Shape-defining alginate shells as semi-permeable culture chambers for soft cell-laden hydrogels

Soft hydrogels have a porous structure that promotes viability and growth of resident cells. However, due to their low structural stability, these materials are fragile and difficult to culture in vitro. Here we present a novel approach for the 3D culture of such materials, where a shape-defining, semi-permeable hydrogel shell is used to provide mechanical stability. These thin hydrogel shells enclose and stabilize the soft materials while still permitting gas and nutrient exchange. Custom alginate-shaped shells were prepared using a thermosetting, ion-eluting hydrogel mold. In a second step, the hydrogel shells were filled with cell-laden infill materials. As an example of the versatility of this technique, materials previously not available for tissue engineering, such as non-annealed microgels or low crosslinked and mechanically unstable hydrogels, were used for tissue culture. Primary human chondrocytes were cultured using this platform, to evaluate its potential for cartilage tissue engineering. To prove the scalability of this technique, anatomically-shaped ears were cultured for 3 weeks. This novel approach has the potential to radically change the material property requirements in the field of tissue engineering: thanks to the shape definition and stability provided by the hydrogel shells, a wide range of materials previously inaccessible for the manufacture of 3D tissue grafts can be re-evaluated.


Introduction
Fabrication of tissues and organs that can correctly replicate the matrix content and mechanical properties of native tissues has long been a challenge. Recent advances in materials science have led to the development and characterization of a variety of hydrogels [1] targeted at providing excellent biocompatibility and mechanical properties for the fabrication and printing of tissue implants. Tissue-engineered constructs produced using hydrogels and bioinks with stiffer mechanical properties are easier to handle and manufacture. For this reason, until recently, soft and uncrosslinked materials with excellent biocompatibility but mechanical stiffness inadequate to retain defined 3D shapes have mostly been neglected in the field of biofabrication [2].
Recent studies [3][4][5][6][7] have shown an inverse correlation between the stiffness of a hydrogel and the bioactivity of the embedded cells; i.e. faster migration, better differentiation, and faster matrix remodeling have been seen in softer hydrogels compared to stiffer alternatives. Many strategies have been proposed for increasing the stability and mechanical properties of biofabricated structures by altering the materials or components of hydrogels to enhance handling and shape retention. For example, different hydrogel concentrations can be used together to provide both structural support and a soft hydrogel matrix for cells to proliferate [8]. Alternatively, adding reinforcement structures to soft hydrogels generates structures with stiff macroscopic properties that still retain a softer microenvironment for cells [9][10][11]. Other approaches focus on the inclusion of nanoparticles or nanofibers to the hydrogel mixture to generate viscous and stiff composite hydrogels [12][13][14]. Finally, biodegradable polymers may be used to support the biofabricated samples by tailoring a scaffold's degradation rate to the maturation of the matrix deposited by the cells [15,16]. However, the inclusion of such components may lead to serious side effects, such as the inability of cells to proliferate or migrate in stiff hydrogels [17,18], toxicity of the nanoparticle components [19], and, in the case of biodegradable reinforcements, a possible accumulation of the degradation byproducts [20].
A novel reinforcement approach that does not alter the hydrogel composition or require reinforcement structures may be achieved by culturing cellular constructs inside a confining insert or chamber. The porosity of such a chamber, however, must not hinder nutrients and molecule transfers or gas exchange [21]. At the same time, the pore size should be limited, to prevent cells and parts of the cultured construct from escaping the confinement. Commercially available transwell plates already fulfill the described requirements by using membranes with typical pore sizes in the range of 0.4-100 µm. Such inserts are typically used for 2D or 2.5D cultures; and although transwell systems have been used for 3D tissue engineering constructs with relevant propagation in the z-axis, created tissues are still sheet-or disc-shaped [22,23] and are therefore not suitable for the manufacture of custom 3D structures.
At the same time, many different tissue-chamber models have been developed to confine implanted tissues in vitro. Implementation, however, has proved to be rather crude, with cylindrical chambers made of inert materials like metal, glass or Teflon [24][25][26][27]. These constructs often have open sides and rather large pores, to permit tissue infiltration. Closed porous chambers used to implant tissues into hosts have been described in literature but are only applicable to flat tissue sections [28]. While these tissue chambers are used to study inflammation, angiogenesis, and vascularization processes, they are unsuitable for the proposed attempt to create de novo tissue in vitro [29,30].
Many studies have demonstrated the possibility of casting tubular multi-layered structures [31][32][33] on the surface of a core structure that has been preloaded with a crosslinker of interest, such as CaCl 2 . As the construct is immersed into baths containing CaCl 2crosslinked hydrogel precursors, the crosslinker diffuses out of the core structure into the bath, and a new hydrogel layer forms on top of the core structure. These methods allow for easy biofabrication of simple structures, but they require high amounts of hydrogel precursor in which the core structure must be immersed. This ultimately leads to a waste of hydrogel precursor and cells for the fabrication of biologically relevant tissues. At the same time, the core structure usually remains trapped within the crosslinked layers, unless an opening is generated by cutting the extremities of the generated tubular structure.
These drawbacks can be avoided by enclosing the cell-laden material within a thin semi-permeable hydrogel shell that retains the desired shape while still permitting the diffusion of nutrients. By exploiting the slow diffusion time of ionically crosslinkable hydrogels such as alginate, a carrageenan-based molding material was generated and preloaded with calcium ions. The formed ion-eluting mold was used to fabricate thin hydrogel shells by crosslinking only the outer section of the cast polymer. By removing the uncrosslinked cast polymer after a specific amount of time, it was possible to generate thin hydrogel shells to be used as culture chambers for cell-laden materials. This semi-permeable shell confined cells and soft materials without hindering gas exchange or transport of nutrients and bioactive cue molecules. In a second step, the alginate shells were sealed to create a tissue chamber and were cultured in vitro. In culture, the alginate shells remained stable and confined cell-laden infill material to inside the chamber. The thickness of the alginate layer is tunable through timecontrolled exposure to the ion-eluting hydrogel mold. At the same time, thanks to the elastic properties of the carrageenan molding material, complex and biologically relevant shapes were realized without additional components or support materials. Manufacture of alginate shells is facile and fast, does not need any specialized equipment, and requires only cheap natural polymers. As a first proof of concept, materials unstable in aqueous solutions, including commercially available dermal fillers, have been used to biofabricate cartilage tissue. Finally, to demonstrate the possibility of using such techniques for large-scale implants, an anatomically shaped ear construct was manufactured and cultured in vitro for 3 weeks.

