Human articular cartilage repair: Sources and detection of cytotoxicity and genotoxicity in photo‐crosslinkable hydrogel bioscaffolds

Abstract Three‐dimensional biofabrication using photo‐crosslinkable hydrogel bioscaffolds has the potential to revolutionize the need for transplants and implants in joints, with articular cartilage being an early target tissue. However, to successfully translate these approaches to clinical practice, several barriers must be overcome. In particular, the photo‐crosslinking process may impact on cell viability and DNA integrity, and consequently on chondrogenic differentiation. In this review, we primarily explore the specific sources of cellular cytotoxicity and genotoxicity inherent to the photo‐crosslinking reaction, the methods to analyze cell death, cell metabolism, and DNA damage within the bioscaffolds, and the possible strategies to overcome these detrimental effects.

providing new capabilities to create, for instance, implantable 3D constructs composed of biomaterials and living cells, intended as bioscaffolds. The biofabrication approach, of which the long-term goal is to switch from nonbiological prosthesis to biological implants has generated promising results in the repair of articular cartilage. 2 Traumatic cartilage (chondral) injuries can result in osteoarthritis, a major source of disability in the developed world. 3 Current treatments to repair chondral lesions, which include autologous chondrocyte implantation, mosaicplasty, and microfracture, are unable to reproduce hyaline cartilage capable of sustaining shear and compressive forces associated with normal joint function. The generation of a bioscaffold, using a combination of biomaterials and cells, is a possible solution for cartilage repair. 4 Currently, this technology has generated promising results both in vitro 5 and in in vivo 6 as well as in preclinical studies. 7 Nevertheless, there are several open questions with respect to its clinical use, 8 especially regarding the safety of the cells implanted, given the multiple sources of cytotoxicity and genotoxicity intrinsic to the bioscaffold generation process. In particular, the chemistries required to generate covalently crosslinked 3D hydrogel environments can have cytotoxic impacts on embedded cells. To our knowledge, the wider literature describing these issues has not previously been critically examined. Thus, the aim of this review is to summarize the sources of cellular damage and provide an indication on the reliable tests to be used to verify the safety of the cells implanted in bioscaffolds. Limitations of current surgical treatments for cartilage repair such as microfracture have prompted the field of cartilage regenerative medicine to integrate engineering and biological principles to promote the growth of new cartilage to replace the damaged tissue. To date, a wide range of scaffolds and cell sources have emerged toward cartilage tissue engineering, with a focus on recapitulating microenvironments present during the human body development or in the adult tissue. These microenvironments should induce the formation of cartilaginous constructs with biochemical and mechanical properties similar to the native tissue. Hydrogels have emerged as a promising scaffold material due to the wide range of properties that are possible to achieve, and the ability to trap cells within the material. 9 2 | PHOTO-CROSSLINKABLE HYDROGEL BIOSCAFFOLDS FOR ARTICULAR CARTILAGE REPAIR: GENERAL OVERVIEW Cartilage biofabrication strategies are designed to overcome the limitations of injection-based stem cell therapies; namely, the massive cell death upon delivery caused by shear stress from the needle, poor engraftment of delivered cells and, as a consequence, limited ability to differentiate into a chondrogenic phenotype. 10,11 Rather than injecting cells directly, one alternative is to deliver stem cell laden hydrogel bioscaffolds to fill the defect. The encapsulating hydrogel has a protective effect on the cells from shear stress and mechanical cytotoxicity coming from the extrusion or the bioprinting process. The 3D microenvironment provided by the hydrogel scaffold supports cell survival and can stimulate differentiation into mature chondrocytes capable of producing their own extracellular matrix. The scaffold is typically designed to degrade over time while it is replaced by newly formed cartilage tissue arising from the cells implanted, producing a tissue which resembles the native articular cartilage. Traditional techniques to generate bioscaffolds include casting of cell-laden hydrogelbased materials in molds of the desired size and structure to be implanted in the defects to be treated, but also inkjet 3D bioprinting, micro-extrusion, in situ bioprinting (Table 2). More advanced biofabrication techniques include the production of neocartilage tissue to better recapitulate the native zonal architecture of the articular cartilage, or the generation of multiphasic constructs with hybrid approaches using different materials, 3D printing techniques, 12 or different cell sources and lineages. 13 To achieve encapsulation, the cells are typically mixed with a liquid hydrogel solution, which is then crosslinked to form a contiguous stable network under physiological conditions.
Since the behavior of chondrocytes is, in part, mediated by the mechanical environment, matching the mechanical properties of the scaffold to that native cartilage also needs to be considered. 14 However, the crosslinking reaction can impact on cell viability and metabolism, and consequently on chondrogenic differentiation. One of the most widely adopted crosslinking strategy uses polymers (naturally derived or synthetic) which have been modified with reactive groups (methacrylate and/or methacrylamide) which can undergo chain polymerization reactions. [15][16][17] The process of protein crosslinking comprises among all chemical, enzymatic, chemo-enzymatic, self-assembly, ionic, thermal formation of new covalent bonds between polypeptides. 18 These reactions allow the site-directed coupling of proteins with distinct properties and the de novo assembly of polymeric protein networks. The chemical photo-crosslinking process is the most investigated and the most common way to achieve a precise spatial and temporal hardening of the hydrogels. At the same time, it is the one that deserves much attention given the presence of different drawbacks that can impair cell viability.
One of the main advantages of photo-crosslinking is the rapid formation of hydrogel networks at ambient temperature under mild conditions, and the tunability of the mechanical properties. The crosslinked site is also ready to be accurately selected, because the photoinitiated polymerization takes place under light exposure and only the irradiated areas are involved in hydrogel crosslinking. The ionic/electrostatic interactions can instead achieve extremely limited mechanical strength. Moreover, the photo-crosslinking process is the preferable choice to perform in situ bioprinting with robotic arms or handheld approaches, 19 which is emerging as a favored bioprinting strategy during certain clinical situations when compared with conventional in vitro bioprinting. Finally, the photocrosslinking strategy allows to a precise temporal and spatial control compatible with the time frame in theater for surgical operations.
In the light-induced crosslinking process, a photoinitiator (PI) molecule is mixed within the reactive hydrogel, and the reaction is initiated through exposure to UV or visible light of a wavelength ( Figure 1). However, such photo-crosslinking reactions can create a transiently cytotoxic environment, which may compromise cell viability and/or phenotype. 20 Other sources of cytotoxicity inherent in the bioscaffold generation process include the shear stresses during the extrusion and the poor diffusion of nutrients or oxygen through the crosslinked hydrogels. 21 Extended "fabrication times" involving exposure to nonphysiological conditions, such as room temperature or the lack of control of oxygen and CO 2 levels within an enclosed bioprinting cartridge, can also have a strong impact on viability in later culture. 22,23 The clinical application of articular cartilage repair strategies will require the identification, quantification, and mitigation of such toxic effects. The scope of this review is to cover the specific sources and detection methods of cellular cytotoxicity and genotoxicity inherent to the photo-crosslinking reaction.
Thus, in the following sections, we will examine: (a) the sources and effects of the photo-crosslinking process on cells viability and DNA integrity; (b) the methods to analyze cell death, metabolism and

