Optimization of Oxygen Delivery Within Hydrogels

Abstract

The field of tissue engineering has been continuously evolving since its inception over three decades ago with numerous new advancements in biomaterials and cell sources and widening applications to most tissues in the body. Despite the substantial promise and great opportunities for the advancement of current medical therapies and procedures, the field has yet to capture wide clinical translation due to some remaining challenges, including oxygen availability within constructs, both in vitro and in vivo. While this insufficiency of nutrients, specifically oxygen, is a limitation within the current frameworks of this field, the literature shows promise in new technological advances to efficiently provide adequate delivery of nutrients to cells. This review attempts to capture the most recent advances in the field of oxygen transport in hydrogel-based tissue engineering, including a comparison of current research as it pertains to the modeling, sensing, and optimization of oxygen within hydrogel constructs as well as new technological innovations to overcome traditional diffusion-based limitations. The application of these findings can further the advancement and development of better hydrogel-based tissue engineered constructs for future clinical translation and adoption.

Introduction

Tissue engineering has the potential to revolutionize modern medical therapies and treatments. The capability to repair and replace damaged tissue with living cells in a scaffold allows for new techniques in treating a wide variety of diseases. Specifically, tissue with limited self-repair capabilities can be restored or engineered tissue can improve the function of organs, helping to ease shortage in donor organ availability. This same technology also allows researchers to create more accurate small-scale to large-scale models in vitro to better study a number of biological processes, such as screening of novel pharmaceuticals. Tissue engineering consists of the embedding of cells within a scaffold, which will then be implanted into a patient. For small-scale experimentation, this process has been investigated with animal models. However, one major challenge that exists regarding engineered tissue is the potential complications associated with scaling the engineering tissue to size that is required for successful transplantation into a large animal model and eventually in humans [1]. Figure 1 illustrates the process of tissue engineering and the transition from animal to human models.

Fig. 1.

The field of tissue engineering is defined as using cultured cells within a scaffold to form a tissue mimic that can be used to treat diseased or damaged tissue. Most work within the field utilizes small-scale animal models; however, clinical application requires translation to large-scale animal models and clinical trials.

Over the past decade, hydrogels scaffolds have risen to the forefront of tissue engineering with some success within animal model studies, paving the way for the translation to clinical research and applications. Hydrogels are defined as three-dimensional polymeric structures that have the capability to crosslink molecules, deliver these molecules to a specified destination, and provide structure for tissue engineered developments. Some advantages of utilizing hydrogels in scientific research include their flexibility, high porosity, and high diffusivity of nutrients, such as oxygen [2,3]. This review specifically investigates anisotropic oxygen diffusion and concentration within hydrogel and how these processes can be optimized to construct advanced models and engineered tissues. Improving oxygen diffusion rates and efficiency within hydrogels allows cells to maximize their energy potential and is crucial for culturing cells used for regeneration in tissue engineering [4]. Without adequate oxygen, cells die or change to a less optimal phenotype. Oxygen insufficiencies often arise due the reliance on diffusion transport within the constructs, which can limit the size of the hydrogel and prevent scalability to larger animal or human use [5]. The intent of this review is to detail recent studies related to oxygen diffusion within hydrogels and highlight new applications. As depicted via the tree diagram shown in Fig. 2, engineered tissue provides a promising alternative for medical application. While many obstacles still exist, this review will focus on the optimization of the oxygen supply in a hydrogel scaffold.

Fig. 2.

Tissue engineering has several advantages of over current medical therapies, yet its implementation still has several obstacles. The subject of this review focuses specifically on oxygen optimization in hydrogels.

In both the human body and in vitro cell and tissue culture systems, gaseous oxygen is dissolved in a liquid—blood or culture medium—to enable its delivery to cells. In cell culture monolayers, oxygen diffusion is efficient and quick enough for adequate transport to supply the cells. However, three-dimensional hydrogels make the diffusion more complex as oxygen must now diffuse through the hydrogel, in addition to the cell culture media. The oxygen movement and concentration can be determined through the governing principles of diffusion, in which oxygen particles will follow a concentration gradient flowing from areas of high concentration to low concentration areas. The diffusion of oxygen in the media and the hydrogel can be characterized using Fick's first law (Eqs. (1) and (2)). Figure 3 demonstrates the principle of Fick's law in relation to monolayer cellular cultures and cells embedded in hydrogels in media. According to this principle, the rate of diffusion (or flux) is determined by the diffusion coefficient, the concentration difference, and inversely to the thickness of the substrate

$$\begin{array}{c}{J}_{A,z}=-D_A\frac{{\partial c}_A}{\partial z}\hspace{0.17em}\end{array}$$ (1)

$$J_A=\hspace{0.17em}-D_A\hspace{0.17em}\nabla C_A\hspace{0.17em}$$ (2)

Fig. 3.

A visual representation of Fick's law applied to a monolayer of cells embedded within a hydrogel cultured in media. D₁ and D₂ represent the diffusivities of the media to hydrogel and hydrogel to the cellular monolayer, respectively. Δz₁ and Δz₂ are the vertical thickness of the hydrogel and the monolayer, respectively.

