Biofabrication of Heterogeneous, Multi‐Layered, and Human‐Scale Tissue Transplants Using Eluting Mold Casting

The creation of multi‐tissue auricular transplants for the treatment of microtia is a challenge due to the complex and layered structure of this anatomical tissue. A novel casting technique for the 3D biofabrication of heterogeneous, multi‐layered, and human‐scale tissue transplants using eluting agarose molds is presented. The molds are generated by casting agarose into custom 3D‐printed containers, termed metamolds, optimized to facilitate the hydrogel casting process based on geometric and topological constraints. Casting yields high resolution (50 µm) and allows for subsequent casting of further hydrogel layers on the transplant. Multi‐layered auricular constructs are fabricated on a cartilage core consisting of a hyaluronic acid‐alginate double network and an adjacent gelatin‐based dermal layer. Bonding between adjacent layers is achieved by orthogonal physical and enzymatic crosslinking of residual functional groups between each layer. Material composition and culture duration are optimized for each layer allowing for maturation into cartilaginous and pre‐vascularized dermal tissues. To demonstrate the scalability of this technique for the biofabrication of human‐sized transplants, bi‐layered human‐sized ears are cast. Overall, this novel casting technique offers a promising approach for the fabrication of complex tissue grafts, overcoming the limitations of other traditional biofabrication methods.


Introduction
Tissue engineering offers a promising approach to fabricating autologous tissues and organs, thereby reducing the risk of DOI: 10.1002/adfm.202305651immunogenic reactions. [1]7] Autologous costal cartilage ear reconstruction remains the preferred technique for outer ear repair due to its low rejection risk and minimal complication rate.However, the aesthetic outcome of autologous rib cartilage reconstruction heavily relies on the surgeon's skills. [8]Alloplastic prostheses can offer more aesthetically pleasing and detailed ear reconstruction, but their inert nature increases the risk of rejection and infection. [9]Additionally, these constructs often fail to replicate the anatomical features and multi-tissue composition of the transplant.Furthermore, damaged tissues, such as those affected by congenital deformations, trauma, or tumors, typically consist of more than two different tissue types and layers.Thus, the biofabrication of heterogeneous multilayered constructs becomes imperative to accurately replicate the anatomical structure of these tissues.
Due to the need for intricate and customized shapes to meet the requirements of medical applications, extensive research has been conducted to develop innovative techniques for producing biomimetic implants. [10]These techniques include extrusion bioprinting, [11] stereolithography, [12] and more recently, volumetric bioprinting. [10,13]Biofabrication methods such as extrusionand volumetric bioprinting offer precise control over the size, shape, and composition of tissue structures.However, these techniques necessitate specialized equipment and have stringent material requirements to ensure the mechanical stability of the printed structures.On the other hand, casting remains a versatile technique for fabricating tissues and implants in tissue engineering due to its simplicity. [14]It can be performed without the need for additional machinery, does not require specific material properties, and can achieve high resolution.Additionally, casting does not subject cells to additional stress or shear and eliminates the need for potentially cytotoxic components like photoinitiators.
In tissue engineering, cell-laden crosslinked hydrogels are commonly used due to their ability to provide an optimal and biomimetic environment for cells.[17] However, the diffusion of ions, enzymes, or cofactors into the molds can be limited.In traditional casting applications, ions can only diffuse from the top opening of the mold, as using other openings would result in material leakage before gelation occurs.This limited diffusion distance leads to challenges when fabricating larger samples, as diffusion time increases exponentially. [18]Consequently, current casting techniques for crosslinked hydrogels are primarily successful in creating small samples, while larger constructs pose significant issues regarding crosslinking time and cell viability. [19]The time required for the entire construct to fully crosslink increases exponentially due to the slower diffusion of ions during polymer crosslinking. [20]Prolonged crosslinking times within the molds are incompatible with hydrogelcontaining cells since cell survival depends on the diffusion of nutrients and gas exchange.Moreover, an uneven distribution of crosslinking can result in variations in the mechanical properties and functionality of the hydrogel, potentially impacting its ability to support cell growth and tissue formation. [21]he fabrication of complex, multi-tissue, and multi-layered constructs in biofabrication relies on bio-orthogonal chemistries that enable attachment between layers and provide biocompatible crosslinking mechanisms to support cell viability.Enzymatically crosslinkable polymers have gained significant attention in tissue engineering due to their physiological crosslinking mechanisms and substrate specificity. [22][24] Polymers modified with tyramine or other hydroxyphenyl groups can undergo gelation in the presence of HRP and H 2 O 2. [22] Moreover, the presence of free tyramine groups on the engineered tissue surface allows for adhesion to native tissues containing hydroxyphenyl groups. [22,25]The adhesion of the tissue-engineered construct to the surrounding tissues is key to its correct integration into the surrounding native tissue. [26]n this study, we introduce a novel casting technique for the biofabrication of a heterogeneous bi-layered construct comprising an avascular cartilaginous core and a pre-vascularized dermal layer.The inclusion of bioengineered microcapillaries aims to support the cartilaginous core by providing nutrients and oxygen beyond the diffusion limit to the surrounding dermal compartment once connected to the host's vasculature. [27]The presented technique is based on the generation of multi-part agarose molds using 3D DLP-printed metamolds, which are designed to respect the geometric and topological constraints involved in the process of casting. [28]The agarose molds were fabricated using a low concentration of agarose (3% w/v) and were preloaded with calcium chloride (CaCl 2 ) and hydrogen peroxide (H 2 O 2 ) for ionic crosslinking of alginate and covalent crosslinking of tyramine modified hyaluronan and gelatin-hydrogels, respectively.The subsequent casting of the cellular hydrogel precursor solution into consecutive agarose molds allowed for the precise generation of multi-layered hydrogels. [29]o demonstrate the scalability and potential of this technique, a two-layered, human-sized ear transplant was fabricated (Figure 1).The auricular cartilage core structure was generated by encapsulating human auricular chondrocytes within a hydrogel made of high-molecular-weight hyaluronan tyramine (HA-Tyr) and alginate (Alg).The second layer, mimicking the dermis, was formed by encapsulating HUVECs and primary human dermal fibroblasts within a gelatin tyramine (Gel-Tyr) hydrogel.This bioorthogonal strategy maintained the avascular nature of hyaline cartilage while also supporting the development of a microcapillary network in the adjacent dermal tissue layer.The use of tyramine crosslinking in both layers further allowed for the adhesion of adjacent cast cartilage core and dermal layer through covalent crosslinking.The possibility of producing patient-specific mold designs and the multi-layered and scalable approach of this technique make eluting molds an attractive solution for the fabrication of human-sized tissues and organs exhibiting a wide range of cell densities.

