Benefits and Risks of Using Implants Coated with a Natural Protein to Help Skin and Bone Heal

Soft and hard tissue injuries usually involve the disruption of blood vessels. To stop the bleeding, a coagulation cascade is initiated within minutes after injury, which requires the activation of platelets. During this process, platelets release large amounts of active TGF-β (predominantly TGF-β1) from their alpha granules. The resulting local increase in active TGF-β helps to attract immune cells to the site of tissue damage and control their function in the following inflammatory phase [25,26].

Within hours after injury, neutrophils are the first immune cells to invade the wound or fracture. They also release active TGF-β (predominantly TGF-β3) from their granules [27]. Typically, within 1 to 3 days after injury monocytes invade the damaged tissue. The local increase in active TGF-β within the wound or fracture then modulates macrophage phenotypes to resolve the inflammation [25,26]. First, pro-inflammatory M1 macrophages aid in clearing the wounded tissue from cellular debris and invading pathogens; after this, pro-fibrotic M2 macrophages are involved in the resolution of inflammation. This also includes the modulation of immune cells from the adaptive immune system that invade the wounded tissue in the following days and weeks.

Usually after 2–5 days the proliferation phase starts. In the following course of healing, less and less active TGF-β is provided by immune cells. Instead, TGF-β needs to be activated from the extracellular reservoir of latent TGF-β by proteolytic cleavage from latent TGF-β binding protein (LTBP) or by mechanical forces or the acidification of pH. In the proliferation phase, TGF-β stimulates fibroblast differentiation into myofibroblasts [28], promotes angiogenesis [29,30], and regulates ECM deposition [18]. While the regulation of angiogenesis (endothelial cells) and the recruitment of tissue fibroblasts is comparable in wound and fracture healing, the additional invading cells and the formed ECM differ.

The remodeling phase aims to restore the pre-injury state of the skin or bone tissue. In this step, balanced TGF-β activity ensures proper ECM turnover via orchestrating collagen deposition and metalloproteinase activity [34,35].

Extracellular matrix protein 1 (ECM1) is a glycoprotein secreted into the ECM. Up to now, two splice variants for ECM1 have been described in humans, ECM1a and ECM1b. ECM1a, encoded by a 1.8 kb mRNA, is mainly expressed in the highly vascularized heart and placenta, but is also present in the liver, ovaries, kidneys, lungs, pancreas, testes, muscle, and colon. ECM1b, encoded by a 1.4 kb mRNA lacking exon 7, is expressed in the skin and tonsils [41]. In these tissues, ECM1 is predominantly expressed by keratinocytes, fibroblasts, and endothelial cells [42]. It regulates skin barrier integrity, angiogenesis, keratinocyte differentiation, and matrix organization [43]. Its clinical relevance is underscored by Lipoid Proteinosis, a genetic disease caused by ECM1 mutations, characterized by scarring and impaired skin integrity [44].

In wound healing, ECM1 has been shown to (i) modulate angiogenesis by interacting with VEGF and perivascular ECM components [17,45]; (ii) bind to perlecan and influence basement membrane assembly and keratinocyte migration [29,46]; (iii) regulate matrix turnover by inhibiting MMP-9 activity, thereby balancing ECM degradation and deposition [17,47].

Regarding bone regeneration and fracture healing, ECM1 has been shown to (i) be expressed in differentiating chondrocytes upon the activation of parathyroid hormone-related peptide signaling [48]; (ii) negatively regulate chondrogenesis and endochondral ossification both in vitro and ex vivo [48,49]; (iii) inhibit skeletal growth when overexpressed in chondrocytes and osteoblasts [48].

In tissue regeneration, ECM1 functions as an integral mediator balancing angiogenic activity and fibrotic suppression, both of which are tightly regulated by TGF-β-dependent molecular mechanisms [15,17,46,47,50,51]. Moreover, ECM1 deficiency correlates with abnormal myofibroblast persistence, excessive collagen I deposition, and impaired re-epithelialization—all classically associated with TGF-β dysregulation [35,52].

