RECENT ADVANCES IN HYDROGELS FOR CARTILAGE TISSUE ENGINEERING

Abstract

Articular cartilage is a load-bearing tissue that lines the surface of bones in diarthrodial joints. Unfortunately, this avascular tissue has a limited capacity for intrinsic repair. Treatment options for articular cartilage defects include microfracture and arthroplasty; however, these strategies fail to generate tissue that adequately restores damaged cartilage. Limitations of current treatments for cartilage defects have prompted the field of cartilage tissue engineering, which seeks to integrate engineering and biological principles to promote the growth of new cartilage to replace damaged tissue. To-date, a wide range of scaffolds and cell sources have emerged towards cartilage tissue engineering, with a focus on recapitulating microenvironments present during development or in adult tissue to induce the formation of cartilaginous constructs with biochemical and mechanical properties of native tissue. Hydrogels have emerged as a promising scaffold due to the wide range of properties that are possible and the ability to entrap cells within the material. Towards improving cartilage repair, hydrogel design has advanced in recent years to improve their utility. Some of these advances include the development of improved network crosslinking (e.g., double-networks), new techniques to process hydrogels (e.g., 3D printing), and the better incorporation of biological signals (e.g., controlled release). This review summarizes these innovative approaches to engineer hydrogels towards cartilage repair, with an eye towards eventual clinical translation.

Introduction

In simplistic terms, articular cartilage is a tissue consisting of a single cell type (chondrocyte) embedded within an extracellular matrix (ECM). However, the structure is more complex and includes three depth-dependent layers: the superficial zone, the middle zone, and the deep zone, with changes in ECM content, structure and chondrocyte behavior with depth. In the middle zone – the largest region, cartilage tissue is stiff, avascular, and features a low density of rounded chondrocytes surrounded by an ECM consisting of glycosaminoglycans (GAGs) and type II collagen (Vunjak-Novakovic and Freed, 1998; Wong and Carter, 2003). Cartilage damage due to trauma typically begins as a focal defect, which later progresses to a full-thickness defect once the lesion comes into contact with surrounding bone. Due to low cellularity, low vascularization, minimal proliferative capacity of residing chondrocytes, and low cell migration to areas of damage, articular cartilage is intrinsically unable to repair itself (Ahmed and Hincke, 2010).

Current strategies to repair focal and full-thickness cartilage defects have evolved from surgery aimed to induce endogenous repair mechanisms (e.g., microfracture), towards osteochondral transplantation, and more recently to cell-based repair techniques, such as autologous chondrocyte implantation (ACI) (Brittberg, 2008; Goldring, 2006; Hunziker, 2002). ACI is a procedure in which patient chondrocytes are implanted into a debrided cartilage lesion (Peterson et al., 2010). Both microfracture and ACI lack 3D scaffolds to define where cells reside and to instruct matrix formation; however, there have been advances in these therapies in recent years. Matrix-induced autologous chondrocyte implantation (MACI) is a cell-based therapy that supplements ACI by providing chondrocytes with a supportive scaffold material for matrix formation (Makris et al., 2015). Autologous matrix-induced chondrogenesis (AMIC) on the other hand supplements microfracture with an acellular scaffold in the lesion in an effort to increase mechanical stability at the site, cell migration, and cartilage production (Benthien and Behrens, 2011). Despite promising results, these clinical techniques are not without their limitations, including the surgical procedures and the quality of the formed cartilage.

Towards further expanding on the use of biomaterials in cartilage repair, hydrogels have gained a great deal of interest in cartilage tissue engineering (Balakrishnan and Banerjee, 2011). Hydrogels are 3-dimensional (3D) polymer networks that are highly swollen and porous on the molecular scale, allowing for the diffusion of various solutes and nutrients. Their fabrication can be cell amenable, allowing for the encapsulation of different cell types (e.g., chondrocytes, stem cells). Lastly, an assortment of hydrogel parameters can also be tuned, including polymer chemistry, crosslinking density, degradation, mechanical properties, and release kinetics of biochemical factors, towards improving their utility in tissue repair (Spiller et al., 2011). Numerous studies have characterized the effects of hydrogel crosslinking – tuned through parameters like macromer concentration - on neocartilage formation (Chung et al., 2006). These studies have shown that matrix distribution can be hindered when a hydrogel with high crosslinking is used. Also, differences in the composition of hydrogels have been investigated towards cartilage formation. For example, several groups report that type II collagen hydrogels induce a higher amount of chondrogenic outputs in comparison to type I collagen hydrogels with similar properties (Lu et al., 2010; Nehrer et al., 1997).

To serve as constructs to replace articular cartilage, hydrogels have been used as either acellular scaffolds or as cell-laden biomaterials (Spiller et al., 2011). For either strategy, hydrogel implants must integrate with surrounding tissue and support the production of cartilage. Acellular constructs are almost exclusively assayed in vivo, where the goal is structural and mechanical properties similar to native cartilage while allowing efficient load transfer or to recruit cells for cartilage repair (Kobayashi and Oka, 2004; Stammen et al., 2001). Cell-laden hydrogels present residing cells with either developmental cues to trigger chondrogenesis, or microenvironmental cues that mimic native cartilage to maintain a chondrogenic phenotype and encourage matrix formation. Hydrogels are advancing to present these signals either as bound to the polymer or through the controlled release of molecules. Also, these cultures may be performed in vitro, such as with bioreactors, or may be implanted directly for tissue growth.

With the rapid advances being made towards hydrogels for cartilage tissue engineering, the goal of this review is to cover current and emerging developments in hydrogel design for cartilage repair. It is not meant to be comprehensive, but rather highlight representative advances in the field in recent years with a particular focus on the hydrogel design. Specific areas include the use of unique crosslinking to produce hydrogels with high mechanical properties, the improved processing of hydrogels into macroporous structures, and the incorporation of biological signals to improve cell behavior.

