Physical and Chemical Factors Influencing the Printability of Hydrogel-based Extrusion Bioinks

Abstract

Bioprinting researchers agree that “printability” is a key characteristic for bioink development, but neither the meaning of the term nor the best way to experimentally measure it has been established. Furthermore, little is known with respect to the underlying mechanisms which determine a bioink’s printability. A thorough understanding of these mechanisms is key to the intentional design of new bioinks. For the purposes of this review, the domain of printability is defined as the bioink requirements which are unique to bioprinting and occur during the printing process. Within this domain, the different aspects of printability and the factors which influence them are reviewed. The extrudability, filament classification, shape fidelity, and printing accuracy of bioinks are examined in detail with respect to their rheological properties, chemical structure, and printing parameters. These relationships are discussed and areas where further research is needed, are identified. This review serves to aid the bioink development process, which will continue to play a major role in the successes and failures of bioprinting, tissue engineering, and regenerative medicine going forward.

Graphical Abstract

1. INTRODUCTION

The ability to bioengineer the various tissues or organs of the human body is a highly sought-after goal within modern research and medicine.^1,2 Tissue engineering strategies utilize cells, signaling molecules, and biomaterials to generate a final, tissue-like product.^3-5 While considerable success and improvements over time have been shown utilizing these strategies, the demand for personalized, large-scale, implantable tissue constructs continuously increased. A key piece to this future is the manufacturing processes that will be utilized. The manufacturing process determines the materials available for use, the architectures which can be created, and the time and cost which goes into the final product.

Additive manufacturing (also known as 3D printing) is causing a dramatic shift in the way society’s products are made. It has also garnered significant interest within tissue engineering strategies. Anatomical structures are often abnormal, complex, and highly variable between patients and disease progressions, advantaging the flexibility of additive manufacturing relative to more traditional techniques such as machining and molding.⁶ There are many different types of additive manufacturing that have been adapted by the tissue engineering field for use with biomaterials and cells, referred to as “bioprinting.” The most common of these are extrusion-based bioprinting.^6-9 Extrusion-based bioprinting can easily incorporate multiple material types, a strategy which is imperative if any regional differences in biomaterials, cell types, or signaling molecules are desired.^10-12

To ensure successful extrusion-based bioprinting, first and foremost are the materials which are a formulation of cells suitable for processing by an automated biofabrication technology that may also contain biologically active components and biomaterials.¹³ These materials, known as bioinks, have a number of difficult requirements they must fulfill.^14,15 First, they must be biocompatible and non-toxic to cells and not promote host immune responses if implanted. Cells must be suspended directly in the material which requires both a thin enough material to mix the cells homogenously and a thick enough material to maintain that distribution and prevent settling of the cells. Cell encapsulation also necessitates bioinks to be primarily composed of water, and for this reason, hydrogels are used.¹⁶ Bioinks must provide bioactivity/cell attachment sites to allow for cell survival, attachment, and proliferation. They must have appropriate mechanical properties, preferably matching that of the tissue application both in terms of driving cell behavior and to withstand mechanical forces during handling and implantation.¹⁷ Appropriate swelling and degradation characteristics are also necessary. Nutrients such as glucose and oxygen must be able to diffuse through the material to allow for cell survival in the deepest portions of the construct.¹⁸ Lastly, bioinks should also serve to direct cellular behavior through the naturally derived polymers, proteins, and binding sites that are found in the ECM of the target tissue.^19,20 These requirements are common to most tissue engineering applications. For the bioprinting process, bioinks should fulfill the three basic requirements: relatively higher viscosity to maintain cell suspension homogeneously and to provide initial structural integrity; strong shear-thinning behavior to minimize shear stress-driven cell damage during the printing process; and rapid crosslinking process after printing.^12,21,22 In this review, we focus on physical and chemical factors impacting the printability of hydrogel-based bioinks specific to the 3D bioprinting process. In particular, we outline the underlying mechanisms which can affect different aspects of printability and present novel hydrogels with dynamic cross-linked networks (physical interactions and dynamic covalent bonds), which can be utilized for cell-based microextrusion bioprinting.

2. PRINTABILITY

There are many requirements of bioinks which are specific to the bioprinting process. These requirements capture the different aspects of bioinks “printability.” In the broadest sense, printability refers simply to the ability to be printed. For the purposes of this review, the following definition will be considered:

“The ability of a material, when subjected to a certain set of printing conditions, to be printed in a way which results in printing outcomes which are desirable for a given application.”

The first aspect of printability a bioink must overcome is extrusion from a micron-scale nozzle tip. This ability is sometimes referred to as a bioink’s “extrudability.” If too much force is used to extrude the material, cells will be permanently damaged via shear stress. Additionally, print times will be longer as higher flowrates cannot be achieved as easily.²³ As a result, if a bioink requires too high of an extrusion force, it cannot be used. After extrusion, a bioink must also be deposited in a predictable manner. Researchers have used a filament classification system to describe the different types of filaments which bioinks can form. Bioinks that form droplets at their tips will spread dramatically and stick to themselves. Continuous filaments are desired and these filaments can either be uniform (smooth) or non-uniform (bumpy, curvy). Smooth filament deposition is easily controlled by the bioprinter, but non-uniform filaments may deviate from the printing path and have variable cross-sections across their length.²⁴ Following bioink deposition, the desired shape must be maintained as multiple layers are deposited on top of it. This “shape fidelity” of bioinks limits the ultimate size and shapes a bioink is able to structurally accomplish.^25,26 Lastly, different printing conditions (nozzle size, print speed, etc.) may be required depending on the application, further constraining the system as a whole. The similarity of a printed construct to the desired dimensions, as influenced by the printing conditions, can be referred to as “print accuracy.” The term printability has been used to describe each of these requirements individually.^22,27-29 However, a bioink is not able to be bioprinted if it fails in any of these areas. Therefore, this review will consider each of these factors as singular aspects of printability.

2.1. Consideration on Bioink Development

With all the constraints and requirements placed upon bioinks, it should come as no surprise that the available bioinks remain a major limitation of bioprinting. Researchers would benefit tremendously from both improvements upon current bioinks and an expanded bioink palette.⁵ A number of hydrogels and hydrogel combinations have been proposed in the literature; however, each has significant room for improvement in at least one area and sometimes multiple areas.¹¹ Contributing to this difficulty, many of the requirements placed on a bioink are in direct opposition. In many cases, bioinks which hold their shape particularly well upon deposition also require the most pressure to extrude.⁵ This tradeoff can be partially circumvented by various strategies including rapid post-printing crosslinking or gelation, the use of support baths, and bioinks which demonstrate particularly high degrees of shear-thinning. Additionally, while great strides are frequently made on the bioactivity and tissue-specific potential of bioinks, these bioinks typically are not well suited for bioprinting and vice versa. For example, Pluronic F127 offers excellent printability for extrusion bioprinting, but it is limited by its biological properties. Meanwhile, decellularized extracellular matrix (dECM) has been converted for use as a bioink due to its high, tissue-specific biological activity, but forms a very weak gel which cannot hold its shape on its own when bioprinted which must be compensated for with complementary strategies.²⁵

Bioink development is a growing field with the main objective of overcoming these various limitations. With respect to printability, bioink development itself is limited by its understanding of the underlying mechanisms which impact printability.¹⁴ The factors which influence printability are numerous, complex, and interrelated.¹⁶ Many relationships between printing conditions, material properties, and printability have been identified. However, many of these findings have been ambiguous or even conflicting and many more relationships remain unstudied. During the bioink development process, this knowledge is used to design, optimize, and improve a bioink.¹⁷ Where it lacks, researchers must work using trial and error and move forward using less than ideal outcomes. The purpose of this review will be to examine these various relationships in an attempt to promote a better understanding and more holistic view of printability and the factors which influence it, in turn improving the bioink development process and therefore the bioinks available for use in bioprinting.

2.2. Measures of Printability

A clear understanding of the measures used to quantify printability is required in order to investigate the factors which influence printability. This topic is covered briefly in this section, but we refer the reader to a recent review by Gillispie et al.³⁰ for a more in-depth discussion. Not all requirements placed upon a bioink, those non-specific to the printing process, should be considered within the realm of printability. With this in mind, the different measures of extrudability, filament type, shape fidelity, and printing accuracy are considered. Table 1 lists the representative measurements for printability.

Table 1.

Representative printability measurements. Reprinted with permission from ref.³⁰ Copyright 2020 IOP Publishing.

Type	Measurement	Limitation/ condition	Ref.
Extrudability
Binary classification	Bioinks "unprintable" if flow could not be achieved at a maximum acceptable pressure	Nozzle detachment	31
	Bioinks "unextrudable" if flow could not be achieved at a maximum acceptable pressure	380 kPa	32
	Bioinks "unprintable" if a flowrate of 0.3 mL/h could not be achieved by their system	Unspecified	33
Pressure under given conditions	Pressure generated by plunger displacement at a constant speed	0.2 mm/s plunger speed	34
	Pressure required to achieve a given amount of material deposition per construct	100 mg construct	21
	Pressure required to achieve a minimum acceptable flowrate (lower bound) & pressure required to achieve a maximum acceptable flowrate (upper bound)	Varied with nozzle size	35
	Minimum pressure required to achieve consistent flow	Unspecified	36
Theoretical shear stress	Proportion of nozzle diameter to extrusion pressure	n/a	37
Filament classification
Submerged	Swelling, equivalent diameter, stretched, rough surface, over-deposited, compressed, discontinuous		38
Filament drop test	Droplet, filament		39
	Droplet, smooth filament, over-gelled filament		40
Qualitative deposition	Straight, curvy		41
	Regular, irregular		32,33
Quantitative deposition	Uniformity ratio (filament perimeter normalized by length)		21
	Pr (pore perimeter normalized by pore area)		40
	Standard deviation of filament heights and widths		42
Shape fidelity
Qualitative	Cross hatch, anatomical shapes		38,41,43-63
Filament spreading (single layer)	Filament width		31,33,34,36,42,46,61,64-70
	Filament height		33,42,61,69
	Spreading ratio (filament width divided by nozzle diameter)		67
	Aspect ratio (filament height divided by width)		71
	Pr (pore perimeter normalized by pore area)		40
Height maintenance	Critical height (maximum achievable height)		71
	Height of cylindrical structure		72
	Height of 5-layer tubular structure		21
Filament collapse	Angle of deflection of unsupported filament		73
	Pore area below an unsupported filament		41
Printing accuracy
Filament dimensions	Observation of broken filaments		31,45,56,65,66
	Filament width		31-33,36,37,41,42,46,61,62,64,69,74,75
	Filament height		33,42,61,69
	micro-CT		76,77
Pore dimensions	Pore area		36,41,62,70,78-83
Filament merging	Minimum distance required between filaments without merging		84
	Overlap distance of a filament printed at an acute angle		62
	Length of fused segment between adjacent filaments with increasing distance between filaments		73
Others
Construct size	Weight of construct		21
Cell mixing	Whether cells could be mixed with pipette		47
Homogeneity	Variability of extrusion force over time		34,58

Extrudability refers simply to how easily a bioink can be extruded through a small diameter nozzle, effecting total processing time and cell viability (Figure 1A). Cell viability during the bioprinting process has been linked to the pressure experienced by the cells in the syringe, the shear stresses experienced by the cells in the nozzle, and the amount of time cells spend in the syringe.^{23,31,40,41,85-88} Because of this, researchers sometimes indirectly quantify the extrudability of their system by live/dead assay immediately following the bioprinting process.^{23,34,69,85,87-90} Others have used extrusion pressure as the measure of extrudability. This can be either the pressure required to achieve a specified flowrate^{21,34,36,37,91}, whether or not a given pressure was able to achieve flow^31,33,71,92, or the minimum pressure which will cause flow^91,93 in their system. More sophisticated measures of a bioink’s extrudability involve rheological measures such as the shear rate-viscosity relationship and power law constants and will be covered in a later section.

Figure 1.

Different aspects of printability. (A) Extrudability can be defined at any point along with the pressure-flowrate relationship. Notably, a minimum flowrate is required to achieve reasonable print times and maximum extrusion force is limited to achieve reasonable cell viability after extrusion. (B) Filament classification has been used to describe the types of filaments which a bioink can form. This example measure from Ouyang et al. shows how the phenomenon can both be observed qualitatively and measured quantitatively. Reprinted with permission from ref.⁴⁰ Copyright 2017 IOP Publishing Ltd. (C) Shape fidelity refers to the ability of a bioink to maintain its structure upon deposition. The example here from Ribeiro et al. tests a bioink’s ability to form lateral pores. Reprinted with permission from ref.⁷³ Copyright 2017 IOP Publishing Ltd. (D) Printing accuracy refers to the similarity of the printed structure to the original design as influenced by the printing conditions. This example from Giuseppe et al. uses zig-zag and cross-hatch structures to compare dimensions. Reprinted with permission from ref.³⁶ Copyright 2017 Elsevier Ltd.

Filament type can easily be determined qualitatively. Several researchers conduct this test by extruding the bioink into the air rather than on a substrate and classify the shape of the extruded material as either droplet or filament (Figure 1B).^35,39-41 Bioinks that form filaments can be further classified by whether that filament is smooth or uniform.⁴⁰ Non-uniform filaments are can be more easily identified after deposition as some other researchers have observed.^33,41,92 However, the most robust measures of filament type are quantitative. After deposition, various dimensions can be made from the filaments including the perimeter of square pores relative to perfect squares,⁴⁰ the perimeter of single filaments relative to a straight line,²¹ and the variability of filament widths.⁴²

The shape fidelity of a bioink is most frequently measured as the width and/or height of a single filament (Figure 1C). While easily to evaluate, this measure is highly prone to influence by the quantity of material deposited and interactions with the surface substrate. Additionally, it does not capture the full multilayer, spatial behavior required of bioinks. Among measures which do, the Pr value developed by Ouyang et al. is an attractive measure. Although it only evaluates two printed layers, it also has the benefit of being able to detect non-uniform filaments and controlling for material deposition by normalizing the perimeter to the area of the pore.⁴⁰ Gao et al. quantified shape fidelity as the height of a 5-layer tubular structure. Bioinks with poor shape fidelity sagged under the weight of multiple layers, resulting in a decreased height measurement ²¹ Lastly, Ribeiro et al. measured the angle of the collapse of an unsupported filament printed between pillars of varying distances apart. Bioinks with poor shape fidelity have larger deflection angles as they cannot span the gaps as well.⁷³

Printing accuracy has been measured in a variety of ways. Some of these measures are very similar to those for shape fidelity (Figure 1D). The key distinction is that print accuracy measurements compare a common bioink under different printing conditions whereas shape fidelity measures compare different bioinks under common printing conditions. Most commonly, filament height, width, and pore size are used to assess printing accuracy. These are simple and important measures that can help ensure similarity between the designed structure and final structure, but they also test the accuracy of the printing system in the least strenuous fashion. Micro-computed tomography (μCT) measurements have been used to compare entire constructs to the design computer-aided design (CAD). This provides a much more robust test but is difficult to conduct and requires expensive imaging equipment and software.^76,77 Weight/volume of the final construct may also be used to confirm the desired amount of material deposition.²¹ Several researchers have also identified unintentional filament merging as an aspect of print accuracy. A particularly robust measure of this phenomenon involves printing a zig-zag pattern with the distance between lines increasing. The outcome measure is the ‘fused segment length’ between two parallel filaments.⁷³ Ultimately, more measures of printing accuracy are needed to capture additional phenomena such as filament discontinuity, turn accuracy, errors in flow initiation and stoppage, and others.

3. RELATIONSHIPS BETWEEN PRINTING OUTCOMES AND RHEOLOGICAL PROPERTIES

Rheology is the branch of physics that studies the flow of matter. It has proven to be extremely useful for bioink development and there are several rheological tests that are applicable to bioprinting. Most hydrogels demonstrate non-Newtonian behavior.⁹⁴ In a Newtonian fluid, the shear rate-shear strain relationship is linear with viscosity remaining constant across strain rates. Hydrogels, on the other hand, are typically shear-thinning, meaning their viscosity decreases as shear rate increases.^95,96 The result is beneficial for the purposes of flow. As shear rate increases, the apparent viscosity of the material decreases, allowing flow to occur with lower pressure differentials and lower shear stresses on the material, and importantly cells, than would have occurred otherwise. This property is often measured via frequency sweeps, where the material is tested at a constant strain across a range of frequencies or shear rates (Figure 2A).^95,96 Also important is the thixotropy of a material. Once a strain rate is applied, thixotropic materials will show a decrease in apparent viscosity over time. This change in viscosity in turn impacts the bioink’s flow with respect to time. Thixotropy is sometimes confused with shear-thinning as both properties result in decreased viscosity, but shear-thinning occurs with an increase in strain rate while thixotropy occurs over time at a given strain rate.⁹⁴ Materials may also be shear-thickening (opposite of shear-thinning) and rheopectic (opposite of thixotropic), but these are highly undesirable for bioprinting.

Figure 2.

Common rheological measures associated with bioink printability (A) Shear-thinning behavior of bioinks (log scale) with viscosity decreasing as shear rate increases. Reprinted with permission from ref.¹⁰⁰ Copyright 2017 BioResources. (B) Viscoelastic and yielding behavior of bioinks. G’ and G” can be averaged from the linear viscoelastic region while yield stress can be defined at the crossover point between G’ and G”. Reprinted with permission from ref.¹⁰⁰ Copyright 2017 BioResources. (C) Recovery behavior of a bioink using different shear rates to model the extrusion phase. Reprinted with permission from ref.¹⁰³ Copyright 2018 Springer-Verlag London Ltd.

Hydrogel materials also demonstrate viscoelastic properties, meaning their behavior is determined both by viscous and elastic components. The elastic component is measured by the storage modulus, denoted by G’, and refers to energy which is stored in a material when deformation is applied. The viscous component is measured as the loss modulus, denoted G”, and refers to energy which is lost when deformation is applied to the material.^95,96 The loss tangent, or tan delta, is the ratio of loss modulus (G”) to storage modulus (G’). The tan delta represents the relative contribution of viscous and elastic components to the material, with materials with values lower than 1 having a higher elastic contribution and materials with values higher than 1 having a higher viscous contribution. Typically, these properties are measured within the linear viscoelastic region (LVR) via strain or stress sweeps.^95,96 Frequency is held constant, and either stress or strain is increased, with storage modulus, loss modulus, and tan delta averaged throughout the LVR where they remain relatively constant (Figure 2B).

Yield stress has been used to describe the initiation of flow and can be defined in multiple ways. Most commonly, a strain or stress sweep is used to gradually increase the stress on the material. As the material begins to flow, the storage modulus will drop dramatically. If dramatic and immediate enough, the stress at this drop off can be defined as the yield stress.⁹⁷ For bioinks where this decrease is more gradual, the stress at which the storage modus intersects with the loss modulus (i.e. where the tan delta is greater than 1 and the viscous properties begin to dominate the material’s behavior) is considered the yield stress (Figure 2B).^33,73,98-100 Others have looked at both static and dynamic yield stress, defining static yield stress as the minimum required to initiate flow and dynamic yield stress as the minimum required to maintain flow.⁹³ While most researchers stick to this concept during bioink development, the measurement and application of yield stress as a rheological measure is less straightforward than these measures may imply.^101,102

Finally, rheology has been used to measure the ability of hydrogels to recover from extrusion. Recovery tests are conducted in three steps. Initially, a very low shear rate is applied to the material, modeling pre-extrusion conditions. This is followed by a high shear rate intended to model extrusion. Finally, the material is returned to the initial shear rate, modeling post-extrusion conditions (Figure 2C). Recovery is typically defined as the end viscosity represented as a percentage of the initial viscosity. Recovery increases over time after the high shear rate is removed, so recovery must be expressed at a certain time (i.e. 10 sec, 60 sec) after the high shear rate has been removed. Compared to other measures, recovery tests model the bioprinting process most closely. However, there is also the least amount of standardization between methods. Recovery can be measured by time to recovery⁹⁹ or percentage recovered.^33,98,103 The recovered parameter can be viscosity^99,103 or G’.^33,98 This can be measured across single^98,99,103 or multiple³³ cycles and the extrusion phase can be modeled via shear rate,^99,103 strain,³³ or stress.⁹⁸

3.1. Extrudability

Among different aspects of printability, the relationships between rheological properties and extrudability are the most well understood. Direct measures of extrudability, such as experimentally derived pressure-flowrate curves, are rarely conducted except to validate a rheological model. As such, most researchers typically present rheological characterizations as a proxy for extrudability. The main rheological measure of extrudability is the viscosity, with higher viscosity resulting in lower extrudability. Due to the non-Newtonian behavior of bioinks, viscosity must be measured across a wide range of shear rates with the upper end preferably being in the range of that experienced by the bioink during extrusion. Many factors can influence a bioink’s viscosity such as temperature, time, solvent, preparation methods, and many more. Hydrogel composition and concentration have been examined most frequently. The addition of material has been demonstrated to result in higher viscosity and therefore lower extrudability in many bioinks, including chitosan, gelatin, alginate, Pluronic F127, κ-carrageenan, laponite, poly(2-hydroxyethyl methacrylate) (PHEMA), methylcellulose, xanthan, and others.^{33,35,54,63,87,89,97,104}

Plotting several bioinks on the same viscosity vs shear rate graph can be sufficient for comparison within a single experiment. However, for cross-study comparisons and further modeling, it is helpful to model this behavior using the power-law relationship. By rheologically testing a bioink over a range of shear rates (γ̇) the resulting data can be fitted to a power-law equation,

$$\tau =K{\stackrel{.}{\gamma }}^n$$

$$\tau =\eta \stackrel{.}{\gamma }$$

$$\eta =K{\stackrel{.}{\gamma }}^{n-1}$$

where τ is the shear stress and η is the viscosity as measured by the rheometer, K is the consistency index, and n is the flow index or power-law index. These two shear-thinning constants are very useful in modeling flow behaviors. K is related to the bioink’s viscosity and is sometimes referred to as the “apparent” or “zero shear” viscosity.^86,105 The degree to which a bioink is shear thinning can be inferred by the value of n, with values closer to 0 demonstrating more shear-thinning and values closer to 1 demonstrating more Newtonian behavior. Both of these values can be useful in comparing the extrudability of bioinks.^{35,86,87,89,91,105,106} For example, Nadgorny et al. developed a novel benzaldehyde-functionalized PHEMA with ethylenediamine (EDA) bioink and used these properties as proxies for extrudability to help optimize for the concentrations of PHEMA and EDA.⁹⁷

Using these constants and the nozzle dimensions, the bioprinting system can be modeled in even further detail via the Hagen-Poiseuille equation. The pressure-flowrate relationship can be derived to predict the flowrate at a given pressure or pressure required to achieve a given flowrate.^{86,89,91,103,107,108} Dávila et al. have applied this concept to the development of an alginate-laponite bioink with a highly desirable viscosity profile.¹⁰³ Additionally, the shear rate and shear stress profiles can be derived to determine the forces cells are experiencing along the radius of the nozzle.^{35,69,89,103,106} While accurate for most purposes, these models incorrectly will predict flow to occur at very low pressures. In reality, minimum stress must first be overcome to initiate flow. By incorporating the yield stress as the determining factor for the initiation of flow, Herschel-Bulkley models can be used to model flow behavior.^93,105 For example, Smith et al. examined Pluronic bioinks with various additives to optimize for chemical and flow characteristics.⁹³ Furthermore, a no wall slip condition is commonly assumed when modeling bioink extrusion with these models. However, there is some evidence that neglecting wall slip will slightly underestimate the flowrate and more complex models that account for this boundary condition are more accurate.^86,105 This is all to say that the influence of rheological properties on extrudability is relatively well understood, at least in terms of the bioink’s flowrate. Several researchers have additionally used the rheological properties of their bioink to predict its deposition based on the flowrate and print speed (i.e. speed ratio).^86,105,107

The influence on cell damage is slightly less understood, although several factors have been identified. Using these models of extrusion, the shear stress experienced by the cells can be estimated, which has been directly related to their viability. Of particular note, shear stresses are highest near the wall of the nozzle, and viability is subsequently lowest in these regions.^23,85,87 Flowrate and nozzle length have been used to estimate the extrusion duration of cells in the nozzle, which was found to decrease the viability of cells.⁸⁷ Other factors that have been implicated include waiting time in the syringe²³ and extensional flow at the syringe to nozzle transition.⁸⁸ Others have found no difference in cell viability at different extrusion pressures or shear stresses.⁶⁹ Complicating the matter further, while the direction of these relationships seems likely to hold regardless, the cell types and hydrogel carriers used may influence the magnitude of these effects. Further research is needed into the different causes of cell damage during the bioprinting process, especially in relation to the rheological properties of the bioink cell carrier. Such studies would aid significantly in future bioink development.

