https://orcid.org/0000-0002-2075-434X
Abstract
Recessive dystrophic epidermolysis bullosa (RDEB) is a severe skin fragility disorder caused by mutations in the Col7a1 gene. Patients with RDEB suffer from recurrent erosions in skin and mucous membranes and have a high risk for developing cutaneous squamous cell carcinoma (cSCCs). TGFβ signaling has been associated with fibrosis and malignancy in RDEB. In this study, the activation of TGFβ signaling was demonstrated in col7a1−/− mice as early as a week after birth starting in the interdigital folds of the paws, accompanied by increased deposition of collagen fibrils and elevated dermal expression of matrix metalloproteinase (MMP)-9 and MMP-13. Furthermore, human cord blood-derived unrestricted somatic stem cells (USSCs) that we previously demonstrated to significantly improve wound healing and prolong the survival of col7a1−/− mice showed the ability to suppress TGFβ signaling and MMP-9 and MMP-13 expression meanwhile upregulating anti-fibrotic TGFβ3 and decorin. In parallel, we cocultured USSCs in a transwell with RDEB patient-derived fibroblasts, keratinocytes, and cSCC, respectively. The patient-derived cells were constitutively active for STAT, but not TGFβ signaling. Moreover, the levels of MMP-9 and MMP-13 were significantly elevated in the patient derived-keratinocytes and cSCCs. Although USSC coculture did not inhibit STAT signaling, it significantly suppressed the secretion of MMP-9 and MMP-13, and interferon (IFN)-γ from RDEB patient-derived cells. Since epithelial expression of these MMPs is a biomarker of malignant transformation and correlates with the degree of tumor invasion, these results suggest a potential role for USSCs in mitigating epithelial malignancy, in addition to their anti-inflammatory and anti-fibrotic functions.
This study demonstrates the activation of TGFβ signaling in neonatal col7a1−/− mice (recessive dystrophic epidermolysis bullosa [RDEB] animal model) starting in the interdigital folds of the paw skin and suggests that an early medical intervention before full activation of TGFβ signaling may be ideal for preventing fibrosis and mitten deformity in patients with RDEB. Administration of cord blood-derived unrestricted somatic stem cells (USSCs) in newborn col7a1−/− mice suppresses TGFβ signaling-mediated fibrosis by decreasing dermal expression of MMP-9 and -13 and increasing anti-fibrotic TGFβ3 and decorin. USSC coculture also suppressed the elevated MMP-9 and MMP-13 expression in RDEB patient-derived keratinocytes and cSCCs, which correlates with tumor invasion and malignant transformation. These preliminary data suggest the potential effects of USSC administration on modulating dermal-cSCC microenvironment in patients with RDEB.
Introduction
Recessive dystrophic epidermolysis bullosa (RDEB) is a severe inherited skin blistering disease caused by mutations in the COL7A1 gene that encodes type VII collagen (C7) [1-3]. Clinical manifestations of RDEB range from mild localized blistering to a severe generalized form (RDEB-sev gen; previously termed Hallopeau-Siemens subtype) characterized by erosions and blistering, mutilating scarring, pseudosyndactyly, and a high risk of developing aggressive and rapidly metastasizing cutaneous squamous cell carcinomas (cSCCs) [2-4].
RDEB-related cSCCs form preferentially within chronic wounds or cutaneous scars [5]. The injury-induced dermal stiffness correlates with a fibrotic microenvironment that is permissive for epithelial malignant initiation and progression [6]. Recent studies demonstrated the association of TGFβ signaling with fibrosis and malignancy in animal models and patients with RDEB [7-9]. TGFβ1 plays an essential role in different phases of wound healing [10]. However, excessive TGFβ1 leads to abnormal deposition of extracellular matrix (ECM), fibrosis, and scar formation [11]. In contrast, TGFβ3, an isoform of TGFβ1, has shown TGFβ1-antagonistic effects in scar formation, presumably by competing with TGFβ1 for binding to TGFβ receptor II (TGFβRII) and initiating a different signaling transduction pathway to promote scar-free healing [10, 12]. Elevated levels of TGFβ1 were identified in both skin wounds and serum of the patients with RDEB [9]. Overexpression of TGFβ1 was also reported in untransformed RDEB-derived keratinocytes and even more so in RDEB-derived cSCC [7]. TGFβ signaling was recently suggested to be a phenotypic modulator in monozygotic twins with identical COL7A1 mutations [8]. Moreover, the level of proteoglycan decorin (DCN), which binds TGFβ1 and neutralizes its biological activities, was significantly upregulated in the less-affected twin. Further supporting the role of TGFβ signaling in fibrosis of RDEB, inhibition of TGFβ signaling with losartan reduced fibrosis and prolonged the progression of fibrotic digit fusion in C7-hypomorphic mice [9]. Inhibition of TGFβ signaling also resulted in decreased endothelial cell tube formation in cSCCs and C7 has been suggested to act as a tumor suppressor by reducing TGFβ signaling [13].
