Bone regeneration is a well-regulated process requiring the recruitment of skeletal stem/progenitor cells (SSPCs) from adjacent tissues. Among them periosteum is a connective tissue that surrounds the external surface of bones and has been widely described as physiologically essential for bone development and healing (Duchamp de Lageneste and Colnot, 2019; Julien et al., 2022; Perrin and Colnot, 2022). This membrane consists of a fibrous outer layer composed of type I collagen and a highly cellular inner layer composed of fibroblasts, osteoprogenitor cells and mesenchymal stem cells (Squier, Ghoneim, and Kremenak, 1990; Dwek, 2010). However, in case of critical bone defect after a massive trauma, a tumoral resection or an osteomyelitis, the surrounding soft tissues, such as periosteum, are usually damaged and the normal physiological healing capacity is insufficient (Vidal et al., 2020).

Current strategies for the reconstruction of large bone defects exploit surgical approaches such as Ilizarov’s (Aktuglu, Erol, and Vahabi, 2019) or Masquelet’s induced-membrane (Masquelet and Begue, 2010) techniques. However, these methods require lengthy and tedious processing. Bone repair therefore remains a major concern for orthopedic surgeons (Evans, Chang, and Knothe Tate, 2013; Delloye et al., 2014; Dalisson et al., 2021) which has led to the emergence of a series of alternative tissue engineering approaches in recent decades (Chen et al., 2015; Lou et al., 2021; Manon et al., 2023). Among these, Manon et al. recently showed the suitability of using human fascia lata matrices as a periosteal-like scaffold for bone regeneration, mainly through their collagen type I composition and appropriate mechanical properties (Manon et al., 2022; Manon et al., 2023). The human fascia lata (HFL) is a dense connective tissue made up of three layers of collagen fiber bundles with little cellularity (Pancheri et al., 2014; Blottner et al., 2019).

Along with bone marrow and muscle, the periosteum is an important source of skeletal stem/progenitor cells (SSPCs) with a high potential for bone regeneration (Zhang et al., 2005; Duchamp de Lageneste et al., 2018). These progenitors reside in a rich environment formed by the periosteal extracellular matrix, where cellular and molecular crosstalk is crucial to achieve bone regeneration (Julien et al., 2022). This three-dimensional scaffold therefore not only provides a structural and mechanical support (Costa-Almeida et al., 2015; Hussey, Dziki, and Badylak, 2018; Sainio and Järveläinen, 2020) but is also a modulator of tissue homeostasis (Theocharis et al., 2016). Consequently, any alteration of this structure can lead to the disruption of tissue function (Theocharis et al., 2016). Collagen constitutes the major ECM component (Shoulders and Raines, 2009; Yamauchi and Sricholpech, 2012) and plays a key role in maintaining biological and architectural integrity of tissues as well as providing physical support (Dong and Lv, 2016). Lin et al. (2018) further emphasized this point by revealing that cell-free collagen scaffold of decellularized periosteum alone efficiently promotes orthotopic remodeling of skeletal defects in vivo (Lin et al., 2018).

Fibril forming collagen type I, the most prevalent one, provides tensile strength to tissues (Theocharis et al., 2016) through its complex hierarchical structure. Encoded by the COL1A1 and COL1A2 genes, collagen precursors undergo series of post-translational modifications (PTMs) to form mature molecules (Gjaltema and Bank, 2017; Rappu et al., 2019), which are stabilized by two main types of intermolecular crosslinks important for their assembly and for collagen functions (Yamauchi and Sricholpech, 2012). First, enzymatic crosslinks, considered a post-translational modification of lysine residues, are mediated by the lysyl hydroxylases and lysyl oxidases through a highly controlled mechanism. Initially formed as immature divalent crosslinking (e.g., dehydrodihydroxylysinonorleucine), some are converted into mature trivalent crosslinks (e.g., pyridinoline) (Thorpe, 2010; Yamauchi and Sricholpech, 2012; Saito and Marumo, 2015; Kamml et al., 2023). Secondly, non-enzymatic crosslinks result from a glycation process forming advanced glycation end products (AGEs), among which pentosidine is the most frequently characterized (Paul and Bailey, 1996; Pritchard and Willett, 2017). All AGEs are formed by a Lysine-Arginine or Lysine-Lysine interaction and predominantly accumulate in collagen because of its low turnover. Their emergence leads to an increase in connective tissue rigidity and insolubility (Paul and Bailey, 1996; Gautieri et al., 2014). Finally, interfibrillar bridges also result from the interactions of non-collagenous components between glycosaminoglycans (GAG)-decorin (DCN) complexes of adjacent fibrils (Scott, 2003; Thorpe, 2010; Costa-Almeida et al., 2015). DCN is a member of class 1 small leucine-rich proteoglycans (SLRPs) essential for the regulation of various tissular processes (Halper, 2014; Gubbiotti et al., 2016) but most notably studied for its involvement in collagen fibrillogenesis and matrix stability. This protein also acts as a reservoir of growth factors and cytokines (Gubbiotti et al., 2016; Zhang et al., 2018).