3.2. Filament Classification

Unfortunately, very few rheological studies have been directly related to filament formation outcomes. Still, some trends can be observed when viewing the body of literature as a whole. Schuurman et al. showed that at 37°C their gelatin methacrylate (GelMA)-only bioink formed droplets even at concentrations up to 20% w/v. The addition of 2.4% w/v hyaluronic acid (HA) resulted in smooth filaments and improved printing outcomes.³⁹ Paxton et al. classified the filaments of different bioinks. They looked at different concentrations of Pluronic F127, different concentrations of alginate, and different degrees of alginate crosslinking. No rheological characterizations were done on bioinks which formed droplets because they were deemed unprintable. However, both Pluronic and alginate transitioned from droplet to smooth filament formation with increasing concentration. Alginate was also able to undergo this transition by increasing its degree of crosslinking via CaCl₂ concentration.³⁵ Davila et al. added Laponite to their 1% alginate bioink and saw a transition from droplet to smooth filament between 4% and 5% Laponite, corresponding with increases in viscosity, G’, and G”.¹⁰³ Habib et al. mixed 4% alginate with varying carboxymethylcellulose (CMC) concentrations and found a transition from droplet to smooth filament formation between 1% and 2% CMC.⁴¹ Ouyang et al. showed the effect of concentration, temperature, and holding time on gelatin-alginate bioinks.⁴⁰ In their work, droplet formation at a given temperature was seen in bioinks which did not gel (defined as a crossover between G’ and G”) at that temperature.

Ouyang et al. also found the formation of non-uniform filaments to occur with higher gelatin concentrations, lower temperatures, and longer holding times. As each of those factors increased gelation, the authors attributed the non-uniform filaments to “over-gelation.”⁴⁰ Gao et al. related rheological measures to filament-type directly.²¹ Using varying concentrations of gelatin and alginate, the researchers showed a transition from smooth to non-uniform filaments as the loss tangent decreased. The transition occurred approximately between a loss tangent of 0.25 and 0.45. However, this window was not found to be predictive when applied to other bioinks. Kiyotake et al. showed their pentanoate-functionalized hyaluronic acid hydrogel (PHA)-based bioinks to increase in viscosity and G’, decrease in recovery percentage, and have higher yield stress with increasing concentration.⁹⁸ At the highest concentrations, the bioinks exhibited non-uniform filaments. They attributed this loss of uniformity to a dramatic increase in yield stress (over 1000 Pa) and poor recovery from a high shear rate (less than 85%). Conversely, Zhu et al. looked at polyion complex (PIC) hydrogels and found lower hydrogel concentrations to result in both decreased viscosity and non-uniform filaments.⁹²

While it is inconclusive from the current studies how rheological properties govern the droplet to uniform filament transition, some notable deductions can be made. Transitions from droplet to smooth filament have been found by increasing hydrogel concentration, ionic crosslinking, and thermal gelation. In general, this transition represents a change in the factor which dominates the bioink behavior. At droplet formation, bioink behavior is dominated by surface tension and water molecule interactions. For filament formation, the polymer network begins to dominate behavior as hydrogel concentration and gelation increases. However, further increases in hydrogel concentration and/or gelation beyond smooth filament formation are typically required to accomplish good shape fidelity.^21,40,41 All bioinks which form droplets will have poor printability, but not all bioinks which form smooth filaments will have good printability. Therefore, droplet formation is better viewed as a pre-screening tool and bare minimum barrier used to quickly eliminate bioink candidates.

The underlying cause of the transition from smooth to non-uniform filaments is even less clear. The contradictory results between gelatin-based and PIC-based bioinks likely mean different mechanisms are at play in the cause of these phenomena between the two hydrogels. Furthermore, some hydrogels (such as Pluronic) do not seem to exhibit this behavior regardless of hydrogel concentration or degree of crosslinking. As a result, further research into this phenomenon is needed before any major conclusions can be drawn.

3.3. Shape Fidelity

It is also difficult to determine which rheological measurements influence shape fidelity. Very few studies have attempted to directly define this relationship. However, many studies have looked at both the rheology and shape fidelity. Most commonly, this is done while optimizing for hydrogel concentration within a bioink.^{33,34,41,54,71,89,90,97-100,103,109} Hydrogel concentration influences many rheological parameters simultaneously, which is a major obstacle in studying the rheology-shape fidelity relationship. The most frequently rheological measure examined is viscosity. As hydrogel concentration increases, so does the viscosity for most materials. In these studies, a subsequent increase in shape fidelity is also typically seen.^{34,41,54,89,90,97-100,109}

However, viscosity encompasses only a single portion of bioink behavior, namely, it’s resistance to flow. The viscoelastic properties of a bioink have also been shown to play an important role. A bioink’s complex modulus (G*) also typically increases with concentration. This increase may be due to an increase in G’, G”, or, most commonly, both.^{71,97,99,100,103,109} Even as both G’ and G” increase, one may come to dominate the behavior as the relative contributions (quantified as the loss tangent) change.²¹ G*, G’, and G” have all been linked to shape fidelity indirectly through bioink concentration. As the measure of elasticity, G’ has received the most attention among them.^{71,97,99,100,103,109}

The yielding behavior of a bioink is also frequently studied alongside changes in shape fidelity. The concept here is that as bioink’s yield strength increases, it can resist greater forces before plastically deforming. As hydrogel concentration increases, so does the amount of stress required to initiate flow. This increase has been linked to shape fidelity in several studies.^71,97-100 Rutz et al. additionally reported on the strain at the yield point, which decreased with increasing concentration and shape fidelity, and G’ at the yield point, which increased with hydrogel concentration and shape fidelity.⁷¹ Additionally, this relationship does not necessarily hold when adding a new hydrogel to the bioink, as Wilson et al. found the addition of laponite to their κ-carrageenan (κCA) bioink to dramatically decrease the yield stress while simultaneously improving its shape fidelity.³³

Lastly, the recovery behavior of a bioink has also been linked indirectly to shape fidelity via hydrogel concentration. Wilson et al. attributed the improved fidelity of their laponite supplemented bioink (despite its decrease in yield stress and G’) to its enhanced recovery behavior (93 to 99% recovery of initial G’ vs 69% for the κCA only bioinks).³³ Also looking at laponite, Peak et al. found a decrease in recovery time corresponded to an improved shape fidelity for their PEG-based bioinks.⁹⁹ Conversely, Kiyotake et al. saw a decrease in recovery percentage with increasing shape fidelity for a pentanoate-functionalized HA-based bioink.⁹⁸

Several studies have attempted directly relate rheological measures to shape fidelity. Diamantides et al. compared several rheological measures to the shape fidelity of their collagen-based bioinks, including G' pre and post UV crosslinking, crosslinking rate, and time required to crosslink. They found that, under their range of testing conditions, G' prior to crosslinking was the best predictor of shape fidelity (Figure 3A).⁷⁰ Jia et al. related the shape fidelity of alginate hydrogels to their viscosity. By varying the alginate concentration and oxidation percentage, they found that higher viscosities resulted in better shape fidelity (Figure 3B).¹¹⁰ Smith et al. investigated four different additives at two concentrations each by incorporating them into their methacrylated Pluronic bioink. Shape fidelity was assessed by the filament diameter upon deposition relative to that of the nozzle diameter. The authors related this measure to their bioink’s dynamic yield stress, with high yield stress correlating to high shape fidelity. However, the researchers did not control for flowrate and it is unclear whether the changes in filament diameter were a result of decreased shape fidelity or simply different volumes of bioink being deposited.⁹³

Figure 3.

Identified relationships between rheology and shape fidelity. (A) Diamantides et al. related the storage modulus of their bioinks to filament width relative to nozzle size. Reprinted with permission from ref.⁷⁰ Copyright 2017 IOP Publishing Ltd. (B) Jia et al. related the viscosity of their bioinks to the area covered by printed dots. Reprinted with permission from ref.¹¹⁰ Copyright 2014 Elsevier Ltd. (C) Gao et al. related the tan delta (loss tangent) of their bioinks to the height of a 5-layer tubular structure. Reprinted with permission from ref.²¹ Copyright 2018 IOP Publishing Ltd. (D) Ribeiro et al. related the yield stress of their bioinks to its angle of deflection across unsupported gaps of varying distances. Reprinted with permission from ref.⁷³ Copyright 2017 IOP Publishing Ltd. (E) Ouyang et al. related the gelation kinetics of their bioinks to the shape of horizontal pores, quantified using their Pr value. Reprinted with permission from ref.⁴⁰ Copyright 2016 IOP Publishing Ltd.

Results from Gao et al. directly relate loss tangent to shape fidelity by measuring the height of a 5-layer tubular structure. The height of the structure was inversely related to the loss tangent (G’/G”) and loss modulus (G”) of their gelatin-alginate bioinks (Figure 3C).²¹ Ribeiro et al. developed a theoretical model for their measure of shape fidelity. They printed Pluronic-PEG bioinks across pillars of varying distances apart, leaving the bioink unsupported between pillars. The following equation was derived to predict the angle of deflection (θ) of the unsupported filament:

$$\theta =si{n}^{-1}(\frac{\rho gL}{{\sigma }_{yield}})$$

where p is the bioink density, g is the gravitational acceleration, L is half the length between pillars, and σ_yield is the yield stress of the bioink as measured on a rheometer. Their results showed a strong negative correlation between yield stress and deflection angle. The model slightly overestimated the angle of deflection but showed similar trends as the experimental data across different bioinks and gap distances (Figure 3D).⁷³ Ouyang et al. showed, using thermosensitive alginate-gelatin bioinks, that gelation time impacted printability. Bioinks which did not gel as quickly did not maintain their shape as well, especially if the gelation time was longer than the time it took to print each layer (Figure 3E).⁴⁰

To summarize, bioink shape fidelity has been positively correlated to hydrogel concentration, thermal crosslinking, viscosity, G’, yield stress, and recovery abilities and negatively correlated with G” and loss tangent. A major limitation in studying this relationship is the difficulty of isolating individual parameters. It seems likely at this point in time that each of these factors may play a role in bioink shape fidelity. More complex models of bioink behavior, such as via finite element analysis,⁴² may be needed to predict printing outcomes.

4. RELATIONSHIPS BETWEEN PRINTING OUTCOMES AND PROCESS PARAMETERS

Some measures of printability are influenced by the printing process. These measures of printability include: line width, corner resolution, corner regularity, pore size, and filament horizontal holding.^35,70,111 Measures of printability are controlled by varying sets of process parameters. These measures of printability have been assessed visually, either qualitatively or quantitatively. Qualitative assessments have been performed by eye and judgment.^41,112 Quantitative assessments involve a photograph from either a camera^{36,41,113,114} or microscope^41,61,114 followed by image analysis with software such as ImageJ.^{36,37,41,42,61,114}

4.1. Consideration on Process Parameters

The most examined measure of printability is the line width. Line width increases with increasing nozzle diameter, extrusion pressure, and nozzle height. Line width also increases with decreasing feedrate and line pitch. These relationships have been demonstrated over limited independent variable ranges and result in limited dependent variable ranges (Table 2). Some independent variable ranges overlap. Differences can be attributed to differences in materials (as different hydrogels were tested in each study), which vary in rheological and material properties, and by differences in other process parameters when isolating correlation among parameters of interest.

Table 2.

Influence of various printing parameters on printing outcomes

	Process parameter	Process parameter values	Range of resulting printability measure	Directiona	Materials
Line width	Nozzle diameter	210 μm (27G), 260 μm (25G), 514 μm (21G)111	220.72 to 3215.85 μm	↑	24.5% Pluronic F127
		30G, 27G, 25G37	270 to 9170 μm	↑	7% alginate + 8% gelatin
		100 μm (32G), 150 μm (30G), 230 μm (27G), 300 μm (25G), 400 μm (23G)36	170 to 570 μm	↑	7% alginate + 8% gelatin
	Extrusion pressure	100, 200, 300 kPa111	220.72 to 3215.85 μm	↑	24.5% Pluronic F127
		100, 150, 200, 250 kPa37	270 to 9170 μm	↑	7% alginate + 8% gelatin
		4, 41, 55, 69, 83, 103 kPa (5, 6, 8, 10, 12, 15 psi)41	~450 to ~1750 μm	↑	4% alginate + CMC
		69, 83, 97, 110, 138 kPa (10, 12, 14, 16, 20 psi)114	~480 to ~1260 μm	↑	4% alginate + CMC + montmorillonite
		40, 60, 80 kPa61	~450 to ~4300 μm	↑	dECM (porcine skin)
	Feedrate	4 to 9 mm/s112	~550 to ~1130 μm	↓	2.5% alginate + 8% gelatin
		10, 20, 30 mm/s111	220.72 to 3215.85 μm	↓	24.5% Pluronic F127
		1, 2, 3, 4, 5, 6 mm/s37	270 to 9170 μm	↓	7% alginate + 8% gelatin
		4, 5, 6, 7, 8, 9, 10 mm/s41	~450 to ~1350 μm	↓	4% alginate + CMC
		0.8, 2.1, 3.3 mm/s 61	~450 to ~4300 μm	↓	dECM (porcine skin)
	Nozzle height	0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1 mm112	~960 to ~1270 μm	↑	2.5% alginate + 8% gelatin
		0.4, 0.7, 0.9, 1.1, 1.3, 1.5 mm41	~570 to ~1270 μm	↑	4% alginate + CMC
		300, 400, 500 μm42	~420 to ~450 μm	-	4% CNF
		300, 400, 500 μm42	~720 μm	-	Ink 6040
	Line pitch	1, 2, 3, 4, 5 mm112	~680 to ~ 860 μm	↓	2.5% alginate + 8% gelatin
Line height	Extrusion pressure	40, 60, 80 kPa61	~250 to ~2250 μm	↑	dECM (porcine skin)
	Feedrate	5, 10, 20 mm/s42	~280 to ~460 μm	↓	4% CNF
Pore size (1 - filament diffusion rate)	Nozzle diameter	100 μm (32G), 150 μm (30G), 230 μm (27G), 300 μm (25G), 400 μm (23G)36	77% to 99%	↓	7% alginate + 8% gelatin
	Line pitch	1, 2, 3, 4, 5 mm112	~80% to ~96%	↑	2.5% alginate + 8% gelatin
		1, 2, 3, 4, 5, 6 mm41	0% to ~90%	↑	4% alginate + CMC
		1, 2, 3, 4, 5, 6 mm114	0% to ~95%	↑	4% alginate + CMC + montmorillonite
		2, 4, 6, 8, 10 mm113	0% to 93.75%	↑	collagen/PPy-b-PCL
Filament collapse	Line pitch	1, 2, 3, 4, 5, 6 mm41	0 – 100%	↑	4% alginate + CMC
		1, 2, 3, 4, 5, 6 mm114	0 – 100%	↑	4% alginate + CMC + montmorillonite
		2, 4, 8, 16 mm73	0.02 - 0.58 [rad]	↑	Poloxamer + PEG

^aDirection of the relationship between printability and printing condition.

Line height can be controlled by extrusion pressure, feedrate, nozzle height, or recovery time. The only report of an extrusion pressure influence on line height employed an over-deposition methodology, whereby the flowrate resulted in an equal or excessive filament height as compared to the layer height. These researchers reported line heights of 1 to 9 times the layer height.⁶¹ Conversely, most printing methodologies constrain the flow rate such that the resulting filament deposition is equal to the layer height. Extrusion pressure, flowrate, and feedrate are freely adjusted within a range that results in filaments equal to the layer height. If extrusion pressure, flowrate, and feedrate yield ideal within-layer printability, but the between-layer printability is unacceptable, nozzle height is then adjusted to match. At no time can the resulting filament height be continuously less than the layer height, as this will result in a cumulative increase in nozzle height and print failure. It may be advantageous for the resulting deposition height to be slightly greater than the nozzle position so as to create interference and better adhesion between layers.

When hydrogel filaments of different orientations overlap (often reported at 90° angles), filaments merge at the point of contact, and surface tension can draw out sharp contact angles.^41,112 The degree of filament diffusion has been correlated with line pitch, in which filament diffusion increases with decreasing pitch. Large line pitch minimizes filament diffusion, but there will always be some small amount which occurs. If pores are too small, filament diffusion can be so severe that the initial pore disappears. Similarly, Ribeiro et al. have developed a filament fusion test which shows how changes in the distance between filaments affect the fusion between filaments. As the line pitch of their cross-hatch design increased, the length of the fused section between filaments decreased.⁷³

As a process parameter, line pitch is user-controlled and determined prior to printing. Therefore, the influence of line pitch on printability is necessary to understand. Line pitch has been shown to influence the filaments deposited on subsequent layers, with greater distances between printed filaments on the supporting layer resulting in higher filament collapse on the supporting layer.^{41,73,114,115} For a given material, some increased line pitch results in 100% filament collapse, and all greater line pitches have complete collapse. For example, a line pitch of 2 mm was achieved with alginate before complete collapse, and the addition of CMC achieved 3 mm at a low concentration and at least 6 mm at higher concentrations.⁴¹ The addition of montmorillonite to alginate decreased the line pitch at which complete collapse occurred, from 3 mm down to 2 mm, which was likely due to the increased weight of the montmorillonite filament without an appreciable increase in yield strength.¹¹⁴ The usage of the filament collapse test has suggested that varying degrees of filament sag are permissible, but so far, acceptable and unacceptable ranges have not been identified or discussed.

Corner resolution describes the sharpness of a corner. Current measures of corner resolution are qualitative, being described as being either sharp or curved/rounded. Nozzle height effects corner resolution: increasing nozzle height reduces corner resolution, shifting the deposition shape from sharp to curved.^41,112 Corner regularity has been used to describe the degree of over-deposition at a corner. As corner angle decreases below a right angle, extrudate increasing overlaps. For sharp corners of small angles measures, extrudate overlaps considerably and causes the corner to bulge. To prevent the buildup of material in sharp angle corners, the rate of material deposition and be decreased or feedrate can be increased.¹¹²

4.2. Improving Printability by Process Parameters

The main researched printability measures effected by process parameters include line width and line height (Figure 4). This distinction is from quantitative-only assessments in the reporting literature; qualitative assessments were not included in Figure 4. However, we have summarized their findings qualitatively, since factors such as differing materials account for significant differences in results. Regardless of the difference in material properties as they play out in different extrusion printing systems, the trends are universal.

Figure 4.

Relationships among printability measures and process parameters to control line width and line height (cross-sectional geometry).

Process parameters have a role in printability. In general, accuracy is improved by reducing line width and increasing line height. Most materials and printing conditions result in filaments that are too wide and that sag over their cross-sectional area (reducing line height). The combined effect is a reduction in pore area (between printed lines). Changes which reduce line width and increase line height will increase pore area, improving printability. Some of the influential process parameters affect more than one printability measure. Recovery time effects both line width and line height. Increasing recovery time decreases line width and increases line height. Fortunately, in the case of recovery time, a single change in this one parameter will improve printability in two different measures (line width and line height). Extrusion pressure and feedrate, however, cause divergences in line width and line height. Increasing feedrate (or decreasing extrusion pressure) decreases line width (an improvement) but decreases line height (a detriment). There is only one report of recovery time affecting filament height as a measure of printability, and this report only investigated alginate and graphene oxide doped alginate.⁸⁹

While some research has identified the connection between flowrate and line width, none has identified a connection between flowrate and line height. Since flowrate is directly proportional to extrusion pressure, the relationship between flowrate and line height can be similar to the one between extrusion pressure and line height. A future study should examine this relationship and use several different materials to show the scope of applicability.

5. CHEMICAL STRATEGIES: PHYSICAL INTERACTIONS

Hydrogels are generally fabricated through cross-linked networks formed by physical interactions or chemical bonds.^116,117 Physically crosslinked hydrogels are characterized by the network being held together with weak noncovalent interactions, including hydrogen-bonding, hydrophobic interaction, ionic interaction, host-guest interaction, metal-ligand interaction, and π-π stacking interaction.^118-120 These reversible interactions allow most physical hydrogels to exhibit good printability due to dynamic rheological properties, which are advantageous for microextrusion-based printing processes.¹²¹ In this section, we focus on (i) hydrogen bonding, (ii) host-guest interactions, and (iii) metal-ligand coordination that have been mostly employed in the recent emerging shear-thinning and self-healing hydrogels due to their versatility and flexibility in chemistry, allowing diverse new design approaches. For a detailed discussion of other classes of physical interactions, such as ionic, hydrophobic, and π-π interactions, we refer the reader to various excellent recent reviews describing design strategies for shear-thinning and/or self-healing physical hydrogel formation.^117,121-125

5.1. Hydrogen Bonding

Hydrogen bonds are noncovalent dynamic interactions in which a hydrogen atom attached to electronegative atoms, such as oxygen, nitrogen, and fluorine, forms a partial intermolecular bonding interaction with other vicinal electronegative atoms.¹²⁶ Hydrogen bonding exhibited a relatively weak bond strength between 0.25 to 15 kcal/mol,¹²⁶ which is several times weaker than covalent bonds and ionic interactions. However, when low molecular weight compounds or polymers are designed to facilitate multiple hydrogen bonding, the overall association force between their structures is greatly improved,^127,128 which enables hydrogen bonds to serve as useful interactions to build up diverse supramolecular hydrogels. To date, various classes of moieties have been employed to develop dynamic hydrogels with shear-thinning or stimuli-responsive properties. Some examples of these moieties are an adenine/thymine base pair,¹²⁹ ureidopyrimidinone (UPy),^130,131 6-aminocaproic acid,¹³² and diaminotriazine,¹³³ and phenolic compounds,^134,135 including gallol or catechol groups. Among physical hydrogels formed by these moieties, we focus on recent reports on shear-thinning dynamic hydrogels constructed via hydrogen bonding interactions based on phenolic compounds and UPy self-complementary dimerization.