Equation (1) is for diffusion in one dimension (z) while Eq. (2) is for multidimensions where $J_A$ is the average diffusion flux (mol/m²·s) for species A, $C_A$ is the concentration (mol/m³) of species A, $D_A$ is the diffusion coefficient (diffusivity) of species A (m²/s), and ∇ is the gradient operator. Using Fick's second law of diffusion (Eq. (3)), concentration over time can be estimated as follows:

$$\frac{\partial C_A}{\partial t}=D_A{\nabla }^2{C}_A\hspace{0.17em}$$ (3)

where $t$ is the time, $D$ is the diffusion coefficient, and ∇² is the Laplacian. For liquids (such as culture media), the diffusion coefficient is dependent on the temperature, viscosity of the fluid, and size of the particles as described in the Stokes–Einstein equation (Eq. (4))

$$D=\frac{k_{b}T}{6\pi \eta r}\hspace{0.17em}$$ (4)

where $k_b$ is the Boltzmann's constant, $T$ is the absolute temperature, η is the dynamic viscosity of the fluid, and $r$ is the radius of the particle (species A). For diffusion through the hydrogel, the diffusivity is more complex and must take into account factors such as the size of voids and porosity of the construct. Often, this diffusion coefficient must be measured (or estimated) for each specific hydrogel. Another approach rather than directly measuring the diffusion coefficient is to apply more detailed diffusion theories including porous medium diffusion [6,7], fractional diffusion [8], self-diffusivity [9], stress-assisted diffusivity [10], and anisotropic diffusivity [9,11] theories to the hydrogels. Stress-assisted diffusivity (Eq. (5)) is of particular importance for modeling oxygen and nutrient concentrations of hydrogel constructs in vivo where they will be subjected to stresses from body motions. In Eq. (5), $d_{ij}$ is the diffusion tensor, ${\sigma }_{ij}$ represents Cauchy stress, ${\delta }_{ij}$ is the identity tensor, $D_0$ is the nondimensional diffusion coefficient, and $D_1$ and $D_2$ are diffusion parameters due to mechanical stress

$$d_{ij}\left({\sigma }_{ij}\right)=\hspace{0.17em}{D}_0{\delta }_{ij}+\hspace{0.17em}{D}_1{\sigma }_{ij}+D_2{\sigma }_{ik}{\sigma }_{kj}\hspace{0.17em}$$ (5)

$$\frac{\partial C(r,t)}{\partial t}=\frac{1}{r}·\frac{\partial }{\partial r}\left[D\left(r,t\right)·r\frac{\partial C\left(r,t\right)}{\partial r}\right]-R\left(r,t\right)$$ (6)

Equation (6) is Fick's second law considering anisotropic diffusivity and a volumetric consumption rate of the solute in cylindrical coordinates, where $D(r,t)$ is the diffusivity as a function of radius and time, $C(r,t)$ is the concentration (of oxygen or solute) as a function of radius and time, $R(r,t)$ is the volumetric oxygen (or solute) consumption rate as a function of radius and time, $r$ is the radius, and $t$ is the time.

This review builds on these basic principles to focus on oxygen diffusion optimization in hydrogel scaffolds for engineered tissues. First described are various computational models of oxygen gradients and concentrations that account for the complexity of hydrogels. These models assess how oxygen's behavior impacts a cell's response or define limitations to current three-dimensional tissue constructs. Next, experiments are highlighted in which researchers modeled and sensed oxygen in engineered tissue using several techniques, such as fluorescent oxygen microsensors, microbeads, and fluorescence-based nano-oxygen particles. Subsequently, several strategies and technologies used to improve oxygen transport and delivery within hydrogels are discussed. Some of these strategies include the use of devices to characterize oxygen within hydrogels, experimentation with perfusion systems and perfluorocarbons, and comparison among hydrogels types for optimal usage and effectiveness. An awareness of studies on oxygen concentration in hydrogels will hopefully lead to more widespread implementation and improvement of oxygen transport within hydrogels that can be applied from small-scale laboratory research toward increased translation and use in humans.

Modeling Oxygen Diffusion

Effective hydrogel constructs for tissue engineering require sufficient oxygen to allow desired cell proliferation, growth, differentiation, and growth factor production. Computational simulations modeling oxygen diffusion gradients and concentrations of cells embedded in hydrogels allow for a greater understanding and prediction of cells' behavior. The advantage of computational models is that they are a more time- and cost-effective way of testing a variety of parameters that may affect oxygen levels. With accurate models, different cell densities, gel geometries, gel densities and porosities, and numerous other factors can be screened before testing in vitro and in vivo. Axpe et al. created a novel model they called a multiscale diffusion model, which combined standard models based on three common diffusion theories: hydrodynamic theory, free volume theory, and obstruction theory [12]. The model more accurately predicted diffusion of solutes with varied particle sizes and mesh sizes in both poly(ethylene glycol) and alginate hydrogels than the individual models alone. Utilizing all three theories, the model was able to determine correlations between each for a unique mechanism. The overall governing equation used for their novel multiscale diffusion model is

$$\begin{array}{c}\frac{D}{D_o}=\left[\mathrm{erf}\left(\frac{r_{\mathrm{FV}}}{r_s}\right)\exp \left(-{\left(\frac{r_s}{r_{\mathrm{FVW}}}\right)}^3\left(\frac{{\phi }_p}{1-{\phi }_p}\right)\right)+\mathrm{erfc}\left(\frac{r_{FV}}{r_s}\right)\exp \left({-\pi \left(\frac{r_s+r_f}{\xi +2r_f}\right)}^2\right)\right]\end{array}$$ (7)

where $\frac{D}{D_o}$ is the relationship between the solute diffusivity and the diffusivity in pure solvent, ${\hspace{0.17em}r}_{FV}$is the average radius of free volume voids, $r_s$ is the hydrodynamic radius of the solute, $r_{\mathrm{FVW}}$ is the radius of free volume voids in the case of water, ${\phi }_p$ is the polymer volume fraction, $r_f$ is the radius of the polymer chains, and $\xi$ is the size of an open space, and the hydrogel is assumed to have reached a swelling equilibrium. One limitation of the initial model is that it assumed the solutes were solid, spherical, 1–40 nm in diameter, and not charged, but the authors assert that it can be adapted for other solutes, such as oxygen and proteins. Malda et al. studied oxygen gradients within a polyethylene glycol terephthalate/polybutylene terephthalate scaffold as a model of cartilage [13,14]. The group used glass micro-electrodes to measure oxygen within the constructs and modeled oxygen gradients in cylindrical constructs used for tissue engineering as well as cartilage explants. They built a mathematical model incorporating cell density and oxygen consumption as well as directly measured diffusion constants. One unique finding from their experimentation is that despite a lower diffusion coefficient of the polymer scaffold, chondrocytes inside of tissue-engineered constructs were able to adapt survive with relatively low oxygen consumption. Demol et al. constructed a mathematical model that investigated whether unequal cell distribution in hydrogels is related to oxygen diffusion within the gel [15]. They validated their model with in vitro experiments of human periosteum-derived cells in fibrin hydrogels. Their model was able to predict the oxygen tension and cell density within the gel over time by modeling the oxygen diffusion coefficients, cell volume, cell oxygen consumption rate, cell proliferation rate, hypoxia-mediated cell death, and surface thickness. In contrast to the studies and model by Malda, human periosteum-derived cells in this study experienced cell death as the oxygen concentration decreased, leading to a very heterogeneous cell distribution. Taken together, these studies suggest that while many models are capable of modeling oxygen gradients, the response of cells in terms of death, growth, and density can vary with cell type. Thus, the oxygen needs and consumption of the specific cell type used are important to incorporate into models.