Multiple and intricate mechanisms appear to underlie the phenotypic manifestations in patients with RDEB. Breitenbach et al. reported that the scar tissue from several RDEB patients with pseudosyndactyly had varied amounts of TGFβ1 and no significant changes in the level of smooth muscle actin (αSMA), a marker for myofibroblast differentiation [14]. The authors postulated that the extracted human tissue might have been at a late-wound healing stage in which fibrotic process had already been terminated [14]. Alternatively, in addition to TGFβ1-induced myofibroblast differentiation, other mechanisms may contribute to RDEB mitten deformity [14]. Albrengues et al. recently demonstrated that leukemia inhibitory factor (LIF), a member of the interleukin (IL)-6 proinflammatory cytokine family, initiates epigenetic switching resulting in constitutive activation of JAK1/STAT3 and proinvasive activation of fibroblasts by both TGFβ-dependent and nondependent mechanisms [15, 16]. As higher plasma levels of IL-6 were observed in patients with RDEB [17], activated JAK–STAT signaling in addition to TGFβ signaling may also play a role in fibrosis and/or fibroblast activation in patients with RDEB.
The RDEB phenotypic variability has also been related to the heterogeneity in the activities of matrix metalloproteinases (MMPs) over tissue inhibitors of MMPs (TIMPs). Abnormal expression of multiple MMPs—such as MMP-1, MMP-2, MMP-3, and MMP-9—has been reported in the skin and blister fluid of patients with RDEB [18-20]. MMP is a family of zinc-dependent endoproteinases capable of degrading ECM components and influencing cellular activities such as cell proliferation, migration, adhesion, and tissue remodeling [21]. MMPs may also in part be a driving factor for cancer progression and metastasis, vascular disease, and inflammatory pathologies [22, 23]. Dermal expression of MMP-13 was identified in the chronic ulcers but not normal healing wounds [24]. However, upon malignant progression in chronic wounds, epithelial localization of MMP-13 has been identified [25]. Moreover, the number of MMP-13 positive cells correlated with the degree of tumor invasion. Therefore, MMP-13 is regarded as a malignant transformation marker in keratinocytes and targeted inhibition of MMP-13 suppressed cSCC growth and invasion [26]. In RDEB-associated cSCCs, MMP-13 and MMP-7 have been identified as tumor cell-specific markers and therapeutic targets [27, 28]. In contrast to MMP-13, MMP-9 was identified in the epithelium of both cSCCs and chronic wounds [25]. Even though MMP-9 expression in keratinocytes is not a specific marker for malignancy, it is correlated with invasion and poor prognosis in cSCCs [29].
We previously demonstrated therapeutic activities of human cord blood-derived unrestricted somatic stem cells (USSCs) in mouse models of wounding and RDEB (col7a1−/−) [30, 31]. USSCs are likely to be the precursors of MSCs, based on USSCs' higher expansion capacity, broader differentiation ability, hypomethylation on pluripotency genes, and differential expression of genes including δ-like 1/preadipocyte factor 1 (DLK1) and the HOX gene clusters [32-34]. Excisional wounds in NOD/SCID mice healed significantly faster and the healed skin was less fibrotic after treatment with USSCs [30]. We also demonstrated that following intrahepatic (i.h.) administration in neonatal col7a1−/− mice, USSCs disseminated into the circulation and migrated to skin and intestines [31]. Specifically, USSCs were identified in the dermis within a week and in the hair follicles within 2 weeks of administration. The median life span of recipient col7a1−/− mice was significantly prolonged and although USSCs were no longer detected in the tissue after a month of administration, C7 was identified at the dermal–epidermal junction for more than 3 months. The study suggested a potential therapeutic efficacy of systemic USSC administration in patients with RDEB. We also speculate that intradermal (i.d.) administration of USSCs in addition to systemic treatment is feasible and may furthermore promote wound healing in patients with RDEB.
The goal of this study was to use col7a1−/− mice to investigate the dynamics of TGFβ signaling and the mechanisms of action of USSCs after systemic and combined systemic and local administration. We furthermore determined the effects of USSC transwell coculture on the production of inflammatory cytokines and MMPs from the fibroblasts, keratinocytes, and cSCCs derived from patients with RDEB. The results suggested dual activities of USSCs in mitigating fibrosis and potentially preventing the malignant transformation of keratinocytes and the development of cSCC in patients with RDEB.
Materials and Methods
RDEB Patient-Derived Cells and Coculture with USSCs
Human primary fibroblasts and keratinocytes were established from biopsies taken from nonlesional, noninflamed skin from two individuals with RDEB, after informed, written consent and in accordance with the ethical standards of the UK National Research Ethics Committee. Patient 1 is a 22-year-old Greek male with generalized severe RDEB who is homozygous for the mutation c425G>A in exon 3 of COL7A. Although this mutation results in the amino acid substitution, p.Lys142Arg, it occurs within a split codon (exons 3/4) and its major consequence is perturbed splicing with generation of out-of-frame transcripts [35]. This is a recurrent loss-of-function mutant COL7A1 allele noted across much of Europe [36]. Patient 2 is a 40-year-old British male with generalized intermediate RDEB with some inversa features [37]. He is a compound heterozygote for an acceptor splice site mutation, IVS30-1G>A, and a nonglycine missense substitution, c.6205C>T (exon 74; p.Arg2069Cys). RDEB-specific cSCC, that is, EB3K was kindly provided by Edel O'Toole, MB, PhD, FRCP at Blizard Institute, UK [13]. Human fibroblasts were cultured in high-glucose DMEM medium containing 10% FBS (Life Technologies, Carlsbad, CA) and penicillin and streptomycin (Corning, Corning, NY) and keratinocytes/cSCCs were cultured in EpiLife Medium (Gibco, Grand Island, NY).