Multi-Layered Metamold Design
A new method for the fabrication of 3D hydrogel implants using multi-part molds was developed.Starting from a manifold triangle mesh of an ear, we developed an algorithm to calculate the "moldability field" by providing a user-defined potential direction for mold separation.The moldability field allowed us to determine the difficulty associated with the removal of the two-part mold once the ear is cast.The algorithm was iterated for different potential parting directions to identify the one associated with the lowest difficulty for removal of the cast implant.Once the optimal direction for removal was found, we generated a set of plastic molds called metamolds (Figure 2A), in which a molten agarose polymer solution is poured and cooled down to room temperature to fabricate a two-part agarose mold (Figure 2B).Next, the agarose molds were extracted from the metamolds and assembled (Figure 2C).A syringe was used to inject cell-laden hydrogel precursor solution into the assembled mold (Figure 2D; Video S1, Supporting Information).Based on the design of the molds, the two molds could be easily separated to reveal the cast implant (Figure 2E).Subsequent layers could be added to the implant using the same method (Figure 2F-L): the process was repeated for the new layer, starting with the top part, and the parting direction was kept the same.During the casting of the top part, the bottom section of the cartilage core remained attached to the initial set of molds used for casting that layer.Only the top mold was replaced with a new one, which included a gap between the cartilage core and the mold.The second mold for the generation of the dermal layer was designed to allow for a uniform gap size with regard to the cartilage core.This gap allowed for the injection of the cellladen polymer, ensuring that the thickness of the new layer was uniformly cast on top of the cartilage core.To cast the second part of the layer, the bottom mold was removed and replaced with a new mold, and the process was repeated (Figure 2F-L).

FEA of Diffusion of Crosslinking Ions and Cofactors into the Agarose Molds
To estimate the time required for the crosslinking of each layer, Finite Element Analysis (FEA) simulations were set up in A polymer solution containing Ha-Tyr/Alg/HA and human auricular chondrocytes is injected into the two-part mold.CaCl 2 and H 2 O 2 diffuse from the agarose molds into the polymer solution, triggering its crosslinking.The top agarose mold is removed, and the formed ear-shaped hydrogel construct is cultured for 42 days.B) The ear-shaped hydrogel is transferred to a new set of agarose molds for the casting of a second layer.The process described in (A) is repeated, where a polymer solution mixture comprising Gel-Tyr, endothelial cells, and fibroblasts is injected into the molds containing only H 2 O 2 .After crosslinking is complete, the top agarose mold is removed, and the two-layered construct is cultured for an additional 7 days.
COMSOL Multiphysics (Figure 3). Figure 3A shows the diffusion of 100 mM CaCl 2 and 0.01% H 2 O 2 , respectively, over time.The simulation estimated a time of 30 min for the 100 mM CaCl 2 to completely diffuse into the cast cartilage structure and 20 min for the 0.01% H 2 O 2 .Figure 3B shows the same simulation for the casting of the second layer.In this case, only H 2 O 2 is required for the crosslinking of the cast material.The time required for the diffusion of H 2 O 2 through the dermal layer is 15 min due to the smaller thickness of the layer.Since the casting of the sec-ond layer is performed in two steps, the overall casting time for a two-layered construct is 1 h (30 min for the cartilage core, 15 min for the first half of the second layer, and an additional 15 min for the remaining half).As a comparison, the bioprinting of a similarly sized construct printed at 100 μL min −1 would require ≈40 min, plus an average of 30 min for its crosslinking.In addition to the FEA simulations, further experimental assessments were performed to confirm a uniform diffusion of crosslinker across the layers.Six millimeters in diameter and 1.5 mm high samples were fabricated employing the casting technique of Figure 2. Subsequent to a crosslinking period of 15 min, the samples were extracted from the molds and sectioned transversely for detailed examination.No differences between the Hertz modulus between the center, middle, and edge were observed (Figure S8, Supporting Information, center: 6.3 ± 1.0 kPa, middle: 6.1 ± 1.2 kPa, and edge: 6.0 ± 1.1 kPa) indicating homogeneous crosslinking throughout the construct.