Another observation that became apparent to the development of models is to account for the cells' use of oxygen via the oxygen consumption rate (OCR). Several studies have addressed determining the OCR through direct measurement and model comparison studies. Brown et al. developed a mathematical model for oxygen concentration in a cylindrical construct with anisotropic diffusion and oxygen consumption within a engineered heart tissue and compared these results to experimental results of neonatal rat ventricular myocytes in a collagen Matrigel disk [11]. They concluded that their finite element model of the oxygen profile was relatively consistent in comparison to the physical model. However, while the profiles and trends matched, the predicted numerical oxygen concentration was often higher than measured. This difference could be due to several parameters, but one likely factor is an incorrect oxygen consumption rate due to a simplified model of cell metabolism. Magliaro et al. attempted to solve this problem by modeling the oxygen consumption rate (OCR) within three-dimensional cylindrical constructs in relation to cell density [16]. The equation they used to define OCR follows:

$$\begin{array}{c}OCR\left(t\right)=\frac{\pi R^2}{V·{\rho }_{\mathrm{cell}}}\int \frac{{\rho }_{\mathrm{cell}}·s\mathrm{OCR}·C\left(z,t\right)}{K_m+C\left(z,t\right)}\partial z=\hspace{0.17em}\frac{1}{H}\int \begin{array}{c}\frac{s\mathrm{OCR}·C\left(z,t\right)}{K_m+C\left(z,t\right)}\partial z\hspace{0.17em}\\ \hspace{0.17em}\end{array}\end{array}$$ (8)

where $\mathrm{OCR}\left(t\right)$ is the oxygen consumption rate as a function of time and is integrated in the $z$ direction from 0 to H. $C\left(z,t\right)$ is the concentration with respect to vertical coordinate $z$ and time, and $H$ is the thickness of the cylindrical gel$.\hspace{0.17em}R$ is the cellular consumption rate per unit volume of oxygen, $V$ is the volume of the cell construct, ${\rho }_{\mathrm{cell}}$ is the cell density, $s\mathrm{OCR}$ is the maximum rate of oxygen consumption for a cell, and $K_m$ is the Michaelis-Menten constant. Their approach is unique as it takes into consideration spatial differences by modeling a three-dimensional cylindrical cell construct in comparison to monolayered cell constructs and including oxygen gradients in the medium as well. Their results concluded that the rate of oxygen consumption is not constant for a specific cell density or cell type, but rather, the oxygen consumption rate is variable depending on the oxygen concentration sensed by the cells, which can vary with geometry, cell density, and time. The authors suggest that rather than assuming a constant OCR, their model can be employed to more accurately determine the oxygen needs of cells at any given spatial location within a construct at any given time. Using a different approach to the same problem of determining oxygen consumption rate, Lambrechts et al. focused on modeling bioluminescence peak intensities from reporter cells, which are used to measure cell activity [17]. Since the luminescent activity is dependent on oxygen and may be affected by decreased oxygen availability within hydrogels, it is important to take that into account when analyzing the results. Thus, this model includes a compensation for spatial oxygen gradients when quantifying the average photon flux from the bioluminescent reporter cells to predict the oxygen metabolism of cells in a hydrogel. The use of this approach in conjunction with reporter cells can yield more accurate measurements of OCR in future models and lead to better hydrogel designs.

Other models study methodologies proposed to increase oxygen in hydrogels beyond traditional diffusion, such as perfusion culture and oxygen releasing compounds in media. Sengers et al. specifically studied the supply of oxygen in engineered tissue by analyzing mass transfer through the use of a computational mass transfer model [18]. This model has the capability to assess diffusion of oxygen, glucose, and lactate in four culture conditions of an agarose gel: petri dish, compression system, suspension, and perfusion. Their work demonstrated that for static media, the first three culture variations performed had large regions of hypoxia. For the case where the media was assumed to be of a large volume, well mixed, and fully oxygenated, the oxygen levels improved greatly in the petri dish and suspension systems (the perfusion system was always assumed to have well mixed media and therefore no large differences). Their results indicate that for optimal oxygen, glucose, and lactate transport within constructs, the media volumes must be large and well mixed. This suggested that the current standard culture system of a hydrogel sitting a dish is not ideal for nutrient transport. They extended their research to model oxygen uptake by chondrocytes using biosensors to measure oxygen concentration in an agarose gel and measuring glucose and lactate at various time points [19]. This numerical model utilized experimental data from glucose, lactate, and oxygen to present a prediction of the nutrients' concentration profiles over time and space for several cell concentrations and both high and low glucose media. While this model is useful in screening a number of different culture conditions, it currently only uses static culture conditions, which are not ideal. Thus, while useful, the model should be expanded to provide more insight for a variety of culture setups. Radisic et al. created a steady-state mathematical model to estimate oxygen concentration profiles within a tissue culture system. This system was comprised of a parallel channel array made of poly(glycerol sebacate) and a perfluorocarbon (Oxygent) in the media [20]. They found the addition of perfluorocarbon (PFC) increased oxygen in the channels but had limited effects in the tissue construct beyond the surface. They also used their model to test the effects of altering experimental parameters, such as flowrate, channel geometry, perfluorocarbon percentage, and cell density, oxygen levels in the tissue construct. Their model implements Michaelis-Menten kinetics to form a governing equation that takes into consideration diffusion and oxygen consumption by cells. This governing equation, which appears in other literature (Iyer et al.) as well, is shown in Eq. (9)