5.1.1. Phenolic Compound-Based Interaction.

Shin et al. reported on rapid spontaneous gelation of gallol-conjugated hyaluronic acid (HA-Ga) in the presence of a gallol-rich cross-linker, oligo-epigallocatechin gallate (OEGCG) (Figure 5A).¹³⁶ The main driving force for hydrogel formation is multiple gallol-to-gallol and gallol-to-HA hydrogen bonding interactions (Figure 5B). Rheological studies showed that the storage modulus (G’) and the loss modulus (G”) were dependent on the [D-glucuronic acid-D-N-acetylglucosamine]/[gallol in OEGCG] stoichiometric ratio. At a ratio of 2, the G′ value was 341.6 ± 53.6 Pa at 1 Hz, whereas, at a ratio of 0.5, the hydrogels showed a G′ value of 1390.5 ± 128.0 Pa. The gallol-involved cross-linking was reversible, and thus this hydrogel exhibited shear-thinning behavior (Figure 5C). Under alternating strain, G′ and G″ were recovered to 92% and 80% of their initial values (Figure 5D), and G′ almost recovered to its initial value after 540 sec of the fraction (red arrow). This hydrogel was able to be injected with a 26G needle (Figure 5E).

Figure 5.

(A) Schematic illustration for preparing gallol-rich, shear-thinning hydrogels of HA-Ga/OEGCG. (B) The proposed multiple hydrogen bond formation (red dashed line) between the gallol-to-gallol moieties and gallol-to-HA backbone. (C) Changes in viscosity as a function of shear rates for HA-Ga/OEGCG hydrogels with the [HA unit]/[gallol in OEGCG] ratio of 7 (blue), 2 (black), or 0.5 (red). (D) The recovery measurement of G′ displaying the hydrogel structure under alternating strain from 0.1% to 10% back down to 0.1%. (E) A photograph showing the injectability of the HA-Ga/OEGCG hydrogel (the ratio = 0.5) using a 26G needle (inner diameter = 0.26 mm). Reprinted with permission from ref¹³⁶. Copyright 2017 American Chemical Society. (F) Schematic illustration of the 3D printing where the gallol ECM hydrogel ink transitions from a shear-thinning hydrogel during printing to one with mechanical stabilization via oxidation after the printing. (G) Printability of the gallol ECM ink with various concentrations (4, 6 wt%) and injectability as a function of the time after gel formation (0.5, 1, 2 h). Scale bars of 4 mm. Reprinted with permission from ref¹³⁸. Copyright 2019 Elsevier.

Gallol-containing compounds, such as tannic acid, have been widely explored for use in supramolecular hydrogels through a combination of hydrogen bonding and metal-coordination interactions.^134,137 In this case, the addition of Fe³⁺ ions to the tannic acid-bound polymer chains was essential to form the cross-linked network via chelation. In contrast, for HA-Ga/OEGCG hydrogels, the addition of metal ions was not needed, because the polymer-bound gallol moieties provided multiple hydrogen bonds that were strong enough to form the gel. Recently, Shin and Burdick et al. designed a gallol-derived extracellular matrix (ECM)-mimetic adhesive bioink exhibiting temporal shear-thinning properties. This bioink formed a gel by noncovalent hydrogen bonding interactions between polymer-bound gallol moieties and peptide linkages in the protein backbones.¹³⁸ The main components for hydrogel formation were gallol-modified HA (HA-Ga) and gelatin-gallol (GEL-Ga). They further studied the stabilization of the hydrogel structure by examining the covalent cross-linking process between gallol aromatic rings through a spontaneous auto-oxidation process (Figure 5F). Gallol groups are known to undergo auto-oxidation to form hydroquinone accompanied by covalent cross-linking over several hours (~120 min).¹³⁹ The printability of this bioink, evaluated by the shape fidelity of the printed filaments, was dependent on the concentration of the gallol ECM hydrogel and time after mixing (Figure 5G). A weight ratio of 1:2 (HA-Ga:GEL-Ga), a concentration of 6 wt%, and a mixing time of 0.5 h resulted in stable filament formation.

A novel injectable hydrogel was prepared through self-assembly of an ABA triblock copolymer comprising catechol-functionalized oligomeric PEG methacrylate (A block) and poly([2-(methacryloyloxy)-ethyl] trimethylammonium iodide) (PMETA) (B block).¹⁴⁰ The aqueous solution of the block copolymer exhibited thermosensitive reversible sol-gel transition behavior. At 37°C, the solution formed hydrogels with shear-thinning and self-healing characteristics, whereas the gel became a sol at 4°C. The main driving force for hydrogel formation was the hydrogen bonding and aromatic interactions between catechol groups. This work demonstrates that the catechol moieties, similar to gallol moieties, can also act as cross-linkers to form dynamic physical networks.

5.1.2. Ureidopyrimidinone Self-complementary Dimerization.

Self-complementary quadruple hydrogen bonding interactions can be used to develop dynamic supramolecular hydrogels. Using a reversible addition-fragmentation chain-transfer (RAFT) polymerization route, Zhang et al. synthesized a water-soluble ABA triblock copolymer (termed UNONU) consisting of a middle poly(ethylene oxide) block (A) and terminal poly(N-isopropylacrylamide) (PNIPAm) block incorporated with Upy moieties (Figure 6A).¹⁴¹ The UPy moieties in the PNIPAM blocks can form self-complementary dimers via quadruple hydrogen bonding interactions^142,143 and have been widely employed to produce diverse supramolecular hydrogels.^{130,131,144-146} The UNONU has the lower critical solution temperature (LCST) of 21°C due to the thermosensitive UPy-conjugated PNIPAM block, which induced the formation of assembled micelles/clusters. UPy moieties in the PNIPAM core domains strongly hold the polymer chains, thereby forming hydrogel cross-linked 3D networks (Figure 6B). The UNONU copolymer showed a temperature-responsive sol-gel transition, whereas the copolymer (NON) without incorporated UPy did not form a gel even at 37°C (Figure 6C). This observation clearly supports the key role of UPy self-complementary dimerization in hydrogel formation. UNONU copolymer hydrogels exhibit rapid self-healing properties after mechanical disruption (Figure 6D). Rheological studies show that G′ is dominant across the whole frequency range (Figure 6E) and that the supramolecular hydrogel is soft but highly elastic (tan δ < 0.2). The viscosity of the hydrogel decreases with increasing shear rate, indicating that the hydrogel can be injected through the extrusion process (Figure 6F). When two individual hydrogel pieces are in contact, rapid healing is observed. Furthermore, the adhered interface withstands stretching (Figure 6G,H).

Figure 6.

(A) Chemical structure of the ABA triblock copolymer. (B) Schematics of the reversible sol-gel transition of the prepared hydrogel under temperature switch. (C) Gelation test of UNONU and NON polymer solutions under cold (4°C) and warm (37°C) conditions via simple tilting. (D) Dynamic strain amplitude cyclic test (γ = 0.5% and 200%) of the hydrogel at 37°C showing rapid self-healing behavior. (E) Frequency-dependent (at a strain of 1%) oscillatory shear rheology of the hydrogel. (F) Viscosity measurement of the hydrogel (inset: injection test of the hydrogel at room temperature). (G) Hydrogels were cut into equal halves by a razor blade. (H) The self-healed hydrogels can also withstand stretching, scale bars: 1 cm. Reprinted with permission from ref¹⁴¹. Copyright 2017 American Chemical Society.

UPy moieties have also been introduced to gelatin-based hydrogel. Zhang et al. prepared UPy-conjugated gelatin-based (Upy-gelatin) hydrogels cross-linked by UPy self-complementary dimerization.¹⁴⁷ The hydrogels formed from a Upy-gelatin solution of 10 wt% at 25°C showed a storage modulus of 1000 Pa, which was ascribed to the ordered arrangement of gelatin chains and the quadruple hydrogen bonding between UPy moieties. The hydrogels exhibited a shear-thinning behavior and an excellent self-healing property at 25°C. They demonstrated that the gelling property and mechanical strength could be further improved by adding Fe³⁺ ions to the solution of Upy-gelatin due to coordination interaction between carboxylate (COO⁻) groups of gelatin and the Fe³⁺ ions.

5.2. Host-Guest Interactions

Two main classes of macrocyclic compounds, cyclodextrins (CDs) and cucurbit[n]urils (CB[n]s), serve as hosts to provide the cavity space for complexation with diverse guest compounds.^148,149 These host compound-based supramolecular interactions have been utilized to produce the various dynamic hydrogels.

5.2.1. Cyclodextrin-Mediated Supramolecular Interaction.

CDs are a family of cyclic oligosaccharides consisting of macrocyclic glucose subunits connected through α−1,4-glycosidic linkages. CDs have a hydrophobic interior cavity with a hydrophilic exterior.¹⁴⁸ The main members of CDs are α-CD, β-CD, and γ-CD, which are composed of six, seven, and eight D-glucose units, respectively. The height of all these CDs is the same (7.8 Å), whereas the minimum diameters (4.4 to 7.4 Å) and the volumes (174 to 427 ų³) of the cavities are dependent on the number of D-glucose units.^148,150 In the aqueous phase, CDs possess a cavity allowing the guests to form inclusion complexes through hydrophobic and van der Waals interactions.^148,151 To date, various design approaches of CD-based host-guest interactions have been demonstrated to generate shear-thinning and/or self-healable supramolecular hydrogels.^152-156 Most of these systems have cross-linked networks formed from the interaction between CD-conjugated polymers and their respective guests.

Xu et al. developed a self-assembled conductive hydrogel formed through the β-CD-adamantane (Ad) host-guest interaction.¹⁵⁷ An electroconductive polymer (PEDOT:S-Alg-Ad) was synthesized by oxidative polymerization of 3,4-ethylenedioxythiophene (EDOT) in the presence of adamantyl-modified sulfated alginate (S-Alg-Ad) (Figure 7A). The PEDOT:S-Alg-Ad polymer formed the hydrogel via host-guest recognition between adamantane moieties in S-Alg-Ad and CDs in poly-β-cyclodextrin (Pβ-CD) (Figure 7B). In a rheological examination, the viscosity of hydrogels decreases with increasing shear rate, indicating that they exhibited shear-thinning behavior (Figure 7C), which could be modulated by the hydrogel composition. This property allows this hydrogel to readily encapsulate cells and to be injected to form 3D constructs. Also, this hydrogel can be rapidly recovered through multiple cycles (G′ > G″), when the applied strain is removed (Figure 7D). The hydrogel exhibited excellent self-healing properties. When the two separated pieces of the hydrogel were in close contact, the gap rapidly disappeared, undergoing complete self-healing within 60 sec (Figure 7E). The physical cross-links in the interfaces between two individual pieces could be reconstructed through host-guest interactions. The mechanical stability of the hydrogels could be improved by the additional π-π stacking and the sulfate anion-π interactions¹⁵⁸. Moreover, this conductive PEDOT:S-Alg-Ad/Pβ-CD hydrogel matrices support the survival, proliferation, and differentiation of encapsulated myoblasts due to the well-established PEDOT properties,^159-161 thereby dramatically promoting myotube-like structure formation. These injectable conductive hydrogel matrices can be used as bioinks to produce 3D tissue constructs that electrically control cellular behavior.^162,163

Figure 7.

(A) Scheme of the synthesis of PEDOT:S-Alg-Ad polymers. (B) Schematic of dynamic cross-link formation utilizing host-guest complexation. (C) Continuous flow experiments showing the shear stress (closed symbols) and viscosity (open symbols) of different host-guest conductive hydrogels. (D) Self-healing property of the conductive hydrogel when the alternate step strain switched from 1 to 500%. (E) Bright-field images of the self-healing process of PEDOT:S-Alg-Ad (2:1)/Pβ-CD hydrogels. Reprinted with permission from ref.¹⁵⁷ Copyright 2019 American Chemical Society.

Noncovalent host-guest interaction between cationic peptide amphiphiles (PAs) bearing either β-CD or adamantane (Ad) groups can be induced to form self-assembling hydrogels. To date, a variety of PA fibrous assemblies have been designed for diverse functional hydrogels. Most PA hydrogels are based on noncovalent interactions, such as hydrogen bonding and van der Waals interactions between β-sheet structures.^164-167 Redondo-Gómez et al. developed PA self-assembled hydrogels.¹⁶⁸ Two PA-conjugates that participated in hydrogel formation were β-CD-PA (host) and Ad-PA (guest). These host-guest PAs are composed of a hydrophobic palmitoyl tail (C₁₆), an oligopeptide sequence (V₃A₃) that induces the β-sheet formation, an ionic trilysine region for further hydrogelation (K₃), a triglycine spacer (G₃), and a 1,2,3-triazole linker near the C-terminus of the PAs to incorporate β-CD or Ad moieties. This dynamic host-guest interaction induced the formation of a hydrogel with PA nanofiber structures comprising β-CD-PA and Ad-PA nanofibers. This PA hydrogel exhibited shear-thinning and self-healing properties. When undergoing strain changes from large (100%) to small (0.1%), the hydrogels underwent a reversible gel-sol transition, and their G′ and G″ values rapidly recovered up to 90%. This peptide-based supramolecular approach may generate a new class of hydrogels with modulated rheological and mechanical properties, which may be achieved not only by manipulating functional groups in nanofiber surfaces, but also by controlling various parameters, such as the concentration and composition of each host-guest PA species.

Wang et al. developed an injectable, host-guest assembled hydrogel between CD-conjugated branched polyethylenimine (CD-PEI) and Ad-conjugated 8-arm PEG (Ad-PEG) for local siRNA delivery.¹⁶⁹ CD-PEI was synthesized by the reaction of PEI with tosylated β-CD and had ~25 CD units per PEI. For the preparation of Ad-PEG, thiolated adamantane was reacted with 8-arm PEG-maleimide through a Michael addition reaction. At a concentration of 20 wt%, CD-PEI and Ad-PEG, with an equimolar ratio of CD and Ad, formed the supramolecular hydrogels through CD-Ad host-guest interactions. The hydrogel showed typical viscoelastic behavior and G’ increased as the frequency increased. The hydrogel also exhibited shear-thinning behavior and rapid self-healing. Under oscillatory strain (250% and 0.5%), the storage modulus at high strain decreased ~40% and was rapidly recovered within 1 sec when returned to low-amplitude strain. The use of cationic PEI as a main polymeric component enables the loading of anionic siRNA into the hydrogel matrix through electrostatic interactions.^170,171 This work is an excellent example suggesting that host-guest recognition-based printable hydrogels can be tailored by selecting relevant chemical structures for specific applications.

5.2.2. Cucurbit[n]uril-Mediated Supramolecular Interaction.

CB[n]s are macrocyclic compounds consisting of glycoluril monomers linked by methylene bridges, where n is the number of glycoluril units.¹⁷² The most frequently used CB[n]s are CB[6], CB[7], and CB[8], synthesized from six, seven, or eight glycoluril monomers, respectively.¹⁵⁰ These CB[n] compounds have the same height of 9.1 Å, while their cavity diameters range from 5.8 to 8.8 Å and the interior volumes are in the range of 164 to 479 ų³.^150,173 CB[n] compounds have a hydrophobic internal cavity bordered by polar carbonyl groups at the cavity entrance. Similar to CD-based recognition, CB[n]-based host-guest interactions have also been employed to produce dynamic supramolecular hydrogels.^174-177 Tan et al. developed a supramolecular hydrogel prepared through dynamic host-guest interactions using a highly branched CB[8]-threaded polyrotaxane (HBP-CB[8]) and a linear naphthyl-functionalized hydroxyethyl cellulose (HECNp) (Figure 8A,B) based on a two-component strategy (HBP-CB[8]@HECNp).¹⁷⁸ HBP-CB[8] which acts as a key building material contains mechanically locked CB[8] host molecules in its branched polymer backbone through inclusion complexation with viologen derivatives.¹⁷⁹ To demonstrate the superiority of this branched architecture in controlling the viscoelastic response, thermal stability/reversibility, and self-healing activity of the hydrogels, a control hydrogel with linear analogs was also prepared using a three-component strategy (LP@CB[8]@HECNp). The HBP-CB[8]@HEC1300Np hydrogels not only exhibited much higher modulus (G′ and G″) values than the LP@CB[8]@HEC1300Np hydrogels but also exhibited a higher viscosity at low shear rates (≈2000 Pa·s at 0.1 s⁻¹), which was more dependent on the shear rate. The higher viscosity and magnitude of the moduli for the HBP-CB[8]@HEC1300Np hydrogels could be ascribed to the more entangled and intertwined network structures compared to the linear polymer-based LP@CB[8]@HEC1300Np hydrogel structures. Both the HBP-CB[8]@HEC1300Np and LP@CB[8]@HEC1300Np hydrogels exhibited shear-thinning behavior. The HBP-CB[8]@HEC1300Np hydrogel showed rapid self-healing performance due to fast CB[8] host-guest association kinetics. The branched architecture of the CB[8]-threaded polyrotaxane motifs employed in this work affords modulation of the dynamics of the supramolecular hydrogel networks, which is probably not expected using the linear architecture-based networks.

Figure 8.

(A) Schematic illustration of (i) synthesis of the highly branched CB[n]-threaded polyrotaxane (HBP-CB[n]) via a semi-batch RAFT polymerization in the presence of CB[n] (CTA : chain transfer agent (benzyltrithiocarbonyl propionic acid) and ACVA : 4,4′-azobis(4-cyanovaleric acid)), (ii) chemical structures of its linear analog (LP), and (iii) naphthyl-functionalized hydroxyethyl cellulose (HECNp). (B) Formation of hydrogel networks through a two-component strategy from HBP-CB[8] polyrotaxane (HBP-CB[8]@HECNp) or a three-component strategy from its linear analog (LP@CB[8]@HECNp). Inset: inverted vial tests for the hydrogel networks. Reprinted with permission from ref¹⁷⁸. Copyright 2018 WILEY-VCH. (C) Light-controlled supramolecular hydrogels. (D) Step-strain rheology alternating between 10% staring and 300% strain for physically crosslinked hydrogels (State I), chemically cross-linked hydrogels (State II), or hydrogels with cross-links reversed by exposure to 254 nm irradiation (State III). (E) Hydrogel swelling and dissipation determined by bathing pre-formed hydrogels in water and hydrogel stability through vial inversion. (F) Hydrogels were patterned by irradiation with 365 nm light using a mask, and the remaining supramolecular network was dissolved in water to leave a patterned covalent hydrogel. Reprinted with permission from ref¹⁸⁷. Copyright 2019 The Royal Society of Chemistry.

Yu et al. demonstrated the microbead-based formation of supramolecular polymer networks assembled from HBP-CB[8] and HEC-Np, based on CB[8]-mediated host-guest interaction.¹⁸⁰ Using a microfluidic device, they developed microbeads with a complementary complex consisting of host HBP-CB[8] and guest HEC-Np. The moldable self-healing supramolecular hydrogels were fabricated by annealing the microparticles at 65°C for 30 min. The resulting hydrogels were self-standing and exhibited self-healing properties due to the dynamic nature of the CB[8]-based host-guest interaction. This microdroplet-based injectable system can be extended for the generation of core-shell microbeads that not only impart dual functions in two domains but also load various bioactive agents in the core region for multifunctional hydrogel bioinks with desirable printability. Zou et al. described a supramolecular hydrogel in which cross-linked host-guest networks can be optically switched between physical and chemical cross-linking states.

Recently, CB[8]-catalyzed [2+2] photodimerization of trans-Brooker’s merocyanine (BM) was observed between two BM guests encapsulated into the cavity of the CB[8] host.¹⁸¹ This unique catalytic activity of CB[8] has afforded the synthesis of supramolecular polymers through the CB[8] host-mediated photodimerization of BM moieties.^182,183 Zou et al. extended the research scope to a host-guest supramolecular system with light-mediated reversibility of the photodimer, which is typically found for photoinduced [2+2] dimerization.^184-186 Figure 8C illustrates that PEG4a-BM and CB[8] form inclusion complex networks, which were reversibly switched between a dynamic and static state under light irradiation with different wavelengths. In the dynamic physical network state, a moldable and self-supporting hydrogel was prepared due to dynamic host-guest interactions. Upon irradiation at 365 nm, the hydrogel became a solid-like material with high elasticity owing to the [2+2] dimerization-induced chemical crosslinking. Subsequent irradiation at 254 nm confirmed the reversibility of the network by returning the system to its initial physical gel state. Step-strain alternating experiments between 10% and 300% strain indicate that the physical gel underwent self-healing behavior under several cycles (Figure 8D, State I). In contrast, the chemically cross-linked hydrogels did not show dynamic recovery behavior, and the hydrogel was mechanically destroyed at high strain (Figure 8D, State II). The dynamic self-healing property could be recovered by irradiating into the hydrogel in State II at 254 nm. Notably, the chemical hydrogel in State II not only maintained the original gel dimensions without significant swelling but was also printed with a high shape fidelity and mechanical stability (Figure 8D,E). The hydrogel with physical host-guest interactions exhibited shear thinning behavior, which can be desirable in bioinks for extrusion-based 3D printing applications (Figure 8F). In addition, the additional post-treatment with light after printing resulted in mechanically robust 3D constructs with high resistance to stress relaxation.

5.3. Metal-Ligand Coordination

Coordination interactions between organic ligands and metallic ions have been widely employed as reversible cross-links to generate dynamic supramolecular hydrogels.^188-190 To date, a diverse selection of metal-ligand pairs has produced various classes of hydrogels with modulated rheological and mechanical properties. In this section, we highlight the recent design strategies for dynamic hydrogels based on several metal-chelating cross-linking approaches in which the organic chelating ligands of interest include bisphosphonate (BP), carboxylate, catechol, and histidine.