In the present study harvested human periosteum (HP) and HFL samples were treated with several protocols of decellularization (Manon et al., 2023) as well as by irradiation. Whereas the general microstructure, composition, and biosafety of the ECM-derived scaffold had received considerable attention, a special analysis of collagen per se is required as the main component contributing to matrix-cell crosstalks and, hence, to the global integrity of the periosteum. Therefore, since decellularization and sterilization processes can have a negative impact on the ECM, the aim of the present study was to characterize the collagen scaffold obtained after these engineering processes. Here, we hypothesized and sought to confirm the global persistence of important ECM components of HP and HFL samples after chemical decellularization and sterilization, and therefore a persistence of their functional and structural properties. The ultimate goal of this study is to confirm the maintenance of HFL matrices properties in order to consider their future use as periosteal scaffold substitutes to promote bone regeneration.

All the decellularization protocols successfully achieved the Crapo’s criteria (Crapo et al., 2011). These standards include an absence of nuclei visible by histological analysis (DAPI, HE staining), a DNA quantity of less than 50 ng/mg dry weight and a length of less than 200 bp for persistent DNA fragments after treatment. In all the samples, DNA concentration felt below the critical threshold of 50 ng/mg of dry weight and no cellular traces were visible on microscopic analyses.

None of the five decellularization protocols (D1 to D5, Table 1) significantly altered collagen arrangement. Detergent-free treatment (D5) disorganized HFL samples to a greater extent, significantly reducing DCN and GAGs, and non-significantly reducing collagen cross-links. HP samples, treated with detergent-based treatment D1, showed collagen disorganization, and a reduction in enzymatic cross-links. Overall, decellularization had minimal impact on the matrix, while the addition of gamma irradiation impact leads to a significant negative effect on the integrity of collagen cross-links.

Structural modifications were therefore most pronounced in HFL when treated with the D5 protocol and in HP with the D1 protocol. This was first observed with the assessment of collagen bundle birefringence using PLM to understand the impact of the engineered treatments on the architectural arrangement (Dayan et al., 1989; Ribeiro et al., 2013; Rittié, 2017). The robustness of the results is confirmed by the convergence of birefringence measurements between various analysis tools. Interestingly, each of these subgroups, HFL-D5 and HP-D1, also showed the lowest levels of enzymatic crosslinks as measured by Raman spectroscopy. In the literature, optical density measures under polarized light were already used as indicator of crosslinked collagen organization (Kirby, Heuerman, and Yellon, 2018; Lien, Chen, and Phan, 2022).

Previous studies confirmed collagen persistence in decellularized scaffolds (Duisit et al., 2017; Manon et al., 2023). Being the major ECM component, collagen biological and architectural functions have already been well characterized (Snedeker and Gautieri, 2014; Dong and Lv, 2016). Ensuring the persistence of enzymatic crosslinks in decellularized collagen scaffolds is therefore crucial for maintaining the functional properties of native matrices, contributing to fibril stability and improving tissue biomechanical competence. Therefore, physical and chemical methods are nowadays used to achieve intermolecular collagen crosslinking in engineered scaffolds (Dong and Lv, 2016).

In addition to collagen enzymatic crosslinks, its overall molecular composition has an impact on its structure and biomechanical properties. Indeed, AGEs are known to alter the mechanical strength of tissues (Saito and Marumo, 2010). In addition, as Unal and Akkus (2015) have shown, the hydration level of bones also plays a role in their mechanical properties. Moreover, it is also known that physical training has an impact on collagen crimps, which can be analyzed by measuring birefringence (Mazon et al., 2018). Therefore, by taking all collagen parameters into account, we could in future investigate a relationship between the structural data of our scaffolds and their mechanical properties.

It is important to note, however, that the samples were taken from old subjects, which may also have an impact on the results. Since the rate of crosslinks is age-related, the slight decrease in enzymatic crosslinks after the engineering protocols could be explained by an age-related decrease, itself associated with a decrease in the mechanical capacity of the tissue. These decreases could therefore possibly be significant in tissues from younger subjects, which should be verified in the future. Moreover, it should be noted that these analyses were carried out on a small number of samples randomly taken from the entire decellularized scaffold, allowing potential tissue heterogeneity to be taken into account.

