Interplay Between Keratinocytes and Fibroblasts: A Systematic Review Providing a New Angle for Understanding Skin Fibrotic Disorders

Background/Objective: Skin fibrosis is the result of aberrant processes leading to abnormal deposition of extracellular matrix (ECM) in the dermis. In healthy skin, keratinocytes participate to maintain skin homeostasis by actively crosstalking with fibroblasts. Within the wide spectrum of fibrotic skin disorders, relatively little attention has been devoted to the role of keratinocytes for their capacity to participate to skin fibrosis. This systematic review aims at summarizing the available knowledge on the reciprocal interplay of keratinocytes with fibroblasts and their soluble mediators in physiological states, mostly wound healing, and conditions associated with skin fibrosis. Methods: We performed a systematic literature search on PubMed to identify in vitro and ex vivo human studies investigating the keratinocyte characteristics and their interplay with fibroblasts in physiological conditions and within fibrotic skin disorders including hypertrophic scars, keloids, and systemic sclerosis. Studies were selected according to pre-specified eligibility criteria. Data on study methods, models, stimuli and outcomes were retrieved and summarized according to pre-specified criteria. Results: Among the 6,271 abstracts retrieved, 73 articles were included, of which 14 were specifically dealing with fibrotic skin pathologies. Fifty-six studies investigated how keratinocyte may affect fibroblast responses in terms of ECM-related genes or protein production, phenotype modification, and cytokine production. Most studies in both physiological conditions and fibrosis demonstrated that keratinocytes stimulate fibroblasts through the production of interleukin 1, inducing keratinocyte growth factor (KGF) and metalloproteinases in the fibroblasts. When the potential of keratinocytes to modulate collagen synthesis by healthy fibroblasts was explored, the results were controversial. Nevertheless, studies investigating keratinocytes from fibrotic skin, including keloids, hypertrophic scar, and scleroderma, suggested their potential involvement in enhancing ECM deposition. Twenty-three papers investigated keratinocyte proliferation differentiation and production of soluble mediators in response to interactions with fibroblasts. Most studies showed that fibroblasts modulate keratinocyte viability, proliferation, and differentiation. The production of KGF by fibroblast was identified as key for these functions. Conclusions: This review condenses evidence for the active interaction between keratinocytes and fibroblasts in maintaining skin homeostasis and the altered homeostatic interplay between keratinocytes and dermal fibroblasts in scleroderma and scleroderma-like disorders.


INTRODUCTION
Fibrosis is a complex process characterized by abnormal deposition of extracellular matrix (ECM), which can lead to altered tissue architecture impacting organ function and survival (1). Fibroblasts are endowed with the full machinery allowing deposition and resorption of ECM, which under homeostatic conditions is continually renewed. Fibroblast synthetic and degradative capacities are modulated by a variety of stimuli, which include soluble factors, cell-to-cell interactions, matrix stiffness, and tensile forces, oxygen levels, epigenetic changes, cell aging, telomere length, and cell survival (2).
Most importantly, fibroblasts are under the influence of a variety of other cell types, which are specifically resident in the tissue undergoing fibrotic changes or professional inflammatory cells recruited in the tissue (3). Soluble mediators of inflammation and, in particular, cytokines and growth factors are deeply involved in regulating fibroblast migration, proliferation, metabolism, and ECM deposition (4). In particular, TGF-β is considered a master mediator of fibrosis (5) relevant for the recruitment and trans-differentiation of cell precursors into myofibroblasts. These are cells with contractile properties associated with the expression of α-smooth actin and with a very high capacity to synthetize and release ECM components such as type I and type III collagen, fibronectin, and tenascin among others (6)(7)(8).
Enhanced ECM deposition is physiologically important and part of the reparative process in damaged tissues. Tissue damage can result from infectious agent assaults, or traumatic wounds, or the effect of physico-chemical injuries. Thus, enhanced ECM deposition is part of normal reparative inflammatory processes, and the characterization of wound healing has historically been fundamental to understand processes leading to fibrosis. What distinguishes controlled ECM enhanced deposition from pathological fibrosis is that the many mechanisms, which are important to halt ECM deposition, are relatively deficient to oppose persistent stimulation (1). Thus, perturbed homeostasis resulting from a variety of origins may explain excessive ECM deposition and pathological tissue fibrosis.
The skin is a tissue that can undergo fibrosis in response to local stimuli but also, while more rarely, as a result of systemic inflammatory disorders. Systemic sclerosis (SSc) or scleroderma is a prototypic condition in which dysregulated inflammation associated with autoimmunity and widespread vascular dysfunction results in skin and internal organs' pathological fibrosis (9). Localized skin fibrosis is observed in Morphea, hypertrophic scars, keloids, and many other conditions with metabolic, vascular, or genetic origins (10). Since long, it is known that the traumatic loss of epithelial cells (keratinocytes) and the following process of re-epithelization are spatially and chronologically important events regulating fibroblast activation and ECM deposition (wound healing) (11). More recent and less developed is our understanding of the role of keratinocytes for their capacity to regulate ECM deposition in non traumatic skin fibrosis. Similarly, relatively little is known about the role of fibroblasts and ECM for their influence on keratinocyte proliferation, differentiation, and epidermis generation. The present work aims to systematically review published evidence on the reciprocal role of keratinocytes and fibroblasts and their soluble products under the angle of human skin fibrosis.