A. Improvements in hydrogel structure

One of the key design criteria for hydrogels towards cartilage regeneration is mechanical integrity. Traditional hydrogel designs based on networks of a single polymer generally result in hydrogel constructs with mechanical properties far inferior to those of native cartilage. Also, the increase in mechanics through increased crosslink density may compromise the viability of encapsulated cells through reduced diffusion. Towards increasing the mechanical properties of hydrogels to approach those of hyaline cartilage, the focus is shifting from conventional hydrogels that use a single polymer for hydrogel fabrication (Fig. 1a), to more complex hydrogel systems with mixtures of multiple polymers, often including two or more independent networks. These systems not only often achieve higher mechanical properties than with networks of single polymers, but may exhibit superior integration with surrounding tissue in vivo. In this section we examine recent advances within the framework of different network types (Fig. 1).

Fig. 1.

Schematic depicting different designs utilized in hydrogels, from (a) traditional single polymer networks to those that include (b,c) multiple networks and (d-f) mixtures of polymers. Double networks may be linked together, but this is not a requirement. Generally, the network design controls properties such as mechanics and degradation.

A1. Interpenetrating Networks

Hydrogels based on interpenetrating networks (IPNs) are comprised of two or more separate crosslinked networks not covalently bound to each other, but rather partially intertwined such that chemical bonds have to be broken to separate the components that make up the networks (Fig. 1b). Consequentially, the mechanical properties of IPN hydrogels tend to be greater than those created with individual component networks, which makes them appealing for cartilage tissue engineering applications. In recent years, the design of IPNs for hydrogel formation has turned to the incorporation of two or more additional networks to either better mimic the physical properties of native tissue or to recapitulate the presentation of bioactive cues available to cells in the hydrogel constructs.

Recent examples of IPNs include work done by Ingavle et al., in which they explored the incorporation of methacrylated chondroitin sulfate, a major component of the cartilage ECM, as the second network after diffusion into an existing agarose-PEGDA network (Ingavle et al., 2013). Incorporating chondroitin sulfate into the network significantly increased the viability of encapsulated chondrocytes for at least six weeks, and promoted greater biosynthesis of collagen and GAGs in the pericellular matrix (Ingavle et al., 2013). Reinforcing networks such as those provided by methacrylated hyaluronic acid (HA) within a fibrin hydrogel have been demonstrated as well (Fig. 2a) (Snyder et al., 2014). Here, the investigators found that the incorporation of a crosslinked HA network increased the compressive moduli of hydrogel constructs and modulated gene expression of encapsulated human mesenchymal stem cells (MSCs) by decreasing expression of dedifferentiation marker type I collagen and increasing expression of the chondrogenic transcription factor Sox9 (Snyder et al., 2014).

Fig. 2.

(a) SEM micrograph of a fibrinogen hydrogel with (bottom) and without (top) interpenetrating methacrylated HA network (Snyder et al., 2014). (b) MSCs encapsulated in gelatin methacrylamide hydrogels exhibit more aggrecan (green) with increasing concentrations of methacrylated HA, which acts as a dual network (Levett et al., 2014). (c) Hydrogel molecular structures and crosslinking schemes can become quite complex, as seen by this schematic depicting supramolecular hydrogels prepared with CB[6]-HA, DAH-HA, and drug conjugated Dexa-CB[6] (Jung et al., 2014).

In addition to IPNs of two networks, several groups have also investigated how tri-component IPNs increase mechanical properties and cartilage formation using both non-cell instructive (Dinescu et al., 2015; Liao et al., 2013) and cell-instructive materials (Guo et al., 2012). In a non-cell instructive scaffold comprised of gelatin, alginate, and polyacrylamide, Dinescu et al. reported higher cell proliferation, lower cytotoxicity, and greater chondrogenic gene expression (Sox9, type II collagen) of human adipose-derived stem cells than in one- or two-component hydrogels (Dinescu et al., 2015). The authors attributed these findings to the tri-component network’s ability to better retain its internal structure and porosity in long-term culture (Dinescu et al., 2015). Guo et al. further used this tri-component IPN concept to better approximate the diversity of biomaterials present in articular cartilage (Guo et al., 2012). The authors demonstrated the feasibility of generating a tri-component IPN using only cell-instructive components (collagen combined with methacrylate-modified chondroitin sulfate and HA) (Guo et al., 2012). By comparing to a semi-IPN using the same components (where the chondroitin sulfate and HA were not methacrylated) and by varying the extent of methacrylation, they demonstrated tunable mechanics with the tri-component IPNs, increased gene expression of chondrogenic markers (i.e., aggrecan, type II collagen, and Sox9), downregulation of dedifferentiation marker type I collagen, and increased collagen and GAG synthesis by encapsulated rabbit chondrocytes that correlated with the bulk mechanics of the IPN constructs (Guo et al., 2012).

A2. Semi-interpenetrating Networks

Unlike IPNs, semi-IPNs consist of a crosslinked network with either linear or branched polymers entrenched within the network (Fig. 1c). As such, the polymers and network can theoretically be separated from one another without breaking chemical bonds. One of the most common macromolecules distributed in these networks is HA. In a recent study, high molecular weight HA was distributed within injectable networks of photopolymerized methacrylated chitosan (MeGC) (Park et al., 2013). An optimum formulation and gelation protocol was developed, and the presence of entangled (but unbound) HA (350 kDa) incorporated into the chitosan network resulted in greater proteoglycan and GAG staining (as measured by alcian blue and safranin O) in the lacunae of chondrocytes encapsulated and cultured for up to 21 days (Park et al., 2013).