In Vitro Characterization of Cartilage Controls
Preliminary studies were performed to determine the optimal material composition for promoting cartilage maturation.Auricular chondrocytes were encapsulated in 6 mm-diameter and 1.5 mm-high discs and cultured in chondrogenic media to assess cartilage tissue maturation over time.At first, a screening experiment was conducted, in which three different hydrogel compositions were tested to evaluate their chondrogenic potential (Table S1 and Figure S1, Supporting Information).The first condition consists of 1.5% HA-Tyr + 0.3% Alg crosslinked with 0.01% H 2 O 2 and 100 mM CaCl 2 .The amount of HA-Tyr was reduced to 0.75% in the second condition, while 0.75% un-crosslinked HA was added and the alginate concentration was kept constant.HATyr-HA-Alg showed a faster and more pronounced stress-relaxation response compared to HATyr-HA as seen in the relaxation halftime and final relaxation percentage (Figure S10, Supporting Information).Condition 3 has the same composition as condition 2 but was crosslinked only by CaCl 2 .As shown in enzymatic degradation tests with hyaluronidase, crosslinking with both crosslinkers increases hydrogel stability compared to crosslinking with either one, thus affecting the hydrogel's network structure (Figure S2, Supporting Information).
The overall viability and histological analysis are shown in Figure S1 (Supporting Information).All conditions showed a cell viability above 90% at all timepoints (Figure S1A, Supporting Information).At day 63, the deposition of collagen fibers was observed (depicted in cyan) through a process known as secondharmonic, characteristic of two-photon imaging [30] (Figure S1A, Supporting Information).A significant increase in stiffness (p < 0.001) was observed between day 1 and day 21, as well as between day 21 and day 63 (p < 0.001) for all conditions (Figure S1C, Supporting Information), as a result of ECM deposition by cells (Figure S1B, Supporting Information).Among the three conditions, condition 2, with uncrosslinked HA and crosslinked with both crosslinkers, showed the best maturation, achieving a compressive modulus of 751.05 ± 80.30 kPa at day 63.This was significantly higher than condition 1 (p < 0.05) and condition 3 (p < 0.01); thus, we opted to use condition 2 for the remaining experiments.
RT-qPCR data confirmed that collagen I (COL1A1) gene expression was not significantly increased across the selected time points (Figure 4F).Conversely, cartilage extracellular ma-trix (ECM) hallmark genes collagen II (COL2A1) and aggrecan (ACAN) showed significant upregulation over time (Figure 4F).Furthermore, we observed steady levels of expression of the chondrocyte's master transcription factor (TF) SRY-Box Transcription Factor 9 (SOX9, Figure 4G).However, we detected a significant upregulation on day 49 of Runt-related transcription factor 2 (RUNX2, Figure 4G).Matrix metalloproteinase 13 (MMP13) and adamalysin-like metalloproteinase with thrombospondin motifs 5 (ADAMTS5) were both significantly upregulated, the former at day 42, and the latter at all time points (Figure 4H).

In Vitro Characterization of Vascularized Dermal Controls
To assess the vasculogenic potential of Gel-Tyr, different polymer concentrations were analyzed with respect to their ability to support vascular outgrowth and fusion of HUVECs, which was indicated by immunofluorescence staining with pan-endothelial CD31 antibody.Gelatin tyramine gels were prepared at concentrations of 3%, 4.5%, and 6%, each containing 1 × 10 6 cells mL −1 HUVECs and 1 × 10 5 cells mL −1 fibroblasts.Both 3% and 4.5% conditions allowed for the rapid development of vascular networks in the 3D matrix (Figure 5A).In contrast, a smaller number of capillaries was observed when using a 6% polymer concentration.Unconfined compression measurements revealed a consistent increase in the elastic modulus with rising polymer content, reaching an elastic modulus of 1709 ± 102 Pa at 6% Gel-Tyr (Figure 5B).The amounts of vascularized area, total vessel length, and junction number consistently showed a decrease with increasing polymer content.Significant differences were observed in the vascularized area and total vessel length between the 3% and 6% Gel-Tyr conditions (Figure 5C).Despite reduced vascular network formation, a concentration of 6% Gel-Tyr was chosen to provide sufficient mechanical stability following crosslinking.Co-localization of CD90 + fibroblasts and microcapillaries was demonstrated using a CD90 (Thy-1)/CD31 co-staining (Figure 5D).Long-term culturing of constructs resulted in the formation of highly interconnected capillary networks (Figure 5E).Lumen formation was clearly visible after 7 days (Figure 5F).

Casting of Human-Sized, Multi-Layered Ear
To show the scalability and potential of the metamold technique, a two-layered, human-sized ear transplant was fabricated.Casting was performed from the inside out, starting from the inner, auricular cartilage core as illustrated in Figure 6A.To generate the auricular cartilage core, 30 × 10 6 cells mL −1 auricular chondrocytes (AuChs) were encapsulated within a high-molecular-weight hyaluronan-based hydrogel composed of 0.75% HA-Tyr, 0.75% HA, and 0.3% Alg.The cell-laden polymer was loaded into a syringe and injected into the first set of agarose molds to cast the cartilage structure.An injection guide was manufactured (Figure S3, Supporting Information) to help with the positioning of the needle in the agarose molds.Crosslinking was initiated by the diffusion of CaCl 2 and H 2 O 2 from the agarose mold into the injected polymer solution for 30 min (Figure 6A, day 0 -cartilage).After crosslinking, one of the two molds was removed, leaving the second mold as a support for the cast construct.The latter was then placed in chondrogenic media and cultured for 4 days.
The casting of the second layer was performed after this preculturing period in two stages, wherein each layer was fabricated in two distinct steps, with only one half being cast at a time.Subsequently, the agarose molds used for the casting of the cartilage core were replaced by a new set of agarose molds.HUVECs (5 × 10 6 cells mL −1 ) and fibroblasts (0.1 × 10 6 cells mL −1 ) were embedded in 6% Gel-Tyr and injected into the void space between the cartilage core and the new set of agarose molds separately for the upper and lower half of the ear, respectively.Between the two steps, the polymer solution was left to crosslink for 15 min by allowing H 2 O 2 to diffuse from the agarose molds (Figure 6A, day 0 -dermis).Finally, the top agarose mold was discarded while the bottom agarose mold served as a resting surface for the multilayered construct during culture (Figure 6A, day 1).The ear was cultured for an additional week.As shown in Figure 6A, excellent shape retention of the cast construct was observed throughout the entire 49-day culturing period, due to the agarose mold, which supported the overhang of the multilayered ear during the initial timepoints when its stiffness was low.Furthermore, the increase in stiffness of the cartilage core further improved the mechanical stability of the construct.To evaluate the mechanical performance of the layered construct, 4 mm-diameter punches were extracted and compressed on days 21 and 42.At the final timepoint, a precise incision was made using a scalpel at the interface of the cartilage and dermal layers to evaluate the increase in compressive modulus of the cartilage core (Figure 6B).The cartilage core showed an increase of 36% in compressive modulus from day 21 (195.67 ± 18.12 kPa) to day 49 (265.12± 46.56 kPa).At the final timepoint, 3 regions from different parts of the matured ear construct (Figure 6C) were evaluated using histological analysis.As shown in Figure 6D, the quality of the produced cartilage was optimal, with strong staining for GAGs and collagen II, while collagen I appeared absent.