$$\begin{array}{c}0=D_t\left[\frac{1}{r}\frac{\partial }{\partial r}\left(r\frac{\partial C_t}{\partial r}\right)+\frac{{\partial }^2{C}_t}{{\partial z}^2}\right]-\frac{V_{O_{2\max }}{C}_t}{K_m+C_t}\hspace{0.17em}\end{array}$$ (9)

where $D_t$ is defined as the diffusion coefficient, $r$ is the radial coordinate, $C_t$ is the oxygen concentration, $z$ is the vertical coordinate, $V_{O_{2\mathrm{m}\mathrm{ax}}}$ is the maximum oxygen consumption, and $K_m$ is the Michaelis-Menten constant [21]. Williams et al. used computational fluid dynamics to model mass transfer including oxygen in engineered constructs and determined the effects of varied oxygen levels on cell growth [22]. Others have published similar models on perfusion bioreactor systems including Yan et al., Pierre et al., Lewis et al., Martin et al., Chung et al., and Croll et al. [23–28]. Many of these models suggest perfusion-based systems improve oxygen delivery and cell density and should be more widely used rather the low current implementation rate. In conclusion, the development of models of oxygen concentration in hydrogels are useful to efficiently screen potential hydrogel constructs; however, in order to improve the accuracy of the results, they must incorporate cell type, oxygen consumption rate, and new culture techniques.

Other recent studies have utilized advances in numerical and computational modeling methods including finite element analysis and computational fluid dynamics to model nutrient transport, shear stresses, mechanical forces, and gel swelling. While computationally expensive the adaption of these techniques can allow for more deliberate designs without the time and expense of in vitro studies on the same wide array of parameters. For example, Lucantonio et al. created finite element models for soft hydrogels to study swelling, bending, elasticity, solvent transport [29–31]. The solvent properties in particular are useful for determining nutrient and oxygen gradients in construct designs. Recent advances in manufacturing techniques have also allowed the creation of more complex scaffold architectures. Manufacturing techniques including three dimensional printing, precision extrusion deposition, selective laser sintering, stereolithography, and fused deposition modeling along with improved computer-aided design software have resulted in constructs in which a wide variety of geometries can be produced [32,33]. These technologies allow the development of patient specific structures as well as the capability of more closely mimicking different tissue types [34]. One study by Malda et al. compared two different scaffold architectures and found that even though oxygen gradients were similar, cell density differed confirming that properly designed scaffolds is important to tissue construct generation [14]. The ability to control and modify parameters including porosity has implications in oxygen availability, nutrient transport, and thus cell density and growth. One limitation of many of these models is that most do not specifically incorporate construct anisotropy or heterogeneity, which are likely present due to production methods. Rather, most models assume bulk properties of a construct rather than spatial variations and the resulting gradients in cell density and oxygen are assumed to arise from a heterogeneous starting point. Additionally, most models assume a steady-state condition and do not account for swelling, remodeling, or mechanical stimuli that would result in from in vivo and long-term culture.

Sensing and Monitoring Oxygen

The ability to sense and monitor oxygen concentrations and gradients within hydrogel constructs allows for the evaluation of local oxygen concentrations that determine a cell's function and survival and can be used to design better constructs to improve cell viability and efficacy. Engineered solutions to monitor and sense oxygen include embedded probes, foils, nanoparticles, and microbeads (Table 1). Figueiredo et al. modeled nutrient transport of oxygen and glucose within a hydrogel (silated-hydroxypropylmethyl-cellulose) with the use of implanted fluorescent oxygen microsensors and glucose sensors into the center of the constructs [35]. Based on their findings, they concluded that the transport of these nutrients could be manipulated with a change in polymer concentration; specifically, there was a strong inverse relationship found between polymer concentration and network density parameters and glucose diffusion, but the relationship was weaker for oxygen diffusion. According to their observations, oxygen levels at the center of the construct seemed to be more dependent on cell density, and oxygen appeared to be the rate limiting factor in cell viability. Kellner et al. were able to noninvasively monitor the partial pressure of oxygen in tissue over time using optical sensor foils containing a luminescent-oxygen-sensitive dye that was placed at the bottom of the well containing a construct [36]. The partial pressure of oxygen at a given location was calculated based on an oxygen-dependent quenching of luminescent light emitted from the sensor. Furthermore, they determined that below an oxygen partial pressure of 11 Torr, tissue development was disrupted. These studies, by directly measuring oxygen levels, provide insight and needed data for the validation of the computational efforts previously discussed.

Table 1.

Oxygen sensors

Sensor type	References
Implantable probe	Figueiredo et al. [35], Simmons et al [42]
Sensor foils	Kellner et al. [36], Westphal et al. [43], Tschiersch et al. [44]
Microbeads	Wang et al. [38], Lesher-Perez et al. [39]
Nano-particles	Koduri et al. [40], Guintini et al. [45], Lee et al. [46]
Directly conjugated probe	Li et al. [41]