Materials
All chemicals were purchased from Sigma-Aldrich and used as received unless stated otherwise. K-carrageenan was received as a free sample from C.P.Kelco, ultrapure alginate (UP LVG, average molecular weight: 192 kDa) was purchased from NovaMatrix, and hyaluronic acid with an average molecular weight of 1.7 MDa was bought from HTL Biotechnology. Injectable hyaluronic acid dermal fillers of the trade name Gloderm were purchased from easinject GmbH.

Preparation of sacrificial hydrogel mold
Protective hydrogel shells were prepared by way of a casting technique that uses a cation-eluting hydrogel mold. To create this eluting hydrogel mold, k-carrageenan and locust bean gum were added in different concentrations (1%-3%) into a 100 mM CaCl 2 solution and fully hydrated at 4 • C. Polymer slurry was then heated to 95 • C until full dissolution. Three-dimensional objects with the desired outer shape of the hydrogel shells were designed in Fusion 360 (v 2.0.8176) and were 3D printed using an Anycubic Photon, LCD-based SLA Printer using PrimaCreator Value White DLP Resin. Threedimensional printed samples were washed with 100% ethanol and solidified by 45 min of UV exposure at 405 nm. Polymer solution heated to 95 • C was poured onto the 3D object and left to cool to room temperature, to form an ion-eluting, tough, flexible hydrogel to be used as a negative mold. For the casting of complex shapes, the mold can be 'sacrificed' , i.e. cut into pieces to allow for the extraction of the casted content.

Preparation of hydrogel shells
To create cell culture shells, a two-step molding procedure was used. As a first step, the main body of the cell culture shells was generated with an opening on one side to permit the removal of residual material and the injection of cell-laden material. In a second step, the opening was closed to form a sealed cellculture shell. In brief, a solution of low-viscous alginate (0.5%-2%) was poured into the cavity of the negative eluting mold. Diffusion of calcium ions from the negative mold into the alginate solution immediately triggered gelation and formation of a thin layer of gelled alginate on the inside of the negative mold. After a defined time, uncrosslinked alginate was removed to stop further thickening of the shell. The crosslinked alginate layer remained in the mold as the uncrosslinked polymer was extracted. Cell-laden material was then injected into the body of the shells and the top opening was covered with alginate solution. A top plate of the calcium-eluting hydrogel was added to crosslink and bind newly added alginate to the main body. After 30 min, the cell-laden material enveloped by the alginate shell was extracted from the negative mold. A step-by-step guide of the complete process can be found in figure 1 while an overview of all conditions used as infill can be found in table 1.

