ABCB5+ dermal mesenchymal stromal cells with favorable skin homing and local immunomodulation for Recessive Dystrophic Epidermolysis Bullosa treatment

Abstract

Recessive dystrophic epidermolysis bullosa (RDEB) is a rare, incurable blistering skin disease caused by biallelic mutations in type VII collagen (C7). Advancements in treatment of RDEB have come from harnessing the immunomodulatory potential of mesenchymal stem cells (MSCs). While human bone-marrow derived MSC (BM-MSC) trials in RDEB demonstrate improvement in clinical severity, the mechanisms of MSC migration to and persistence in injured skin and their contributions to wound healing are not completely understood. A unique subset of MSCs expressing ATP-binding cassette subfamily member 5 (ABCB5) resides in the reticular dermis and exhibits similar immunomodulatory characteristics to BM-MSCs. Our work aimed to test the hypothesis that skin-derived ABCB5+ dermal MSCs (DSCs) possess superior skin homing ability compared to BM-MSCs in immunodeficient NOD-scid IL2rgamma^null (NSG) mice. Compared to BM-MSCs, peripherally injected ABCB5+ DSCs demonstrated superior homing and engraftment of wounds. Further, ABCB5+ DSCs versus BM-MSCs co-cultured with macrophages induced less anti-inflammatory interleukin-1 receptor antagonist (IL-1RA) production. RNA sequencing of ABCB5+ DSCs compared to BM-MSCs showed unique expression of major histocompatibility complex class II and Homeobox (Hox) genes, specifically HOXA3. Critical to inducing migration of endothelial and epithelial cells for wound repair, increased expression of HOXA3 may explain superior skin homing properties of ABCB5+ DSCs. Further discernment of the immunomodulatory mechanisms amongst MSC populations could have broader regenerative medicine implications beyond RDEB treatment.

GRAPHICAL ABSTRACT

Schema of ABCB5+ dermal MSCs (DSCs) characteristics compared to BM-MSCs. Potential mechanisms of superior skin homing of ABCB5+ DSCs for use in RDEB and regenerative medicine applications. Similarities in immunomodulatory capacity of ABCB5+ DSCs and BM-MSCs are shown in the middle column including anti-inflammatory M1>M2 macrophage skewing.

INTRODUCTION

The skin is the body’s first line of defense against the outside world and is constantly regenerating. This barrier is disrupted in recessive dystrophic epidermolysis bullosa (RDEB), a devastating, blistering skin disease resulting from biallelic mutations in the COL7A1 gene encoding type VII collagen (C7)¹. C7 is a key protein in the basement membrane zone that is essential for skin integrity. It plays a critical role in wound healing by promoting keratinocyte re-epithelization through interaction with laminin-332 in hemidesmosomes, supporting dermal fibroblast migration, and regulating cytokine production². Without adequate expression of C7, patients suffer from painful blisters of the skin, esophagus, and lower gastrointestinal tract¹. No cure exists for RDEB, but several clinical trials investigating the use of bone marrow transplant (BMT)^3,4, skin grafting⁵, and intradermal injections of MSC have shown promise^6,7. Although the mechanisms of clinical improvement are still being investigated, it has been hypothesized that donor cells can home to the affected tissue, engraft, and deposit functional C7.

C7 production is not limited to keratinocytes, as MSCs and fibroblasts secrete C7 upon homing to wounded areas⁸. Several studies have investigated BM-MSCs and stromal cells as cellular therapies for RDEB and demonstrate local, transient, wound healing improvement^9–13. Intradermal injections of BM-MSCs have been shown to improve skin integrity and facilitate wound healing in an RDEB mouse model via de novo formation of immature anchoring fibrils lasting for up to 12 weeks¹⁴. Potential mechanisms driving improvement in clinical phenotype include BM-MSC-derived extracellular vesicles delivering C7 to wounded tissue¹⁵ and MSC immunosuppressive functions¹⁶. BM-MSCs can indirectly downregulate inflammatory mediators by acting upon tissue-resident macrophages, skewing them from pro-inflammatory (M1) to anti-inflammatory macrophages (M2), thereby promoting wound healing¹⁷.