Recent studies have developed novel microbead, nanoparticle, and directly conjugated sensors that can be incorporated directly and uniformly into hydrogels. Microbeads are a common device used to locally monitor, image, or detect the transport of nutrients and other processes within a given system [37]. Wang et al. designed biocompatible oxygen-sensing microbeads, which contained a silica core, an oxygen-sensitive dye (Ru(Ph₂phen3)Cl₂), an oxygen insensitive reference dye (Nile Blue), and a polydimethylsiloxane (PDMS) shell [38]. They were able to effectively monitor the oxygen levels by converting the florescent intensity measured by the beads normalized to the reference dye into partial pressures of oxygen. Lesher-Perez et al. developed a method to create a digital “map” of oxygen concentration gradients within three commonly used three-dimensional (3D) cell culture methods—cell spheroids in well, hanging drop cell spheroids, and cell-laden hydrogel constructs [39]. They created this mapping using microbeads they fabricated from PDMS microbeads and Ru(Dpp) (tris(4,7-diphenyl-1,10-phenanthroline) ruthenium(II) dichloride), and phase fluorimetry detection techniques. The authors purport the use phase fluorimetry provided more accurate and consistent oxygen readings than fluorescence intensity detection methods. They also suggest that detecting and understanding oxygen heterogeneities is important when selecting the optimal culture method. Koduri et al. created novel fluorescence-based nano-oxygen particles (FNOPs) to monitor oxygen in 3D cultures [40]. The FNOPs were synthesized from commercially available polystyrene beads, Pluronic F127 as a linker, and oxygen-sensitive dye Ru(Dpp)₃Cl₂. The nanoparticles were tested in alginate spheres containing RIN-m5F and HeLa cells. The FNOPs were nontoxic to the cells and had the sensitivity and range to detect special differences in oxygen across spheres 700–1000 μm in diameter. In order to noninvasively sense the oxygen concentration within tissues over time, Li et al. created injectable hydrogel probes that can be used for electron paramagnetic resonance imaging. A mixture of N-Isopropylacrylamide (NIPAAm), 2- hydroxyethyl methacrylate (HEMA), dimethyl-γ-butyrolactone acrylate (DBA), and bromoethyl methacrylate polymers conjugated with a tetrathiatriarylmethyl radical was used to make the hydrogel probes. These probes offer advantages over other approaches in that conjugating tetrathiatriarylmethyl to the hydrogel increases it stability for up to 4 weeks without diminished oxygen sensitivity, while also having the advantage of noninvasive injectable delivery and degradation products that are bio-eliminable so retrieval of the probes is not required [41]. These new technologies of incorporating sensors directly into hydrogels have the advantage over earlier electrode-based sensors in that they are less invasive and can be used for small-scale cultures.

Perfusion Systems to Improve and Control Oxygen

The ability to control oxygen supply within engineered systems allows for better viability and function of the cells. To more precisely control oxygen and prevent large gradients that can occur with diffusion, several groups have used perfusion system approaches for in vitro cell culture. Another major advantage of these systems is creating a more uniform distribution of cells throughout the thickness of a construct. While standard hydrogel approaches follow simple diffusion with a driving force of concentration gradients, perfusion systems combine the concepts of diffusion and convection to model the flux of fluid entering the hydrogel. Convection-diffusion correlates to Fick's law of simple diffusion but considers velocity and source and sink quantities. Figure 4 creates a comparison of schematics and governing equations of how convection-diffusion in perfusion systems differ from standard diffusion systems. The governing equation of convection-diffusion is

$$\begin{array}{c}\frac{\partial C_A}{\partial t}=D_A{\nabla }^2{C}_A-v·\nabla C_A+R_A\hspace{0.17em}\end{array}$$ (10)

Fig. 4.

A comparison between a simple diffusion experimental setup versus a convection-diffusion setup, which are typically used for perfusion systems. Each type of diffusion follows a governing equation. (a) Simple diffusion follows Fick's second law where C is the concentration of species $t$ is time, and $D$ is the diffusion coefficient. (b) Perfusion systems use the convection- diffusion governing equation where $C_A$ is the concentration of species A, $D$ is the diffusion coefficient, $t$ is the time, $v$ is the velocity, and $R$ is the source/sink quantities of species A: (a) simple diffusion and (b) convection diffusion.

where $C_A$ is the concentration of species A, $D$ is the diffusion coefficient, $t$ is the time, $v$ is the velocity, and $R$ is the source/sink quantities of species A.

Carrier et al. explored how engineered cardiac muscle tissue is influenced based on perfusion rate and oxygen partial pressure using an in vitro construct of rat cardiac myocytes and polyglycolic acid hydrogel [47,48]. They found that perfusion of oxygenated medium through a construct versus suspension of the construct in a spinning flask to mix media improved the concentration of oxygen and other parameters, such as pH, nutrients, and metabolites, as well as resulted in a more uniform cell distribution and higher expression of cardiac makers. They then studied the effects altering of oxygen partial pressure in different constructs via the adjustment of the flowrate and constructs in parallel. They concluded that cardiac muscle's optimal environment for efficiency consisted of high perfusion rates and oxygen partial pressure, likely due to the promotion of cell metabolism. Similarly, Wendt et al. developed an integrated bioreactor system to deliver cells via perfusion to ensure an even initial cell distribution and then to perfuse media during culture to maintain adequate oxygen and nutrients [49]. The cells were subjected to direct perfusion and a normoxic environment, and the resulting oxygen tension was observed using internal and external oxygen sensors. They concluded that in relatively thick engineered tissue, cells exposed to perfusion maintain their even distribution in comparison to static culture of constructs perfusion seeded in the same manner. Further work with the group added oxygen sensors before and after the construct and to create a model to noninvasively predict cell number within construct at the end of culture period based on the current oxygen consumption rate [50]. Another group used a similar approach to noninvasively estimate cell number though their simplified system only require one oxygen sensor [51]. The systems, which can determine cell numbers non-invasively through the dectection of oxygen, are valuable as the field moves toward clinical use beacuse they can be used to evaluate the quality of constructs before implantation. Schmid et al. also developed a perfusion microbioreactor system, which has the unique ability to distribute cells within the engineered construct through an automated perfusion approach, which resulted in a more even distribution of cells when compared to traditional static seeding [52]. The three-dimensional construct maintained a steady oxygen concentration through oxygen microsensors inside of the scaffold and a feedback loop, which adjusted flow to maintain constant oxygen. They predict this bioreactor can pave the way for creating highly functional cell cultures and engineered tissues in the future using their automated, parallelizable, and constant oxygen maintaining device. Other groups have shown similar improvements in cell viability and function in dynamic and perfusion culture approaches including Janssen et al., Volmer et al., and Yeatts et al. [53–55]. It is important to note, as several authors did, that the perfusion systems do subject the cells to some degree of shear stress, which may have different effects depending on the cell type. However, the improved uniformity in cell distribution from increased oxygen is often more beneficial than any potential negative effects of sheer stress. In addition, flow parameters can be tweaked to find a balance between sheer effects and oxygen levels. While modeling and now these studies widely suggest that perfusion improves oxygen delivery to cells, the method still lacks wide adoption. This is likely due to the increased complexity and cost of designing and building a perfusion system over standard cell culture conditions; however, as oxygen levels remain one of the current hurdles to translation of hydrogel technology into clinical use, perhaps more work in the future will employ this strategy.