Measurement of thickness
Alginate solutions of varying concentrations (0.5% (w/v), 1% (w/v), 2% (w/v)) were prepared and poured into eluting molds prepared with 100 mM CaCl 2 to initiate gelling of cell culture shells. Uncrosslinked alginate solution was removed after defined crosslinking times (20 s, 60 s, 180 s), and cell culture shells were washed with deionized H 2 O twice to remove any uncrosslinked residual alginate solution from the inside of the cell culture shells. The thickness of the formed shells was imaged with a zoom microscope (Leica M205 FA). The area of the crosssection was measured with Photoshop (Adobe, version 10.0.18362.329) to calculate average thickness of the samples.

Alginate sulfate
The synthesis of alginate sulfate was carried out as previously reported [34]. Briefly, ultrapure, highviscosity alginate was dissolved in formamide under continuous stirring. 96% chlorosulfonic acid was added dropwise to achieve a final concentration of 2% (v/v). The solution was left to react for 2.5 h at 60 • C under continuous stirring. The alginate was precipitated by addition of ice-cold acetone, and the precipitate was isolated by centrifugation. The precipitate was re-dissolved in ultrapure water and constantly neutralized with 5 M NaOH. After the polymer was completely solubilized and pH stabilized, the solution was dialyzed against 100 mM NaCl and deionized water before samples were snap frozen and lyophilized. Dried samples were stored at −20 • C until use.

Hyaluronic acid methacrylate (HA-MA)
High-molecular-weight hyaluronic acid was added to ultrapure water and kept at 4 • C until the polymer was completely solubilized. Pre-cooled dimethylformamide (DMF) was added under continuous stirring to reach a final ratio of water to DMF of 3:2. The reaction was started by the addition of methacrylic anhydride. 10 M NaOH was added as required to keep the pH between 8 and 9. After 4 h, solid sodium chloride was added (final concentration 0.5 M) and the polymer was precipitated with pure ethanol. The precipitate was washed, dried and re-dissolved in ultrapure water. Further purification was done with a diafiltration unit (Äkta 3, 10 NMWC hollow fiber), before the product was lyophilized and stored at −20 • C until use. To analyze the degree of substitution, polymer was dissolved in deuterium (Cambridge Isotope Laboratories) and characterized with 1H nuclear magnetic resonance (NMR) spectroscopy. A Bruker AV-NEO 600 MHz spectrometer equipped with a TCI CryoProbe was used. Spectra were obtained with 1024 scans using a 5 s recycle delay. To determine the degree of substitution, the ratio of the sum of the integrated peaks of the methacrylate protons (peaks at ∼6.1 and ∼5.6) and the integrated peak of the methyl protons of HA (∼1.9 ppm) were compared. Degree of substitution was found to be 0.28. To create cell-laden filling materials, cells were mixed with microgels and polymer solutions using a doublebarrel syringe and a static mixer system (DN2.0x16, Medmix), to achieve a final concentration of 10 million cells ml −1 .

Mechanical evaluation
Mechanical evaluation was performed with a Bioindenter (UNHT 3 Bio, Antoon Paar) equipped with a 500 µm spherical ruby indentation probe. After peeling the alginate shell off from each sample using tweezers, replicates were cut in half along either the coronal or transverse axis (for more details refer to figure S4). One half of the sample was used for indentation and the other half for histological analysis. Samples were kept in PBS at 4 • C until use, before being immobilized on a 35 mm-diameter Petri dish and submerged in 0.9% NaCl for indentation. For each sample, indentation was performed at three different locations: in the center, middle and edge regions. The measurement for each region was repeated three times. The indentation protocol was set up as follows: first, a 15 µN force was applied to detect the surface of the sample. Subsequently, samples were indented 75 µm within 5 s and measured force was acquired. Data was analyzed in the Anton Paar's software to determine the Hertz modulus of each tested location. Data was plotted in Graphpad Prism (v 7.04) by averaging the three replicates at each location.

