Molecular Therapeutics in Development for Epidermolysis Bullosa: Update 2020

Many laboratories, in both academia and the pharmaceutical sector, have focused on gene-replacement approaches, particularly delivery of COL7A1 to the skin of patients with RDEB (Table 1). For details on methods, hurdles and risks, we refer to recent review articles [27, 28]. Some studies have tested the efficacy of topical application of an expression vector harboring full-length COL7A1 complementary DNA (cDNA), which would then allow expression of the proα1(VII) polypeptides in the skin, followed by their incorporation into trimeric type VII collagen molecules and supramolecular assembly into functional anchoring fibrils.

Another approach for delivery of COL7A1 into patients with RDEB entails introduction of the transgene into the patient’s own cells in cultures, with subsequent delivery of the corrected cells back into the skin. One such approach (Castle Creek Biosciences, Inc.) corrects autologous fibroblasts in culture with a vector that leads to integration of the transgene into the genome, followed by direct injection of the corrected cells to the edges of the wounds. This approach necessitates multiple injections, which can be painful, to the eroded areas of skin, and how long the corrected resident fibroblasts are present and remain active in situ is unclear. Another company (Abeona Therapeutics, Inc.) cultures autologous keratinocytes from the skin of patients with RDEB, followed by genetic correction ex vivo and development of epidermal sheaths, which can then be grafted to the denuded areas of the patient’s skin [29, 30]. These studies have utilized a lentiviral vector that allows incorporation of the COL7A1 cDNA into the genome of the recipient cells. Preliminary studies have revealed that the gene-corrected keratinocytes in the graft are capable of expressing type VII collagen, and there is evidence of assembly of anchoring fibrils. Of some concern is the durability of this approach since the early published data suggested that the expression of the collagen gene may fade over time, possibly attesting to the fact that the transgene is driven by a viral promoter or that the number of stem cells initially isolated and targeted were too low to sustain expression or possibly the graft [31]. Furthermore, in some cases, the graft fails after a certain time, potentially necessitating regrafting of the area. In this context, a recent study demonstrated that correction of both keratinocytes and fibroblasts (cell types that in normal skin both synthesize type VII collagen) is required for optimal assembly of the anchoring fibrils [32]. In this study, skin grafts were made by combining gene-corrected keratinocytes with type VII collagen-deficient fibroblasts, or, conversely, type VII collagen-deficient keratinocytes in combination with gene-corrected fibroblasts, or both cell types being gene corrected. All these three combinations of cells in skin-equivalent explant culture ex vivo expressed type VII collagen, but only the grafts in which both keratinocytes and fibroblasts were gene corrected showed assembly of functional anchoring fibrils. In keeping with these data, there is an ongoing effort to develop skin grafts (GENEGRAFT) that combine keratinocytes and fibroblasts in which both cell types have been corrected with a self-inactivating viral vector expressing type VII collagen [33].

Epidermal grafts have been shown to be highly successful in correcting the underlying defect in patients with intermediate or localized JEB with defects in one of the laminin 332 genes, LAMB3 [33‐35]. The difference with these studies and those published on grafting in patients with RDEB is that the grafts for patients with JEB were made of a holoclone stem cell population, thus ensuring the longevity of the cells after transplantation. In fact, in one patient with JEB, the functionality of the skin grafts was retained several years after the transplantation [36]. The difference in the outcome between the epidermal cells corrected for transplantation into patients with RDEB and JEB may also reside in the observation that, in normal skin, the laminin 332 polypeptide subunit genes, LAMA3, LAMB3 and LAMC2, provide a selective advantage for retaining stem cells in vivo [37, 38].

In addition to gene-replacement approaches, a number of studies have attempted gene repair, such as correction of the mutation by CRISPR/Cas9 editing technology [39‐43] (Table 2). For details on gene-editing strategies and their appropriate selection depending on mutation, gene and disease type, we refer to a recent overview article [44]. Improvements in this technology have resulted in a high yield of gene correction in keratinocytes or fibroblasts derived from patients with RDEB, with subsequent development of skin grafts that have been transplanted into immunocompromised mice, demonstrating their capability of functional type VII collagen synthesis. One of the potential limitations of this approach is the requirement for a large number of cultured cells for graft production, and an innovative way to circumvent this limitation would be the development of iPSCs that, following the gene correction, can be differentiated into either keratinocytes or fibroblasts [42, 45].

An intriguing possibility for cell-based therapy in EB is provided by revertant mosaicism, a phenomenon described as “natural gene therapy,” in which a number of skin cells undergo spontaneous reversal of the mutation to wild-type genotype, resulting in areas of normal skin [46, 47]. The mechanisms of the mutation reversion are multiple, including mitotic recombinations, back mutations and second site mutations [48‐50].

Revertant mosaicism in patients with RDEB has been primarily documented in keratinocytes, but evidence of revertant mosaic fibroblasts has also recently been reported [51]. Attempts to establish long-term cultures of keratinocytes from areas of skin with revertant mosaicism have been mostly unsuccessful, primarily due to depletion of stem cells in the affected skin. However, punch grafting of revertant skin to isolated lesions in patients with laminin 332-deficient JEB has been reported [52]. Transplantation of the biopsy specimens resulted in re-epithelialization of the wounds, but punch grafting allowed only a limited expansion of the revertant skin [53]. However, expansion of the treated area can be achieved by generation of iPSCs from revertant keratinocytes, and these cells can then be differentiated into genetically corrected keratinocytes [45]. These revertant iPSCs can be used to create three-dimensional skin equivalents ex vivo and reconstitute human skin in vivo, developing cell-based therapeutic approaches for EB.