A subset of dermal stromal cells expressing ATP-binding cassette subfamily member 5 (ABCB5) exhibit MSC-like properties, skewing macrophages toward a wound-healing phenotype both in vitro and in vivo^16,18. They exhibit plastic adherence in culture, can be differentiated into three mesenchymal lineages, express high levels of MSC markers, and lack other skin-cell specific markers^16,18. This provides evidence that ABCB5+ dermal cells are a distinct cell population and meet defining criteria for MSC nomenclature¹⁹. Henceforth, they will be termed ABCB5+ dermal MSCs (DSCs).

ABCB5+ DSCs have shown promise in the field of regenerative medicine, which makes them of interest in treatment for RDEB^20–22. Intradermal injections of ABCB5+ DSCs promoted healing in a chronic venous ulcer murine model, with an increase in M2 macrophages in the wound bed¹⁶. Our group has shown that neonatal injections of ABCB5+ DSCs in Col7a1^−/− RDEB murine model significantly prolonged survival compared to phosphate buffered saline (PBS) control injected animals, with ABCB5+ DSC-treated mice showing decreased M1 macrophage skin infiltration²³. These preclinical studies have supported the development of early phase clinical trials of ABCB5+ DSCs, including local therapy for chronic venous ulcers and systemic administration for treatment of RDEB (NCT03529877; EudraCT 2018-001009-98). Results from the first human data, which included using local administration of ABCB5+ DSCs in 12 wounds from 9 patients with chronic venous ulcers showed safety, tolerability, median wound size reduction of 63% and perceived decrease in pain²⁰.

While BMT and administration of MSCs have shown promise in dampening aberrant immune responses and improving clinical response in RDEB, the mechanism(s) of stromal cell migration to injured skin, persistence, hastened wound healing, and C7 deposition, is not completely understood. Therefore, we further investigated human ABCB5+ DSCs in the treatment of RDEB in a murine xenograft model with focus on the skin homing ability and genetic differences of human ABCB5+ DSCs compared to BM-MSCs.

MATERIALS AND METHODS

Animal experiments:

NOD-scid IL2rγ^null (NSG) mice (n=3 per group) were anesthetized and given full-thickness dorsal wounds and injected in the facial vein with 4x10⁵ human cells: BM-MSCs, ABCB5+ DSCs, donor-matched ABCB5− cells, or PBS vehicle control on day 0. Wounded skin and lung tissues were taken at days 1, 7 and 14. For competitive polymerase chain reaction (PCR), 200 ng of wound DNA was examined²⁴. Percent of human DNA was back-calculated using SYBR-green (Thermo Fischer) experimental counts and a standard curve (Figure S1).

Cell Source:

Human ABCB5+ DSCs (>95% ABCB5+ expression) and donor-matched ABCB5− cell fractions (<2% ABCB5+ expression) were isolated by TICEBA. ABCB5+ DSCs were enriched and isolated from reticular dermis primary cell cultures of healthy skin donors as previously described^21,22, while the remaining negative cell fraction (2-3-fold negative selection) was used as the ABCB5− control confirmed by flow cytometry and immunofluorescence staining. BM-MSCs were isolated from fresh bone marrow aspirate and expanded per standard practice²⁵.

RNA sequencing:

RNA was isolated using the RNeasy Plus kit (Qiagen). Sequencing libraries were prepared using the Illumina TruSeq Stranded mRNA kit (20020594; Illumina). Sequencing was performed on the MiniSeq Sequencing System (Illumina) using 150 cycle kits (FC-420-1002; Illumina). RNA-seq reads were processed using a pipeline in Galaxy and differential gene expression was analyzed using edgeR package in R software 3.6.0. Briefly, lowly expressed genes were removed, trimmed mean of M-values (TMM) normalization performed, dispersion estimated using the quantile-adjusted conditional maximum likelihood method (qCML) method, differential gene expression analyses using exact test function and multiple testing corrected using Benjamini-Hochberg (BH) method with medium stringency.