Histology
Samples were washed repeatedly with 0.9% NaCl and fixed in 4% paraformaldehyde for 2 h at room temperature. Dehydration was done by serial baths of

Diffusion study through the alginate shells
Umbelliferone was purchased from Sigma-Aldrich. Green fluorescent protein (GFP) was synthetized as previously reported [35]. Briefly, superfolder GFP (sfGFP) was recombinantly produced using a pET28b vector in BL21 (DE3) Escherichia coli (New England Biolabs). Bacterial cultures were grown at 37 • C and 220 rpm in 500 ml shaker flask culture overnight using ZYM-5052 autoinduction medium supplemented with 50 ug ml −1 kanamycin (Sigma-Aldrich). After spinning down the cultures at 3500 rcf and 4 • C for 10 min, the bacterial pellet was resuspended in 5 ml of BugBuster ® HT Protein Extraction Reagent (Novagen). The suspension was incubated at room temperature under gentle agitation for 20 min, and the insoluble contaminants were removed by centrifugation at 15 000 rcf and 4 • C for 20 min. The sfGFP supernatant was used as received and the GFP concentration quantified using UV/Vis spectroscopy (Plate Reader, Synergy H1, BioTek Instrument). The GFP and Umbelliferone solutions were then added to a container containing a pre-crosslinked 1% Alginate Hydrogel, and their diffusion through the sample was recorded using a digital USB microscope (figure S3).
Mean ± standard deviation is reported for all data acquired.

Ion-eluting molds enable custom-shaped hydrogel shell creation
To create custom-shaped alginate shells, the 3Dprinted positive of a half-sphere joined to a cylinder was placed into a container filled with a calciumloaded carrageenan solution and cooled down to form an ion-eluting hydrogel mold. Next, the ioneluting mold was cut close to the object to permit extraction of the positive. Due to the toughness and flexibility of the carrageenan hydrogel, even intricate models such as the cartilage part of a human ear could be removed from the mold without damaging the mold. Manufacture of the alginate shell was conducted in a second process after the mold was created. First, alginate was poured into the mold (figures 1(B) and (C)). The calcium ions present within the ioneluting mold started to diffuse into the alginate polymer solution, crosslinking a thin alginate shell at the interface with the ion-eluting mold ( figure 1(D)). Once the desired shell thickness was reached, the non-crosslinked alginate was removed, leaving a thin alginate shell in the shape of the initial 3D-printed positive on the walls of the mold ( figure 1(E)). The opening left by the removal of the uncrosslinked alginate was used to fill the alginate shells with cellladen material (figures 1(F) and (G)). To fully seal the infill material within the hydrogel shell, an additional layer of alginate solution was poured onto the infill ( figure 1(H)) and covered with the closing layer of the ion-eluting mold ( figure 1(I)). The alginate solution crosslinked and bonded with the already-gelled alginate shell, sealing the construct. The structure created by enclosing an infill material with an alginate hydrogel shell will henceforth be referred to as a culture chamber.

Alginate shell thickness can be tuned with concentration and gelation time
The thickness of the hydrogel shell plays an important role in the culture of cellular materials. Thicker shells have a higher mechanical stiffness, which helps with the handling of the shell during fabrication and to keep the infill materials sealed during culture. At the same time, thin shells are desirable, to minimize the diffusion barrier between the culture media and the infill material. Using diffusion coefficients of CaCl 2 in aqueous solution previously reported [36], a finite element analysis (FEA) simulation was performed to predict the diffusion of calcium chloride over time (figure S1). Figure S2 displays a scatter point graph, predicting the diffusion distance at different times based on the FEA. The simulations were validated by producing a range of alginate shells (figure 2(A)) at different crosslinking times (20 s, 60 s, 180 s). Additionally, the polymer concentration (0.5%, 1%, 2%) was varied, to test the mechanical stability of the generated shells. All the alginate shells formed with 0.5% alginate and all samples crosslinked for 20 s were difficult to handle without damaging the hydrogel shell. Thickness variations between different alginate concentrations were found to be meaningful but can only explain 3.5% of the variation (p = 0.027). Time, however, was a highly meaningful factor (p < 0.001). Samples crosslinked for 180 s had significantly increased thicknesses compared to those crosslinked for 60 s (p < 0.001; 60 s: 308 ± 47 µm; 180 s: 609 ± 64 µm). No significant difference in thickness was found for samples crosslinked for 20 s or 60 s (p = 0.45). The FEA simulations correctly predicted the diffusion distance and consequent crosslinking thickness due to 100 mM CaCl 2 solution diffusing in the alginate gel from the surrounding eluting mold. The lowest concentration and crosslinking time that still resulted in stable and easily processable hydrogel shells was found to be 1% at 60 s with an average thickness of 242.2 ± 26 µm. This condition was used for further experiments.