Literature Search
We searched the literature on PubMed up to, and including, August 31, 2019. We conducted our search using a combination of free terms and controlled vocabulary terms by Boolean operators (AND, OR). The terms used were: ["myofibroblast" (Mesh) OR "mesangial cells" (Mesh) OR "fibroblast * " (tiab) OR "fibro * " (tiab)] AND ("dermis" (Mesh) OR "derm * " (tiab) OR "skin" (Mesh) OR "cutis * " (tiab) OR "cutan * " (tiab) OR "epithel * " (tiab) OR "keratin * " (tiab)]. Keywords were detected in titles and abstracts. We also reviewed reference lists of the included full text and of other reviews on the topic to find additional reports. The systematic review was performed according to the PRISMA guidelines (12).

Inclusion Criteria
Studies fulfilling the following inclusion criteria were included in the present review: in vitro or ex vivo studies on cells or tissues of human origin from healthy donors or individuals affected by fibrotic pathologies with the exclusion of tumors or cancers. The focus was put on the keratinocyte-fibroblast interactions and the methods used to investigate these interactions, with no restrictions.

Exclusion Criteria
We excluded studies on animals, animal tissues, animal cells, and animal pathologies. We excluded human studies on hair follicles, neoplastic conditions or neoplastic cell lines, as well as inflammatory skin pathologies with no evident fibrotic component. We excluded reviews and commentaries. We excluded studies not describing keratinocyte-fibroblast interactions. We excluded studies when the full text was not available and when the language was other than English.

Data Extraction
We used standardized data extraction forms. For each study, the following items were collected: first author, year of publication, type of experimental models and methods used for investigating keratinocyte-fibroblast interactions, type of culture medium, type of stimuli and their outcomes, and mediators potentially responsible for the observed effect. Initially, titles and abstracts of all identified citations were reviewed. Full text of potentially relevant articles was screened and checked for eligibility. Disagreements about the inclusion of articles were resolved by two of the authors (BR, CC). In detail, abstract and full texts were reviewed together by the authors to reach a shared decision in case of disagreement.

Summarizing and Interpreting the Data
Data were subdivided according to the main objective of the identified studies in two categories: studies reporting mainly the effects of keratinocytes on fibroblasts ( Table 1), studies reporting mainly the effects of fibroblasts on keratinocytes ( Table 2). Studies specifically addressing fibrotic skin disorders are summarized in Table 3. Reporting was focused on cell proliferation, differentiation, and migration, extracellular matrix components, and turnover, identification of soluble factors of inflammation and growth factors, skin pathology, type of activating stimuli. The studies describing reciprocal effects on both cells types were listed in both categories and tables.