Significance statement
Several hurdles need to be addressed before the clinical translation of articular cartilage regeneration procedures using photo-crosslinkable hydrogels. Cellular cytotoxicity and genotoxicity need to be identified and carefully detected to provide an indication of the safety of the repair treatment approach in patients.
DNA damage within the bioscaffold; and (c) potential solutions to overcome these detrimental effects.

| PHOTO-CROSSLINKABLE HYDROGEL BIOSCAFFOLDS FOR ARTICULAR CARTILAGE REPAIR: COMPONENTS AND PROCEDURES
The assessment of the cytotoxic and genotoxic effects of the photocrosslinking process requires preliminary consideration of the type of hydrogels and cells that constitute the bio-ink and therefore the final bioscaffold. Although the type of cells may differ in their susceptibility to cyto-and genotoxic effects, the hydrogel itself can influence cell survival and behavior through the presence of functional groups and bioactive moieties that favor its hardening. Finally, the method of delivering the bioscaffold may influence the crosslinking conditions.

| Sources of cells
Adult mesenchymal stem cells possess self-renewing abilities and inherent chondrogenic properties which lend to be the elective cell type in cartilage regeneration. 24 In particular, human adipose-derived stem cells (hADSCs) have been incorporated into many different scaffold-based systems and have shown promising results in cartilage tissue engineering. 25 The two major sources of hADSCs are abdominal fat and infrapatellar fat pad (IFP). 26 The IFP can be opportunistically harvested during routine surgical procedures such as knee arthroplasty or arthroscopy, and is known to have high chondrogenic potential. 27 Bone marrow-derived stem cells share common properties with hADSCs but are limited due to low tissue availability and cell number, and inferior chondrogenic potential compared with IFP-derived stem cells. 28