For the coculture of USSCs with human cells, 2 × 104 human fibroblasts and keratinocytes/cSCCs were first seeded in each well of the six-well culture plate (Corning) for 24 hours in their respective culture medium. After washing and replacing with basal medium (for fibroblasts) and fresh EpiLife (for keratinocytes and cSCCs), 0.2 × 106 USSCs were added to the transwell inserts and cocultured for 18 hours before the addition of 5 ng/ml TGFβ1(BD Bioscience, Bedford, MA). Cell lysates were prepared 1, 3, 6, and 48 hours after addition of TGFβ1, for Western blotting analysis. The 48-hour medium of fibroblasts cultured in basal condition and keratinocytes/cSCCs in EpiLife, with or without USSC coculture were also collected, for the MMP and cytokine array analyses. Triplicate samples were set up for each experimental condition.
Antibody and Cytokines Array Analyses
Sera from WT and col7a1−/− mice, with or without USSC administration, were diluted 1:3 in sample diluent and added to the Quantibody Mouse Cytokine Array 1 (RayBiotech, Norcross, GA), according to manufacturer's recommendation. Basal medium recovered from transwells of human fibroblasts and keratinocytes/cSCCs cultured in the presence and absence of USSCs were assayed to quantitate the amount of MMPs and inflammatory cytokines, with the Human MMP antibody array (Abcam, Cambridge, MA) and Quantibody Human Inflammation array (RayBiotech), respectively, according to manufacturer's protocols. Slides from the arrays were read with a GenePix 4000B laser scanner to determine pixel intensity of quadruplicate spots which were averaged and compared to a standard curve to estimate cytokine concentrations.
Statistical Analysis
All numerical data were presented as mean ± standard deviation. Statistical evaluation was performed using an unpaired student's t test analysis. The values were considered significant when the probability (p) was <.05.
Additional methods can be found in Supporting Information.
Results
Activation of TGFβ Signaling via Phosphorylation of Smad2/3 in col7a1−/− Mice and Modulation by USSC Administration
We recently showed that repeated injection of dextran/human serum albumin (D/HSA) buffer were able to extend the survival of some col7a1−/− mice, most likely by adjusting fluid balance, although it had no significant effect on the median life span [38]. The mice that survived enabled the current investigation on the dynamic development of TGFβ signaling in this animal model. Here, we demonstrated that phosphorylated (p)-Smad2/3 was absent in the paw skin at birth and started to appear in a week, most prominently in the dermis between the digits (Fig. 1A, shown by arrows). In 2 weeks, significantly more pSmad2/3-positive cells progressed from the interdigital folds to the overall skin and were evenly distributed in the dermis in 4 weeks (Fig. 1A, 1B). Interestingly, unlike the nuclear localization of the pSmad2/3 that is typically observed in the cultured fibroblasts upon stimulation with TGFβ1 or in the fibroblasts and keratinocytes of acutely wounded skin, the pSmad2/3 in col7a1−/− mouse skin was localized around nuclei in the cytoplasm (Supporting Information, Fig. S1A, S1B). Moreover, most of the pSmad2/3-positive cells were marked by the intermediate filament protein vimentin, suggestive of the identity as fibroblasts (Fig. 1C). In addition, some of the pSmad2/3-positive cells also overlapped with the anti-CD68 staining, which marks resident macrophages (Fig. 1C). As phosphorylation of p38MAPK and STAT3 was not observed in any of these skin biopsies (Supporting Information, Fig. S1C), these preliminary data suggest that both fibroblasts and immune cells in the skin contribute to TGFβ signaling, via phosphorylation of Smad2/3, in col7a1−/− mice.
TGFβ signaling is activated over time in col7a1−/− mice and USSC intrahepatic (i.h.) administration suppressed pSmad2/3-mediated TGFβ signaling. (A): pSmad2/3 immunostaining (green) in the paws of the WT and age-matched col7a1−/− mice that received either D/HSA or USSC i.h. administration at birth. Nuclei were counter stained by DAPI. Images were acquired with a 4× objective. The inset at the interdigital fold of 1w-D/HSA-injected col7a1−/− mice was acquired with a 20× objective in Supporting Information, Fig. S1A. Scale bar: 200 μm. (B): Quantitation on the mean number of pSmad2/3-positive cells per section (≥6 sections per biopsy) under each condition. Two to four mice were analyzed for each condition, except that only one col7a1−/− mouse survived for 4 weeks after D/HSA injection due to limited survival of the knockout mice. *, p ≤ .05: 1w RDEB D/HSA versus 2w and 4w RDEB D/HSA; age-matched RDEB D/HSA versus RDEB USSCs. (C): Costaining of pSmad2/3 with antibody against Vimentin (red, left) and CD68 (green, right), respectively, in 2w-D/HSA-injected col7a1−/− mice. Scale bar: 20 μm.