Western blots:

ABCB5+ DSCs, fibroblasts, keratinocytes and BM-MSCs were either lysed on ice for 10 minutes in RIPA Lysis Buffer for whole cell lysate (Santa Cruz Biotechnology) or cell lysis solution for extracellular matrix (ECM) protein extraction (5ml 1M NH4OH + 1.25ml Triton-100 + H₂O to 250 mLs) at room temperature for one minute. ECM proteins were then dissolved in 4% SDS in 0.1M Tris·HCl. Equal quantities of protein (40 μg) were loaded onto a Novex WedgeWell 4–12% Tris-Glycine Mini Gels (Thermo Fisher). Following electrophoresis, protein was transferred onto a nitrocellulose membrane and incubated with anti-ABCB5 antibody (bs-1604R; Bioss), anti-C7antibody (sc-20774; Santa Cruz or gift from Dr. Woodley), anti-laminin alpha-3 (ab242197; Abcam), anti-laminin 332 (ab78286, Abcam), anti-laminin beta-3 (OTI3A2; Bio-rad) or anti-beta-actin antibody (A2547; Millipore Sigma) at 4°C overnight. Membranes were then incubated with goat anti-rabbit HRP-conjugated secondary antibody (sc-2004; Santa Cruz Biotechnology) or anti-mouse HRP-conjugated secondary antibody (sc-2005; Santa Cruz Biotechnology). Blots were developed using SuperSignal West Femto Chemiluminescent Substrate (Thermo Fisher) for C7, laminin alpha-3, laminin 332 and laminin beta-3 or using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher) for beta-actin and developed on X-ray film.

Co-culture experiments:

THP-1 monocyte cells (tib-202tm; ATCC) were cultured in RPMI-1640 medium (12633012; Thermo Fischer) with 0.05 mM 2-mercaptoethanol and 10% fetal bovine serum (FBS). THP-1 cells were stimulated with phorbol 12-myristate 13-acetate (PMA) for 24 hours to induce macrophage phenotype as assessed by adherence to tissue culture flask and expression of CD11b integrin (Figure S2). For co-culture, we followed an established protocol^16,22. Briefly, ABCB5+ DSCs or donor-matched ABCB5− fractions were plated at 2x10⁴ cells per well in 24-well plates in 0.5 ml Dulbecco’s modified Eagle’s medium (DMEM) with 10% FBS, 100 U/ml penicillin/streptomycin and 2 mM L-glutamine. After 24 hours, THP-1 derived macrophages were seeded on top or in transwell inserts (CLS3464; Corning) at 1×10⁵ cells per well. Co-cultures were incubated with 50 U/ml recombinant human interferon-gamma (IFN-γ) (285-IF-100; R&D Systems) for 24 hours and then stimulated with 20 ng/ml LPS (L3755; Sigma-Aldrich) and 50 U/ml IFN-γ for another 24-hour period before supernatants were harvested and analyzed by enzyme linked immunosorbent assay (ELISA) for IL-1RA (DRA00B; R&D Systems). IL-1RA levels were analyzed between conditions with unpaired t-tests.

RESULTS

Superior skin engraftment potential of ABCB5+ DSCs

To compare skin homing properties of ABCB5+ DSCs and BM-MSCs, we used competitive PCR to determine the relative abundance of systemically injected human cells (ABCB5+ DSCs, donor-matched ABCB5− cells, BM-MSCs or PBS vehicle control) in full-thickness dorsal murine wounds on days 1, 7 and 14 after administration. We found superior engraftment of ABCB5+ DSCs compared to both BM-MSCs and ABCB5− cells 14 days after wounding (0.85%, 0.32% and 0%, respectively of human DNA in wound, p < 0.001) (Figure 1A, B). Human DNA was found in NSG mouse lungs injected with cells on day 1, while absent on day 14, most likely due to first pass effect (Figure S3). All mice survived to all analysis timepoints. To identify potential molecular mechanisms of cell homing and persistence in the wounds, we performed RNA sequencing on ABCB5+ DSCs and BM-MSCs. The analyses revealed increased expression of 17 homeobox (Hox) genes and various ECM genes in ABCB5+ DSCs. The ten top differentially expressed genes in ABCB5+ DSCs compared to BM-MSCs are shown in Figure 1C. ABCB5+ DSCs also express multiple MHC-II genes, with only HLA-DPB1 significantly upregulated (p = 4.98e-7) (Figure 1D).