Discussion
A human-sized multilayered cellular ear was biofabricated using a novel and biocompatible casting strategy employing eluting molds.This method for producing hydrogel constructs has several advantages compared to common casting techniques used in tissue engineering.Most importantly, the process of extracting the cast object from a mold usually requires deciding on a direction in which the mold pieces can be separated without getting caught in overhanging parts of the object.This process, which typically involves manual work from an experienced designer, has been automated.Additionally, by optimizing the placement of the mold partitions, we have further reduced the risk of rupture of the cast implant while allowing for the subsequent casting of additional layers.We successfully demonstrated the fabrication of complex 3D multi-layered constructs using ionically and enzymatically crosslinked hydrogels by developing agarose molds loaded with CaCl 2 and H 2 O 2 .
Multi-layered grafts with precise tissue thickness were created by sequentially repeating the casting process.We validated this technique by fabricating a two-layered, cartilage-dermis auricular transplant, maturing into auricular cartilage.The use of agarose for the fabrication of the two-part ear-casting molds offers several advantages in this application.Agarose molds produced using metamolds respect the geometric and topological constraints involved in the process of casting for easy extraction of complex shapes.This bypasses the mechanical properties usually required in traditional casting to avoid deformation of the cast implant upon extraction from the molds.By using agarose as a porous molding material, we could pre-load our mold with crosslinkers of interest, to realize crosslinking of the cast hydrogel by diffusion through the mold.Preloading the agarose molds with CaCl 2 and H 2 O 2 enabled ionically and enzymatically crosslinked cast hydrogels.The large surface area of the agarose molds in contact glycosaminoglycans (Safranin O), Collagen I, and Collagen II deposition at day 49 (n = 3).An auricular cartilage sample was used as a positive control (n = 3).Scale bars: 100 μm.C) Viability quantification of Figure 4A (n = 3).D) Quantification of Safranin O, Collagen I, and Collagen II histological staining (n = 3).Normalized to human auricular tissue serving as control.E) Compression testing of 6 mm 1% HA, 0.5% HA-Tyr, and 0.3% Alginate discs over 49 days.F) qPCR of cartilage ECM hallmark genes shows optimal matrix maturation across the different timepoints (n = 3).G) qPCR of master transcription factors SOX9 and RUNX2.SOX9 levels did not significantly vary during sample maturations, while RUNX2 levels on day 49 were significantly increased compared to day 1 (n = 3).H) qPCR of matrix remodeling enzymes MMP13 and ADAMTS5.MMP13 was significantly increased only at day 42, while ADAMTS5 was significantly upregulated across all time points (n = 3).with the hydrogel precursor ensured uniform and fast diffusion of the crosslinker into the hydrogel, eliminating potential stress induced by manual mixing of the crosslinker with the hydrogel precursor.Moreover, the steady diffusion of the crosslinker from the agarose molds improved the crosslinking homogeneity of the cast material.Lastly, the agarose mold served as a neutral resting surface for the cultured ear, helping to retain its shape while permitting the diffusion of nutrients and growth factors from the culture media.Simultaneously, the poor cell attachment and migration into the agarose constructs reported in the literature [31] as advantageous in our setup as they inhibited cell migration from the cast construct into the agarose mold over time.It is worth noting that the ear could also be easily separated from the mold and transferred to a grid without any complications (Figure S7, Supporting Information).This approach requires additional manipulation of the construct but would ensure a higher diffusion of nutrients and gases to the part of the ear that otherwise rests on the agarose mold.
The presented technique offers several advantages over traditional mold casting.First, it allows for the casting of multilayer structures, enabling the creation of complex tissue architectures.[34][35][36][37][38][39] Furthermore, the gradual diffusion of crosslinkers through agarose molds of this technique reduces the stress exerted on cells caused by the mixing of crosslinkers with hydrogel precursors, a step commonly performed in traditional casting. [40]Another advantage lies in the porosity of agarose molds, which not only provides support for the cast implant during cell culture but also allow for the diffusion of gases and nutrients to the cells.This feature proves invaluable for casting delicate and soft hydrogels that may be easily damaged if transferred to a culture vessel without any support.Moreover, the presented technique surpasses one of the main limitations of bioprinting by enabling the casting of thin and soft biomimetic materials, which are highly conducive to cell culturing but pose challenges in terms of bioink properties such as rheology and crosslinking behavior. [41,42]Finally, the superior casting resolution achieved through this technique ensures precise replication of tissue structures, promoting optimal cell attachment and growth.This outcome is facilitated by orthogonal crosslinking, involving the use of enzymatically and ionically crosslinkable hydrogels in combination with H 2 O 2 /CaCl 2 diffusion through the agarose molds.Significantly, the resolution we obtain with our method, ≈50 μm (Figure S9, Supporting Information), is remarkable, particularly considering the technique's simplicity and cost-effectiveness.Extrusion bioprinting, on the one hand, typically achieves resolutions in the range of 100-500 μm, depending on the bioink properties and nozzle diameter. [43]On the other hand, microscopic biofabrication methods typically allow for resolutions in the order of 1 μm or less [44] depending on the specific technique and parameters used.For example, laser-assisted bioprinting and stereolithography, both under the umbrella of 3D bioprinting, can achieve resolutions as high as 10 and 1 μm, respectively. [44]Similarly, advanced micropatterning methods such as photolithography or soft lithography can reach resolutions below 1 μm. [45]It is important to note, however, that higher resolution often comes with increased complexity, time, and cost, highlighting the ongoing trade-off between resolution and cell viability in biofabrication.
We used results from previous studies [46] and computational tools to better predict the time required for each crosslinking step, thus further optimizing the casting procedure.The FEA diffusion study has multiple benefits: first, it ensures that the cast layer is completely crosslinked before the molds are removed, hence maintaining the stability and shape fidelity of the construct.Furthermore, it reduces unnecessary exposure of cells to CaCl 2 and H 2 O 2 for longer periods of time.Finally, it shortens the time required for the casting of the multilayered implant, allowing it to be transferred into a chondrogenic culture medium as soon as possible, which reduces the time cells are left without nutrients and growth factors.This increases cell viability and reduces the stress cells are subjected to.The use of FEA was particularly useful due to the complex geometry of the cast layers which would have required several rounds of trial and error and wasted materials to identify the optimal crosslinking time.