Literature Search
The literature search resulted in 6,250 hits from PubMed and 21 from reference screening. After the screening of titles, abstracts, and full texts, 73 articles were included in the present review as reported in the flowchart (Figure 1). Six studies examined simultaneously the reciprocal effect of keratinocytes on fibroblasts and of fibroblasts on keratinocytes. Fifty-six papers explored the effects of keratinocytes on fibroblasts, 19 of which investigating soluble factors of inflammation and growth factors, 17 reporting cell proliferation, differentiation, and migration, extracellular matrix components, and turnover, 10 reporting responses to soluble factors and physical stress, 14 referring to specific skin pathologies including SSc, keloids, hypertrophic scars.
Twenty-three studies investigated the effects of fibroblasts on keratinocytes, 14 of which investigated keratinocyte proliferation, differentiation, activation, survival, and adhesion; two investigated keratinocyte production of inflammatory mediators. Six papers focused on responses to soluble factors or altered expression of transcription factors. Two papers investigating the effect of fibroblasts on keratinocytes focused on pathological conditions (SSc and keloids).

Experimental Models Used to Assess the Crosstalk Between Keratinocytes and Fibroblasts
The experimental models used to assess the crosstalk between keratinocytes and fibroblasts are schematically reproduced in Figure 2, and analytically reported in Tables 1, 2. Many papers combined two or more experimental models. The simplest and straightforward experimental approach used in 11 papers was based on the use of the conditioned medium (CM) to be transferred from a cell type to the other (Figure 2A). A potential drawback may be related to differential media requirements for optimal survival, proliferation, and differentiation of keratinocytes and fibroblasts. Two papers have used centrifugation of CM to enrich for keratinocyte microvesicles or exosomes to be tested on fibroblasts (20,38). Physical coculture of keratinocytes with fibroblasts was used in 14 papers ( Figure 2B). The Transwell technology has been adapted to assess many different cell combinations for a total of 16 papers. Thus, keratinocytes put in the upper well-could have been cultured in monolayers or could be grown to reach stratification and differentiation to become epidermal equivalents (EE) ( Figure 2C). Similarly, fibroblasts put in the lower well could have been grown in monolayers adherent to plastic or embedded in a matrix, thus generating a dermal equivalent (DE) (Figure 2C). Skin equivalents generated in cultures based on air/liquid interphase have been used in 24 papers ( Figure 2D). Finally, full skin organotypic culture has been used in three papers ( Figure 2E) (13,28,60). The proportion of studies using these culture methods is reported in Figure 2F.

Effects of Keratinocytes on Fibroblast Proliferation, Differentiation, Migration, Extracellular Matrix Components, and Turnover
Type-I collagen (col-I) production by fibroblasts submitted to the influence of keratinocytes was found enhanced in four studies (13,16,32,66) and decreased in eight of 12 studies in HD (26,35,39,41,43,50,61,65). Fibronectin (FN) production was reported to be enhanced in five of six studies (33,44,45,62,66) and decreased in one of six (35). Other ECM components including tenascin, versican, lumican, and thrombospondin were variably reported to be increased in two or decreased in one of three studies (20,33,49). The production of matrix metalloproteinases (MMP) including MMP-1, MMP-2, MMP-3, MMP-8, MMP-9, and MMP-12 when investigated was always found to be increased in fibroblast under the influence of keratinocytes for a total of 12 studies (13,18,20,34,35,38,40,44,(46)(47)(48)51). The production of tissue inhibitor of metalloproteinases (TIMP)-1, TIMP-2, and TIMP-3 was variably reported to be increased in two (34,48) or decreased in one of three studies (35). Decreased production of cathepsins was reported in one paper (50). The production of plasminogen activator inhibitor (PAI) was variably reported to be increased in two (44,50) and decreased in one of three studies (35). Uroplasminogen (uPA) was increased in two of two papers (35,40). Decreased proliferation of HD fibroblasts was reported in two (45,57) and unchanged in one of three (55). Enhanced fibroblast migration was reported in three of three papers (20,25,27). Four papers reported enhanced HD fibroblast proliferation mediated by keratinocyte produced S100A9 (24), or by UVB-exposed keratinocytes cultured in estradiol (30) or enhanced migration (25) mediated by microvesicles released by keratinocytes (20). Cell adhesion and cadherin expression were found increased in two papers each (20,44), and gel contraction increased in three of three (20,49,62). The expression of α-smooth muscle actin (αSMA) was variably reported to be increased in two (49,50) or decreased in one of three studies (35).