In contrast to the escalation of pSmad2/3 in the D/HSA-injected mice, significantly fewer pSmad2/3-postive cells were identified in age-matched RDEB mice that received a USSC i.h. administration at birth (Fig. 1). Moreover, these pSmad2/3-positive cells were deeper in the dermis and significantly weaker in the intensity of immunostaining (Fig. 1, shown by arrow heads).
In accordance with the activation of TGFβ signaling, the w4-D/HSA RDEB skin exhibited extensive collagen fibrils and activated (myo)fibroblasts, demonstrated by immunostaining of Collagen I (Col I) and αSMA, respectively (Fig. 2). MMP-9 and MMP-13 were also strongly positive by immunostaining in the dermis (Fig. 2). Importantly, the level of all these aberrantly elevated molecules was significantly less in the age-matched USSC-treated RDEB skin, except for MMP-9, which showed the trend of diminishment, but was not significantly ameliorated by USSC administration (Fig. 2). Moreover, TGFβ3 that was barely detectable in the D/HSA-RDEB skin was significantly upregulated after USSC administration (Fig. 2). These preliminary data suggest in part that USSC administration suppress TGFβ signaling induced fibrosis.
USSC administration suppressed fibrotic markers and increased anti-fibrotic TGFβ3 in the w4-paw skin of col7a1−/− mice. Immunohistochemical analyses with antibodies against Col I, αSMA, MMP-9, MMP-13, and TGFβ3 (green) were performed on paw skin from w4 D/HSA (n = 1) and USSC (n = 2) administered RDEB mice. Nuclei were counter stained by DAPI. The bar graphs on the right indicate quantification of the mean immunostaining intensity per field acquired with a 20× objective. More than or equal to six fields were acquired per section and at least four sections were analyzed per biopsy. * and *** denote p ≤ .05 and .001, respectively. Scale bar: 50 μm.
We have to mention, however, that pSmad2/3 indeed occurred in the RDEB skin 14 weeks after USSC administration, when USSCs were no longer detected and the mice developed pseudosyndactyly (Supporting Information, Fig. S1D). Therefore, it appeared that the TGFβ suppressive effect was contingent on the presence of USSCs and/or USSC-secreted factors.
To validate this speculation, we next determined whether an additional intradermal (i.d.) administration would further enhance the suppressive effect of USSCs on TGFβ signaling and we focused on the fibrosis-related molecules in the skin of RDEB mice at the time that TGFβ signaling starts to be activated, that is, 1 week after birth. i.d. administration was performed in the dorsal skin in the newborn RDEB mice that also received USSC i.h. administration. RT-PCR and qPCR analyses on the skin lysates revealed that the expression of MMP-9 (p < .05), MMP-13 (p < .05), and IL-8 (p < .01) was significantly lower in the skin of the RDEB mice that received i.h. USSCs, compared to the mock control (Fig. 3A, 3B). Importantly, the level of MMP-9 and MMP-13 mRNA was further significantly decreased in the RDEB skin with i.h./i.d. USSCs (p < .01). In the USSC-i.h./i.d. skin, we noted a modest (1.69 ± 0.14-fold, p < .05) increase in TGFβ1 mRNA and more than sixfold (p < .01) increase on the expression of the receptor TGFβRII, which recruits and phosphorylates TGFβ receptor I to initiate TGFβ signaling. However, ELISA analysis demonstrated that the active form of TGFβ1 in the lysate of the USSC i.h./i.d.-skin was only 61.9% (p < .05) of the level in the mock injected RDEB skin, suggesting that active TGFβ1 production was indeed significantly suppressed after USSC administration (Fig. 3C). Remarkably, the expression of TGFβ3 that competes with TGFβ1 in binding to TGFβRII was more than 20-fold higher in the USSC- i.h./i.d. skin (p < .001) (Fig. 3A). As the primers used for PCR amplification were designed to be mouse-specific and did not amplify the transcripts from human USSCs (Supporting Information, Fig. S2A), we concluded that the increased TGFβ3 mRNA originated from recipient mice but not human-derived USSCs. In addition to TGFβ3, the expression of DCN, which also antagonizes TGFβ signaling was 2.4-fold higher in the USSC- i.h./i.d. RDEB skin as compared to the mock control (p < .01) (Fig. 3A). Consistent with the noted mRNA changes, immunohistochemical staining demonstrated a distinctive decrease in αSMA, MMP-9, and MMP-13, as well as a remarkable increase in TGFβ3, in the dermis of USSC i.h./i.d. injected RDEB skin (Supporting Information, Fig. S2B). The collagen content as evaluated by picro-sirus red staining was also significantly less in both i.h. and i.h./i.d. USSC injected RDEB skin (Supporting Information, Fig. S2B, S2C).