For the fabrication of the multi-layered tissue, we opted for an enzyme-mediated crosslinker (horseradish peroxidase) that enabled enzyme-catalyzed covalent bond formation between tyramine-modified biopolymers.The bio-orthogonality of this process does not interfere with cellular processes, [47] enabled hydrogel formation in each of the layers separately, and allowed for the attachment of two adjacent layers through crosslinking of free tyramine residues in each layer.The design of biopolymers and choice of encapsulated cells are optimized to maintain the avascular nature of hyaline cartilage while allowing the development of a microcapillary network in the adjacent dermal tissue layer.The material composition was optimized regarding various parameters such as the need for fast gelation, induction or inhibition of vascularization, availability of diffusible crosslinkers, and viscous behavior to prevent cell sedimentation.In addition, cell densities were optimized for each layer using small, 6.0 mm-diameter discs to assess cell viability and tissue maturation.Regarding cartilage tissue engineering, hyaluronic acid was used, as it is a key component of the ECM of many tissues including cartilage tissue, making it an ideal biopolymer for tissue engineering applications. [48]In addition, a small percentage of alginate was incorporated to replicate the negative charge characteristic of the ECM environment of chondrocytes.The polysaccharide nature of alginate can interact with proteoglycans and polyelectrolytes, promoting cell attachment and growth. [49]The addition of alginate enabled the tuning of the stiffness of the base material.This modulation of matrix stiffness plays an essential role in the correct maturation and growth of chondrocytes. [50]he addition of alginate particularly affects the viscoelastic properties of the material, thus providing stress-relaxing properties to the cells' microenvironment, similar to the native ECM of chondrocytes.In addition, studies have shown that these stressrelaxing properties promote a chondrogenic phenotype and cartilage tissue maturation. [51,52]The addition of alginate particularly affects the viscoelastic properties of the material, thus providing stress-relaxing properties to the cells' microenvironment, similar to the native ECM of chondrocytes.In addition, studies have shown that these stress-relaxing properties promote a chondrogenic phenotype and cartilage tissue maturation. [51,52]hile alginate is non-degradable in humans due to the lack of specific enzymes, [53] the viscoelastic properties of alginate allow cells to remodel their microenvironment. [54]Through reversible ionic crosslinks present in alginate sensitive to stresses exerted by cells, chondrocytes can proliferate and form a more interconnected ECM. [55,56]In particular, it was shown that the faster this process occurs, i.e., the faster the stress-relaxation response, the more cartilage ECM is being deposited and the higher the proliferation rates.Conversely, slow relaxing gels restricted proliferation, ECM deposition, and induced cell death and a catabolic phenotype. [57]n our screening experiment of scaffold composition for optimal cartilage tissue maturation, we explored the addition of unmodified hyaluronan to the HA-Tyr/Alg-based hydrogel (conditions 1 and 2 in Figure S1, Supporting Information).Comparing the final compressive modulus of the two constructs, we observed that the addition of uncrosslinked HA promoted matrix deposition and tissue maturation (Figure S1A, Supporting Information).This effect might be related to the added viscoelasticity by high molecular weight hyaluronan as well as the additional CD44 binding sites provided to chondrocytes.[58] Furthermore, we checked whether double crosslinking with both H 2 O 2 and CaCl 2 makes a difference in tissue maturation compared to single crosslinking with CaCl 2 (conditions 2 and 3 in Figure S1, Supporting Information).Herein, we also investigated whether H 2 O 2 has a negative effect on encapsulated cells and impairs their ability to produce ECM.The higher compressive modulus of constructs for double crosslinked samples at both day 21 and day 63 showed not only that H 2 O 2 had no cytotoxic effect on cells, but also that the double network structure of the hydrogel provided more support for cells, enhancing their metabolic activity and ECM secretion.The final compressive modulus of the scaffolds after 63 days of culture was 751 kPa, approaching the compressive modulus of native cartilage (≈1.6 MPa).[59] To facilitate the culture process, we reduced the culture time in our second experiment from 63 to 49 days and increased cell density from 10 × 10 6 to 30 × 10 6 cells mL −1 .The increased cell density had no negative effect on cell viability or ECM production, even though it resulted in a higher compressive modulus (880 kPa) of the construct at an earlier timepoint.
RT-qPCR data confirmed optimal cartilage maturation.The significant increase in COL2A1 and ACAN, coupled with low expression of COL1A1, indicated proper maturation of the samples.Downregulation of SOX9 has been reported in the literature as a direct consequence of chondrocyte dedifferentiation. [60,61]In this study, the constant levels of SOX9 expression indicate correct maintenance of the chondrocytes' phenotype.On the other hand, RUNX2 is essential for endochondral bone formation, and its expression in cartilage tissue is correlated with chondrocyte hypertrophy and loss of chondrogenic phenotype. [62]While RUNX2 levels were significantly upregulated, we acknowledge that the levels reached here are low compared to COL2A1, and they are not correlated with more abundant collagen I production. [63]n addition, hypertrophic chondrocytes were not observed in the histological analysis of the samples, suggesting that cartilage mineralization would require a higher level of RUNX2 expression.MMPs and ADAMTS are classes of proteolytic enzymes capable of degrading ECM proteins such as collagen and aggrecan. [64,65]While they are normally involved in tissue remodeling, uncontrolled activity can lead to pathological tissue destruction.The significant upregulation of MMP13 and ADAMTS5 that we observed was not associated with a downregulation of COL1A1/COL2A1 and ACAN gene expression.In addition, histological stains confirmed that no degradation occurred throughout sample maturation.Therefore, we hypothesized that the high levels of these enzymes reflect the highly dynamic process of ECM remodeling, rather than being an indication of matrix degradation.
To fabricate the dermal compartment of a human-sized ear, we chose gelatin-tyramine, as it can be covalently crosslinked to HA-Tyr in the presence of the enzyme horseradish peroxidase.[68] Importantly, capillary formation was reported even at relatively high (3-5% (w/v)) polymer contents. [69]Our co-culture experiments involving fibroblasts and HUVECs have demonstrated that lumenized vascular networks were readily formed within 7 days of culture.Results show similar neovascularization as previously reported for co-culture systems based on fibroblasts and HUVECs. [70]Even though lower polymer contents (3%, 4.5%) of Gel-Tyr exhibited better overall vascularization, the successful casting of the dermal layer required an increase in hydrogel stability.73] After optimization of the mold setup and cast hydrogel composition, we fabricated a two-layered human-sized ear transplant using our metamold system.The anatomical ear model was successfully fabricated using this method, and the harvesting of the construct after crosslinking within the mold did not impair the shape of the ear.Furthermore, the construct matured over time, and cells actively produced ECM, resulting in increased compressive modulus of the construct from day 1 to 21 and day 21 to 42.Compared to the cylindrical controls cultured previously, the stiffness of the hydrogel within the multilayered ear at the day 42 timepoint was lower, which was expected because of the increased diffusion distance to the material and slower diffusion rate due to the agarose mold support.Importantly, the tissue maturation process did not impair the shape of the ear, and the construct exhibited excellent shape retention after 6 weeks of culture.Following a 6-week long culturing of the cartilage core, the second layer was cast on top of the now matured ear construct using a new set of agarose molds.The second layer was uniform and successfully formed on top of the cartilage core.The adhesion between the two layers was realized through covalent crosslinking of gelatin-tyramine to the residual phenol groups present in either HA-Tyr or cell-derived ECM in the cartilage core.Histological analysis of the construct at the final timepoint showed that the quality of the produced cartilage in different parts of the construct is optimal and uniform.In addition, the dermal layer firmly adhered to the previously matured cartilage core.In terms of mechanical stability, the ears matured substantially (265.12 ± 46.56 kPa) but fell short of the modulus of human auricular cartilage (≈1.6 MPa). [59]Since the flexibility of the ear largely depends on its cartilage structure, the mechanical properties of its cartilage reflect the overall elasticity of the ear.It is therefore crucial to enhance the elasticity of the cartilage structure of the bi-layered construct in an attempt to match the properties of the normal human ear.To achieve this goal, the diffusion distances for nutrients and oxygen need to be enhanced and the culture conditions adjusted, particularly to match the tissue maturation of cylindrical controls.