Microfluidic Models to Sense and Control Oxygen

Microfluidic systems represent a new technology that has been employed to characterize the effects of oxygen at a cellular level, as these systems provide better control in comparison to standard cell culture models. Orcheston-Findlay et al. utilized a microfluidic system to adjust and measure oxygen supply within a two-dimensional cell culture system [56]. This culture was subjected to different levels of oxygen concentrations ranging from hypoxia to hyperoxia. They were able to monitor the oxygen profiles using an integrated sensor film containing an oxygen-sensitive fluorescent dye. With the ability to control oxygen concentration in cell cultures and provide a constant gradient, this microfluidic system allows for more effective imaging and processing of new medical drugs. Ayuso et al. studied tumor models in relation to how oxygen and nutrient starvation influences cell behavior and viability [57]. They created a device they call the microfluidic tumor slice model, in which tumor cells were embedded in a collagen hydrogel with a channel delivering media on the side to mimic a blood vessel. This model has the capability to independently alter nutrients like oxygen and pH levels, as well allow for microscopic imaging during the experiment and retrieval of cells from specific regions for further analysis. Careful control of oxygen and nutrient access to cancer cells will allow for the development of new medical treatments for cancer patients and will have implications in a variety of other research areas. Another study created a PDMS microfluidic device structured with three channels: one for cell culture, one for oxygen scavenging reactions, and one for oxygen-generating reactions [58]. This allows oxygen to be precisely controlled without the use of large volumes of chemicals or pressurized gas cylinders to control a larger incubator environment. This system can be applied to a wide range of experimental applications including modulating carbon dioxide or nitric oxide and is microscope compatible to allow noninvasive assessment of cell response and behavior to different oxygen concentrations. Taken together, these and similar microfluidic systems allow the investigation of effects of varied oxygen levels at a cellular level through the ability to carefully control gradients as well as image the cells with microscopy in real-time. These systems also require smaller spaces and media volumes than standard hypoxia chambers or oxygenated perfusion systems. Data generated from these systems will allow us to more accurately determine factors such as oxygen consumption rate and in turn to develop more accurate models.

Oxygen Delivery by Hydrogels

One approach to overcoming diffusion limitations, which impedes oxygen's availability, is to develop hydrogels that can deliver oxygen to the cells inside the construct and surrounding tissues. The most common ways to increase oxygen delivery are the use of perfuorocarbons (PFCs) and peroxide-containing compounds. Perfluorocarbons can improve proper oxygen delivery to cells due to their higher capacity for oxygen compared to that of water [59,60]. The general molecular structure of PFCs is C_xF_y. They are hydrophobic, and thus are often mixed with a surfactant emulsifier for incorporation into media or hydrogels. Early work explored the use of a perfluorocarbon-based synthetic oxygen carrier, Oxygent^TM, in the culture media to improve oxygen delivery to engineered constructs [21]. In an experimental study with rat cardiomyocytes and fibroblasts, they found that oxygen consumption and cell density were increased and the excitation threshold for contraction was decreased in the PFC media constructs indicating improved cell conditions. They also created a model to fully analyze perfluorocarbons' effect on the oxygen diffusion within the construct allowing them to explore different geometries and flow rates to optimize future constructs. However, while PFCs in the media did increase oxygen diffusion rate and delivery, the center of the constructs still experienced low oxygen environments. To address this, several groups incorporated PFCs into hydrogels to more directly increase oxygen within the constructs. Chin et al. studied perfluorooctyl bromide (PFOB) immobilized in alginate beads to analyze oxygen's transport specifically within liver cells [61,62]. They found that cells in the PFOB containing beads had increased metabolism in comparison to alginate-only beads. They also developed a model for oxygen levels within the beads and found that the predictions aligned closely with their experimental results. White et al. also investigated an alginate hydrogel with addition of the perfluorocarbon PFOB for oxygen delivery [63]. Specifically, they focused on analyzing the mechanical properties and transport of various molecules. They found that the mechanical behavior at small strains was unchanged by the addition of PFOB (and surfactant), but the fracture stress valuedecreased in higher percentage PFOB constructs when subjected to higher compressive strains. They also found that transport of small hydrophilic molecules was not changed but the transport of small hydrophobic and larger proteins diffusion was changed. These findings suggest that while PFCs are able to increase oxygen, other parameters must be considered when designing hydrogels. Taking a different approach, Niu et al. created an injectable hydrogel that is able to deliver cells in a noninvasive manner and incorporate perfluorocarbons to deliver oxygen [64]. This hydrogel was a mixture of N-isopropylacrylamide (NIPAAm) with acrylate-oligolactide, HEMA, and a PFC macromer—methacrylate-poly(ethylene glycol)-perfluorooctane (MAPEGPFC). Oxygen partial pressure in the gels was measured at 1% O₂ culture conditions, and the pO₂ of the gels with PFC was consistently higher than the gel alone (gels with 10% PFC being greater than the 5% gels). MSCs cultured in the gels had increased proliferation over 14 days in the gels with PFC in both normoxic and hypoxic conditions (with 10% PFC the highest). However, it is also important to note that the mechanical properties of the gels differed with the 10% PFC gel significantly less stiff (lower Young's modulus).