Alginate shells allow for the diffusion of molecules
To assess whether the addition of the alginate shell would have any effect in terms of the diffusion of chemical molecules from the culture media to the cultured material, the diffusion of two fluorescent molecules through a piece of 1% alginate was recorded using the setup of figure S3. Umbelliferone (≈162 g mol −1 ) and GFP (≈27 000 g mol −1 ) were chosen, as they possess similar molecular weights and diffusion dynamics compared to the major components present in culture media, such as D(+)-glucose (≈180 g mol −1 ) and TGF-β3 (≈12 300 g mol −1 ). As shown in figure 2(B), both molecules could successfully diffuse through the alginate, reaching a diffusion distance of at least 0.5 mm in under 5 min. Umbelliferone's smaller molecular size allowed it to diffuse roughly 1.5 mm into the alginate in 15 min, while GFP required nearly 7.5 times longer to reach the same penetration depth. Nevertheless, since the diffusion distance through the alginate shells is below 300 µm, a fast diffusion of both small and large molecules from the culture media to the inside of the culture chambers is to be expected. It is also anticipated that the diffusion of gases, such as O 2 and CO 2 , will not be impeded by the alginate shell [37], owing to their relatively small molar mass of 32 g mol −1 and 44 g mol −1 which is more than an order of magnitude smaller than that of Umbelliferone.

Culture chambers are stable and sealed environments in vitro
Culture chambers can be filled with cell-laden materials that are otherwise unstable in cell culture medium. Three categories of infill materials were investigated to demonstrate the power of this system: very weak crosslinked hydrogels, a cell-only solution and nonannealed granular microgels. An overview of all conditions used as infill can be found in table 1. Infill materials were combined with primary, human articular chondrocytes to investigate cartilaginous tissue development. Since cells were obtained from young patients, they could be extensively expanded without losing their chondrogenic potential, overcoming the problem of dedifferentiation observed in aged chondrocytes. Being immunosuppressive and therefore suitable for allogeneic use, they provided an excellent off-the-shelf cell source for cartilage regeneration [38]. Cells were used at a concentration of 30 Mio ml −1 in all conditions, apart from the pure cells condition. All infill materials could be successfully incorporated inside the alginate shell and enclosed by a top layer of alginate to form a culture chamber. Culture chambers were cultured in vitro for 3 weeks in chondrogenic medium, after which the alginate shell was peeled off and tissue formation was characterized.
No loss of integrity of the alginate shells was observed over the 3 week culture time and shells were not infiltrated by cells, nor did the infill material escape the culture chamber ( figure 2(C)). Visible differences among the different infill materials were already observed after 1 week of culture. Culture chambers containing alginate-based infill materials (Alg & Alg-S) and the pure cell condition retained their original shape. Contrarily, samples of Glo20, Glo30 and HA-MA exhibited swelling, slightly deforming the culture chamber. As a consequence, the thickness of alginate shells in those three conditions was reduced. After 3 weeks, samples were harvested and visually inspected ( figure S4). All conditions had an opaque, yellowish appearance. The initial shape was retained for all conditions except for the microgel-based infill materials, which had a slightly swollen appearance. Therefore, samples were cut open to investigate their cross-sections. Granular and weak hydrogels matured into stable structures and were further processed. The pure cells condition, however, had not solidified into a stable material inside the culture chamber, and consequently, the material lost shape after the hydrogel shell was cut open. As a result, this condition was excluded from further analysis.