| Photo-crosslinkable hydrogels and crosslinking process
The ideal hydrogel for cartilage regeneration is one that resembles the natural extracellular matrix of cells to support cell survival and differentiation and thus to form functional articular tissue. 36 Natural hydrogels such as gelatin display high biocompatibility and biodegradability. 37 Gelatin is composed of hydrolyzed collagen and retains abundant Arginine-Glycine-Aspartate sequence motifs which serve as cell attachment sites, and it contains matrix metalloproteinase sensitive degradation sites. These bioactive motifs facilitate cell adhesion, proliferation, and differentiation via integrin-mediated cell adhesion and cellmediated enzymatic degradation. However, in order to be used as a biomaterial, gelatin needs to be modified to gain irreversible crosslinking, necessary strength, and precise mechanical tunability. 38 This problem can be overcome by the addition of functional groups such as methacrylate/methacrylamide, which can be crosslinked after the activation of a PI, to form gelatin methacryloyl (GelMa) (Figure 2). This achieves F I G U R E 1 Schematic representation of the photo-crosslinking process of a hydrogel laden with cells. Although the bio-ink is extruded in gel form, it hardens following exposure to UV light (A). The photoinitiator molecule (eg, lithium phenyl-2,4,6-trimethylbenzoylphosphinate, LAP) mixed within the hydrogel (B) is cleaved and forms two free radicals (C), which are responsible for the formation of highly resistant covalent bonds between polymer chains in the hydrogel, but at the same time can lead to DNA damage and cell death Synthetic hydrogels such as polyethylene glycol (PEG) exhibit limited biocompatibility compared with natural hydrogels but are useful for their superior mechanical properties. 43 Introduction of acrylate functional groups forming polymerizable PEG diacrylate (PEGDA) as well as the addition of other moieties to improve the biological properties has propelled PEGDA to become a popular hydrogel choice in cartilage tissue engineering. [44][45][46] Hydrogel crosslinking can be achieved via a photoinitiated but also enzymatic system. Enzymatic crosslinking commonly utilizes transglutaminases, tyrosinases, or peroxidases to catalyze the formation of highly resistant covalent bonds between polymer chains. 47 The main drawbacks are the instability of some of the enzyme types, especially transglutaminases and tyrosinases, and the limited mechanical properties of the gels formed. As described above, photo-crosslinking has been demonstrated to provide excellent temporal and spatial control over the process hence allows greater control of the mechanical properties of the resultant matrix compared with the indirect enzymatic process. 48 The reaction involves the usage of light of a specific wavelength to strike the PI, which in turn is cleaved into two free radicals ( Figure 3).
One or both of these free radicals then radicalize a nearby reactive functional group (methacrylate or methacrylate) which propagates the polymerization chain reaction.
As the reaction proceeds the number of the crosslinks in the system increases exponentially resulting in a biopolymer network linked through polymethacryloyl chains. 43 The degree of crosslinking and therefore the degree of mechanical stiffness is a result of PI type and concentration, light intensity, wavelength, exposure time, and degree of methacrylation. 50 Although the photo-crosslinking process is efficient and is somewhat controllable, it presents three major potential sources of cellular toxicity: generation of free radicals, exposure to the PI molecule itself, and exposure to light. Ultimately, the toxicity introduced by the photo-crosslinking process must be minimized to achieve a crosslinked hydrogel with optimal mechanical stiffness, maximal cell viability, and minimal DNA damage.
F I G U R E 2 Schematic representation of the methacrylation of gelatin to form GelMa. Functional side chains of the GelMa molecule can be photo-crosslinked by adding a specific photoinitiator (PI) and light irradiation, to form a network contributing to the stiffness of the resulting scaffold. Source: Adapted from Caballero Aguilar et al, 39 reproduced with permission of Royal Society of Chemistry via Copyright Clearance Center F I G U R E 3 Examples of two commonly used PI molecules: Irgacure 2959 (2-hydroxy-4 0 -(2-hydroxyethoxy)-2-methylpropiophenone) and LAP (lithium phenyl-2,4,6-trimethylbenzoylphosphinate), generating two free radicals after cleavage with UV light (hv). These two free radicals will then attack functional groups on the hydrogel to initiate the polymerization reaction. Source: Adapted from Fairbanks et al 49