Another approach towards utilizing the properties of semi-IPNs is exploiting their ability to leach low-molecular weight macromolecules over time, as demonstrated by Skaalure et al. in a degradable poly(ethylene glycol) (PEG) hydrogel with low-molecular weight HA (29 kDa) (Skaalure et al., 2014). By comparing PEG networks infiltrated by either low- or high-molecular weight HA (2 MDa), the authors found that leaching of low molecular weight HA over a 28-day period led to the greatest soluble GAG deposition during construct maturation, while collagen biosynthesis was comparable to that generated in constructs with high molecular weight HA (which remained in the network during this period) (Skaalure et al., 2014). Conceptually similar approaches taken by Little et al. suggest that low-molecular weight HA and chondroitin sulfate in a fibrin-alginate hydrogel may exert post-transcriptional effects on collagen expression to influence construct composition as cells deposit matrix over time (Little et al., 2014).

A3. Double Networks

A double network consists of two networks with significantly different mechanical properties crosslinked together (Fig. 1d). Typically, the first network provides a rigid structure and the second network is ductile, resulting in greater toughness than the corresponding single networks alone would have achieved since the network can yield under mechanical load (Gong, 2010). These types of networks have gained interest in cartilage tissue engineering due to their superior mechanical properties over traditional hydrogels, including those that can approach the mechanics of native hyaline cartilage (Arnold et al., 2011). Double networks are also conceptually appealing since cartilage and other skeletal tissues inherently incorporate double networks into their ECM in order to achieve their robust mechanical properties (Arnold et al., 2011).

Commonly utilized double networks for cartilage tissue engineering consist of a combination of two acrylamide polymers, poly(2-acrylamido-2-methylpropanesulfonic acid) (PAMPs), poly(acrylamide) (PAAm), and/or poly(N,N-dimethylacrylamide) (PDMAAm) (Yasuda et al., 2009). Work with these double network hydrogels has recently extended to in vivo studies in both rabbit and sheep models of critical size defects, in which acellular plugs comprised of this double network implanted in defects resulted in enriched GAG and type 2 collagen content as measured by histological staining after 4 weeks (Kitamura et al., 2016). In all cases, the double-network plug was not infiltrated by cells, and instead seemed to serve as a support material for tissue regeneration above and around it.

A direct comparison between double network hydrogels and traditional single network hydrogels of either only PAMPS or PDMAAm demonstrated that double network hydrogel constructs resulted in superior cartilage repair by histological scoring (Ogawa et al., 2012). The tissue formed in the presence of double networks also exhibited similar gene expression profiles (Imabuchi et al., 2011) and tissue surface roughness to native cartilage, even as observed by confocal laser scanning microscopy. Furthermore, combinatorial therapies using double network hydrogels and intra-articular injections (e.g., HA) have shown potential in tissue quality as measured by histological scoring and the volume of cartilage generated by double network hydrogels implanted in cylindrical osteochondral defects in rabbit femoral trochlea at both 4 and 12 weeks—again, using acellular plugs of the PAMPS/PDMAAm double network (Fukui et al., 2014).

With this said, the application of novel and cell-instructive materials for double network construct design in cartilage tissue engineering has not been widely explored until recently. Polymers used for double networks are often non-degradable, and while this stability renders them as potentially useful materials for mechanical support in tissue defects, this stable and rigid network may limit cell infiltration and impede long-term matrix deposition and distribution. Despite these properties of double-network hydrogels, Levett et al. developed a double network consisting of a combination of gelatin-methacrylamide and hyaluronic acid-methacrylate (Levett et al., 2014). This system leveraged the relatively higher reactivity of methacrylate groups to create a double network, thus generating greater increases in compressive modulus and cartilage matrix component synthesis by encapsulated human chondrocytes (Levett et al., 2014). New hydrogels making use of this type of network are continuously being developed, such as an injectable and cytocompatible double network hydrogel based on HA that is formed through the combination of dynamic (i.e., guest host pairs) and stable (i.e., covalent) crosslinks (Rodell et al., 2016).

A4. Dual Networks

Unlike double networks that use two materials with different mechanical properties, dual networks are defined as two materials crosslinked together into the same network and with similar crosslinking mechanisms (Fig. 1e). Although dual networks do not possess the toughness of double networks, each material in dual networks can imbue other useful properties to the hydrogel. For instance, one material can enable effective integration with the surrounding tissue, while the other can attract cells and encourage migration into the hydrogel. Teixeira et al. utilized a dextran-tyramine and heparin-tyramine dual network hydrogel to encapsulate bovine chondrocytes in vitro and reported improved cell viability and proliferation (Moreira Teixeira et al., 2012). Additionally, they observed increased deposition of chondroitin sulfate and collagen compared to cells in a single component dextran-tyramine hydrogel (Moreira Teixeira et al., 2012). Jin et al. reported similar findings using the same dual network polymers ex vivo with bovine articular cartilage explants (0.5 × 0.5 × 0.2 cm) to determine tissue-adhesion of the hydrogel constructs (Jin et al., 2011). In a more recent study, in situ forming ethylenediamino-functionalized HA (HA-EDA) and divinylsulfone-inulin (INU-DV) dual network hydrogels were used to encapsulate chondrocytes, with the first material selected due to its native presence in articular cartilage and the second for its biocompatibility and status as a widely-used FDA-approved polymer (Palumbo et al., 2015).

Along the same lines of using HA as the biologically relevant component, Pirinen et al. developed a dual network hydrogel using high molecular weight HA (>1,600 kDa) and low molecular weight PVA (27 kDa) functionalized with aldehydes and primary amines for crosslinking (Pirinen et al., 2015). This dual hydrogel system was amenable to tunable swelling properties by varying the size of the smaller PVA component, and encapsulation of bovine knee chondrocytes showed favorable cell viability for at least 2 weeks in culture (Pirinen et al., 2015). Similar hydrogel systems (e.g., with HA and gelatin) have also produced favorable results in construct formation and maturation. Levett et al. reported that the addition of HA as the second component resulted in retention of a rounded chondrocyte morphology, greater aggrecan deposition and compressive moduli, and suppression of type I collagen accumulation over the course of a 56-day study (Fig. 2b) (Levett et al., 2014).