5.3.1. Bisphosphonates.

BPs, with two phosphonate groups (PO(OH)₂) covalently linked to a central carbon atom, have served as an organic ligand that exhibits an excellent binding affinity for diverse multivalent metallic cations such as calcium (Ca²⁺) and magnesium ions (Mg²⁺).^191-195 Some recent studies have shown that water-soluble polymers bearing conjugated BP moieties can form supramolecular dynamic hydrogels in the presence of metallic cations and metal-based inorganic particles.^196-198 The BP-based metal-ligand interaction has been extensively pursued as one of the main driving forces to produce dynamic cross-links in shear-thinning self-healing hydrogels. Shi et al. reported on a supramolecular hydrogel through the metal-ligand coordination between BP-conjugated hyaluronic acid (HA-BP) and Ca²⁺ ions.¹⁹⁹ HA-BP was synthesized based on a thiol-ene photo-addition reaction between thiolated HA and acrylamide-functionalized BP.^197,200 Immediate hydrogel formation, through a coordination interaction between ionized BP and Ca²⁺ ions, was observed when HA-BP was mixed with Ca²⁺ ions in water. The hydrogels showed the shear-thinning and rapid self-healing characteristics, which could be attributed to the reversible dynamic nature of the BP-Ca²⁺ coordinated cross-links. The G′ value recovered immediately when the strain decreased from a high value (400%) to a low value (1%). This dynamic property of the hydrogel was supported by the self-healing behavior in which two pieces of hydrogels could be joined together. When extruded through a printer nozzle on the glass surface, the HA-BP/Ca²⁺ hydrogel formed self-standing 3D structures, indicating its potential as a bioink with good printability. Similarly, Zhang et al. demonstrated BP-Mg²⁺ coordination-induced hydrogel network formation.¹⁴¹ When acrylated BP (Ac-BP) and Mg²⁺ ions were mixed with HA-BP in the aqueous phase, assembled Ac-BP-Mg nanoparticles (Ac-BP-Mg-NPs) were formed, which acted as a cross-linker to stably hold the networks of HA polymer chains. This in situ cross-linking process resulted in hydrogel formation in 20 min. Due to the reversibility of the BP-Mg²⁺ multivalent cross-links, the hydrogel exhibited both excellent injectability and self-healing properties. This characteristic was supported by the almost instantaneously recovered G’ value of the hydrogels upon switching from high (100% or 200%) to low shear strain (1%). In addition, the Ac-BP present in the hydrogel network can be used for post-photopolymerization of acrylate groups, thereby resulting in the formation of chemical cross-links. This dynamic hydrogel possessing the potential for post-treatment may offer a benefit in providing robust and stable 3D constructs through 3D printing processes.

Recently, researchers reported on a series of HA-based supramolecular hydrogels that were also stabilized by in situ formed NPs and were assembled through interactions between the BP ligand and various divalent metal ions (M).¹⁹⁰ The self-assembled HA-BP-M hydrogels were formed by the simple mixing of HA-BP, free BP, and various metallic ions under aqueous conditions (Figure 9A-D). The divalent cations of various alkaline earth metals (AEMs) or transition metals (TMs) were coordinated to multiple BP ligands to form BP-M NP cross-links for establishing dynamic hydrogel networks. Interestingly, the G’ and G” values were dependent on the diameter of the NP cross-links (Figure 9E,F). The moduli of the hydrogels decreased from the Mg²⁺ system to the Ba²⁺ system as the size of the BP-M NPs increased. For the TM-based NPs, the HA-BP-Co hydrogels, cross-liked with the smallest NPs, showed the largest G’ and G” values. For the HA-BP-M hydrogels, at a high shear strain, G’ decreased dramatically to a level below that of G”, revealing a shear-thinning behavior of the hydrogels. The excellent recovery ability was also observed using oscillatory strain experiments (Figure 9G). This work provides a useful strategy for controlling the dynamics of metal-ligand coordinated hydrogels. Considering their printability as bioinks, the rheological and mechanical properties of these hydrogels may be easily optimized depending on the class of metallic ions.

Figure 9.

(A) Schematic illustration and molecular structure of bisphosphonate-modified hyaluronic acid (HA-BP). (B) Efficient in situ self-assembly of BP-M NPs surrounding grafted BP groups of HA-BP macromers via BP-M coordination. (C) Hydrogel networks are stabilized by the BP-M NPs. (D) Representative elements (AEM and TM) and diameters of corresponding divalent cations (numbers under the name of elements, unit: pm) used for hydrogel fabrication. (E,F) Representative oscillatory rheological analysis results of the nanocomposite hydrogels prepared with a series of AEM ions and TM ions, respectively. All hydrogels were prepared with identical concentrations of ions, free BP, and HA-BP. (G) Rheological data for the HA-BP-M nanocomposite hydrogels under alternating high (20%) and low shear (0.1%). Reprinted with permission from ref¹⁹⁰. Copyright 2019 WILEY-VCH.

5.3.2. Carboxylates.

Coordination between carboxylates and metal ions has also been diversely considered to prepare dynamic physical hydrogels.^201-205 Zhang et al. reported on a key role for the coordination of ferric ions (Fe³⁺) and carboxylates in the gelation of UPy-functionalized gelatin.¹⁴⁷ The formation of shear-thinning hydrogels was dominated by the Fe^3+-carboxylate coordination, and the quadruple hydrogen bonds formed by each UPy moiety improved the mechanical strength and self-healing ability by providing additional physical cross-linking. Shao et al. developed physically cross-linked networks of poly(acrylic acid) (PAA)-based nanocomposite hydrogels through hydrogen bonding and dual metal-carboxylate coordination interactions.²⁰⁶ The Fe³⁺ ions and oxidized cellulose nanofibrils (CNFs) with carboxylates mainly acted as cross-linkers to produce the shear-thinning and self-healing poly(acrylic acid) (PAA)-based hydrogels. The Fe³⁺ ions formed three types of metal-ligand coordination bonds in PAA-PAA, PAA-CNF, and CNF-CNF. As the Fe³⁺ concentration increased, the dynamic moduli and the stiffness of the physical hydrogels increased due to the synergetic relaxation of multiple coordination bonds.

Liu et al. described metal-ligand coordination-triggered hierarchical folate/zinc supramolecular hydrogels.²⁰⁷ The folate molecules were found to form tetramers with a planar structure through intermolecular hydrogen bonding, and the subsequent π-π stacking interactions between tetramers formed the supramolecular fibrous structures (Figure 10A). This assembly behavior of folate is based on the liquid crystal-inducing pterin ring structure.^208-210 The addition of Zn²⁺ ions induced the cross-linking of stacked tetramers to build the fibrillar structure through carboxylate-Zn²⁺ interactions, resulting in the hydrogel formation. The hydrogels were formed by simple mixing of aqueous solutions of folate and Zn²⁺ ions in an alkaline pH range (pH > 10) (Figure 10B). The G′ value was largely dependent on the molar ratio of folate/Zn²⁺ and was increased by approximately 5 orders of magnitude (from 10 to 10⁶ Pa) when the ratio increased from 1.7 to 2.0. This result showed that the increased density of cross-linkers greatly improved the mechanical properties of the hydrogel. The multiple long-range interactions between Zn²⁺ and the hierarchical folate structures resulted in the hydrogel showing shear thinning behavior (Figure 10C). Continuous step-stain measurements showed the excellent recovery ability of the folate/Zn²⁺ hydrogels (Figure 10D). Moreover, the hydrogel could be injected and printed, into various self-supporting 3D structures with good shape fidelity.

Figure 10.

(A) Illustration of the hierarchical self-assembly process during gelation and the Printed hollow triangle structure. (B) Gelation in the folate/Zn²⁺ system. (C) Viscosity of folate/Zn²⁺ hydrogel at an increased shear rate followed by reverse shear rate decrease in continuous flow experiments. (D) Continuous step-stain measurements, which were carried out in steps of 50 and 0.5% oscillatory strain for four cycles. (Hydrogels used in (C) and (D) were [folate] = 15 mM, folate/Zn²⁺ = 1/1.8.). Reprinted with permission from ref²⁰⁷. Copyright 2018 American Chemical Society.

5.3.3. Catechols.

Catechol-metal coordination bonds have generated a variety of dynamic hydrogels with self-healing properties.^211-215 Among various catechol-metal combinations, catechol-Fe³⁺ complexes have been the primary focus in the development of supramolecular hydrogels, because a vast number of studies elucidated their formation and dynamics through intensive research on the cuticle of mussel byssal threads, whose hardness and extensibility are largely dependent on catechol-Fe³⁺ coordination bonds.^216-221 Hence, catechol-Fe³⁺ complexes have been an important choice as reversible cross-links to produce diverse classes of dynamic polymeric networks.^211,222-225 Li et al. prepared a dynamic hydrogel by incorporating iron oxide (Fe₃O₄) nanoparticles into a catechol-modified 4-arm PEG matrix through cross-linking by reversible catechol-Fe³⁺ coordination bonds on Fe₃O₄ nanoparticle surfaces.²²⁶ The presence of Fe₃O₄ nanoparticles resulted in dynamic rheological and mechanical behaviors, which are normally observed with supramolecular physical hydrogels. The catechol-Fe³⁺ coordination on the nanoparticle surfaces provides a solid-like reversible injectable hydrogel behavior.

Azevedo et al. developed chitosan-based double-network (DN) or doubly cross-linked (DC) hydrogels formed from noncovalent catechol-Fe³⁺ coordination interactions and chemical bonds (Figure 11).²²⁷ The main concept of this work is similar to that of mussel-inspired doubly cross-linked tissue adhesives, as reported by Fan and co-workers.²²⁸ They used 3,4-dihydroxyphenylalanine (DOPA)-conjugated chitosan (DOPA-CHT) as the main building block for hydrogel formation. The DOPA groups bearing catechol groups act as cross-linking points for network formation through catechol-Fe³⁺ coordination bonds.²²⁹ The dynamic nature of catechol-Fe³⁺ coordination bonds provided the hydrogel with a self-healing ability, which can allow the hydrogel to rapidly recover its initial mechanical properties. The additional chemical cross-linking by genipin mechanically improved the stiffness of the hydrogels. These hydrogels could be formed using a one-pot injection system. A dynamic oscillatory rheology study indicates the good self-healing ability of the hydrogels. In terms of printability, this property needs to be addressed to optimize the injection time period of the hydrogel precursors in a mixed state. An injection speed that is too fast may result in dispersion on the printed surfaces without forming 3D constructs with shape fidelity, while an injection speed which is too slow may induce poor injectability due to premature chemical cross-link formation.

Figure 11.

Schematic representation of the method developed to fabricate the DC and DN hydrogels. Besides the DOPA-CHT, GP, and Fe³⁺ ions, DN hydrogels are also composed of MMW-CHT. The inset depicts a picture of the obtained hydrogels (DC and DN showed a similar appearance). Two crosslinking processes were employed to produce the hydrogels, namely a covalent cross-linking using GP and a physical cross-inking through coordination bonds in the presence of Fe³⁺ ions. Therefore, DC hydrogels can be composed of bis- and tris-complexes Fe:DOPA-CHT (a) or/and covalent bonds between two DOPA-CHT chains (b). Additionally, DN hydrogel can also establish covalent bonds between a chain of DOPA-CHT and a chain of M_MW-CHT (c) or/and between two chains of MMW-CHT (d). Reprinted with permission from ref²²⁷. Copyright 2017 WILEY-VCH.

Liu et al. developed self-healing hydrogels with doubly cross-linked networks and demonstrated the improvement of mechanical strength of wheat gluten hydrogels by in situ catechol-Fe³⁺ coordination.²³⁰ Gluten molecules formed the first cross-linked network through their intermolecular physical interactions and the second network was formed through coordination interactions between Fe³⁺ and proanthocyanidins (PACs). PACs, a group of polyphenol compounds, acted as catechol-rich polymers and can form metal-catechol coordination complexes with various metal ions.^223,231 UV-Vis spectroscopic analyses showed that bis- and tris-catechol-Fe³⁺ complexes were identified in hydrogel matrices. Oscillatory strain experiments demonstrated that the hydrogels exhibited good self-healing properties. Dynamic shear rheology experiments showed that the hydrogels exhibited stronger mechanical properties at the optimized condition for hydrogel preparation (0.5 mM Fe³⁺ and 2 mM PACs (pH 6.0) than those obtained at other conditions. Similar to DOPA, PACs can also be conjugated to polymer chains to produce catechol-Fe³⁺-crosslinked networks of various dynamic hydrogels.

5.3.4. Histidine.

Histidine (His)-metal coordination is one of the key interactions in mechanically stiffening natural materials, such as sandworm jaws and mussel byssus.^232,233 For example, it was reported that the presence of dynamic His-Zn²⁺ coordination bonds was significantly involved in the dynamic course of mussel byssal threads, in which His-Zn²⁺ bonds were broken during thread yield and are recovered immediately upon thread relaxation.^234,235 These His-metal coordination bonds have been employed as reversible cross-links to produce diverse self-healing hydrogels.^236-238 The rheological and mechanical properties could be modulated by using different transition metal ions²³⁷ or in situ control of the metal oxidation state.^239,240 Grindy et al. demonstrated that His formed complexes with transition metal ions (such as Ni²⁺, Cu²⁺, and Co²⁺), which were cross-linked with the His-conjugated 4-arm PEG chains to produce supramolecular hydrogels (Figure 12A).²⁴¹ The viscoelastic properties of the hydrogels were controlled under ultraviolet (UV) irradiation in the presence of photoinitiator, LAP (lithium phenyl-2,4,6-trimethylbenzoylphosphinate). His-Cu²⁺-coordinated hydrogels lost their initial rigidity, whereas the increased stiffness was found for His-Co²⁺-complexed hydrogels (Figure 12B). For His-Ni²⁺ cross-linked hydrogels, the initial viscoelastic properties were maintained. These results indicate that the His-metal ion complexes exhibited different reactivities for the radicals generated by photo-triggered dissociation of LAP. Enke et al. demonstrated that the type of counter ion of the zinc salt tuned the mechanical and self-healing properties of His-Zn²⁺-based hydrogels.²⁴² Using a RAFT polymerization technique, a His-bearing acrylamide/acrylate copolymer was synthesized and complexed with Zn²⁺ ions, which were ionized from ZnCl₂ and Zn(OAc)₂. When using Zn(OAc)₂ for hydrogel preparation, the stoichiometry of the His:Zn²⁺ was found to have one more His than when using ZnCl₂. From the chelation of more His moieties to Zn²⁺ ions, the presence of acetate counter anions produced hydrogels with improved moduli and hardness compared to that with chloride counterions.

Figure 12.

(A) 4-Arm PEG-His forms hydrogel networks with Ni²⁺, Cu²⁺, or Co²⁺ ions based on His-M²⁺ coordination complexes formation. (B) Viscoelastic properties of PEG-His-M²⁺ hydrogels controlled by longwave low-intensity UV irradiation (365 nm). Reprinted with permission from ref²⁴¹. Copyright 2017 The Royal Society of Chemistry. (C) Schematic showing the protein network assembled by SpyTag/SpyCatcher chemistry and Co/His6-tag coordination. AA, SpyTag-ELP-SpyTag. BB, SpyCatcher-ELP-SpyCatcher. ELP, elastin-like polypeptide. (D, E) Dynamic frequency sweep tests on the products of AA + BB + Co²⁺ and AA + BB + Co²⁺ + NaIO₄ with a fixed strain of 5%. The concentration of Co²⁺ is 3 mM. The insets show the products of AA + BB + Co²⁺ in the absence and presence of NaIO₄. Reprinted with permission from ref²³⁹. Copyright 2019 American Chemical Society.

Recently, Kou et al. developed recombinant protein-based hydrogels based on His-cobalt coordination interactions.²³⁹ Two His6-tagged telechelic proteins (His6-SpyTag-ELP-RGD-ELP-SpyTag (AA) and His6-SpyCatcher-ELP-RGD-ELP-SpyCatcher (BB)) were synthesized in Escherichia coli.²⁴³ His6-tags in the proteins acted as cross-linkers through chelation to Co²⁺ ions (Figure 12C). The resulting hydrogel lacked stiffness, showing a very low G’ value (~20 Pa), close to that of G″ (Figure 12D). Interestingly, the oxidation of Co²⁺ into Co³⁺ by NaIO₄ rapidly resulted in the formation of an elastic solid with a greatly improved G′ value of ~1.4 kPa (Figure 12E). Wegner et al. also reported that 4-arm His-modified PEG could be cross-linked with cobalt ions, and the in situ transition of the oxidation state of cobalt ions from Co²⁺ to Co³⁺ switched the gel state from a viscoelastic liquid to a highly elastic solid.²⁴⁰ Taken together, the rheological and mechanical properties of His-metal coordination-based hydrogels can be tailored to develop novel bioinks with attractive printability for 3D bioprinting.

6. CHEMICAL STRATEGIES: DYNAMIC COVALENT BONDS

While physically cross-linkable hydrogels have many advantages for microextrusion bioprinting, the mechanical stability of the printed hydrogel-based constructs is usually low, and this weak resistance to deformation frequently requires post-treatment for stabilizing 3D tissue constructs. In contrast, chemical hydrogels cross-linked with a permanent covalent bond are mechanically stronger, but the network is irreversible once the cross-links are formed.¹¹⁷ This nondynamic characteristics of chemical networks limit their applications as bioinks for 3D bioprinting. Current fabrication approaches of 3D tissue constructs via chemical gelation are normally based on the extrusion of mixed solutions containing two-component complementary reactive polymeric species or light/temperature-mediated polymerizable compounds.^244-250 Because the chemical networks form during the extrusion mixing state, it is critical to consider the various factors that affect the gelation behavior, including the homogeneity in the mixed solution as well as the mixing rate.¹²⁵ Excessively rapid cross-linking results in the clotting of the nozzles, whereas excessively late cross-linking induces spreading after extrusion and thus cannot produce self-standing 3D constructs.

Therefore, dynamic covalent bonds have attracted great attention in the generation of novel classes of hydrogels.^117,251 Dynamic covalent bonds are exceptional chemical bonds and are in a constant equilibrium state between bonded and dissociated states.¹²² It is noteworthy that these bonds combine the strength of chemical bonds with the reversible properties of physical interactions, which can display the dynamic rheological behavior.²⁵² Chemistries for network formation through dynamic covalent bonds involve the formation of imine, hydrazone, oxime, disulfide, and boronic ester bonds, as well as Diels-Alder cycloaddition.^253,254 Hydrogels cross-linked with these dynamic bonds have advantages over those with permanent covalent cross-links in terms of fabrication of cell-laden 3D constructs because the dynamic rearrangement of these bonds allows cells to readily move and spread into the hydrogel structure.

Aiming toward new design approaches for the dynamic hydrogels, diverse chemical strategies have been suggested based on supramolecular assembly, macromolecular chemistry, colloidal nanoparticle-incorporated composite approaches, and so on.^119,255,256 A number of studies have afforded vast information on the structure-property relationship that determines the rheological properties of dynamic hydrogels. Thus, the diversity of design approaches may offer new insight into ideal bioinks with tunable rheological properties. In addition, newly developed design strategies may break down the limited scope of conventional hydrogels and may pave the way for the design of ideal bioinks for 3D bioprinting applications. In this review, we focus on dynamic hydrogels that exhibit their inherent properties at physiological pH and temperature. Hence, dynamic covalent bond-based dynamic hydrogels for which exchange reactions are activated by external stimuli, such as pH or UV light, are not discussed in this section. Examples of these dynamic covalent bonds include phenylboronate-salicylhydroxamate bonds,²⁵⁷ phenylboronate-catechol bonds,²⁵⁸, and reshuffling reactions of trithiocarbonate units.²⁵⁹

6.1. Imine Bonds

The formation of a Schiff base commonly occurs through a reaction between a primary amine and an aldehyde to produce an imine linkage under mild physiological conditions. The imine bonds can be reversibly hydrolyzed to return to the starting aldehyde and amine reactants;²⁶⁰ they are in equilibrium, showing the dynamics between the bonded and dissociated states. This dynamic nature of the imine bonds has been used to produce a number of injectable and self-healing hydrogels.^261-264 For imine-based hydrogels, two different polymeric species with a primary amine or aldehyde are reacted in an aqueous phase. For example, a combination of amine-bearing chitosan and dibenzaldehyde-functionalized PEG (DF-PEG) with terminal benzaldehyde groups have been frequently selected to produce imine-based dynamic hydrogels.^264-266 Wang et al. reported on a dynamic hydrogel based on glycol chitosan (GC) and DF-PEG as a cross-linker.²⁶⁷ The simple mixing of aqueous solutions of GC and DF-PEG resulted in hydrogel formation within 1 min. The hydrogels could be injected through a syringe needle (shear-thinning behavior) and had self-healing properties, which were ascribed to the dynamic equilibrium of the imine bonds between the bound and unbound state.^268-270 This hydrogel system showed tunable mechanical properties by varying the composition of GC and the DF-PEG cross-linker which modulated the cell proliferation behavior. They also suggested the versatile utility of this hydrogel by showing its potential as an injectable depot for cancer chemotherapy.^271,272 Xie et al. demonstrated excellent self-healing properties of the GC/DF-PEG hydrogels via a continuous step strain (from 1% to 300% and then 300% to 1%) evaluation.²⁷³ At a higher dynamic strain of 300%, the G’ value decreased from 1.2 kPa to 15 Pa due to the dissociation of dynamic imine linkages.²⁶³ The G’ value was rapidly recovered to its initial level as the strain changed from 300% to 1%. Qu et al. reported that the hydrogel formed from N-carboxyethyl chitosan (CEC) and DF-PEG showed injectable and self-healing characteristics.²⁷⁴ Zhao et al. developed another class of injectable self-healing hydrogels by reacting quaternized chitosan-g-polyaniline with benzaldehyde-functionalized PEG-co-poly(glycerol sebacate) under physiological conditions.²⁷⁵ The incorporation of electroactive polyaniline segments enabled the hydrogel to exhibit not only wound healing activity, but also antioxidant activity. Generally, imine-based hydrogels have normally been built on the simple mixing of aldehyde-functionalized polymers and primary amine-bearing polymers. Control of the rheological behavior is mainly dependent on the composition of each polymer, whereas there is a lack of novel chemical strategies for the fabrication of imine-based hydrogels.

6.2. Hydrazone Bonds

Imine bonds are known to be susceptible to hydrolysis, which may be recognized as a weakness in terms of maintaining in vivo structural and mechanical stability of hydrogels.²⁵⁴ Hydrazone bonds, as a close relative to imine bonds, are formed by reactions between aldehyde and hydrazide and show improved hydrolytic stability compared to that of imines. Hydrazone formation is based on a “click reaction”²⁷⁶ and has been extensively pursued to form diverse hydrogels with dynamic cross-linked networks.^277-281 Wang et al. demonstrated injectable hydrogels that undergo a first network formation through hydrazone bond formation by the reaction between hydrazine-modified elastin-like protein (ELP-HYD) and aldehyde-modified HA (HA-ALD) (Figure 13A,B).²⁸² Upon heating to physiological temperature (37°C), thermoresponsive phase segregation of ELP served as an additional physical cross-linking, which reinforced the network structure. The G’ value of the ELP-HA hydrogel becomes approximately 1000 Pa less than 10 sec after the mixing of ELP-HYD and HA-ALD (Figure 13D). The hydrogels from ELP-HYD with a larger portion of ELP domains showed a higher increase in the storage modulus (Figure 13E). The injectability and self-healing properties after injection through a 28G needle were supported by the shear-thinning property of the ELP-HA hydrogels at physiological temperature (Figure 13C,F). Due to the reformation ability of dynamic covalent hydrazone bonds, the hydrogel networks exhibited rapid and reversible recovery behavior, which was shown by alternating the shear rate between 0.1 and 10 s⁻¹ (Figure 13G). The ELP-HA hydrogels may meet the requirements of a bioink with good printability. Their rheological properties can be further modulated by adjusting the degree of cross-linking formed from the dynamic covalent hydrazone bonds.