Exon skipping makes use of antisense oligonucleotides (ASOs) to modify the splicing of pre-messenger RNA (mRNA) and consequently eliminate the mutation responsible for a disease. Antisense-mediated exon skipping was initially developed for the treatment of Duchenne muscular dystrophy (DMD) and evolved into clinical trials to target DMD (dystrophin gene) exons with recurrent mutations, such as exons 45, 53 and 51. Eteplirsen (Exondys 51) is the first approved antisense therapy for DMD in the USA and provides a treatment option for the ~ 14% of patients with DMD who are amenable to exon 51 skipping [54].

The rationale for using exon skipping in EB comes from studies of genotype–phenotype correlations showing that exon skipping can ameliorate the severity of JEB and DEB [4, 55‐57] and from the fact that most COL17A1 and COL7A1 exons are in frame. Several preclinical studies have demonstrated that skipping of exons 13, 70, 73, 80 or 105 in COL7A1 results in slightly shortened, partially functional protein, which is deposited at the DEJZ [58‐61] (Table 2). A clinical trial testing topical administration of QR-313, a water-based gel (hydrogel) containing the ASO targeting exon 73 that will be applied directly onto DEB wounds for COL7A1 correction is currently ongoing (Wings Therapeutics; NCT03605069).

Preclinical studies have shown that recombinant type VII collagen, injected locally or intravenously, homes to the DEJZ and promotes wound healing [62, 63]. Recombinant type VII collagen has obtained fast drug designation from the US FDA, and a phase I/II clinical trial evaluating its safety and tolerability, as well as clinical proof of concept in adults with RDEB, is ongoing in the USA (Phoenix Tissue Repair, Inc.; NCT03752905). The advantages of systemically administered recombinant type VII collagen include the possibility that the protein, in addition to homing into skin, will reach extracutaneous tissues affected by RDEB, such as the gastrointestinal track and the cornea of the eye, with subsequent repair.

Beside this protein-replacement approach, a therapy with a high mobility group box 1 (HMGB1) fragment is under development (redasemtide trifluoroacetate phase II, Shinogi & Co., Ltd.). HMGB1 can mobilize the Lin(−)/PDGFRα(+) cells from bone marrow to damaged tissue and facilitate tissue repair [64], probably by suppressing the inflammation of injured skin [65]. In animal models, systemic administration of the HMGB1 fragment has shown benefits, preventing deterioration of cardiac performance in the delta-sarcoglycan-deficient hamster [66] and ameliorating cutaneous and non-cutaneous manifestations in a dystrophic EB model mouse [67].

The read-through approach involves the use of small-molecular-weight compounds that allow the translational machinery to suppress nonsense mutations by incorporating an amino acid in place of a stop codon and results in synthesis of full-length protein.

The original prototype of such read-through molecules, PTC124, was shown to read-through only pathogenic PTCs but not through naturally occurring endogenous stop codons of translation because the nucleic acid context and the intron–exon organization of the gene provide a very robust and strong stop signal. This read-through molecule has been tested on a number of genes, principally the dystrophin gene in DMD [68], but some studies have failed to identify PTC124 as being able to efficiently read-through COL7A1 PTCs as compared with the prototypic PTC read-through drug, the aminoglycoside gentamicin [20, 23, 69]. Work has shown that gentamicin can effectively read-through COL7A1 and LAMB3 PTCs [20, 21], and clinical studies in RDEB have shown promising results [22]. Long-term systemic treatment with this antibiotic has a number of potential issues, including the risk of renal toxicity [70], and topical application might be a favorable delivery route in this context. Nevertheless, clinical studies testing the effect of intravenous injections of gentamicin are being conducted for RDEB (NCT03392909) and JEB (NCT03526159).

Another compound recently shown to induce read-through of COL7A1 is amlexanox [71], a drug with a number of different activities and targets currently in trial for a wide range of indications. This drug could potentially be used long term because of its very favorable toxicity profile and clinical history. However, the current manufacturer of amlexanox, Takeda Pharmaceutical, has removed it from the market, and an alternative supply will be needed to pursue clinical application. One interesting observation from the amlexanox study in EB was that it enhanced read-through only in cells with a low level of endogenous full-length protein, whereas cells with undetectable protein did not respond [23]. These data suggest that certain cellular conditions required for read-through need to be identified and exploited for more efficient and effective approaches.

Fibrosis is a major pathological complication of RDEB, and work with patient cells in culture, animal models and patients have all identified TGFβ signaling as a major driver of fibrosis and disease severity [9, 72‐75]. TGFβ is a primary mediator of fibrosis driving extracellular matrix (ECM) deposition in numerous pathological situations, and considerable effort has focused on understanding and inhibiting TGFβ in a number of contexts [76]. However, as TGFβ receptors participate in a signaling pathways that control many aspects of mammalian development and tissue homeostasis, global inhibition has proven problematic [77]. Indirect targeting of molecules that activate or inhibit the pathway in specific contexts will likely have improved efficacy compared with global inhibition. The first example of such an approach in RDEB is the use of losartan, a drug approved for the treatment of high blood pressure. Work in animal models identified a significant reduction of fibrosis in the paws of hypomorphic mice with losartan treatment, and this drug is now in a clinical trial (EudraCT number: 2015-003670-32) [78]. The mechanism here is thought to be a reduction in blood pressure, which reduces the bioavailability or release of TGFβ ligand from the fibrotic ECM. Other preclinical efforts have used viral delivery of decorin, an inhibitor of TGFβ, to show reduced paw fibrosis in mouse models [79], whereas in vitro work with patient cells has suggested that inhibition of thrombospondin-1, a potent activator of TGFβ, may have clinical efficacy [72]. In addition to these academic-led initiatives, a number of anti-TGFβ antibodies are in development in the commercial space that have potential to reach clinical trial in the coming years.