A5. Guest-host networks

Injectable hydrogels are an attractive approach for cartilage tissue engineering since they can be delivered via a direct injection or arthroscopically (Guvendiren et al., 2012). Shear-thinning hydrogels are an attractive method towards this as a hydrogel can be disrupted during the injection process and then self-heal upon injection. One shear-thinning HA system developed is based on the reversible bonds between guest (adamantane modified HA) and host (β-cyclodextrin modified HA) polymers, termed guest-host interactions (Fig. 1f) (Rodell et al., 2013). These types of interactions rapidly form hydrogels via non-covalent interactions in the guest-host bonds. In addition to cyclodextrin-adamantane guest-host pairs, guest-host interactions between cucurbit[6]uril and diaminohexane have also been investigated (Jung et al., 2014). Both of these interactions are reversible and give rise to shear-thinning and remoldable hydrogels.

Wei et al. described the assembly of a two-component guest-host hydrogel with adamantane-functionalized HA as the first component and monoacrylated β-cyclodextrin (assembled into clusters by photopolymerization) as the second component (Wei et al., 2016). The hydrogels formed from these components exhibited nanoclustered interactions that allowed for robust drying and re-swelling without changes in water content or shape; they also enabled the robust chondrogenesis of encapsulated human MSCs with greater collagen deposition compared to a covalently crosslinked methacrylated HA network, although the differences in biophysical properties and network structure limit a direct comparison (Wei et al., 2016). Jung et al. demonstrated the use of guest-host interactions between curcubit[6]uril and diaminohexane to create hydrogels for the encapsulation of human MSCs as well as simultaneous release of dexamethasone for chondrogenesis. Cells in this injectable guest-host system, especially with the additional function of controlled dexamethasone release, exhibited increased cell proliferation, GAG synthesis, chondrogenic gene expression (i.e., type II collagen, COMP, aggrecan, Sox9), and neocartilage formation in an in vivo subcutaneous study (Fig. 2c) (Jung et al., 2014).

B. Advances in the processing of hydrogel scaffolds

Parallel advances in manufacturing technologies and material design have paved the way for new possibilities in how hydrogels can be manipulated and formed into 3D macroporous scaffolds. Architectures that aid in the viability, retention, and chondrogenic induction of cells have been achieved using several techniques including 3D printing, spinning, and doping hydrogels with degradable porogens or microspheres (Fig. 3). Due to their ability to generate unique microenvironments for cells and better recapitulate tissue structure on the macroscale, these techniques have become more prominent in tissue engineering in general and in cartilage engineering in particular. Recent studies making use of some of these more widely adopted methods are outlined in the subsections below.

Fig. 3.

Overview of different macroporous scaffold structures used for cartilage tissue engineering. To create hydrogel fibers, 3D printing and spinning techniques have been employed (blue box). In contrast, porous hydrogels and complementary microsphere hydrogels can also be fabricated (red box). To recapitulate native cartilage structures (e.g., different regions of cartilage, the osteochondral interface), multi-layer hydrogels incorporating several fabrication techniques can also be utilized (bottom).

B1. Hydrogel fibers

Fiber fabrication techniques at length scales ranging from 150 nm to 1600 μm are typically generated by either spinning or 3D printing. The resultant fibrous networks offer a number of potential advantages, from more facile cell infiltration and greater hydrogel surface area (and thus improved diffusion of nutrients and other soluble factors) to increased mechanical properties. The techniques involve both the encapsulation of cells within macro-fibers to the seeding of meshes formed from nanofibers. Numerous groups have engineered hydrogel fibers using material composites and copolymers that allow for subsequent photo and ionic crosslinking, resulting in a wide spectrum of material properties.

3D printing

3D printing, or additive manufacturing, typically involves the building of 3D structures layer-by-layer with the controlled deposition of materials. A chief concern with 3D printing of cell-laden hydrogels is the optimization of print parameters and control over material properties to enable efficient printing while preserving cell viability and phenotype. Schuurman et al. adapted a methacrylated gelatin formulation by introducing HA to increase the viscosity of the precursor material and demonstrated print fidelity and chondrocyte viability with post-print photocrosslinking (Schuurman et al., 2013). Conceptually similar fabrication techniques have been explored with other methods, including the use of ionically crosslinkable materials in lieu of photocrosslinking. For example, Fedorovich et al. used chondrocyte-laden alginate crosslinked with calcium chloride (Fedorovich et al., 2012). In this case, constructs were printed with varying nozzle diameter, print speed, and fiber orientation and then cultured or implanted subcutaneously to demonstrate cell viability and tissue formation (Fedorovich et al., 2012). Building upon this approach, Markstedt et al. engineered additional shear-thinning properties into this ionically crosslinkable network by incorporating nanofibrillar cellulose into the alginate precursors to give the material additional shear-thinning properties (Markstedt et al., 2015). Cui et al. examined the timing of crosslinking in this context as well, demonstrating direct 3D printing into osteochondral defects using a PEG dimethacrylate (PEGMA) hydrogel in an explant model (Cui et al., 2012). The authors found greater chondrocyte viability and more uniform cell distribution with tandem (simultaneous) crosslinking, where crosslinking occurs at the same time as gel extrusion, rather than with traditional post-print crosslinking in which the construct can be crosslinked after printing (Cui et al., 2012).