Figure 13.

(A) ELP-HA is composed of hydrazine-modified elastin-like protein (ELP-HYD) and aldehyde-modified hyaluronic acid (HA-ALD). (B) Schematic of ELP-HA hydrogel formation. (C) Photographs demonstrating the injectability and rapid self-healing of ELP-HA hydrogels. (D) Oscillatory time sweep of ELP-HYD (4 wt%) and HA-ALD (2 wt%) before mixing and after mixing to form the ELP–HA (2, 1 wt%) hydrogel. Storage modulus (G′) shown with filled symbols and loss modulus (G″) shown with empty symbols; tested at 25 °C. (E) Oscillatory time sweep of ELP–HA hydrogels for 5 min at 25°C and 5 min at 37°C. (F) ELP-HA hydrogel viscosity as a function of shear rate at 37°C under continuous flow. (G) Shear-thinning and self-healing behavior of ELP–HA hydrogel under alternating shear rates of 0.1 and 10 s⁻¹ at 37°C. Reprinted with permission from ref²⁸². Copyright 2017 WILEY-VCH.

Wang and Burdick et al. also developed an injectable hydrogel cross-linked through dynamic covalent hydrazone bonds and estimated its applicability as a bioink for 3D bioprinting.²⁸³ Upon mixing aqueous solutions of HA-hydrazide and HA-ALD, hydrogels that are resistant to flow were formed through hydrazone-based network formation. Contrary to physical cross-links that often undergo hydrogel relaxation over time (several minutes or hours),^153,284 the cross-links formed by dynamic covalent bonds normally have higher strength. In addition, the equilibrium of hydrazone strongly favors the bound state.²⁸⁰ The hydrogel showed shear-thinning ability similar to that of previously reported hydrazone bond-based hydrogels²⁸² and could be ejected through a 27G × 1/2 in. syringe. Shear oscillatory rheometry studies showed that rapid self-recovery behavior, indicating the self-healing ability of the hydrogels. Additional photo-stiffening using thiol-ene-based interpenetrating networks, which was obtained by incorporating HA-norbornene and tetrathiol cross-linkers, increased the modulus of the hydrogel constructs up to ~300%. Li et al. reported on a dynamic hydrogel (HA-az-F127 hydrogel) formed by the reaction between hydrazide-modified HA (HAAD) and benzaldehyde-functionalized F127 triblock copolymer (BAF127) micelles.²⁸⁵ The BAD127 copolymer underwent thermoresponsive self-assembly to form core-shell type micelles, surface benzaldehyde groups of which could react with hydrazide moieties in the HA backbones to form dynamic acylhydrazone bonds (Figure 14A). The micellization of BAF127 is a key factor contributing to improving the mechanical strength of the HA-az-F127 hydrogel because HA-az-TPEG (the four-armed PEG with benzaldehyde terminals), which lacks the micelle-forming ability, shows lower strength (Figure 14B). The micellization of BAF127 resulted in the injectable dynamic hydrogel exhibiting shear-thinning properties (Figure 14C), and the oscillatory strain change shows that Gel3 exhibited rapid recovery and excellent self-healing properties. This reflects that the dynamic acylhydrazone covalent linkages dynamically underwent switching between the uncoupling and recoupling state in the hydrogel matrix (Figure 14D). When two individual half disks of the hydrogel, stained with Rhodamine B and methylene blue, were brought into contact, they combined into one piece of hydrogel with good extensibility under tensile force and showed excellent recovery ability to the original dimension after removal of the strain.(Figure 14E) The researchers suggested the potential of using this acylhydrazone-based dynamic hydrogel as an injectable material for skin burn wound healing due to its tissue adhesiveness and good absorption properties.

Figure 14.

(A) Double cross-linked HA-az-F127 hydrogel. (b) Rheological results on (A) time sweep tests. (B) Temperature sweep tests in the temperature range of 4-60 °C at a frequency of 1 Hz, (G3 is the hydrogel with solid content of 15 wt% and the ratio of 5:5 (hydrazine to aldehyde)). (C) Viscosity as a function of shear rate. (D) G′ and G″ of Gel3 from the continuous step strain measurements (1% → 300% → 1%) under 1 Hz. (E) Photographs of self-healing process: (e1) as-formed hydrogel; (e2) two halves of the hydrogel in PTFE mold; (e3, e4) self-healed hydrogel after 30 min; (e5) stretching with forceps; and (e6) recovered shape. Reprinted with permission from ref²⁸⁵. Copyright 2018 American Chemical Society.

6.3. Oxime Bonds

As another dynamic covalent bond similar to imine bonds, oxime bonds are formed through a “click reaction” between hydroxylamine/alkoxyamine and aldehyde or ketone.^276,286,287 Compared to that of imine and hydrazone bonds, the reaction equilibrium greatly favors bound oxime linkages and thus the hydrolytic stability of oxime bonds is known to be much higher.²⁵⁴ Oxime bonds have also been used for the formation of reversible cross-links to prepare diverse dynamic hydrogels.^288-291 Most frequently, studied oxime-based cross-linked hydrogels are based on aldehyde-bearing oxidized HA or alginate, which are cross-linked by bifunctional alkoxyamine compounds. Baker et al. reported on cell-friendly tissue mimetic hydrogel constructs from bioorthogonal oxime bond formation between HA-aldehyde and PEG bis(oxyamine).²⁹² Wang et al. examined the catalytic influence of monovalent or bivalent salts on oxime reaction kinetics to obtain dynamic HA hydrogels. They found that the addition of monovalent or divalent salts to the mixed solution of HA-ALD and HA-aminooxy significantly reduced the gelling time compared to that of salt-free buffer solution.²⁹³ Hafeez et al. prepared oxime-cross-linked injectable shear-thinning hydrogels through the reaction of aldehyde-bearing alginate with O,O’−1,3-propanediylbishydroxylamine.²⁹⁰ Nadgorny et al. prepared an oxime-based hydrogel through a cross-linking reaction between poly(n-hydroxyethyl acrylamide-co-methyl vinyl ketone) (PHEAA-co-PMVK) and tetraethylene glycol bishydroxylamine (TEG-BHA).²⁹⁴ The Schiff base reaction of bifunctional hydroxylamines in TEG-BHA with carbonyl groups in PHEAA-co-PMVK forms stable oxime linkages, which strongly hold the hydrogel networks (Figure 15A). The oxime-based hydrogels showed shear thinning behavior, which is suitable for 3D printing (Figure 15B). These dynamic hydrogels can be stably extruded under shear force (Figure 15C). The 3D-printed oxime-based hydrogels were treated by thermally induced phase separation (TIPS), which facilitated the formation of hydrogen bonds and oxime cross-links. The physical cross-linking through cryogelation-induced hydrogen bonding efficiently reinforced the oxime-based dynamic hydrogels. This post-treatment after 3D printing improved the mechanical strength of the printed oxime-cross-linked hydrogels by up to ~1900%. Notably, the hydrogels exhibited an immediate recovery from destruction under high strain (Figure 15D). When two separate pieces of the hydrogels are in contact, an excellent self-healing property is observed (Figure 15E-G). Ghosh et al. reported on the synthesis of star-PEO-based hydrogels by cross-linking with oxime bonds.²⁹⁵ End group-functionalized PEO₈-benzaldehyde (PEO₈-PhCHO) and PEO₈-oxyamine (PEO₈-ONH₂) were reacted to form tough hydrogel materials. The rheological and mechanical properties were proportional to the cross-linking density of the hydrogels. Although the main purpose of this work did not include studying the dynamic behavior, hydrogels that showed shear-thinning and self-healing properties could be fabricated by control of the reaction parameters, such as the cross-linking density, the unit ratio of aldehyde:oxyamine, and the mixing composition of each functional polymer for hydrogel preparation.

Figure 15.

(A) A schematic of the methodology used to form 3D-printable, doubly dynamic self-healing cryogels. The inset demonstrates the chemical structure of the cryogels and their macroporous morphology. (B) Apparent viscosity curves of oxime-based hydrogels 25°C (black squares) and 80°C (red triangles). The gel exhibits shear thinning behavior, suitable for 3D printing. (C) A stable, shape-retaining extruded filament of oxime-based hydrogels. (D) Self-healing rheology of oxime-based hydrogels. (E) 3D-printed oxime-based hydrogel scaffold after 3 cryogelation cycles at −10°C and induction of macroscopic cuts (a representative cryogel is demonstrated). (F) The damaged halves of the scaffold were brought into contact, to facilitate the healing process. (G) The damaged scaffold recovered from the cut and could be lifted as a single self-supporting unit after ~3 h. Reproduced from the licensed article²⁹⁴ of a Creative Commons Attribution 3.0 Unported License.

6.4. Disulfide Bonds

The disulfide bond is another class of dynamic covalent bonds that can generate the dynamic hydrogels with fast gelation kinetics. At neutral or alkaline pH, disulfide bonds are broken and reformed by the presence of free thiol groups, which produces dynamic and reversible hydrogel networks.^296-298 The dynamics of these hydrogels are usually dependent on thiol/disulfide exchange reactions, which are sensitive to high temperature, UV irradiation, or mechanical stress.¹²² Yu et al. reported on the formation of thermosensitive dynamic hydrogels by simply mixing thiol-functionalized Pluronic F127 and dithiolane-modified PEG.²⁹⁹ The thiol/disulfide exchange reaction is a key mechanism for dynamic network formation. The hydrogels as-prepared could be injected through a syringe needle and showed self-healing properties over a broad pH range from mildly acidic to alkaline conditions. The rheological properties of these hydrogels could be modulated by adjusting the mole ratio of thiol to dithiolane functional groups. Zhang et al. developed a self-healing hydrogel based on the self-assembly and gelation behavior of an ABA triblock copolymer containing a PEG middle block (B block) and terminal polycarbonate blocks (A block) with pendant 1,2-dithiolane groups.²⁸¹ This block copolymer self-assembled to form interconnected flower-type micelles that formed physically cross-linked networks. The dithiolanes in the core region of the micelles underwent thiol-initiated ring-opening polymerization to form dynamic and reversible networks (Figure16). The rheological and mechanical properties of these dithiolane-derived hydrogels could be modulated by varying the ratio of the dithiolane moieties in the block copolymer. Oscillatory shear rheometry measurements showed that the disulfide-cross-linked hydrogels exhibited rapid and complete self-recovery behavior in contrast to the micellization-based physically cross-linked hydrogel that demonstrated poor self-recovery behavior.

Figure 16.

Proposed gelation mechanism for dithiolane-containing triblock copolymers in the presence of thiols. The physical cross-linking via the bridging PEG chains was shown in red. The chemical cross-linking via the thiol-initiated ring-opening polymerization of dithiolanes was shown in blue. CMC: critical micelle concentration, N_agg: aggregation number, D_H: hydrodynamic diameter. Reprinted with permission from ref²⁸¹. Copyright 2017 American Chemical Society.

It is known that the thiol-disulfide exchange reaction suffers from uncontrolled aerial oxidation of thiols to disulfides. As a thiol/disulfide exchange reaction, the exchange between Au-thiolate species and disulfide at biological pH has also been a motif for diverse injectable self-healing hydrogels.^300,301 Pérez-San Vicente et al. developed an injectable dynamic hydrogel formed from an Au-thiolate/disulfide exchange reaction.^302,303 The purpose of using Au-thiolate species is to efficiently protect the nucleophilicity of thiolates, which is required for the exchange reaction with disulfide. Dynamic hydrogels were fabricated by mixing an aqueous solution of thiol-terminated 4-arm PEG with aqueous HAuCl₄. The aerial oxidation of thiol groups in PEG chains at alkaline pH was not sufficient for self-standing hydrogels. The Au(I) ions induced hydrogel formation by acting as a cross-linker for thiolated PEG chains. The hydrogel showed increased mechanical properties as the frequency increased at physiological pH, which was ascribed to the permanent Au-thiolate/disulfide exchange.^300,301 Rheology experiments showed that this metal ion-involved disulfide-based dynamic hydrogel exhibited injectability and fast self-recovery properties. Although the disulfide bond is employed as one type of dynamic covalent bond for hydrogel network formation, its chemistry requires the presence of an alkaline pH above 8.5^304,305 or strong oxidants, such as iodine and hydrogen peroxide.^305,306 The pKa of thiols is in the range of 8~10, and thus, basic conditions (pH > 8.5) were required to drive the reaction toward disulfide formation.³⁰⁴

Recently, Bermejo-Velasco et al. developed injectable disulfide cross-linked HA hydrogels by reducing the pKa of the thiol group through the incorporation of an electron-withdrawing group at the β-position of thiol in cysteine or N-acetyl-L-cysteine.³⁰⁷ Interestingly, the reduced thiol pKa facilitated the formation of thiolate anions at physiological pH, allowing the formation of oxygen-catalyzed disulfide-cross-liked hydrogels at physiological pH within 3.5 min. This work may suggest that the chemical approach of controlling the pKa of thiols has diverse utility in preparing various disulfide-based injectable (shear thinning) and self-healing hydrogels in a short time period at physiological pH without the need for oxidants.

6.5. Boronic Ester Bonds

The boronic ester bond is a popular dynamic covalent bond and has been used to produce diverse supramolecular assemblies and hydrogels.^308-311 This reversible bond is formed by the condensation of boronic acids with cis-1,2 or cis-1,3-diol compounds.^312,313 In particular, the reversible complexation of phenylboronic acid (PBA) (one example of arylboronic acids) and cis-diols has been widely employed as cross-links to form dynamic hydrogels.^314-316 Owing to having pKa values in the range of 8~9, arylboronic acids have a limitation with respect to producing stable boronic ester-based cross-links at physiological pH.^308,309 Thus, the optimum pH for the formation of boronic esters is known to be approximately 9.0. However, it was reported that there still exist ionizable boronic acid groups (approximately 6% of the total boronate moieties) sufficient for binding to cis-diols at pH 7.4.^317,318 Pettignano et al. demonstrated the formation of boronic ester-based alginate hydrogels without adding external diol compounds or divalent cations, such as Ca²⁺ ions.³¹⁴ PBA-conjugated alginate formed a hydrogel under basic conditions (pH 9.0). The hydrogel showed thixotropic behavior and self-healing properties. Hong et al. also prepared identical dynamic hydrogel at physiological pH.³¹⁹ When dissolving PBA-bearing alginate (2.3 wt%) in the aqueous media (pH 7.4), the hydrogel was formed within 1 min. The dynamic rearrangement of boronic ester bonds provides hydrogels with self-healing and shear thinning properties. These works show the advantage of using a single-component gelling material over two- or three-component systems, which often suffer from inhomogeneous mixing and inconsistent, batch-dependent gelling and rheological behavior.

One recent approach that modulates the rheological and mechanical properties of boronic ester-based hydrogels involves incorporating another dynamic cross-linking motif. For example, Li et al. reported on a double dynamic network hydrogel, which was formed through two dynamic covalent bonds, borate ester, and imine bonds.³²⁰ As a key component, the researchers synthesized multifunctional PEG (MF-PEG) with a benzaldehyde group and a PBA group at each chain terminus.³²⁰ MF-PEG formed a doubly cross-linked network through a borate ester-forming reaction with poly(vinyl alcohol) (PVA) and an imine-forming reaction with GC to produce a dynamic hydrogel in seconds under mild conditions (pH ≈ 7, 25°C). Notably, the control of various parameters, such as the ratios of PVA/GC, amine/aldehyde, and diol/PBA, could modulate the rheological and mechanical behavior of the hydrogels.

As discussed above, due to the pKa value of PBA (~8.8), efficient hydrogel formation is impeded at physiological pH. Moreover, considering in vivo use, the hydrogels may suffer from poor stability due to the dissociation of boronic ester bonds at physiological pH. For this reason, PBA analogs with pKa values close to physiological pH have been employed to endow hydrogels with feasible stability under physiological conditions.^312,321 Benzoxaborole is a cyclic analog of PBA with a pKa value lower than physiological pH (~7.2) and forms stable five-membered boronate rings through complexation with cis-diols.^322,323 Benzoxaborole has been a preferred choice for forming diverse dynamic injectable and self-healing hydrogels applicable in physiological conditions.^324,325 Chen et al. developed a self-healing hydrogel using benzoxaborole-catechol dynamic covalent chemistry.³²⁶ They synthesized two kinds of phosphorylcholine-containing copolymers, which have pendent benzoxaborole and catechol groups, respectively. Upon mixing aqueous solutions of these copolymers (PBS at pH 7.4), the hydrogel formed quickly within 30 sec, which was attributed to the pKa value of benzoxaborole (~7.2) being close to physiological pH.³¹² The hydrogel exhibited high stability even after 24 h immersion in PBS solutions, indicating that the equilibrium favored the complexed tetrahedral boronates at physiological pH. The hydrogel showed rheological behavior typically observed for dynamic hydrogels, exhibiting shear-thinning and self-healing characteristics. Compared to PBA-based dynamic chemistry, the use of benzoxaborole-bearing polymers for complexation with cis-1,2 or cis-1,3-diol compounds may allow the production of hydrogel bioinks with excellent shape fidelity and mechanical stability. Chen et al. designed a dynamic hydrogel exhibiting shear-thinning and self-healing characteristics based on a thermosensitive copolymer with pendent galactose groups (PLDL) and a benzoxaborole-containing copolymer (PAB) (Figure 17).³²⁷ Dynamic benzoxaborole-galactose complexation is a driving phenomenon for reversible network formation in a hydrogel matrix. Rheological analyses showed that the mechanical properties could be controlled by varying the galactose/benzoxaborole molar ratio. The hydrogels exhibited excellent self-healing ability, which could be explained by the dynamic nature of the boronic ester bonds. Step-strain experiments demonstrated that the rapid recovery of the hydrogel network. The viscosity of the hydrogel decreased with increasing shear rate, indicating a shear-thinning property desirable for injectable bioinks in 3D printing applications. The same research groups further identified the parameters that control the mechanical properties of similar benzoxaborole-galactose complex-based dynamic hydrogels.³²⁸ By controlling the sugar content, the cross-linking density in the hydrogel was modulated, which tuned the mechanical strength of the hydrogels. The G′ value could be controlled in the range of 261.2 Pa (20% sugar at pH 7.4) ~ 2638.1 Pa (80% sugar at pH 9.4), suggesting a potential for designing bioinks with tailored mechanical properties optimized for 3D bioprinting.

Figure 17.

(A) Schematic illustration of injectable, self-healing, and multi-responsive hydrogel cross-linked by benzoxaborole-galactose complexation. (B) Chemical structures of PLDL (PLAEMA-b-PDEGMA-b-PLAEMA: poly(2-lactobioamidoethyl methacrylamide-b-di(ethylene glycol) methyl ether methacrylate-b-2-lactobioamidoethyl methacrylamide)) and PAB (P(AAm-st-MAABP): poly(acrylamide-st-5-methacrylamido-1,2-benzoxaborole)). (C) Images showing the self-healing process by reconnecting two pieces of hydrogel together. (D) Storage modulus and loss modulus value change in response to the imposed strain. (E) The corresponding strain variation: increasing first from 0.1% to 500% to break the gel and then dropping back to a small strain of 1% for gel recovery. (F) Shear-thinning property of the hydrogel. Reprinted with permission from ref³²⁷. Copyright 2019 American Chemical Society.

7. CHEMICAL STRATEGIES: NANOCOMPOSITE HYDROGELS

Another category of significant emphasis is nanocomposite hydrogels, which are fabricated by incorporating nanomaterials into polymeric materials. This hybrid strategy using nanomaterials is innovative in generating multifunctional hydrogels with diverse functionalities.^255,329,330 The latest studies show that most of these nanocomposite hydrogels exhibit shear-thinning rheological properties.²⁵⁶ The main modulator of this rheological property involves dynamic interactions of various physical bonds between the surface of nanomaterials and the hydrogels. Recently, various nanomaterials, including calcium carbon nanotubes (CNTs),^331-333 graphene oxide (GO) nanosheets,^334,335 calcium carbonate, Laponites,^336,337 silica nanoparticles, and gold and silver nanoparticles^338-340 are employed in either unmodified or modified state. The rational combination of these nanomaterials with native or functionally modified polymers may also provide a deep understanding regarding a wide variety of choices in combinations of chemical species for 3D printable hydrogel bioinks. Herein, we provide a critical review of the design approaches of the most recent hydrogel systems in terms of their applicability in 3D bioprinting. Although the surface chemistry of some polymeric nanoparticles, including polymer latexes and micelles,³²⁹ and cellulose nanocrystals,^341,342 has also been investigated for the formation of nanocomposite hydrogels, most chemical strategies have focused on the surface of inorganic nanomaterials due to the convenience and diversity in functionalization for developing nanocomposite hydrogels that exhibit shear-thinning and self-healing properties. In this section, we provide a detailed discussion on the role of inorganic nanomaterials exploitable for the novel design of printable bioinks for 3D bioprinting.

7.1. Nanocarbon-Based Hydrogels

Carbon nanotubes (CNTs) and graphene oxide (GO) nanosheets have been incorporated into physically cross-linkable hydrogels and have provided superior properties compared to those of the native hydrogels.^343-345 The addition of nanocarbon has modulated the rheological and mechanical characteristics of the nanocomposite hydrogels through a diverse single or multiple interactions between nanocarbon and organic low molecular weight compounds or polymers, which are displayed by π-π, hydrogen bonding, electrostatic, and van der Waals interactions, among others.^332,334,345

7.1.1. Carbon Nanotubes.

Mansukhani et al. demonstrated single-walled carbon nanotube (SWCNT)-based nanocomposite hydrogels based on optimized deoxyribonucleic acid (DNA) linker chemistry.³⁴⁶ They employed DNA base pairing as the cross-linking interaction in the formation of 3D networks in the nanocomposite hydrogels. The linker DNA strands consist of two regions: one is complementary to the SWCNT-adsorbed DNA (sequence S1), and the other is (CG) repeats of variable length (Figure 18A). A single adenosine (A) base is inserted between these two regions to enhance the flexibility of the linker strand.³⁴⁷ For hydrogel formation, SWCNTs were dispersed in an aqueous solution of single-stranded DNA containing 12 adenosine cytosine (AC) repeats attached to S1. Due to the self-complementary feature of the CG region, a DNA cross-linked network of SWCNTs was produced through duplex formation between the SWCNT-bound linker DNA strands (Figure 18B). An SEM image shows that the lyophilized SWCNT/DNA hydrogel has a porous structure typically observed for supramolecular hydrogels (Figure 18C). This CNT-based hydrogel shows the sol-gel transition (Figure 18D). A shear stress sweep showed that the viscosity of SWCNT/DNA hydrogels was 900 Pa s at 1.8 mg/mL. When a noncomplementary linker was used, the mixed dispersion of SWCNT/DNA did not form hydrogels, indicating that the duplex linker was a key interaction for network formation. The addition of SWCNTs produced nanocomposite hydrogels with shear-thinning properties (Figure 18E). The G’ value of the hydrogels could be controlled by the SWCNT concentration. A G’ value of 200 Pa was measured at an SWCNT concentration of 3.2 mg/mL (Figure 18F), and the total number of inter-SWCNT cross-links can be modulated as a function of the SWCNT concentration (Figure 18G). Moreover, the SWCNT/DNA hydrogels underwent near-infrared (NIR)-controlled thermoreversible gelation behavior, which can be displayed by the photothermal property of the SWCNTs.^348,349 Under NIR irradiation, the viscosity of the hydrogels significantly decreased, whereas recovery to the original gel state was observed upon cessation of NIR irradiation. This work is an excellent example of using the unique complementary interaction of DNA to produce SWCNT-based shear-thinning injectable hydrogels.