Effect of Soluble Factors and Physical Stress Acting on Keratinocytes for Their Influence on Fibroblasts
Within the context of wound healing, the effects on fibroblasts of dehydration or hyper-hydration of partially stratified keratinocytes was investigated in six papers (15,19,22,34,52). Robustly, five of them examining col-I production reported enhanced col-I production when keratinocytes were dehydrated compared to their normally hydrated counterpart. When tested, consistently with the results on col-I, αSMA expression was increased by dehydration in three of three papers (15,19,22). One of these papers focused on MMPs and TIMP-1 showing that hydration enhances MMP1, MMP8, and MMP13 and decreases TIMP-1 production (34). One paper reported that hyper-hydration of partially stratified keratinocytes enhances the production of KGF by fibroblasts (34). Physical stimuli were investigated in two papers. One focusing on keratinocyte exposure to UVB in the presence of estradiol (E2) reported that when exposed, fibroblasts responded by enhanced proliferation and EGF and lumican production (30). Another paper reported that keratinocytes exposed to micro-particles enhanced the fibroblast production of IL-1β, IL-6, TNF-α, PGE2, MMP-1, and MMP2 (14).

Specific Skin Pathologies
Eight papers investigated the influence of keratinocytes on fibroblasts in keloids. Three reported enhanced fibroblast proliferation (53,54,57) of which one also enhanced TGF-β production (53). One paper pointed to a reduced expression of S100A7 and S100A15, which were suggested to act as inhibitors (21); one concentrated on enhanced col-I production (56). A single paper reported enhanced production of TGF-β, oncostatin M (OSM), fibroblast-activating protein (FAP), αSMA, and laminin, compared to HD (37). Enhanced production of KGF by keloid keratinocytes resulting in enhanced release by fibroblasts of OSM, and col I was reported in one paper (28). One paper reported enhanced IL-18 production by keratinocytes resulting in enhanced production of IL-6, IL-8, and col-I by fibroblasts (32).
Two papers investigating the influence of keratinocytes on fibroblasts in hypertrophic scars reported an increase in matrix thickness, PDGF, and bFGF production compared to HD (31,58).

Effect of Fibroblasts on Keratinocyte Proliferation, Survival, Adhesion, Keratin Expression
The effect of fibroblasts on keratinocyte proliferation was investigated in nine papers and reported to be increased in all of them (72, 74, 76, 82-84, 86, 87). Keratinocyte survival was reported to be enhanced in the presence of fibroblasts for reduced apoptosis, reduced expression of Bcl2, and enhanced expression of p53 (72,78,80). Keratinocyte adhesion and cadherin expression were reported to be enhanced in the presence of fibroblasts in two papers (72,74). Keratinocyte differentiation was robustly reported to be enhanced in the presence of fibroblasts in four of four papers (73,(80)(81)(82), two of which specifically attributed this effect to fibroblasts from papillary compared to superficial dermis (73,81). The influence of fibroblasts on keratin expression by keratinocytes was studied in two papers, one reporting enhanced expression of keratin 8 (29) and two others a reduced expression of keratins 6, 16, and 17 (80,83). Fibroblasts were reported to enhance the deposition of basal membrane components by keratinocytes in three papers (73,80,81).