Thermally responsive hydrogels use temperature to modulate their gelation behavior, allowing for the transition from liquid to a hydrogel as a function of temperature exclusively. This is particularly appealing in cartilage tissue engineering, since the change from ambient temperature to physiological conditions can induce rapid gel formation (Klouda and Mikos, 2008) or stabilize materials in 3D printing processes. A common composite choice for thermoresponsive “bioinks” consists of HA or chondroitin sulfate mixed with a thermoresponsive polymer. Numerous groups have tested different thermoresponsive materials in this regard, from poly(N-isopropylacrylamide) (pNIPAAM) (Kesti et al., 2015) to triblock copolymers composed of PEG linked to N-(2-hydroxypropyl) methacrylamide (HPMA) (Boere et al., 2015).

In exploring possible effects of thermoresponsive material structures at the nanoscale, Muller et al. showed that nanostructuring a thermoresponsive hydrogel allowed for high chondrocyte viability in a printed multilayer mesh and other more physiologically-relevant constructs including an ear (Fig. 4e) and a sheep meniscus (Fig. 4f-g) (Muller et al., 2015). In this case, the authors incorporated both unmodified and acrylated Pluronic, crosslinked, and then subsequently washed out the unmodified fraction. In each case mentioned, methacrylated HA or chondroitin sulfate was either used as a crosslinker or as an independent reinforcing network to compensate for the generally poor mechanics of the thermoresponsive component.

Fig. 4.

SEM micrographs of PBLG microsphere hydrogels fabricated at a gelatin concentration of (a,b) 1.9 % and (c,d) 3.25 % (Fang et al., 2015). 3D printing was used to print (e) a human ear and (f,g) sheep meniscus, as seen from different angles with Ink8020 after crosslinking (Muller et al., 2015).

Spinning hydrogel fibers

A number of techniques that use spinning technology have been explored towards fabricating hydrogel fibers. A wet-spinning technique was introduced by Han and colleagues in which an 8-arm methacrylated PEG precursor was injected through a syringe pump into a tris(2-aminoethyl) amine (TAEA) bath (Han et al., 2014). Subsequent crosslinking yielded microribbons that could be further crosslinked together to form hydrogels with complex geometries at both the micro and macro scales, and these hydrogels supported adipose-derived stem cell viability as well as proliferation and spreading under various conditions with the potential of recapitulating the laminar matrix organization of cartilage tissue (Han et al., 2014).

Electrospinning has also been used to generate fibrous constructs from methacrylated HA that supported human MSC interactions and chondrogenesis, without outcomes dependent on the hydrogel fiber properties such as mechanics and adhesion (Kim et al., 2013). Multi-polymer fibrous scaffolds were fabricated using HA hydrogels and implanted in combination with microfracture for tissue repair in a mini-pig model. The multi-polymer design allowed for stable fibers to maintain the fiber structure and degradable fibers to release chondroinductive factors. Here, the transforming growth factor beta 3 (TGF-β3) spun into degradable HA fibers was verified to be active upon release, and in vivo it resulted in improved histological scores and type 2 collagen content in the repaired defects (Kim et al., 2015). Fiber spinning and printing have also been combined in cell-laden constructs, where, for instance, alternating rounds of electrospinning of polycaprolactone (PCL) fibers and printing of fibrin-collagen hydrogels were performed to yield scaffolds with greater mechanical properties and improved histological staining of GAGs and type 2 collagen in 8-week subcutaneous implant studies in mice (Xu et al., 2013).

B2. Porous hydrogels

Porous hydrogels

While hydrogel fibers and scaffolds comprised of fibrous networks permit the incorporation of pores of varying shapes and sizes, several other methods for generating porous hydrogel constructs within the context of cartilage tissue engineering also exist. These may be advantageous in some cases since they do not need specialized printing or spinning equipment and may produce scaffolds more rapidly. More recently, the design of porous hydrogels has turned to dynamic control of macroporosity as well as dynamic control of bulk material properties.

Han et al. developed stimuli-responsive porogens from alginate, gelatin, and HA that respond to chelation, temperature, and enzymatic activity (Han et al., 2013). Here, the alginate porogen was also used to deliver chondrocytes into the hydrogel, and treatment with changes in temperature, EDTA, or hyaluronidase activity released chondrocytes from the alginate components and subsequently increased hydrogel macroporosity (Han et al., 2013). While this controlled degree of macroporosity affected bulk mechanical properties, approaches have also been developed where mechanics can be dynamically increased with photocrosslinking, such as from 2.6 to 12.5 kPa, without changing porosity (Marklein et al., 2012). In this case, across the range of mechanics tested, chondrogenic gene expression by human MSCs in porous constructs was slightly upregulated even in the absence of chondrogenic media (Marklein et al., 2012). In a different approach, Ahrem et al. treated whole bacterial nanocellulose hydrogels with 3D laser perforation to produce a porous construct that promoted chondrocyte ingrowth and proliferation (Ahrem et al., 2014). Control of hydrogel porosity and functionalization of pore interiors with relevant molecules such as type II collagen have also been explored in tandem with the use of stimuli-responsive materials. For example, Almeida et al. recently showed that “shape-memory” properties can be attained by covalent crosslinking of alginate scaffolds by making use of carbodiimide chemistry, which along with type II collagen incorporation led to higher sulfated GAG (sGAG) and collagen production in comparison to scaffolds functionalized with type I collagen (Almeida et al., 2016).

Microsphere hydrogels

As a complement or alternative to porous bulk hydrogels, scaffolds comprised of microspheres from a range of materials have also been developed. Such scaffolds have been studied most often as delivery vehicles or depots for small molecules or proteins, but they have also been used as potential cell carriers (i.e., cell microencapsulations). In one case, a xanthan gum derivative was selected as the encapsulation material because of its established safety record in pharmaceuticals, food, and cosmetics, as well as its observed protective effects on joint cartilage via an intra-articular injection (Hamcerencu et al., 2007; Katzbauer, 1998). Murine chondrocytes were viable in these microgels as observed for up to 21 days (Mendes et al., 2012). Another interesting scaffold comprised of the synthetic peptide poly(g-benzyl-L-glutamate) (PBLG) achieved tunable porosity in microspheres by varying amounts of gelatin porogen followed by its removal, producing spherical hydrogel scaffolds that could be seeded with chondrocytes throughout their interior (Fig. 4a-d); in vitro, they retained chondrocyte roundness after seeding and supported ECM deposition in culture for 3 to 7 days, and when injected subcutaneously in vivo they formed tissue with improved histological scoring and type 2 collagen content compared to cell-only injections (Fang et al., 2015).