| Light source and irradiation
The photo-crosslinking process commonly employs an ultraviolet (UV) light source which is by itself a potential source of cytotoxicity, due to UV induced apoptosis and most importantly, genotoxicity when DNA damaged cells are not eliminated. Long-wave UV (A, 315-400 nm) is widely accepted as a mutagen owing to its ability to induce cellular DNA damage, and shortwave UV (B, 280-315 nm) irradiation can lead to DNA base lesions such as cyclobutane pyrimidine dimers (CPDs), and pyrimidine 6-4 pyrimidone photo-products. 51 Higher wavelength UV-A (315-400 nm) induced cytotoxicity occurs mostly via indirect mechanisms, whereby cellular chromophores act as photo-sensitizers to generate reactive oxygen species (ROS) which causes insult to proteins, lipids and DNA, the main lesion being the oxidized base 8-oxo-7,8-dihydroguanine (8-oxoG). 52,53 Similarly to CPDs, 8-oxoG can pair with adenine and cause a guanine:cytosine to thymine:adenine transversion, but can also result in DSBs if inserted during DNA replication. 54 Other studies identified an action spectra to determine cell killing and mutations by monochromatic ultraviolet and visible radiations (254-434 nm) in human epithelial cells. 55 More recently, the cytotoxicity of UV-A1 radiation was tested in human mesenchymal stem cells and data show that a prolonged 2-hour exposure to high intensity (370 ± 5 nm; 788 kJ/m 2 ) in the absence of any PI, results in a significant reduction of cell viability of up to 50% compared with cells exposed to visible light only. 56 Nevertheless, it has also been demonstrated that low-dose and long-wave UV-A light do not affect their gene expression, making ideal cells candidate for their low susceptibility to cytotoxic and genotoxic effects derived from a photo-crosslinking reaction. 57 Double-stranded breaks (DSBs), which are considered the most deleterious type of DNA damage, can be caused as a direct result of these lesions and indirectly through the production ROS. 58 Inbuilt cellular mechanisms to avoid mutagenesis include DNA repair, apoptosis or cell cycle arrest. However, should these mechanisms fail, these genomic lesions can result in tumor formation 59,60 (Figure 4).

| Photoinitiator molecules
The optimal PI is one that generates phase transformation of hydrogels from gel to solid to withstand required compressive forces for the target tissue while remaining minimally toxic to the encapsulated cells. The PI itself has intrinsic toxic effects although this varies between PIs and can be minimized by selecting the lowest practicable concentration; although reducing the PI concentration necessitates an increase to the light exposure time or intensity, it needs to achieve equivalent mechanical properties. 50 The toxicity of PI molecule relates to its chemical structure, especially its hydrophobicity, which increases its potential to cross the cellular membrane. A comparative study between three PIs, Irgacure 2959 alone, but not to UV light itself ( Figure 5). This was one of the main aspects that prompted the exploration of protective elements for the cells, such as, for example, their encapsulation in hydrogel or a spatial separation from the PI (ie, coaxial printing techniques that can separate cells from direct contact with the PI). These strategies have been demonstrated to significantly reduce the cytotoxic effect of these chemicals.

| Free radicals
As discussed above, photoinduced free radicals are highly reactive species, chosen for their ability to trigger a radical polymerization reaction. However, they can also interact with double bonds within cellular components such as membranes, proteins, and DNA, thus threaten cell viability, metabolism, and DNA integrity.
The toxicity of free radical can arise through direct effects, as well as indirect effects, such as the formation of ROS upon reaction of a free radical with the environmental oxygen. 67 Oxidative degradation of lipids which constitute the cell and mitochondrial membranes produce toxic aldehyde end products such as 4-hydroxynonenal (4-HNE).
4-HNE is particularly cytotoxic and mediates this effect through depletion of glutathione, a potent antioxidant that has a role in mitochondrial redox reactions, and the formation of mitochondrial protein adducts. 68 The subsequent disruption of mitochondrial function activates intrinsic apoptotic pathways, although it should be noted that at very high concentrations, acute cell death by necrosis can occur. 69 In terms of genotoxicity, free radical-induced DNA damage can take the form of base lesions, damage to the sugar moiety, tandem lesions, DNA-protein crosslinks, single, and double strand breaks. 70 Of the bases, guanine is most susceptible to oxidative stress leading most commonly to the formation of 8-oxoG lesions as discussed above, but also to other products such as imidazolone and spirodihydantoin. 71  Finally, during DNA replication, the presence of single strand breaks or other DNA lesions such as interstrand crosslinks or DNA-protein crosslinks can hinder the normal replicative process leading to a collapse of the replication fork and DSBs formation. 76 As such, the formation of less harmful lesions such as base lesions described in the paragraph above have the potential to form these highly risky DSBs.
The reduction in cell viability due to cytotoxic effects of free radical photoinitiation has been well characterized across different PI

types. Fedorovich et al demonstrated that the combination of UV light
with Irgacure 2959 resulted in the highest cytotoxicity compared with the two modalities alone. 20 Similar results were generated with LAP and RB indicating that the PI toxicity is drastically exacerbated by photoactivation. 62,65,66 More concerning than cell death is the damage to cells that survive despite free radical induced toxicity. Evidently, the significant drop in cell metabolic activity immediately after high intensity UV crosslinking, and the progressive decline over the following week, suggests that engendered free radicals from light-induced PI degradation causes irreparable damage to cellular processes. 50