Conclusion
A novel biofabrication method based on advanced casting strategies that employ eluting molds was used to fabricate large tissue transplants.Compared to traditional casting approaches, the method enables the biofabrication of more complex structures involving different layers composed of various materials and cell types.As there are no strict rheological requirements, such as in, e.g., extrusion bioprinting, this new biofabrication method allows a broad range of biomaterials to be used.Furthermore, different materials and cell types can be combined in a layerwise manner, independent of the chosen design.This could be of particular interest for biofabrication of complex structures with high resolution, which is not possible using current biofabrication methods. [74]Fabrication of the metamolds can be performed inexpensively by way of a service provider, with no upfront investment required for the purchase of a dedicated bioprinter.Also, except for the plastic metamolds, all other components used during the casting process can be purchased sterile and certified for medical use or are undergoing clinical trials. [75,76]In this work, we provide proof-of-concept for this technique, using two different hydrogels and multiple cell-types for the biofabrication of human-sized cell-laden constructs.Future work will focus on more intensive characterization of each layer after casting and on improving the culture condition for such large constructs in order to mitigate the effects of the diffusive limit through the agarose mold.This could be done using custom-made plastic grid supports that retain the construct without introducing physical barriers.
Agarose Molds and Metamold Cast Design: Metamolds were designed using MATLAB and Rhino 3D following a previously published methodology. [28]A flexible mold was developed to replicate the shape of a human ear pinna for this study.Achieving the intricate geometry of the pinna with a two-part mold became feasible by fabricating the molds from an elastic and stretchable material.The mold design utilized a computational approach capable of calculating the molds' parting surfaces while minimizing the necessary deformations required to extract the cast shape.To achieve this, multiple scalar fields were initially computed across the entire surface, corresponding to a range of candidate parting directions.The concept of "moldability costs," which reflected the level of challenge or complexity associated with creating a mold along a specific parting direction, was then quantified.This approach helped to streamline the parting direction with minimal moldability costs across all options.Finally, the STL files for the different components of the metamold were generated and prepared for printing.
Preparation of Agarose Molds: Metamolds were fabricated using a DLP Prusa SL1 printer (Prusa 3D) and subsequently assembled using autoclaving tape.After being sterilized with H 2 O 2 gas, a sterile solution containing 3% agarose, supplemented with 100 mM CaCl 2 , was poured into the metamolds under sterile conditions immediately following autoclaving.The solution was then left undisturbed to cool to room temperature until it solidified.Once solidified, the agarose molds designed for the cartilage core were soaked either in a solution of 100 mM CaCl2 and 0.01% H 2 O 2 or solely in 0.01% H 2 O 2 for 30 min before usage.Conversely, molds intended for the dermal layer were exclusively immersed in a 0.01% H 2 O 2 solution for the same duration prior to use.To confirm H 2 O 2 equilibrium throughout the agarose molds, random incisions were made in the molds at various locations, and no noticeable differences in crosslinking times were observed once gels were cast.Furthermore, considering the larger volume of the mold (119.6 mL) in comparison to the skin layer (4.2 mL), a consistent diffusion of H 2 O 2 at the interface between the mold and the skin layer was expected to remain constant.
FEA Simulations: FEA simulations were performed in COMSOL Multiphysics (v5.6)The STL file of the cartilage core of the average ear used for the generation of the agarose molds was imported in COMSOL.To represent the diffusion boundary between the agarose mold and the injected crosslinked hydrogel, a boundary condition was set on the outer surface of the ear with 100 mM CaCl 2 and 0.01% H 2 O 2, respectively, at time zero.
The diffusion coefficient for CaCl 2 and H 2 O 2 in agarose was taken from the literature. [77,78]Element size was set to 0.8 mm.The simulation was run for 30 min.A similar process was utilized to simulate the diffusion into the layer cast on top of the cartilage core.
HA-Tyr Synthesis: Hyaluronic acid tyramine (HA-Tyr) was synthesized using a previously reported protocol [79] with some modifications.Sodium hyaluronate (1.0 g, 2.5 mmol of COOH groups) was dissolved in 200 mL of MilliQ-water overnight with gentle stirring.EDC (0.95 g, 5 mmol) and NHS (1.15 g, 10 mmol) were added, and the solution was stirred for 30 min at room temperature.The aqueous solution of Tyr-HCL (0.86 g, 5 mmol) was added to the mixture to initiate the reaction; the pH value was controlled at 4.75, and the reaction was carried out overnight.Then the pH value was brought to 7.0 by adding 1 M NaOH solution to terminate the reaction.The product was precipitated in cold ethanol and dried in a vacuum oven.The dried polymer was dissolved in distilled and purified water and further purified by dialysis for 4 days against distilled water to get rid of residual unreacted reagents.The resultant HA-Tyr conjugate was collected as white foam after freeze-drying.The degree of substitution (DS), defined as the number of substituted groups per 100 disaccharide units, was calculated from 1 H-NMR (D 2 O) spectral data (Figure S4, Supporting Information) by comparing the integral values of the aromatic protons of tyramine (peaks at 6.86 and 7.17 ppm) and the methyl protons of HA (1.9 ppm).HA-Tyr with a degree of substitution (DS) of 10% was used in this study.
Gel-Tyr Synthesis: Gelatin-tyramine (Gel-Tyr) was synthesized as previously reported, [80] with minor modifications.Briefly, 4 g of gelatin type A (300 bloom) was dissolved in 200 mL of 50 mM MES solution and then heated to 60 °C to dissolve the gelatin.Afterward, the solution was cooled to room temperature.1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC; 1.5 × 10 −3 mol L −1 , Sigma) and N-hydroxysuccinimide (NHS; 0.75 × 10 −3 mol L −1 , Sigma) were added and the gelatin was allowed to activate for 15 mins.One gram of tyramine was added, and the mixture was reacted for 24 h at room temperature with gentle stirring.The modified gelatin solution was dialyzed against deionized water and using regenerated cellulose dialysis tubing (MWCO: 10 kDa, Sigma) for 5 days, with a minimal number of 6 dialysis changes.The polymer solution was subsequently lyophilized.The DS of the resultant Gel-Tyr conjugate was assessed from 1 H-NMR (D 2 O) spectral data (Figure S5, Supporting Information) containing 2 M NaCl and 0.5 mg mL −1 DSS.The tyramine proton peaks detected at ≈6.6-7.17ppm were adjusted for the phenylalanine signal detected in unmodified gelatin.To calculate the degree of substitution (in mmol g −1 ), the resulting integral was normalized to the methyl protons of 3 (Trimethylsilyl)−1-propanesulfonic acid (DSS) used as an internal standard (from ≈0.5 to −0.5 ppm).The degree of substitution of the gelatin tyramine used in this study was determined to be 0.074 mmol g −1 .
Enzymatic Degradation: Fifty microliters hydrogels were cast into PDMS molds and crosslinked using filter paper soaked in crosslinker solution comprising 100 mM CaCl 2 and 0.05% H 2 O 2 for 15 min.After crosslinking, the molds were removed, samples weighed and then placed in PBS containing 10 U mL −1 hyaluronidase (hyaluronidase from bovine testes).Samples were kept at 37 °C and fresh enzyme solution was added every day.Samples were weighed daily, and degradation was calculated by dividing the weight recorded on each day by the initial weight of the sample post-crosslinking (Figure S2, Supporting Information).n = 3 samples per condition were tested.
Rheological Characterization: Rheological analysis was carried out on an Anton Paar MCR 301 rheometer equipped with a 20 mm parallel plate geometry (PP20, Anton Paar) and a Peltier element with thermal hood (H-PTD 200, Anton Paar).Tests were performed at 25 °C and humidity in the thermal hood was controlled by placing a wet tissue inside to prevent the sample from drying out.Crosslinking dynamics were characterized by performing oscillatory measurements at a shear strain of 1% and a frequency of 1 Hz.Tests were carried out at a gap of 200 μm and measurement points were recorded every 10 s.
Cartilage Sample Preparation and Culture Agarose Mold Preparation: AuChs were embedded at a concentration of 10 or 30 × 10 6 cells mL −1 into a precursor mixture consisting of 0.75% HA-Tyr, 0.75%HA, and 0.3% alginate, except for the material screening experiment, where concentrations differed as stated.The viability before encapsulation was measured by trypan blue exclusion on a Countess device (Thermo).Gelation was achieved as described by Loebel et al. [83] Gel precursors were mixed in TRIS buffer (300 mm D-Glucose, 50 mm TRIS, pH 7.4) containing 3% (v/v) of 1 mg mL −1 or HRP.Crosslinking was initiated by adding 0.01% H 2 O 2 .Gels were cast in sterilized PDMS (SYLGARD 184) molds of 6 mm diameter and 1.5 mm height placed in a 12-well plate.Gels were left to crosslink for 15 min at 37 °C, after which chondrogenic medium was added.After one day, the PDMS molds were lifted to allow the gels to float freely.Samples were cultured in 12-well plates with DMEM supplemented with 10 ng mL −1 transforming growth factor 3 (TGF-3, Peprotech), 50 μg mL −1 L-ascorbate-2-phosphate, 40 μg mL −1 L-proline, 50 μg mL −1 gentamicin (Gibco), and 1% ITS+ Premix (Corning).The well plates were placed in tissue culture incubators at 37 °C and 5% CO 2 .Cultures were maintained for up to 49 days and media changes were performed twice a week.
Dermis Sample Preparation and Culture: Gel-Tyr precursor solutions were prepared by dissolving lyophilized Gel-Tyr at final concentrations of 3%, 4.5%, or 6% (w/v) in TRIS buffer (300 mm D-Glucose, 50 mm TRIS, pH 7.4) containing 3% (v/v) of 1 mg mL −1 HRP.HUVECs and FBs were embedded at final concentrations of 1 × 10 6 and 1 × 10 5 cells mL −1 , respectively.Gels were cast in sterilized PDMS (SYLGARD 184) molds of 6 mm diameter and 1.5 mm height placed in a 12-well plate.Crosslinking was initiated by placing a filter paper drenched in aqueous 0.01% H 2 O 2 on top of the precursor solution.Gels were then left to crosslink for 10 min on ice, after which EGM-2 was added to cover the gels.
Multilayer Ear Casting: Dimensions of the layered, cast ear were 57 mm x 39 mm x 26 mm, with a total volume of 5.1 mL for the cartilage core and 4.2 mL for the dermal layer.Casting of the multilayered ear was performed starting from the cartilage core.AuChs (30 × 10 6 mL −1 ) were encapsulated in 5.1 mL of 0.75% HA-Tyr, 0.75% HA, and 0.3% Alg.The solution was loaded into a 10 mL syringe and injected into the first set of agarose molds.An injection guide was used to help with the positioning of the needle during injection (Figure S3, Supporting Information).Crosslinking was performed for 30 min, where CaCl 2 and H 2 O 2 diffused from the agarose mold into the cast polymer solution.After crosslinking, one of the two parts of the mold was removed, leaving the second part of the mold to serve as a support for the cast cartilage core.The latter was then placed in chondrogenic media (DMEM supplemented with 10 ng mL −1 transforming growth factor 3 (TGF-3, Peprotech),