B3. Multilayered scaffolds

The final scale of architectural complexity in cartilage engineering is one that attempts to recapitulate the zonal architecture of cartilage in the joint. Regardless of the particular technique used to deposit or arrange the material, numerous groups have worked to establish two-to-three-layered hydrogel constructs with differential characteristics to mimic the zones likely damaged by osteochondral defects. Feasibility of a 3D printed bilayer hydrogel construct has been demonstrated, both by 3D extrusion of material (Cui et al., 2012) and by projection stereolithography (Sun et al., 2015). As for trilayer hydrogels, Nguyen et al. described a PEG-based hydrogel with the superficial layer containing chondroitin sulfate and an matrix metalloproteinase (MMP)-sensitive peptide, the middle layer containing PEG and chondroitin sulfate, and the deep layer composed of PEG and HA (Nguyen et al., 2011). In such a construct the investigators showed that with a single stem cell population evenly distributed in all layers, type II collagen deposition decreased gradually from superficial to deep layers along with an increase in type X collagen and proteoglycans, resulting in a gradient of compressive modulus across the construct.

With all of this said, there are a number of different approaches to zonal design of hydrogel constructs, with alternatives including the use of different cell types (e.g., chondrocytes, osteogenic progenitors) (Fedorovich et al., 2012) or culture conditions in each layer rather than substantially modifying the material characteristics of the layer itself. It remains to be seen how advances in the design of hydrogels with this degree of control over architecture will compare with other diverse strategies for hydrogel engineering towards cartilage repair.

C. Controlled presentation and delivery of biochemical factors

Cartilage formation is regulated by the dynamic spatial and temporal presentation of an assortment of biochemical factors that either interact with receptors at the cell surface, or are internalized by cells, resulting in the induction of chondrogenesis. Thus, hydrogels are being designed with both biomolecules that are directly tethered to the hydrogel and those that are released to encapsulated or surrounding cells. This control over biomolecule spatial introduction and timing can aid in the production of heterogeneous constructs to mimic developing or adult tissues and to enhance the dynamic processes of differentiation.

The spatial and temporal presentation of biochemical factors is extremely important, since uncontrolled exposure may not only reduce effects on chondrogenesis and cartilage formation, but can also induce adverse effects. For example, supra-physiological levels of TGF-β delivered via injections to murine knee joints resulted in a decrease in proteoglycan synthesis, synovial fibrosis, and endochondral ossification (Bakker et al., 2001; van Beuningen et al., 2000). To regulate the dose and timing of biochemical factors that interact with cell surface receptors, several techniques including protease-degradable tethers and presenting factors in their latent form (Place et al., 2012; Re’em et al., 2012) have been recently proposed. To achieve controlled release of cartilage-inducing factors that are internalized by cells, novel design techniques towards the improved design of carrier vehicles (e.g., microparticles (MPs) or nanoparticles (NPs)) loaded with biochemical factors have been proposed. In this section we will highlight several recent advances of biochemical factor presentation using these mechanisms and their in vivo and in vitro cartilage regeneration potential using cell-laden and acellular hydrogels (Fig. 5).

Fig. 5.

Overview of controlled presentation of biochemical factors. These include cell-cell and cell-matrix interactions, as well as growth factors and other molecules that can be either tethered to the hydrogels via affinity or heparin binding, or encapsulated in MPs and NPs for controlled release.

C1. Covalent conjugation and affinity binding of biochemical factors

In order to present cells with biochemical cues that interact with integrins or receptors at the cell surface, these signals are typically incorporated into the hydrogel itself, either by covalent tethering or affinity binding via electrostatic interactions as is observed within the ECM. Chemical coupling of peptides to hydrogels is one of the simplest approaches that can be taken to modify hydrogels with biochemical signals.

During limb development, the first instance of cartilage formation occurs within a highly condensed cell-rich aggregate that is largely devoid of ECM (Bobick et al., 2009). Cell-cell interactions in this context are mediated by the adhesion molecule N-cadherin (DeLise and Tuan, 2002a; Delise and Tuan, 2002b; Oberlender and Tuan, 1994), and these interactions regulate signaling events that are critical to the initiation of chondrogenesis. Of note, these cell-cell adhesion signals are not present throughout development, but rather arise and peak during this condensation phase, after which adhesion-based signaling from the formed ECM dominates (Singh and Schwarzbauer, 2012). To this end, Bian et al. recently developed an HA hydrogel system that incorporates an N-cadherin mimetic peptide (i.e., HAV) (Fig. 6a), and found that the peptide conjugation promoted both early chondrogenesis of human MSCs, as well as cartilage-specific matrix production (Fig. 6b) (Bian et al., 2013). This outcome was even observed in vivo and it was shown recently that the presence of the N-cadherin signal from the hydrogel also influenced β-catenin signaling in human MSCs (Fig. 6c-d) (Vega et al., 2016).

Fig. 6.