TGF-β, with its three isoforms (TGF-β1, -β2, and -β3), is by far the most abundant cytokine within the TGF-β superfamily. TGF-β is expressed as a latent protein, the so-called latency-associated peptide (LAP). The LAP is covalently bound to LTBP through disulfide bonds in the endoplasmic reticulum. The resulting large latent complex (LLC) is then incorporated into the extracellular matrix, waiting to be activated either enzymatically by proteolytic cleavage, chemically by a drop in pH, or mechanically by interaction with integrins [53]. The released active TGF-β dimers transduce their signals through two types of serine/threonine kinase receptors, termed type I and type II [54]. The type II receptors are constitutively active kinases, which phosphorylate type I receptors upon ligand binding. Seven type I receptors termed activin receptor-like kinase (Alk)-1 through -7 were identified in mammals, of which TGF-β1-3 preferably bind Alk-4 and Alk-5. Upon activation by the type II receptor, Alks activate (phosphorylate) Smad transcription factors in the cytoplasm. TGF-β1-3-activated signaling is mainly mediated via the receptor-regulated Smads (R-Smads) -2 and -3, which upon phosphorylation complex with the common partner Smad (Co-Smad) 4 to enter the nucleus and regulate the expression of target genes, including genes participating in feedback mechanisms of the signaling pathway itself [36,54]—for overview see Figure 2.

Mechanistically, ECM1 modulates TGF-β signaling through the regulation of the growth factor bioavailability. By binding matrix proteoglycans and interacting with collagen/elastin networks [46,47,50], ECM1 influences how latent TGF-β binding proteins (LTBPs) store and present TGF-β in tissues [51]. There are different modes of actions described [55]. Integrins, such as αvβ3, αvβ5, and αvβ6, can mechanically activate TGF-β by connecting the LLC to the contractile cytoskeleton. This requires binding the integrins to the RGD (arginine–glycine–aspartic acid) motif in the LAP. It has been reported that ECM1 protects TGF-β from activation by competitively binding to this RGD sequence, thus preventing the interaction of LAP with the integrins [55]. ECM1 was further reported to interfere with the proteolytic activation of TGF-β by blunting the activity of matrix metalloproteinases 2 and 9 (MMP2 and MMP9), ADAMTS1 (a disintegrin and metalloproteinase with thrombospondin motifs 1), and TSP1 (thrombospondin 1) by interacting with their intrinsic KTRF (lysine–tryptophan–arginine–phenylalanine) and KRFK (lysine–phenylalanine–arginine–lysine) motifs, respectively [45].

Recombinant human ECM1 at an implant surface should prevent excessive TGF-β activation. The early hemostasis and inflammatory phase might not be affected by such a coating, as their active TGF-β is released from platelets and immune cells, such as neutrophils. However, during the proliferation phase a lack of TGF-β activation might hinder re-epithelialization, angiogenesis, and the formation of the provisional ECM, as observed when all three TGF-beta isoforms are neutralized with neutralizing antibodies [81].

In order to exert its biological effect, ECM1 has to be presented within the extracellular space; therefore, surface-coating techniques facilitating non-covalent and covalent protein bonding with the implant surface may be a promising strategy. Non-covalent coating techniques include techniques such as electrophoretic deposition, precipitation, and dip or spin coating. These methods use physical adsorption by van der Waals forces or electrostatic interactions to bind the protein (dissolved in a liquid) to the implant's surface. Due to the reversible adhesive forces, non-covalently bound proteins may be rapidly released from the implant surface. In contrast, covalent bonding approaches create a stronger and thus more permanent bond between the protein and the implant surface. To facilitate the covalent bonding of proteins with the implant surface, the implant surface has to be chemically modified by creating reactive sites for proteins by silanization, polyethylene glycol (PEG) coating, or via chemicals that allow ultraviolet light (UV)-mediated crosslinking. Biomaterial scaffolds mimicking the ECM or incorporating ECM fragments have been proposed to enhance wound repair outcomes [82,83]. Therefore, indirect implant coating by peptide bonds or mineral coating that allows the timely release of the ECM1 [84,85] would be preferable to direct coating techniques, such as spray or dip coating. In contrast, the late remodeling phase application of ECM1 might be beneficial to prevent scar or keloid formation in respective risk patients. With this, the recombinant human ECM1 might be delivered later in the healing phase incorporated in hydrogel wound patches or as peptide-based nanoparticles [86,87,88]. However, in both cases the presentation of ECM1 within the extracellular space needs to be provided to ensure its biological function.