Figure 1 .
Figure 1.Experiment overview: A) Assembly of agarose molds containing CaCl 2 and H 2 O 2 .A polymer solution containing Ha-Tyr/Alg/HA and human auricular chondrocytes is injected into the two-part mold.CaCl 2 and H 2 O 2 diffuse from the agarose molds into the polymer solution, triggering its crosslinking.The top agarose mold is removed, and the formed ear-shaped hydrogel construct is cultured for 42 days.B) The ear-shaped hydrogel is transferred to a new set of agarose molds for the casting of a second layer.The process described in (A) is repeated, where a polymer solution mixture comprising Gel-Tyr, endothelial cells, and fibroblasts is injected into the molds containing only H 2 O 2 .After crosslinking is complete, the top agarose mold is removed, and the two-layered construct is cultured for an additional 7 days.

Figure 2 .
Figure 2. Fabrication of metamolds and agarose molds: A moldability field was calculated for different mold parting directions to find the surface for the easiest extraction of the cast implant.A) Using this algorithm, plastic molds called metamolds were generated and 3D printed using a commercial DLP printer.B) The plastic metamolds were filled with a pre-heated (80 °C) agarose gel precursor which solidifies at room temperature.C) The agarose molds were then extracted from the plastic metamolds, soaked for 30 min in a solution of CaCl 2 and H 2 O 2 , and assembled.D) Cell-laden hydrogel precursors were injected into the void between the agarose molds and crosslinked by diffusion of CaCl 2 and H 2 O 2 from the agarose mold.E) The cast implant was extracted along the optimal parting direction of the agarose molds.F) Agarose molds for additional layers are manufactured as shown in A and B but are soaked for 30 min in H 2 O 2 only.Additional layers are cast by replacing the two agarose molds used in steps C, D, and E in separate steps.G) Half of the additional layer is cast.H) The construct is flipped, and the second agarose mold is replaced.I) The second half of the additional layer is injected.L) The cast implant is extracted as in E. The construct can be placed in culture or additional layers can be further cast by repeating steps F-L with new agarose molds.