Figure 18.

(A) DNA sequence (AC)₁₂S1 (black) is used to disperse SWCNTs. The linker DNA (green) has two regions: one complementary to S1 to hybridize with the SWCNT and one self-complementary (CG) repeat of variable length, separated by a single adenosine base. (B) Proposed structure upon gel formation, wherein (AC)₁₂S1 helically wraps around the SWCNT while the linker DNA forms duplexes with both S1 and other linker strands, thereby providing the cross-linking mechanism. (C) Cryo-SEM of a lyophilized SWCNT-DNA gel, showing a network of SWCNTs amid aggregated buffer salts and DNA. (D) Photographs of the inverted-vial test to demonstrate the sol-gel transition (solution on left and gel on right). (E) Viscosity versus shear stress for a typical gel, compared to an ungelled sample. (F) Storage (G′) and loss (G″) moduli as a function of SWCNT concentration. Linker DNA concentration is increased in proportion to SWCNT concentration. Dashed black and red lines indicate moduli for a 5 mg mL−1 van der Waals gel. (G) Number of inter-SWCNT and intra-SWCNT cross-links in simulation, as a function of the number of SWCNTs for gel systems formed at slow and fast annealing rates. Reprinted with permission from ref³⁴⁶. Copyright 2018 WILEY-VCH.

Rehman et al. described CNT-incorporated poly(acryloyl-6-amino caproic acid) (PAACA)-based nanocomposite hydrogels.³⁵⁰ They used oxidized CNTs to form cross-links with PAACA chains through hydrogen bonding interactions between amide and carboxyl groups in PAACA side chains and oxidized CNTs. These CNT/PAACA hydrogels could be prepared without any organic cross-linkers. They exhibited self-healing recovery behavior. The presence of CNTs in hydrogel matrices increased the gel strength from 60 kPa (PAACA hydrogels) to 320 kPa (CNT/PAACA hydrogels). The higher the CNT amount, the stronger strength the hydrogel exhibited. Liao et al. developed the functionalized SWCNT-containing dynamic hydrogels with reliable self-healing capability and string adhesiveness.³⁵¹ The hybrid network of these hydrogels were formed by dynamic supramolecular cross-linking among functional SWCNTs, PVA, and polydopamine (PDA), where hydrogen-bonding and π–π stacking interactions were involved. The dynamic cross-links could be easily ruptured and reformed, imparting self-healing abilities to these hydrogels. The hydrogels exhibited rapid self-healing properties within 2 s and high self-healing efficiency of 99%.

7.1.2. Graphene-Based Nanomaterials.

Like CNTs, graphene oxide (GO) has been introduced to improve the mechanical and rheological properties of the hydrogel-based bioinks for 3D bioprinting. Liu et al. reported on the fabrication of a 3D printable and mechanically robust hydrogel based on alginate and GO nanosheets.³⁵² They focused on the key role of the alginate-functionalized GO (A-aGO) in modulating the mechanical and rheological properties of A-aGO/SA hydrogels. For the synthesis of A-aGO, GO dispersed in alginate-containing water was first aminated using the Bucherer reaction, which partially converts phenolic oxygens on the GO to primary amines (-NH₂).^353,354 The aminated aGO formed A-aGO with anionic alginate through both electrostatic and hydrogen-bonding interactions (Figure 19A-C). The A-aGO/SA physical hydrogel was fabricated by adding Ca²⁺ ions to the A-aGO/SA solution. A-aGO played a role as a co-cross-linking species to provide a hierarchically developed hydrogel structure. Rheological studies show that the A-aGO_0.2/SA_Ca-6 (where the subscripts indicate the weight percentages of A-aGO and CaCl₂, based on SA) hydrogel showed a shear-thinning property from 0.01 to 500 s⁻¹ (Figure 19D). It was found that, at a fixed shear rate, A-aGO_0.2/SA_Ca-6 hydrogel showed higher viscosity than that of the control groups, GO_0.2/SA_Ca-6 and SA_Ca-6 hydrogels. This result reflects the obvious enhancement effect of A-aGO. In addition, the A-aGO_0.2/SA_Ca-6 hydrogel was more sensitive to oscillation strain. The G′ value rapidly decreased when the strain increased to 3% and intersected with G″ at 23.5% (Figure 19E). This finding indicates that reversible dynamic absorption and desorption of SA chains from the A-aGO surface would impart the hydrogel with shear-thinning and excellent self-healing properties (rapid recovery of the gel structure). Under alternating oscillation strain, G′ and G″ recovered to 92% and 80% of their initial values (Figure 19F), showing much better reversibility in the transition from gel to the quasi-liquid state than that of the SA_Ca-6 hydrogel. To evaluate the capability for bioprinting, the A-aGO/SA hydrogel could be further chemically cross-linked using UV-mediated polymerization in the presence of acrylamide (AAm) and N,N’-methylene bisacrylamide (MBA), a cross-linker. The A-aGO/SA/PAAm nanocomposite hydrogel was fabricated into various 3D structures. Because the A-aGO/SA hydrogels have good printability based on their shear-thinning and self-healing properties suitable for 3D printing, the final UV-treated constructs had excellent structural resolution with good mechanical stability. For example, the A-aGO_0.2/SA_Ca-6/PAAm hydrogel could be printed to construct hollow pentagonal structures with a self-weight supporting ability (Figure 19G). In contrast, the identical structure constructed from SA_Ca-6/AAm and GO_0.2/SAC_a-6/AAm hydrogels showed low self-standing ability and quickly collapsed (Figure 19H,I). From a bioprinting point of view, the main idea of this work is the use of an approach to aminate GO in the presence of alginate before hydrogel formation. Using this chemistry, alginate was effectively adsorbed in situ on the surface of GO nanosheets through noncovalent interactions. This prevented A-aGO from agglomerating, thereby affording excellent colloidal stability in water, reflecting that A-aGO can be dispersed at the individual nano-level in the aqueous phase. Hence, the separately dispersed A-aGO could establish a strong cohesion through the co-cross-linking interactions between alginate and the A-aGO. The A-aGO could act as an efficient nano cross-linker for hydrogel formation and serve as a key modulating factor for the printability of A-aGO/SA gels. The rheological properties supported the significance of A-aGO/SA in controlling the printability for constructs with well-defined dimensions and morphologies. Considering the chemistry, if the degree of amination of the GO nanosheets can be controlled, the rheological and mechanical properties that determine the printability of the hydrogels may be more readily adjustable.

Figure 19.

(A-C) Schematic diagrams for fabrication of the A-aGO/SA/PAAm nanocomposite hydrogel: (A) Homogeneous aqueous solution of SA, A-aGO, acrylamide (AAm), N,N’-methylene bisacrylamide (MBA), and photoinitiator. (B) Ca²⁺ ions inducing ionic association of SA chains. (C) UV exposure resulting in the final A-aGO/SA/PAAm nanocomposite hydrogel. (D) Apparent viscosity as a function of shear rate for the A-aGO_0.2/SA_Ca-6 and SAC_a-6 hydrogels as well as the SA solution without Ca²⁺ ions. (E) Storage modulus G′ and loss modulus G″ as a function of oscillation strain at an angular frequency of 1 Hz for the A-aGO_0.2/SAC_a-6 and SAC_a-6 hydrogels as well as the SA solution without Ca²⁺ ions. (F) Changes of G′ and G″ with time at alternant oscillation strains of 2% and 300% and at an angular frequency of 1 Hz for the A-aGO_0.2/SA_Ca-6 hydrogel, where the red numbers represent the average modulus at the oscillation strain of 2%. (G) Hollow pentagon printed from the A-aGO_0.2/SAC_a-6/AAm hydrogel bioinks (15 layers) viewed from the top and side. 3D printed hollow pentagon patterns from (H) the SA_Ca-6/AAm hydrogel and (I) the GO_0.2/SAC_a-6/AAm hydrogel inks, respectively. Reprinted with permission from ref ³⁵². Copyright 2017 American Chemical Society.

Khabibullin et al. reported on injectable nanofibrillar hydrogels formed by the combination of cellulose nanocrystals (CNCs) and graphene quantum dots (GQDs).³⁵⁵ Amphiphilic rod-like CNCs (average length of 183 nm and diameter of 23 nm) were used as the main building block and had negatively charged surfaces due to the presence of half-ester sulfate groups (ζ-potential of −60 mV at pH 7.0).^356,357 Disk-shaped GQDs served as cross-linkers for 3D network formation with CNCs to prepare nanocomposite hydrogels. GQDs have an amphiphilic character because they exhibit hydrophobicity in their basal plane and hydrophilicity at the carboxylic acid group-containing edges.³⁵⁸ The nanocomposite hydrogels were formed by simply adding GQDs into the aqueous dispersion of CNCs. When combining these two nanospecies, three types of noncovalent interactions are involved in network formation; (i) hydrogen bonding interactions between the hydroxyl groups on the CNCs and the carboxylic acid groups of GQDs, (ii) hydrophobic interactions between the hydrophobic faces of the CNCs and the GQD basal plane, and (iii) electrostatic repulsion between the half-ester sulfate groups of the CNCs and the carboxylic acid groups of GQDs. Considering the identical anionic nature of the two nanomaterials, it is interesting to observe physical gel formation. As a mechanism for hydrogel formation, it has been suggested that interspecies hydrogen bonding and hydrophobic interactions were more dominant and thus could overcome the electrostatic repulsive forces. Rheological data showed that, at >10% strain, the CNC-GQD nanofibrillar hydrogels exhibited a nonlinear viscoelastic behavior caused by the disruption of the physically cross-linked network.³⁵⁹ The mechanical properties of the hydrogels could be controlled by adjusting the amount of either CNCs or GQDs. For example, at C_CNC = 50 mg/mL, the G′ value for the gel was 266 Pa when C_GQD was 10 mg/mL. Meanwhile, at C_GQD = 5 mg/mL, G′ was decreased by approximately 5-fold (54 Pa). This observation indicates that a higher concentration of GQD cross-linkers resulted in the formation of stronger networks. In addition, the higher the CNC content was, the stronger the hydrogels that were obtained. More importantly, the resulting CNC-GQD hydrogels exhibited shear-thinning behavior, which is desirable for extrusion-based 3D bioprinting applications. This viscosity-decreasing behavior under shear force was verified by observing the G’ value under oscillation strain from 1 to 50% strain. For the hydrogel with G’ = 80 Pa, the G′ value decreased to 2 Pa upon decreasing the strain to 50% and dropped below the G″ value, reflecting gel liquification. The G′ value recovered completely within 12 min at the end of each cycle. To evaluate the printability of the hydrogels, they were extruded from a syringe to form a threadlike shape with a predesigned pattern. Notably, the GQD-containing 3D hydrogel is attractive because it can exhibit photoluminescence properties. In addition, the extruded hydrogel exhibited birefringence, as visualized by polarized optical microscopy. This phenomenon is probably due to the shear-induced alignment of the CNCs during extrusion. It was suggested that the anisotropic structure of the hydrogels may mimic the anisotropic morphology of various tissues, such as cartilage and cornea, which is crucial for cell guidance, proliferation, and differentiation.

A GO-based supramolecular hydrogel can be formed by host-guest interactions between α-cyclodextrin (α-CD) and PEG-functionalized GO nanosheets. Zhang et al. used pyrene-conjugated PEG (Py-PEG).³³⁴ Py-PEG chains were tethered on both sides of GO surfaces through π-π stacking interaction between GO and polyaromatic pyrene moieties of Py-PEG. The addition of α-CD into the Py-PEG-modified GO induced hydrogel formation through the formation of an inclusion complex of α-CD and PEG. This PEG/CD complexation generated a 3D network consisting of alternating layered structures of GO nanosheets and PEG/CD inclusion complexes. In this supramolecular assembly, the GO nanosheets acted as physical cross-linkers to provide the hydrogel with mechanical stability. The GO played a crucial role not only in promoting the formation of the supramolecular nanocomposite hydrogel but also in increasing the mechanical strength of the resulting hydrogel. The G’ value of the GO-based hydrogel was more than 1000 Pa, and a higher G’ value was observed with increasing amounts of α-CD. The π-π stacking interaction between the GO cross-linkers and PEG/CD complex is a dynamic reversible interaction. Thus, the hydrogels showed shear-thinning, which is observed for many supramolecular hydrogels.^360-362

Very recently, Huang et al. developed injectable self-healable nanocomposite hydrogels containing PDA-coated chemically reduced graphene oxide (rGO@PDA).³³⁵ rGO@PDA of 3.3 nm thickness (with 2 nm of PDA coating) was fabricated by alkaline polymerization of dopamine on the rGO surface, and the surface of the resulting rGO@PDA contains functional catecholic diol groups. The PDA coating chemistry used in this work has been widely explored to produce novel nanomaterials with diverse surface functionalities.^363-368 It is known that the PDA coating layer has numerous catechol groups and possesses electrophilic o-quinones, which are further reactive toward both amine- and thiol-bearing species through Schiff base formation and/or Michael-type addition reactions.^369-372 In this work, the further chemical reaction of PDA coatings was not considered, and the catechol groups on the surface of rGO@PDA were used to form dynamic covalent networks with PBA-containing polymers in the hydrogel matrix. The nanocomposite hydrogel was fabricated by simple mixing of poly(N,N-dimethylacrylamide-stat-3-acrylamidophenylboronic acid) (PDMA-stat-PAPBA), poly(glycerol monomethacrylate) (PGMA), and rGO@PDA in an alkaline aqueous solution (pH ~10). The cross-linked network formed in the hydrogel was based on boronic ester-based dynamic covalent bonds (Figure 20A). rGO@PDA was homogeneously dispersed within the hydrogel and acted as a cross-linker through boronic ester bond formation between the surface catecholic diols and PBA of PDMA-stat-PAPBA chains. Rheology analysis indicated that the incorporation of rGO@PDA endowed the nanocomposite hydrogels with improved mechanical strength, as evidenced by the significant increase in G′ over that of the rGO@PDA-free hydrogel. Depending on the amount of rGO@PDA, the nanocomposite hydrogels exhibited G’ values in the range of 40,000~60,000 Pa, which was 7~11 fold higher than those of the hydrogel without rGO@PDA. To estimate the self-healing property, the two pieces of separately prepared hydrogels (denoted as a and b) were allowed to come into contact for 20 sec. An intimately merged hydrogel was formed without any boundary (Figure 20B), and the hydrogel behaved as one single and could be stretched without rupture at the interface where each hydrogel had initially come into contact. This self-healable ability could be ascribed to the reversible nature of boronic ester dynamic covalent bond-based networks. The self-healing behavior was also investigated by dynamic strain sweep (γ = 0.1% and 400%) at a fixed frequency (1 rad/sec). Rheological studies showed that both G′ and G″ showed a significant drop with increasing strain, and beyond ~245% of strain, G″ became greater than G′, indicating the complete destruction of the hydrogels (Figure 20C). The continuous step-strain sweeps showed that the G’ value largely decreased from ~68,000 Pa to ~700 Pa at 400% of applied strain (Figure 20D). After the strain returned to 1.0%, G′ and G″ were found to almost recover to their initial levels. This recoverability of hydrogels is typically observed in most dynamic hydrogels cross-linked with reversible interactions.^308,352 The rGO@PDA-incorporated hydrogel exhibits shear thinning behavior, making it applicable as a bioink for 3D bioprinting (Figure 20E). Figure 20F shows the successful printability of the hydrogel by demonstrating its good injectability and rapid recovery of sufficient mechanical fidelity to support the next printed layer. As described, the chemical approach using the boronic ester dynamic covalent bonds enabled dynamic interactions between rGO@PDA and the polymer network, resulting in the novel hydrogels with controllable self-healing and injectable capabilities for 3D bioprinting applications. The surface chemistry used for endowing GO with catechol functionalities appears to be very effective in generating this kind of dynamic hydrogel.

Figure 20.

(A) Illustration of the synthesis of nanocomposite hydrogels. (B) Demonstration of the self-healing at room temperature. (C) Dependence of moduli on strain amplitude sweep (γ = 0.1–400%) at a fixed frequency of 1 rad/s. (D) Step-strain test at a fixed frequency of 1 rad/s (1% or 400% of strain). (E) Viscosity measurement at 1% of strain. (F) Printed cone and hollow cylinder 3D patterns. All these data were obtained using the hydrogel with the ration of PDMA-stat-PAPBA (wt %) : PGMA (wt %) : rGO@PDA (wt %) : PBA/diol (molar ratio) as 5.00 : 2.50 : 0.50 : 1/2. Reprinted with permission from ref³³⁵. Copyright 2019 American Chemical Society.

As an alternative type of chemistry that can install catechol groups on GO surfaces, poly(norepinephrine) (PNE) coating chemistry is recommended.^373,374 Norepinephrine (NE) is a small-molecule neurotransmitter possessing a catecholamine structure;³⁷⁵ It is a dopamine analog with an additional alkyl hydroxyl group.³⁷⁶ PNE coatings are also formed through oxidative polymerization in alkaline solutions. A number of reports have demonstrated the advantages of PNE coating techniques over PDA coating, in terms of coating uniformity. Compared to that with PDA, the PNE coating provides a more uniform and well-defined nanolevel coating on the surfaces.^374,377,378 Hence, it is often recommended as a more preferable chemical approach for material surface coatings.^375,379,380 It is expected that the surface of the resulting PNE-coated nanomaterials may exhibit more reproducible and consistent characteristics in forming dynamic covalent bonds with the polymeric networks.

7.2. Metal Carbonate-based Hydrogels

Biomineralization processes for fabrication of organic-inorganic composites have been explored to generate high-performance biofunctional nanoparticles and hydrogels.^381-386 As representative examples of biominerals, calcium carbonate (CaCO₃) and calcium phosphate (CaP) have been incorporated into various hydrogel matrices in either a nano/microparticle form or an in situ mineralized forms.^387-390 Although the incorporated biominerals effectively improve the mechanical properties, they usually exist as fillers and not as cross-linkers that can strongly hold the polymer networks. Thus, the gelling and rheological properties are mostly dependent on the matrix polymer itself.^391,392 For example, Huang et al. prepared PVA-based magnetic nanocomposite hydrogels containing nanohydroxyapatite (n-HA) and magnetic nanoparticles (Fe₂O₃).³⁹¹ Although n-HA improved the mechanical properties of the hydrogel, it did not present any effects in modulating the gelling properties, such as shapeability or injectability. This is because the network formation is due to tiny crystalline PVA regions formed by a repeated freeze-thaw process. Ren et al. reported on injectable and degradable polysaccharide-based hydrogels integrated with n-HA and CaCO₃ microspheres.³⁹³ Likewise, the addition of n-HA and CaCO₃ was not the main contributor in controlling the injectability of the hydrogel, because the hydrogel was formed by network formation through the Schiff base reaction between amine-bearing carboxymethyl chitosan and aldehyde-functionalized oxidized alginate.

7.2.1. Calcium Carbonates.

A well-defined ultrafine amorphous CaCO₃ (ACC)-incorporated nanocomposite hydrogel was constructed by simply mixing CaCl₂, Na₂CO₃, and PAA in an aqueous phase.^394,395 Very small ACC nanoparticles (~ 3 nm) were formed through in situ mineralization in the aqueous PAA solution. The main idea of this work is not the direct addition of ACC nanoparticles into PAA solution but rather the formation of ACC nanoparticles in the presence of PAA in water. The former approach may not provide effective interaction between the surface of ACC and PAA chains due to the limited dispersion of ACC in the PAA solution. Moreover, this simple mixing process results in the limited adsorption of PAA chains on ACC surfaces. For this reason, it is unlikely to expect strong PAA-ACC interspecies interactions. In contrast, the in situ formation of ACC was initiated by sequentially adding Ca²⁺ ions and CO₃^2- ions into PAA solution (Figure 21A). By this in situ mineralization route, the binding of PAA to ACC nanoparticles may occur in the following two possible ways: (i) Ca²⁺ ions bind to anionic carboxylic acid groups of PAA and then the ACC grows around the PAA chains, or (ii) ACC crystals are first formed and PAA chains are then incorporated onto the ACC surface. This mineralization behavior, in the presence of organic species, was reasonably supported by a previous report that described the in situ formation of thiol-functionalized hydroxyapatite nanoparticles.³⁹⁶ This process provides a strong hybrid structure, in which the PAA chains are tightly incorporated into and/or onto the ACC nanoparticles. Thus, it is expected that the ACC nanoparticles can serve as excellent cross-linkers for PAA network formation. It is noted that PAA showed the chain-length-dependent gelling ability as well as concentration-dependent gelation behavior. If the length of PAA was too short (~2000 g/mol) or too long (~ 450000 g/mol), the PAA chains cannot produce a gel structure and instead form weak precipitates in water. In contrast, for the PAA of ~100,000 g/mol, the hydrogel was stable and was not dispersed again in water (Figure 21B). This phenomenon is understood in that Ca²⁺ cross-links with PAA chains were dependent on the chain length of PAA. There may exist a critical point of the cross-links that hinders the redispersion of the formed hydrogels. Furthermore, the minimal concentration of PAA necessary to form a hydrogel was found to be approximately 0.08 M. The lower concentration resulted in precipitates of crystalline CaCO₃, which were formed through uncontrolled nucleation and growth. Hence, optimization of the length and concentration of PAA chains was key in fabricating the stable hydrogels. The ACC-cross-linked hydrogels were moldable into various shapes and stretchable into long fibers without any elastic recovery (Figure 21C,D). Interestingly, the hydrogel shows rapid self-healing behavior, when two separate hydrogels are linked together (Figure 21E). This behavior appears to be useful for the construction of multifunctional 3D constructs by linking several pieces of small hydrogel blocks with different biofunctions, such as cell-laden hydrogels (possibly different cell types) and growth factor-releasing hydrogels. ACC/PAA hydrogels were observed to have a typical hydrogel structure with porosity (Figure 21F). The TEM image shows the presence of ACC nanoparticles of 1.5~3 nm, which reflects the complex structure of ACC nanoparticles physically cross-linked with PAA (Figure 21G). The ACC/PAA hydrogels show the angular frequency-dependent viscoelastic behavior, which is normally observed for hydrogels with physical cross-links (Figure 21H). The hydrogel exhibits shear thinning behavior: with increasing shear rate, a reduction in viscosity is found (Figure 21I). Hysteresis of the shear rate sweep cycle indicates that the hydrogel is thixotropic (Figure 21J). These rheological properties provide the hydrogels with good injectability and shapeability. In a recent report, it was demonstrated that the ACC/PAA nanocomposite hydrogels had good injectability.³⁹⁵ The temperature dependence indicates that G’ becomes higher than G” with increasing temperature, reflecting the formation of harder gels due to the more dynamic interaction of PAA chains on ACC nanoparticles (Figure 21K).