Figure 1:

Skin engraftment quantified by competitive PCR in NSG wounds. Human ABCB5+ DSCs home to wounded mouse skin to a higher extent than BM-MSCs and ABCB5− fractions. A) Results from SYBR-green competitive PCR show percent of human DNA in mouse wound B) PCR products from competitive mouse vs. human DNA PCR. **p < 0.001. C) The top ten upregulated genes in ABCB5+ dermal MSCs compared to BM-MSCs. D) Differential expression of HLA-genes MHC class II in ABCB5+ DSCs compared to BM-MSCs. FC = Fold change, CPM = Counts Per Million. FDR = False Discovery Rate.

ABCB5+ DSCs express skin basement membrane proteins C7 and laminin-322

Basement membrane proteins C7 and heterotrimeric laminin-332 (comprised of laminin-α3,-β3, and -γ2), mutated in RDEB and various forms of junctional epidermolysis bullosa (JEB), respectively, are critical for cutaneous wound healing. After confirming ABCB5 expression in ABCB5+ DSCs (n=3 human donors; Figure 2A), we assessed baseline C7 and laminin protein production showing expression of laminin 332 (α3 and β3 subunits) in ABCB5+ DSCs, while BM-MSC expression of C7 and laminin-α3 and β3 were absent (Figure 2). These findings suggest that the ability to secrete skin basement membrane proteins C7 and laminin 332 subunits might contribute to superior therapeutic potential of ABCB5+ DSCs.

Figure 2:

Expression of skin proteins in BM-MSCs, ABCB5+ DSCs and ABCB5− fibroblasts. A) Western blot characterizing expression of Collagen VII and ABCB5 in ABCB5+ DSCs and BM-MSCs. BM-MSCs have varying levels of ABCB5 expression and do not express C7 at baseline. B) Expression of skin proteins in keratinocytes, fibroblasts and ABCB5+ dermal cells. ABCB5+ cells express C7 and laminin-332 (also known as laminin 5) β3 subunit proteins. C) Extracellular matrix (ECM) proteins isolated from BM-MSCs, ABCB5− fibroblasts and ABCB5+ DSCs shows expression of laminin α3 and β3 subunits in ABCB5− fibroblasts and ABCB5+ DSCs, and lack of expression in BM-MSCs.

Immunosuppressive ability of MSCs when co-cultured with macrophages measured by IL-1RA secretion

Wound healing is influenced by tissue-infiltrating or resident immune cells, such as macrophages. Previous studies have shown that THP-1 induced macrophages can secrete anti-inflammatory IL-1RA when cultured with MSCs (citation). In the current study we examined IL-1RA secretion by ABCB5+ DSCs, donor-matched ABCB5− fibroblasts or BM-MSCs co-cultured with THP-1 activated macrophages either directly or separated by transwell inserts (Figure 3). IL-1RA was detected when macrophage-derived cells were seeded with the ABCB5+ DSCs or BM-MSCs in both co-culture or transwell conditions (Figure 3). IL-1RA was not detected when ABCB5+ DSCs, MSCs or ABCB5− fibroblasts were cultured alone or when ABCB5− donor-matched fractions were co-cultured with macrophages. BM-MSCs showed increased anti-inflammatory M2-macrophage skewing ability in both direct and transwell co-cultures compared to ABCB5+ dermal MSCs (p = 0.003 and p = 0.044, respectively).

Figure 3:

ABCB5+ DSCs can skew macrophages towards M2 phenotype to a greater extent when culture in the same well vs. transwell. Interleukin- 1 receptor antagonist (IL-1RA). *p < 0.05. ABCB5 = ABCB5+ DSCS, MSC = BM-MSCs, MAC = THP-1 derived macrophages.

DISCUSSION:

We have demonstrated that using skin-derived ABCB5+ DSCs over BM-MSCs is advantageous for enhanced localization in wounded skin, potentially driven by their derivation from the skin and hox genes. Human DNA from ABCB5+ DSCs and BM-MSCs was not detected until 14 days after injection. We hypothesize this time course reflects processes of cell migration and persistence.