USSC administration(s) modulated the expression of fibrotic and inflammatory markers in col7a1−/− mice at the time (1 week) that TGFβ signaling starts to be activated. Quantitative PCR (A) and representative RT-PCR (B) analyses on the relative expression of MMP-9, MMP-13, IL-8, TGFβ1, TGFβRII, TGFβ3, and DCN in the skin of 1-week WT, RDEB untreated, RDEB with USSC intrahepatic (i.h.) or combined intrahepatic and intradermal injections (i.h./i.d). Three biological samples with technical triplicates were used for qPCR analysis. (C): ELISA analysis on the level of TGFβ1 in the skin lysate of 1-week WT, RDEB untreated, RDEB USSC i.h. and RDEB USSC i.h./id. (D): Quantitation on inflammatory cytokines in the sera of D0 and 1-week-old WT and RDEB mice with and without USSC i.h. and i.h./i.d administration, respectively, using Quantibody Mouse Cytokine Array 1. *, **, and *** denote p ≤ .05, .01, and .001, respectively.
We also compared the levels of cytokines in the serum of the 1-week-old WT, mock- and USSC-administered RDEB mice using Quantibody Mouse Cytokine Array (Fig. 3D). As compared to the mock control, the level of proinflammatory cytokines, including IL-6, IFN-γ, TNF-α, and IL-17A, was significantly decreased in the RDEB mice that received either i.h. or i.h./i.d USSC administration. In contrast, the level of anti-inflammatory cytokines, including IL-10 and IL-13, was significantly higher in the serum of the RDEB mice that received USSC administration.
Suppression of Fibrotic MMP and Upregulation of Anti-Fibrotic Gene Expression in Fibroblasts Derived from Neonatal col7a1−/− Mice by Coculture with USSCs
To determine whether USSC-secreted factors could directly influence fibrotic or anti-fibrotic gene expression, we isolated the fibroblasts from neonatal col7a1−/− and WT mice and co-cultured with USSCs in a transwell system. Consistent with the results that TGFβ signaling was not activated at newborn stage, fibroblasts derived from neonatal col7a1−/− mice were not different than the WT in the expression of TGFβ1 and fibrotic markers such as αSMA under both growth (in the presence of 10% FBS) and serum-free (basal) conditions (Fig. 4 and data not shown). To further dissect the response of the mouse fibroblasts to TGFβ signaling in the presence of USSCs, the coculture was set up under both basal condition and following TGFβ1 addition. qPCR analyses demonstrated that exogenous TGFβ1 increased the expression of TGFβ1 and MMP-9 and 13 in both WT and RDEB fibroblasts (Fig. 4). While USSC coculture had no significant effect on the expression of TGFβ1, it significantly decreased the expression of MMP-9 in RDEB fibroblasts under both basal and TGFβ1 activated conditions and the expression of MMP-13 in both WT and RDEB fibroblasts under basal condition (Fig. 4). Moreover, under TGFβ-activated conditions, coculture with USSCs significantly upregulated the expression of DCN and TGFβ3, in support of the effects observed above in the RDEB mouse skin.
Transwell coculture with USSCs significantly diminished fibrotic MMP expression and elevated anti-fibrotic gene expression in fibroblasts derived from neonatal col7a1−/− mice. qPCR analysis for the relative expression of TGFβ1, αSMA, MMP-9, MMP-13, TGFβ3, and decorin (DCN) under basal or TGFβ1activated condition, with and without USSC coculture. Experiments were repeated twice with technical triplicates. *, **, and *** denote p ≤ .05, .01, and .001, respectively.
Suppression of MMPs and Inflammatory Cytokines in RDEB Patient-Derived Cells by Coculture with USSCs
We next investigated whether USSC transwell coculture has similar regulatory activities on cells derived from patients with RDEB. Fibroblasts and keratinocytes were derived from uninvolved skin of two adult RDEB patients with generalized RDEB manifestation. The level of C7 in patient-derived fibroblasts and keratinocytes was lower compared to the normal control (Fig. 5A). Moreover, the intracellular distribution of patient C7 appeared to be aggregated, implying that the folding of the protein is likely to be abnormal (Fig. 5A). An RDEB-specific cSCC line, namely, EB3K [13], had no expression of C7 and was also included in this study together with fibroblasts and keratinocytes from healthy donors and patients with RDEB.
RDEB patient-derived cells responded similarly to TGFβ activation and were constitutively active for STAT3 signaling. (a): COL7A1 immunostaining (green) in the normal- and RDEB patients (pt1 and pt2)-derived fibroblasts and keratinocytes and RDEB-specific cSCC, EB3K. Scale bar: 50 μm. (B): Western blot analysis to determine phosphorylated and total Smad-2/3 and STAT3 in the protein lysate of Pt1-fibroblasts and keratinocytes, RDEB SCC, and normal keratinocytes, before and after (1 and 3 hours, respectively) the addition of TGFβ1, with or without USSC coculture in the transwell. The amount of GAPDH was used as the loading control.
The level of TGFβ1 was undetectable in the conditioned medium of fibroblasts and was not significantly different between the normal and the RDEB patient-derived keratinocytes, although there was a modest and significant increase in EB3k cSCCs (Supporting Information, Fig. S3A). In addition, neither Smad2/3 nor p38 was phosphorylated in basal condition (Fig. 5B and data not shown). Upon exogenous addition of TGFβ1, pSmad2/3 was detected in an hour and decreased in 3 hours among all the samples, suggesting that TGFβ signaling may not be aberrantly activated in these patient samples. However, a constitutively active STAT signaling, revealed by phosphorylation of STAT3, was observed in the patient-derived cells in both basal and TGFβ-activated conditions (Fig. 5B). The pattern of phosphorylation was not significantly altered when the cells were preincubated overnight with USSCs, before the addition of TGFβ1.