Figure 3 .
Figure 3. FEA simulation of the diffusion of 100 mM CaCl 2 and 0.01% H 2 O 2 from the agarose molds into the cast hydrogels: A) 30 min are required to achieve a concentration of 100 mM CaCl 2 throughout the cartilage structure, while only 20 min to achieve a concentration of 0.01% H 2 O 2 .B) Crosslinking of the dermal layer.Only H 2 O 2 is required for the crosslinking of the dermal layer.Since the thickness of this layer is smaller than the cartilage core, only 15 min are required for the 0.01% H 2 O 2 to diffuse throughout the whole layer.

Figure 4 .
Figure 4. A) Viability imaging was performed using a 2-photon microscope (green: viable, red: dead cells).Scalebar is 100 μm.The second harmonic generation shows collagen fiber deposition (cyan) from day 21.Excellent viability was observed at all timepoints.B) Histological analysis showing

Figure 5 .
Figure 5. Vascularization in dermal constructs A) Immunofluorescence stainings showing the extent of vascularization at D7 with different Gel-Tyr concentrations.B) Measurement of elastic modulus at different polymer concentrations (n = 3).C) A consistent negative trend in the vascularized area, total vessel length, and number of vascular junctions is observed as the polymer concentration is increased (n = 3).D) Co-localization of HUVECs and FBs, as revealed by endogenous GFP expression of HUVECs and CD90 immunofluorescence staining for FBs.E) Typical vascularization and maturation pattern observed for HUVECs at D1, D3, and D7, respectively.F) HUVECs form extensive networks with highly interconnected and widened, lumenized channels.Scale bars: 100 μm.

Figure 6 .
Figure 6.A) Casting of a two-layered, human-sized ear using metamolds.The cartilage core is cast (day 0).The construct was left on the bottom agarose mold used to fabricate the cartilage core and transferred to a culture vessel.The construct was cultured for 42 days.Scale bar: 1 cm B) Compressive modulus of 4 mm punches taken at D21 and D49 for the cartilage core (n = 3).C) Schematic description of ear sections shown in (D).D) Histological evaluation of D42 (a) cartilage-only sections as well as D49 (a,b,c) cartilage + dermis sections (n = 3).Scale bar: 100 μm.