Effect of Soluble Factors or Altered Expression of Transcription Factors in Fibroblasts for Their Influence on Keratinocytes
Fibroblasts exposed to bFGF were reported to enhance keratinocyte differentiation in one study (70) and to reduce keratinocyte production of TGF-β in another study (75). Fibroblasts exposed to stromal cell-derived factor (SDF)-1 were reported to enhance keratinocyte proliferation and stratification in one study (71). Fibroblasts exposed to testosterone were reported to decrease keratinocyte differentiation in one study (69). Fibroblasts with inhibited expression of both peroxisome proliferator-activated receptor (PPAR)α and PPARδ were reported to enhance keratinocyte proliferation and their production of IL-1 and activator protein (AP)-1-targeted genes in one paper (79). (Table 3) Fibroblasts from keloids and SSc were reported to enhance the production of oncostatin M (OSM) by HD keratinocytes in one paper (28).

Soluble Mediators of Inflammation Influencing the Crosstalk of Keratinocytes With Fibroblasts
Among the soluble mediators of inflammation produced by keratinocytes affecting fibroblast responses, IL-1 is robustly reported to be a relevant keratinocyte-derived mediator inducing fibroblast activation in eight of eight papers addressing this aspect (13,17,23,36,(63)(64)(65)82). Conversely, three of three papers reported that fibroblasts regulate epidermal homeostasis (proliferation and differentiation) through the secretion of KGF (28,70,83). Keratinocyte production of TGF-β by itself or in association with other mediators including IL-1 and ET-1 was reported to enhance col-I production by fibroblasts in three of four papers (13,36,49). The role of keratinocyte-derived stratifin, also known under the name 14.3.3 sigma, has been extensively investigated by one group that showed its role in enhanced MMPs and reduced col-I production by fibroblasts (33,38,39,44,45,47,51). Further, in the presence of fibroblasts, the same group showed enhanced stratifin production by keratinocytes (77). Keratinocytes were shown to produce fibronectin resulting in enhanced fibroblast migration (25). Keratinocytes were reported to produce IL-19, which resulted in enhanced KGF production by fibroblasts. In its turn, KGF enhanced the IL-19 production by keratinocytes. Chemokine (C-C motif) ligand (CCL)26 (eotaxin-3) production by keratinocytes was reported to enhance fibroblast proliferation and motility (27). High mobility group box 1 (HMGB1) production by keratinocytes was reported to enhance fibroblast activation and αSMA expression (15). One study reported that parathyroid hormone-related protein (PTHrP) released by keratinocytes enhanced the production of KGF by fibroblasts (85). Finally, two studies reported that keratinocyte production of vesicles (whether microvesicles or exosomes) enhanced fibroblast activation with higher production of MMPs and a number of other mediators detected by microarrays (20,38,45).

Specific Skin Pathologies
In SSc, one paper reported enhanced col-I production induced by keratinocytes in a TGF-β-independent fashion (16), while a role for TGF-β was reported in two (13,36). Expression of calprotectin, also known as S100A8/A9, was reported to be increased in keratinocytes from hypertrophic scars and SSc, resulting in enhanced fibroblast production of col-I and CTGF in two studies (19,24). Psoriasin, also known as S100A7, was reported to be decreased in keratinocytes from keloids, a finding associated with increased col-I production by fibroblasts (21). Single papers have addressed the role of several other mediators. Collectively, the production by keratinocytes of IL-18 in keloids (32), PGE2 in dehydration (22), reduced TIMP production in hypertrophic scars (31), was associated with enhanced col-I production.