The other common strategy for designing oxygen delivering materials is to incorporate a peroxide containing compound that will release oxygen as it breaks down. The common peroxide compounds are calcium peroxide, sodium percarbonate, and magnesium peroxide. Equations (11)–(14) detail the chemical reactions in which peroxide releases oxygen within a hydrogel [68].

$$\begin{array}{c}\text{Ca}{\mathrm{O}}_{2\left(\mathrm{s}\right)}+2{\mathrm{H}}_{2}O\to \text{Ca}{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}+{\mathrm{H}}_2{\mathrm{O}}_2\hspace{0.17em}\end{array}$$ (11)

$$\begin{array}{c}Mg{\mathrm{O}}_2+2{\mathrm{H}}_{2}O\to Mg{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}+{\mathrm{H}}_2{\mathrm{O}}_{2\hspace{0.17em}}\end{array}$$ (12)

$$\begin{array}{c}{\left[{\mathrm{Na}}_2{\mathrm{CO}}_3\right]}_2.3{\mathrm{H}}_2{\mathrm{O}}_2\to 2{\mathrm{Na}}^{+}+2{\mathrm{CO}}_3^{-}+3{\mathrm{H}}_2{\mathrm{O}}_2\hspace{0.17em}\end{array}$$ (13)

$$\begin{array}{c}2{\mathrm{H}}_2{\mathrm{O}}_2\to {\mathrm{O}}_2+2{\mathrm{H}}_{2}O\hspace{0.17em}\end{array}$$ (14)

Using this strategy, Harrison et al. created Poly(D,L-lactide-co-glycolide) (PLGA) films containing sodium percarbonate (2 Na₂CO₃ · 3 H₂O₂) that released oxygen over 24 h in vitro and decreased tissue necrosis in an ischemic skin flap model in rats [65]. Oh et al. created PLGA scaffolds containing calcium peroxide (CaO₂), which reacts with water to generate H₂O_2, which further degrades releasing oxygen (O₂) [66]. In vitro testing demonstrated the hydrogel released oxygen up to 10 days and improved cell viability in hypoxic culture conditions. Fan et al. created an injectable hydrogel comprised of NIPAAm/HEMA/acrylate-oligolactide polymers that was able to be injected through a 28-gage needing and gel in less than 7 s at 37 °C [67]. The oxygen was delivered via microspheres, which has a PLGA shell and poly(N-vinylpyrrolidone)/H₂O₂ and catalase in the core. Oxygen was released as catalase converted the H₂O₂ to oxygen and water. They tested the oxygen delivery system in a myocardial infarction model and found it provided sufficient oxygen to improve outcomes in infarcted tissue in terms of cell viability and heart function. Table 2 provides a summary of these different oxygen delivery strategies with additional references. In summary, the incorporation of perfluorocarbons in media and gels or enzymatic release of oxygen offers advantages to cells embedded within hydrogel scaffolds including higher oxygen diffusion rates, which allows for an increase of cell viability and proliferation. Figure 5 details the main approaches for optimization of oxygen delivery within hydrogels.

Table 2.

Oxygen delivery techniques

Delivery type	References
PFCs in media	Iyer et al. [21]
PFCS in gels	Chin et al. [61], Khuttak et al. [62], White et al. [63], Niu et al. [64]
Peroxide containing compounds	Harrison et al. [65], Oh et al. [66], Fan et al. [67], Pedraza et al. [69], Li et al. [70], Ng, et al. [71]

Fig. 5.

A schematic detailing the different approaches to optimize oxygen delivery with the use of PFCs. Different experimental set-ups include PFCs within the media (a), PFCs inside of a hydrogel (b), and peroxide molecules embedded within a hydrogel (c).

Effects of Different Gels on Oxygen

While most studies examine only one hydrogel type, several studies directly compare and contrast different hydrogels, which is useful when designing new constructs. Guaccio et al. investigated the effects of two different hydrogels, collagen type I and agarose, on oxygen consumption of bovine chondrocytes [72]. By assessing the relative oxygen concentration within each hydrogel via the use of phosphorescence quenching microscopy, they found that the oxygen supply depleted faster in the agarose hydrogel compared to the collagen type I. The diffusion within the two gels was not predicted to be different based on diffusivity calculations; thus, they concluded that the material of a hydrogel can highly affect cell behavior and activity and ultimately influence proper tissue development due to other factors including mechanosensing, construct stiffness, and cell adhesion/integrin binding. Another study compared two different scaffold structures (compression molded sponge and 3D deposited fiber) and found no differences in oxygen levels, but they did see differences in cell density in the interior between the two groups [14]. Both scaffolds had significant decreases in oxygen over the first 14 days of culture, which then plateaued at a nonzero value. Stoppel et al. studied the incorporation of the surfactant F68 in alginate hydrogels and its effects on the transport of molecules, such as oxygen and proteins [73]. They extended their previous work of F68 and similar nonionic surfactants, which were used to adjust mechanical parameters and to aid delivery of molecules, such as perfluorocarbons, for oxygen delivery. This most recent investigation specifically examined the effects of the surfactant on the transport on two prototypical cargos, small molecules (riboflavin) and large proteins. They found that it had little effect on small molecule transport but variable effects on larger protein transport depending on concentration and gelation method. The addition of F68 also had significant effects on the water retention (swelling) and mechanical properties of the hydrogel. This suggests that attention should be given when using the surfactant in gels to deliver oxygen with PFCs as described above. These studies show that it is important to consider the limitations and benefits of specific gel types when designing a new tissue engineering approach and that further research is required to understand the underlying mechanisms behind differences in gel types and production methods.