QPCR analysis demonstrated that the expression of DCN and TGFβ3 was undetectable in the patient-derived cells (Supporting Information, Fig. S3B). The expression of MMP-1 and MMP-3 was significantly elevated in the fibroblasts of RDEB patients. Significantly, although undetectable in the fibroblasts, MMP-13 expression was more than 500- and 50-fold higher in cSCCs and patient keratinocytes respectively than the normal control, consistent with its correlation with epithelial malignant transformation [26]. The expression of inflammatory cytokines, including IL-6 and IL-8, was also more upregulated in patient keratinocytes and cSCCs (Supporting Information, Fig. S3B). Consistently, quantitative measurement of human inflammatory cytokines with a multiplex antibody array showed significantly elevated levels of IL-6 and IL-8 in the medium of patient keratinocytes and cSCCs (Fig. 6A). However, as USSCs constitutively secrete these cytokines, we cannot conclude whether coculture with USSC affected their production from the skin cells. The level of IFN-γ was also elevated in all the patient-derived cells. Importantly, while USSCs also secrete IFN-γ, the level of IFN-γ was significantly less in the USSC-cocultured medium (Fig. 6A). As for the other cytokines included in the multiplex array, IL-1β, IL-10, IL-13, TNF-α, and IL-4 were undetectable in all the samples, despite a higher level of IL-1β mRNA in cSCCs (Supporting Information, Fig. S3B and data not shown). IL-1α was marginally detectable in the medium of Pt1 fibroblasts and was at much higher levels in both normal and patient keratinocytes and cSCCs. In the presence of USSC coculture, the level of IL-1α was significantly decreased in all the samples, except for cSCCs (Fig. 6A). In addition, elevated levels of monocyte chemoattractant protein-1 (MCP-1) were detected in patient fibroblasts as compared to the normal control. However, as USSCs also constitutively secrete MCP-1, we speculate that the higher level of MCP-1 in the coculture medium was due to the secretion from USSCs.
Transwell coculture with USSCs modulated the secretion of inflammatory cytokines and MMPs from RDEB patient-derived cells. Quantitation on the level of (A) inflammatory cytokines and (B) MMPs in the 48-hour medium of RDEB patient-derived cells and normal controls, cultured in low-glucose DMEM (for fibroblasts) and EpiLife (for keratinocytes and cSCCs) medium, respectively, in the presence or absence of USSCs in transwells. Experiments were repeated twice with technical triplicates for each condition. *, **, and *** denote p ≤ .05, .01, and .001, respectively.
We also utilized a human MMP antibody multiplex array to compare the secretion of MMPs and TIMPs from these cells with and without USSC coculture. Consistent with the gene expression profile, MMP-13 was identified in the medium of patient keratinocytes and was much more elevated in cSCCs, but was undetectable in patient fibroblasts, normal controls, and USSCs (Fig. 6B). Importantly, USSC coculture significantly decreased the amount of MMP-13 in the medium of both cell types (p < .01). The suppressive effect of USSCs on MMP-13 production was further validated by the significantly decreased level of mRNA (p < .01) and protein (p < .05) in the lysate of cSCCs cocultured with USSCs (Supporting Information, Fig. S4A). A similar suppression by USSCs was also observed for MMP-10 (Fig. 6B). In addition, although the MMP-9 mRNA level was not upregulated in patient-derived cells based on QRT-PCR analysis, MMP-9 protein was identified in the medium of keratinocytes and cSCCs and was significantly diminished following coculture with USSCs (Fig. 6B). These results suggest USSCs also suppress the production of MMP-9. As for the other MMPs and TIMPs included in the array, the levels of MMP-8 and TIMP-4 were below the assay's sensitivity of detection (data not shown). The remaining targets, that is, MMP-1, MMP-2, MMP-3, TIMP-1, and TIMP-2, were detected in the medium of USSCs alone (Supporting Information, Fig. S4B). All these molecules are secreted proteins except for MMP-2, which could be detected in whole cell lysates and its level was significantly less in the lysate of cSCC after coculture with USSCs, suggesting that USSCs also suppressed the production of MMP-2 (Supporting Information, Fig. S4A). In summary, although the secretion of some of these factors by USSCs limited our interpretation on the effects of USSCs, it is evident that coculture with USSCs had a significant impact on the molecules that are most dysregulated in the RDEB patient-derived cSCC and keratinocytes.
Another factor that was worthy of investigation but was not included in the MMP array was MMP-7, as its mRNA was upregulated in the RDEB patient keratinocytes and cSCCs (Supporting Information, Fig. S3B). However, neither ELISA nor Western blotting analyses detected any positive signal of MMP-7 from the medium or the cell lysate, suggesting no significant generation of MMP-7 protein in the patient samples included in this study (Supporting Information, Fig. S5).