Human MSCs were (a) photoencapsulated in hydrogels containing either N-cadherin mimics or scrambled sequence controls. (b) After 4 weeks of in vitro culture, N-cadherin mimics enhanced chondroitin sulfate (CS) and type II collagen (COL2) by human MSCs, as seen by immunohistochemical staining (Bian et al., 2013). (c-d) Single cell analysis of MSCs in these hydrogel environments showed an increase in N-cadherin mediated β-catenin signaling after 3 days in culture. (c) Cross-sectional images of MSCs stained for β-catenin (green) show that N-cadherin mimics recruit β-catenin to the cell membrane. Additionally, N-cadherin mimics induced an increase in nuclear β-catenin, as confirmed by (d) representative maximum (top) and average (bottom) projections of single MSCs stained for actin (red), nucleus (blue) and β-catenin (green) (Vega et al., 2016). Scale bars: b = 50 μm; c,d = 25 μm.

Due to the importance of the temporal presentation of biochemical signals, there has been increased interest in regulating the presentation of these biochemical cues via enzymatic regulation. Towards this, Salinas et al. used thiol-acrylate photopolymerization to tether RGD peptides with an MMP-13 cleavable linker in order to induce MSCs towards a chondrogenic phenotype (Salinas and Anseth, 2008). The authors found that MSCs encapsulated in these hydrogels produced active MMP-13 between 9 and 14 days in culture, resulting in an over ten-fold increase of GAG deposition in comparison to MSCs exposed to non-cleavable RGD peptides (Salinas and Anseth, 2008).

Besides peptides, growth factors, such as those from the TGF-β superfamily have also been tethered to hydrogels. Although TGF-β plays a key role in promoting chondrogenesis, its therapeutic utility is limited by its inherent protein instability, requiring high amounts of protein that can cause adverse side effects with inefficient cartilage formation. Choi and coworkers compared hydrogels containing TGF-β1 covalently tethered to visible blue light-inducible chitosan (MeGC) hydrogels prior to photopolymerizing with type II collagen versus MeGC hydrogels with positively-charged TGF-β1 ionically conjugated to type II collagen (Choi et al., 2015). The authors found that the cumulative release of TGF-β1 was significantly higher in the covalently bound system, resulting in more sGAG production by adipose-derived stem cells in vitro, as well as improved integration with surrounding tissue of hydrogels implanted in a rat cartilage defect model (Choi et al., 2015).

In vivo, control over the bioavailability of biomolecules is dependent on numerous factors. For example, growth factors including TGF-β are present as a complex which includes a latency associated peptide (LAP), which inactivates TGF-β by masking the receptor-binding domains (Shi et al., 2011). As such, members of the TGF-β superfamily remain inactive and protected from degradation until LAP is removed by one of several mechanisms including proteolytic cleavage of LAP. Laos et al. sought to recapitulate this natural mechanism by coupling TGF-β1 in its latent form to PEGDA prior to crosslinking with thiolated HA hydrogels via a Michael addition reaction (Place et al., 2012). The authors found that encapsulated chondrocytes in these hydrogels were viable for up to 34 days and produced type II collagen without the presence of soluble TGF-β1, as confirmed by immunocytochemistry (Place et al., 2012).

In addition to covalent conjugation, affinity binding of macromolecules to growth factors is an alternative method towards the controlled release of growth factors. To this end, dendrimers, a particular type of branched macromolecules, can be functionalized with peptides that selectively bind to different molecules (Simon-Gracia et al., 2013). Seelbach et al. recently investigated the use of dendrimers with affinity binding peptides towards the non-burst release of bone morphometric protein 2 (BMP-2) and TGF-β1 (Seelbach et al., 2015). Briefly, dendrimers containing four binding peptides presenting sequences targeting BMP-2 or TGF-β1 protein binding were covalently conjugated to HA, mixed with poly(N-isopropylacrylamide)-grafted HA (Hyal-pN) brush copolymers, and then loaded with BMP-2 or TGF-β1. The authors found a dependence on the release of these growth factors based on the length of the dendrimer arms.

GAGs are highly negatively charged polysaccharides that can modulate macromolecular binding, particularly with positively-charged molecules (Lindahl and Hook, 1978). By regulating the degree of GAG sulfation, the extent of this charge can be tuned to control the presentation of positively-charged growth factors (e.g., TGF-β1). To this extent, Lim et al. showed that MSCs cultured in TGF-β1 containing media and encapsulated in desulfated chondroitin hydrogels featured significantly upregulated gene expression of type II collagen and aggrecan when compared to a PEG-based control hydrogel (Lim and Temenoff, 2013).

C2. Particles for controlled release of biochemical factors

Carrier vehicles such as MPs and NPs as a method to deliver encapsulated factors have been extensively studied and optimized to achieve continuous delivery of their payload by controlling the chemical (e.g., charge, degradability) and physical (e.g., carrier size, shell thickness) properties of the carriers. In contrast to simple diffusion of soluble cues into hydrogel constructs, loaded MPs encapsulated in hydrogels protect biochemical factors from degradation and help prevent rapid diffusion and clearance from hydrogels, allowing for greater control over the release kinetics of biochemical cues into the hydrogel. Although this technique has been used and investigated for decades, the utility of MPs (Reyes et al., 2013a; Reyes et al., 2013b) and NPs (Ertan et al., 2013; Lim et al., 2011; Siu et al., 2012; Whitmire et al., 2012) has seen recent advances in the use of delivery vehicles for cartilage repair.

To achieve a more sustained delivery of growth factors, Spiller and coworkers recently designed a hybrid scaffold where insulin-like growth factor 1 (IGF-1) was loaded into a poly(vinyl alcohol) (PVA) and poly(lactic-co-glycolic acid) (PLGA) hydrogel system (Spiller et al., 2012). The authors here used a novel double emulsion technique such that PLGA MPs (11.3 ± 6.4 μm) containing IGF-1 directly formed and evenly dispersed throughout the PVA hydrogel. The authors showed that the IGF-1 release was linear and sustained for at least 45 days, and in vivo studies showed that the hydrogel resulted in the formation of thick cartilage layers and exhibited good integration between the formed cartilage and the surrounding neocartilage.