Raman spectra of the HFL samples differed slightly from those of the periosteal ones, particularly in the amide I region (1670/1690 cm⁻¹). We suspected interference from tissues other than periosteum, adding some background in the spectra. The samples were harvested from old subjects and, despite efforts to remove muscle and fat, some residues remained due to the thinness of the HP. Indeed, anomalies in the spectra were observed for all protocols, including native tissue, with the exception of protocol D5, the only one to include a degreasing step. Therefore, in line with the literature, an additional peak (1657 cm⁻¹) was added for deconvolution of this amide region, known to correct for potential interference from lipid vibrations (Mandair et al., 2020). In addition, we observed increased hydration in samples after decellularization, probably because decellularization disrupts the natural barriers regulating hydration. This leads to increased water uptake during abundant rinsing steps. Irradiation, however, leads to a partial loss of this increased hydration, without returning to baseline levels.

Since these decellularized tissues are intended to regenerate in vivo damaged structures, sterilization processes are required to prevent pathogen transmission (Dziedzic-Goclawska et al., 2005; Seto, Gatt, and Dunn, 2008; Harrell et al., 2018). Althoug gamma irradiation is commonly used in bone banks for sterilization, it causes radiolysis of water molecules, releasing free radicals (Nguyen et al., 2007) that damage collagen (Seto, Gatt, and Dunn, 2008; Harrell et al., 2018). Previous studies on bones showed a negative synergic effect on mechanical resistance when irradiation was added to freeze-dried samples (Cornu et al., 2000; Cornu et al., 2011), attributed to a significant reduction in enzymatic crosslinks density (Cornu, 2012). This is consistent with our results revealing a more pronounced alteration of crosslinks post-irradiation. Future studies should explore decellularization procedures under totally sterile conditions to avoid the detrimental effects of subsequent irradiation on tissue properties.

We did not investigate irradiated periosteum since the priority, for the creation of the new regenerative membrane, was the use of irradiated decellularized HFL (Manon et al., 2023).

DCN-GAG complexes are essential for the ECM organization, especially in collagen fibrillogenesis, as evidenced by the alteration of the latter in DCN-deficient mice (Gubbiotti et al., 2016; Zhang et al., 2018). This PG is also involved in stress transfer between collagen fibrils (Thorpe, 2010) and in the bioavailability of growth factors such as TGF-β (Gubbiotti et al., 2016; Zhang et al., 2018). Alteration in decorin content vary according to the tissue type, with HFL samples globally maintaining their DCN concentration after decellularization, except for D5, while HP samples show a decrease in all D-protocols.

Furthermore, it has already been demonstrated that detergent-based decellularization affects the GAG composition of ECM (Duisit et al., 2017; Liguori et al., 2020; Uhl et al., 2020), as confirmed by our quantitative measurements, with potential dysfunction of persistent GAGs (Uhl et al., 2020). However, the maintenance of functional GAGs is necessary. Abundantly distributed on cell surface and in the ECM, they modulate various physiological processes. They provide biochemical and mechanical support in the ECM and influence cell behavior (adhesion, proliferation, differentiation) (Salbach et al., 2012; Menezes et al., 2022).

Despite the lower proportions of DCN and GAG in decellularized samples compared to native ones, no correlation could be demonstrated in their alteration after chemical treatments. Since GAGs are ubiquitous on the cell surface, their removal by the decellularization protocol is expected. Therefore, proportionally, the loss of GAGs could be greater than that of DCN.

The present study confronts different decellularization approaches. Protocols D1 to D4 are based on the use of Triton, SDS and DNase and ensure cell membrane lysis, protein solubilization and residual DNA elimination. Triton and SDS are known for their efficiency in decellularization, but also for altering the structure of the ECM (Crapo et al., 2011). Protocol D5 consists of a succession of solvent- and detergent-based baths, which has also already demonstrated its decellularization potential (Manon et al., 2023), including the monitoring of Crapo’s criteria (Crapo et al., 2011). In this protocol, solvents such as ethanol and acetone are used for cell lysis and lipid removal but are both also known as altering the ECM ultrastructure. The additional advantage of this protocol is that it also neutralizes prions, which the other protocols do not allow.

As qualitatively shown in the Figure 1 on the SR stained sections, more undulating collagen structure are observed in decellularized tissues than in native tissues. The crosslink alteration could modifiy the fibers alignment. In the future, analysis of the collagen distribution could support the observations made in this study. Approaches such as transmission electron microscopy (Adeoye et al., 2022), histology and contrast-enhanced microCT (Maes et al., 2022) could be used and be gathered in order to precisely cartograph the entire HFL structure.

In conclusion, a relationship was observed between the ECM structure and its molecular composition. The decellularization approaches demonstrated no significant alteration of the matrices, which justifies, along with the analysis of other parameters, the future clinical use of these processed matrices for bone regeneration. However, the combination of D5 treatment with subsequent irradiation appears the most deleterious for tissue properties. In the future, an analysis of the entire sample surface would provide more accurate information about potential sample heterogeneity, and precise quantitative analyses of the crosslink content could further validate our results. It would also be interesting to correlate all the observations of this study with the mechanical properties of the tissues.