Strategies for Clinical Applications

One area of promise for translation of hydrogel materials to clinical applications is advanced wound-healing dressings. This medical application can release necessary drugs into the wound to aid the healing process, absorb any excess fluid around the wound, and allow for oxygen permeability to a specified area to support cell growth and allow for a rapid recovery. Singh et al. created several novel hydrogel films from gum acacia polysaccharide, Carbopol (polyacrylic acid-based polymer), and the antibiotic gentamicin and the polysaccharide sterculia crosslinked to poly(VA) or poly(VA-co-AAm) (poly(VA)-AAm) and antibiotics tetracycline and gentamicin [74]. They found that the films had sufficient oxygen permeability and drug release in addition to other factors, such as blood compatibility and mechanical properties. The hydrogel wound-healing dressings also significantly improved wound recovery quicker than open wounds. The Leipzig research group also developed a hydrogel for wound healing; however, this group studied the addition of three different perfluorocarbons to determine, which best controls oxygen release into the wound for optimal dermal healing [75,76]. They created hydrogels from methacrylamide chitosan (MAC) with either pentafluoropropionic anhydride (MAC(Ali₅)F), pentaflourobenzaldehyde (MAC(Ar₅)F), or pentadecafluorooctanoyl chloride (MAC(Ali₁₅)F). They found that the formulation MAC(Ali₁₅)F with the most fluorines had the highest oxygen uptake and release, a higher cell number, and an increased cell metabolism. Because this formulation was also the stiffest, it is suggested that the effects on multiple parameters should be considered when selecting the best formulation for each application. They also compared the MACF gels and the MACF gels saturated with oxygen before application to the commercially available Derma-Gel^TM in a porcine model of wound healing. They analyzed the healing benefits in terms of re-epithelialization, collagen synthesis, neovascularization, and several other parameters. They deemed the oxygenated hydrogel to be effective for wound closure due to its stimulation of blood vessel formation and its unique collagen synthesis pathway. Another study by Jee et al. created a carbopol and poly(ethylene glycol) hydrogel system that incorporated skin-permeable growth factors, anti-oxidants (Quercetin), and PFCs to test if the construct can expedite tissue repair [77]. Their results showed that their system indeed promoted quicker diabetic wound healing in mouse model as well as in a number of in vitro studies. The use of novel wound repair materials is very promising, especially in the setting of diabetes. While one of these studies was in a porcine large-animal model, more studies need to be done facilitate a more rapid translation of these advances in hydrogel technology from small-scale animal models to clinical applications.

One of the major hurdles that remains as constructs are scaled up to larger sizes for use in humans is insufficient oxygen supply. While adaptations, such as perfusion culture systems discussed above have helped solve the size problems with in vitro culture, there still remains a need to create better solutions for in vivo applications. Approaches to this problem include improving mass transfer properties within the gels themselves and promoting faster vascularization of the tissue in vivo. Recent advances in hydrogel design and fabrication have shown promise toward achieving better outcomes and clinical use. Madden et al. created poly(2-hydroxythyl methacrylate-comethacrylic acid) hydrogel with interconnected spherical pores that formed a larger channel structure [78]. The dimensions of the properties were designed to optimize passive mass transfer, vasculature integration, and cardiomyocyte organization. They found that human embryonic stem cell derived from cardiomyocytes survived throughout the construct for two weeks of static culture. Acellular constructs implanted in rat hearts showed neovascularization within the construct at four weeks. Thomson et al. used a similar structural design to fabricate fibrin hydrogels for cardiac tissue engineering [79]. An advantage of these constructs is that fibrin is a natural polymer with tunable degradation and mechanical properties. It degrades into bioproducts that can promote angiogenesis. Recent work by Lee et al. studied three-dimensional microchannel networks in enzyme cross-linkable gelatin hydrogels through the dissolving of poly(N-isopropylacrylamide) fibers in both micro- and macro-sized channel arrays [80]. This hydrogel network was implanted in a mouse model of ischemia. The study found that the microchannel network, which was similar in size to capillaries, improved recovery by several measures in the hindlimb ischemia model and in the future could be development into more advanced therapeutics. Arakawa et al. used a novel approach of creating three-dimensional vascular networks in a poly(ethylene glycol) tetrabicyclononyne hydrogel that contained an RGDS peptide sequence (RGPQGIWGQGRGDSGK) as well as ortho-nitrobenzyle ester crosslinker [81]. This crosslinker allows the hydrogel to be photodegradable by infrared light. Thus, using multiphoton lithography, they were able to create complex networks with channels of various sizes mimicking the vasculature within the hydrogels. These gels were also capable of supporting cells seeded throughout the gel during polymerization and endothelial cells lining the lumen. The authors hope this technique allows the creation of complex three-dimensional networks that will hopefully improve vascularization once implanted in vivo. These recent advances in the field show that promise that hydrogel can be translated into more widespread clinical use.

Conclusion

A stable oxygen supply is crucial for a cell's proper growth and function. While many studies utilize hydrogels due to their flexibility and porous nature, one big challenge hydrogels face is the limited supply of oxygen and nutrients in cells as the construct size increases. This review outlined studies that tackle this challenge by providing techniques for analyzing oxygen's behavior within hydrogels and detailing design applications. This review mapped and classified the various experimental, computational, sensing, and product-driven literature. First introduced were studies modeling oxygen within hydrogels to create a more efficient and effective design for hydrogel constructs. Next, studies were highlighted that sense and analyze oxygen concentration and gradients to confirm findings of the computational models and to determine relative conditions within the hydrogels. Finally, several strategies and technologies that improve oxygen transport within hydrogels to allow for the development of successful treatments were discussed including perfusion culture systems and oxygen releasing gels. While individual research efforts have made significant progress toward improving oxygen supply to both in vitro and in vivo cultures, some of the findings have yet to be widely accepted, and thus we hope this comprehensive overview of the topic will promote increase consideration of oxygen parameters in future research. In order for progress and more widespread translation of hydrogel tissue engineering constructs to be implemented, oxygen supply and vascularization should be considered as one of the key parameters when designing therapies. The papers described in this review suggest that a wide variety of approaches are available, but more studies and widespread adoption of these strategies are required before tissue engineering can truly revolutionize the current medical therapies.