DISCUSSION
Our systematic review has retrieved 73 published papers investigating the interplay between keratinocytes and fibroblasts. Our main aim was to focus on fibrosis. While only 14 papers specifically aimed at skin fibrotic disorders, many focused on wound healing, which is a physiological condition considered to have several analogies with skin fibrosis, at least during the initial proliferative and synthetic phase (1). In this respect, it has to be underlined that the relatively little number of papers dedicated to this topic reflects, at least in part, the complexity of experimental settings needed to investigate the interactions between these two cell types with different requirements for optimal in vitro survival. This is particularly true for keratinocytes, which may undergo proliferation and differentiation under specific and mutually exclusive culture conditions. Not unexpectedly, compared to their undifferentiated counterpart, keratinocytes undergoing differentiation synthetize a distinct panel of proteins and soluble mediators and react differentially to exogenous stimuli, including those potentially provided by fibroblasts. For instance, a number of papers investigating the effect of keratinocytes on fibroblasts have used culture supernatants as effectors on fibroblasts. Furthermore, the supernatants may have been generated from non-primary keratinocyte cell lines, undifferentiated primary keratinocyte lines, and, in some instances, differentiated and stratified keratinocytes. Likely, the most physiologically relevant approaches to address the crosstalk between keratinocytes with fibroblasts were based on the use of epidermal equivalents or skin equivalents of full skin approaches. This information is provided in Tables 1, 2, and Figure 2F. However, as limitation of our review, we have not weighted the relevance of the reported results based on the experimental assay used. Of particular importance to critically apprise the mutual relationship between keratinocytes and fibroblasts is the role of the basement membrane, which separates, holding together, the epidermis and dermis in vivo (88). In this respect, the papers specifically studying the structure and composition of the in vitro-generated basement membrane acquire additional value (73,80,81). The experimental settings leading to the results here reviewed were mostly based on the use of conditioned medium, skin equivalents, transwells, and cocultures each contributing to about one fifth of the total. Additional approaches took advantage on combinations of methods, histology on skin tissues, and more demanding skin explants. Given the existence of a basement membrane separating the epidermis from dermis, cell-to-cell contact effects between keratinocytes and fibroblasts could have limited physiological relevance. An additional point to consider is the possibility that dermal fibroblasts adapting to in vitro culture growth may lose some of their tissue-specific characteristics, then impacting on their effects on keratinocytes.
Notwithstanding these considerations, the wealth of retrieved papers clearly highlights the interest in the problematics of keratinocyte to fibroblast crosstalk and the capacity of these cell types to mutually influence each other. The majority of the retrieved papers investigated how keratinocytes interact with fibroblasts in the context of wound healing, using keratinocytes and fibroblasts generated from healthy individuals, with 14 papers investigating how this interaction is modified and characteristic of pathologic conditions. Overall, a large agreement characterizes the results indicating that also in homeostatic conditions, the crosstalk between keratinocytes and fibroblasts has an impact on both cell types and ultimately on the structure of both epidermis and dermis. However, the outcome of the interactions and the factors contributing to the crosstalk were heterogeneously investigated, and in some cases, the reported results were inconsistent.
Strong evidence supports a role for keratinocyte-produced IL-1 in inducing fibroblast production of KGF, GM-CSF, TGFα, IL-6, IL-8, IL-1, the expression of COX2 and PGE2 production (Figures 3A-D). In its turn, KGF, GM-CSF, and PGE2 promote keratinocyte proliferation and favor proper keratinocyte differentiation (Figures 3A,B,D). Simultaneously, TGFα enhances the expression on keratinocytes of both receptors for KGF (FGFR2b) and GM-CSF (GM-CSF-R), thus favoring keratinocyte responses to these ligands ( Figure 3B). Furthermore, PGE2, IL-6, GM-CSF, and KGF produced by fibroblasts enhance IL-1 production by keratinocytes, thus promoting a positive forward amplification loop ( Figure 3D). Not last, the autocrine production of IL-1 by fibroblasts may amplify fibroblast production of several mediators including KGF and GM-CSF (Figures 3A,D). It is of interest to notice that the circuitries here reported and highlighted in Figure 3 all propose Synergistic effect of IL-1 with IL-6 family members (in red) on the crosstalk. Arrowheads indicate enhancement. bFGF, basic fibroblast growth factor; COX2, Cyclooxygenase 2; CXCL, chemokine containing the CXC motif; EGF, epidermal growth factor which comprises multiple mediators including transforming growth factor-α, amphiregulin, heparin binding-EGF, and epiregulin; ENA-78, Epithelial