In addition to IGF-1, there are other biochemical cues that enhance cartilage formation and can benefit from a suitable delivery vehicle for their sustained release. Nell-like molecule 1 (Nell-1) has previously been shown to induce differentiation and growth towards bone and cartilage tissue in vivo (Siu et al., 2011). In order to showcase the importance of sustained release of Nell-1, Siu et al. compared the release of Nell-1 from alginate hydrogels either directly encapsulated in the bulk hydrogel, or loaded into chitosan NPs (100 – 300 nm in size) (Siu et al., 2012). The authors found that Nell-1 containing NPs induced a significantly more gradual release of the protein in comparison to hydrogels where Nell-1 was directly encapsulated. Additionally, Nell-1 NP containing hydrogels induced an increase in GAG synthesis and proteoglycan accumulation in a critical size osteochondral defect 12 weeks post-implantation (Siu et al., 2012). This study showed not only the importance of Nell-1 in cartilage formation, but also the need for NPs for controlled release.

Although design parameters used to control the release kinetics of biochemical factors loaded in MPs and NPs have been thoroughly investigated, not much work has been done to study the influence of the encapsulating hydrogel on the release profiles of biochemical cues loaded in MPs and NPs. To this end, Ahearne and coworkers studied how different hydrogel macromers (i.e., fibrin, agarose, and gellan gum) affected the efficiency of TGF-β3 loaded MPs to induce chondrogenesis of encapsulated MSCs. Piglet MSCs and TGF-β3 loaded gelatin MPs (50 – 70 μm) were encapsulated in fibrin, agarose, or gellan gum hydrogels, and the role of the MPs in inducing cartilage formation from the MSCs over the course of 21 days in vitro was evaluated (Ahearne and Kelly, 2013). The authors observed a significant increase in sGAG accumulation in the agarose and gellan gum hydrogels in comparison to the fibrin hydrogels. Additionally, histological staining with alcian blue showed a stronger and more homogeneous distribution of sGAG in the gellan gum hydrogels, in contrast to a more pericellular presentation of sGAG in the agarose hydrogels. This study shows that not just MPs and NPs, but the hydrogel macromers as well, are important determinants of the release profiles of biochemical factors loaded in these vehicle carriers.

Summary and Future Directions

Although hyaline cartilage is perceived as a simple tissue, developing biomaterials that can achieve mechanical properties of native cartilage and complete integration with surrounding tissues remains a challenge. Cartilage tissue engineering has seen a rapid advance in fabrication techniques that have resulted in hydrogel constructs that are improving the quality of produced cartilage (summarized in Table 1). To increase the mechanical properties of hydrogels, traditional single network hydrogels are being supplemented with either additional networks or mixtures of polymers. These techniques have also been used to fabricate hydrogels which can induce integration with surrounding tissue while promoting chondrogenesis in vivo. Novel processing techniques have also been employed towards fabricating hydrogel fibers and porous hydrogels for improved cartilage formation, as well as the fabrication of multilayered hydrogels that mimic the zonal architecture of native cartilage. In addition to these advances, the field has also seen an increased interest in presenting biochemical cues in a controllable and temporal fashion.

Table 1.

Summary of hydrogel fabrication techniques for cartilage tissue engineering.

Improvements in hydrogel structure		Advances in the processing of hydrogel scaffolds
IPNs	(Dinescu et al., 2015; Guo et al., 2012; Ingavle et al., 2013; Snyder et al., 2014)	3D printing	(Fedorovich et al., 2012; Kesti et al., 2015; Markstedt et al., 2015; Muller et al., 2015; Schuurman et al., 2013)
Semi-IPNs	(Little et al., 2014; Park et al., 2013; Skaalure et al., 2014)	Spinning	(Han et al., 2014; Xu et al., 2013)
Double networks	(Arnold et al., 2011; Kitamura et al., 2016)	Porosity	(Ahrem et al., 2014; Han et al., 2013; Marklein et al., 2012)
Dual networks	(Jin et al., 2011; Levett et al., 2014; Moreira Teixeira et al., 2012; Palumbo et al., 2015; Pirinen et al., 2015)	Microspheres	(Fang et al., 2015; Mendes et al., 2012)
Guest-host networks	(Jung et al., 2014; Wei et al., 2016)	Multilayered	(Cui et al., 2012; Fedorovich et al., 2012; Nguyen et al., 2011; Sun et al., 2015)
Controlled presentation of biochemical factors
Mimetic peptides	(Bian et al., 2013; Salinas and Anseth, 2008; Vega et al., 2016)
MPs / NPs	(Ahearne and Kelly, 2013; Siu et al., 2012; Spiller et al., 2012)
ECM affinity binding	(Choi et al., 2015; Seelbach et al., 2015)

Looking ahead, combinatorial screening approaches to robustly identify synergies between the various hydrogel parameters presented in this review would enable the development of cell-laden and acellular hydrogels that will eventually adequately recapitulate native cartilage. The technology to achieve this kind of screening has been explored for the discovery of hydrogels with a variety of properties such as cell adhesion and proliferation, and foreign body responses (Le et al., 2016; Vegas et al., 2016; Zant and Grijpma, 2016a; Zant and Grijpma, 2016b). So far this has been achieved by changing the extent of material modification, the combinations of different materials at different ratios, or the amount of a peptide of interest in the precursor. Although none of these combinatorial hydrogel platforms have been directly applied to probe chondrogenesis or cartilage matrix synthesis, they are extremely promising for this application. Additionally, a better understanding of cartilage both during development and in its mature state will be critical to the success of advancing these exciting techniques towards developing hydrogel systems that robustly recapitulate native cartilage. Ultimately, as technology is developed with new polymers synthesized and new processing techniques advanced, this field will continue to expand.