Figure 21.

(A) Schematic synthesis of the ACC/PAA supramolecular hydrogel. (B) ACC/PAA hydrogel is stable in water. (C) ACC/PAA hydrogel is plastic, which can be made in different shapes. (D) ACC/PAA hydrogel is stretchable. (E) Self-adhesion of ACC/PAA hydrogel. Dye molecules (rhodamine B and methylene blue) were introduced to produce the colors. (F) SEM image of the freeze-dried ACC/PAA hydrogel. (G) TEM images of ACC/PAA dry gel. The insets are the corresponding electron diffraction pattern and an enlarged view of the area highlighted by the red square illustrating the presence of very small ACC nanoparticles (highlighted by green circles), (H-K) Rheological behavior of the ACC/PAA hydrogel. (H) Frequency dependencies of the storage (G’) and loss (G’’) moduli. (I) Viscosity as a function of shear rate. (J) Thixotropic loop measurement. (K) Temperature dependencies of the storage (G’) and loss (G’’) moduli. Reprinted with permission from ref.³⁹⁴ Copyright 2016 WILEY-VCH. (L) The ATR–FTIR spectra of all mineral plastics indicate the complexation between carboxylates and respective metal ions (around 1530 cm⁻¹) and the deprotonation degree of PAA (around 1700 cm⁻¹). (M) Separated MnCO₃/PAA hydrogels in the swollen state. The dark red color results from the addition of Toluylene Red, and (N and O) upon connection, the gels heal themselves within minutes. Reprinted with permission from ref.³⁹⁷ Copyright 2018 The Royal Society of Chemistry.

7.2.2. Other Alkaline Earth Metal and Transition Metal Carbonates.

The PAA-based shapeable and injectable nanocomposite hydrogels could also be fabricated from the substitution of Ca²⁺ by other alkaline earth and transition metal ions (Mg²⁺, Sr²⁺, Ba²⁺, Mn²⁺, Ni²⁺, and Zn²⁺).³⁹⁷ For hydrogel formation, it was critical to control the amount of added metal ions and the deprotonation degree of the carboxylic acid groups of PAA. ATR-FTIR analyses revealed the formation of M²⁺-COO⁻ complexes upon the addition of Na₂CO₃, except for BaCO₃/PAA hydrogels (Figure 21L). Rheological analyses of the swollen hydrogels indicate that these metal carbonate/PAA nanocomposite hydrogels exhibited behavior analogous to that of the already reported ACC/PAA hydrogels.³⁹⁴ All these physical hydrogels show viscoelastic behavior dependent on the angular frequency. All the metal carbonate/PAA hydrogels exhibited high G’ and G” values (10³~10⁵ Pa). Notably, the hydrogels exhibit shear thinning behavior. This work showed that G’ and G” values and viscosity of the hydrogels could be modulated by the choice of metal ions. These adjustable mechanical and rheological characteristics of metal carbonate/PAA hydrogels may find diverse applicability with optimized printability for 3D bioprinting. In addition, similar to the ACC/PAA hydrogel, the diverse classes of metal carbonate/PAA hydrogels (SrCO₃/PAA, NiCO₃/PAA, MnCO₃/PAA, and ZnCO₃/PAA) exhibited pastelike plasticity and rapid self-healing in the swollen state. For example, MnCO₃/PAA hydrogels exhibited the excellent self-healing ability within minutes, which was ascribed to the dynamic physical cross-linking of the PAA chains (Figure 21M-O). Although these current studies are only based on PAA, there are many choices of COOH-containing biopolymers (poly-COOH), including hyaluronic acid, alginic acid, and carboxyl methylcellulose. Therefore, it is worth developing diverse classes of metal carbonate/poly-COOH systems in an attempt to find useful bioinks with tailored printability for the construction of specific 3D tissue constructs.

7.3. Nanoclay and Silica-Based Hydrogels

Nanoclays or silica-based nanoparticles have been broadly used as reinforcements for polymeric composites to improve mechanical, thermal, and anti-corrosion properties. This section presents the current update on developing nanoclay or silica nanoparticle-reinforced hydrogel systems for biomedical applications.

7.3.1. Laponites.

Laponite nanosilicates (Na_0.7Si₈Mg_5.5Li_0.3O₂₀(OH)₄) are synthetic clays that have similar compositions of bioactive glasses and have a nanodisk-shape with an average diameter of 25 nm and a thickness of 1 nm.^398-400 One feature that distinguishes Laponites from natural clays is their excellent uniform dispersibility in the aqueous phase.⁴⁰¹ Laponites have surfaces and edges that are negatively and positively charged, respectively. Due to their ionic properties, these clays are preferentially exfoliated at low concentrations (ca. ~1.5 wt%), whereas at high concentrations (~3 wt%), they form a ‘house of cards’ structure due to interplatelet edge-to-face ionic interactions.^402,403 Over the past decade, a vast number of Laponite-based hybrid materials have been developed, and the role of Laponite in the nanocomposites have been addressed by several excellent review papers.^404,405 In particular, Laponites are employed as a major colloidal filler for tissue-forming scaffolds in the field of regenerative medicine because they not only can reinforce the scaffolds but can also provide bioactivity for osteogenic differentiation of human mesenchymal stem cells (hMSCs).^406-408 With respect to 3D bioprinting, Laponites have played a pivotal role in modulating the rheological and mechanical properties of many classes of polymer hydrogels. They can modulate these properties by acting as cross-linkers that can form 3D networks with the matrix polymer chains.^400,409,410 In this section, we highlight the recent use of Laponites as cross-linking species to control the rheological and mechanical characteristics of Laponite-based nanocomposite hydrogels. Wilson et al. reported on a new class of the nanocomposite bioink consisting of kappa-carrageenan (κCA) and 2D Laponite nanosilicates (nSi, Laponite XLS).³³⁰ They focused on the role of nSi in improving the self-healing property and mechanical stiffness of the hydrogels. The nSi used in this work had a negatively charged surface, where the positive charge on its surface was shielded using inorganic phosphate. The κCA/nSi nanocomposite hydrogels are fabricated mainly based on thermal gelation and subsequent ionic gelation, representing the inherent gelling behavior of κCA. The main role of nSi as an additional cross-linker was suggested through the formation of hydrogen bonds between the nSi surface and the OH groups of κCA.⁴¹¹ The heated mixed solution of κCA/nSi was cooled to induce hydrogel formation, which occurs due to the formation of double helices through hydrogen bonding between galactose units of κCA. The subsequent addition of K⁺ ions induces additional interactions with the κCA chains, thereby further stabilizing the thermo-gelling hydrogels. Although the κCA thermoreversible hydrogel is known to exhibit shear-thinning property, it lacks desirable printability for 3D printing due to its poor self-healing properties and low mechanical strength. These weaknesses may result in final 3D constructs with low shape fidelity and self-standing ability. Thus, the main purpose of this work is to overcome these drawbacks by incorporating nSi into the κCA hydrogel matrix. Figure 22A shows a three-step route in which κCA/nSi nanocomposite hydrogels are subjected to 3D printing. The random coils of κCA at 37°C are converted to the shear-aligned structure, which induces shear thinning behavior and finally is ionically stabilized to form stable hydrogels. Figure 22B shows that both κCA and κCA/nSi exhibit similar shear thinning behavior, as indicated by the power-law flow index (n) numbers of 0.56, 0.57, 0.58, and 0.54 for various amounts of nSi. The flow index (n) is a parameter of the power-law equation (η = Kγ̇(n−1)) (where η is viscosity, K is the consistency index, γ̇ is the shear rate, and n is the flow index (shear thinning index)) and describes the rheological behavior of liquids, such as non-Newtonian (shear thinning and shear thickening) and Newtonian behaviors. A flow index (n) of n < 1 is indicative of materials with the shear-thinning property.⁴¹² The almost identical viscosity decrease under shear force indicated a negligible effect of nSi on the shear-thinning ability of κCA, indicating that nSi did not interfere with the ability of κCA polymer junction to undergo for shear-thinning characteristics. The effect of nSi on the recovery of hydrogel structure after the removal of shear stress was examined by subjecting the hydrogel precursor solutions to alternating low (1%) and high (100%) strain conditions. κCA presented a low percent recovery of G’ (69%), while nSi-incorporated hydrogels exhibit a higher percent recovery range of 93~99%, depending on the nSi content (Figure 22C). Importantly, it was found that nSi addition afforded a faster recovery (below approximately 5 sec) after cessation of shear stress. This observation indicates that nSi greatly improved the self-healing ability of the hydrogels due to the reversible dynamic interaction between nSi and κCA, which was due to the OH groups of κCA interacting with the surface of nSi. The researchers demonstrated the ability of the κCA/nSi bioink to print diverse complex constructs with high shape fidelity. The printed structure has high shape retention, even under the loading of many successive layers (~30 layers). This excellent processability may not require additional stabilization steps, such as the post-chemical cross-linking process. Overall, the presence of nSi enabled κCA/nSi to form self-supporting anatomical size constructs, which can be achieved by the rapid self-healing property and high stiffness of these nanocomposite hydrogels. Recently, the same research group reported on a nanoengineered ionic-covalent entanglement (NICE) bioink for the fabrication of mechanically stiff and elastomeric 3D structures.⁴¹³ The main component of this system was a gelatin methacrylate/κCA (GelMA/κCA) ICE network, which was reinforced by the incorporation of nSi having both of negatively charged faces and positively charged edges. In contrast to the use of nSi with only negative charges, the reversible interaction between doubly charged nSi and the charged polymer backbone allowed controllable shear thinning behavior (Figure 22D). In this work, a Herschel-Bulkley computational model was employed to simulate the behavior of the hydrogel bioinks and predict the cell stress during bioprinting by incorporating rheological data including shear thinning characteristics. The simulation study indicated that nSi and κCA exhibited synergy in increasing the shear thinning and yield stress in the NICE bioinks. The NICE bioink has promising potential due to the combination of superior printability, improved mechanical properties, and excellent bioactivity.

Figure 22.

(A) Synthesis of thermo-responsive and shear thinning bioinks from nSi and κCA. Schematic showing the dual cross-linking process of thermoreversible gelation and ionic gelation of the κCA network. Nanosilicate-stabilized cross-linked network due to physical interactions with κCA and ions to improve the mechanical stability of 3D printed anatomical-size structures. κCA undergoes thermo-reversible gelation upon heating and cooling which results in the formation of double-helical structures that can then be ionically cross-linked with the introduction of K⁺ ions to form a stable network. (B) Shear rate sweeps revealing the shear-thinning nature of κCA and κCA–nanosilicate bioinks. (C) Time sweep at 37°C demonstrating percent recovery of bioink’s storage modulus with the introduction of nSi. Reprinted with permission from ref.³³⁰ Copyright 2017 American Chemical Society. (D) The shear stress sweeps measure viscosity changes with increasing shear stress, allowing visual comparison of the yield regions of each bioink (NICE yield region shaded on graph). Shear rate sweeps illustrate the shear thinning characteristics of pre-cross-linked gels. Reprinted with permission from ref.⁴¹³ Copyright 2018 American Chemical Society.

Basu et al. designed silicate nanodisk (nSi, Laponite XLG)-incorporated DNA-based nanocomposite hydrogels that show shear-thinning and rapid self-healing properties.¹⁵ The DNA-nSi nanocomposite hydrogel was prepared following a two-step fabrication process (Figure 23A). First, DNA (4 w/v%) with a double helix structure was denatured by heating at 90°C and the sequential cooling to 37°C resulted in randomly rehybridized DNA networks (Type A network) through the formation of multiple hydrogen-bonding interactions between complementary base pairs of neighboring DNA strands. This heating and cooling process produced an intermediate weak pregel. Second, the addition of nSi into the pregel produces another physical network (Type B network), which is formed through electrostatic interactions between the positively charged edges of nSi and the negatively charged phosphate groups of DNA. This anisotropically charged, two-dimensional, ultrathin structure with a high surface-to-volume ratio facilitated physically cross-linked networks for Type B network points.^414-416 The specific electrostatic interaction between nSi and DNA was thoroughly investigated using X-ray photoelectron spectroscopy (XPS) analysis. The additional Type B network formed by the nSi-DNA interaction provided the hydrogel with shear-thinning property and injectability (Figure 23B,C). The rheological property is a crucial requisite for injectable hydrogels, for which the viscosity decreases with increasing shear rate.⁴¹⁷ Noticeably, nSi-DNA interactions modulated the elasticity of the nanocomposite hydrogel, and an increase in storage modulus values was found in the hydrogels containing a larger amount of nSi (Figure 23D). In addition, the yield stress increased from 29.45 ± 3.75 to 73.98 ± 9.42 Pa as the nSi concentration increased from 0 to 0.5%, respectively. Due to this improved reinforcement, the nanocomposite hydrogel could retain its shape in comparison to that of the starting DNA solution. To evaluate whether the nSi-DNA nanocomposite hydrogels could exhibit rapid recovery after the removal of shear stress, oscillatory strain amplitude studies were performed. The result showed that the G’ value was completely recovered within seconds, and this behavior was observed for the formulations both with and without nSi (Figure 23E). However, the nanocomposite hydrogels incorporating nSi have a higher value of G′. Taken together, the nSi-reinforced DNA hydrogels show potential as a new type of bioink with desirable printability for 3D bioprinting. The strength of using DNA as the main building block for injectable nanocomposite hydrogel fabrications may lie in the many unique properties of DNA, including high molecular weight, numerous negatively charged phosphate groups, and highly specific base-pair interactions.

Figure 23.

(A) Schematic representation of the design strategy for the development of multifunctional nanocomposite hydrogels. DNA–nSi injectable hydrogels are formed via a two-step gelation method. The first step consists of an intermediate weak gel (pregel) formation by heating and subsequent cooling of double-stranded DNA. The denaturation of double-stranded DNA followed by rehybridization in a random fashion facilitates the development of interconnections between adjacent DNA strands (type A network points) via complementary base pairing. Introduction of nSi in the second step of the gelation process increases the number of network points (type B) via electrostatic interaction with the DNA backbone, resulting in a shear-thinning injectable hydrogel. (B) Viscosity vs shear rate plots illustrate an increase in the viscosity due to the presence of nSi. All the formulations display the typical shearthinning behavior with a reduction in the viscosity as the shear rate increases. (C) Image showing the injection of the blue colored hydrogel through a 22G needle. (D) Frequency sweep experiments performed in the range of 0.01 to10 Hz indicate an increase in storage modulus as the concentration of nSi was increased. (E) Recovery data obtained by monitoring the storage modulus of the nanocomposite hydrogels while subjecting them to alternating high (100%) and low (1%) strain conditions. Both 0% nSi (i.e., DNA gel without nSi) and 0.5% nSi (i.e., DNA gel with 0.5% nSi) exhibited more than 95% recovery. Reprinted with permission from ref¹⁵. Copyright 2018 American Chemical Society.

Liu et al. developed moldable self-healing nanocomposite hydrogels based on dopamine-modified PEG and Laponite nanosilicates.⁴¹⁸ The multiarm PEG conjugated with dopamine moieties (PEG-D8) could form 3D networks with Laponite (Figure 24A), which were constructed through interfacial interactions between Laponite and catechol units of dopamines in PEG-D8 (Figure 24B). The dopamine-Laponite interaction was known to be reversible and thus has been used for various nanocomposite hydrogels.^419,420 Due to the dynamic property of the cross-linked networks, the hydrogels could be molded into diverse shapes (Figure 24C). When the oscillatory shear strain was applied, the hydrogel exhibited re-establishment of its network through strong catechol-Laponite interactions, indicating good self-healing properties (Figure 24D). The dopamine of the Laponite/PEG-D8 hydrogels underwent the time-dependent auto-oxidation to dopamine quinone and formed permanent covalent cross-links through the dimerization of dopamine units.^421,422 This spontaneous chemical cross-linking resulted in increased stiffness of the hydrogels, as estimated by the increased G’ value. The incubation time for saturated chemical cross-linking was approximately 4 days, indicating that the formation of chemical networks was very slow. If the rate of oxidation-induced chemical cross-linking is controllable, this class of hydrogels may have potential as bioinks due to their injectability, self-healing properties, and mechanical stability after printing.

Figure 24.

(A) Laponite was combined with PEG-D to prepare the adhesive hydrogel. (B) the hydrogel consists of a reversible interaction formed between dopamine and Laponite. (C) The hydrogel is remolded to different shapes (F; scale bars: 10 mm). (D) Normalized G’ values of D8-15/Lapo-5 (incubated for 1 day) subjected to repeated cycles of 1 and 1000% shear strain with 10 s resting time in between. Reprinted with permission from ref⁴¹⁸. Copyright 2017 WILEY-VCH.

PEG and PEG derivatives are also used for the formation of hydrogel matrices of which rheological and mechanical properties can be modulated by incorporating Laponite nanosheets. Peak et al. fabricated injectable PEG/Laponite nanocomposite hydrogels.⁴²³ The addition of Laponite as a cross-linker reinforced the PEG network and increased the viscosity and storage modulus of the hydrogels. When Laponite (2~8 w/v%) was added to the PEG solution (10 w/v%), the mixture exhibited shear thinning properties. This viscosity decrease with increasing shear rate was supported by estimating the shear thinning index (n). PEG showed almost Newtonian behavior (n = 0.96), whereas the PEG/Laponite nanocomposite hydrogels had n values of 0.22~0.29, indicating that Laponite induced shear thinning behavior. Nojoomi et al. developed Laponite-incorporated PEG-co-pyromellitic dianhydride (PEG-co-PMDA) copolymer-based injectable nanocomposite hydrogels.³³⁷ The addition of Laponite into the copolymer hydrogel precursor changed the rheological behavior from Newtonian to thixotropic. The physical interaction between the Laponite surface and PEG-co-PMDA restricted the mobility of the polymer chains. In addition, matrix-assisted oppositely charged edge-to-face inter-Laponite interactions formed a ‘house of cards’ structure at low shear rates and consequently increase the viscosity of the nanocomposites.^424,425 Under shear force, the Laponite platelets may align with the applied stress increasing the mobility of the polymer chains, and finally decreasing the viscosity with increasing shear rate. The Carreau-Yasuda model predicted the shear-thinning properties of the nanocomposite hydrogel precursor. This precursor could be chemically cross-linked through the reaction of dianhydrides of PMDA with diamine cross-linkers. This posttreatment step could provide the nanocomposite hydrogel with increased mechanical stability.

7.3.2. Silica Nanoparticles.

Like Laponite nanoclays, silica nanoparticles are also good candidates that modulate the rheological properties of nanocomposite hydrogels. The silanol groups (-Si-OH) of silica surfaces can be readily modified with alkoxide-containing reactive compounds, such as aminopropyl triethoxysilane (APTES).^426,427 Further modification with diverse polymers is also feasible to provide additional functionality.^428,429 The rheological and mechanical properties of silica nanoparticle-containing hydrogels have been adjusted by physical interactions, including host-guest⁴³⁰ and electrostatic interactions.⁴³¹ Lee et al. incorporated sub-100 nm cationic silica nanoparticles (SiNPs) in an anionic alginate/gellan (Alg/gellan) polymer mixture to produce nanocomposite hydrogels with improved mechanical stiffness, printability, and printing fidelity.⁴³¹ The addition of aminated SiNPs (AmNPs) (6 wt %) into the Alg/gellan mixture resulted in significantly enhanced shear thinning with much higher zero shear viscosity (2930 Pa s, 1062% increase) than that of the Alg/gellan inks (252 Pa·s). This behavior was due to the electrostatic interactions between the positively charged surfaces of AmNPs and negatively charged Alg/gellan. The mechanical properties could be modulated by the concentration and surface chemistry of the nanoparticles as well as the length of the anionic polymer chains. The major effect expected from the addition of AmNPs is the suppression of swelling of the printed constructs compared with the inks without AmNPs. This role of AmNPs may result in the high printing fidelity of the final 3D structures. Oscillatory strain amplitude studies showed that Alg/gellan inks, irrespective of the presence of AmNPs, exhibited rapid shear recovery behaviors. The presence of AmNPs increased the storage modulus (G′ at 1 rad/s) by 486% compared with that without AmNPs. The increased G′ vaue was attributed to AmNP-mediated physical cross-linking. Compared to Alg/gellan inks lacking AmNPs, Alg/gellan containing AmNPs could be printed without any collapse. In addition, AmNPs efficiently suppressed the swelling and shrinking of printed constructs, which was ascribed to the decreased hydrophilicity and highly cross-linked network of the AmNP-incorporated hydrogels. Serres-Gómez et al. reported on nanocomposite hydrogels formed through α-CD threading on PEGylated silica nanoparticles (~ 50 nm), which was prepared by the reaction of thiolated silica nanoparticles with maleimide-functionalized methoxy PEG (Mw: 750 or 5,000 g/mol).⁴³² The supramolecular hydrogel was formed at relatively high CD (6~10 wt%) and PEG (6~12 wt%) concentrations. The threaded α-CDs on silica surfaces acted as physical cross-links to form dynamic 3D networks, which were formed by the aggregation of CD/PEG inclusion complexes. Rheology studies showed a solid-like behavior of the hydrogels, with a G′ value of up to 25 kPa, which was higher by 1 order of magnitude than that of the hydrogel prepared without the silica nanoparticles. This shows a good example of the host-guest complexation approach that can develop diverse dynamic hydrogels. Various surface chemistries may generate useful dynamic hydrogels by allowing CD or CB[n]-based host-guest events on the surface of many classes of nanoparticles, irrespective of their surface elements or dimensions. Although most current studies of silica-based nanocomposite hydrogels have focused on non-porous solid silica nanoparticles, mesoporous silica nanoparticles need to be considered as alternatives for not only modulating the rheological properties of hydrogels but also releasing bioactive molecules.

7.4. Metal-Based Hydrogels

Gold nanoparticles (GNPs) and silver nanoparticles (AgNPs) have been employed as main building materials for the generation of a variety of nanocomposite hydrogels with diverse functionalities. GNPs and AgNPs are incorporated into the hydrogel network not only to simply provide the inherent property of each nanoparticle, such as photothermal activity^433,434 or antibacterial property,^339,435,436 respectively, but also to mainly control the rheological or mechanical properties by participating as cross-linkers in the hydrogel matrix.^{338,340,437-439} In this review, we focus on recent studies of the latter cases, which are relevant to the printability of the hydrogels.