O'Connell et al demonstrated in
fact that although metabolic activity declined, cell survival remained high (>90%) which raises concern that damaged cells could contain DNA-base lesions. As discussed earlier, depending on the genes where these lesions occur, tumor formation within the generated bioscaffold could result, rendering this technology unsafe for clinical application.  thermofisher.com/au/en/home/brands/molecular-probes/key-molecular-probes-products/live-dead-viability-brand-page.html) utilizes green- foregoes 3D spatial information afforded by microscopy. 92,93 An alternative genotoxicity assay which could be adapted to analyze bioscaffolds is the comet assay, a microgel electrophoresis technique where cells with single and DSB migrate toward the anode in the shape of a comet. 94 The degree of migration or length of the comet tail is proportional to the degree of DNA damage. This assay has been used across various cell types including mesenchymal stem cells. Although this technique has been used for analysis of cells in tissue engineered skin, 95 it has not been used in photoencapsulated cells and would require a protocol to extract the cells from the scaffold without inflicting further DNA damage. As such, further investigation to develop a protocol for the detection of DSBs in cells within bioscaffolds is required.

| STRATEGIES TO REDUCE CYTOTOXICITY
Both cytotoxicity and genotoxicity can be minimized by targeting the choice and the concentration of the PI or alternatively by the separation of cells from free radicals.

| Photoinitiator choice and concentration
PI choice is typically based upon intrinsic properties such as their absorption spectra and its solubility limit in water. The efficiency of a PI at a wavelength of light is then determined by (a) its molar absorptivity at that wavelength, (b) the quantum yield of photolysis (the fraction of absorbed photons which produce free radicals), and (c) the PI efficiency (the ratio of initiation events to radicals generated). Unfortunately, a lack of such intrinsic data in the literature to date has severely hampered a rational comparison of relevant PIs.
The most commonly used PI in GelMa crosslinking is Irgacure 2959 which has been acclaimed for its water solubility and relatively low cytotoxicity. 61 Irgacure 2959 has a peak absorption of 260 nm and although it can be activated at a higher wavelength of 365 nm, it has a cross-linking time that is too long for clinical application as the extended period of UV exposure poses a risk of cytotoxicity. 96 Across the literature, there are reports of PI concentrations ranging from 0.05% to 0.5% w/v with little consensus on the optimal concentration for cartilage tissue engineering. 61,97 Arguably, the optimal concentration will vary for the hydrogel being cross-linked and the susceptibility of the cell type. Furthermore, Bartnikowski

| Coaxial extrusion
Given the detrimental effects of free radicals engendered by the photo-crosslinking process, reducing cell exposure to the PI and its activation products is a possible solution to increase cell viability and protect the DNA molecule from genotoxic effects of free radicals.
Coaxial extrusion allows the deposition of cells and the cross-linking solution through separate internal and external needles, respectively.
This compartmentalization of cells within an inner, non-crosslinked hydrogel "Core" surrounded by a photo-crosslinked "Shell" is aimed at protecting vulnerable cells from the toxic effects intrinsic to the PI molecule and more importantly, from free radicals. The shielding effect of cells by the shell compartment increases the scope of materials that can be used for fabricating tissue constructs with a high cell viability. In cartilage tissue engineering, the coaxial method was found to be superior to the monoaxial configuration where cells are embedded throughout the hydrogel and exposed to the toxic photoinitiation process triggered by a fast high irradiance exposure to UV light. In Duchi et al, the cytotoxicity of bioprinted hADSCs upon 10 second irradiation with 365 nm at 700 mW/cm 2 was found to be negligible, when cells were segregated from the PI, and the viability of UV only irradiated cells was comparable to the nonirradiated control group throughout 7 days. 62 Moreover, high levels of cell survival immediately after printing and crosslinking with an increase in cell number 10 days following extrusion.
In contrast monoaxial printed cells experienced a significant downturn in survival following photoinitiation and a further decrease over time.

| CONCLUSIONS
The hydrogel material and the other additives required to harden the hydrogel itself constitutes the barriers to successful clinical translation of articular cartilage repair techniques using bioscaffolds. The photocrosslinking process is an efficient way to achieve a precise spatial and temporal hardening of the hydrogels, but, if not optimized, becomes a major source of cytotoxicity and genotoxicity within the process. These

CONFLICT OF INTEREST
The authors indicated no potential conflicts of interest.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.