Discussion
Mitten deformities of the hands and feet occur in nearly every patient with severe generalized form of RDEB [39]. Even after surgical intervention, the deformities frequently reoccur. The current investigation on col7a1−/− mice demonstrated that TGFβ signaling via phosphorylation of Smad2/3 occurs as early as a week after birth, starting in the interdigital folds of the paws. This suggests that intrinsic molecular abnormalities at a very early stage in the dermal microenvironment may have predisposed the development of mitten deformity in patients with RDEB. Paradoxically, elevated deposition of collagen fibers and dermal expression of MMP-9 and MMP-13 co-exist in the col7a1−/− mouse skin. As the dermal expression of MMP-13 is associated with chronic erosions but not acute normal healing wounds [24], these preliminary data suggest in part a chronic wounding environment in this RDEB animal model in the absence of external lesions. This may explain the difference in the subcellular localization of pSmad2/3 between the col7a1−/− mouse skin and acutely wounded skin. Indeed, cytoplasmic localization of pSmad2/3 was observed in subtypes of human trophoblast, as well as healthy human fetal skin [40, 41]. The implication of the cytoplasmic localization of pSmad2/3 in the pathological development of RDEB skin requires further investigation. Nevertheless, the data are consistent with an altered wound healing in RDEB [42] and meanwhile suggests that the RDEB skin may adopt a varied canonic TGFβ signaling, in response to the inherent separation of the epidermis and dermis.
Nyström et al. previously demonstrated that excisional wounds in the adult C7 hypomorphic mice and human RDEB wounds display similar molecular alterations to human chronic wounds, with regard to the laminin-332/integrin α6β4 signaling axis that leads to suprabasal JNK and STAT3 activation [43]. The pSTAT3 was not observed in this study and we postulate that it may be activated at a later stage or upon an external wounding. However, such investigation is not feasible in this animal model due to poor survival (median life span of 2 days). In this study, the level of TGFβ1 was not significantly elevated in the RDEB patient-derived cells, except for the cSCCs. As only one cSCC line and two RDEB patient-derived cells were included in this study, the data may reflect the heterogeneity in the mechanisms underlying the pathological development in patients with RDEB, as reported before [14]. Importantly, it should be noted that the fibroblasts and keratinocytes were derived from nonlesional and noninflamed skin. Based on the animal studies, immune cells also contribute to TGFβ signaling. Alternatively, as all the patient-derived cells were constitutively active for STAT3 signaling, in accordance with high level of IL6, the data may imply that the patient skin may have undergone TGFβ-mediated ECM remodeling and progressed into a proinvasive tumor microenvironment. Consistently, as will be discussed in more detail below, more significant abnormalities were observed in the patient-derived keratinocytes than the fibroblasts, particularly in the expression of MMPs that influence tumor invasion.
In this study, administration of USSCs effectively suppressed TGFβ signaling in neonatal col7a1−/− mice. However, USSCs were unable to prevent the acute activation of TGFβ signaling or inhibit the constitutively active STAT signaling in the patient cells using the transwell-coculture system. Future experiments under more physiological conditions, for example, 3D skin equivalents and with more patients-derived biological samples, will be required to validate the effects of USSCs on TGFβ and STAT signaling in the RDEB patients-derived cells.
One effect that USSCs acted similarly in the mouse and human systems is suppression of MMP-9 and MMP-13 production in the dermis (and fibroblasts) of the col7a1−/− mice and from the keratinocytes and cSCCs of the patients with RDEB, respectively. It has been recognized that the functions of MMPs are not limited to degrading ECM and a precise spatiotemporal regulation of MMPs is essential to maintain a proper homeostasis of extracellular and pericellular environment [23, 44]. Recent studies indicated a bidirectional regulatory loop between TGFβ1 and MMPs. TGFβ1 can induce or activate the transcription of both MMP-9 and MMP-13 molecules through different mediators [45-47]. Thus, with an escalation of TGFβ signaling in the dermis of col7a1−/− mice, an increase in the level of MMP-9 and MMP-13 was observed and following the inhibition of pSMAD2/3 by USSC administration, the levels of MMP-9 and MMP-13 were significantly decreased. Reciprocally, TGFβ1 is normally maintained in an inactive state by a noncovalent association with a latency associated peptide (LAP). Its activation involves proteolytic cleavage of the LAP by soluble MMP-9 or MMP-9 bound to CD44 in the cell surface [48, 49]. Consistently, the amount of active TGFβ1 in the skin of RDEB mice after USSC i.h./i.d. administration was indeed decreased based on ELISA analysis, even though the TGFβ1 mRNA was slightly increased. In this study, there was no significant difference in the levels of major TIMPs between the normal and RDEB patients-derived cells. However, both MMP-13 and MMP-9 were significantly upregulated in not only RDEB-specific cSCCs, but unexpectedly also in RDEB patient-derived keratinocytes. As epithelial expression of MMP-13 marks malignant transformation and MMP-9 is correlated with poor prognosis in SCCs, the data imply that the epithelium of the RDEB patient may have started certain features of malignant transformation. The suppression of MMP-9 and MMP-13 production in RDEB keratinocytes and cSCCs by USSC coculture is thus exciting and suggest that USSCs may play a role in modulating RDEB keratinocyte migration, malignant transformation, and invasion.