neutrophil-activating protein 78; FGF, fibroblast growth factor; GM-CSF, granulocyte-monocyte colony stimulating factor IL, interleukin; KGF, keratinocyte growth factor (also known as FGF7); LIF, leukemia inhibitory factor; MCP-1, monocyte chemotactic protein-1; OSM, oncostatin M;PGE2, Prostaglandin E2; TGF, Transforming growth factor; VEGF, vascular endothelial growth factor. positive feedforward effects. It is very unlikely that this reflects the reality since biological systems have inbuilt physiological modulators and inhibitors. Thus, further homeostatic factors and inhibitory mechanisms important in the crosstalk between keratinocytes and fibroblasts likely will be identified in future work. It is, however, true that feed-forward mechanisms may participate in pathological processes.
Several papers retrieved in our systematic review address the effect of keratinocytes on ECM component production by fibroblasts. Controversial are the results reported on collagen deposition and other ECM components. Thus, while a majority of studies (eight of 12) demonstrate an inhibitory role of keratinocytes, four of 12 papers reported an enhancing effect of keratinocytes on collagen production ( Figure 4A). One paper proposed for the enhanced production of collagen a TGF-βindependent keratinocyte contribution (16), the others via TGFβ. It is difficult to reconcile these contradictory results; however, substantial differences in the experimental settings including the culture medium composition, the differentiation status of keratinocytes, as well as the methods used to quantify collagen may explain the differences observed. For future studies, it will be important to standardize further the experimental settings to allow robust comparisons across results.
CTGF was reported to be downregulated by keratinocytes. However, keratinocytes form keloids, hypertrophic scars, and SSc that distinctly showed enhanced expression of the alarmin S100A8/A9, which directly favored collagen and CTGF production and αSMA expression by fibroblasts, thus pointing to pathology-associated differences compared to controls ( Figure 4C). Further, among the states of perturbed homeostasis, keratinocyte dehydration was frequently investigated and consistently found to favor profibrotic responses in fibroblasts ( Figure 4C). Finally, decreased expression of S100A7 and S100A15 in keloids may mechanistically be linked to enhanced collagen production since they were reported to be inhibitory (21).
In contrast with the inconsistent results reported on collagen production, there was a strong agreement among reports showing that keratinocytes enhance MMP production by fibroblasts ( Figure 4B). One paper reported, in addition, a decreased production of TIMP by fibroblasts under the influence of keratinocytes. Overall, the picture that emerges from these studies supports a model in which keratinocytes favor ECM turnover by favoring MMP over a concomitant decreased or alternatively increased collagen production by fibroblasts. Of interest, one paper exploring this issue reports that in SSc compared to healthy controls, the ratio of collagen over MMP-1 is distinctly in favor of enhanced deposition, such as an effect not being present in HD (13). IL-1 stands out among the soluble factors produced by keratinocytes involved in the enhanced production of MMPs by fibroblasts ( Figure 4B). In addition, one group has devoted enormous attention to the role of stratifin expressed by keratinocytes in inducing MMP and decreasing collagen, CTGF, insulin-like growth factor (IGF), bFGF, glycosaminoglycan (GAG) production as well as the expression of αSMA by fibroblasts (Figures 4A,B). Of further interest, the possibility that exosomes released by keratinocytes may be, at least in part, mediators of this effect (38) and that soluble factors released by fibroblasts may modulate stratifin production by keratinocytes (77).
Concerning the influence of fibroblasts on keratinocytes, the literature provides solid and consistent evidence that, in the presence of fibroblasts, keratinocytes show enhanced proliferation, reduced apoptosis, physiological differentiation, enhanced basement membrane deposition (Figure 5). These effects are mediated mostly by KGF (Figure 3A). Other important soluble factors are HGF and PGE2 (Figure 5). Of interest, TGF-β and myofibroblasts exert an inhibitory role particularly on keratinocyte differentiation and proliferation. Further, the deficiency of PPARα and PPARδ in fibroblasts promotes keratinocyte proliferation and, among others, enhanced IL-1 production ( Figure 5). However, only two papers provided data on the effect of fibroblasts on keratinocytes in fibrotic disorders (28,37) showing stronger effects of fibroblasts from fibrotic disorders. The paucity of studies exploring this topic most likely may be explained by the fact that keratinocytes are not currently integrated in physiopathological models of fibrosis development. However, the recent documentation of altered keratinocyte differentiation and inflammatory response in skin fibrosis begs the question whether these abnormalities are primary or secondary to dermal fibrosis. Thus, at the moment, it remains an interesting area of research to investigate whether fibroblasts generated form SSc may affect keratinocyte behavior.