Cell chambers enable the use of novel materials for cartilage tissue engineering
Tissue maturation of infill materials in culture chambers was evaluated by mechanical ( figure 3) and histological analysis ( figure 4). For each sample, Hertz modulus was probed at three different regions (center, middle, edge-as shown in figure 3(A) in grey, red and green respectively) to investigate possible inhomogeneities within the scaffold. The Hertz modulus measured for the different hydrogel conditions after 3 weeks of culture in chondrogenic medium averaged at 17.2 ± 9.8 kPa for the alginate condition, at 10.3 ± 4.0 kPa for the alginate-sulfate condition, at 6.1 ± 2.8 kPa and 9.6 ± 4.4 kPa for the Glo20 and Glo30 conditions respectively and at 16.8 ± 14.8 kPa for the HA-MA microstrands condition. Indeed, alginate and HA-MA samples showed an increased modulus at the edge region compared to the center, but no statistical significance could be identified between the center, middle and edge locations probed within each material.
Starting from a compressive modulus of 4.3 ± 2.1 kPa for the alginate, 3.9 ± 2.7 kPa for the alginate sulfate, 3.2 ± 2.2 kPa for the Glo20, 3.8 ± 1.9 kPa for the Glo30 and 5.9 ± 3.2 kPa for the HA-MA gels, this translates to an average increase of the Hertz modulus after 3 weeks of 300% for the alginate, 164% for the alginate-sulfate, 90% for the Glo20, 153% for the Glo30 and 185% for the HA-MA gels ( figure 3(B)). This stiffness increase is remarkable considering that its only contribution is due to the matrix deposition of the embedded cells and the fact that these materials were not mechanically stable at the start of the maturation process.
Histological evaluation verified the onset of chondrogenic maturation where glycosaminoglycans and proteins of the cartilaginous extracellular matrix were deposited in all conditions (figure 4). In weakly crosslinked alginate hydrogels (Alg and Alg-S), a uniform distribution of glycosaminoglycans, collagen type I and collagen type II were observed. In granular material, hydrogel particles and the extracellular  matrix deposited in the void space in between were clearly distinguishable. Microgels prepared from HA-MA were smaller and more regular in shape and appearance compared to gel particles in Glo20 and Glo30. First infiltration of extracellular matrix (ECM) inside the microgels was visible in both conditions with dermal fillers but not in HA-MA microgels. Spatial organization of ECM was again similar to that of native cartilage, with glycosaminoglycans especially pronounced in the center and collagenous matrix at the edge region.
Even though staining for collagen I as a fibrocartilaginous marker was present, collagen II, the predominant type in healthy cartilage, was also present in all conditions. In alginate and alginate-sulfate samples, an artifact from the production process was visible in the middle of the cross-section. Since the polymer was filled inside the alginate shell as a liquid precursor and immediately sealed with more alginate on top, some of the alginate added to form the shell penetrated the cell-laden infill. This resulted in the formation of a plug of pure alginate in the central region, originating from the top part of the alginate shell. Since the top section was used for histology, these artifacts are visible in the staining. In the alginate-sulfate gels, major structural damage became visible. Even though the outside shell kept the hydrogel in shape, several cracks were visible through the cross-section of the hydrogel. This degradation happened even though ECM was deposited throughout the whole construct. In the alginate and alginate-sulfate samples, staining for glycosaminoglycans and collagens could be observed throughout the cross-section.

Cell chambers enable the casting of large cellular constructs
To demonstrate the scalability of this technique, we fabricated ear-shaped samples comprising of overhangs and fine features (figure 5(A)) using the technique described above. As before, the infill materials were poured into the ear-shaped culture chambers (n = 3). Once sealed, the samples were cultured for 3 weeks. Since literature has shown great potential for low-percentage HA-based polymers [39][40][41], this experiment was performed using a low-percentage enzymatically crosslinked HA hydrogel (HA-TG), with and without the addition of uncrosslinked HA, and low alginate percentage (table 2). To ensure comparability between the histological and mechanical results obtained in this new set of materials, human infant chondrocytes were embedded at a concentration of 30 Mio ml −1 , as done in the previous experiments.
Tissue maturation of ear constructs was evaluated by indentation and histological analysis. Hertz modulus was probed on each ear ( figure 5(B)). The 0.25% alginate samples showed an average modulus of 5.41 ± 0.11 kPa; 1% HA-TG samples had an average modulus of 7.93 ± 1.19 kPa, while 1% HA-TG+1%HA samples showed an average modulus of 14.51 ± 2.96 kPa. The Hertz modulus measured for the new set of hydrogels in table 2 is lower than the previously measured moduli in table 1. Although excellent biocompatibility and matrix maturation has already been reported in literature for these materials [39,41], the larger size of the constructs has a great impact on the overall maturation time of the construct. As a matter of fact,  figure 5(C). This revealed the presence of an extracellular matrix rich in glycosaminoglycans and similar depositions of both collagen I and collagen II in all three conditions. The excellent biocompatibility, matrix maturation, and consequent stiffness increase observed for these materials are comparable to previously reported results [39,41]. Even though the alginate shell thickness has proved not to hinder the diffusion of nutrients to the infill material, as the overall size of the construct and the number of embedded cells increases, the passive diffusion of nutrients from the media may turn out to be too slow. The implementation of a hydrostatic bioreactor could improve the overall culture and stiffness maturation of these larger samples.