7.4.1. Gold Nanoparticles.

Qin et al. designed gold nanoparticle (GNP)-incorporated near-infrared (NIR)-responsive self-healing nanocomposite hydrogels.⁴⁴⁰ For hydrogel fabrication, the surface of the GNPs was first functionalized with acrylamide groups through the formation of dynamic RS-Au coordination bonds (Figure 25A). Then, the aqueous free radical polymerization of N-isopropylacrylamide (NIPAM) in the presence of acrylamide-modified GNPs produced nanocomposite hydrogels (GNP/PNIPAM), in which the GNPs act as large cross-linkers to tightly hold the PNPAM chains in the 3D networks (Figure 25B). Combined with the photothermal activity of GNPs,^441-444 this nanocomposite hydrogel employed the dynamic reversible RS-Au coordination interaction as a self-healing motif.^445-447 The photothermal property of GNPs triggered by NIR irradiation allows the RS-Au coordination bonds to be reversibly switched on and off (Figure 25C). Under NIR irradiation, this coordination interaction was disrupted, whereas it was reformed upon cessation of irradiation. As expected, in the ‘Laser ON’ state, the nanocomposite hydrogels may lose the cross-linking interactions between the GNPs and polymer matrix, resulting in soft materials that are moldable or injectable under shear force. More importantly, in the ‘Laser off’ state, the cross-links are rapidly recovered from the hydrogel of the original state. The NIR-treated GNP/PNIPAM hydrogels could be injected using a syringe, indicating that the photothermal effect disrupted the RS-Au coordination interactions (Figure 25D). When the NIR laser irradiated the hydrogels, the temperature of the sample increased rapidly to almost 50°C in the initial 2 min. This thermal responsiveness disrupted the cross-linked networks, endowing the hydrogel with injectability. The GNP/PNAIPAM-based hydrogel system exhibited the excellent NIR-induced self-healing properties. Under NIR irradiation, the two pieces of GNP-incorporated hydrogels (GNP-15) in close contact became fused into one piece, presenting good stretchability (Figure 25E). This healing ability is well supported by the schematic illustration in Figure 21h, which shows the dynamic reconstruction of RS-Au interactions at the interface of two separate hydrogels. In addition, when two individual pieces of GNP-15 and GNP-0 (without GNP) were brought together, rapid self-healing was observed in a few minutes (Figure 25F). Interestingly, GNPs could act as an adhesive species for two separate pieces of GNP-free hydrogels (Figure 25G). The fractured surface of one piece was first brushed with GNPs to form a thin GNP layer, and this piece was placed in contact with fresh hydrogel (GNP-0). Under NIR irradiation at the interface, each piece of the hydrogel was reconnected in a few minutes to establish the good tensile properties. The healed GNP-15 hydrogels exhibited a stress-strain curve nearly identical to that of the initially prepared sample, indicating the excellent recovered extensibility. Interestingly, the mechanical properties of the healed hydrogels were retained with a healing efficiency of 90%, even after ten cycles of the healing process. The NIR-triggered dynamic RS-Au interactions are key to the excellent maintenance of these mechanical characteristics. To optimize the printability for bioprinting applications, further detailed studies on NIR-controlled rheological behaviors are needed. Overall, the dynamic coordinated self-healing mechanism described in this work may produce a robust injectable nanocomposite hydrogel with high impact in the field of 3D printing because it is possible to assemble a geometrically complex 3D structure by simply contacting of many pieces of the same and/or different classes of printed hydrogels with diverse shapes and dimensions, various biochemical and biological properties, and different mechanical characteristics.

Figure 25.

(A) Structure of gold NPs and a list of co-monomers used for polymerization of hydrogels (NIPAM : N-isopropylacrylamide and BACA : N,N-bis(acryloyl)cystamine (BACA)). (B) Formation of GNP hydrogel with modified gold NPs as large cross-linker in the in situ free-radical polymerization. (C) Mechanism for dynamic and reversible RS-Au bonding with NIR laser irradiation (808 nm). (D) Optical image of the hydrogel injected using a syringe with a 20-gauge needle. (E) The self-healing procedure between two separated GNP-15 hydrogel pieces (with a gold NP concentration of 750 ppm) under an NIR laser. (F) The self-healing process between GNP-15 and GNP-0 hydrogel pieces under an NIR laser. (G) The self-healing process between GNP-0 hydrogel pieces with the aid of gold NPs under an NIR laser. Reprinted with permission from ref⁴⁴⁰. Copyright 2017 Cell Press.

Niu et al. reported on injectable supramolecular α-CD/PEG polypseudorotaxane-based hydrogels hybridized with GNPs.³⁴⁰ The cross-linked 3D networks of these hydrogels were based on CD-based host-guest polypseudorotaxane motifs. For GNP-incorporated hydrogels, PEGylated GNPs (mean diameter = 8 nm) were first prepared by reducing anionic HAuCl₄ in the aqueous solution of positively charged poly(poly(ethylene glycol) methyl ether acrylate)-grafted poly(2-(dimethylamino)ethyl methacrylate) copolymers (PPEGMA-co-PDMAEMA). GNP hybrid supramolecular hydrogels were then fabricated by simply mixing α-CDs with PPEGMA-co-PDMAEMA in water. The α-CDs formed inclusion complexes with PEG on the GNP surfaces, and the PEG/CD threaded complexes formed the physical network. It was reported that polypseudorotaxanes generated by threading α-CDs onto PEG preferentially formed aggregates, which served as physical cross-links to form hydrogels.^334,448 In addition, GNPs act as cross-linkers to physically hold the CD aggregate networks together. Frequency sweep and dynamic step-strain rheological tests showed that the nanocomposite hydrogels exhibited both typical thixotropic behavior and excellent self-healing properties.

7.4.2. Silver Nanoparticles.

AgNPs have also been diversely used as cross-linking materials to produce various dynamic hydrogels exhibiting self-healing or shear-thinning characteristics.^{52,338,449,450} Niu et al. developed dynamic α-CD/PEG polypseudorotaxane-based hydrogels incorporated with silver nanoparticles.³³⁸ The cross-linking of these hydrogels was based on CD-based host-guest interactions. For injectable AgNP composite hydrogels, PEGylated AgNPs were prepared by ionic complexation of AgNO₃ with the random copolymer of PEG methyl ether methacrylate (PEGMA) and PAA via electrostatic interactions, followed by in situ reduction (Figure 26A).

Figure 26.

(A) Schematic for the preparation of AgNP hybrid supramolecular hydrogels. (B) Dynamic frequency sweep and (C) Shear-thinning behaviors of the Gel-0 (formed at pH 4) and AGel-1 (formed at pH 4). (D) Dynamic step-strain rheological test results of the AgNP hybrid AGel-2 (formed at pH 7). Reprinted with permission from ref³³⁸. Copyright 2018 The Royal Society of Chemistry.

The AgNPs decorated with surface PEG chains had a mean diameter of 4 nm. The AgNP hybrid supramolecular hydrogels were then fabricated through host-guest inclusion interactions between PEG chains on AgNPs and α-CDs in the aqueous solution (Figure 26A). As described for GNP-based supramolecular hydrogels, these AgNP cross-linked hydrogels were physically cross-linked by both polypseudorotaxane structures and AgNPs. In addition, the synergistic effect of polypseudorotaxane and AgNP cross-links induces the formation of nanocomposite hydrogels from pH 4 to 9. For both Gel-0 and AGel-1, the G’ value was higher than the corresponding G” value and the moduli remained constant with increasing frequency in the detected region, indicating the quasi-solid state of these nanocomposite hydrogels (Figure 26B). Furthermore, the G’ value of AGel-1 was 500 kPa, tenfold higher than that of Gel-0, which was ascribed to the additional cross-linking generated by the PEGylated AgNPs. Notably, the hydrogel with AgNPs (AGel-1, 0.14 wt%) showed fourfold higher viscosity than that of the hydrogel without AgNPs (Gel-0). This result reflects the higher crosslinking density in the presence of AgNPs compared to that of the hydrogel with only polypseudorotaxane cross-links. The viscosity of the Gel-0 and AGel-1 hydrogels decreased with increasing shear rate, indicating that the hydrogels showed shear-thinning behavior and thus can be applied as injectable bioinks (Figure 26C). Furthermore, the hydrogels were self-healed by strain, indicating that they could be injected under extrusion and could be self-restored upon cessation of the shear force (Figure 26D). The CD-based polypseudorotaxane structure in the GNP- and or AgNP-based nanocomposite hydrogels was known to undergo thermoreversible threading/dethreading of α-CD from the polymer chains.^451,452 At 58°C, α-CDs are dethreaded from the PEG chains, resulting in disruption of the hydrogel structure into the transparent sol. Conversely, upon cooling, the sol gradually turned into a hydrogel at room temperature. This temperature-controlled sol-gel transition behavior may be considered one of the main factors in designing bioinks with favorable printability for the construction of 3D materials. Jiang et al. reported on dynamic hybrid hydrogels formed through coordination between AgNPs and thiol or acetyl thioester-containing polymers.⁵² AgNPs acted as cross-linking sites to form self-healing dynamic hydrogel networks. This work suggested an advanced system from the researcher’s previous work for thermo-sensitive dynamic hydrogels based on strong chelation between polythioether dendrons and Ag nanoparticles.⁴⁵⁰ The amount of AgNPs affected the mechanical properties of resulting hydrogels. When the feed ratio of Ag/S approached 0.9, the hydrogel exhibited the best mechanical properties. Thiol-containing hydrogels showed better self-recovery performance than the hydrogels with acetyl thioester groups.

Table 3 summarizes the testing status of dynamic cross-linked hydrogel systems that can potentially be used as hydrogel-based extrusion bioinks. Even though several hydrogel systems have not been tested for the 3D bioprinting process, their rheological properties, particularly shear-thinning behavior, and injectability have great potential for developing novel hydrogel-based bioink systems for cell-based microextrusion bioprinting.

Table 3.

Dynamic cross-linked hydrogel systems for potential and recent hydrogel-based extrusion bioinks

	Crosslinking mechanism		Hydrogels	Testing status			Ref.
				Shear- thinning	Injectabilitya	Printabilityb
Physical interaction	Hydrogen bonding	Phenolic compound-based interaction	Gallol-containing hydrogels	○	○ (26 or 25 G)	○	136,138
			Catechol-functionalized ABA triblock copolymer hydrogels	○	○ (23 G× 3/4”)	✖	140
		Ureidopyrimidinone self-complementary dimerization	UPy-containing thermosensitive ABA triblock copolymer hydrogels	○	○ (20 G)	✖	453
	Host-guest interaction	Cyclodextrin-mediated supramolecular interaction	poly-β-cyclodextrin (Pβ-CD)/ poly(3,4-ethylenedioxythiophene)(PEDOT):adamantyl-modified sulfated alginate (S-Alg-Ad) hydrogels	○	○ (27 G)	✖	157
		Cucurbit[n]uril-Mediated supramolecular interaction	highly branched CB[n]-threaded polyrotaxane (HBP-CB[n])/naphthyl-functionalized hydroxyethyl cellulose (HECNp) hydrogels	○	✖	✖	178
	Metal-ligand coordination	Bisphosphonates	bisphosphonate-modified hyaluronic acid/metal ion hydrogels	○	○ (21 G)	O	141,190,199
		Carboxylates	folate/Zn2+ hydrogels	○	○	✖	207
			Fe3+-carboxylate in UPy-functionalized gelatin hydrogels	○	✖	✖	147
		Catechols	DOPA-conjugated chitosan/Fe3+ ion hydrogels	✖	○	✖	227
		Histidine	histidine-bearing acrylamide/acrylate copolymer hydrogels crosslinked with Zn2+ ions	✖	✖	✖	242
Dynamic covalent bond	Imine bond		Chitosan/dibenzaldehyde-functionalized PEG hydrogels	○	○ (22 G, 26 G, or 32 G)	✖	271,273,274
	Hydrazone bond		hydrazine-modified elastin-like protein/ aldehyde-modified hyaluronic acid hydrogels	○	○ (28 G)	✖	282
			hydrazide-modified hyaluronic acid/benzaldehyde terminated F127 triblock copolymers hydrogels	○	○	✖	285
	Oxime bond		poly(n-hydroxyethyl acrylamide-co-methyl vinyl ketone) hydrogels crosslinked with a bifunctional hydroxylamine.	○	○ (0.6 mm)	○ (400 μm, 3.0 bar, and 12 mm/s)	294
	Disulfide bond		thiol functionalized F127/dithiolane modified PEG hydrogels	✖	○ (23 G× 3/4”)	✖	299
	Boronic ester bond		glycopolymer chain-containing copolymer/benzoxaborole-containing copolymer hydrogels	○	○ (21 G)	✖	327,328
Nanocomposite hydrogels	Nanocarbon-based hydrogels	Carbon nanotubes	Single-walled carbon nanotube (SWCNT)/DNA supramolecular hydrogels	○	✖	✖	346
		Graphene-based nanomaterials	aminated-graphene oxide/sodium alginate/polyacrylamide/ nanocomposite hydrogel	○	✖	○ (1.2 bar and 500 mm/min)	352
			phenyl boronic acid-containing copolymer/polydopamine-coated graphene oxide hydrogels	○	○ (19 G)	✖	335
	Metal carbonate-based hydrogels	Calcium carbonates	Poly(acrylic acid)/amorphous CaCO3 hydrogels	○	✖	✖	394
		Other alkaline earth metal (except for Ca2+ ions) and transition metal carbonates	Poly(acrylic acid)/alkaline earth or transition metal carbonate hydrogels	○	✖	✖	397
	Nanoclay and silica-based hydrogels	Laponites	κ-carrageenan/laponite nanosilicate-based hydrogels	○	✖	○ (400 μm or 23 G and 20 or 0.5~8 mm/s)	330,413
			DNA/laponite nanosilicate hydrogels	○	○ (22 G)	○ (feedrate: 39 mm/s and flowrate: 10 mm/s)	15
		Silica nanoparticles	anionic alginate/gellan polymer mixture/cationic silica nanoparticle composite hydrogels	○	✖	○ (400 μm, 40-50 kPa, and 10~20 mm/s)	431
	Metal-based Hydrogels	Gold nanoparticles	poly(N-isopropylacrylamide)/gold nanoparticle composite hydrogels	○	○ (20 G)	✖	440
		Silver nanoparticles	α-CD/PEG polypseudorotaxane/siliver nanoparticle composite hydrogels	○	✖	✖	338

^aInformation for the hydrogels subjected to the injection through syringe needles. The sizes of needle gauges are presented in the parenthesis; in case the information was provided.

^bThe hydrogels were tested by 3D printing are marked with open circles. The printing parameters are presented; in case the information was provided.

8. CONCLUSIONS AND OUTLOOK

The ability for hydrogels to achieve desired printing outcomes, referred to as their printability, continues to be of high interest to the field. Researchers have many tools at their disposal to improve the printability of their bioinks, including modifying the chemistry, hydrogel composition, rheological properties, and printing parameters. One of the most challenging aspects of bioink design is the interrelatedness of each of these factors. Composition and chemistry influence the hydrogel’s response to external forces (rheological properties) and the window of operable printing conditions which may be appropriate to use. Even the optimal hydrogel will result in poor printing outcomes if the correct printing conditions are not utilized. Meanwhile, although many studies have looked at how chemical modifications influence the rheological properties of different hydrogels, the translation to printability is not always a direct one. It has been established that shear-thinning, yielding, and quick recovery/self-healing behaviors are desirable from a printability perspective. However, some hydrogels may exhibit these properties and still not be sufficient to achieve desired printing outcomes. Further research is needed to improve the predictability of these relationships and allow for more intentional design during the bioink development process.

The current hydrogels as bioinks employed chemical cross-linking before and/or after printing not only to modulate printability but also to stabilize final printed constructs. However, there are several shortcomings, including the limited rheological properties and diversity of these hydrogels. Therefore, a rapidly growing demand exists for newly designed hydrogels that can modulate rheological and mechanical properties for desired printability. In this review, we have presented evolving design strategies developed to control the rheological and mechanical properties of diverse classes of dynamic hydrogels. These hydrogels were formed through different chemical designs to build reversible, cross-linked networks based on physical bonds or dynamic covalent bonds. The rational choice or combination of these bonds is the most important factor in determining the gelation kinetics, and dynamic and/or mechanical behaviors of the resulting hydrogels. From a comprehensive review of recent dynamic hydrogels that show shear-thinning and self-healing characteristics, we have collected a variety of useful information on chemical designs, which may establish general criteria that should be taken into account for the development of optimized bioinks with desirable printability. Through various design approaches, many ideas have been suggested to identify the major control parameters affecting the printability of hydrogel-based bioinks for microextrusion bioprinting. For physical interaction-based hydrogels, the rheological and mechanical properties could be tuned not only by selecting the appropriate class of complimentary pairs for physical interactions, but also by adjusting the concentration and composition ratio of each complementary species for physical interactions. For hydrogels with dynamic covalent cross-links, the hydrogel characteristics could also be modulated by several factors including the reaction class, reaction solution pH, reactant concentrations, and ratio between the reactants. Similarly, nanocomposite-based supramolecular hydrogels could also be controlled in terms of dynamics and mechanics by considering the parameters described for physical and dynamic covalent bond-based hydrogels. In addition, the nanoparticle class is another key parameter that determines the physical properties of the corresponding nanocomposite hydrogels. These various control parameters affect the final characteristics of the resulting hydrogels by tuning the molecular interactions in aqueous phases or on nanoparticle surfaces, the reaction kinetics, and the cross-linking densities in the hydrogel matrix. From the clever combination of these various control parameters, one may develop new dynamic hydrogels with tailored rheological and mechanical properties for use as bioinks optimized for specific 3D constructs.

For future developments in bioinks, chemistry-based approaches may offer important clues because the design strategy has the most important influence on the hydrogel properties over other factors, both in terms of gelation kinetics and rheological and mechanical properties. Although recent chemical approaches have shown promise in developing useful hydrogel bioinks that meet the requirements for 3D bioprinting, persistent efforts from a chemistry standpoint should be made to improve the performance of hydrogels that can be exploited for human use. In addition to new chemical strategies, a system design in which multiple bonds are combined in a single system is crucial to efficiently control or improve the printability of hydrogels. Using this approach, one may expect the synergistic role of multiple interactions to enhance the mechanical strength as well as induce rapid recovery after extrusion, which cannot be achieved using a single class of dynamic bonds. Several combinations of bonds for cross-linked networks can be considered possible between different classes of physical bonds, between physical bonds and dynamic covalent bonds, or between different classes of dynamic covalent bonds. As an example, we have discussed the reinforcing effect of oxime dynamic covalent bond-based hydrogels through the formation of hydrogen bonding-induced cross-linking by the cryogelation process. Currently, diverse combinations of multiple interactions have been reported, and this design approach will also be one of the mainstream strategies to control the printability of the various dynamic hydrogels.

For nanocomposite-based dynamic hydrogels, there are a number of considerations for system designs. The choice of the additives can be extended to other nanoparticles that have never been considered before. Although the classes of nanoparticles presented in this review were included based on their frequent use, various surfaces of diverse polymeric, inorganic (metal and metal oxide), and organic-inorganic nanomaterials can be considered for new chemical approaches to produce dynamic hydrogels. Nanoparticle shapes (spherical, rod-like, cubic, or sheet-like) and structures (nonporous solid, porous, or hollow) and their specific surface areas should also be considered to develop novel dynamic nanocomposite hydrogels. The currently used nanoparticles are nearly all based on non-elastic inorganic hard spheres, irrespective of their atomic composition and structures. However, soft and flexible nanostructures, such as nanoemulsions, nanobubbles, and nanogels, have been of interest. These deformable nanoparticles may provide some unprecedented insights in designing dynamic hydrogels, because they may significantly affect the rheological properties of hydrogels during shear stress-involved extrusion processes. Along with basic research from a materials point of view, a systematic simulation study of the effect of nanoparticle deformation on the bulk dynamics of hydrogels should also be conducted. As discussed above, the surface chemistry at nanoparticle surfaces can also involve a diverse combination of physical interactions and dynamic covalent bonds. The rational combination of multiple types of bonds may allow the printability of hydrogels to reach a level desirable for producing specific 3D constructs.

By focusing on the various suggestions presented above, we are sure that continuing efforts will lead to great promise for the next stage of development of bioinks, allowing better and more accurate control of optimized rheological and mechanical properties. Finally, the advanced design criteria established by various chemical strategies will undoubtedly contribute to the future development of bioinks with practical applicability in clinical use.

Biography

Dr. Sang Cheon Lee is received his B.S. (1995) in Polymer Science and Engineering and M.S. (1997) and Ph.D. (2001) in Polymer Chemistry at Inha University, Incheon, South Korea. During the Ph.D. program, he worked as a research trainee at the Korea Institute of Science and Technology (KIST). He took a postdoctoral fellowship in the School of Pharmacy at Purdue University (2001-2003) and the School of Materials Science at the Japan Advanced Institute of Science and Technology (JAIST) (2004). He was a senior research scientist at the Korea Institute of Ceramic Engineering and Technology (KICET) (2004-2009). He is currently a Professor in the School of Dentistry at Kyung Hee University, Seoul, South Korea. In 2019, he was a Visiting Professor at the Wake Forest Institute for Regenerative Medicine (WFIRM), Wake Forest School of Medicine, Winston-Salem, North Carolina. His research interests are focused on the development of mineralized organic-inorganic nanohybrids for applications in drug delivery, molecular imaging, bone tissue engineering, and bioactive implants.

Dr. Gregory Gillispie received his Ph.D. from the joint Virginia Tech - Wake Forest University School of Biomedical Engineering Sciences in 2020. Under the advisement of Dr. Sang Jin Lee at the Wake Forest Institute for Regenerative Medicine (WFIRM), his research focused on the printability of extrusion-based bioinks. This work aimed to improve the methodologies which are used to assess printability and to study the underlying mechanisms which impact printing outcomes. Additionally, he has pursued the development of tissue-specific bioinks targeting the regeneration of orthopedic tissues such as bone, tooth, and cartilage.

Dr. Peter Prim received his Ph.D. in Industrial and Systems Engineering at the North Carolina State University (NCSU) in 2016 and took a postdoctoral fellowship under the advisement of Dr. Sang Jin Lee at Wake Forest Institute for Regenerative Medicine (WFIRM).

Dr. Sang Jin Lee is a tenured Associate Professor and cross-appointed to the Virginia Tech - Wake Forest University School of Biomedical Engineering Sciences. Dr. Lee received his Ph.D. in Chemical Engineering at Hanyang University, Seoul, South Korea in 2003 and took a postdoctoral fellowship in the Laboratories for Tissue Engineering and Cellular Therapeutics at Harvard Medical School and Children’s Hospital Boston and the Wake Forest Institute for Regenerative Medicine (WFIRM) where he is currently a faculty member. He has authored more than 140 scientific publications and reviews, has edited 2 textbooks, and has written 35 chapters in several books. His current research works focus on: designing tissue-engineered scaffolding system (skin, bone, cartilage, skeletal muscle, tendon/ligament, blood vessel, heart, and kidney); 3D integrated tissue-organ printing (ITOP) system for bioengineering complex, composite tissues, and solid organs; bioink development and printability; bio-adhesion ad biointegration between biomaterials and cell/tissue; drug/protein delivery system; bioconjugation and polymer synthesis; bioreactor system for preconditioning; in vitro microphysiological system; and NIR fluorescence-based monitoring system.