The knowledge of specific motifs either in the ECM1 protein or the interacting proteases offers great perspectives for the modulation of ECM1 effects. Much of this knowledge is obtained from murine studies on the effect of TGF-β signaling in liver fibrosis. On the one hand, the hepatocyte-specific knock-out of ECM1 caused latent TGF-β1 activation and spontaneously induced liver fibrosis with rapid mortality [45,89]. On the other hand, the overexpression of ECM1 was able to attenuate liver cirrhosis in mouse models [90,91]. In patients with chronic liver disease (CLD), ECM1 expression is inversely associated with the levels of TSP1, ADAMTS1, MMP-2, MMP-9, and LTGF-β1 activation [45,92].

By investigating the underlying effects, ECM1 was shown to suppress the activity of the proteases TSP1, ADAMTS1, MMP-2, and MMP-9, which are all involved in latent TGF-β1 activation. Immunoprecipitation experiments proved that ECM1 binds directly to these enzymes. In the case of TSP1 and ADAMTS1, specific tetrapeptide sequences were identified that control this interaction—the KRFK and KTRF motifs [45]. Both motifs are proposed to induce a conformational change in the LTGF-β1 molecule, resulting in the release of the active TGF-β1 ligand without the need for additional proteolysis. In line with this, the application of KRFK tetrapeptides, which is characteristic for the interaction with TSP1, induced the activation of latent TGF-β. This effect was partly attenuated by the overexpression of ECM1 [45]. In contrast, the application of KTRF tetrapeptides suppressed the activation of latent TGF-β, even in ECM1 knock-out mice [45].

The knowledge of this opposite effect of these two tetrapeptides offers great perspectives for the biofunctionalization of implants. In particular, due to their very small size, tetrapeptides can be applied to implants in larger amounts than the recombinant human ECM1 by using the same application methods described before. As the KTRF tetrapeptide simulates the ECM1 effect, its application might be beneficial later in the healing cascade when a resolution of the inflammation is required to prevent excessive scar formation. The KRFK tetrapeptides, which were shown to antagonize the ECM1 effects, might exert beneficial effects early in the proliferation phase, such as improving re-epithelialization, angiogenesis, and ECM production.

Using RNA- or DNA-based methods might have the advantage that the target gene needs to be expressed and secreted by the local cells, which causes a natural delay in the proposed effects. This is advantageous when ECM1 will be overexpressed, as this is supposed to interfere with the early phases of wound healing. Given this required delay of the proposed effect, this argues for the DNA-based rather than the RNA-based overexpression of ECM1—preferably delivered as episomes or plasmids that allow the transient expression of the transferred gene without integration into the genome.

In contrast, the suppression of ECM1 might be beneficial in the early proliferation phase of the healing cascade, where active TGF-β supports re-epithelialization, angiogenesis, and ECM production. There a timelier effect is required, an RNA-based rather than a DNA-based knock-down of ECM1 can be argued for. However, similar to the DNA-based methods, the delivery technique may have a significant effect on the speed and duration of the response. Usually, lnRNAs and aptamers provide the most rapid response with the shortest duration of the effect lasting between hours and days. This is due to their short half-life, which for lnRNAs is defined mainly by their primary and secondary structures, but also by their interaction with other proteins within the cellular context [93]. The half-life of aptamers, in contrast, is defined by renal clearance and nuclease degradation. However, their half-life and therapeutic effect can be extended through modifications such as PEGylation, which increases their size and resistance to nucleases [94]. An intermediate effect duration (~5 days) is expected from miRNAs, which have an average half-life of about 5 days. Their stability is enhanced when forming a complex with the argonaute proteins as part of the RNA-induced silencing complex (RISC) [95]. Synthetic miRNAs follow the same degradation kinetics as siRNAs; thus, the expected effect duration is also comparable (5 to 7 days). A more persistent effect is obtained by shRNAs that are often expressed from a plasmid or viral vector by the host cells. This might even be permanent, when the delivery method includes integration into the hosts' genome [96]; however, this is unfavorable for therapeutic use.