Conclusion
Soft hydrogels with low polymer content are rarely employed in tissue engineering due to their low shape retention and mechanical stability, which render the fabrication of complex 3D structures challenging. To tackle such issues, commercially available semipermeable membranes, e.g. diafiltration membranes, could be used to confine the low-crosslinked, cellladen material and to separate it from the remaining cell culture reservoir. However, such a technique would pose significant limitations in terms of potentially manufacturable 3D shapes and sizes.
In this study, a new biofabrication technique for the facile generation of large and highly detailed constructs is introduced. With respect to commonly used tissue culture methods, this technique is based on culturing and confining cell-laden materials inside a custom-shaped alginate shell. This shell is prepared using a sacrificial molding technique capable of achieving complex structures, overhangs, and high resolution with no additional steps. The hydrogel shell function is twofold, as it allows for soft hydrogels to be stabilized in a desired shape, while at the same time acting as a semi-permeable membrane. This second property allows for the confinement of cells and high molecular weight polymers to the inside of the construct, while still permitting gas and nutrient exchange with the outside. Although additional studies are required to ensure that the shells have no impact on the diffusion of nutrients and gasses, culture chambers can be submerged in culture media and processed like any other biofabricated construct.
As shown in the histology stainings, cells remained alive and metabolically active even in the inner parts of constructs, where the diffusion is most difficult. Moreover, biochemical cues could efficiently diffuse through the alginate shell, reaching the inner part of the construct and instructing cells to produce the desired ECM. The possibility of using various biological components as an infill that can be incorporated with a low degree of substitution is a key point of this technique. Adding the necessary components in a biologically accurate composition provides the cells with the necessary biochemical clues and adhesion sites to start maturing in the desired way and helps to maintain their phenotype. The low level of structure of the inner scaffold permits a more efficient rearrangement of the scaffold by the cells, which can then easily start ECM deposition.
A further crucial role played by the alginate shells is the structural support and shape definition they provide. Even though several biofabrication methods for preparing custom-shaped scaffolds have been developed over recent years, each comes with its own disadvantages, such as severe material restrictions and/or the requirement of a highly specialized setup. These two issues present major hurdles for the adoption of such techniques by the scientific community and for clinical applications. The method of using sacrificial hydrogel molds is straightforward and comparable to the use of PDMS casting, a simple and established technique. Custom-shaped hydrogel shells can be created and filled with a material of choice, with no need for specialized equipment. Due to the semi-permeable nature of the shell, the infilled material can be cultured in a defined shape, and cells can be provided with nutrients and biological cues from the culture media, which stimulate the remodeling of the infill into a specified tissue. Finally, alginate shells could be readily removed after a culture period of three weeks, which could prove beneficial in cases where longer culture periods are required and improved nutrient diffusion is desirable.
Following this approach, a first test using several materials designed for cartilage tissue engineering was conducted. The focus was placed on materials rarely used for 3D cultures. These materials are, in fact, not stable on their own, as they would either dissolve due to their granular composition or exhibit instability in culture due to their low polymer content. Interestingly, after only 3 weeks of culture, all granular materials (HA-MA microstrands, Gloderm20, Gloderm30) were stabilized into one coherent structure through extracellular matrix deposition of embedded cells. This also marks the first use of a commercially available, injectable tissue filler as a tissue-engineering matrix. Even though microgels of hyaluronic acid gel swelled greatly, resulting in rather large microgel particles and a bloated tissue structure, such adaption of already applied solutions used in clinics has the potential to accelerate innovation of tissue-culture polymers.
The scalability of the presented technique was assessed, confirming its potential for the generation of centimeter-scale-sized implants, although issues such as the lower compressive stiffness should be addressed by increasing the maturation rate or time of the constructs. Although further studies are required to evaluate the effects of the alginate shells on the cultured materials, the use of alginate shells in biofabrication may open up unprecedented possibilities for tissue engineering and 3D cell culture.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: http:// hdl.handle.net/20.500.11850/599599.