta2) determines less or none scarring [36][37][38][39][40][41][42].
It is known that the major source of TGFbeta in a wound is represented by platelets; platelets release TGFbeta from their secretory granules in inactive form; part of it will be immediately activated (in the first moments of wound appearance) by trombospondin-1 and released together with TGFbeta from platelets secretory granules and also some time later by plasmin which disintegrates blood clot [43,44]. The second wave of TGFbeta activation occurs after macrophages also occupy the wound territory via plasmin (macrophages secrete plasminogen activators) [45,46].
TGFbeta is involved in all phases of wound healing. In the first phase (inflammation), TGFbeta are present from the very beginning. They are supplied in their inactive forms by platelets and are activated by thrombospondin1 and plasmin; in this stage, they act as chemotactic cytokines for neutrophils, monocytes, and macrophages [43,47]. Later on, activated macrophages (local cells or differentiated from recruited circulating monocytes) release supplementary quantities of TGFbeta that stimulates angiogenesis, fibroblasts chemotaxis, and proliferation in second phase (proliferation, "granulation tissue formation") [48]. TGFbeta stimulates the differentiation of fibroblasts into myofibroblasts [40,49] and stimulates the production of components of extracellular matrix (ECM): collagen, fibronectin, and fibronectin receptor [50]; also, by inhibiting both production and activity of matrix metalloproteinases (MMPs) and stimulating the expression of tissue inhibitors of matrix metalloproteinases (TIMPs), they determine a reduction of ECM components degradation [5]. TGFbeta activity in angiogenesis involves all aspects of the process; they promote endothelial cell migration and differentiation and also tubule formation [7].
In maturation phase, TGFbeta intervenes in a delicate balance between MMPs and TIMPs activation and also in increasing lysyl oxidases responsible of cross-linking of collagen with subsequent increase of tensile strength of the wound [51]. Studies performed on animals showed that TGFbeta1 occurs in the wound almost immediately after wound occurrence (5 minutes), then TGFbeta2 and TGFbeta3 occur, surpassing by far TGFbeta1 levels at 24 hours after wound occurrence; there is another peak of TGFbeta1 at 5 days after wound occurrence [52][53][54]. Also, in embryos, wounds express mainly TGFbeta3 (both in epithelial cells and fibroblasts), while TGFbeta1 and TGFbeta2 have very low levels of expression; in adults, wounds express mainly TGFbeta1 and TGFbeta2. Since wound in embryo heal without scarring while those in adults do not, it is a sound presumption that TGFbeta3 favor scar-free healing opening the gates for new therapies [55,56].

4.2.
TNFalpha. In our study, TNFalpha was negative in all cells but positive in plasma cells in oral specimens, while almost half of the cutaneous specimens showed TNFalpha positivity in macrophages; also, some skin cells (both mesenchymal and endothelial cells) had some positivity for TNFalpha also with a tendency towards expression in not so young or old scars. In week 4, all the patients (males, head, trunk, or axilla scars, resection after tumor without residual tumor present) showed TNFalpha expression in endothelial cells.
TNFalpha enhances the inflammation in wound repair, hence contributing to an impaired healing both in chronic and acute wounds. Studies revealed and increased TNFalpha levels (locally, in the wound territory, and systemically) in otherwise healthy elderly patients with active chronic venous ulcers. Also, in animals, experiments with secretory leukocyte protease inhibitor (SLPI) null mice showed that wound healing in these animals is deficient, and TNFalpha is increased in wound area (demonstrated both by RT-PCR and immunohistochemically in local inflammatory cells and epithelia); moreover, local TNFalpha inhibition with anti-TNFalpha antibody accelerates the rate of healing in a dose-dependent manner in SLPI null mice and also improve healing in wild-type mice used as controls [57].
TNFalpha induces the expression of MMP2 and MMP9 and inhibits the local accumulation of fibroblasts either by direct inhibition of chemoattraction or by attracting an impressive number of inflammatory cells in the wound area; both these actions diminish the collagen deposition and impair the wound healing [58,59]  PDGF is a dimeric glycoprotein with mitogenic effects on mesenchymal cells such as fibroblasts, osteoblasts, and smooth muscle cells; also, it is implicated in angiogenesis and fibrosis. Two types of subunits are identified A and B, three variants of PDGF being thus possible: PDGF-AA, PDGF-BB, and PDGF-AB [60]. PDGF stimulates reepithelialization, revascularization, and complete wound closure in ischemic skin and hyperglicemic mice, mainly due to its mitogenic effects on both keratinocytes and endothelial cells [61]. The association of PDGF-BB with TGF-alpha in topic application on wounds in genetically diabetic (C57BL/KsJ-db/db) mice induces the acceleration of healing to a level almost similar to nondiabetic mice [62]. The treatment of leg ulcers in diabetic patients with becaplermin gel (a recombinant human PDGF approved for topical applications) stimulates complete healing and shortens the time to healing [63]. However, caution should be exerted in becaplermin treatment due to the increased risk of both local infections (infected skin ulcer, cellulitis, and osteomyelitis) and death due to malignancies, currently the drug being no longer authorized [64]. 4.4. FGF1. FGF1 expression was noted in our study in similar manner as PDGF BB-oral scars lack FGF1 in mesenchymal cells, while 50% of the skin scars had FGF1 mesenchymal cells positivity, statistically associated with the limbs location and residual tumor present in the vicinity of the scar. More numerous cases of skin scars lack FGF1 expression in endothelial cells and/or macrophages than PDGF BB, but all oral scars show FGF1 expression in these types of cells.
Fibroblast growth factor (FGF) family includes 22 members designated by numbers from FGF1 to FGF23 (FGF15 is lacking in humans). FGF1 (acidic FGF) has a large variety of functions; the most important ones being related to angiogenesis. It induces proliferation of endothelial cells with subsequent organization in tubes and formation of new vessels in myocardium [65]; its function as angiogenic factor overcomes that of consecrated angiogenic factors as vascular endothelial growth factor (VEGF) or platelet-derived growth factor (PDGF) [66]. Based on its angiogenic function, FGF1 is a major player in wound healing [67,68]; it has decreased gene expression in early phases of wound healing in diabetic patients [69], and it has a proven effect on reepitelization in NONcNZO10/LtJ mouse (model for impaired wound healing in type 2 diabetes) [70]. Also, FGF1 is involved in bone regeneration, and it is a promising biomolecule to be used in humans [71].

Conclusions
We identified the differences in the expression of profibrotic and antifibrotic factors in oral mucosa and skin scars; TGFbeta2 was positive in cutaneous multinucleated giant cells, TNFalpha was positive in cutaneous macrophages, and both were negative in oral mucosa while TGFbeta3 was positive in oral macrophages and mostly negative in cutaneous ones. PDGF BB and FGF1 were positive in oral endothelial cells and oral macrophages and negative in macrophages with opposite positivity pattern in cutaneous scars. Based on these findings, macrophage seems to be the key player in modulating pro-and antifibrotic processes in wound regeneration. Further studies are needed in order to establish the mechanisms favoring scarless healing and subsequent application in daily practice.

Data Availability
The datasets used and/or analyzed during the present study are included within the article; if supplemental information is needed, it is available from the corresponding author upon request except confidential data to whom access is restricted in order to protect patients' privacy.

Disclosure
The presented study will be integrated in the original part of the PhD thesis of author Mihai Bucur.