Interestingly, 17 hox transcription factors were significantly upregulated in ABCB5+ DSCs relative to BM-MSCs. Hox genes are leading candidates in regenerative medicine and wound healing due to their critical role in morphogenesis. Specific hox genes are anatomically restricted where they play an essential role in tissue remodeling and angiogenesis. HOXA3, which was significantly upregulated in ABCB5+ DSCs compared to BM-MSCs, has been shown to be a master coordinator of wound repair by inducing migration of endothelial and epithelial cells^26,27. ABCB5+ DSCs also had increased expression of the vascular cell adhesion molecule VCAM-1. We hypothesize surface VCAM-1 expression by ABCB5+ DSCs to support homing of these cells to the perivascular niche in skin and persistence in wounds. Additionally, it has been previously reported that murine Abcb5+ DSCs express the skin-homing molecules CCR4, CCR10 and E-selectin¹⁸. It is speculated that human ABCB5+ DSCs express these as well and should be the topic of future skin-homing studies along with investigations VCAM-1. ABCB5+ DSCs also expressed lower levels of some inflammatory cytokines such as IL-1β. It is possible that epigenetic regulation differs between BM-MSCs and ABCB5+ DSCs, which could have an impact on Notch signaling or exosome formation and secretion, impacting the ability of these cells to provide local immunoregulation²⁸.

Expression of MHC-II genes may allow ABCB5+ DSCs to evade the immune system and persist in an MHC-mismatched host¹⁸. It has been previously shown that ABCB5+ DSCs can skew macrophages towards an M2 anti-inflammatory phenotype although it is still unknown whether this occurs through a paracrine or direct mechanism. This phenomenon results in increased expression of the anti-inflammatory cytokine IL-1RA (receptor antagonist) leading to suppression of pro-inflammatory cytokines. IL-1RA expression was greater when MSCs and macrophages were in direct co-culture relative to the transwell system, suggesting that MSCs influence macrophage skewing by close proximity of secreted factors or a direct contact. While BM-MSCs were also able to increase the amount of secreted IL-1RA in vitro, the added ability of the ABCB5+ DSCs to home to wounded skin suggests their potential advantage over BM-MSCs for development as a cellular wound healing therapy.

CONCLUSION

Here, we have demonstrated the superior skin homing ability of ABCB5+ DSCs over BM-MSCs with implication for treatment of wounds in RDEB. The anti-inflammatory properties of ABCB5+ DSCs suggest that direct interaction with macrophages in the local dermal environment could provide greatest clinical benefit. Further studies will focus on the gene candidates HOXA3 and VCAM-1 in the mechanism of skin homing and wound healing properties of ABCB5+ DSCs. We also aim to discern the key secreted factors of ABCB5+ DSCs and the extent of their direct or paracrine interactions with macrophages. Elucidating additional ABCB5+ DSC immunomodulatory mechanisms could have broader regenerative medicine implications beyond RDEB treatment.

Supplementary Material

Suppl Figure S1

Figure S1: A) Standard curve for known mix of human and mouse DNA shows specificity of competitive PCR for presence of human DNA. B) PCR products of known mixes of human ABCB5+ dermal MSC and mouse skin DNA.

Suppl Figure S2

Figure S2: Characterization of THP-1-derived macrophages by CD11b expression after 24 hours of stimulation with PMA. A) CD11b expression on unstimulated THP-1 cells. B) CD11b expression on THP-1-derived macrophages after 24-hour stimulation with PMA.

Supple Figure S3

Figure S3: Human DNA was found in mice lungs at day 1 but not day 14. A) Graphical representation of percent human DNA in mouse lung tissue calculated from standard curve. B) PCR products of lung tissue at day 1 and day 14. NS = not significant.

SIGNIFICANCE STATEMENT.

Recessive dystrophic epidermolysis bullosa (RDEB) is caused by a defect in the COL7A1 gene resulting in the inability to produce type VII collagen (C7). C7 anchors the epidermis to the underlying dermis and supports skin integrity. Mesenchymal stromal cells (MSCs) are of interest in RDEB treatment due to their immunosuppressive role. A subset of MSC-like dermal cells (DSCs) identified by expression of ATP-binding cassette subfamily B5 (ABCB5) shows superior skin homing ability compared to bone marrow derived-MSCs. ABCB5+ DSCs express skin proteins and Hox genes, potentially responsible for enhanced homing to wounded skin. These novel findings provide mechanistic insight into the contribution of MSCs to treatment of chronic wounds.

DATA AVAILABILITY:

Data available from the corresponding author upon reasonable request.