Another effect that was common in the mouse and human systems is attenuation on IFN-γ secretion in the col7a1−/− mice after administration with USSCs and from RDEB patient-derived cells after coculture with USSCs. Annicchiarico et al. showed that the level of IFN-γ was significantly higher in the sera of the patients with RDEB than that of healthy controls or patients with EB simplex form [17]. Moreover, the level of IFN-γ was positively correlated with the titers of autoantibodies, including anti-C7. Blockade of IFN-γ has been proposed as a possible pharmacological approach to modulate the adaptive immune response in patients with RDEB [50]. Therefore, modulation of IFN-γ by USSCs represents another therapeutic merit of USSCs in the treatment of RDEB.
This study also suggests in part that USSCs antagonize fibrosis by upregulating the dermal (and fibroblast) expression of two anti-fibrotic molecules, that is, DCN and TGFβ3. The anti-fibrotic role of DCN in RDEB was implicated in the study on the monozygotic twins [8]. Jἂrvinen et al. recently demonstrated that engineered DCN-targeting peptide fusion protein in vitro neutralized activities of TGFβ1 but not TGFβ3 and after systemic injection in vivo, selectively accumulated in wounds and promoted a scar-less wound healing [51]. DCN can also bind to and downregulate receptors such as EGFR, IGF-IR, and VEGFR2 that are often overexpressed in cancer cells [52]. On the other hand, the expression of DCN can be downregulated by TGFβ1 in an additive manner with TNF-α [53]. Such a negative feedback loop was indeed revealed in the current coculture study, which demonstrated that the expression of DCN in the mouse fibroblasts was decreased after addition of TGFβ1. As USSC coculture only significantly influenced the expression of DCN in TGFβ1-activated condition, it is likely that USSCs restored its expression by relieving the suppressive effect of TGFβ signaling. As for TGFβ3, although the mechanism for its increased expression by USSCs remains to be determined, the outcome after USSC administration appears to be associated with a scar-free wound healing microenvironment.
The results from transwell coculture system suggest in part that USSCs can suppress MMP production and stimulate anti-fibrotic DCN and TGFβ3 expression via a paracrine effect. Secretome analysis has demonstrated that USSCs secrete trophic factors involved in a spectrum of biological processes such as cell adhesion, cytoskeleton organization, and extracellular matrix organization [54]. USSCs have also been shown to constitutively produce different cytokines such as stem cell factors (SCF), vascular endothelial growth factor (VEGF), stromal cell derived factor (SDF)-1α, hepatocyte growth factor (HGF), and so forth [55]. When USSCs were stimulated with IL-1β, granulocyte (G)-CSF was released [55]. G-CSF has been associated with improved wound healing in patients with toxic epidermal necrolysis and in radiation-induced moist desquamation [56, 57]. It also enhanced wound healing in a pilot trial in patients with DEB [58]. However, which secreted factors from USSCs played a role remains to be determined.
The enhanced effects after an additional i.d. injection of USSCs in the skin of col7a1−/− mice supports the rationale for potential combined systemic and local administration of USSCs in patients with RDEB. As USSCs persisted only short term in the recipient mice, future studies will be designed to investigate whether continuous administration of USSCs will lead to long-term suppression of TGFβ signaling, thus prevention of the development of mitten deformity.
Conclusion
USSCs have the ability to modulate the fibrotic and inflammatory microenvironment of the RDEB skin. USSC administration in col7a−/− mice effectively suppressed TGFβ signaling, proinflammatory cytokine expression, and dermal MMP-9 and MMP-13 production, meanwhile upregulating the level of anti-fibrotic DCN and TGFβ3. USSC coculture also significantly suppressed the production of MMP-9 and MMP-13 from RDEB patient-derived keratinocytes and cSCCs. These preliminary data suggest in part dual roles of USSCs in preventing fibrosis and potentially modulating keratinocyte malignant transformation and cSCC invasion.
Acknowledgments
We are grateful to Dr. Jouni Uitto at Jefferson Medical College for the col7a1+/− mice. We also thank Edel O'Toole, MB, PhD, FRCP at Blizard Institute, UK for providing EB3K. We acknowledge the scientific discussion from colleagues at the Pediatric Cancer Research Laboratory at NYMC. We also thank Erin Morris, RN, Janet Ayello, MS, ASCP, and Miguel Muniz, AAS for their assistance in the preparation of this manuscript. This work was supported by grants from Pediatric Cancer Research Foundation to M.S.C., DEBRA International funding to M.S.C., and NYMC/Touro SEED funding program to Y.L.
Author Contributions
Y.L.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; L.I. and H.Z.: collection and assembly of data, data analysis and interpretation, manuscript writing; T.P., B.M., and A.G.: collection and assembly of data. C.H. and A.M.C.: data analysis and interpretation; J.A.M.: provision of study material or patients, and data analysis and interpretation; M.S.C.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of Potential Conflicts of Interest
A.M.C. disclosed consultant and ownership interest with Aclaris Therapeutics Dermira, Inc. and research funding with Pfizer. The other authors indicated no potential conflicts of interest.