CONCLUSIONS
Evidence generated in recent years and reviewed here strengthen a role for keratinocytes in participating in dermal fibrosis. Whether this is a modulatory role rather than an initiation role remains to be established firmly. Murine models support the possibility that keratinocytes may indeed instruct fibroblast to enhance ECM deposition. For instance, Brakebusch et al., observed the development of dermal fibrosis in a mouse deficient for the β1 integrin subunit in keratinocytes (89). Similarly, the keratinocyte-specific genetic deletion of Friend leukemia virus integration 1 (Fli1) induced in mice a SSc-like phenotype with skin, esophageal, and lung involvement (90). To further FIGURE 5 | Effects of fibroblasts on keratinocytes. Arrowheads indicate enhancement. Blunted heads indicate inhibition. BM, basal membrane; HB-EGF, heparin binding EGF like growth factor; HGF, hepatocyte growth factor; IL, interleukin; KGF, keratinocyte growth factor; PDGF, platelet derived growth factor; PGE2, prostaglandin E2; PPAR, peroxisome proliferator-activated receptor; TGF, transforming growth factor. strengthen this point, a system level analysis based on consensus clustering of genes expressed in human SSc skin revealed that keratinocytes make major connections with the inflammation network, thus highlighting their role in SSc (91).
Further work is required to better understand the reciprocal role of keratinocytes and fibroblasts and their interactions at initiation and stabilization of skin fibrosis. In this respect, novel sophisticated technical approaches may provide important new information. For instance, the generation of human skin equivalents where keratinocytes, dermal fibroblasts, and endothelial cells are grown on a biological scaffold and perfused at physiological pressure have very recently been shown to respond to fibrotic stimuli (92). Thus, vascularized skin equivalents can replicate key features of fibrotic skin and may serve as a platform to better understand the interplay between different cell types including keratinocytes and fibroblasts in pathophysiologically relevant human setting. Skin generated from stem cells and human organoids or humanized mouse models may provide additional tools for approaching similar questions (93). A complementary ex vivo approach would be the use of precision cut slices of healthy and diseased human skin, which would recapitulate the organ architecture then analyzed by advanced imaging techniques (94). Further, single-cell mRNA studies from cells freshly obtained from healthy and diseased skin will expand our knowledge, particularly comparing wound healing to fibrotic skin disoders. These approaches will possibly capture the subtle mechanisms involved in rapid termination of ECM deposition, which very likely distinguish physiological reparative processes from pathological fibrosis. This may lead to the development of novel therapeutic strategies.

AUTHOR CONTRIBUTIONS
BR performed the systematic review of the literature. BR, NB, and CC drafted the manuscript, reviewed its contents, and approved its final version.

FUNDING
This work was supported, in part, by grant 310030-159999 from the Swiss National Science Foundation and by a grant from sclerodermie.ch (Swiss Scleroderma Patient organization) to CC. BR was supported, in part, by a grant from the Ernst and Lucie Schmidheiny Foundation.